diff --git a/README.md b/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..a7c0b1baf5b4703a3eb487e8c5f6033586ab7d57
--- /dev/null
+++ b/README.md
@@ -0,0 +1,88 @@
+---
+tags:
+- espnet
+- audio
+- automatic-speech-recognition
+- speech-translation
+language: multilingual
+datasets:
+- owsm_v3.1
+license: cc-by-4.0
+---
+
+## OWSM: Open Whisper-style Speech Model
+
+[OWSM](https://arxiv.org/abs/2309.13876) is an Open Whisper-style Speech Model from [CMU WAVLab](https://www.wavlab.org/). It reproduces Whisper-style training using publicly available data and an open-source toolkit [ESPnet](https://github.com/espnet/espnet).
+
+Our demo is available [here](https://huggingface.co/spaces/pyf98/OWSM_v3_demo). The [project page](https://www.wavlab.org/activities/2024/owsm/) contains various resources.
+
+**[OWSM v3.1](https://arxiv.org/abs/2401.16658) is an improved version of OWSM v3. It significantly outperforms OWSM v3 in almost all evaluation benchmarks.**
+We do not include any new training data. Instead, we utilize a state-of-the-art speech encoder, [E-Branchformer](https://arxiv.org/abs/2210.00077).
+
+This is a small-sized model with 367M parameters. It is trained on 180k hours of public speech data. Specifically, it supports the following speech-to-text tasks:
+- Speech recognition
+- Any-to-any-language speech translation
+- Utterance-level alignment
+- Long-form transcription
+- Language identification
+
+
+### Citing OWSM, Branchformers and ESPnet
+
+```BibTex
+@misc{peng2024owsm,
+      title={OWSM v3.1: Better and Faster Open Whisper-Style Speech Models based on E-Branchformer}, 
+      author={Yifan Peng and Jinchuan Tian and William Chen and Siddhant Arora and Brian Yan and Yui Sudo and Muhammad Shakeel and Kwanghee Choi and Jiatong Shi and Xuankai Chang and Jee-weon Jung and Shinji Watanabe},
+      year={2024},
+      eprint={2401.16658},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL}
+}
+@INPROCEEDINGS{owsm-asru23,
+  author={Peng, Yifan and Tian, Jinchuan and Yan, Brian and Berrebbi, Dan and Chang, Xuankai and Li, Xinjian and Shi, Jiatong and Arora, Siddhant and Chen, William and Sharma, Roshan and Zhang, Wangyou and Sudo, Yui and Shakeel, Muhammad and Jung, Jee-Weon and Maiti, Soumi and Watanabe, Shinji},
+  booktitle={2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)}, 
+  title={Reproducing Whisper-Style Training Using An Open-Source Toolkit And Publicly Available Data}, 
+  year={2023},
+  doi={10.1109/ASRU57964.2023.10389676}
+}
+@inproceedings{peng23b_interspeech,
+  author={Yifan Peng and Kwangyoun Kim and Felix Wu and Brian Yan and Siddhant Arora and William Chen and Jiyang Tang and Suwon Shon and Prashant Sridhar and Shinji Watanabe},
+  title={{A Comparative Study on E-Branchformer vs Conformer in Speech Recognition, Translation, and Understanding Tasks}},
+  year=2023,
+  booktitle={Proc. INTERSPEECH 2023},
+  pages={2208--2212},
+  doi={10.21437/Interspeech.2023-1194}
+}
+@inproceedings{kim2023branchformer,
+  title={E-branchformer: Branchformer with enhanced merging for speech recognition},
+  author={Kim, Kwangyoun and Wu, Felix and Peng, Yifan and Pan, Jing and Sridhar, Prashant and Han, Kyu J and Watanabe, Shinji},
+  booktitle={2022 IEEE Spoken Language Technology Workshop (SLT)},
+  pages={84--91},
+  year={2023},
+  organization={IEEE}
+}
+@InProceedings{pmlr-v162-peng22a,
+  title = 	 {Branchformer: Parallel {MLP}-Attention Architectures to Capture Local and Global Context for Speech Recognition and Understanding},
+  author =       {Peng, Yifan and Dalmia, Siddharth and Lane, Ian and Watanabe, Shinji},
+  booktitle = 	 {Proceedings of the 39th International Conference on Machine Learning},
+  pages = 	 {17627--17643},
+  year = 	 {2022},
+  editor = 	 {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan},
+  volume = 	 {162},
+  series = 	 {Proceedings of Machine Learning Research},
+  month = 	 {17--23 Jul},
+  publisher =    {PMLR},
+  pdf = 	 {https://proceedings.mlr.press/v162/peng22a/peng22a.pdf},
+  url = 	 {https://proceedings.mlr.press/v162/peng22a.html},
+  abstract = 	 {Conformer has proven to be effective in many speech processing tasks. It combines the benefits of extracting local dependencies using convolutions and global dependencies using self-attention. Inspired by this, we propose a more flexible, interpretable and customizable encoder alternative, Branchformer, with parallel branches for modeling various ranged dependencies in end-to-end speech processing. In each encoder layer, one branch employs self-attention or its variant to capture long-range dependencies, while the other branch utilizes an MLP module with convolutional gating (cgMLP) to extract local relationships. We conduct experiments on several speech recognition and spoken language understanding benchmarks. Results show that our model outperforms both Transformer and cgMLP. It also matches with or outperforms state-of-the-art results achieved by Conformer. Furthermore, we show various strategies to reduce computation thanks to the two-branch architecture, including the ability to have variable inference complexity in a single trained model. The weights learned for merging branches indicate how local and global dependencies are utilized in different layers, which benefits model designing.}
+}
+@inproceedings{watanabe2018espnet,
+  author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
+  title={{ESPnet}: End-to-End Speech Processing Toolkit},
+  year={2018},
+  booktitle={Proceedings of Interspeech},
+  pages={2207--2211},
+  doi={10.21437/Interspeech.2018-1456},
+  url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
+}
+```
diff --git a/data/token_list/bpe_unigram50000/bpe.model b/data/token_list/bpe_unigram50000/bpe.model
new file mode 100644
index 0000000000000000000000000000000000000000..3f386604ac50541de0d4500913350e158dbce5b8
--- /dev/null
+++ b/data/token_list/bpe_unigram50000/bpe.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d6327da127e870bcb8c737dceb3bd47ccbce63da74ddb094f64afe313d68c8c
+size 1041297
diff --git a/exp/s2t_stats_raw_bpe50000/train/feats_stats.npz b/exp/s2t_stats_raw_bpe50000/train/feats_stats.npz
new file mode 100644
index 0000000000000000000000000000000000000000..4ef154c285deb458cb537da751909790aea294d2
--- /dev/null
+++ b/exp/s2t_stats_raw_bpe50000/train/feats_stats.npz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ef4b5e465110edf32eec024cf2427eedd677f5733bb87d6b2131e6984a6e13f
+size 1402
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/RESULTS.md b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/RESULTS.md
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..38fb582b58b61c9ae44380b843722473a91c513d
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
@@ -0,0 +1,50271 @@
+config: conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml
+print_config: false
+log_level: INFO
+drop_last_iter: false
+dry_run: false
+iterator_type: sequence
+valid_iterator_type: null
+output_dir: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000
+ngpu: 1
+seed: 42
+num_workers: 4
+num_att_plot: 0
+dist_backend: nccl
+dist_init_method: file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6
+dist_world_size: 16
+dist_rank: 0
+local_rank: 0
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: slurm
+multiprocessing_distributed: true
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: false
+write_collected_feats: false
+max_epoch: 45
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - valid
+    - acc
+    - max
+-   - valid
+    - total_count
+    - max
+keep_nbest_models: 5
+nbest_averaging_interval: 0
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: true
+train_dtype: float32
+use_amp: true
+log_interval: null
+use_matplotlib: true
+use_tensorboard: true
+create_graph_in_tensorboard: false
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+use_lora: false
+save_lora_only: true
+lora_conf: {}
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: 15000
+batch_size: 256
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/s2t_stats_raw_bpe50000/splits12/speech_shape
+- exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe
+- exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe
+- exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe
+valid_shape_file:
+- exp/s2t_stats_raw_bpe50000/valid/speech_shape
+- exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe
+- exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe
+- exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe
+batch_type: unsorted
+valid_batch_type: null
+fold_length:
+- 80000
+- 150
+- 150
+- 150
+sort_in_batch: descending
+shuffle_within_batch: false
+sort_batch: descending
+multiple_iterator: true
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+chunk_excluded_key_prefixes: []
+chunk_default_fs: null
+train_data_path_and_name_and_type:
+-   - exp/s2t_stats_raw_bpe50000/splits12/wav.scp
+    - speech
+    - kaldi_ark
+-   - exp/s2t_stats_raw_bpe50000/splits12/text.prev
+    - text_prev
+    - text
+-   - exp/s2t_stats_raw_bpe50000/splits12/text.ctc
+    - text_ctc
+    - text
+-   - exp/s2t_stats_raw_bpe50000/splits12/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/dev_v3/wav.scp
+    - speech
+    - kaldi_ark
+-   - dump/raw/dev_v3/text.prev
+    - text_prev
+    - text
+-   - dump/raw/dev_v3/text.ctc
+    - text_ctc
+    - text
+-   - dump/raw/dev_v3/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+allow_multi_rates: false
+valid_max_cache_size: null
+exclude_weight_decay: false
+exclude_weight_decay_conf: {}
+optim: adamw
+optim_conf:
+    lr: 0.0005
+    betas:
+    - 0.9
+    - 0.98
+    eps: 1.0e-06
+    weight_decay: 0.0
+scheduler: piecewiselinearwarmuplr
+scheduler_conf:
+    warmup_steps_list:
+    - 0
+    - 30000
+    - 60000
+    warmup_lr_list:
+    - 0.0
+    - 5.0e-05
+    - 0.0005
+token_list:
+- <blank>
+- <unk>
+- <na>
+- <nospeech>
+- <abk>
+- <afr>
+- <amh>
+- <ara>
+- <asm>
+- <ast>
+- <aze>
+- <bak>
+- <bas>
+- <bel>
+- <ben>
+- <bos>
+- <bre>
+- <bul>
+- <cat>
+- <ceb>
+- <ces>
+- <chv>
+- <ckb>
+- <cmn>
+- <cnh>
+- <cym>
+- <dan>
+- <deu>
+- <dgd>
+- <div>
+- <ell>
+- <eng>
+- <epo>
+- <est>
+- <eus>
+- <fas>
+- <fil>
+- <fin>
+- <fra>
+- <frr>
+- <ful>
+- <gle>
+- <glg>
+- <grn>
+- <guj>
+- <hat>
+- <hau>
+- <heb>
+- <hin>
+- <hrv>
+- <hsb>
+- <hun>
+- <hye>
+- <ibo>
+- <ina>
+- <ind>
+- <isl>
+- <ita>
+- <jav>
+- <jpn>
+- <kab>
+- <kam>
+- <kan>
+- <kat>
+- <kaz>
+- <kea>
+- <khm>
+- <kin>
+- <kir>
+- <kmr>
+- <kor>
+- <lao>
+- <lav>
+- <lga>
+- <lin>
+- <lit>
+- <ltz>
+- <lug>
+- <luo>
+- <mal>
+- <mar>
+- <mas>
+- <mdf>
+- <mhr>
+- <mkd>
+- <mlt>
+- <mon>
+- <mri>
+- <mrj>
+- <mya>
+- <myv>
+- <nan>
+- <nep>
+- <nld>
+- <nno>
+- <nob>
+- <npi>
+- <nso>
+- <nya>
+- <oci>
+- <ori>
+- <orm>
+- <ory>
+- <pan>
+- <pol>
+- <por>
+- <pus>
+- <quy>
+- <roh>
+- <ron>
+- <rus>
+- <sah>
+- <sat>
+- <sin>
+- <skr>
+- <slk>
+- <slv>
+- <sna>
+- <snd>
+- <som>
+- <sot>
+- <spa>
+- <srd>
+- <srp>
+- <sun>
+- <swa>
+- <swe>
+- <swh>
+- <tam>
+- <tat>
+- <tel>
+- <tgk>
+- <tgl>
+- <tha>
+- <tig>
+- <tir>
+- <tok>
+- <tpi>
+- <tsn>
+- <tuk>
+- <tur>
+- <twi>
+- <uig>
+- <ukr>
+- <umb>
+- <urd>
+- <uzb>
+- <vie>
+- <vot>
+- <wol>
+- <xho>
+- <yor>
+- <yue>
+- <zho>
+- <zul>
+- <asr>
+- <st_ara>
+- <st_cat>
+- <st_ces>
+- <st_cym>
+- <st_deu>
+- <st_eng>
+- <st_est>
+- <st_fas>
+- <st_fra>
+- <st_ind>
+- <st_ita>
+- <st_jpn>
+- <st_lav>
+- <st_mon>
+- <st_nld>
+- <st_por>
+- <st_ron>
+- <st_rus>
+- <st_slv>
+- <st_spa>
+- <st_swe>
+- <st_tam>
+- <st_tur>
+- <st_vie>
+- <st_zho>
+- <notimestamps>
+- <0.00>
+- <0.02>
+- <0.04>
+- <0.06>
+- <0.08>
+- <0.10>
+- <0.12>
+- <0.14>
+- <0.16>
+- <0.18>
+- <0.20>
+- <0.22>
+- <0.24>
+- <0.26>
+- <0.28>
+- <0.30>
+- <0.32>
+- <0.34>
+- <0.36>
+- <0.38>
+- <0.40>
+- <0.42>
+- <0.44>
+- <0.46>
+- <0.48>
+- <0.50>
+- <0.52>
+- <0.54>
+- <0.56>
+- <0.58>
+- <0.60>
+- <0.62>
+- <0.64>
+- <0.66>
+- <0.68>
+- <0.70>
+- <0.72>
+- <0.74>
+- <0.76>
+- <0.78>
+- <0.80>
+- <0.82>
+- <0.84>
+- <0.86>
+- <0.88>
+- <0.90>
+- <0.92>
+- <0.94>
+- <0.96>
+- <0.98>
+- <1.00>
+- <1.02>
+- <1.04>
+- <1.06>
+- <1.08>
+- <1.10>
+- <1.12>
+- <1.14>
+- <1.16>
+- <1.18>
+- <1.20>
+- <1.22>
+- <1.24>
+- <1.26>
+- <1.28>
+- <1.30>
+- <1.32>
+- <1.34>
+- <1.36>
+- <1.38>
+- <1.40>
+- <1.42>
+- <1.44>
+- <1.46>
+- <1.48>
+- <1.50>
+- <1.52>
+- <1.54>
+- <1.56>
+- <1.58>
+- <1.60>
+- <1.62>
+- <1.64>
+- <1.66>
+- <1.68>
+- <1.70>
+- <1.72>
+- <1.74>
+- <1.76>
+- <1.78>
+- <1.80>
+- <1.82>
+- <1.84>
+- <1.86>
+- <1.88>
+- <1.90>
+- <1.92>
+- <1.94>
+- <1.96>
+- <1.98>
+- <2.00>
+- <2.02>
+- <2.04>
+- <2.06>
+- <2.08>
+- <2.10>
+- <2.12>
+- <2.14>
+- <2.16>
+- <2.18>
+- <2.20>
+- <2.22>
+- <2.24>
+- <2.26>
+- <2.28>
+- <2.30>
+- <2.32>
+- <2.34>
+- <2.36>
+- <2.38>
+- <2.40>
+- <2.42>
+- <2.44>
+- <2.46>
+- <2.48>
+- <2.50>
+- <2.52>
+- <2.54>
+- <2.56>
+- <2.58>
+- <2.60>
+- <2.62>
+- <2.64>
+- <2.66>
+- <2.68>
+- <2.70>
+- <2.72>
+- <2.74>
+- <2.76>
+- <2.78>
+- <2.80>
+- <2.82>
+- <2.84>
+- <2.86>
+- <2.88>
+- <2.90>
+- <2.92>
+- <2.94>
+- <2.96>
+- <2.98>
+- <3.00>
+- <3.02>
+- <3.04>
+- <3.06>
+- <3.08>
+- <3.10>
+- <3.12>
+- <3.14>
+- <3.16>
+- <3.18>
+- <3.20>
+- <3.22>
+- <3.24>
+- <3.26>
+- <3.28>
+- <3.30>
+- <3.32>
+- <3.34>
+- <3.36>
+- <3.38>
+- <3.40>
+- <3.42>
+- <3.44>
+- <3.46>
+- <3.48>
+- <3.50>
+- <3.52>
+- <3.54>
+- <3.56>
+- <3.58>
+- <3.60>
+- <3.62>
+- <3.64>
+- <3.66>
+- <3.68>
+- <3.70>
+- <3.72>
+- <3.74>
+- <3.76>
+- <3.78>
+- <3.80>
+- <3.82>
+- <3.84>
+- <3.86>
+- <3.88>
+- <3.90>
+- <3.92>
+- <3.94>
+- <3.96>
+- <3.98>
+- <4.00>
+- <4.02>
+- <4.04>
+- <4.06>
+- <4.08>
+- <4.10>
+- <4.12>
+- <4.14>
+- <4.16>
+- <4.18>
+- <4.20>
+- <4.22>
+- <4.24>
+- <4.26>
+- <4.28>
+- <4.30>
+- <4.32>
+- <4.34>
+- <4.36>
+- <4.38>
+- <4.40>
+- <4.42>
+- <4.44>
+- <4.46>
+- <4.48>
+- <4.50>
+- <4.52>
+- <4.54>
+- <4.56>
+- <4.58>
+- <4.60>
+- <4.62>
+- <4.64>
+- <4.66>
+- <4.68>
+- <4.70>
+- <4.72>
+- <4.74>
+- <4.76>
+- <4.78>
+- <4.80>
+- <4.82>
+- <4.84>
+- <4.86>
+- <4.88>
+- <4.90>
+- <4.92>
+- <4.94>
+- <4.96>
+- <4.98>
+- <5.00>
+- <5.02>
+- <5.04>
+- <5.06>
+- <5.08>
+- <5.10>
+- <5.12>
+- <5.14>
+- <5.16>
+- <5.18>
+- <5.20>
+- <5.22>
+- <5.24>
+- <5.26>
+- <5.28>
+- <5.30>
+- <5.32>
+- <5.34>
+- <5.36>
+- <5.38>
+- <5.40>
+- <5.42>
+- <5.44>
+- <5.46>
+- <5.48>
+- <5.50>
+- <5.52>
+- <5.54>
+- <5.56>
+- <5.58>
+- <5.60>
+- <5.62>
+- <5.64>
+- <5.66>
+- <5.68>
+- <5.70>
+- <5.72>
+- <5.74>
+- <5.76>
+- <5.78>
+- <5.80>
+- <5.82>
+- <5.84>
+- <5.86>
+- <5.88>
+- <5.90>
+- <5.92>
+- <5.94>
+- <5.96>
+- <5.98>
+- <6.00>
+- <6.02>
+- <6.04>
+- <6.06>
+- <6.08>
+- <6.10>
+- <6.12>
+- <6.14>
+- <6.16>
+- <6.18>
+- <6.20>
+- <6.22>
+- <6.24>
+- <6.26>
+- <6.28>
+- <6.30>
+- <6.32>
+- <6.34>
+- <6.36>
+- <6.38>
+- <6.40>
+- <6.42>
+- <6.44>
+- <6.46>
+- <6.48>
+- <6.50>
+- <6.52>
+- <6.54>
+- <6.56>
+- <6.58>
+- <6.60>
+- <6.62>
+- <6.64>
+- <6.66>
+- <6.68>
+- <6.70>
+- <6.72>
+- <6.74>
+- <6.76>
+- <6.78>
+- <6.80>
+- <6.82>
+- <6.84>
+- <6.86>
+- <6.88>
+- <6.90>
+- <6.92>
+- <6.94>
+- <6.96>
+- <6.98>
+- <7.00>
+- <7.02>
+- <7.04>
+- <7.06>
+- <7.08>
+- <7.10>
+- <7.12>
+- <7.14>
+- <7.16>
+- <7.18>
+- <7.20>
+- <7.22>
+- <7.24>
+- <7.26>
+- <7.28>
+- <7.30>
+- <7.32>
+- <7.34>
+- <7.36>
+- <7.38>
+- <7.40>
+- <7.42>
+- <7.44>
+- <7.46>
+- <7.48>
+- <7.50>
+- <7.52>
+- <7.54>
+- <7.56>
+- <7.58>
+- <7.60>
+- <7.62>
+- <7.64>
+- <7.66>
+- <7.68>
+- <7.70>
+- <7.72>
+- <7.74>
+- <7.76>
+- <7.78>
+- <7.80>
+- <7.82>
+- <7.84>
+- <7.86>
+- <7.88>
+- <7.90>
+- <7.92>
+- <7.94>
+- <7.96>
+- <7.98>
+- <8.00>
+- <8.02>
+- <8.04>
+- <8.06>
+- <8.08>
+- <8.10>
+- <8.12>
+- <8.14>
+- <8.16>
+- <8.18>
+- <8.20>
+- <8.22>
+- <8.24>
+- <8.26>
+- <8.28>
+- <8.30>
+- <8.32>
+- <8.34>
+- <8.36>
+- <8.38>
+- <8.40>
+- <8.42>
+- <8.44>
+- <8.46>
+- <8.48>
+- <8.50>
+- <8.52>
+- <8.54>
+- <8.56>
+- <8.58>
+- <8.60>
+- <8.62>
+- <8.64>
+- <8.66>
+- <8.68>
+- <8.70>
+- <8.72>
+- <8.74>
+- <8.76>
+- <8.78>
+- <8.80>
+- <8.82>
+- <8.84>
+- <8.86>
+- <8.88>
+- <8.90>
+- <8.92>
+- <8.94>
+- <8.96>
+- <8.98>
+- <9.00>
+- <9.02>
+- <9.04>
+- <9.06>
+- <9.08>
+- <9.10>
+- <9.12>
+- <9.14>
+- <9.16>
+- <9.18>
+- <9.20>
+- <9.22>
+- <9.24>
+- <9.26>
+- <9.28>
+- <9.30>
+- <9.32>
+- <9.34>
+- <9.36>
+- <9.38>
+- <9.40>
+- <9.42>
+- <9.44>
+- <9.46>
+- <9.48>
+- <9.50>
+- <9.52>
+- <9.54>
+- <9.56>
+- <9.58>
+- <9.60>
+- <9.62>
+- <9.64>
+- <9.66>
+- <9.68>
+- <9.70>
+- <9.72>
+- <9.74>
+- <9.76>
+- <9.78>
+- <9.80>
+- <9.82>
+- <9.84>
+- <9.86>
+- <9.88>
+- <9.90>
+- <9.92>
+- <9.94>
+- <9.96>
+- <9.98>
+- <10.00>
+- <10.02>
+- <10.04>
+- <10.06>
+- <10.08>
+- <10.10>
+- <10.12>
+- <10.14>
+- <10.16>
+- <10.18>
+- <10.20>
+- <10.22>
+- <10.24>
+- <10.26>
+- <10.28>
+- <10.30>
+- <10.32>
+- <10.34>
+- <10.36>
+- <10.38>
+- <10.40>
+- <10.42>
+- <10.44>
+- <10.46>
+- <10.48>
+- <10.50>
+- <10.52>
+- <10.54>
+- <10.56>
+- <10.58>
+- <10.60>
+- <10.62>
+- <10.64>
+- <10.66>
+- <10.68>
+- <10.70>
+- <10.72>
+- <10.74>
+- <10.76>
+- <10.78>
+- <10.80>
+- <10.82>
+- <10.84>
+- <10.86>
+- <10.88>
+- <10.90>
+- <10.92>
+- <10.94>
+- <10.96>
+- <10.98>
+- <11.00>
+- <11.02>
+- <11.04>
+- <11.06>
+- <11.08>
+- <11.10>
+- <11.12>
+- <11.14>
+- <11.16>
+- <11.18>
+- <11.20>
+- <11.22>
+- <11.24>
+- <11.26>
+- <11.28>
+- <11.30>
+- <11.32>
+- <11.34>
+- <11.36>
+- <11.38>
+- <11.40>
+- <11.42>
+- <11.44>
+- <11.46>
+- <11.48>
+- <11.50>
+- <11.52>
+- <11.54>
+- <11.56>
+- <11.58>
+- <11.60>
+- <11.62>
+- <11.64>
+- <11.66>
+- <11.68>
+- <11.70>
+- <11.72>
+- <11.74>
+- <11.76>
+- <11.78>
+- <11.80>
+- <11.82>
+- <11.84>
+- <11.86>
+- <11.88>
+- <11.90>
+- <11.92>
+- <11.94>
+- <11.96>
+- <11.98>
+- <12.00>
+- <12.02>
+- <12.04>
+- <12.06>
+- <12.08>
+- <12.10>
+- <12.12>
+- <12.14>
+- <12.16>
+- <12.18>
+- <12.20>
+- <12.22>
+- <12.24>
+- <12.26>
+- <12.28>
+- <12.30>
+- <12.32>
+- <12.34>
+- <12.36>
+- <12.38>
+- <12.40>
+- <12.42>
+- <12.44>
+- <12.46>
+- <12.48>
+- <12.50>
+- <12.52>
+- <12.54>
+- <12.56>
+- <12.58>
+- <12.60>
+- <12.62>
+- <12.64>
+- <12.66>
+- <12.68>
+- <12.70>
+- <12.72>
+- <12.74>
+- <12.76>
+- <12.78>
+- <12.80>
+- <12.82>
+- <12.84>
+- <12.86>
+- <12.88>
+- <12.90>
+- <12.92>
+- <12.94>
+- <12.96>
+- <12.98>
+- <13.00>
+- <13.02>
+- <13.04>
+- <13.06>
+- <13.08>
+- <13.10>
+- <13.12>
+- <13.14>
+- <13.16>
+- <13.18>
+- <13.20>
+- <13.22>
+- <13.24>
+- <13.26>
+- <13.28>
+- <13.30>
+- <13.32>
+- <13.34>
+- <13.36>
+- <13.38>
+- <13.40>
+- <13.42>
+- <13.44>
+- <13.46>
+- <13.48>
+- <13.50>
+- <13.52>
+- <13.54>
+- <13.56>
+- <13.58>
+- <13.60>
+- <13.62>
+- <13.64>
+- <13.66>
+- <13.68>
+- <13.70>
+- <13.72>
+- <13.74>
+- <13.76>
+- <13.78>
+- <13.80>
+- <13.82>
+- <13.84>
+- <13.86>
+- <13.88>
+- <13.90>
+- <13.92>
+- <13.94>
+- <13.96>
+- <13.98>
+- <14.00>
+- <14.02>
+- <14.04>
+- <14.06>
+- <14.08>
+- <14.10>
+- <14.12>
+- <14.14>
+- <14.16>
+- <14.18>
+- <14.20>
+- <14.22>
+- <14.24>
+- <14.26>
+- <14.28>
+- <14.30>
+- <14.32>
+- <14.34>
+- <14.36>
+- <14.38>
+- <14.40>
+- <14.42>
+- <14.44>
+- <14.46>
+- <14.48>
+- <14.50>
+- <14.52>
+- <14.54>
+- <14.56>
+- <14.58>
+- <14.60>
+- <14.62>
+- <14.64>
+- <14.66>
+- <14.68>
+- <14.70>
+- <14.72>
+- <14.74>
+- <14.76>
+- <14.78>
+- <14.80>
+- <14.82>
+- <14.84>
+- <14.86>
+- <14.88>
+- <14.90>
+- <14.92>
+- <14.94>
+- <14.96>
+- <14.98>
+- <15.00>
+- <15.02>
+- <15.04>
+- <15.06>
+- <15.08>
+- <15.10>
+- <15.12>
+- <15.14>
+- <15.16>
+- <15.18>
+- <15.20>
+- <15.22>
+- <15.24>
+- <15.26>
+- <15.28>
+- <15.30>
+- <15.32>
+- <15.34>
+- <15.36>
+- <15.38>
+- <15.40>
+- <15.42>
+- <15.44>
+- <15.46>
+- <15.48>
+- <15.50>
+- <15.52>
+- <15.54>
+- <15.56>
+- <15.58>
+- <15.60>
+- <15.62>
+- <15.64>
+- <15.66>
+- <15.68>
+- <15.70>
+- <15.72>
+- <15.74>
+- <15.76>
+- <15.78>
+- <15.80>
+- <15.82>
+- <15.84>
+- <15.86>
+- <15.88>
+- <15.90>
+- <15.92>
+- <15.94>
+- <15.96>
+- <15.98>
+- <16.00>
+- <16.02>
+- <16.04>
+- <16.06>
+- <16.08>
+- <16.10>
+- <16.12>
+- <16.14>
+- <16.16>
+- <16.18>
+- <16.20>
+- <16.22>
+- <16.24>
+- <16.26>
+- <16.28>
+- <16.30>
+- <16.32>
+- <16.34>
+- <16.36>
+- <16.38>
+- <16.40>
+- <16.42>
+- <16.44>
+- <16.46>
+- <16.48>
+- <16.50>
+- <16.52>
+- <16.54>
+- <16.56>
+- <16.58>
+- <16.60>
+- <16.62>
+- <16.64>
+- <16.66>
+- <16.68>
+- <16.70>
+- <16.72>
+- <16.74>
+- <16.76>
+- <16.78>
+- <16.80>
+- <16.82>
+- <16.84>
+- <16.86>
+- <16.88>
+- <16.90>
+- <16.92>
+- <16.94>
+- <16.96>
+- <16.98>
+- <17.00>
+- <17.02>
+- <17.04>
+- <17.06>
+- <17.08>
+- <17.10>
+- <17.12>
+- <17.14>
+- <17.16>
+- <17.18>
+- <17.20>
+- <17.22>
+- <17.24>
+- <17.26>
+- <17.28>
+- <17.30>
+- <17.32>
+- <17.34>
+- <17.36>
+- <17.38>
+- <17.40>
+- <17.42>
+- <17.44>
+- <17.46>
+- <17.48>
+- <17.50>
+- <17.52>
+- <17.54>
+- <17.56>
+- <17.58>
+- <17.60>
+- <17.62>
+- <17.64>
+- <17.66>
+- <17.68>
+- <17.70>
+- <17.72>
+- <17.74>
+- <17.76>
+- <17.78>
+- <17.80>
+- <17.82>
+- <17.84>
+- <17.86>
+- <17.88>
+- <17.90>
+- <17.92>
+- <17.94>
+- <17.96>
+- <17.98>
+- <18.00>
+- <18.02>
+- <18.04>
+- <18.06>
+- <18.08>
+- <18.10>
+- <18.12>
+- <18.14>
+- <18.16>
+- <18.18>
+- <18.20>
+- <18.22>
+- <18.24>
+- <18.26>
+- <18.28>
+- <18.30>
+- <18.32>
+- <18.34>
+- <18.36>
+- <18.38>
+- <18.40>
+- <18.42>
+- <18.44>
+- <18.46>
+- <18.48>
+- <18.50>
+- <18.52>
+- <18.54>
+- <18.56>
+- <18.58>
+- <18.60>
+- <18.62>
+- <18.64>
+- <18.66>
+- <18.68>
+- <18.70>
+- <18.72>
+- <18.74>
+- <18.76>
+- <18.78>
+- <18.80>
+- <18.82>
+- <18.84>
+- <18.86>
+- <18.88>
+- <18.90>
+- <18.92>
+- <18.94>
+- <18.96>
+- <18.98>
+- <19.00>
+- <19.02>
+- <19.04>
+- <19.06>
+- <19.08>
+- <19.10>
+- <19.12>
+- <19.14>
+- <19.16>
+- <19.18>
+- <19.20>
+- <19.22>
+- <19.24>
+- <19.26>
+- <19.28>
+- <19.30>
+- <19.32>
+- <19.34>
+- <19.36>
+- <19.38>
+- <19.40>
+- <19.42>
+- <19.44>
+- <19.46>
+- <19.48>
+- <19.50>
+- <19.52>
+- <19.54>
+- <19.56>
+- <19.58>
+- <19.60>
+- <19.62>
+- <19.64>
+- <19.66>
+- <19.68>
+- <19.70>
+- <19.72>
+- <19.74>
+- <19.76>
+- <19.78>
+- <19.80>
+- <19.82>
+- <19.84>
+- <19.86>
+- <19.88>
+- <19.90>
+- <19.92>
+- <19.94>
+- <19.96>
+- <19.98>
+- <20.00>
+- <20.02>
+- <20.04>
+- <20.06>
+- <20.08>
+- <20.10>
+- <20.12>
+- <20.14>
+- <20.16>
+- <20.18>
+- <20.20>
+- <20.22>
+- <20.24>
+- <20.26>
+- <20.28>
+- <20.30>
+- <20.32>
+- <20.34>
+- <20.36>
+- <20.38>
+- <20.40>
+- <20.42>
+- <20.44>
+- <20.46>
+- <20.48>
+- <20.50>
+- <20.52>
+- <20.54>
+- <20.56>
+- <20.58>
+- <20.60>
+- <20.62>
+- <20.64>
+- <20.66>
+- <20.68>
+- <20.70>
+- <20.72>
+- <20.74>
+- <20.76>
+- <20.78>
+- <20.80>
+- <20.82>
+- <20.84>
+- <20.86>
+- <20.88>
+- <20.90>
+- <20.92>
+- <20.94>
+- <20.96>
+- <20.98>
+- <21.00>
+- <21.02>
+- <21.04>
+- <21.06>
+- <21.08>
+- <21.10>
+- <21.12>
+- <21.14>
+- <21.16>
+- <21.18>
+- <21.20>
+- <21.22>
+- <21.24>
+- <21.26>
+- <21.28>
+- <21.30>
+- <21.32>
+- <21.34>
+- <21.36>
+- <21.38>
+- <21.40>
+- <21.42>
+- <21.44>
+- <21.46>
+- <21.48>
+- <21.50>
+- <21.52>
+- <21.54>
+- <21.56>
+- <21.58>
+- <21.60>
+- <21.62>
+- <21.64>
+- <21.66>
+- <21.68>
+- <21.70>
+- <21.72>
+- <21.74>
+- <21.76>
+- <21.78>
+- <21.80>
+- <21.82>
+- <21.84>
+- <21.86>
+- <21.88>
+- <21.90>
+- <21.92>
+- <21.94>
+- <21.96>
+- <21.98>
+- <22.00>
+- <22.02>
+- <22.04>
+- <22.06>
+- <22.08>
+- <22.10>
+- <22.12>
+- <22.14>
+- <22.16>
+- <22.18>
+- <22.20>
+- <22.22>
+- <22.24>
+- <22.26>
+- <22.28>
+- <22.30>
+- <22.32>
+- <22.34>
+- <22.36>
+- <22.38>
+- <22.40>
+- <22.42>
+- <22.44>
+- <22.46>
+- <22.48>
+- <22.50>
+- <22.52>
+- <22.54>
+- <22.56>
+- <22.58>
+- <22.60>
+- <22.62>
+- <22.64>
+- <22.66>
+- <22.68>
+- <22.70>
+- <22.72>
+- <22.74>
+- <22.76>
+- <22.78>
+- <22.80>
+- <22.82>
+- <22.84>
+- <22.86>
+- <22.88>
+- <22.90>
+- <22.92>
+- <22.94>
+- <22.96>
+- <22.98>
+- <23.00>
+- <23.02>
+- <23.04>
+- <23.06>
+- <23.08>
+- <23.10>
+- <23.12>
+- <23.14>
+- <23.16>
+- <23.18>
+- <23.20>
+- <23.22>
+- <23.24>
+- <23.26>
+- <23.28>
+- <23.30>
+- <23.32>
+- <23.34>
+- <23.36>
+- <23.38>
+- <23.40>
+- <23.42>
+- <23.44>
+- <23.46>
+- <23.48>
+- <23.50>
+- <23.52>
+- <23.54>
+- <23.56>
+- <23.58>
+- <23.60>
+- <23.62>
+- <23.64>
+- <23.66>
+- <23.68>
+- <23.70>
+- <23.72>
+- <23.74>
+- <23.76>
+- <23.78>
+- <23.80>
+- <23.82>
+- <23.84>
+- <23.86>
+- <23.88>
+- <23.90>
+- <23.92>
+- <23.94>
+- <23.96>
+- <23.98>
+- <24.00>
+- <24.02>
+- <24.04>
+- <24.06>
+- <24.08>
+- <24.10>
+- <24.12>
+- <24.14>
+- <24.16>
+- <24.18>
+- <24.20>
+- <24.22>
+- <24.24>
+- <24.26>
+- <24.28>
+- <24.30>
+- <24.32>
+- <24.34>
+- <24.36>
+- <24.38>
+- <24.40>
+- <24.42>
+- <24.44>
+- <24.46>
+- <24.48>
+- <24.50>
+- <24.52>
+- <24.54>
+- <24.56>
+- <24.58>
+- <24.60>
+- <24.62>
+- <24.64>
+- <24.66>
+- <24.68>
+- <24.70>
+- <24.72>
+- <24.74>
+- <24.76>
+- <24.78>
+- <24.80>
+- <24.82>
+- <24.84>
+- <24.86>
+- <24.88>
+- <24.90>
+- <24.92>
+- <24.94>
+- <24.96>
+- <24.98>
+- <25.00>
+- <25.02>
+- <25.04>
+- <25.06>
+- <25.08>
+- <25.10>
+- <25.12>
+- <25.14>
+- <25.16>
+- <25.18>
+- <25.20>
+- <25.22>
+- <25.24>
+- <25.26>
+- <25.28>
+- <25.30>
+- <25.32>
+- <25.34>
+- <25.36>
+- <25.38>
+- <25.40>
+- <25.42>
+- <25.44>
+- <25.46>
+- <25.48>
+- <25.50>
+- <25.52>
+- <25.54>
+- <25.56>
+- <25.58>
+- <25.60>
+- <25.62>
+- <25.64>
+- <25.66>
+- <25.68>
+- <25.70>
+- <25.72>
+- <25.74>
+- <25.76>
+- <25.78>
+- <25.80>
+- <25.82>
+- <25.84>
+- <25.86>
+- <25.88>
+- <25.90>
+- <25.92>
+- <25.94>
+- <25.96>
+- <25.98>
+- <26.00>
+- <26.02>
+- <26.04>
+- <26.06>
+- <26.08>
+- <26.10>
+- <26.12>
+- <26.14>
+- <26.16>
+- <26.18>
+- <26.20>
+- <26.22>
+- <26.24>
+- <26.26>
+- <26.28>
+- <26.30>
+- <26.32>
+- <26.34>
+- <26.36>
+- <26.38>
+- <26.40>
+- <26.42>
+- <26.44>
+- <26.46>
+- <26.48>
+- <26.50>
+- <26.52>
+- <26.54>
+- <26.56>
+- <26.58>
+- <26.60>
+- <26.62>
+- <26.64>
+- <26.66>
+- <26.68>
+- <26.70>
+- <26.72>
+- <26.74>
+- <26.76>
+- <26.78>
+- <26.80>
+- <26.82>
+- <26.84>
+- <26.86>
+- <26.88>
+- <26.90>
+- <26.92>
+- <26.94>
+- <26.96>
+- <26.98>
+- <27.00>
+- <27.02>
+- <27.04>
+- <27.06>
+- <27.08>
+- <27.10>
+- <27.12>
+- <27.14>
+- <27.16>
+- <27.18>
+- <27.20>
+- <27.22>
+- <27.24>
+- <27.26>
+- <27.28>
+- <27.30>
+- <27.32>
+- <27.34>
+- <27.36>
+- <27.38>
+- <27.40>
+- <27.42>
+- <27.44>
+- <27.46>
+- <27.48>
+- <27.50>
+- <27.52>
+- <27.54>
+- <27.56>
+- <27.58>
+- <27.60>
+- <27.62>
+- <27.64>
+- <27.66>
+- <27.68>
+- <27.70>
+- <27.72>
+- <27.74>
+- <27.76>
+- <27.78>
+- <27.80>
+- <27.82>
+- <27.84>
+- <27.86>
+- <27.88>
+- <27.90>
+- <27.92>
+- <27.94>
+- <27.96>
+- <27.98>
+- <28.00>
+- <28.02>
+- <28.04>
+- <28.06>
+- <28.08>
+- <28.10>
+- <28.12>
+- <28.14>
+- <28.16>
+- <28.18>
+- <28.20>
+- <28.22>
+- <28.24>
+- <28.26>
+- <28.28>
+- <28.30>
+- <28.32>
+- <28.34>
+- <28.36>
+- <28.38>
+- <28.40>
+- <28.42>
+- <28.44>
+- <28.46>
+- <28.48>
+- <28.50>
+- <28.52>
+- <28.54>
+- <28.56>
+- <28.58>
+- <28.60>
+- <28.62>
+- <28.64>
+- <28.66>
+- <28.68>
+- <28.70>
+- <28.72>
+- <28.74>
+- <28.76>
+- <28.78>
+- <28.80>
+- <28.82>
+- <28.84>
+- <28.86>
+- <28.88>
+- <28.90>
+- <28.92>
+- <28.94>
+- <28.96>
+- <28.98>
+- <29.00>
+- <29.02>
+- <29.04>
+- <29.06>
+- <29.08>
+- <29.10>
+- <29.12>
+- <29.14>
+- <29.16>
+- <29.18>
+- <29.20>
+- <29.22>
+- <29.24>
+- <29.26>
+- <29.28>
+- <29.30>
+- <29.32>
+- <29.34>
+- <29.36>
+- <29.38>
+- <29.40>
+- <29.42>
+- <29.44>
+- <29.46>
+- <29.48>
+- <29.50>
+- <29.52>
+- <29.54>
+- <29.56>
+- <29.58>
+- <29.60>
+- <29.62>
+- <29.64>
+- <29.66>
+- <29.68>
+- <29.70>
+- <29.72>
+- <29.74>
+- <29.76>
+- <29.78>
+- <29.80>
+- <29.82>
+- <29.84>
+- <29.86>
+- <29.88>
+- <29.90>
+- <29.92>
+- <29.94>
+- <29.96>
+- <29.98>
+- <30.00>
+- ▁
+- ','
+- ▁the
+- .
+- s
+- ▁and
+- ▁of
+- ▁to
+- ▁a
+- 。
+- ▁in
+- ''''
+- ▁that
+- ▁i
+- ▁it
+- ▁was
+- en
+- e
+- t
+- ▁you
+- ▁he
+- ▁is
+- 的
+- '-'
+- ▁for
+- ▁de
+- ▁with
+- ▁be
+- n
+- d
+- ▁as
+- ▁his
+- ▁we
+- の
+- ▁on
+- 、
+- ▁die
+- ▁had
+- a
+- ▁but
+- ▁so
+- '?'
+- o
+- ▁not
+- ▁at
+- ▁have
+- ▁this
+- ▁und
+- er
+- ▁her
+- i
+- ▁they
+- m
+- ed
+- re
+- ▁she
+- ▁an
+- ▁by
+- ▁der
+- ing
+- ▁all
+- ▁are
+- ▁la
+- が
+- を
+- ▁from
+- ▁me
+- ▁which
+- ▁my
+- es
+- ▁one
+- は
+- ▁no
+- ▁there
+- 了
+- ▁him
+- ▁or
+- ▁were
+- に
+- y
+- ▁do
+- ▁what
+- r
+- ▁zu
+- ▁our
+- u
+- ▁if
+- ly
+- ▁ist
+- ▁would
+- ▁when
+- ▁their
+- ▁will
+- ▁said
+- で
+- ▁es
+- ▁like
+- ▁das
+- ▁Sie
+- ▁them
+- 在
+- ▁can
+- ▁who
+- ▁out
+- ▁know
+- te
+- ▁been
+- ▁ich
+- ▁about
+- ▁up
+- ▁I
+- '!'
+- ▁more
+- ▁man
+- ve
+- 我
+- ▁que
+- ▁un
+- 是
+- ▁sie
+- 和
+- ▁then
+- ▁your
+- ▁ein
+- ▁some
+- ▁den
+- а
+- ▁now
+- 你
+- ▁von
+- ▁nicht
+- ▁very
+- と
+- ▁has
+- ▁into
+- in
+- ▁time
+- ▁в
+- ar
+- ▁just
+- ▁не
+- ▁um
+- k
+- ▁could
+- ▁dass
+- ▁Und
+- 人
+- ▁think
+- ▁auf
+- al
+- de
+- ▁и
+- ▁war
+- ▁eine
+- ▁over
+- ▁mit
+- ▁well
+- も
+- 他
+- ▁see
+- ▁also
+- ▁wir
+- ▁other
+- ▁des
+- le
+- ▁how
+- ll
+- ▁than
+- 不
+- 啊
+- ▁these
+- ▁little
+- ▁sich
+- ▁"
+- ten
+- c
+- an
+- ▁us
+- 上
+- ▁l
+- ▁two
+- ▁any
+- 有
+- ▁don
+- ▁go
+- ▁did
+- ▁people
+- ▁only
+- е
+- na
+- ▁good
+- l
+- st
+- ▁where
+- ▁на
+- ▁el
+- ▁se
+- z
+- to
+- 中
+- ▁first
+- g
+- ta
+- ▁na
+- ▁di
+- ▁als
+- is
+- ▁come
+- ▁Ich
+- ▁much
+- ▁für
+- ne
+- ▁с
+- 我们
+- 大
+- ▁get
+- ▁here
+- ▁down
+- у
+- ▁du
+- ▁le
+- ▁что
+- から
+- ▁way
+- h
+- ▁y
+- ▁wie
+- ▁should
+- ▁before
+- ▁am
+- ▁made
+- ▁those
+- ▁after
+- 一
+- 'on'
+- ▁upon
+- ▁because
+- ▁back
+- ▁right
+- ▁haben
+- se
+- ▁great
+- 他们
+- ▁say
+- ▁going
+- です
+- 这个
+- ▁dem
+- и
+- 就
+- f
+- ▁по
+- ▁make
+- ra
+- 来
+- ▁The
+- م
+- ▁er
+- ▁its
+- ▁such
+- ▁ver
+- ▁may
+- w
+- ge
+- ▁even
+- la
+- ▁men
+- 说
+- ':'
+- ▁new
+- ▁im
+- 地
+- é
+- da
+- ch
+- ▁through
+- 呢
+- ▁long
+- な
+- me
+- ▁never
+- ▁most
+- ▁Es
+- as
+- il
+- p
+- ▁A
+- ▁must
+- '"'
+- ▁sind
+- ▁day
+- し
+- 一个
+- ▁really
+- ▁per
+- b
+- ▁came
+- ▁con
+- '1'
+- 这
+- man
+- ▁And
+- ▁life
+- ▁many
+- м
+- ▁old
+- est
+- я
+- j
+- ▁я
+- ▁yeah
+- 'no'
+- ro
+- ▁again
+- т
+- ▁things
+- ▁La
+- ▁hat
+- be
+- ma
+- ▁mu
+- ▁own
+- 她
+- den
+- ▁take
+- ci
+- ▁et
+- ▁aus
+- ▁being
+- ▁je
+- ga
+- os
+- 它
+- ni
+- we
+- ▁mr
+- ▁still
+- ir
+- ▁我
+- '2'
+- ▁last
+- か
+- 小
+- ▁too
+- ▁les
+- ▁might
+- 会
+- ▁work
+- us
+- ▁у
+- ce
+- や
+- ▁every
+- un
+- ▁We
+- 子
+- ▁ku
+- ▁al
+- 去
+- ▁за
+- ▁So
+- ы
+- at
+- ▁want
+- る
+- ی
+- ▁Die
+- 到
+- it
+- ▁went
+- ▁look
+- ▁über
+- х
+- ▁In
+- ti
+- ▁years
+- 对
+- ’
+- 吗
+- ▁это
+- ▁same
+- ▁three
+- ▁something
+- ▁da
+- do
+- li
+- ▁werden
+- ▁let
+- ▁thought
+- ▁himself
+- ▁ne
+- й
+- ▁ah
+- 被
+- ▁van
+- って
+- ▁got
+- ▁ge
+- д
+- ▁world
+- ね
+- い
+- ▁wenn
+- va
+- 好
+- ▁einen
+- 要
+- ▁thing
+- ▁while
+- ▁à
+- ▁part
+- 都
+- ه
+- с
+- 着
+- ▁year
+- el
+- ▁del
+- ▁oh
+- ▁away
+- ▁hand
+- ▁Er
+- em
+- et
+- ho
+- 下
+- た
+- ▁Das
+- ▁auch
+- ka
+- ▁ya
+- ter
+- ▁kind
+- ▁without
+- ▁place
+- ж
+- ▁put
+- 出
+- lo
+- ba
+- ment
+- ▁oder
+- ▁under
+- 没有
+- した
+- л
+- ▁вы
+- ▁una
+- ▁off
+- th
+- ▁то
+- ri
+- 里
+- ▁yet
+- ze
+- н
+- する
+- ▁diese
+- ▁another
+- ers
+- ▁why
+- je
+- ▁found
+- 从
+- 时
+- ko
+- 就是
+- sa
+- ng
+- ▁S
+- ▁shall
+- ▁aber
+- ▁sein
+- or
+- ▁а
+- 点
+- '3'
+- ur
+- ка
+- ▁tell
+- ▁god
+- お
+- ▁و
+- ▁Aber
+- ke
+- ness
+- 也
+- ▁far
+- able
+- ▁give
+- si
+- して
+- ▁though
+- ▁nothing
+- mo
+- ▁eyes
+- 手
+- ▁once
+- 家
+- x
+- ver
+- ▁always
+- ▁het
+- 日
+- ▁saw
+- ▁house
+- men
+- 用
+- di
+- ▁ha
+- この
+- ja
+- go
+- ▁wird
+- ▁ever
+- ▁een
+- ▁face
+- о
+- )
+- ▁vor
+- ▁你
+- 年
+- ▁love
+- am
+- 想
+- ung
+- ent
+- ▁uns
+- ▁home
+- he
+- на
+- ▁habe
+- ation
+- ▁head
+- ▁nach
+- ли
+- ▁nur
+- mi
+- ▁few
+- zi
+- ▁young
+- но
+- ن
+- て
+- ▁mean
+- ▁mich
+- ki
+- ▁better
+- ▁each
+- ▁einem
+- ▁find
+- ha
+- ▁si
+- ▁business
+- ш
+- ت
+- では
+- ▁left
+- ▁both
+- ton
+- ▁It
+- ▁v
+- der
+- v
+- ▁einer
+- 那
+- 为
+- wa
+- ▁night
+- ·
+- 过
+- ▁took
+- ▁(
+- ▁können
+- ▁mind
+- ▁—
+- ya
+- ▁father
+- ▁mir
+- ▁moment
+- ▁done
+- 得
+- ▁va
+- ca
+- د
+- 水
+- ▁su
+- 」
+- ▁against
+- ▁need
+- ry
+- 吧
+- 的人
+- ▁end
+- ▁yes
+- sch
+- ▁lot
+- ity
+- という
+- um
+- ▁course
+- co
+- ▁dat
+- ▁half
+- ▁between
+- 高
+- ▁Wir
+- ▁من
+- ant
+- り
+- ▁next
+- ste
+- 当
+- ▁noch
+- ▁pro
+- ▁He
+- 多
+- 做
+- ▁E
+- gen
+- 本
+- 这些
+- 事
+- でも
+- ▁می
+- ▁door
+- к
+- ▁called
+- ted
+- son
+- ▁looked
+- za
+- ▁ja
+- ul
+- 前
+- tu
+- ▁ko
+- ▁wurde
+- ye
+- ス
+- ،
+- 什么
+- ▁por
+- ▁kann
+- ▁los
+- ▁asked
+- ش
+- ▁seen
+- ▁как
+- 回
+- ла
+- 「
+- ▁дробь
+- ▁улица
+- 心
+- ▁side
+- ▁told
+- ▁does
+- ら
+- '4'
+- 者
+- lu
+- ▁son
+- ▁having
+- 生
+- ▁water
+- 分
+- ol
+- 三
+- ▁sehr
+- ▁quite
+- ▁hatte
+- 看
+- ▁El
+- ▁heard
+- ▁whole
+- ▁heart
+- ▁believe
+- '".'
+- ▁te
+- ▁met
+- ч
+- 打
+- 与
+- ▁name
+- ▁uh
+- 的时候
+- ▁ni
+- 而
+- ▁mother
+- ▁در
+- ▁ال
+- vo
+- 把
+- ▁bei
+- land
+- 能
+- ▁ob
+- ▁knew
+- ▁о
+- ▁op
+- ▁ma
+- ▁different
+- ▁around
+- ▁best
+- ▁به
+- ▁call
+- 可以
+- ▁mi
+- ▁enough
+- ▁il
+- ▁second
+- ▁para
+- く
+- 行
+- 自己
+- ▁par
+- 性
+- ie
+- ▁point
+- ▁seemed
+- ю
+- ます
+- 后
+- ▁ba
+- ▁set
+- ▁four
+- 所
+- ました
+- ة
+- ▁pas
+- ▁mrs
+- so
+- には
+- ▁woman
+- ia
+- ven
+- ▁room
+- ▁да
+- 山
+- ting
+- その
+- ・
+- bo
+- S
+- ▁five
+- ▁che
+- ion
+- 이
+- ▁high
+- 走
+- ▁所以
+- ▁sure
+- ▁但是
+- ▁durch
+- 还
+- '5'
+- ▁hundred
+- ▁country
+- ▁light
+- ▁sir
+- та
+- ▁use
+- ▁hier
+- ▁anything
+- ▁sa
+- г
+- ▁к
+- ▁actually
+- 现在
+- ad
+- by
+- ▁bir
+- ku
+- ▁он
+- ▁quarter
+- ▁wa
+- ▁po
+- ó
+- ▁при
+- ▁almost
+- ▁про
+- ▁days
+- ▁от
+- ▁他
+- 天
+- ken
+- ▁help
+- 最
+- ▁care
+- ▁sort
+- ر
+- 您
+- ▁dieser
+- ▁Der
+- po
+- 特
+- id
+- 新
+- ▁L
+- ▁ab
+- ですね
+- ▁gibt
+- ▁soon
+- ler
+- 国
+- ▁com
+- ▁з
+- ▁Be
+- 我的
+- ▁white
+- ▁alle
+- ▁dann
+- du
+- ling
+- ▁small
+- 力
+- ي
+- less
+- ▁together
+- ▁fact
+- im
+- ▁since
+- ▁money
+- ak
+- ▁мы
+- 很
+- à
+- 个
+- 你的
+- ▁qui
+- ▁doing
+- さん
+- hi
+- 更
+- ▁until
+- 给
+- ine
+- 斯
+- ▁talk
+- ▁question
+- ▁nor
+- 老
+- き
+- ty
+- ▁twenty
+- ▁didn
+- ▁used
+- ik
+- vi
+- ▁bit
+- ▁hard
+- ▁miss
+- zu
+- ▁big
+- ▁full
+- ▁Le
+- ▁Re
+- ▁yn
+- ▁De
+- ist
+- よ
+- 非常
+- au
+- ▁gave
+- 不是
+- ▁morning
+- さ
+- з
+- ▁mehr
+- ов
+- 将
+- ку
+- ▁keep
+- pa
+- ut
+- 可能
+- ▁lord
+- ▁rest
+- ▁number
+- ▁An
+- que
+- ▁real
+- 가
+- ▁words
+- ▁began
+- and
+- ки
+- 呀
+- ▁ik
+- ▁immer
+- ▁however
+- ▁hands
+- ▁open
+- ▁king
+- 头
+- ▁bin
+- ▁ب
+- A
+- ▁lo
+- bi
+- ▁looking
+- ▁раз
+- ▁pe
+- led
+- ل
+- ▁па
+- ▁c
+- п
+- ig
+- р
+- ▁turned
+- ▁feel
+- ▁En
+- 那个
+- ▁waren
+- 可
+- lar
+- ic
+- ▁і
+- ▁felt
+- ▁так
+- 比
+- ▁zum
+- ▁bu
+- ها
+- 再
+- ▁في
+- ▁wirklich
+- 像
+- в
+- 外
+- ▁etwas
+- 开始
+- 已经
+- 长
+- ちょっと
+- ▁poor
+- ▁pour
+- ▁D
+- ▁power
+- 或
+- 名
+- ▁word
+- ▁among
+- gi
+- 度
+- ▁myself
+- ▁children
+- ▁during
+- ا
+- ▁whom
+- ▁ka
+- しました
+- ▁这
+- ▁lady
+- ▁large
+- ▁matter
+- ▁death
+- ▁Ver
+- ▁Ja
+- ▁vi
+- pe
+- ▁true
+- '6'
+- ▁certain
+- ate
+- ом
+- kan
+- nt
+- ▁state
+- ▁pa
+- 并
+- ь
+- ▁till
+- ▁که
+- ▁dans
+- ▁person
+- 月
+- 金
+- ▁present
+- ▁general
+- les
+- ▁tun
+- ▁O
+- tion
+- ă
+- 向
+- ▁boy
+- 因为
+- ▁live
+- ▁case
+- ▁est
+- ▁brought
+- 以
+- ▁black
+- 成
+- 又
+- 你们
+- ek
+- ▁order
+- ▁dis
+- oj
+- ▁rather
+- mu
+- だ
+- ▁Se
+- 太
+- ▁voice
+- ▁w
+- ▁все
+- ▁taken
+- zo
+- 道
+- ▁Menschen
+- 先
+- jo
+- ть
+- ▁Wenn
+- 让
+- ▁perhaps
+- ▁given
+- ない
+- 内
+- ▁hear
+- ▁already
+- ju
+- 所以
+- ся
+- ▁une
+- dy
+- ▁important
+- ▁girl
+- ▁human
+- ▁show
+- 发
+- ▁вот
+- 这种
+- ▁ihre
+- ▁را
+- ▁C
+- wi
+- ▁hij
+- á
+- ▁ihr
+- ▁B
+- ▁qu
+- ▁ihn
+- ▁K
+- ▁company
+- lan
+- ▁meine
+- ▁strong
+- ren
+- ▁thousand
+- the
+- った
+- га
+- ▁seine
+- す
+- ▁air
+- ▁coming
+- ▁But
+- 面
+- ▁friend
+- ▁--
+- ben
+- ▁within
+- ▁round
+- 海
+- 方
+- ▁Li
+- ▁change
+- ва
+- ji
+- ard
+- ▁six
+- み
+- 二
+- ai
+- ▁often
+- ▁everything
+- ▁sea
+- ع
+- 没
+- ▁along
+- nie
+- ▁sent
+- ▁Ein
+- ▁whether
+- 他的
+- '7'
+- ▁wissen
+- 无
+- う
+- 这是
+- とか
+- 尔
+- per
+- ▁estas
+- 但
+- ▁но
+- ieren
+- ▁This
+- ▁car
+- ive
+- ▁themselves
+- ب
+- 车
+- 口
+- ▁forward
+- 入
+- 学
+- ▁co
+- ▁N
+- ado
+- ▁Na
+- 西
+- 何
+- ▁others
+- ▁za
+- í
+- つ
+- го
+- ▁我们
+- ▁idea
+- 开
+- ▁keur
+- س
+- ▁across
+- ▁dear
+- ▁stood
+- ▁school
+- ▁那
+- ▁understand
+- ▁sagte
+- ▁making
+- ▁dan
+- ck
+- bu
+- ▁family
+- ▁fire
+- ▁nature
+- ▁near
+- ▁из
+- 于
+- ем
+- もう
+- ▁times
+- да
+- ▁herself
+- ▁public
+- ▁niet
+- ▁b
+- ▁Zeit
+- ▁read
+- ▁cannot
+- ▁pre
+- ▁hope
+- 花
+- ▁turn
+- ▁city
+- ル
+- ▁این
+- cu
+- ▁würde
+- è
+- ▁است
+- ▁zijn
+- ger
+- 死
+- ▁از
+- sta
+- ны
+- ▁viel
+- ▁li
+- ner
+- ▁women
+- ▁gone
+- ▁sagen
+- ster
+- ▁non
+- ▁sur
+- 听
+- ▁然后
+- fe
+- ▁ser
+- ▁means
+- ▁thus
+- いい
+- ▁tu
+- ▁child
+- ▁least
+- ق
+- ▁sehen
+- ▁T
+- ▁nu
+- 化
+- ▁pretty
+- aba
+- ▁ki
+- ты
+- ▁sense
+- ▁says
+- ada
+- ▁أ
+- fa
+- 工作
+- ので
+- ▁reason
+- 这样
+- ▁einfach
+- ▁keine
+- ▁behind
+- 日本
+- ▁この
+- ▁wife
+- ▁Il
+- T
+- E
+- いた
+- che
+- ▁voor
+- ▁short
+- ▁bi
+- 一些
+- ▁body
+- ▁indeed
+- 先生
+- ▁ce
+- ▁feet
+- wo
+- 一样
+- '8'
+- ▁M
+- 就像
+- ▁wo
+- 还是
+- 德
+- نا
+- ▁до
+- ▁Ma
+- ▁sat
+- ▁continue
+- ま
+- ts
+- ▁unter
+- age
+- q
+- ▁ask
+- 儿
+- ▁U
+- ▁close
+- ить
+- ம்
+- ▁several
+- 放
+- ▁won
+- ▁om
+- ية
+- ▁whose
+- не
+- ▁market
+- 之
+- ▁red
+- ▁possible
+- C
+- up
+- 你知道
+- ▁amb
+- イ
+- わ
+- 部
+- ▁leave
+- tes
+- など
+- 는
+- として
+- ▁less
+- ile
+- ク
+- ▁become
+- ля
+- ▁later
+- 重
+- 拉
+- ان
+- 吃
+- ▁known
+- ▁Un
+- ف
+- ▁growth
+- zen
+- ä
+- P
+- ft
+- ▁Ge
+- ▁friends
+- و
+- lich
+- ▁either
+- ny
+- gu
+- ▁story
+- ▁probably
+- 体
+- ous
+- ana
+- ▁ihm
+- ー
+- ▁Sch
+- ▁able
+- ▁thou
+- ▁else
+- ▁wieder
+- io
+- 白
+- ц
+- 跟
+- 万
+- ▁run
+- 光
+- 位
+- ▁diesem
+- ▁bad
+- б
+- ▁alone
+- 但是
+- ات
+- ▁そして
+- ▁gut
+- ز
+- ▁start
+- 法
+- ▁past
+- 路
+- ▁itself
+- iert
+- aj
+- ▁în
+- š
+- ▁Art
+- ле
+- ▁jetzt
+- I
+- ran
+- і
+- ▁therefore
+- ア
+- ▁free
+- ▁sometimes
+- ▁passed
+- ў
+- ▁speak
+- mos
+- 数
+- 问题
+- yo
+- ▁book
+- ▁line
+- ▁andere
+- ▁mal
+- chen
+- ми
+- ▁Bu
+- ▁above
+- ▁became
+- ▁Al
+- bar
+- 我们的
+- ▁wanted
+- des
+- ال
+- way
+- 利
+- era
+- ك
+- ▁bo
+- ▁Je
+- ▁cried
+- '9'
+- ban
+- ▁today
+- tro
+- ▁au
+- 今
+- 고
+- ▁getting
+- 别
+- ▁okay
+- ley
+- ▁kaj
+- ham
+- 马
+- ん
+- fer
+- sten
+- 지
+- ▁saying
+- ▁No
+- ▁bring
+- ▁zwei
+- 四
+- ated
+- 需要
+- ▁p
+- 田
+- ح
+- ▁weil
+- ▁las
+- ▁z
+- mer
+- gel
+- ▁és
+- 世界
+- '10'
+- ber
+- 孩子
+- んだ
+- ет
+- end
+- 都是
+- ▁ta
+- ▁그
+- だった
+- 王
+- リ
+- ▁form
+- ▁dead
+- ▁hour
+- ▁future
+- ма
+- ▁bis
+- ▁machen
+- 物
+- پ
+- лі
+- ▁earth
+- ▁zur
+- 自
+- nd
+- ful
+- カ
+- ▁là
+- ▁government
+- ▁Do
+- 钱
+- ▁،
+- му
+- nu
+- ▁play
+- ▁remember
+- ▁land
+- ▁bed
+- ▁vous
+- ere
+- ▁plus
+- 知道
+- 两
+- 带
+- 真
+- ▁fast
+- ish
+- ning
+- 快
+- 身
+- ▁fell
+- ▁dark
+- ▁held
+- 时间
+- rs
+- 定
+- им
+- ▁As
+- ▁started
+- ▁lost
+- op
+- 主
+- ▁hi
+- ▁como
+- 爱
+- ▁ت
+- ▁answer
+- ▁Was
+- 一下
+- ▁bar
+- ▁один
+- tor
+- します
+- gan
+- ▁cost
+- ная
+- ▁fa
+- nya
+- ang
+- ▁сто
+- ой
+- ▁early
+- ari
+- min
+- ご
+- ▁müssen
+- su
+- ▁ground
+- lin
+- ▁answered
+- ▁ago
+- red
+- ▁anderen
+- ▁lay
+- 明
+- ▁Vi
+- ▁ў
+- ▁Ba
+- con
+- ▁stand
+- 正
+- ▁但
+- 安
+- 知
+- 加
+- ▁под
+- 合
+- ▁selbst
+- 谁
+- 起
+- han
+- ▁они
+- ▁em
+- ра
+- ▁Ha
+- ç
+- ']'
+- tt
+- vu
+- 五
+- ▁truth
+- ▁Ka
+- 美
+- ▁mar
+- ot
+- ド
+- ▁мне
+- 看到
+- ▁alles
+- 流
+- ▁denke
+- ▁try
+- ac
+- iz
+- ait
+- 人们
+- ker
+- cy
+- ru
+- ▁clear
+- ко
+- 同
+- ▁Leute
+- ▁working
+- 克
+- ▁Ne
+- ▁front
+- ▁towards
+- ▁replied
+- ト
+- port
+- ▁taking
+- ▁Ihnen
+- ▁viele
+- 火
+- 然后
+- ▁же
+- 怎么
+- her
+- え
+- র
+- ▁further
+- 等
+- 那么
+- day
+- 的话
+- ин
+- 叫
+- ▁Dinge
+- ▁fine
+- ▁ad
+- 市
+- வ
+- ▁kon
+- ▁ну
+- ам
+- 只是
+- om
+- ▁zurück
+- ▁trying
+- ."
+- tan
+- ism
+- ▁law
+- O
+- ▁town
+- ▁было
+- ▁для
+- 感
+- ▁fear
+- ff
+- ▁Da
+- 门
+- lle
+- ez
+- B
+- 全
+- 气
+- これ
+- にも
+- ▁его
+- ini
+- ern
+- まで
+- ire
+- 目
+- ▁ты
+- 真的
+- ラ
+- ;
+- ▁thank
+- ▁со
+- ▁table
+- are
+- ▁ye
+- ؟
+- ▁Y
+- 使用
+- 中国
+- ▁есть
+- ▁third
+- ▁church
+- ▁dieses
+- ▁expect
+- ú
+- 作
+- います
+- ex
+- ▁Du
+- ▁art
+- 平
+- ▁Sa
+- ▁diesen
+- ный
+- ▁sun
+- ▁doubt
+- ▁бы
+- ▁return
+- qui
+- ▁system
+- ▁pay
+- ap
+- ▁sin
+- ▁era
+- ▁[
+- ü
+- ance
+- のは
+- ▁plan
+- 도
+- め
+- ▁maybe
+- ▁soul
+- ர்
+- コ
+- 野
+- ▁dr
+- ita
+- tre
+- ina
+- ло
+- ▁muss
+- ▁الم
+- sen
+- 通
+- ▁aan
+- 请
+- ▁sound
+- ring
+- ▁거
+- ند
+- one
+- ▁Wie
+- 南
+- ▁beautiful
+- ▁seems
+- 見
+- ах
+- ▁week
+- ▁brother
+- ▁два
+- ▁با
+- ▁kept
+- sha
+- 時
+- elle
+- あ
+- ary
+- gar
+- ▁certainly
+- ▁continued
+- ▁comes
+- ▁P
+- ▁top
+- ▁position
+- ча
+- 어
+- '20'
+- sto
+- ▁там
+- ит
+- ▁gu
+- ▁нас
+- K
+- ▁suddenly
+- ▁seven
+- ▁thinking
+- ▁А
+- している
+- 에
+- ab
+- ▁wish
+- ▁talking
+- ▁ready
+- ▁Mi
+- ▁master
+- F
+- ▁service
+- 自己的
+- ন
+- ▁evening
+- しています
+- ▁view
+- ▁cap
+- ▁thy
+- 那些
+- ▁nie
+- ک
+- ▁ما
+- sel
+- ▁subject
+- 通过
+- ▁Per
+- '...'
+- 应该
+- ▁Ta
+- ▁eines
+- ▁cold
+- ure
+- és
+- pi
+- ▁dé
+- 文
+- ▁longer
+- ▁letter
+- ▁cause
+- ▁Ro
+- ▁happened
+- kin
+- 生活
+- ды
+- ▁deep
+- ▁nous
+- ▁если
+- dan
+- ▁common
+- ding
+- 立
+- 使
+- ▁happy
+- ▁post
+- van
+- ▁zij
+- bel
+- でした
+- ン
+- tar
+- ▁дом
+- ▁ca
+- んで
+- ▁low
+- ▁arm
+- 区
+- ▁gi
+- ▁cu
+- 动
+- ▁pass
+- ти
+- ни
+- то
+- ble
+- ▁interest
+- ▁dit
+- 格
+- ah
+- ▁schon
+- cia
+- ix
+- 这么
+- ê
+- igen
+- ▁ihnen
+- 只
+- ▁john
+- 米
+- une
+- ▁Ihre
+- ▁manner
+- ▁living
+- ies
+- ▁To
+- 它们
+- 声
+- து
+- ama
+- ▁natural
+- ある
+- 送
+- いて
+- ▁hold
+- ▁thirty
+- ▁eight
+- ▁「
+- ▁б
+- ▁Welt
+- mp
+- 研究
+- ▁spirit
+- fi
+- ▁months
+- された
+- ▁river
+- ▁அ
+- ين
+- ▁unsere
+- 出来
+- ▁husband
+- ▁watch
+- G
+- ক
+- ▁capital
+- ite
+- sh
+- 店
+- 不会
+- mal
+- ris
+- 问
+- ▁road
+- 意
+- 住
+- ▁rose
+- ▁fi
+- los
+- ▁guess
+- val
+- 总
+- ▁Ab
+- 的是
+- ▁seeing
+- ▁support
+- 喜欢
+- ня
+- ной
+- ▁result
+- ub
+- ▁Ni
+- 有一个
+- 我想
+- 原
+- ▁You
+- ▁value
+- во
+- ط
+- ▁deal
+- ▁meet
+- ▁results
+- 号
+- ▁wurden
+- 長
+- bra
+- 今天
+- tic
+- 木
+- 不能
+- ▁street
+- 次
+- 希望
+- ▁vol
+- 情
+- 给我
+- ▁green
+- ар
+- ▁denn
+- ▁social
+- ▁hours
+- 或者
+- 科
+- ▁returned
+- ▁cut
+- ил
+- ▁experience
+- vis
+- ian
+- ون
+- len
+- ▁sub
+- ungen
+- ▁strange
+- ud
+- 元
+- ▁werde
+- 戦
+- кі
+- いる
+- ая
+- 是的
+- ▁whatever
+- ▁einige
+- ▁sight
+- 之前
+- ▁mein
+- 公司
+- 半
+- 的事情
+- ▁drei
+- 言
+- ▁doesn
+- 公
+- ischen
+- car
+- ▁その
+- ▁Aus
+- ей
+- те
+- 也是
+- 是一个
+- ▁să
+- ▁fair
+- ▁У
+- ▁terms
+- ▁food
+- 気
+- ▁minutes
+- ef
+- 的东西
+- ▁seiner
+- ▁아
+- 能够
+- ль
+- خ
+- ▁ex
+- ons
+- sche
+- ▁Te
+- ▁court
+- ▁process
+- ナ
+- tel
+- ▁account
+- 发现
+- 该
+- 品
+- 林
+- ▁received
+- ▁blood
+- ▁меня
+- ных
+- ence
+- ag
+- D
+- ▁followed
+- ▁Ra
+- ▁ho
+- ▁Mo
+- 干
+- ▁pri
+- ロ
+- 张
+- がある
+- ò
+- ▁feeling
+- ну
+- ici
+- ▁Co
+- 州
+- 觉得
+- ▁particular
+- las
+- ல்
+- マ
+- ▁Di
+- ▁makes
+- ні
+- lie
+- ▁はい
+- ▁yo
+- ▁due
+- 他们的
+- ▁move
+- ▁یک
+- ▁english
+- 第
+- line
+- ā
+- 女
+- ▁أن
+- ▁mais
+- ▁seem
+- 讲
+- 代
+- 八
+- ▁Vor
+- ende
+- ▁damit
+- ما
+- 线
+- ▁act
+- ▁beyond
+- ria
+- れ
+- nen
+- ▁captain
+- zer
+- ▁single
+- ▁hair
+- ▁об
+- té
+- ▁arms
+- ▁spoke
+- ▁They
+- tra
+- ▁suppose
+- ▁có
+- tas
+- க்
+- ர
+- ▁self
+- ▁Also
+- ▁information
+- ▁data
+- ▁simple
+- وا
+- ста
+- 之后
+- ша
+- ▁wild
+- ▁horse
+- ige
+- ▁R
+- 还有
+- 一点
+- M
+- king
+- ▁tried
+- ▁sleep
+- č
+- 相
+- ▁mine
+- 을
+- 石
+- ée
+- 社会
+- ndo
+- 色
+- _
+- mes
+- 量
+- 人が
+- 神
+- ▁wrong
+- 大家
+- ▁effect
+- ▁wind
+- ▁history
+- ▁sit
+- ▁yourself
+- ور
+- 才
+- 集
+- ▁eye
+- 球
+- 理
+- ▁walk
+- ▁fifty
+- ブ
+- ▁states
+- 军
+- cer
+- ▁foot
+- ▁Um
+- 们
+- レ
+- hu
+- cht
+- ▁очень
+- 台
+- だけ
+- じ
+- 形
+- ▁inter
+- ▁Les
+- ▁problem
+- gri
+- ▁although
+- ▁fer
+- ▁maar
+- こ
+- 空
+- ме
+- 不要
+- ▁gold
+- ische
+- N
+- 找
+- ▁G
+- ▁mo
+- fu
+- ▁这是
+- ura
+- mente
+- ally
+- ▁F
+- 关于
+- 北
+- if
+- 很多
+- î
+- ▁space
+- ست
+- ▁когда
+- ▁period
+- なかった
+- ▁ei
+- 所有
+- ▁könnte
+- ています
+- んです
+- ▁glad
+- こと
+- タ
+- ▁weiß
+- 六
+- ▁example
+- ▁двадцать
+- ▁stop
+- ▁stuff
+- 私
+- ▁els
+- ron
+- ▁reached
+- 认为
+- ▁doctor
+- ę
+- 政府
+- 让我
+- ▁none
+- nde
+- ano
+- ම
+- ▁daughter
+- ▁send
+- ▁force
+- のか
+- use
+- 进
+- tter
+- ▁chance
+- ras
+- ▁attention
+- че
+- 见
+- ▁W
+- 买
+- ▁fall
+- ▁pot
+- 教
+- ম
+- room
+- ▁Man
+- ş
+- について
+- ▁она
+- ▁blue
+- ▁exactly
+- ale
+- 如果
+- ▁New
+- 东西
+- né
+- ▁dir
+- 時間
+- ▁Ku
+- ▁camp
+- ▁stay
+- wer
+- ▁party
+- ▁Si
+- ▁especially
+- ض
+- 完全
+- 川
+- wn
+- af
+- ando
+- fen
+- ди
+- ▁character
+- ▁dich
+- ▁except
+- вер
+- 布
+- ▁please
+- my
+- ок
+- ▁Pa
+- そう
+- ▁Als
+- ▁share
+- ▁hatten
+- ия
+- フ
+- 事情
+- ▁Leben
+- ▁higher
+- ▁tôi
+- 阿
+- 何か
+- ement
+- 受
+- になる
+- ▁mis
+- ▁outside
+- ▁following
+- ▁hu
+- ▁level
+- ▁Me
+- ные
+- ▁geht
+- 信
+- ру
+- ▁couple
+- ▁lui
+- ▁necessary
+- ok
+- ▁Auf
+- ele
+- 器
+- 给你
+- 卡
+- 味
+- やっぱり
+- 起来
+- це
+- 话
+- ▁instead
+- ▁và
+- ▁opportunity
+- ▁carried
+- ▁nun
+- eg
+- nk
+- ▁appeared
+- ▁toward
+- 食
+- 如何
+- ▁gab
+- cho
+- ▁respect
+- 风
+- ati
+- ▁south
+- ▁Teil
+- 美国
+- シ
+- ▁added
+- ž
+- ▁window
+- ▁如果
+- nda
+- 正在
+- ▁او
+- är
+- ன்
+- ö
+- ▁control
+- ating
+- もの
+- ▁england
+- ор
+- ф
+- за
+- ▁seinen
+- ▁пере
+- ▁nine
+- па
+- ▁questions
+- ই
+- 成为
+- 越
+- オ
+- ▁ought
+- 期
+- ▁más
+- ▁save
+- 九
+- 朝
+- del
+- ць
+- ▁Mu
+- ▁late
+- ▁lower
+- ▁нь
+- 民
+- ira
+- 以后
+- ▁lived
+- ▁health
+- ▁knowledge
+- ▁ke
+- ▁Lo
+- 写
+- ▁што
+- ns
+- ▁нет
+- ▁现在
+- res
+- са
+- ▁customers
+- ▁tra
+- он
+- 嘛
+- ▁dollars
+- ый
+- ё
+- ▁Mae
+- ▁Bo
+- ▁gehen
+- cha
+- ▁wat
+- ▁Po
+- र
+- 歌
+- ▁opened
+- ks
+- ار
+- 取
+- ▁Im
+- chi
+- ▁miles
+- ▁grand
+- eth
+- let
+- また
+- 少
+- ▁haar
+- ている
+- ▁pen
+- ▁wall
+- ach
+- க்க
+- ▁è
+- 学校
+- 选择
+- ▁comp
+- ▁eighteen
+- ▁step
+- ▁easy
+- tur
+- 哪
+- heit
+- ば
+- ▁quickly
+- don
+- ▁Mal
+- mar
+- ▁wait
+- ging
+- 书
+- っていう
+- ky
+- 回来
+- 员
+- eur
+- ба
+- ▁Pro
+- 城
+- ▁Bi
+- ien
+- ▁neither
+- ▁уже
+- ре
+- tal
+- ▁fellow
+- ise
+- ▁group
+- ▁product
+- lla
+- ▁Ri
+- ام
+- 人の
+- ▁building
+- ▁чтобы
+- ide
+- ▁因为
+- ▁goes
+- ▁north
+- ▁chi
+- ▁jeung
+- ▁Z
+- ▁vielleicht
+- ▁music
+- ு
+- 一种
+- ▁main
+- ▁три
+- una
+- ▁paper
+- 周
+- ри
+- ▁bright
+- ▁beginning
+- க
+- ным
+- out
+- ▁V
+- ры
+- tin
+- 几
+- ▁hall
+- L
+- ▁ningal
+- ▁wonder
+- ▁diwawancara
+- ▁forth
+- ▁olohok
+- ▁office
+- ▁christ
+- ▁sweet
+- ▁möchte
+- ミ
+- 後
+- ذ
+- ▁wollen
+- stre
+- stand
+- ▁ihren
+- ities
+- H
+- ned
+- yi
+- ▁var
+- ▁wäre
+- ▁figure
+- ▁mhm
+- ▁Ki
+- 为什么
+- ▁entre
+- ▁nearly
+- ▁ran
+- ▁dipoto
+- ன
+- ▁wartawan
+- ▁avec
+- ▁forty
+- 任何
+- 制
+- ▁пра
+- 十
+- kt
+- 在这里
+- ке
+- ▁team
+- ali
+- time
+- ▁mon
+- ida
+- 夜
+- 清
+- bre
+- 进行
+- ▁Mann
+- lor
+- med
+- 的一个
+- ▁ganz
+- ج
+- دا
+- 由
+- ▁bien
+- 机
+- 本当に
+- 拿
+- と思います
+- 近
+- ▁Con
+- 大学
+- ▁kam
+- ğ
+- ges
+- wy
+- ▁rich
+- 直
+- ийн
+- ▁interesting
+- 的な
+- ▁Mit
+- ду
+- ▁There
+- ▁local
+- ▁konnte
+- или
+- ▁sollte
+- 急
+- ▁Am
+- 波
+- 特别
+- ▁purpose
+- 其他
+- 倒
+- mon
+- ▁este
+- ▁lives
+- ல
+- ということです
+- ▁будет
+- ан
+- ▁daß
+- 的地方
+- ▁Ad
+- ту
+- ville
+- ▁key
+- kel
+- pan
+- ص
+- 巴
+- ▁một
+- 村
+- 得到
+- 表
+- od
+- ert
+- ▁heute
+- けど
+- ল
+- ▁Com
+- ▁근데
+- ça
+- 解
+- ▁game
+- 接
+- ▁news
+- ▁peace
+- ▁weiter
+- ய
+- ▁pleasure
+- sti
+- ▁darüber
+- for
+- ▁situation
+- ▁learn
+- ▁ou
+- ▁ihrer
+- 局
+- 笑
+- ▁lie
+- ▁안
+- バ
+- 是什么
+- tie
+- ▁على
+- ▁silence
+- ▁grow
+- R
+- 是不是
+- '30'
+- ▁sister
+- tri
+- ▁sei
+- 安全
+- ther
+- த
+- ▁class
+- ید
+- 音
+- 派
+- 甚至
+- ▁zoo
+- ▁pan
+- ▁moved
+- ▁strength
+- ib
+- ili
+- ▁sta
+- ということで
+- ▁modern
+- ▁success
+- ante
+- side
+- 雪
+- ▁secret
+- log
+- ▁Va
+- ł
+- 君
+- 初
+- 哥
+- ▁dog
+- ▁follow
+- 指
+- ani
+- cher
+- се
+- 船
+- 切
+- 继续
+- ▁board
+- 男
+- ▁comme
+- ▁ru
+- ▁mag
+- 提供
+- ▁뭐
+- vel
+- 古
+- nes
+- U
+- ▁och
+- ▁davon
+- 다
+- 中的
+- 国家
+- 雨
+- 来说
+- ▁french
+- ッ
+- 早
+- ▁gegen
+- ро
+- mit
+- lichen
+- ▁middle
+- 两个
+- ▁project
+- ▁С
+- lang
+- ▁uit
+- ▁afraid
+- どう
+- ía
+- ▁eat
+- ▁expected
+- ▁dar
+- cal
+- への
+- ▁bear
+- む
+- 实际上
+- 管
+- غ
+- nce
+- 落
+- ▁Я
+- よく
+- ▁performance
+- ▁Arbeit
+- っ
+- 难
+- ца
+- ▁wouldn
+- ▁ara
+- tte
+- ▁army
+- こう
+- ▁entered
+- uri
+- 到了
+- mas
+- 字
+- ▁spring
+- ▁fit
+- ▁Ho
+- mor
+- ize
+- 如此
+- ier
+- ▁impact
+- ▁object
+- 七
+- 帮助
+- ▁national
+- can
+- ▁drive
+- rea
+- ▁cool
+- ▁private
+- 片
+- ”
+- ▁difficult
+- 在一起
+- 咱们
+- 土
+- 曲
+- ▁joy
+- ized
+- ப்
+- ү
+- ▁standing
+- 其实
+- ▁progress
+- ▁trouble
+- 低
+- ▁floor
+- tres
+- ▁pu
+- 選手
+- ▁только
+- 東京
+- ▁ship
+- ▁Nun
+- かな
+- 哈
+- ▁Su
+- ▁break
+- чи
+- 一直
+- 命
+- form
+- ▁mai
+- ول
+- ▁zusammen
+- ئ
+- ple
+- ▁av
+- ▁cho
+- エ
+- ズ
+- ▁provide
+- 救
+- pen
+- ▁She
+- ами
+- ▁happen
+- ving
+- 病
+- ▁din
+- 感じ
+- 作为
+- zeit
+- 活
+- になって
+- ▁nice
+- ウ
+- ▁ways
+- ▁american
+- ▁everybody
+- ▁special
+- ▁smile
+- же
+- ▁major
+- 率
+- ▁ju
+- sse
+- ▁chief
+- ▁hem
+- 理解
+- 一起
+- およそ
+- 处
+- ▁sont
+- ত
+- ▁distance
+- 罗
+- ▁И
+- 看看
+- чы
+- 电
+- ció
+- vin
+- ▁prince
+- ▁dies
+- ▁sign
+- لا
+- ▁fight
+- ▁greater
+- ▁That
+- ен
+- ▁mouth
+- 朋友
+- ▁Jahren
+- dig
+- ido
+- ▁kein
+- ▁consider
+- ▁demand
+- ව
+- 反
+- ▁sitting
+- ого
+- ▁written
+- ▁lang
+- ▁oku
+- ▁Wa
+- ní
+- cla
+- ▁remain
+- ▁couldn
+- 李
+- art
+- dd
+- ▁Eine
+- 은
+- 街
+- 场
+- ▁design
+- ▁spot
+- ▁потому
+- house
+- cken
+- 不知道
+- ▁vom
+- すると
+- ▁length
+- nis
+- win
+- ▁ба
+- 〉
+- ▁job
+- ▁пять
+- すごい
+- ship
+- 岁
+- hen
+- ▁heaven
+- де
+- ა
+- san
+- 必须
+- 河
+- ▁books
+- 条
+- ▁final
+- of
+- ▁grew
+- ▁base
+- 一次
+- ие
+- ▁walked
+- ▁isn
+- ▁increase
+- 官
+- ▁просто
+- 奥
+- ▁society
+- ▁amount
+- ▁On
+- 断
+- ero
+- cur
+- гу
+- ▁trees
+- ▁program
+- ▁これ
+- ې
+- ▁был
+- न
+- 有点
+- ى
+- ▁ம
+- ations
+- â
+- iza
+- ▁faith
+- vor
+- ▁heavy
+- ▁immediately
+- ▁battle
+- ▁president
+- ین
+- ▁financial
+- ▁What
+- ▁〈
+- のが
+- ▁dort
+- ▁четыре
+- ▁running
+- ▁begin
+- ▁рас
+- par
+- 間
+- サ
+- ▁personal
+- へ
+- ▁action
+- ха
+- wan
+- ▁cor
+- ▁тридцать
+- sie
+- ▁places
+- ▁development
+- ▁visit
+- ▁во
+- ▁train
+- ▁eu
+- ▁mary
+- ▁learned
+- ▁không
+- ▁write
+- ▁cre
+- ▁ohne
+- ですよね
+- ▁Gu
+- ст
+- 一个人
+- 东
+- 最近
+- 首
+- ется
+- ▁darauf
+- nas
+- ▁hot
+- さんが
+- ▁died
+- 地方
+- 支持
+- ую
+- ▁laid
+- ▁perfect
+- 她的
+- ces
+- ries
+- ▁queen
+- 工
+- 却
+- そして
+- ▁Ar
+- 决定
+- できる
+- ▁einmal
+- 서
+- 足
+- ▁command
+- ▁safe
+- ии
+- yu
+- される
+- ▁slowly
+- 连
+- dra
+- ә
+- ▁price
+- ▁Ti
+- ▁super
+- 玩
+- ı
+- ▁Mar
+- ий
+- ▁quick
+- ▁wood
+- ▁glaube
+- ▁сейчас
+- ▁add
+- 女性
+- ram
+- ndi
+- ▁estis
+- liche
+- ▁build
+- 只有
+- 实
+- っている
+- ▁sales
+- ▁nga
+- ▁Zu
+- ▁stopped
+- ą
+- 队
+- dro
+- ▁sobre
+- キ
+- 森
+- ▁嗯
+- ▁finally
+- 红
+- 思
+- nia
+- ▁various
+- ▁conversation
+- 差
+- 当然
+- ▁color
+- ▁changed
+- ▁age
+- 推
+- ө
+- цы
+- ▁married
+- 远
+- ▁west
+- ▁caught
+- mm
+- 酒
+- ▁gentleman
+- vid
+- 深
+- mus
+- ▁més
+- ▁cette
+- lik
+- 战
+- ▁hast
+- clock
+- uch
+- 所有的
+- 强
+- ве
+- ▁или
+- ▁film
+- ▁desire
+- shi
+- mel
+- ▁straight
+- ▁study
+- ins
+- メ
+- ▁pi
+- isi
+- ▁может
+- вы
+- ▁model
+- ▁warm
+- ▁ging
+- rie
+- sion
+- ▁которые
+- ▁tree
+- 当时
+- ▁ill
+- த்த
+- ▁cross
+- ▁hay
+- ▁ад
+- dem
+- ▁built
+- ▁Land
+- ▁science
+- ▁бол
+- ▁J
+- ▁bra
+- ட
+- ans
+- 亚
+- ところ
+- ▁danger
+- 商
+- 努力
+- ska
+- ▁kwa
+- 啦
+- 每
+- ▁talked
+- ▁remained
+- 部分
+- 老师
+- 士
+- 室
+- 变
+- sis
+- 番
+- ▁paar
+- schen
+- ▁giving
+- gang
+- 話
+- そうです
+- ▁gonna
+- ▁All
+- pp
+- ité
+- ▁quiet
+- 让我们
+- ▁rate
+- ного
+- 转
+- ▁york
+- ▁meiner
+- 热
+- ▁их
+- rus
+- チ
+- ▁الأ
+- ɣ
+- ▁tre
+- 我是
+- 事件
+- ▁wrote
+- ▁picture
+- ▁loved
+- ip
+- yan
+- аў
+- ▁Diese
+- ps
+- 整个
+- ▁Ur
+- ▁tr
+- ▁Ko
+- ▁snow
+- ▁gemacht
+- ▁waiting
+- ▁uncle
+- ث
+- tive
+- ▁семь
+- 肉
+- ▁那么
+- ▁fort
+- ▁yang
+- ▁race
+- ito
+- 失
+- ▁nichts
+- ▁For
+- ▁london
+- ▁кто
+- мо
+- ▁これは
+- ▁minute
+- ать
+- ino
+- ▁boys
+- 报
+- ski
+- бы
+- 千
+- ▁dinner
+- til
+- ▁gar
+- ▁march
+- ය
+- хо
+- 房
+- ▁fourth
+- rem
+- 边
+- ▁works
+- ▁sprechen
+- ▁broken
+- tus
+- 关
+- ▁tatsächlich
+- ち
+- ▁below
+- 有什么
+- ▁đ
+- ties
+- 感觉
+- ona
+- ▁piece
+- xi
+- ▁similar
+- より
+- ▁according
+- 保
+- ▁simply
+- 最后
+- ▁trust
+- 往
+- 达
+- なく
+- ம
+- ▁seinem
+- ḥ
+- ▁regard
+- ug
+- ▁hardly
+- 如果你
+- ▁zwischen
+- ▁direction
+- ▁seu
+- ▁drew
+- 世
+- lä
+- ▁mil
+- сан
+- ▁brain
+- ▁corner
+- 真正
+- cle
+- 种
+- 以上
+- 关系
+- ▁using
+- 这里
+- ▁summer
+- اس
+- ▁buy
+- ich
+- ▁language
+- ▁Ke
+- ▁impossible
+- ▁sus
+- 不同
+- چ
+- ше
+- eren
+- ▁Tu
+- ▁bon
+- 权
+- ▁hätte
+- ade
+- ▁Tag
+- ▁beauty
+- ▁lips
+- ue
+- 画
+- 第一
+- ▁evil
+- ▁Now
+- õ
+- 单
+- ▁france
+- ▁garden
+- ▁بود
+- ▁month
+- bil
+- ▁station
+- ec
+- さんの
+- ev
+- 进入
+- ▁lead
+- ré
+- int
+- 包
+- 県
+- sco
+- 기
+- ▁Ga
+- gehen
+- ▁denen
+- ▁presence
+- amente
+- ▁girls
+- ▁opinion
+- ▁touch
+- dia
+- そういう
+- ▁gran
+- 的问题
+- ▁fresh
+- ▁آن
+- 过去
+- ▁united
+- 系统
+- field
+- ▁Hu
+- gra
+- 야
+- rit
+- ▁mas
+- ▁political
+- ▁showed
+- とは
+- ят
+- fo
+- یم
+- ▁ма
+- ▁г
+- ser
+- ▁looks
+- ться
+- 眼
+- ず
+- ▁Frage
+- 以及
+- را
+- ▁vier
+- ▁나
+- ▁гэта
+- ▁placed
+- ney
+- パ
+- pu
+- ▁brown
+- 冷
+- 発
+- ▁сорок
+- ал
+- cra
+- 根
+- met
+- ▁Pe
+- ato
+- ▁我想
+- 星
+- ava
+- ción
+- ▁growing
+- ei
+- ions
+- 省
+- ▁risk
+- 了一个
+- ▁că
+- ▁В
+- ▁sick
+- ツ
+- ▁bank
+- ▁переулок
+- ▁current
+- ▁decided
+- ▁soft
+- ▁village
+- 板
+- ですが
+- 夫
+- 我就
+- 今年
+- ব
+- ▁filled
+- cent
+- 青
+- 好的
+- ▁hit
+- 実
+- ▁еще
+- ▁speaking
+- nh
+- ih
+- ют
+- ▁dazu
+- ice
+- گ
+- 其
+- 图
+- ral
+- น
+- V
+- ▁вас
+- ර
+- ▁hin
+- ▁Hi
+- ▁Ca
+- ▁led
+- 来了
+- kon
+- ▁ஆ
+- ou
+- ▁برای
+- 装
+- schaft
+- ▁wasn
+- ▁cent
+- ▁har
+- ▁muri
+- iti
+- эр
+- ▁weeks
+- 上的
+- 找到
+- ▁zich
+- ▁pero
+- 非
+- well
+- ▁inside
+- ial
+- like
+- ▁Sta
+- ▁Ihr
+- ▁writing
+- berg
+- ▁mentioned
+- 在那里
+- ▁expression
+- 出去
+- ▁trade
+- 脚
+- ▁born
+- fin
+- ▁pain
+- ▁á
+- ▁worked
+- ika
+- 風
+- ament
+- 比较
+- 交
+- ▁twelve
+- 这样的
+- ▁type
+- ▁cy
+- ده
+- ▁kan
+- ▁Jahre
+- ▁revenue
+- nte
+- ▁rain
+- ▁serve
+- ова
+- ute
+- ▁reach
+- よね
+- ▁H
+- oli
+- ▁meant
+- ▁würden
+- har
+- 妈妈
+- ▁direct
+- ▁இ
+- ң
+- やって
+- ▁plain
+- ▁för
+- ▁Ru
+- ▁pleased
+- ▁win
+- ▁winter
+- ▁influence
+- ▁parts
+- ▁doch
+- isch
+- kat
+- uk
+- ▁mij
+- ▁Lu
+- 車
+- 你可以
+- ▁naar
+- ▁worth
+- via
+- 这一
+- ▁community
+- ▁thoughts
+- 突然
+- 》
+- 包括
+- ▁kommen
+- hin
+- ▁しかし
+- ▁raised
+- ▁wide
+- ▁col
+- ▁große
+- ▁của
+- дзе
+- eb
+- 传
+- リー
+- ▁haven
+- ▁chair
+- ▁bri
+- ▁field
+- fall
+- 亲
+- ▁umu
+- 松
+- 里面
+- wood
+- 存在
+- 让你
+- 皮
+- 零
+- ▁boat
+- ▁forget
+- bro
+- ▁exclaimed
+- 代表
+- uka
+- ded
+- 听到
+- '11'
+- 血
+- ут
+- ▁carry
+- ▁chúng
+- mmer
+- mis
+- das
+- 华
+- 别人
+- ió
+- ▁Geschichte
+- ford
+- bb
+- ▁broad
+- ▁od
+- ▁paid
+- ▁likely
+- ▁condition
+- ▁ersten
+- ▁afternoon
+- 以前
+- ▁someone
+- ▁grave
+- ▁și
+- put
+- ای
+- ▁genau
+- 々
+- ▁dream
+- сы
+- ▁Fa
+- ых
+- න
+- ▁мо
+- ▁На
+- 说话
+- ком
+- ▁ourselves
+- ▁waar
+- ô
+- 草
+- ▁band
+- 我觉得
+- ▁content
+- ▁wenig
+- ora
+- 自然
+- 家庭
+- 方面
+- ে
+- 黑
+- ▁imp
+- ▁industry
+- ▁terrible
+- dos
+- 尼
+- ым
+- rig
+- ▁report
+- ▁somebody
+- 要求
+- ▁allowed
+- さんは
+- ▁path
+- rá
+- ▁pale
+- 那里
+- ▁appearance
+- ssen
+- ▁mark
+- ▁worden
+- umu
+- よう
+- 学生
+- ments
+- мен
+- 院
+- ▁க
+- أ
+- sur
+- 円
+- ▁journey
+- жа
+- ▁usually
+- ▁دو
+- 卖
+- ную
+- ves
+- ите
+- gli
+- 的方式
+- だと
+- 调
+- ▁listen
+- 警察
+- び
+- wen
+- ▁Frau
+- 兰
+- tik
+- グ
+- ī
+- 温
+- '50'
+- 線
+- んですよ
+- ▁vast
+- يا
+- ▁anar
+- ▁material
+- ە
+- ▁generally
+- ට
+- ▁fait
+- ▁enemy
+- пер
+- ▁area
+- ▁struck
+- dar
+- aient
+- 重要
+- tir
+- ▁rock
+- ような
+- 喝
+- 的に
+- ▁shot
+- ると
+- ▁finden
+- ▁sorry
+- ▁gra
+- ▁evidence
+- ▁آ
+- া
+- ний
+- 我认为
+- ход
+- ula
+- sam
+- 了解
+- ▁daran
+- wu
+- uma
+- ▁Gi
+- ▁Nach
+- いました
+- ков
+- ▁easily
+- 件
+- ▁individual
+- 节
+- ▁duty
+- 座
+- ▁tout
+- mie
+- 过来
+- ▁những
+- ▁unless
+- まだ
+- 它的
+- ু
+- ros
+- ▁bill
+- ▁fifteen
+- heid
+- mann
+- ▁Her
+- lt
+- 成功
+- iv
+- pri
+- اد
+- iya
+- それ
+- lia
+- ▁laughed
+- ▁шесть
+- ▁надо
+- ▁clean
+- ▁normal
+- sar
+- ▁ن
+- ▁tea
+- ala
+- ▁kar
+- ami
+- var
+- ながら
+- 恩
+- oc
+- ▁gesagt
+- llen
+- 是否
+- ▁meeting
+- 纳
+- ▁create
+- ▁ஒரு
+- ҡ
+- ▁Dann
+- ▁ré
+- ▁products
+- ▁Et
+- ▁вам
+- tis
+- ▁fun
+- ▁während
+- 了吗
+- ▁focus
+- ses
+- あの
+- ▁fond
+- ▁sí
+- んですけど
+- あと
+- ▁star
+- ▁Weise
+- 健康
+- ▁considered
+- َ
+- ▁não
+- 油
+- स
+- せ
+- んですね
+- ▁Weg
+- ▁broke
+- stra
+- ▁moving
+- ▁sah
+- set
+- ▁быть
+- 此
+- ▁leaving
+- ▁supposed
+- stan
+- ぶ
+- ▁million
+- ▁services
+- ひ
+- ▁charge
+- ▁ber
+- ▁drink
+- ▁test
+- ская
+- ▁enter
+- ▁gerade
+- 技术
+- ▁sudden
+- 式
+- wr
+- ги
+- ista
+- ▁wonderful
+- त
+- ▁season
+- ▁어
+- '",'
+- ▁difference
+- ▁jo
+- ▁আ
+- 未
+- اء
+- ▁knows
+- ▁geben
+- ▁rep
+- きました
+- ified
+- 为了
+- leg
+- iri
+- ▁mad
+- ▁positive
+- ▁grace
+- ▁пры
+- dri
+- ▁stone
+- なんか
+- ▁spent
+- 我在
+- ▁everyone
+- W
+- als
+- られる
+- ē
+- ▁bat
+- ▁Kinder
+- 修
+- 方法
+- э
+- ▁principal
+- 怎么样
+- ▁sagt
+- ▁iron
+- dor
+- pt
+- ▁том
+- ก
+- 坐
+- ▁Ve
+- アメリカ
+- ▁ball
+- ▁fünf
+- ▁Jo
+- ▁Fall
+- ped
+- ▁energy
+- ▁sixty
+- ▁honor
+- 雷
+- ius
+- لی
+- ▁min
+- رو
+- ジ
+- ▁ganze
+- ▁بر
+- というのは
+- 间
+- чу
+- ▁ப
+- ▁imagine
+- zar
+- ţi
+- ▁covered
+- ▁exist
+- 人的
+- ▁latter
+- ▁receive
+- ▁wollte
+- ▁cast
+- 任
+- ▁daar
+- 不同的
+- nor
+- ▁moral
+- vers
+- ▁global
+- ▁fo
+- ▁esta
+- pro
+- 強
+- eu
+- ように
+- たい
+- تر
+- ▁natürlich
+- ぐらい
+- bed
+- сти
+- uru
+- ▁allow
+- ே
+- ▁closed
+- ▁cases
+- 离
+- nos
+- ▁bound
+- ஸ்
+- but
+- 把它
+- ▁kun
+- ▁shut
+- 左
+- ▁dafür
+- ▁arrived
+- ▁huge
+- pf
+- ▁technology
+- ▁besides
+- ▁term
+- ா
+- ット
+- ことが
+- ph
+- 自由
+- 改变
+- ▁Kon
+- 服务
+- 二十
+- ニ
+- 许多
+- 注意
+- ろ
+- ay
+- ▁At
+- cin
+- ▁entirely
+- ▁tears
+- ですか
+- ojn
+- 教育
+- ▁particularly
+- ▁pod
+- ▁девять
+- 有人
+- ey
+- ▁hill
+- 直接
+- 试图
+- ▁ل
+- ▁großen
+- ão
+- ▁Ze
+- ▁而且
+- ella
+- ▁actual
+- looking
+- ata
+- bli
+- по
+- のです
+- all
+- ▁І
+- 方式
+- 状況
+- ants
+- ▁anti
+- pre
+- ▁Our
+- ▁turning
+- ació
+- しか
+- هم
+- 学习
+- ▁ко
+- 一天
+- 步
+- yn
+- ▁lassen
+- んですが
+- rt
+- yon
+- 准备
+- ビ
+- 维
+- 药
+- ▁cat
+- ward
+- ▁percent
+- ▁ahead
+- ▁notice
+- spe
+- 离开
+- 人类
+- ▁justice
+- ▁sollten
+- iga
+- ▁spend
+- pul
+- ▁significant
+- ▁fest
+- ▁ي
+- ▁needs
+- 收
+- hand
+- ай
+- ical
+- ▁sand
+- だから
+- ட்
+- лю
+- ▁sky
+- ▁killed
+- 语
+- ▁dos
+- ▁bak
+- nder
+- eva
+- 를
+- ▁Beispiel
+- rà
+- лу
+- 留
+- 考え
+- ▁holy
+- 宝
+- 百
+- ▁Dr
+- ▁Pi
+- ▁Go
+- ส
+- ▁europe
+- ▁rise
+- ▁animal
+- های
+- 完
+- ▁macht
+- ение
+- ▁jeder
+- ▁labor
+- ▁бо
+- ▁John
+- ▁deux
+- 参加
+- ▁island
+- ▁kids
+- ▁size
+- ▁occasion
+- cio
+- ▁beside
+- ぎ
+- iki
+- ▁park
+- ул
+- illa
+- ви
+- og
+- 中心
+- され
+- 福
+- board
+- 島
+- ▁jack
+- ▁complete
+- 提
+- ▁east
+- ▁religion
+- ▁note
+- ein
+- ▁usual
+- ture
+- ▁horses
+- ▁police
+- ▁youth
+- ▁unserer
+- vil
+- ▁uma
+- ▁wise
+- 短
+- ハ
+- ón
+- ola
+- 角
+- tru
+- ▁crowd
+- ▁george
+- ▁welche
+- ▁пред
+- ḍ
+- ▁gute
+- ▁silent
+- 吉
+- 兵
+- 跑
+- していた
+- 影响
+- ム
+- qu
+- ▁пятьдесят
+- 查
+- ▁opportunities
+- hr
+- ▁tend
+- lon
+- ▁peter
+- av
+- になった
+- ima
+- 师
+- ▁education
+- 見て
+- 好好
+- ▁К
+- ▁«
+- лен
+- bert
+- ți
+- ▁countries
+- ボ
+- ably
+- ▁sam
+- ability
+- ▁нам
+- 你会
+- கள்
+- ▁blind
+- ど
+- ▁trans
+- '40'
+- ▁восемь
+- ▁former
+- ▁مع
+- vy
+- nan
+- ▁ただ
+- ▁molt
+- ▁tut
+- ▁су
+- ▁“
+- 完成
+- ▁sicher
+- ▁prop
+- ▁noble
+- ▁ubu
+- ▁tax
+- ▁della
+- ▁meinen
+- lé
+- mun
+- 案
+- mba
+- 客
+- ▁ziemlich
+- เ
+- тор
+- ▁Hand
+- ▁warum
+- ▁lange
+- ман
+- ▁letters
+- ▁была
+- ▁reading
+- ▁Denn
+- ▁parents
+- 演
+- 配
+- ▁Car
+- rei
+- ▁attack
+- ▁чем
+- лся
+- teil
+- ▁surprise
+- 来自
+- にある
+- 破
+- ▁heraus
+- pon
+- 文化
+- 走了
+- ▁aunt
+- ▁needed
+- ▁ngo
+- ▁slow
+- low
+- 数据
+- ▁offer
+- といいます
+- 至
+- ▁pleasant
+- ▁list
+- ett
+- arbeit
+- 罪
+- ▁entire
+- ৰ
+- 无法
+- uta
+- ▁lines
+- sent
+- ▁Dies
+- 만
+- ▁merely
+- ▁cry
+- ▁está
+- ▁ac
+- ierte
+- iro
+- tha
+- mbo
+- ▁worse
+- ▁pra
+- 待
+- kla
+- ▁cha
+- 另一个
+- huh
+- tag
+- ▁unto
+- があります
+- ▁absolutely
+- ▁marriage
+- tim
+- 跳
+- ened
+- ▁п
+- sia
+- 算
+- ▁gesehen
+- ▁ни
+- ▁original
+- ▁этом
+- nou
+- ▁tall
+- ネ
+- ▁based
+- aka
+- ร
+- ov
+- 管理
+- 似乎
+- ▁ideas
+- ador
+- mut
+- bla
+- ▁этого
+- 我会
+- ые
+- نی
+- ▁هو
+- ▁store
+- 取り
+- 全部
+- 也不
+- ▁kandi
+- 获得
+- ▁somewhat
+- 대
+- ▁kleine
+- ▁watching
+- 我要
+- 经济
+- ▁forest
+- 原因
+- ▁weak
+- kommen
+- ats
+- 問題
+- ▁takes
+- шы
+- 总是
+- தி
+- سی
+- ▁steps
+- ▁nog
+- ▁всё
+- rat
+- 利用
+- ▁mà
+- 一定
+- 政治
+- ▁Or
+- ▁đó
+- ▁members
+- ▁silver
+- ▁discovered
+- 你在
+- ге
+- án
+- ▁ع
+- 割
+- า
+- 好了
+- ▁including
+- 科学
+- ский
+- tta
+- эл
+- ▁tan
+- ▁property
+- тер
+- ▁bow
+- け
+- ▁appear
+- 计划
+- gal
+- itat
+- 通常
+- ▁draw
+- 飞
+- 守
+- 掉
+- 谷
+- 未来
+- ▁serious
+- ▁bottom
+- ils
+- rn
+- night
+- ▁circumstances
+- ▁是的
+- vre
+- ▁Geld
+- اب
+- ▁escape
+- ▁chapter
+- ét
+- ▁movement
+- J
+- ▁proud
+- ell
+- “
+- ▁fortune
+- ▁customer
+- そ
+- nden
+- tung
+- ▁iz
+- лы
+- ▁breath
+- ů
+- ▁san
+- 满
+- ▁ph
+- ▁Jahr
+- werk
+- 发生
+- 好像
+- ▁Wo
+- ▁وال
+- ▁thick
+- ▁mountain
+- 党
+- tz
+- 可能会
+- ū
+- ▁changes
+- ения
+- ав
+- ▁sondern
+- 故事
+- бо
+- ი
+- ந்த
+- ケ
+- ▁prepared
+- ▁sad
+- ▁Stadt
+- ▁completely
+- 母
+- 感染
+- 迪
+- ▁fish
+- dis
+- ▁mort
+- ete
+- до
+- ▁points
+- 击
+- ▁besser
+- ▁gw
+- oni
+- rwa
+- ▁dels
+- oma
+- ев
+- ▁increased
+- ▁bij
+- ▁america
+- ▁Herr
+- ।
+- ▁instant
+- ica
+- ▁Moment
+- ▁Über
+- ▁где
+- rin
+- ▁erste
+- ▁named
+- ரி
+- 如
+- モ
+- ▁conditions
+- 停
+- ▁guard
+- 리
+- lic
+- ▁здесь
+- ▁Ya
+- bit
+- ис
+- ස
+- 共
+- tek
+- 夏
+- ep
+- ын
+- ُ
+- ▁mijn
+- これは
+- ظ
+- 几乎
+- ட்ட
+- 精
+- ▁effort
+- ▁judge
+- 仍然
+- ▁habit
+- ▁credit
+- sim
+- 牛
+- ▁jesus
+- 江
+- ▁тем
+- ▁afterwards
+- ม
+- ▁ok
+- lung
+- 信息
+- 追
+- ▁cal
+- най
+- ▁sono
+- 皆さん
+- ное
+- ▁gives
+- aga
+- 忙
+- '12'
+- ▁advantage
+- ně
+- rum
+- 谈论
+- itu
+- ray
+- 毛
+- 素
+- ret
+- ▁research
+- ▁meaning
+- ator
+- ▁zou
+- ці
+- lay
+- ック
+- ▁Grund
+- ▁glass
+- ▁nobody
+- ▁video
+- kle
+- ▁Ende
+- gy
+- 程
+- 感到
+- イン
+- 养
+- හ
+- ▁Ort
+- ▁cash
+- hir
+- ▁count
+- ▁inform
+- ▁また
+- ▁neue
+- encia
+- 歳
+- lü
+- ▁taste
+- ▁porque
+- ▁box
+- ▁fin
+- ற
+- ことを
+- ân
+- ▁watched
+- ▁college
+- ▁막
+- ▁York
+- ▁того
+- ▁storm
+- ▁mass
+- みんな
+- 约
+- 我们在
+- ▁weather
+- 拍
+- े
+- å
+- 排
+- ▁scene
+- ▁laugh
+- ▁dry
+- ▁total
+- 确实
+- ということ
+- zin
+- 对于
+- ▁beiden
+- ▁memory
+- ▁się
+- ▁Els
+- ▁madame
+- 其中
+- '15'
+- তা
+- mak
+- ▁events
+- 'off'
+- ▁greatest
+- тай
+- ▁aba
+- 界
+- ping
+- 型
+- press
+- ▁Sache
+- ▁speech
+- ▁pick
+- ados
+- ▁played
+- ▁companies
+- idad
+- ▁passage
+- 下来
+- こんな
+- ▁weg
+- ▁benefit
+- ▁alt
+- ▁War
+- 退
+- light
+- 视
+- ▁economic
+- ▁тут
+- ▁edge
+- ▁international
+- gue
+- net
+- 晚上
+- 照
+- 医生
+- ▁Bar
+- ева
+- ▁record
+- bol
+- ロシア
+- zel
+- 屋
+- ▁fallen
+- ▁tak
+- ▁animals
+- 超
+- ż
+- ほど
+- ▁تا
+- odd
+- ▁我是
+- ▁kill
+- 菜
+- ක
+- ille
+- ación
+- ს
+- 今日
+- ▁app
+- ▁tone
+- cos
+- ▁shore
+- ▁balance
+- abo
+- ▁إلى
+- ▁dia
+- ▁solid
+- ın
+- 香
+- 勝
+- 思い
+- wed
+- ো
+- 肯定
+- zy
+- ری
+- ぐ
+- ▁denken
+- ux
+- rang
+- kul
+- äh
+- ▁spread
+- kar
+- ノ
+- ily
+- table
+- ▁moon
+- 班
+- 娘
+- 端
+- ад
+- over
+- ua
+- эн
+- じゃない
+- 昨日
+- なら
+- wing
+- 服
+- die
+- ще
+- 因
+- ▁için
+- ▁sharp
+- tem
+- いく
+- ▁без
+- ▁jeden
+- sk
+- 苦
+- vad
+- eta
+- fre
+- 시
+- ▁comfort
+- ப
+- mb
+- ɛ
+- ран
+- 云
+- 回到
+- ▁understood
+- 機
+- ▁الت
+- ▁Haus
+- 称
+- 树
+- ガ
+- ▁ihrem
+- ▁های
+- ▁можно
+- ▁search
+- iko
+- тра
+- vé
+- 费
+- ▁bag
+- られた
+- ▁honour
+- 经常
+- ▁tom
+- ría
+- டி
+- 無
+- ▁casa
+- 洗
+- テ
+- ▁bekommen
+- ▁shape
+- 必要
+- rent
+- 望
+- ▁fla
+- 非常に
+- ▁cover
+- ▁даже
+- 井
+- 員
+- 换
+- ▁kommt
+- ved
+- ちゃん
+- 武
+- 是个
+- ▁exp
+- ▁interested
+- ▁prove
+- 나
+- qua
+- ▁ben
+- 全国
+- ▁believed
+- ▁fixed
+- म
+- ▁brave
+- law
+- ▁surface
+- рэ
+- ▁double
+- ▁reply
+- 場
+- प
+- dir
+- ني
+- ▁Daten
+- ▁dress
+- ▁Gefühl
+- от
+- pped
+- 的时间
+- ▁song
+- some
+- ▁sau
+- ▁stock
+- ▁If
+- मा
+- ▁nó
+- ara
+- ย
+- ▁laws
+- ▁square
+- ▁bru
+- ير
+- ここ
+- ▁gehört
+- ▁professor
+- ▁emp
+- ▁excited
+- ▁لل
+- ▁Unternehmen
+- stone
+- 玉
+- 市场
+- ▁finding
+- ▁sing
+- ith
+- ▁environment
+- gg
+- ▁schnell
+- enden
+- gera
+- 最终
+- 刘
+- ▁popular
+- ▁ont
+- ▁ancient
+- স
+- 联系
+- asi
+- 帮
+- ▁royal
+- 右
+- ▁pie
+- став
+- 身上
+- ▁leben
+- стра
+- ▁seventy
+- ▁parte
+- 个人
+- ▁throughout
+- ▁Fi
+- ▁いや
+- ▁ces
+- 大きな
+- ер
+- ▁smoke
+- 求
+- bri
+- 基
+- ນ
+- டு
+- ▁markets
+- ▁These
+- ため
+- ▁lake
+- '0'
+- ▁bal
+- ▁mere
+- 接受
+- ▁driven
+- 密
+- 登
+- 之间
+- ▁persons
+- ▁người
+- ▁explain
+- ▁seat
+- ▁putting
+- ▁af
+- 鱼
+- мы
+- imi
+- ார்
+- ▁pos
+- ▁telling
+- ▁час
+- ▁其实
+- ▁بال
+- 龙
+- ▁region
+- ▁center
+- ▁Ĝi
+- 系
+- ▁walls
+- ▁century
+- ▁mor
+- ▁mid
+- ว
+- ▁club
+- nel
+- 便
+- ▁pat
+- ▁range
+- ▁determined
+- ering
+- ▁Han
+- ▁prevent
+- ▁author
+- ப்ப
+- ▁sold
+- 呃
+- ▁எ
+- 音乐
+- ▁civil
+- ▁bas
+- com
+- aŭ
+- 黄
+- ▁accept
+- ▁stage
+- din
+- ▁narrow
+- ▁foreign
+- ob
+- ▁neck
+- 害
+- ▁رو
+- ▁issue
+- hor
+- 解决
+- 发展
+- tig
+- ▁production
+- ▁yellow
+- ▁soll
+- 证
+- ▁ladies
+- ▁hurt
+- ▁perfectly
+- ▁fill
+- und
+- 各
+- tant
+- 法律
+- 就会
+- jen
+- 杀
+- ол
+- 告
+- их
+- ▁german
+- ▁你知道
+- tura
+- 跟我
+- 付
+- ▁plant
+- 頭
+- ▁wel
+- ина
+- 想要
+- 大会
+- ِ
+- мі
+- вед
+- ▁approach
+- recht
+- gas
+- ▁access
+- ▁St
+- ▁familiar
+- ▁Ce
+- 去了
+- ▁proper
+- 常
+- ▁tri
+- との
+- ▁hotel
+- 伤
+- শ
+- ▁Dar
+- ▁learning
+- ▁Ju
+- 塔
+- ▁numbers
+- ▁fan
+- ▁bla
+- bin
+- ast
+- とも
+- rar
+- 有很多
+- ▁stick
+- ual
+- ▁hebben
+- gestellt
+- ▁因此
+- ী
+- pin
+- ත
+- ▁students
+- 対
+- lli
+- 站
+- りました
+- 投
+- ▁wished
+- ▁paris
+- ▁iki
+- ▁offered
+- ▁19
+- 关注
+- ▁cloud
+- тар
+- ▁grass
+- 城市
+- سا
+- rec
+- ур
+- war
+- ▁observed
+- ▁هم
+- ▁tam
+- ii
+- 之一
+- ▁favor
+- れば
+- ▁meinem
+- 乱
+- ته
+- ▁confidence
+- ▁agree
+- 父
+- 女人
+- cour
+- ?"
+- でしょう
+- 少し
+- ▁som
+- twa
+- osa
+- ea
+- ▁keinen
+- له
+- ▁degree
+- ▁который
+- ▁beneath
+- 秋
+- ▁était
+- فر
+- ▁mental
+- ▁Problem
+- work
+- 場所
+- tā
+- eux
+- biri
+- ▁marry
+- 微
+- ün
+- ged
+- ▁quality
+- ▁Seite
+- ła
+- 结果
+- лись
+- ▁firm
+- ▁baby
+- 奇
+- ları
+- rah
+- ▁clearly
+- ▁oil
+- 做的
+- ▁promise
+- ▁roman
+- ▁한
+- gin
+- lid
+- 伊
+- チーム
+- れる
+- '100'
+- ද
+- ▁farm
+- ▁человек
+- ▁finished
+- ًا
+- ▁darkness
+- ego
+- leri
+- lum
+- ▁clothes
+- ্
+- '60'
+- 死亡
+- 久
+- مان
+- цца
+- 保持
+- っと
+- 迷
+- 俺
+- ▁produce
+- ▁happiness
+- ▁Elle
+- ▁rule
+- ▁کار
+- 是我
+- みたいな
+- ▁playing
+- ▁gas
+- 列
+- ▁шестьдесят
+- 节目
+- たら
+- tet
+- uz
+- ane
+- ▁تو
+- وم
+- को
+- 家族
+- ▁golden
+- ▁あっ
+- 相信
+- ▁உ
+- ▁rein
+- 选
+- ▁scarcely
+- ▁seit
+- 你就
+- 几个
+- 友
+- 一般
+- ▁equal
+- ▁knowing
+- 回家
+- tern
+- ▁Ex
+- ச
+- 役
+- 回答
+- 果
+- ▁management
+- ▁lose
+- ール
+- ná
+- ▁brand
+- ▁sal
+- 啥
+- ▁könnten
+- ▁可是
+- зна
+- ▁opening
+- 試合
+- ター
+- 人生
+- morrow
+- ▁drawing
+- デ
+- ▁operating
+- ▁bạn
+- ▁profit
+- ▁Gra
+- ▁paul
+- 让他
+- rak
+- оч
+- glo
+- ette
+- ß
+- ▁loss
+- ▁stories
+- ▁wine
+- ▁sell
+- があった
+- 电话
+- hm
+- 脸
+- ▁attempt
+- ▁address
+- ▁bay
+- uku
+- 町
+- ▁leaves
+- ▁courage
+- spir
+- ▁media
+- ▁leading
+- ▁schwer
+- ▁ses
+- 的小
+- ▁efforts
+- ▁soldiers
+- ▁surely
+- ▁San
+- hel
+- lip
+- rü
+- iger
+- ▁Hier
+- ві
+- pot
+- 电影
+- ▁är
+- ▁passiert
+- ▁Але
+- än
+- ▁etwa
+- 他在
+- 春
+- ow
+- non
+- ▁produced
+- ▁flowers
+- ▁valley
+- ▁earlier
+- ▁Ik
+- cie
+- 人は
+- ▁holding
+- ▁eigenen
+- 理由
+- لي
+- ▁Mr
+- ▁sword
+- ▁supply
+- bot
+- ▁được
+- ▁weight
+- 副
+- ▁さあ
+- 주
+- 网
+- 视频
+- 妈
+- ி
+- ▁check
+- ▁walking
+- ▁flu
+- 連
+- 告诉我
+- ▁Europe
+- 枪
+- ▁additional
+- 这一点
+- ls
+- ▁pla
+- ிய
+- 助
+- pit
+- ▁fra
+- 因为我
+- ena
+- ification
+- eau
+- ako
+- ▁choice
+- による
+- 骨
+- бу
+- ▁vain
+- 自分の
+- ▁decision
+- cing
+- ▁如果你
+- 背
+- ▁Б
+- ▁gro
+- 라
+- ▁powerful
+- 弱
+- 地区
+- лан
+- 陈
+- きた
+- 拥有
+- 東
+- 这样做
+- ▁welcome
+- ▁recent
+- lam
+- 残
+- ▁happens
+- ▁structure
+- 在这个
+- fan
+- lit
+- وی
+- ▁formed
+- ▁han
+- ▁speed
+- ddy
+- zione
+- ▁meisten
+- ా
+- 是在
+- '14'
+- 狗
+- lí
+- 로
+- ▁gli
+- ▁seek
+- 一番
+- 组织
+- rio
+- ▁nation
+- tat
+- ▁houses
+- ▁travel
+- ▁uz
+- ▁тоже
+- ▁bird
+- ▁passion
+- ▁directly
+- 去年
+- ▁그냥
+- である
+- ious
+- lı
+- mul
+- вод
+- ▁basis
+- 明天
+- прав
+- ▁Can
+- ▁platform
+- ▁як
+- aire
+- どんな
+- 的这个
+- cker
+- The
+- ists
+- 治
+- gna
+- serv
+- ▁threw
+- ▁dropped
+- ▁divine
+- ▁deze
+- ▁그런
+- 这次
+- んですか
+- 读
+- ▁flat
+- ง
+- ▁Buch
+- 考
+- ир
+- uza
+- 考虑
+- dol
+- න්
+- ▁deliver
+- ак
+- vá
+- かった
+- 는데
+- ▁شد
+- スト
+- 」。
+- mat
+- fully
+- ugu
+- hör
+- ▁military
+- ▁letzten
+- 睡
+- ▁birds
+- ▁sake
+- ▁lying
+- ください
+- ▁Fe
+- жи
+- ▁required
+- ▁existence
+- ▁ничего
+- ▁addition
+- 横
+- 気持ち
+- ▁contract
+- 爸爸
+- ्
+- ள
+- バー
+- stu
+- из
+- bie
+- бе
+- ▁passing
+- 不了
+- try
+- 若
+- ▁ibi
+- ▁declared
+- セ
+- 唱
+- ▁Par
+- уч
+- ▁конечно
+- ▁эти
+- ▁anybody
+- 身体
+- ▁coast
+- aux
+- tum
+- きょう
+- си
+- ▁cri
+- কে
+- ▁내가
+- ныя
+- ▁available
+- rich
+- क
+- 建
+- '18'
+- ▁glance
+- ▁track
+- ▁unser
+- 尽
+- ▁page
+- بر
+- ること
+- bl
+- 良
+- chten
+- mic
+- ▁understanding
+- ▁orders
+- ▁practice
+- 源
+- су
+- 带着
+- ▁seva
+- ▁wichtig
+- vas
+- sin
+- 三个
+- bau
+- zwe
+- ole
+- kuru
+- 愿
+- 人家
+- stat
+- mond
+- こちら
+- she
+- ▁你们
+- ▁william
+- ▁caused
+- stein
+- ▁christian
+- ible
+- ient
+- ▁Mas
+- ▁catch
+- cas
+- ▁Wasser
+- ▁guy
+- ction
+- 時に
+- ▁uno
+- ▁vint
+- 脱
+- 轻
+- ▁eighty
+- ▁allen
+- 封
+- ▁şi
+- ▁host
+- ▁fly
+- ▁–
+- rate
+- rick
+- ▁Д
+- gil
+- uko
+- ▁bedeutet
+- ▁hung
+- ▁ears
+- ▁prison
+- ▁più
+- ▁ф
+- ▁двести
+- more
+- 例
+- ▁muy
+- 前に
+- ▁text
+- ▁centre
+- ▁presently
+- 熱
+- 情况
+- ▁wahrscheinlich
+- gle
+- ポイント
+- によって
+- ▁Augen
+- вал
+- genommen
+- 普
+- อ
+- chan
+- gre
+- 社区
+- ▁že
+- ▁này
+- 托
+- ▁curious
+- になります
+- ön
+- 해
+- ▁larger
+- jar
+- ▁этот
+- ▁pel
+- ін
+- ▁shows
+- ▁voi
+- 新的
+- つの
+- geben
+- 类
+- 夫人
+- 处理
+- ket
+- ▁instance
+- ▁colonel
+- ▁Okay
+- quer
+- ungs
+- ▁net
+- ▁indi
+- 比如
+- 年前
+- thi
+- 错
+- ▁ре
+- すごく
+- nne
+- 第二
+- ▁hath
+- 자
+- க்கு
+- 及
+- ▁pr
+- 这是一个
+- 首先
+- igkeit
+- ▁population
+- 면
+- ▁pentru
+- 了我
+- cen
+- ▁vision
+- 的事
+- ▁accident
+- じゃ
+- ▁matters
+- ▁elle
+- 解释
+- 返
+- ▁richtig
+- 也许
+- 群
+- व
+- 极
+- ▁spoken
+- ▁indian
+- ▁kwi
+- ▁coat
+- ▁hearing
+- っています
+- ▁gets
+- ざ
+- nic
+- 级
+- ▁shown
+- ▁هذا
+- ▁hospital
+- kes
+- での
+- би
+- ロー
+- ी
+- ▁officers
+- ▁port
+- ais
+- 一切
+- ग
+- ▁kleinen
+- ▁statements
+- 回去
+- ▁cro
+- 刚
+- ▁Bei
+- 데
+- ative
+- ▁gre
+- ним
+- '19'
+- 不好
+- 办
+- 乐
+- 沙
+- ▁shadow
+- ہ
+- ▁fool
+- 团
+- ▁今天
+- vert
+- ▁calm
+- 担心
+- bye
+- ▁motion
+- stru
+- プ
+- 現在
+- ▁ook
+- ▁faire
+- ▁aux
+- ▁thin
+- ▁basically
+- まして
+- ▁shook
+- 事故
+- 保护
+- গ
+- ▁henry
+- ▁member
+- lis
+- ▁thanks
+- ▁aware
+- ▁время
+- あります
+- されました
+- eri
+- ▁opposite
+- ო
+- ▁pocket
+- ▁task
+- ▁willing
+- ▁successful
+- فت
+- 秒
+- ▁bread
+- uw
+- চ
+- ▁ama
+- 变得
+- ▁tem
+- なので
+- ▁trong
+- ▁extra
+- ▁religious
+- ▁люди
+- ▁strategy
+- ▁Har
+- یه
+- ▁remains
+- ▁cousin
+- ус
+- ▁إ
+- 票
+- 挺
+- などの
+- 不用
+- ぞ
+- ▁fre
+- ですよ
+- ‘
+- 游戏
+- ▁beat
+- ese
+- dde
+- 而不是
+- 苏
+- ▁guys
+- ▁Aba
+- ▁central
+- 最高
+- それを
+- ches
+- jan
+- 容
+- ▁officer
+- pic
+- ▁wants
+- ched
+- ▁related
+- ▁bw
+- ふ
+- ▁noise
+- 杰
+- ▁possession
+- 開
+- lat
+- ▁pure
+- そんな
+- ▁Person
+- 造
+- ▁nti
+- 語
+- ▁otherwise
+- ▁gray
+- ▁carefully
+- ▁sogar
+- 産
+- вид
+- 一年
+- 项目
+- ண
+- ග
+- ▁thể
+- kom
+- ние
+- ด
+- 個
+- ▁daily
+- 衣
+- ▁comment
+- ▁我觉得
+- mek
+- ෙ
+- ors
+- 咱
+- ▁10
+- 穿
+- bat
+- ▁areas
+- mme
+- vol
+- ś
+- ▁pres
+- ▁sum
+- 防
+- cause
+- ですから
+- ற்ற
+- ▁stream
+- ▁今
+- ▁были
+- 手机
+- 段
+- igi
+- ▁ear
+- ▁scale
+- ▁activity
+- ▁drawn
+- uh
+- ▁smiled
+- ▁Oder
+- 出现
+- 控制
+- pat
+- ▁conduct
+- られ
+- istic
+- ▁ninety
+- ston
+- ▁surprised
+- ▁са
+- ▁cas
+- ▁engaged
+- pie
+- ▁你说
+- ▁Ent
+- nc
+- ▁fate
+- 動
+- 相当
+- ▁policy
+- 你想
+- ▁Form
+- ▁شما
+- あった
+- ரு
+- ▁aid
+- ▁bur
+- nar
+- ▁repeated
+- ▁jedoch
+- 休
+- ▁throw
+- 给他
+- 才能
+- ▁excellent
+- 結構
+- plo
+- ▁event
+- ▁Ben
+- ign
+- 我们可以
+- ▁семьдесят
+- دي
+- ▁больше
+- 渡
+- eh
+- ▁apart
+- 快乐
+- али
+- 改
+- ▁neu
+- ▁district
+- ij
+- sed
+- 男人
+- ▁gun
+- kte
+- lem
+- ▁versch
+- 应
+- ▁এ
+- يد
+- 《
+- ▁ay
+- мер
+- 様
+- ▁Ye
+- 层
+- ▁fue
+- dat
+- 饭
+- ismus
+- kä
+- ▁spite
+- ▁dire
+- ▁skin
+- ▁castle
+- sz
+- hl
+- ▁خ
+- ▁bil
+- ▁My
+- lig
+- ▁digital
+- 已
+- ▁dick
+- ும்
+- ▁monsieur
+- ▁species
+- ▁О
+- iku
+- ▁machine
+- 愛
+- ▁reform
+- ▁gentlemen
+- ▁fur
+- '16'
+- ▁regular
+- ши
+- far
+- ु
+- ▁mountains
+- ▁fat
+- ć
+- ас
+- ▁på
+- ▁fruit
+- ierung
+- ▁correct
+- 直到
+- 分享
+- ▁difficulty
+- ▁نه
+- こちらの
+- ি
+- ▁Spiel
+- ▁image
+- setzen
+- ▁review
+- ▁todo
+- ▁getan
+- 楼
+- об
+- heim
+- ım
+- ▁hoe
+- ï
+- ດ
+- вар
+- ▁advance
+- 放在
+- dic
+- 底
+- ▁reality
+- 根据
+- ▁portion
+- 強い
+- sor
+- ска
+- ▁ausge
+- ▁commercial
+- ▁created
+- ▁spa
+- 不想
+- tí
+- če
+- ▁还有
+- ela
+- ном
+- ▁temple
+- 同じ
+- ポ
+- ▁anyone
+- 姐
+- ▁보
+- 持
+- ▁keeping
+- udi
+- ▁message
+- 看着
+- ල
+- 数字
+- row
+- یت
+- ▁später
+- isa
+- 并不
+- 上帝
+- ▁crime
+- ▁forced
+- ▁noticed
+- ▁tired
+- ▁choose
+- ▁pieces
+- ▁đã
+- 想象
+- ソ
+- place
+- 活动
+- ▁creature
+- dom
+- ▁union
+- 茶
+- 父母
+- 更多的
+- 兄弟
+- legen
+- ▁wahr
+- '25'
+- 初めて
+- кой
+- fra
+- ques
+- anya
+- 我说
+- ник
+- 没事
+- ずっと
+- ▁proved
+- ▁plans
+- gh
+- ▁legal
+- 每个
+- ▁improve
+- لو
+- ▁pride
+- ▁bevor
+- 午後
+- 组
+- ▁erhalten
+- ▁знаю
+- ゴ
+- hy
+- ▁عن
+- ▁ту
+- ▁كان
+- 아
+- 大人
+- nem
+- ▁suffering
+- rad
+- 楽
+- ▁Well
+- ▁кон
+- ▁duke
+- ダ
+- ▁served
+- ▁sol
+- ša
+- пи
+- 我可以
+- ys
+- ạ
+- yle
+- 小姐
+- ▁dachte
+- fel
+- ▁obviously
+- 再次
+- ▁suit
+- ▁fully
+- ▁frank
+- 作品
+- ▁Kar
+- ピ
+- 性的
+- kim
+- 双
+- こういう
+- ▁bridge
+- pus
+- mont
+- ▁eh
+- 除
+- ▁res
+- ▁mention
+- ▁sounds
+- されている
+- ێ
+- 有些
+- 限
+- 情報
+- ▁لا
+- its
+- ▁blow
+- 精神
+- 做了
+- ▁été
+- ▁然而
+- ▁だから
+- nye
+- kwa
+- ▁peculiar
+- ore
+- ▁burst
+- ▁planet
+- ▁faint
+- ▁jemand
+- 样
+- сто
+- 伯
+- ▁nineteen
+- ▁tur
+- ▁asking
+- ▁pounds
+- ▁bringen
+- んですよね
+- ▁allem
+- 的工作
+- bona
+- ري
+- ▁starting
+- 明白
+- ヤ
+- ▁press
+- 款
+- ▁avait
+- ▁pray
+- 圣
+- 可能性
+- ▁gij
+- 紧
+- ▁height
+- 女孩
+- 告诉
+- nova
+- kir
+- eɣ
+- 梅
+- parti
+- ▁bereits
+- ▁att
+- ▁අ
+- wel
+- ▁david
+- ▁Thank
+- ▁measure
+- ▁date
+- ▁진짜
+- ▁nom
+- fru
+- 痛
+- stro
+- gt
+- 引
+- てる
+- hood
+- নি
+- dal
+- ▁forma
+- 我也
+- 余
+- ิ
+- ▁servant
+- 结
+- ங்க
+- fle
+- ▁stranger
+- ву
+- wana
+- '17'
+- ▁Frauen
+- ▁blo
+- ric
+- ▁Tom
+- auf
+- ней
+- ▁enjoy
+- 料
+- 書
+- 한
+- してる
+- ▁Unter
+- ▁moi
+- wie
+- nni
+- 他说
+- ▁Nu
+- ▁names
+- ult
+- ▁provided
+- ▁breast
+- vez
+- 政策
+- ▁palace
+- となる
+- 大约
+- ल
+- লে
+- ▁costs
+- ▁tap
+- тан
+- amu
+- 男性
+- 而且
+- ▁devil
+- ▁freedom
+- ▁charles
+- ھ
+- 存
+- ▁authority
+- bone
+- mble
+- ▁defend
+- ▁hoch
+- pens
+- ▁бар
+- ▁reasons
+- しい
+- 业
+- ▁Sam
+- ▁sechs
+- ません
+- ▁fancy
+- ▁fashion
+- 毒
+- ос
+- ▁intention
+- ▁alive
+- 倍
+- py
+- 静
+- 今回
+- ▁pair
+- 设计
+- ▁impression
+- 儿子
+- ör
+- ▁sex
+- 伦
+- ellen
+- ▁cyane
+- ▁desert
+- ▁gentle
+- 认识
+- ▁bell
+- ۆ
+- 잖아
+- ала
+- 多少
+- 压
+- 都有
+- ▁style
+- ṛ
+- ▁forms
+- ▁anxious
+- ▁teach
+- 哦
+- 词
+- ▁Pu
+- 更多
+- vent
+- 的生活
+- ▁genug
+- 怕
+- 很好
+- род
+- voll
+- ▁sup
+- ▁Menge
+- 假
+- 方向
+- mark
+- ▁circle
+- ▁weit
+- ▁тебя
+- ddi
+- cul
+- 自分
+- ▁twice
+- তে
+- 政
+- قا
+- ▁ordered
+- ▁gla
+- ▁mer
+- ▁honest
+- ▁сказал
+- のお
+- ▁Zi
+- ▁stra
+- ▁possibly
+- ▁angry
+- ▁س
+- ▁council
+- sol
+- loo
+- dam
+- ▁vida
+- 能力
+- gara
+- ši
+- ▁brief
+- ▁کرد
+- eln
+- 基本
+- ▁лет
+- ró
+- 打开
+- ▁20
+- 犯
+- nik
+- br
+- ▁leur
+- ▁gy
+- mine
+- ration
+- uli
+- ▁ocean
+- ▁origin
+- ▁Sant
+- ▁З
+- ▁geen
+- icht
+- 英国
+- ▁patient
+- ▁ordinary
+- ▁bent
+- ැ
+- ▁british
+- alt
+- ▁ای
+- ▁importance
+- ▁qua
+- ▁són
+- 喜
+- vie
+- ▁яго
+- 多くの
+- ▁legs
+- ство
+- ния
+- kins
+- ▁university
+- long
+- ▁вопрос
+- ▁eigentlich
+- ▁юм
+- ▁hören
+- 好吧
+- shed
+- lijk
+- rag
+- 姆
+- 令人
+- 인
+- 结婚
+- ▁rising
+- дер
+- leid
+- ▁Vater
+- ▁culture
+- ▁或者
+- ▁tongue
+- ▁murder
+- els
+- はい
+- '200'
+- 当中
+- ▁Kopf
+- arm
+- ▁somewhere
+- ▁volume
+- ▁heads
+- ega
+- ▁qual
+- ▁với
+- گر
+- nnen
+- 清楚
+- ▁safety
+- halt
+- ▁ganzen
+- কা
+- ▁zo
+- ades
+- 支
+- pend
+- 同时
+- 嗯
+- 스
+- ▁Mutter
+- 激
+- 選
+- تی
+- сть
+- ví
+- pet
+- ▁problems
+- ▁agreed
+- ▁physical
+- ▁investment
+- 内容
+- ▁truly
+- gus
+- wal
+- ▁row
+- わけ
+- ि
+- ▁security
+- iĝis
+- hab
+- ээ
+- ▁fe
+- ▁тебе
+- ology
+- ▁hate
+- ▁shoulder
+- ▁forgotten
+- ord
+- ▁breakfast
+- ▁rode
+- пла
+- ▁seventeen
+- ▁那你
+- ▁taught
+- ▁ему
+- 你是
+- ▁roof
+- vir
+- ▁aim
+- ▁cook
+- 迫
+- tä
+- ▁med
+- ▁mistake
+- ▁promised
+- ▁много
+- ▁dance
+- 运动
+- ▁сам
+- ▁Cha
+- されています
+- pun
+- ود
+- ▁China
+- ыш
+- ▁join
+- še
+- geb
+- 感谢
+- ▁versuchen
+- ▁waited
+- elijk
+- ▁exercise
+- ▁roll
+- ight
+- uda
+- ▁ability
+- ▁anne
+- rou
+- ▁altogether
+- ▁dare
+- izi
+- ▁remembered
+- plan
+- 息
+- нов
+- 氏
+- гүй
+- люб
+- 主要
+- ▁Jetzt
+- ▁kor
+- ▁би
+- ▁calling
+- fl
+- ▁làm
+- ▁Ch
+- lent
+- ▁fingers
+- ▁dangerous
+- சி
+- ▁weniger
+- ▁hart
+- 不可
+- ▁feelings
+- аль
+- ▁Tra
+- ▁kuri
+- ▁target
+- வா
+- ていた
+- ▁потом
+- old
+- 三十
+- ▁tent
+- ▁salt
+- part
+- 质
+- jn
+- けれども
+- ▁failed
+- 这不是
+- ▁glory
+- ule
+- 每天
+- از
+- ▁восемьдесят
+- 永远
+- ▁chose
+- uit
+- ▁そう
+- اق
+- ▁бу
+- 曾经
+- ▁streets
+- 汉
+- கி
+- ▁specific
+- nje
+- ▁wil
+- раз
+- 関係
+- any
+- を見
+- ▁вообще
+- tch
+- ▁terror
+- tausend
+- すること
+- 一位
+- 税
+- ▁bist
+- ға
+- 跟你
+- ▁empty
+- lief
+- 景
+- ▁helped
+- ▁method
+- ▁james
+- 刚刚
+- ▁vielen
+- zon
+- uni
+- bili
+- jā
+- stellen
+- ▁rot
+- ▁quietly
+- bon
+- ねえ
+- ▁famous
+- pping
+- パー
+- rim
+- 社
+- gor
+- ▁drop
+- 一条
+- ▁fil
+- ▁auto
+- 调查
+- urs
+- ▁express
+- zig
+- ▁grant
+- ziehen
+- ▁active
+- 都会
+- 消
+- ▁wear
+- ▁после
+- bal
+- 婚
+- 亮
+- ▁байна
+- iye
+- ▁пре
+- য়
+- ▁princess
+- ▁limit
+- 后来
+- 万円
+- ld
+- ▁ха
+- ło
+- ▁wij
+- 生命
+- keit
+- の中で
+- ائ
+- ▁ச
+- ▁bul
+- 探
+- ▁Für
+- ▁kingdom
+- cke
+- ▁series
+- ▁sufficient
+- 越来越
+- ▁busy
+- ▁grown
+- бор
+- '13'
+- 有一些
+- ▁Мы
+- ▁aussi
+- tika
+- 企业
+- ▁các
+- 行动
+- ▁issues
+- wert
+- bus
+- ▁rapid
+- зы
+- 告诉你
+- ூ
+- ▁Za
+- ▁شده
+- ▁такой
+- 赛
+- eli
+- もある
+- rz
+- ▁delight
+- ▁sang
+- なんです
+- ▁struggle
+- 堂
+- 在我
+- ▁online
+- ▁그거
+- tori
+- ▁explained
+- ño
+- 軍
+- про
+- ▁fault
+- aw
+- ▁deg
+- প
+- ім
+- rom
+- ▁ended
+- ▁guide
+- 目前
+- ート
+- ▁signal
+- ▁heat
+- či
+- ▁flow
+- 日本の
+- ▁そうですね
+- phy
+- nin
+- 念
+- 旧
+- tto
+- ▁aside
+- ▁gr
+- û
+- ium
+- ▁그래서
+- ▁liegt
+- 散
+- ▁És
+- ▁potential
+- ▁yesterday
+- ▁rights
+- ▁девяносто
+- ▁false
+- gia
+- 医
+- يل
+- nze
+- 違う
+- ▁pré
+- ▁sorrow
+- ▁highest
+- 医院
+- sp
+- されて
+- ▁hinter
+- ▁Wer
+- kā
+- ▁extremely
+- ▁Dia
+- ▁experiment
+- Laughter
+- иться
+- ▁judgment
+- 网络
+- ▁exact
+- ▁сегодня
+- اف
+- ▁fel
+- ▁hell
+- 你说
+- aya
+- 绝对
+- ▁dev
+- ▁verstehen
+- 駅
+- ▁listening
+- おいしい
+- 第一次
+- 的孩子
+- ▁strike
+- ▁connection
+- ▁sentence
+- ▁income
+- 赤
+- 居
+- ล
+- ▁European
+- ▁kuba
+- ▁pou
+- ▁greatly
+- ▁Па
+- 房子
+- ▁described
+- ского
+- したい
+- ກ
+- ▁klar
+- ▁recently
+- 萨
+- ▁priest
+- 旅行
+- ి
+- ▁margin
+- ое
+- ▁pressure
+- ▁touched
+- ▁trial
+- haus
+- ▁separate
+- ▁network
+- ▁refer
+- وي
+- était
+- tle
+- ▁fix
+- ▁previous
+- ▁sides
+- 요
+- 出了
+- 降
+- ▁companion
+- uf
+- uff
+- спе
+- 洛
+- dio
+- ▁때
+- ▁Fragen
+- 真是
+- ▁vers
+- ▁де
+- 那种
+- ▁older
+- ▁disease
+- ▁почему
+- 大的
+- ▁angel
+- lá
+- нем
+- ▁zehn
+- ▁beim
+- 進
+- ▁settled
+- হ
+- oso
+- ude
+- ▁gay
+- rung
+- ▁ride
+- 合作
+- ▁grande
+- ются
+- horn
+- cies
+- ▁becomes
+- ym
+- 剧
+- pla
+- 置
+- からの
+- দ
+- tom
+- 你要
+- ▁eleven
+- ▁When
+- bs
+- ▁людей
+- ▁meer
+- یا
+- ْ
+- ▁map
+- ▁gleich
+- ▁apparently
+- ▁naturally
+- ства
+- ҙ
+- 那是
+- gun
+- دی
+- 很难
+- 来的
+- ▁sage
+- ime
+- ▁onder
+- 英
+- ▁гор
+- ▁Prozent
+- 致
+- ming
+- ▁superior
+- aus
+- tă
+- ▁имени
+- sem
+- 游
+- ウクライナ
+- ▁Lebens
+- 即使
+- ▁hun
+- ▁intended
+- ▁philip
+- lio
+- ▁sixteen
+- сон
+- ▁involved
+- ыя
+- эд
+- ился
+- ▁liked
+- ▁capacity
+- 诺
+- まず
+- 轮
+- ▁vie
+- ▁Ci
+- iste
+- wand
+- ана
+- ▁interview
+- ▁troops
+- ▁Auto
+- eye
+- antes
+- лась
+- ▁même
+- ées
+- ▁belief
+- ້
+- 一家
+- gem
+- cara
+- າ
+- nal
+- ▁него
+- ▁title
+- nge
+- 我们将
+- 参与
+- 你能
+- 机会
+- 除了
+- 产
+- amos
+- ▁bald
+- ▁release
+- ▁corn
+- 谈
+- 粉
+- 创造
+- 親
+- 冲
+- でしょうか
+- ▁ice
+- ▁مت
+- ▁driving
+- ▁stars
+- щ
+- rry
+- ▁student
+- hundert
+- ▁highly
+- 展示
+- 止
+- ▁vote
+- ▁shop
+- ht
+- nehmen
+- ande
+- нд
+- ▁operations
+- eller
+- inde
+- ash
+- ек
+- lot
+- ruh
+- ա
+- ▁career
+- ct
+- ▁definitely
+- ▁chamber
+- ▁argument
+- ▁happening
+- tier
+- fr
+- phi
+- bes
+- 晚
+- гийн
+- ▁sans
+- 讨论
+- ono
+- ▁admit
+- 抓
+- ましょう
+- olo
+- zwa
+- 料理
+- ila
+- ▁ap
+- той
+- 収
+- ε
+- ▁avoid
+- burg
+- ▁thrown
+- vä
+- 実は
+- 校
+- ив
+- によりますと
+- ▁todos
+- ▁contra
+- 架
+- ▁aller
+- 整
+- やはり
+- ▁flo
+- 富
+- ▁suffer
+- ▁shoulders
+- 认
+- ▁minister
+- ▁Th
+- gro
+- ès
+- ▁match
+- ▁ந
+- tud
+- 滑
+- ▁bereit
+- spiel
+- tür
+- 俩
+- 期待
+- ▁concerned
+- 弄
+- ▁windows
+- 一定要
+- ийг
+- ▁harm
+- ▁cup
+- ori
+- hora
+- 末
+- বে
+- ज
+- ario
+- ▁thousands
+- ▁kennen
+- тур
+- লা
+- ▁pointed
+- ▁nay
+- ▁His
+- get
+- ▁ideal
+- 下去
+- 它是
+- ži
+- 箱
+- sing
+- rd
+- '80'
+- ▁fairly
+- ▁standard
+- ▁ее
+- 只能
+- ▁nose
+- 的故事
+- lus
+- kal
+- iamo
+- ▁spiritual
+- ▁해
+- ▁Pri
+- 田さん
+- dale
+- ▁carriage
+- яв
+- رس
+- ▁verwenden
+- ▁kinds
+- 要是
+- ▁Gott
+- 呼
+- tten
+- 块
+- ▁loud
+- ▁helfen
+- ▁joined
+- ▁себя
+- ▁bro
+- 吃饭
+- ▁kid
+- 索
+- 하는
+- 报告
+- vit
+- 可能是
+- ▁novel
+- tain
+- 县
+- 不管
+- ▁weitere
+- gam
+- ejo
+- anza
+- stri
+- ▁claim
+- reich
+- ▁excuse
+- த்
+- வு
+- 复
+- 叶
+- 就在
+- ▁fighting
+- ▁mari
+- пу
+- ▁ones
+- ▁temps
+- 猫
+- ▁affairs
+- gwa
+- ust
+- kana
+- ▁wealth
+- 症
+- 重新
+- ▁useful
+- dé
+- 百分之
+- ▁presented
+- ▁flight
+- 鲁
+- kam
+- 灯
+- ▁grey
+- cut
+- ▁пад
+- 记
+- 每个人都
+- cz
+- ▁papa
+- ique
+- ▁fri
+- rik
+- できない
+- ▁mix
+- lock
+- ▁loose
+- ность
+- fort
+- ▁Cor
+- بي
+- ▁niemand
+- ян
+- ▁cuando
+- ▁prayer
+- ▁faces
+- pad
+- bare
+- 歳の
+- ▁debt
+- ▁evidently
+- abi
+- ▁unseren
+- ▁möglich
+- まあ
+- ▁suggest
+- 付け
+- pha
+- ща
+- 动物
+- ▁woods
+- का
+- 你不
+- ▁Х
+- vien
+- ▁pull
+- ▁develop
+- 当你
+- ▁cruel
+- ▁そんな
+- ità
+- igung
+- 借
+- したら
+- ▁anders
+- ▁ف
+- بل
+- மா
+- 令
+- dus
+- 因此
+- 市の
+- になりました
+- ▁indem
+- ▁Regierung
+- ▁الح
+- ▁average
+- 时代
+- شر
+- ▁native
+- 速
+- ked
+- ▁ange
+- ▁خود
+- ▁X
+- ▁아니
+- ▁tin
+- ි
+- tü
+- 梦
+- hau
+- ила
+- тал
+- 愿意
+- ▁worst
+- por
+- வி
+- じゃないですか
+- 続いて
+- თ
+- bwa
+- wick
+- ssa
+- ▁lack
+- 表示
+- ▁considerable
+- ▁нужно
+- olu
+- hn
+- 志
+- ▁trail
+- ace
+- rup
+- ▁dogs
+- 勒
+- ath
+- ▁suggested
+- ▁torn
+- パン
+- ▁nos
+- ▁pal
+- 程度
+- 也有
+- ify
+- ▁bid
+- mwe
+- eix
+- 慢
+- ▁raz
+- ụ
+- 都在
+- tent
+- ▁пока
+- 父亲
+- 贵
+- haft
+- 条件
+- ск
+- र्
+- ▁Dan
+- 注目
+- ▁arbeiten
+- lichkeit
+- 是谁
+- bas
+- ▁gate
+- ▁மற்றும்
+- ▁male
+- uel
+- нь
+- ▁dabei
+- ▁sought
+- ▁raise
+- ters
+- elles
+- ▁какой
+- それで
+- ▁이제
+- ▁gain
+- 的想法
+- ▁bodies
+- book
+- ▁wohl
+- ▁それは
+- ゆ
+- hind
+- ыр
+- ▁constant
+- 明日
+- ▁heb
+- ▁好吧
+- 威
+- 心里
+- وت
+- ▁sacrifice
+- 然
+- ▁Männer
+- 普通
+- ▁labour
+- 产品
+- ете
+- ▁bekannt
+- 分析
+- 込み
+- 看起来
+- とき
+- 莫
+- ▁refused
+- 占
+- gé
+- 顿
+- comp
+- しない
+- ▁slightly
+- ▁хорошо
+- 道路
+- ▁fail
+- しっかり
+- ▁sheet
+- 招
+- ▁source
+- ▁Meinung
+- ▁dot
+- ▁Dank
+- 经
+- sign
+- 是你
+- ening
+- ум
+- ▁birth
+- ▁també
+- bury
+- コン
+- ▁hills
+- 顔
+- ui
+- ▁rapidly
+- ▁pur
+- ▁becoming
+- 我知道
+- 谢谢
+- 盘
+- ▁saved
+- kken
+- hang
+- ▁поэтому
+- ▁bringing
+- 谢谢你
+- ▁Nicht
+- mia
+- ▁tender
+- ▁showing
+- дел
+- える
+- 以来
+- pal
+- 也没有
+- ской
+- ▁computer
+- নে
+- ギ
+- ▁virtue
+- 曼
+- ▁nehmen
+- ره
+- 硬
+- ▁unknown
+- ▁tradition
+- fon
+- cr
+- ws
+- なる
+- ▁finger
+- ▁advanced
+- inte
+- мат
+- 句
+- сэн
+- ▁Film
+- 到底
+- avi
+- ▁papers
+- ▁governor
+- trag
+- ▁upper
+- 就可以
+- ▁じゃあ
+- cé
+- produ
+- кам
+- ▁Post
+- ▁audience
+- たち
+- anga
+- 切り
+- با
+- ▁extent
+- ▁peut
+- ▁pin
+- なぜ
+- halten
+- ▁solche
+- ▁gift
+- ▁э
+- endo
+- 側
+- ▁neuen
+- லை
+- ▁pity
+- ▁theory
+- sal
+- 提出
+- ▁Em
+- ▁¿
+- ▁alte
+- 場合
+- ▁division
+- 历史
+- ▁everywhere
+- iva
+- '70'
+- ▁Bir
+- 들
+- なって
+- pli
+- 北京
+- ization
+- 沉
+- ▁çok
+- ▁praise
+- ▁cities
+- term
+- yer
+- ▁Des
+- ▁Vo
+- ▁portfolio
+- ▁Tre
+- これから
+- று
+- 会社
+- 价
+- ▁первый
+- ▁dressed
+- ▁Auch
+- punkt
+- ел
+- 教授
+- ▁servants
+- 建立
+- ▁charm
+- ▁nächsten
+- ▁cell
+- ▁System
+- ベ
+- 靠
+- 育
+- зи
+- стр
+- cept
+- хи
+- ▁See
+- ▁talent
+- ▁Idee
+- wir
+- ▁où
+- 你有
+- right
+- ál
+- 魔
+- ▁кор
+- 射
+- ▁trip
+- ▁established
+- ül
+- ▁sua
+- ▁features
+- ▁الس
+- ▁distant
+- 至少
+- ▁rome
+- ▁Punkt
+- ▁kitchen
+- ced
+- ets
+- 都不
+- stellung
+- ▁wake
+- yl
+- 具
+- ецца
+- ▁advice
+- ешь
+- xa
+- 国内
+- なんて
+- ▁quarters
+- 坏
+- ibi
+- না
+- 折
+- 一百
+- ▁gewesen
+- ▁occurred
+- esa
+- эх
+- ▁Hause
+- kus
+- 結果
+- 刀
+- ér
+- ▁respond
+- 満
+- ▁reflect
+- phe
+- (
+- ır
+- ▁moments
+- ▁Park
+- isten
+- ワ
+- ේ
+- ▁yr
+- ▁beg
+- ▁аб
+- کن
+- ▁mode
+- ▁ذلك
+- jou
+- ▁forces
+- 家里
+- ised
+- ▁committed
+- ▁hy
+- ▁Г
+- 是如何
+- ña
+- zza
+- 判断
+- mber
+- その後
+- ▁announced
+- ▁Tri
+- ▁quatre
+- ▁bought
+- дар
+- уд
+- ности
+- 空间
+- 立ち
+- cro
+- 核
+- 怪
+- 刚才
+- ▁думаю
+- ständig
+- main
+- ▁robert
+- ▁discover
+- ара
+- ▁pace
+- ▁guidance
+- раб
+- ▁include
+- 業
+- ▁kiss
+- mbe
+- 藤
+- ▁pulled
+- ▁había
+- ▁بعد
+- cions
+- 言葉
+- ▁Gesicht
+- بار
+- nell
+- ▁darin
+- сі
+- 恶
+- lf
+- ▁teeth
+- ▁How
+- ▁пе
+- ej
+- を受け
+- 活動
+- ▁lamp
+- '24'
+- ▁smart
+- ▁leg
+- ீ
+- 礼
+- rac
+- ▁lovely
+- pose
+- きます
+- বা
+- த்து
+- 总统
+- 都没有
+- зе
+- ает
+- 所有这些
+- ▁card
+- ▁jane
+- 警
+- ▁facts
+- dur
+- ▁ankaŭ
+- ▁我就
+- mir
+- 身边
+- ▁sunday
+- 投票
+- ▁prefer
+- ră
+- 黒
+- ema
+- ▁detail
+- ▁shame
+- ▁oli
+- 電
+- 마
+- ▁fut
+- str
+- 主义
+- ож
+- ▁block
+- บ
+- ▁Cu
+- 世纪
+- 番組
+- 航
+- ▁هذه
+- ▁frei
+- ▁этой
+- ře
+- ▁closely
+- ▁intelligence
+- ▁erst
+- 卫
+- 意识到
+- ì
+- ▁hang
+- mma
+- ▁hearts
+- ▁leicht
+- ▁drove
+- app
+- nga
+- нос
+- 支援
+- ske
+- 厚
+- ▁bud
+- ▁அவர்
+- ▁One
+- 进来
+- ▁afford
+- back
+- ▁zal
+- 什么时候
+- ▁whispered
+- 現場
+- ▁medical
+- 旅
+- 真正的
+- pra
+- être
+- kor
+- ▁pop
+- 建议
+- tı
+- 増
+- ico
+- ▁Warum
+- ▁போ
+- ▁protection
+- ▁details
+- ▁Ol
+- ▁milk
+- 搞
+- 顶
+- eza
+- いろんな
+- ▁comfortable
+- ことは
+- ▁Ob
+- 论
+- ▁section
+- 属
+- oon
+- ▁department
+- ат
+- ▁다
+- 博
+- ▁gathered
+- 证明
+- ▁existing
+- ން
+- 게
+- von
+- 尾
+- izing
+- ▁affection
+- かなり
+- ech
+- ▁metal
+- ▁liebe
+- 蒙
+- sko
+- ここで
+- 他是
+- ür
+- pé
+- 势
+- ▁dozen
+- фи
+- 影
+- тро
+- 最好的
+- ▁secure
+- lab
+- ▁extraordinary
+- ▁anyway
+- ▁cosa
+- ▁listened
+- 货
+- sus
+- ▁brauchen
+- ▁slight
+- nta
+- ▁Namen
+- ますね
+- ▁conclusion
+- ▁female
+- нее
+- سم
+- 抗
+- пол
+- tos
+- ▁gene
+- 观
+- ▁laughing
+- ▁represent
+- ▁manage
+- 鬼
+- ▁syn
+- ▁solo
+- ▁crown
+- ▁handsome
+- ▁Pre
+- town
+- ▁Video
+- 领
+- ▁ở
+- ▁heeft
+- 我们有
+- 延
+- ▁bitter
+- ▁carrying
+- 一会儿
+- ▁daha
+- 我有
+- ▁„
+- 感情
+- ▁nurse
+- ు
+- 胡
+- ▁smooth
+- ▁iran
+- ▁él
+- 運
+- 経済
+- ▁indians
+- 是为了
+- sicht
+- 位置
+- eḍ
+- ▁donc
+- ▁больш
+- ▁Hal
+- მ
+- nek
+- 象
+- ▁treatment
+- 我已经
+- っていうのは
+- medi
+- ත්
+- ▁অ
+- டை
+- なんだ
+- ப்பு
+- ▁consideration
+- ▁த
+- 制作
+- тэй
+- tions
+- 不到
+- ▁appears
+- ong
+- дал
+- 的一些
+- fini
+- ▁Wissen
+- ▁当然
+- 今回の
+- ▁developed
+- 宫
+- 运
+- ▁careful
+- 新闻
+- ▁obliged
+- ế
+- ▁projects
+- もっと
+- ▁treat
+- kh
+- ŝ
+- ের
+- ▁discussion
+- ▁clar
+- ▁sự
+- べ
+- ▁về
+- zie
+- 思考
+- ▁reward
+- 环境
+- ヒ
+- ▁П
+- kö
+- ▁дело
+- 参
+- ボール
+- гі
+- 冬
+- ente
+- ▁resources
+- ▁வி
+- ▁второй
+- 日に
+- 네
+- ència
+- cks
+- şi
+- ා
+- ▁independent
+- ▁professional
+- さらに
+- ▁Is
+- 地说
+- 各种
+- blick
+- пе
+- ▁staff
+- ▁Raum
+- iyo
+- 铁
+- 是因为
+- zan
+- 十分
+- ▁role
+- 阳
+- 引き
+- ▁тогда
+- pl
+- ティ
+- ▁rough
+- 举
+- ່
+- 冰
+- ▁nta
+- ▁jest
+- ▁focused
+- ▁drug
+- ▁mighty
+- 記録
+- ▁ප
+- द
+- fli
+- aha
+- ctor
+- bor
+- ên
+- bank
+- 日の
+- az
+- ो
+- ▁Q
+- 了吧
+- ▁transport
+- hält
+- ▁lies
+- ▁wow
+- ▁schools
+- wit
+- ▁reg
+- ▁smaller
+- それは
+- 最大
+- ▁Woche
+- ▁кол
+- rap
+- 想到
+- ን
+- ▁systems
+- ▁mé
+- ball
+- あれ
+- なかなか
+- 幸福
+- cion
+- のに
+- ▁response
+- tá
+- ▁compared
+- ▁Mon
+- ▁imagination
+- ▁pare
+- 是吧
+- ค
+- رف
+- ▁200
+- nyi
+- 做出
+- рав
+- ▁louis
+- ppen
+- 曾
+- 会有
+- ▁realized
+- каз
+- ▁Wort
+- ▁Chris
+- ▁artist
+- ção
+- くる
+- ▁element
+- sho
+- ▁sieht
+- 归
+- 说了
+- gl
+- este
+- تا
+- ý
+- gate
+- einander
+- dí
+- ciones
+- 随着
+- ต
+- 欢迎
+- 年代
+- uga
+- 岛
+- 独
+- nom
+- nken
+- げ
+- 职
+- ▁kurz
+- ▁permit
+- 尽管
+- ▁organization
+- vier
+- ère
+- ▁extreme
+- 危险
+- ژ
+- fold
+- ▁training
+- 现
+- hat
+- 寻找
+- ▁spr
+- ings
+- ▁University
+- 行为
+- カー
+- ▁Dollar
+- ▁incident
+- mbi
+- gon
+- 塞
+- ▁David
+- ▁instrument
+- ▁hal
+- star
+- ▁questo
+- ▁rare
+- でしょ
+- 价值
+- ▁Schw
+- 嫌
+- ▁cop
+- ã
+- უ
+- ▁organis
+- 把我
+- ▁Neu
+- யா
+- られて
+- ▁cya
+- ität
+- ▁leader
+- 歩
+- 仕事
+- ▁smiling
+- ▁まずは
+- imo
+- ▁Kind
+- бер
+- ం
+- ▁beide
+- peri
+- 受到
+- 暗
+- ▁contrary
+- ▁saint
+- ▁deine
+- ▁treasure
+- llo
+- ▁hidden
+- ierten
+- iyor
+- ▁patients
+- 状態
+- ▁mess
+- ▁unbe
+- ▁skill
+- nza
+- ▁eso
+- ▁joe
+- ▁principle
+- 示
+- ▁realize
+- 领域
+- ▁EU
+- ▁relief
+- ▁ain
+- 的歌
+- يت
+- ▁Pen
+- ちゃんと
+- ▁arab
+- فا
+- مر
+- ▁Che
+- lim
+- ▁succeeded
+- pos
+- ▁fleet
+- wind
+- өр
+- 同样
+- ▁measures
+- 豆
+- 公共
+- ▁actions
+- ▁temper
+- жы
+- ▁不过
+- һ
+- ▁gan
+- ayo
+- ってる
+- ▁carrer
+- ▁liberty
+- রা
+- ▁ئا
+- ▁setting
+- оў
+- ट
+- ▁protect
+- 藏
+- 我是说
+- ▁colour
+- ▁более
+- 刺
+- ▁function
+- blo
+- ▁estate
+- ▁invest
+- sid
+- ▁families
+- ▁Sha
+- ыл
+- ун
+- ▁wore
+- nat
+- ▁steht
+- 同意
+- 劳
+- ppe
+- ▁falling
+- 必
+- 私は
+- 委
+- 简单
+- ▁suffered
+- に対して
+- тэ
+- টা
+- 随
+- ▁Sp
+- ▁frequently
+- hol
+- ▁somehow
+- 不可能
+- 十二
+- ▁partner
+- itude
+- ▁flesh
+- ▁powers
+- اع
+- 你看
+- ▁olarak
+- ▁ships
+- '500'
+- cus
+- 』
+- ▁yi
+- اه
+- ▁ج
+- 唯一
+- ист
+- ▁stellen
+- 有时
+- ▁campaign
+- uje
+- 『
+- ▁solution
+- aron
+- 上了
+- ubu
+- ery
+- shy
+- ydd
+- ▁nevertheless
+- 投资
+- ▁overall
+- 床
+- 対応
+- ▁affair
+- 我不
+- ▁awful
+- ▁Europa
+- ▁như
+- ▁stark
+- ▁begann
+- ▁Recht
+- ▁picked
+- ▁Ihrer
+- kazi
+- 基本上
+- 说的
+- bia
+- ▁accepted
+- ▁sp
+- ▁request
+- фе
+- ▁Freund
+- ency
+- 功
+- vari
+- shaka
+- ▁favour
+- ▁direkt
+- ▁euch
+- ක්
+- ▁plenty
+- ▁ĉi
+- 有了
+- unt
+- ём
+- 等等
+- ▁businesses
+- ▁seized
+- もあります
+- dien
+- ów
+- ▁grim
+- 特に
+- 不过
+- ▁glauben
+- hal
+- ▁kra
+- üh
+- ▁conference
+- ▁zeigen
+- ▁motor
+- ▁пера
+- 患者
+- юць
+- ▁peu
+- ▁thế
+- 以为
+- 薬
+- sey
+- 产生
+- ▁faithful
+- ▁satisfied
+- fic
+- spect
+- yor
+- lly
+- ▁dust
+- ▁dreadful
+- 即
+- ▁하
+- 私の
+- ▁うん
+- ▁jim
+- ▁pet
+- 套
+- 一直在
+- 에서
+- ▁funny
+- viv
+- 股
+- weg
+- cs
+- ▁expressed
+- ▁مو
+- 累
+- 上げ
+- mara
+- ▁foundation
+- аны
+- ont
+- ▁tip
+- gul
+- وس
+- 職
+- 把他
+- さんに
+- uv
+- にして
+- ▁Sal
+- ▁remarked
+- ここに
+- ▁appeal
+- pel
+- dul
+- ं
+- ▁meal
+- ▁Internet
+- ▁bob
+- ▁fields
+- 前の
+- 生物
+- ▁possess
+- ▁soldier
+- эт
+- つけ
+- ▁horror
+- ▁alla
+- лет
+- 変
+- んだよ
+- ▁kal
+- 你也
+- veni
+- 还没有
+- cri
+- 贝
+- می
+- 当我们
+- лар
+- ome
+- 洋
+- дан
+- ▁воз
+- ▁ක
+- gie
+- 吸
+- 预
+- ▁excitement
+- ▁거야
+- 是一种
+- ▁contact
+- ▁rules
+- 両
+- ▁mac
+- ▁negative
+- ▁Ist
+- 敢
+- ▁Гэта
+- ▁connected
+- ▁universal
+- ▁Gar
+- ▁irgendwie
+- pil
+- ▁majority
+- ▁destroy
+- ▁Los
+- 蒂
+- hod
+- 我去
+- prob
+- kol
+- がありました
+- ▁Den
+- ோ
+- ▁sé
+- ▁relationship
+- со
+- ▁bore
+- ▁lifted
+- 编
+- tory
+- ▁Körper
+- ▁fu
+- ▁whenever
+- かもしれない
+- ▁sprang
+- வே
+- と思う
+- 了一
+- cat
+- ▁我要
+- これを
+- top
+- 汽车
+- зі
+- 害怕
+- '90'
+- ▁iyi
+- ▁With
+- ▁şey
+- ▁qué
+- ▁emperor
+- ▁lock
+- oh
+- ▁Và
+- ▁flag
+- ▁Reihe
+- 结束
+- ▁tail
+- ▁pardon
+- ians
+- ред
+- ▁proof
+- ▁fal
+- ▁protest
+- ▁parties
+- 地域
+- 死了
+- rian
+- 你必须
+- яр
+- ▁rid
+- ▁amazing
+- kas
+- మ
+- nig
+- 袋
+- Z
+- ▁hello
+- ட்டு
+- ké
+- кова
+- ▁challenge
+- 但我
+- сад
+- ▁Pan
+- act
+- isto
+- 我没有
+- زی
+- ▁savage
+- tisch
+- ▁Angst
+- ▁spo
+- 丁
+- mise
+- 弹
+- 我都
+- 可是
+- ▁prior
+- だって
+- ▁West
+- ▁adam
+- ▁nest
+- 我还
+- ▁resist
+- ▁Antwort
+- ▁rev
+- ▁수
+- ▁mot
+- Y
+- ▁இந்த
+- ▁decide
+- ▁wondering
+- ▁phone
+- ▁所以我
+- 境
+- ▁crossed
+- down
+- 疑
+- ▁radio
+- 母亲
+- 印象
+- ▁Saint
+- ▁те
+- mobil
+- ▁wisdom
+- ▁để
+- 战争
+- nna
+- ▁anger
+- ップ
+- ▁flower
+- ▁Familie
+- kli
+- ▁zei
+- مل
+- ▁Не
+- rze
+- ▁screen
+- aniran
+- ington
+- ▁کو
+- ▁frame
+- 食べ
+- ktor
+- ange
+- kü
+- '""'
+- 露
+- ▁đi
+- ▁occupied
+- ả
+- ▁それでは
+- nit
+- 翻
+- ▁despair
+- ▁washington
+- five
+- 人に
+- 務
+- 和你
+- ▁kunnen
+- ров
+- ▁demanded
+- 里的
+- ose
+- 的名字
+- uti
+- 天气
+- col
+- シャ
+- ▁Liebe
+- бай
+- ▁dawn
+- 烟
+- colo
+- 做什么
+- ▁schi
+- 最初
+- ▁statement
+- دار
+- nam
+- نى
+- させて
+- 商品
+- 까
+- band
+- 杨
+- position
+- tage
+- 土地
+- ▁gerne
+- ▁ghost
+- 谢
+- ▁пол
+- ▁fundamental
+- ▁managed
+- 池
+- 治疗
+- duc
+- ▁Ihren
+- 人员
+- ▁enemies
+- eurs
+- ▁School
+- ▁kur
+- ▁rank
+- ель
+- nah
+- なんですが
+- fern
+- ▁yer
+- 衣服
+- 凯
+- ▁communication
+- ▁agreement
+- ▁marked
+- treten
+- ெ
+- tti
+- ▁Son
+- бра
+- 共同
+- 赶
+- みたい
+- ▁一方
+- ▁increasing
+- य
+- 니까
+- ▁Let
+- ▁removed
+- وب
+- 浮
+- 発表
+- 有没有
+- をして
+- ▁dying
+- ▁slave
+- 关键
+- ▁remarkable
+- 进去
+- ▁Krieg
+- 権
+- dzi
+- tó
+- iş
+- ▁Het
+- ▁теперь
+- ▁supper
+- ▁ari
+- ஜ
+- ▁cow
+- न्
+- 他们在
+- 年の
+- ▁improvement
+- ▁mistress
+- 计
+- 舞台
+- 团队
+- ු
+- 面前
+- ered
+- ▁equally
+- ▁суд
+- ▁jak
+- iem
+- ▁violence
+- بی
+- ▁strategic
+- ▁burning
+- öl
+- 没有人
+- 今の
+- くらい
+- ▁шу
+- liegen
+- ▁très
+- ▁schien
+- ▁However
+- ▁mü
+- などを
+- ▁poet
+- ▁الك
+- ▁bishop
+- ▁clo
+- ▁deck
+- お願いします
+- ▁baron
+- ▁Mor
+- stig
+- 有多
+- ▁farther
+- 皇
+- 课
+- 恋
+- кая
+- ень
+- ▁primer
+- 的声音
+- 人民
+- ▁말
+- ▁regarded
+- ▁Spe
+- 你好
+- 值
+- ▁groups
+- ▁asleep
+- 尤其是
+- ▁Э
+- 組
+- пом
+- ▁kom
+- 戴
+- ▁effective
+- ት
+- stin
+- sky
+- ▁mile
+- ▁verschiedene
+- ▁Alle
+- யில்
+- 机构
+- 如果我
+- ında
+- の方
+- 聞いて
+- ▁educa
+- jas
+- ▁code
+- ▁kas
+- ▁Cap
+- stellt
+- ▁Ste
+- chter
+- テレビ
+- ▁generation
+- 坐在
+- 秘密
+- од
+- もし
+- ▁wedi
+- роз
+- พ
+- vita
+- ▁sprach
+- ▁dengan
+- ▁cab
+- ▁describe
+- ▁route
+- ಿ
+- ▁weißt
+- ▁nahm
+- ед
+- いない
+- ▁좀
+- grad
+- ▁esto
+- 원
+- ▁calls
+- 务
+- ▁deeply
+- că
+- انی
+- ▁continues
+- னை
+- オリンピック
+- 了很多
+- ▁meat
+- お前
+- 的大
+- bul
+- thy
+- 先ほど
+- ngu
+- ▁gew
+- uba
+- ▁pack
+- сты
+- wyd
+- 丸
+- ▁arch
+- ▁播放
+- 怀疑
+- 了你
+- weise
+- ага
+- آ
+- ▁topic
+- oz
+- ган
+- ▁herum
+- 看见
+- 和我
+- ▁Amerika
+- 移
+- ▁behold
+- schau
+- 这个人
+- ▁inner
+- 营
+- ▁import
+- angle
+- rice
+- ▁capable
+- гла
+- 女儿
+- ▁nervous
+- ▁Kra
+- 介绍
+- ▁flying
+- 熊
+- ▁hacer
+- ▁chain
+- 唐
+- ▁minds
+- がない
+- whi
+- 妹
+- 混
+- tische
+- 姑娘
+- mah
+- ▁acht
+- ▁будзе
+- 地球
+- ▁أو
+- ▁innocent
+- gui
+- run
+- ▁Men
+- vě
+- ▁software
+- ffer
+- 背景
+- ific
+- らない
+- bet
+- களை
+- تى
+- 宁
+- ▁begun
+- uzi
+- ▁levels
+- vă
+- 年轻
+- ▁yield
+- lap
+- よりも
+- ▁Informationen
+- ▁بی
+- ▁spirits
+- ▁alarm
+- ▁проезд
+- ▁machte
+- ▁explanation
+- 对我
+- 不仅
+- ▁jam
+- zone
+- ▁younger
+- র্
+- 优
+- لت
+- гал
+- toj
+- ▁hide
+- ▁buried
+- マン
+- 攻撃
+- ▁worthy
+- 希
+- 吹
+- ▁Commission
+- ▁konnten
+- ा
+- ▁gods
+- tors
+- lama
+- ▁aspect
+- ▁eene
+- このあと
+- dum
+- 那样
+- ▁pred
+- ison
+- org
+- 并且
+- ▁rear
+- ரா
+- 具体
+- ▁wave
+- 加入
+- клад
+- ▁complex
+- 你现在
+- ▁Arm
+- を見て
+- loc
+- ▁southern
+- ▁mayor
+- ງ
+- gs
+- тель
+- ▁doors
+- лә
+- ▁citizens
+- 麻
+- 眼睛
+- ▁forgive
+- نت
+- ▁evident
+- 事儿
+- 人で
+- 帮我
+- வை
+- ▁commission
+- ▁site
+- 自身
+- ▁included
+- igt
+- 宮
+- 莉
+- 不太
+- 莱
+- ▁unique
+- setzt
+- ▁अ
+- 第三
+- ▁sons
+- 展
+- ▁dull
+- pass
+- ▁我说
+- ▁bottle
+- 容疑者
+- ▁dari
+- ▁الع
+- wydd
+- ▁verschiedenen
+- ▁bull
+- 一只
+- ▁stehen
+- 関
+- っていた
+- ▁کا
+- ▁чего
+- pers
+- ν
+- ступ
+- anda
+- 高い
+- তো
+- worth
+- 你还
+- ▁dim
+- ▁tower
+- ▁millions
+- ▁satisfaction
+- ▁effects
+- ▁rates
+- ▁gener
+- 表明
+- ▁虽然
+- 灵
+- 전
+- kuwa
+- 叫做
+- 下面
+- ▁cheer
+- من
+- کر
+- 领导
+- 变化
+- عد
+- 正常
+- 第一个
+- ▁clever
+- ▁treated
+- ▁divers
+- water
+- lös
+- ő
+- ▁rocks
+- 路上
+- ▁necessity
+- もちろん
+- 的一部分
+- licht
+- ingly
+- ۇ
+- 我们会
+- yard
+- 了他
+- аг
+- ▁ново
+- oka
+- ▁zero
+- ▁fund
+- ▁Natur
+- 接種
+- ridge
+- ▁pipe
+- 老板
+- قد
+- ▁figures
+- 丽
+- vant
+- 在哪里
+- maze
+- だけで
+- 替
+- lah
+- ▁edward
+- ▁shade
+- 人間
+- 乡
+- 在他
+- nim
+- ĉ
+- 备
+- それが
+- 新しい
+- ▁friendship
+- oi
+- ホ
+- 你这
+- nud
+- ▁liber
+- ▁sheep
+- ▁вер
+- ▁pictures
+- 义
+- ▁election
+- bung
+- gira
+- ▁Tur
+- эм
+- alo
+- ▁attitude
+- ▁них
+- jes
+- 橋
+- ▁conscious
+- že
+- ▁proportion
+- ▁ruin
+- cil
+- ▁себе
+- cyo
+- 其他人
+- ▁document
+- ▁western
+- ▁oo
+- ез
+- ▁economy
+- ▁twee
+- cient
+- ư
+- ▁palm
+- gua
+- elli
+- ств
+- と思って
+- raz
+- ▁execution
+- 視
+- 麦
+- ными
+- ▁entrance
+- 食物
+- кт
+- ▁manifest
+- ци
+- ▁perspective
+- ▁nations
+- ▁ху
+- lea
+- anti
+- ▁Bre
+- ▁plate
+- ▁desired
+- ço
+- ペ
+- ▁alten
+- ▁الب
+- ▁expense
+- 女子
+- そこ
+- vou
+- 目的
+- 坚持
+- ▁tiny
+- voir
+- ▁allein
+- ▁confi
+- шел
+- 导
+- ります
+- tica
+- ▁gradually
+- ▁Chi
+- cial
+- ▁pli
+- 壁
+- ▁mem
+- tab
+- 会议
+- wis
+- ış
+- 私が
+- ▁smith
+- 一人
+- ▁obtained
+- في
+- ▁relation
+- ▁grup
+- ▁gaze
+- رب
+- rous
+- していました
+- か月
+- 彩
+- rk
+- 许
+- hä
+- ▁tour
+- ▁giant
+- ▁perform
+- いや
+- ▁lad
+- ▁triumph
+- ▁finish
+- oku
+- ニュース
+- 艺术
+- ▁бер
+- 席
+- ▁всех
+- やすい
+- hung
+- ▁mí
+- jí
+- ▁sail
+- ▁require
+- ▁core
+- ▁einzige
+- 調査
+- рам
+- 马上
+- ▁сказать
+- 影響
+- ▁wounded
+- griff
+- 大哥
+- ▁Tür
+- 完了
+- ▁Ziel
+- 一场
+- 一旦
+- gent
+- ▁untuk
+- ▁criminal
+- nny
+- 企業
+- ▁revolution
+- 还要
+- ▁mystery
+- ▁mercy
+- aban
+- ▁constantly
+- cli
+- ▁teaching
+- taj
+- shobora
+- rod
+- cast
+- roll
+- ▁teacher
+- итель
+- ologie
+- bly
+- ▁update
+- 僕
+- ▁acquaintance
+- 休息
+- 今月
+- ▁article
+- 挑战
+- ▁эта
+- ▁bisschen
+- 某
+- ▁あの
+- fri
+- 照顾
+- 酸
+- ▁nói
+- وق
+- ▁sul
+- ▁visible
+- ▁instantly
+- gegangen
+- ▁dei
+- ▁それ
+- ▁значит
+- ▁versucht
+- 対策
+- ▁работа
+- өө
+- ▁через
+- ▁Ed
+- 测试
+- 护
+- year
+- というのが
+- ▁counsel
+- сла
+- ҫ
+- ですけど
+- 接下来
+- ▁你看
+- 難しい
+- ▁tres
+- 实现
+- ▁триста
+- last
+- 按
+- 觉
+- 宗教
+- ▁Plan
+- کی
+- 这么多
+- 确定
+- 悪
+- ▁letzte
+- ец
+- ▁violent
+- ▁knees
+- ▁knight
+- 在我们
+- 如果我们
+- isha
+- aria
+- ını
+- ▁пу
+- '21'
+- gru
+- ▁nel
+- ▁crack
+- ്
+- 因为他
+- ttle
+- ▁splendid
+- ▁richard
+- ▁Mont
+- ▁brilliant
+- ▁assured
+- script
+- 对你
+- rel
+- ▁mill
+- sca
+- imu
+- 減
+- ▁jede
+- ▁какие
+- 责任
+- apo
+- ▁être
+- ▁az
+- 补
+- 就是说
+- ▁earnings
+- 新たな
+- ▁fragte
+- ვ
+- ▁necessarily
+- tiv
+- ▁wit
+- ▁critical
+- ▁harry
+- 一张
+- gol
+- 赶紧
+- 潜
+- ではない
+- cou
+- 一緒に
+- ▁confident
+- ▁lag
+- 本来
+- ew
+- In
+- лася
+- 言って
+- 建筑
+- ާ
+- 上がって
+- ▁multi
+- ically
+- ▁turns
+- 惊
+- ию
+- ▁website
+- よかった
+- ▁worship
+- ▁unable
+- ▁throat
+- ӱ
+- ▁grief
+- lement
+- ここから
+- مي
+- ▁relations
+- iɣ
+- ▁كل
+- ière
+- 额
+- ▁published
+- 纸
+- dina
+- ▁vis
+- umi
+- ▁suspicion
+- 陆
+- test
+- ▁دارد
+- ўся
+- ▁sẽ
+- ▁Det
+- 也会
+- dad
+- eten
+- ▁petit
+- در
+- জ
+- 我们要
+- ▁darum
+- ▁maintain
+- ▁director
+- ▁الن
+- غا
+- 这个问题
+- 馬
+- 我很
+- aja
+- ▁Paul
+- 库
+- fy
+- ▁official
+- rí
+- ▁delivered
+- ▁cart
+- ▁goed
+- 上面
+- ▁schön
+- ▁prisoner
+- enga
+- 让他们
+- ▁universe
+- 挑
+- ▁rooms
+- оп
+- ▁bija
+- аць
+- 站在
+- pper
+- ایی
+- ▁оста
+- ▁regret
+- ▁differ
+- ▁나는
+- даг
+- ▁Col
+- んですけれども
+- ene
+- ந்து
+- 这儿
+- トップ
+- ▁mat
+- 言う
+- ▁puis
+- tail
+- だろう
+- ▁実は
+- 来到
+- 起こ
+- ▁tal
+- ▁objection
+- ▁Not
+- üm
+- ▁reden
+- 我自己
+- 经历
+- ▁midst
+- ▁admitted
+- ▁እ
+- 耳
+- ▁نمی
+- にかけて
+- ится
+- ▁threat
+- ás
+- ▁stronger
+- ingen
+- 考えて
+- ხ
+- كم
+- Т
+- ▁county
+- zem
+- ▁halten
+- 遇到
+- eff
+- 聊
+- ▁worry
+- 错误
+- fla
+- 观察
+- fte
+- ▁proposed
+- 免
+- ▁hence
+- ▁attend
+- لة
+- ume
+- ▁cada
+- ▁nearer
+- ▁Mädchen
+- 联
+- 大事
+- 甲
+- lassen
+- өн
+- ▁такое
+- ▁roz
+- の中
+- ▁guns
+- ик
+- ট
+- cade
+- 牌
+- 岡
+- ▁description
+- ▁Ка
+- ▁lots
+- ないと
+- ▁web
+- 男孩
+- ▁잘
+- 最大的
+- اند
+- 争
+- ▁adalah
+- рад
+- ▁concerning
+- ▁singing
+- ▁blame
+- lè
+- ▁affected
+- ▁folks
+- ▁verwendet
+- kira
+- ા
+- fas
+- ▁shock
+- ▁さらに
+- 容易
+- ick
+- 日本人
+- ously
+- 你自己
+- ▁possessed
+- mittel
+- ▁cabin
+- бі
+- ▁liberal
+- ▁我认为
+- ▁stones
+- 根本
+- 缺
+- ▁engine
+- graf
+- vat
+- ▁vice
+- ▁plants
+- 变成
+- dung
+- 录
+- lerin
+- ▁Bro
+- 唔
+- ▁mc
+- dü
+- ▁бул
+- ▁changing
+- ▁scientific
+- ท
+- ▁immense
+- ▁val
+- zug
+- ▁sport
+- ▁largest
+- über
+- lk
+- ▁magic
+- ▁гар
+- meter
+- 届
+- gger
+- ე
+- 笔
+- ▁私は
+- ▁absence
+- 京
+- ▁absolute
+- ▁дела
+- லி
+- cons
+- 媒体
+- ▁пи
+- 这件事
+- 原来
+- ▁wy
+- 丹
+- قل
+- aries
+- ▁Kan
+- ▁conflict
+- تم
+- 对吗
+- 府
+- ▁над
+- ischer
+- 園
+- ▁brothers
+- しく
+- oth
+- னி
+- ▁proceeded
+- ▁sem
+- nak
+- ▁intent
+- ▁کند
+- ▁disappeared
+- 只要
+- ово
+- bag
+- chte
+- ▁govern
+- ugh
+- きのう
+- 秀
+- yar
+- ▁mortal
+- 停止
+- ▁voices
+- 之间的
+- ▁nächste
+- ディ
+- ▁Au
+- ▁rue
+- گی
+- ප
+- zz
+- ▁bleiben
+- ▁لم
+- ▁favorite
+- pak
+- zähl
+- ▁hoped
+- ▁Nein
+- ▁mae
+- なり
+- 声音
+- ▁seated
+- дет
+- ▁bold
+- meye
+- ▁affect
+- ▁tempo
+- ે
+- ▁soil
+- 乔
+- 宇宙
+- ह
+- 镇
+- 骗
+- یک
+- ▁мар
+- 暴
+- aan
+- ▁waters
+- 中で
+- ั
+- ucht
+- мет
+- ▁awake
+- elo
+- ▁expectations
+- ▁sieben
+- знач
+- মা
+- としては
+- 撃
+- 工具
+- usa
+- 记录
+- 일
+- ▁arrival
+- atu
+- 在这
+- ▁Que
+- ▁hole
+- ffe
+- 港
+- ▁coffee
+- 選挙
+- gleich
+- ▁studies
+- 幅
+- ▁gente
+- 嘴
+- komp
+- ▁wusste
+- tric
+- ▁library
+- 增加
+- nzi
+- ▁delicate
+- ▁domestic
+- mina
+- ▁scheme
+- ▁перед
+- rö
+- bild
+- ▁profession
+- ▁experienced
+- ▁jedes
+- ▁interests
+- 后面
+- ▁Kom
+- 经过
+- ▁elizabeth
+- enda
+- プロ
+- ▁vessel
+- hold
+- ▁eating
+- ▁quando
+- czy
+- abil
+- merk
+- ▁remind
+- 盛
+- ▁channel
+- ▁Nacht
+- ▁না
+- 有个
+- 也没
+- 意思
+- ▁link
+- ▁تر
+- ▁numerous
+- zt
+- のある
+- 今後
+- bun
+- نگ
+- ▁recognized
+- ▁scheint
+- ▁الج
+- 这就是
+- рын
+- ▁custom
+- ▁mundo
+- ▁occur
+- nau
+- 的原因
+- 所以我
+- lect
+- сці
+- そうな
+- ▁vez
+- 行业
+- ▁bath
+- ▁Gehirn
+- ax
+- 由于
+- 描述
+- ▁pol
+- 陪
+- ▁silk
+- ort
+- ▁crew
+- ▁Wi
+- bt
+- ▁frightened
+- ▁всего
+- 现实
+- dies
+- 娜
+- ザ
+- ▁seines
+- たくさん
+- ▁えっ
+- ▁hor
+- 我不知道
+- ▁Bereich
+- ▁ceased
+- 湖
+- 爹
+- ▁israel
+- ▁wondered
+- ▁objects
+- 是我们
+- ▁prime
+- 丝
+- ▁apartment
+- ▁steel
+- ▁ещё
+- ただ
+- 我相信
+- ▁resolved
+- ▁Nor
+- ▁Bur
+- 固
+- Al
+- 洞
+- ▁更に
+- 中央
+- 舞
+- ▁china
+- ▁cert
+- ▁Mein
+- 确保
+- mini
+- 怀
+- ▁stepped
+- 依
+- ▁shelter
+- ▁fourteen
+- ching
+- ▁bou
+- ▁failure
+- ▁investments
+- амі
+- кон
+- ▁christmas
+- ▁signs
+- 看到了
+- ▁khi
+- ref
+- ছে
+- ▁groß
+- மை
+- ▁committee
+- autre
+- ▁anywhere
+- ▁organ
+- ▁게
+- ▁apply
+- ບ
+- いません
+- ▁chosen
+- ook
+- тә
+- ▁buck
+- んだけど
+- 음
+- ▁Wal
+- ular
+- erung
+- 優勝
+- ▁mysterious
+- ▁پر
+- 決め
+- ĝ
+- ▁friendly
+- 上がり
+- 打ち
+- rez
+- ▁Abend
+- ▁tous
+- xe
+- 万人
+- ▁Энэ
+- ▁кар
+- ぬ
+- سل
+- prav
+- scher
+- adi
+- ▁ён
+- lyn
+- スター
+- jun
+- ை
+- 一部
+- බ
+- thing
+- ▁principles
+- ▁너
+- eka
+- ອ
+- ▁voyage
+- ラン
+- nut
+- 最好
+- ▁rent
+- 方が
+- ▁Tod
+- ▁Staaten
+- ▁دە
+- 奖
+- ▁majesty
+- wij
+- 易
+- ▁tiene
+- ▁закон
+- 麻烦
+- ▁Rat
+- 过了
+- 達
+- ▁حال
+- 传统
+- ۈ
+- iyi
+- 週間
+- つく
+- radi
+- ▁Haupt
+- hil
+- ▁bet
+- 对不起
+- ▁prices
+- ▁manchmal
+- ▁manera
+- 习惯
+- ▁construction
+- gat
+- X
+- 的所有
+- ▁error
+- 抱
+- ▁està
+- なくて
+- ▁heißt
+- weit
+- 我现在
+- ▁mission
+- nica
+- àn
+- cor
+- zna
+- scha
+- ときに
+- hui
+- ago
+- ▁ee
+- ▁herr
+- 模式
+- 人気
+- 高兴
+- 准
+- ▁为什么
+- どこ
+- стан
+- 学院
+- 我看
+- 結
+- gne
+- fahren
+- ▁foolish
+- 枚
+- ə
+- ที่
+- ifi
+- já
+- homme
+- 检查
+- 银行
+- 允许
+- дзі
+- ▁pause
+- ▁rief
+- ො
+- 醒
+- uto
+- というか
+- টি
+- list
+- 午前
+- ▁problema
+- char
+- 勢
+- ▁Licht
+- nov
+- 乗
+- ▁Pla
+- ▁char
+- ▁internet
+- ▁coach
+- кан
+- 还没
+- enza
+- ▁gently
+- ▁beach
+- fried
+- 标准
+- ñ
+- ▁currently
+- 不行
+- ▁encore
+- 圈
+- ለ
+- iß
+- fect
+- 年轻人
+- ▁College
+- ọ
+- ▁discovery
+- ▁practical
+- ▁ĝi
+- 电视
+- poli
+- ▁oben
+- 命令
+- дзя
+- 艾
+- ▁stands
+- 虽然
+- ▁besteht
+- ▁Möglichkeit
+- ▁git
+- ▁paused
+- ▁sooner
+- ksi
+- ▁Mais
+- unk
+- ▁사
+- த்தில்
+- iden
+- cc
+- ▁这就是
+- زن
+- 名字
+- кра
+- ▁trois
+- 您的
+- door
+- 刻
+- ▁mille
+- ген
+- スタート
+- ▁cui
+- ▁Will
+- ▁sacred
+- やり
+- च
+- зу
+- system
+- ▁mud
+- inter
+- ▁одна
+- ▁quand
+- dienst
+- ▁hon
+- แ
+- 这位
+- ▁blessed
+- ▁pushed
+- ▁reco
+- 瓦
+- ▁ariko
+- 大多数
+- ▁наш
+- ▁Wahl
+- ▁reign
+- 장
+- air
+- kum
+- 来看
+- دة
+- ▁delay
+- стро
+- жу
+- ▁mont
+- ▁fought
+- seite
+- ог
+- ▁reported
+- chester
+- 他就
+- gged
+- ▁bare
+- ▁sigh
+- 薄
+- ▁thunder
+- شی
+- 年間
+- いつも
+- アン
+- ▁Yes
+- ▁Präsident
+- ▁northern
+- ▁Sir
+- ले
+- 当地
+- dik
+- sy
+- tia
+- liv
+- عمل
+- essa
+- 軽
+- 议
+- ▁modest
+- 蛋
+- ngi
+- gesetzt
+- oto
+- ▁empire
+- geführt
+- 人たち
+- nej
+- ▁Ihrem
+- ъ
+- ▁waste
+- ыг
+- 被害
+- ▁İ
+- bile
+- ▁delle
+- 弗
+- でき
+- 当我
+- ▁quan
+- ▁collection
+- ▁கா
+- ▁prima
+- 杯
+- ▁photograph
+- ▁veel
+- 压力
+- tif
+- 典
+- ▁wings
+- мә
+- ச்
+- rij
+- ▁dreams
+- 司
+- ▁dur
+- 予想
+- ▁gained
+- 怎么了
+- при
+- ▁내
+- ▁سر
+- ▁horrible
+- 人物
+- пар
+- 我能
+- ▁fled
+- 超过
+- 干什么
+- cí
+- rais
+- фа
+- 調
+- vision
+- ăm
+- ars
+- ▁easier
+- ▁могу
+- ▁Probleme
+- 있
+- 顺
+- facebook
+- ▁sofort
+- たま
+- not
+- zog
+- 戏
+- isme
+- ▁آنها
+- ే
+- 显示
+- ▁funktioniert
+- wali
+- なんと
+- 本身
+- 姿
+- 永
+- ▁stairs
+- ▁sale
+- ulo
+- ép
+- roj
+- ▁Port
+- ▁proceed
+- 甘
+- ▁copy
+- ▁burn
+- ▁bearing
+- 確認
+- ▁consequence
+- ▁completed
+- 不错
+- 在一个
+- ▁се
+- fred
+- 我喜欢
+- ▁employed
+- 瑞
+- ▁Oh
+- ▁banks
+- ห
+- 棒
+- ▁enormous
+- 烧
+- ▁sympathy
+- ▁Lage
+- vet
+- ▁Some
+- ▁concern
+- cré
+- ங்கள்
+- α
+- faced
+- richtet
+- ▁factors
+- اح
+- 哭
+- ▁nada
+- ▁Tor
+- ▁philosophy
+- ▁clouds
+- レー
+- ▁gesprochen
+- ▁tight
+- asta
+- born
+- 既
+- head
+- indi
+- ▁By
+- ▁instinct
+- ▁algo
+- 幕
+- ゲーム
+- 困难
+- rr
+- 我这
+- ҡа
+- ▁elements
+- の中に
+- lık
+- ▁congress
+- ▁你要
+- ▁album
+- ▁London
+- ▁relative
+- typ
+- cted
+- 专业
+- ▁connect
+- ▁restaurant
+- lier
+- ▁climate
+- ▁goal
+- ▁Gespräch
+- 则
+- ▁mood
+- ▁classes
+- ▁introduced
+- ▁gì
+- üt
+- ο
+- ▁schwierig
+- ▁anche
+- лег
+- ▁maid
+- 持续
+- ава
+- ▁König
+- 振
+- tique
+- dda
+- ▁odd
+- cord
+- tit
+- 试
+- dit
+- ▁segment
+- ņ
+- 転
+- ▁conscience
+- ▁retreat
+- ▁По
+- ▁jar
+- 難
+- ▁earn
+- ▁Por
+- sión
+- 我真的
+- 盗
+- それから
+- ▁التي
+- kro
+- عة
+- '!?'
+- ▁version
+- 施設
+- ▁Fin
+- 那就是
+- 徒
+- 对我来说
+- ▁cave
+- ▁medicine
+- ▁application
+- ции
+- 约翰
+- ▁Bild
+- ▁informed
+- zio
+- rant
+- ▁gör
+- 说你
+- ▁谢谢
+- 大量
+- ajn
+- 腿
+- جا
+- 傷
+- аж
+- ▁previously
+- pho
+- ▁immediate
+- 痛苦
+- 敬
+- 他们会
+- ▁witness
+- ▁interpret
+- 避難
+- gebracht
+- じゃあ
+- ▁benefits
+- ▁morgen
+- ▁reference
+- ▁feed
+- ▁chu
+- 终于
+- mmel
+- 年に
+- 하고
+- ▁එ
+- ▁anxiety
+- ▁severe
+- ▁via
+- اش
+- 移动
+- れて
+- пы
+- ẓ
+- ▁discuss
+- 到达
+- のように
+- ▁但是我
+- ▁returning
+- オー
+- нес
+- 捕
+- ▁recognize
+- ▁Vielleicht
+- ▁nord
+- 取材
+- 怎么办
+- 打电话
+- ▁falsch
+- ช
+- ▁Mark
+- gomba
+- ▁cheap
+- ишь
+- 突
+- ▁recon
+- 動き
+- 好吗
+- 炎
+- ▁kit
+- ▁Of
+- ▁ease
+- ▁yards
+- рь
+- ▁engagement
+- だと思います
+- oro
+- 大丈夫
+- ubi
+- ▁games
+- ▁Musik
+- となりました
+- 状
+- 季
+- ものを
+- ▁slide
+- aza
+- segu
+- führen
+- 注
+- さんと
+- 他会
+- 想法
+- ▁hurried
+- termin
+- تي
+- ▁mostly
+- sun
+- める
+- ▁wheel
+- kem
+- 反对
+- ▁intend
+- いって
+- 知识
+- ▁amongst
+- ún
+- ▁countenance
+- fite
+- ワクチン
+- 全然
+- ▁variety
+- ▁thomas
+- 形式
+- ▁بە
+- ▁intelligent
+- force
+- 值得
+- acht
+- jem
+- 用户
+- ▁الش
+- нага
+- quin
+- rri
+- hle
+- ▁rush
+- тов
+- 说什么
+- ▁అ
+- овать
+- ▁Sin
+- ▁کنم
+- anto
+- 吓
+- いっぱい
+- ▁partly
+- ▁hinaus
+- ▁guilty
+- isse
+- ▁fox
+- stead
+- 確
+- ▁rope
+- 出場
+- ▁вос
+- 面对
+- ▁assistance
+- ▁gesch
+- ▁Fo
+- いつ
+- கொண்ட
+- tech
+- 据
+- ▁Miss
+- ▁tools
+- lau
+- 逃
+- 大臣
+- тив
+- лов
+- ▁которая
+- ▁Ber
+- なんで
+- 显然
+- ▁solemn
+- 楽しみ
+- 辛
+- ▁geworden
+- ald
+- ▁eager
+- ▁counter
+- 我们现在
+- mana
+- ▁consciousness
+- iendo
+- ▁为了
+- iba
+- ▁너무
+- 胸
+- ▁També
+- ▁scho
+- ▁vent
+- xon
+- ▁candle
+- ▁Ш
+- ▁lion
+- ▁combat
+- 关心
+- ije
+- ▁located
+- ▁شود
+- staat
+- ▁resolution
+- 検査
+- ▁august
+- ▁disse
+- 交通
+- گو
+- ▁depend
+- ▁Von
+- かい
+- ▁Gen
+- 不安
+- はこの
+- ▁你是
+- chu
+- ▁Programm
+- aku
+- luc
+- ▁joke
+- مه
+- lands
+- эг
+- ▁adult
+- ▁Да
+- が出
+- 成员
+- 姐姐
+- 照片
+- ▁Ap
+- ▁consent
+- rer
+- ▁tief
+- ▁rub
+- 察
+- ып
+- 全球
+- 族
+- 落ち
+- ▁ließ
+- ▁High
+- feld
+- tul
+- cl
+- 事实
+- 하
+- 今日は
+- 响
+- rés
+- ▁Ger
+- рт
+- ও
+- ▁Herz
+- 你别
+- ▁سال
+- 的朋友
+- ▁neces
+- ▁pitch
+- gno
+- ▁tai
+- kie
+- ▁notion
+- ▁yu
+- eks
+- kora
+- ▁victory
+- hur
+- ▁Bas
+- agi
+- のでしょうか
+- 映画
+- 评论
+- کە
+- ▁Pol
+- どんどん
+- ▁enable
+- ▁marketing
+- ▁número
+- 摩
+- ▁reduce
+- ▁cela
+- වි
+- ్
+- ▁cru
+- byo
+- ril
+- ▁push
+- ▁ເ
+- yen
+- ▁bre
+- ▁flash
+- やる
+- ▁خو
+- すぐ
+- ▁improved
+- 持ち
+- まる
+- ät
+- had
+- ря
+- 糖
+- かもしれません
+- 负
+- دم
+- anyi
+- baza
+- ▁überhaupt
+- 鸡
+- れた
+- ▁rushed
+- 回来了
+- оль
+- ▁Una
+- ▁obtain
+- 사
+- ▁refuse
+- 语言
+- fangen
+- ▁Eu
+- 選手が
+- kka
+- 魚
+- ▁wider
+- 骑
+- 善
+- ▁eternal
+- قر
+- ▁trick
+- ▁así
+- ongo
+- ▁worn
+- ▁stores
+- рос
+- 制造
+- lied
+- rica
+- raf
+- 十年
+- omo
+- 、2
+- ▁smell
+- rav
+- ▁pensa
+- ▁continent
+- ▁stupid
+- となった
+- 接触
+- ▁werd
+- ▁Para
+- 闹
+- ▁stir
+- ▁score
+- 구
+- とても
+- 者の
+- ▁こちら
+- 属于
+- çi
+- том
+- 説明
+- 受け
+- ▁gest
+- サン
+- ului
+- ▁slip
+- 없
+- 増え
+- ▁apple
+- のかな
+- 偏
+- 承认
+- ▁surrounded
+- ▁Zukunft
+- ▁valuable
+- ▁führen
+- 答案
+- ▁hätten
+- teilen
+- ▁einigen
+- 孙
+- ▁rw
+- handel
+- ▁approached
+- cis
+- ▁exchange
+- 、1
+- ▁internal
+- ▁мал
+- まり
+- ラー
+- ▁gegenüber
+- 埃
+- lain
+- ▁absurd
+- 全体
+- ция
+- ▁phil
+- 上海
+- ▁associated
+- ▁điều
+- ▁cam
+- ▁William
+- ▁dont
+- ▁পা
+- шла
+- 技
+- ここまで
+- ▁abantu
+- kna
+- 章
+- hem
+- 例如
+- ▁Dass
+- ▁softly
+- qa
+- ▁Pra
+- 不再
+- 抽
+- ▁họ
+- sehen
+- ▁sunt
+- rada
+- を見せ
+- ▁நா
+- lou
+- ▁kindly
+- رة
+- уб
+- 巨大的
+- 作り
+- bling
+- ▁fet
+- ▁moreover
+- quel
+- ▁uniform
+- laufen
+- ▁stable
+- ▁байсан
+- ▁dal
+- ▁clients
+- wol
+- だよ
+- ismo
+- コロナ
+- 材料
+- ▁bitte
+- ▁genius
+- mā
+- があって
+- fat
+- ▁escaped
+- guard
+- 男子
+- 在你
+- ▁beginnen
+- ▁substance
+- ▁bush
+- 小心
+- 呆
+- ▁suggestion
+- دان
+- 跟他
+- ▁limited
+- ーン
+- ▁offen
+- ▁unserem
+- 伤害
+- cel
+- ▁يمكن
+- mate
+- 独立
+- giye
+- ▁mur
+- ▁robot
+- etti
+- esc
+- grund
+- ▁Bra
+- 负责
+- ▁situa
+- 开心
+- ▁Don
+- 羽
+- ▁dessen
+- ▁plane
+- ことで
+- gü
+- ▁куда
+- ▁closer
+- ▁precious
+- 거든
+- овская
+- است
+- ▁convinced
+- ▁bomb
+- tà
+- จ
+- 種
+- ▁früher
+- май
+- ał
+- ▁confess
+- そこに
+- ▁жа
+- 因为我们
+- ▁phải
+- вой
+- кс
+- ▁curiosity
+- نو
+- 図
+- 別
+- ▁десять
+- tting
+- йте
+- ▁kiel
+- ▁Meine
+- できた
+- ▁sap
+- 大きく
+- ▁addressed
+- 康
+- ota
+- சா
+- نه
+- kali
+- тым
+- ▁shouted
+- ▁этим
+- гә
+- ▁сер
+- gere
+- 総
+- ▁우리
+- ▁concept
+- here
+- のために
+- ▁unten
+- ▁National
+- sek
+- ▁crea
+- ▁hare
+- ▁button
+- なのか
+- উ
+- これまで
+- 関東
+- quet
+- 当他
+- 弟
+- ▁نو
+- ▁நி
+- 博士
+- stal
+- 佐
+- amo
+- ▁gal
+- mé
+- ár
+- zing
+- ▁zwanzig
+- ▁nachdem
+- بو
+- していく
+- ▁pressed
+- ▁neza
+- 有关
+- ▁dignity
+- ▁meanwhile
+- ▁Millionen
+- ▁causes
+- ▁poison
+- を取り
+- pir
+- ▁arrive
+- ▁mia
+- ▁flood
+- ory
+- ▁thu
+- 放弃
+- 早く
+- ckt
+- 的那个
+- anar
+- 虚
+- ▁lights
+- 雲
+- wart
+- ▁같애
+- مو
+- ▁doen
+- ▁phrase
+- そうですね
+- 的一
+- ▁cum
+- யை
+- Co
+- 込んで
+- ▁tiu
+- idos
+- سب
+- ▁今日は
+- ▁musste
+- 似
+- қ
+- ▁flew
+- ▁Mittel
+- 给我们
+- ライ
+- ▁bigger
+- prov
+- ▁Qui
+- ▁hanging
+- schrift
+- してください
+- ▁За
+- 犯罪
+- ▁purchase
+- ▁quarrel
+- üs
+- ▁seg
+- ণ
+- ▁federal
+- ී
+- 说我
+- ▁también
+- ▁yari
+- apa
+- nja
+- ▁succeed
+- ▁hielt
+- ▁recht
+- Ma
+- 正如
+- 焼き
+- ▁besten
+- fam
+- 論
+- おります
+- ุ
+- くて
+- tation
+- 住宅
+- وه
+- ▁mod
+- ▁coal
+- ▁italian
+- ほ
+- пан
+- tine
+- ▁però
+- ▁render
+- ▁South
+- 欠
+- 广告
+- реш
+- をする
+- кий
+- schuldig
+- ▁programs
+- рен
+- ぼ
+- hé
+- ▁cancer
+- ▁verk
+- ük
+- ▁miserable
+- श
+- azione
+- ▁bunch
+- ık
+- ▁fierce
+- fie
+- ▁appropriate
+- ▁healthy
+- ▁subjects
+- ▁kre
+- qué
+- ▁día
+- ▁Ama
+- ych
+- ▁fatal
+- 馆
+- kiri
+- ▁لە
+- 菲
+- 每一个
+- ились
+- форм
+- hoo
+- вор
+- vingt
+- ▁assume
+- 消息
+- ▁swift
+- ▁سو
+- elt
+- ▁competitive
+- ▁hasta
+- ▁aquí
+- 小时
+- 既然
+- させ
+- gene
+- stoff
+- ▁opera
+- ▁М
+- shing
+- lies
+- 連続
+- voy
+- ▁mamma
+- めて
+- unda
+- ▁hint
+- ▁offering
+- 吴
+- ▁luck
+- ▁Medi
+- iu
+- 賞
+- lte
+- 一份
+- 国际
+- ▁crisis
+- ▁creatures
+- raum
+- لك
+- ▁Roedd
+- lut
+- ▁tremendous
+- ▁Esta
+- swa
+- রে
+- ために
+- ρ
+- ▁micro
+- 发布
+- kov
+- ▁monde
+- ▁shoot
+- ▁rage
+- ín
+- ன்ற
+- ident
+- ▁reflection
+- ▁stern
+- 部屋
+- 太多
+- ▁forgot
+- 殺
+- ▁arthur
+- 広
+- ▁zwar
+- lina
+- を使って
+- ▁rob
+- эй
+- ▁destroyed
+- idi
+- ello
+- ▁activities
+- cole
+- ▁nein
+- ▁dell
+- pada
+- 여
+- ▁Ser
+- ▁tells
+- ▁painful
+- ones
+- 子供
+- ▁他说
+- ▁politics
+- 还在
+- 观众
+- ▁Seine
+- ines
+- 记住
+- ▁branches
+- 僕は
+- 更加
+- wala
+- ▁Nord
+- satz
+- ▁appointed
+- ▁że
+- きて
+- რა
+- uring
+- 決
+- äng
+- ▁sens
+- tischen
+- 表达
+- pia
+- ▁acquisition
+- ▁تم
+- 疾病
+- ▁directed
+- ▁bwa
+- 持って
+- 某种
+- ▁chap
+- یر
+- хан
+- ▁keen
+- はない
+- ▁lover
+- ▁nou
+- 有一
+- ▁Sicherheit
+- ▁hum
+- ▁observe
+- 岩
+- ▁accompanied
+- ▁laughter
+- ▁ஒ
+- jor
+- دە
+- ▁Hel
+- rā
+- ▁whilst
+- 時代
+- ▁waves
+- 意味
+- ▁survey
+- ид
+- 作業
+- 伝え
+- ▁Sprache
+- schrei
+- ▁distinct
+- 沈
+- един
+- يم
+- வர்
+- dzie
+- 赵
+- ▁spin
+- amba
+- 雅
+- ▁دست
+- 很多人
+- 北海道
+- ありがとうございました
+- schließen
+- ▁porta
+- லா
+- vic
+- ▁tren
+- hum
+- ▁Eltern
+- பெ
+- 澳大利亚
+- なんですね
+- ▁많이
+- ▁branch
+- 咋
+- 商业
+- ▁throne
+- ▁burden
+- ▁martin
+- ▁alors
+- fertig
+- ▁rela
+- 情绪
+- 抓住
+- 留下
+- 因为你
+- ▁totally
+- 并没有
+- ▁sector
+- ▁launch
+- ▁leaders
+- ▁shouldn
+- 众
+- ▁cái
+- 形成
+- 寺
+- ▁decline
+- ▁Thema
+- ▁fifth
+- 从未
+- ▁thoroughly
+- ▁Prozess
+- La
+- ệ
+- ▁Bla
+- 一块
+- box
+- will
+- なんですけど
+- 终
+- ▁sugar
+- ří
+- роб
+- ▁»
+- én
+- ▁votre
+- 撮影
+- ▁warning
+- ▁hundreds
+- ▁Это
+- ▁divided
+- からは
+- 重要的
+- 我希望
+- ▁string
+- 合い
+- دن
+- ▁fand
+- ▁Politik
+- dı
+- ▁stated
+- 充
+- 映像
+- 目标
+- سي
+- ▁household
+- ▁suis
+- ण
+- ▁hut
+- ι
+- 几天
+- ▁dollar
+- 鸟
+- 病院
+- 在这儿
+- ís
+- rid
+- riye
+- 我对
+- プレー
+- ▁punt
+- 戸
+- ▁City
+- ologi
+- صد
+- ▁tear
+- رد
+- emos
+- ணி
+- turn
+- ▁pea
+- 设
+- ье
+- 齐
+- からね
+- 表现
+- 早上
+- уул
+- 盖
+- ▁degrees
+- 四个
+- ▁hurry
+- 爆
+- ▁প
+- ▁plej
+- ▁forever
+- ė
+- пад
+- ders
+- ▁دی
+- ▁remote
+- ▁mount
+- ▁privilege
+- uj
+- bata
+- のこと
+- ぜひ
+- cket
+- ▁hari
+- ▁critic
+- ▁Fer
+- ▁hopes
+- மான
+- ▁applied
+- かけ
+- تون
+- ▁kwam
+- ▁cyangwa
+- ▁Macht
+- ▁lift
+- 流れ
+- gram
+- 平台
+- 写真
+- 拒绝
+- гор
+- ು
+- ▁Gemeinschaft
+- ников
+- ▁regarding
+- ▁reduced
+- 生产
+- 頂
+- ▁joseph
+- ▁Boden
+- ▁shining
+- pela
+- ວ
+- dacht
+- 植物
+- ▁heel
+- 학
+- これが
+- ったら
+- 是有
+- ▁Während
+- ▁weary
+- 大概
+- 坂
+- ▁völlig
+- ▁años
+- chy
+- ▁день
+- ▁wound
+- えて
+- ▁combination
+- oy
+- autres
+- инская
+- ван
+- asse
+- кол
+- ▁States
+- ▁эту
+- دو
+- ▁meu
+- ▁sollen
+- 出て
+- 人口
+- ል
+- 我们需要
+- ▁humble
+- به
+- ▁bang
+- most
+- 自我
+- inda
+- ▁rolled
+- 报道
+- 睡觉
+- 暖
+- ے
+- ▁crazy
+- ▁seriously
+- 较
+- 니
+- 我只是
+- 虫
+- ▁remaining
+- rra
+- ▁Ten
+- unter
+- pr
+- 入り
+- ngo
+- ▁punishment
+- ▁episode
+- ▁zeide
+- kop
+- ▁sank
+- いきます
+- ▁file
+- ▁Durch
+- ▁Selbst
+- ▁Projekt
+- ▁ring
+- ▁prisoners
+- ▁pes
+- ▁Fu
+- 絶
+- ால்
+- 言い
+- ▁народ
+- cca
+- ▁dad
+- ▁hero
+- ▁vos
+- ▁Mus
+- ▁helen
+- ▁생각
+- ▁Schritt
+- ていました
+- shire
+- adas
+- ▁pricing
+- ▁Entwicklung
+- 子ども
+- 的情况
+- dim
+- 码
+- gur
+- gging
+- vul
+- ▁pays
+- ліся
+- 夢
+- ▁holiday
+- hus
+- ▁remarks
+- ▁Rolle
+- ған
+- ほとんど
+- ිය
+- '23'
+- 专
+- ▁strongly
+- ▁solutions
+- tang
+- ▁اما
+- ake
+- tara
+- 比赛
+- lad
+- 律师
+- ▁uw
+- ап
+- 団
+- bur
+- ▁types
+- 欧
+- ▁assist
+- 海外
+- ▁california
+- ▁pig
+- ▁trap
+- ▁gall
+- ös
+- dica
+- gled
+- ▁architect
+- ▁Ph
+- ▁Blick
+- 自宅
+- ▁onze
+- ▁我也
+- ▁Geist
+- 把你
+- ▁wicked
+- ▁wolf
+- 結婚
+- 记得
+- ▁ああ
+- ▁breaking
+- şe
+- 尝试
+- ound
+- ▁views
+- 每个人
+- ское
+- তি
+- ▁pool
+- 過
+- ▁trat
+- ▁attended
+- عا
+- ▁profound
+- 大部分
+- 应该是
+- ▁humanity
+- în
+- mad
+- ここは
+- 調べ
+- ▁loving
+- 流行
+- ▁arose
+- 高校
+- 智
+- あっ
+- 不仅仅是
+- 国会
+- 最も
+- てきた
+- ▁sous
+- ▁Но
+- зд
+- ▁movie
+- camp
+- lieb
+- шу
+- ▁kindness
+- ς
+- ▁我知道
+- gut
+- ▁literature
+- 斯特
+- tam
+- ▁presentation
+- ことし
+- ▁designed
+- わけです
+- km
+- ▁expenses
+- oko
+- 我们都
+- ▁George
+- 優
+- 交流
+- 献
+- 失败
+- ひと
+- ▁vy
+- ▁border
+- ▁pilot
+- 集中
+- ▁tegen
+- ም
+- 基础
+- ▁inquired
+- つか
+- ▁assets
+- ▁ibyo
+- ▁slept
+- ▁mismo
+- ▁zien
+- 登場
+- iḍ
+- ▁interrupted
+- ▁constitution
+- ▁nodded
+- archi
+- ключ
+- 还不
+- ▁هي
+- ▁alas
+- ▁merit
+- 비
+- ▁начал
+- ▁буду
+- ▁Nur
+- ▁thrust
+- luci
+- ▁الق
+- ▁adventure
+- ▁puede
+- ▁debate
+- пуст
+- ▁paying
+- 诗
+- mata
+- four
+- hari
+- mě
+- ▁enterprise
+- については
+- ▁Ei
+- fit
+- প্র
+- সা
+- ▁atmosphere
+- ண்ட
+- ▁இது
+- 煮
+- ▁cottage
+- アル
+- ▁ஏ
+- ▁cattle
+- ில்
+- arte
+- ▁pole
+- luk
+- ▁secretary
+- লি
+- ▁leven
+- 因为他们
+- stä
+- 困
+- ▁Aquest
+- 豪
+- 功能
+- ▁peer
+- вол
+- 两个人
+- ມ
+- ▁earl
+- 偷
+- ▁möchten
+- ▁kuko
+- ▁گ
+- ost
+- тен
+- ▁прав
+- ▁manager
+- ▁пыта
+- ▁Robert
+- த்தை
+- ject
+- iten
+- ғы
+- 本当
+- 异
+- ▁faster
+- ▁interior
+- füg
+- вя
+- ▁Wert
+- ▁cel
+- frage
+- مة
+- baka
+- 目の
+- ▁Energie
+- ▁leap
+- muntu
+- 的新
+- ▁margaret
+- ▁weakness
+- ▁steam
+- ▁souls
+- 정
+- ▁danke
+- ▁Ма
+- ▁eventually
+- ché
+- 我们就
+- 哥哥
+- ▁약간
+- 安排
+- lant
+- 扔
+- ▁guests
+- ско
+- ▁click
+- 頃
+- ▁Gruppe
+- organiz
+- ნ
+- 水平
+- lop
+- ▁tanto
+- vý
+- பி
+- ▁imagin
+- ▁Peter
+- ▁rosa
+- ▁Tat
+- ▁bai
+- 成了
+- uur
+- gio
+- ▁Ac
+- haw
+- が多い
+- esse
+- そうだ
+- ▁کنند
+- avoir
+- ▁seus
+- 国際
+- ា
+- نىڭ
+- 売
+- ▁spanish
+- 宽
+- 聪明
+- 反应
+- ▁widow
+- 不敢
+- ▁unit
+- アメリカの
+- ▁suc
+- 質
+- ▁unhappy
+- ▁flame
+- 对吧
+- ▁tide
+- ▁dijo
+- ậ
+- ▁shift
+- 帕
+- 韓国
+- azi
+- ▁déc
+- 经验
+- 达到
+- ▁notes
+- 랑
+- ▁fare
+- 怒
+- ▁catholic
+- 技術
+- 资源
+- ▁Nous
+- bila
+- ▁stared
+- ▁tied
+- quest
+- 各位
+- ▁thirteen
+- vano
+- 对方
+- ağı
+- ▁abge
+- ssel
+- 因为它
+- ▁veil
+- ▁chat
+- ▁또
+- establish
+- 以外
+- ި
+- raj
+- மாக
+- ▁auszu
+- გ
+- lance
+- ▁anzu
+- غير
+- kara
+- 答え
+- 晓
+- மே
+- ▁essential
+- 当時
+- 번
+- nyuma
+- ▁Stimme
+- lage
+- ै
+- ört
+- を持って
+- ▁anderes
+- 亿
+- ▁Sol
+- игр
+- 兴
+- リン
+- த்தி
+- wachsen
+- ってこと
+- ▁plu
+- ▁partners
+- ▁eggs
+- ▁largely
+- ▁hadn
+- ▁enjoyed
+- ው
+- aid
+- 상
+- ▁despite
+- ▁Os
+- 房间
+- ▁Union
+- кр
+- 五十
+- ▁distribution
+- ▁pues
+- 人々
+- 客户
+- ▁intellectual
+- ▁cred
+- ▁buildings
+- 国王
+- ▁Show
+- ios
+- ▁baş
+- ▁wholly
+- ▁parent
+- ▁neat
+- лын
+- ▁ehhe
+- ▁braucht
+- ▁stud
+- 入れ
+- ▁verloren
+- ▁guest
+- wyr
+- ちゃう
+- enti
+- sak
+- ران
+- ▁kr
+- аа
+- 简
+- شا
+- gegeben
+- ▁mate
+- 亲爱的
+- 增长
+- 意义
+- あり
+- ▁đang
+- ▁delighted
+- 婚姻
+- trau
+- ▁handle
+- ▁مح
+- ゲ
+- 伟大的
+- مت
+- ▁steady
+- 健
+- ▁Cal
+- ▁Bank
+- ances
+- ▁Tôi
+- ▁physician
+- his
+- pren
+- ambi
+- ڭ
+- laden
+- ▁stayed
+- などで
+- ▁patience
+- ▁utterly
+- ▁오
+- ▁Dieser
+- 牙
+- ▁primarily
+- ▁Grunde
+- ▁esper
+- كون
+- י
+- ony
+- ▁context
+- kaj
+- தா
+- ▁sud
+- là
+- Le
+- ▁іх
+- ▁Mer
+- ▁annual
+- сил
+- 中の
+- ▁fro
+- цу
+- вали
+- ▁contrast
+- ▁Technologie
+- 爷
+- ▁lernen
+- 泰
+- burn
+- 有一种
+- দি
+- 禁
+- An
+- 和他
+- uld
+- ▁будут
+- 幸
+- 史
+- coming
+- ▁entwickelt
+- ache
+- 呼吸
+- hara
+- ▁trump
+- ▁hey
+- post
+- 子さん
+- ▁beste
+- asa
+- ▁Zo
+- füll
+- 这么说
+- 挂
+- ▁arranged
+- 悲
+- 一つ
+- ics
+- лек
+- ▁coup
+- 办法
+- மி
+- があり
+- 一边
+- ▁pink
+- ▁расс
+- 保证
+- ▁follows
+- ▁scar
+- ировать
+- 御
+- ▁retail
+- 另一
+- ダー
+- эс
+- 文件
+- ▁symbol
+- جر
+- 爸
+- 现场
+- 透
+- diri
+- след
+- teur
+- σ
+- というふうに
+- 给了
+- ▁contained
+- 放送
+- ddle
+- ▁budget
+- ▁""
+- өл
+- ▁kamen
+- ▁Met
+- wag
+- ▁کل
+- ▁biggest
+- ▁gross
+- ▁feature
+- Л
+- fun
+- 圆
+- ▁distinguished
+- ▁studied
+- kunda
+- ▁dispose
+- 人を
+- ▁Hy
+- 得到了
+- ▁Af
+- Ar
+- fus
+- バス
+- rai
+- ▁obwohl
+- innen
+- ▁shoes
+- eme
+- そうなんです
+- او
+- kra
+- ▁sensation
+- mission
+- ▁всегда
+- raga
+- ▁hoffe
+- 心理
+- ▁ŝi
+- 했
+- ▁properly
+- ▁pala
+- 那边
+- 大統領
+- ▁prophet
+- ▁parliament
+- нова
+- tera
+- fil
+- というところ
+- 就不
+- hou
+- ▁consum
+- ▁tomb
+- ު
+- uze
+- ていく
+- 羊
+- ▁lit
+- 适合
+- ▁Gesellschaft
+- としています
+- ▁созда
+- ▁desk
+- ▁হ
+- ▁አ
+- rek
+- ▁đến
+- 带来
+- твор
+- 题
+- год
+- ▁russian
+- に出
+- ▁Luft
+- '22'
+- dank
+- 再说
+- 斗
+- వ
+- ▁jedem
+- ▁பா
+- ▁knife
+- 反正
+- 開発
+- ▁crying
+- ▁moet
+- sser
+- ▁operation
+- ▁confusion
+- ರ
+- に入って
+- master
+- 缩
+- ▁этих
+- ▁continuing
+- ▁rival
+- ły
+- ▁articles
+- ▁кажется
+- ▁remark
+- ▁damage
+- ご覧
+- nto
+- ▁Google
+- ▁strip
+- 銀
+- お店
+- tse
+- ▁iets
+- ▁devoted
+- 絵
+- ▁jump
+- 园
+- 女士
+- 世界上
+- ▁log
+- 丈夫
+- ▁traffic
+- ▁所以我们
+- ünde
+- ▁deu
+- ▁abandon
+- ▁tale
+- 审
+- şa
+- ▁bah
+- ▁тот
+- bble
+- 寄
+- ▁tis
+- ▁musical
+- 开发
+- ▁deposit
+- 阅读
+- two
+- のような
+- كن
+- 项
+- 决
+- ▁num
+- tiva
+- 中に
+- ▁چه
+- ıyor
+- ▁persönlich
+- kta
+- ён
+- ▁Mc
+- that
+- ander
+- ▁perceived
+- 東京都
+- ▁Ter
+- ▁hunting
+- ▁companions
+- ▁emotion
+- cious
+- 부
+- 升
+- さまざまな
+- 记者
+- 绝
+- ್
+- ▁ends
+- ▁الإ
+- 细
+- ias
+- thro
+- ▁bor
+- fei
+- ▁Inter
+- 的感觉
+- 任务
+- ▁surrounding
+- ▁хочу
+- ▁tar
+- ▁mixed
+- ▁Schüler
+- ▁স
+- さま
+- ▁spending
+- ▁Parlament
+- see
+- пор
+- 一段
+- 我和
+- guru
+- ▁beau
+- еш
+- аш
+- ▁என்று
+- zor
+- ▁sisters
+- вать
+- ის
+- bei
+- ites
+- ▁yours
+- лось
+- ▁друг
+- ▁Just
+- ▁expedition
+- ▁utter
+- ▁செ
+- ▁transform
+- ▁hollow
+- lev
+- ▁lonely
+- ▁sonra
+- క
+- ▁zijne
+- үү
+- ▁erkennen
+- ▁té
+- ▁هر
+- ▁komen
+- ktion
+- сло
+- 利益
+- 真的是
+- ड
+- tzen
+- ▁mare
+- late
+- ▁shake
+- ▁servi
+- ▁homes
+- しながら
+- भ
+- 法院
+- ▁multiple
+- ▁Fri
+- あまり
+- ▁exception
+- ▁reader
+- 潮
+- ▁discussed
+- 什么样的
+- ▁trop
+- ▁две
+- nä
+- irwa
+- ▁uko
+- 塁
+- hing
+- sack
+- いただきます
+- 肯
+- 盛り
+- ological
+- ▁Facebook
+- ▁poco
+- cles
+- schauen
+- 出し
+- تن
+- 她说
+- чит
+- ▁spi
+- zelf
+- 沢
+- ▁helping
+- খ
+- ▁Dat
+- 价格
+- 看到的
+- ▁деле
+- ▁kap
+- ▁accordingly
+- ӹ
+- ▁operate
+- ▁wer
+- 委员会
+- ▁meng
+- ▁literally
+- tari
+- 左右
+- лом
+- ▁bus
+- ▁permanent
+- 效
+- 四十
+- хий
+- pol
+- vous
+- rè
+- こうした
+- ▁arise
+- こっち
+- ▁lại
+- ্যা
+- teri
+- ▁driver
+- 十五
+- ▁American
+- соб
+- ù
+- ▁heap
+- ▁barn
+- nger
+- ▁Cas
+- すれば
+- ▁rifle
+- ▁Produkt
+- ▁After
+- ▁まず
+- 中间
+- ▁stem
+- ▁background
+- stehen
+- ▁Tage
+- bij
+- ▁schließlich
+- ▁কি
+- rte
+- кай
+- ▁writer
+- 額
+- ▁duties
+- ▁hungry
+- ▁wegen
+- ▁piano
+- ▁pretend
+- 運転
+- ▁sleeping
+- mpa
+- 大脑
+- iw
+- 部门
+- ▁doctrine
+- ▁define
+- dian
+- 美元
+- mmen
+- 提醒
+- ▁efficient
+- ▁続いては
+- ו
+- eo
+- ▁plötzlich
+- ສ
+- ▁несколько
+- きょうは
+- denken
+- дә
+- ▁President
+- ▁occasionally
+- 不喜欢
+- ▁James
+- ▁perd
+- ▁eigen
+- 今日の
+- ▁nombre
+- under
+- 津
+- ▁estaba
+- ▁бел
+- ▁rod
+- ▁mala
+- ▁sovereign
+- 今回は
+- anno
+- ▁понял
+- مە
+- ▁tiempo
+- 平均
+- ▁initiatives
+- ▁Tro
+- ▁тре
+- got
+- ▁industrial
+- uro
+- ▁Weil
+- ▁beast
+- ▁spell
+- ことに
+- ▁conviction
+- ▁sorts
+- ▁sistema
+- ▁visited
+- lines
+- ▁olma
+- enz
+- 寝
+- 伴
+- ▁《
+- 状态
+- ▁tele
+- ▁quit
+- ▁poder
+- registr
+- ▁你怎么
+- 将军
+- 战斗
+- leben
+- 途
+- ্য
+- 触
+- ▁absolut
+- 朱
+- bis
+- 懂
+- ▁cells
+- 里边
+- stadt
+- ▁kings
+- はず
+- ▁divi
+- ▁bless
+- ▁einge
+- 不需要
+- ်
+- ▁queer
+- bir
+- ▁свет
+- ич
+- alter
+- çe
+- 先月
+- ▁lawyer
+- ▁кур
+- now
+- еж
+- ▁formal
+- ▁город
+- ▁extended
+- ▁poverty
+- 許
+- буд
+- 湿
+- bringen
+- 贴
+- ▁từ
+- ▁verlassen
+- ▁Folge
+- ▁lest
+- 那天
+- ▁composed
+- ▁innovation
+- quen
+- ▁вз
+- ▁striking
+- 锁
+- ▁fellows
+- サー
+- ▁Team
+- ▁lieutenant
+- ряд
+- ▁loro
+- ▁gefunden
+- ▁Freunde
+- ▁Minuten
+- 地震
+- ▁これが
+- ▁generous
+- ▁толькі
+- rib
+- 経験
+- ヘ
+- hap
+- 家的
+- pid
+- iente
+- 丢
+- chel
+- mod
+- ▁sagten
+- enge
+- ▁exciting
+- 되
+- 月に
+- ▁spielen
+- ▁destruction
+- ▁すごい
+- 接近
+- ович
+- ▁stiff
+- constru
+- pur
+- 想想
+- ▁Wochen
+- 岸
+- tsi
+- ▁premier
+- 见过
+- ▁Non
+- ▁possibility
+- ▁responsible
+- ší
+- ▁expand
+- স্
+- ị
+- Be
+- ▁strain
+- ڵ
+- ▁интерес
+- ister
+- dau
+- ▁singular
+- ▁location
+- ▁prospect
+- 変化
+- ▁suo
+- ▁genannt
+- ▁stress
+- лд
+- ubwo
+- 検
+- るの
+- бли
+- жээ
+- ▁Sein
+- ▁cả
+- 相手
+- ▁вторая
+- ▁Morgen
+- comb
+- ▁troba
+- вин
+- igo
+- 那么多
+- ▁должны
+- いで
+- ās
+- ▁Dis
+- ▁那我
+- 昨天
+- geld
+- ▁acts
+- ▁Trump
+- ▁boats
+- ジャ
+- ▁eigene
+- 隔
+- 输
+- іць
+- лин
+- ▁Schule
+- ▁যা
+- пре
+- ▁association
+- ▁trembling
+- 猜
+- вая
+- ரை
+- loop
+- ▁recall
+- どうぞ
+- buch
+- '300'
+- ▁gibi
+- ▁stretched
+- ட்டி
+- ▁россии
+- ▁нэг
+- 5%
+- ▁performed
+- グループ
+- ıl
+- ▁proposal
+- ▁semi
+- 捜査
+- ▁sehe
+- 隆
+- ▁accustomed
+- したのは
+- ▁schlecht
+- 长大
+- ▁unusual
+- ress
+- 很快
+- ▁frequent
+- activ
+- gezogen
+- ▁germany
+- ▁ça
+- bound
+- ▁باش
+- ▁Fla
+- ifica
+- ▁anni
+- یو
+- ▁brings
+- ▁вельмі
+- лог
+- ▁individuals
+- maz
+- »
+- 걸
+- ▁Ordnung
+- лё
+- 大阪
+- ▁یا
+- ேன்
+- erte
+- 不断
+- தான்
+- ▁ale
+- кла
+- ▁glanced
+- jya
+- pis
+- 是不
+- aro
+- ▁Yo
+- 我们必须
+- sprech
+- چی
+- ▁nav
+- graph
+- ▁großer
+- uten
+- 温度
+- ▁Cre
+- ▁modo
+- ▁mewn
+- ▁wearing
+- figur
+- uye
+- ▁North
+- chung
+- ▁kay
+- ▁european
+- るのは
+- ▁billion
+- служ
+- 細
+- ▁hunt
+- ▁responsibility
+- ▁ও
+- ▁говорит
+- ▁grateful
+- 太阳
+- ▁fragen
+- 抢
+- ▁অঁ
+- ▁dice
+- ▁riding
+- ▁tard
+- dition
+- ファン
+- ▁있어
+- ▁sports
+- π
+- ▁india
+- ья
+- বি
+- cum
+- 妻
+- 无论
+- 意外
+- produkt
+- やっぱ
+- 而是
+- ありません
+- 症状
+- ▁Op
+- bio
+- ▁victim
+- ▁wanna
+- owa
+- ▁Ak
+- ▁самом
+- ▁Glück
+- rse
+- سه
+- 我们已经
+- rede
+- sept
+- 霍
+- ▁Gre
+- dwa
+- nant
+- 仲
+- rack
+- ▁thì
+- 我将
+- ▁butter
+- ▁hace
+- ▁Š
+- 特朗普
+- át
+- bera
+- ▁делать
+- 力量
+- ▁yok
+- ▁داد
+- het
+- gis
+- ▁Leuten
+- 다고
+- för
+- ▁agent
+- ป
+- 監督
+- تح
+- 誰
+- anc
+- ▁mehrere
+- 作为一个
+- ▁değil
+- 瓶
+- ▁지금
+- 下午
+- ▁katika
+- 同学
+- ▁سا
+- ▁invited
+- ▁spare
+- ▁represented
+- eld
+- 印
+- 撞
+- ▁swept
+- ▁wishes
+- 之外
+- などが
+- ▁maiden
+- ▁الف
+- ▁prize
+- ase
+- phone
+- 問
+- خر
+- ▁concluded
+- ▁repair
+- gos
+- ню
+- liz
+- ее
+- 有可能
+- ▁ак
+- ▁neben
+- 造成
+- ▁nennen
+- ▁Cy
+- ▁oak
+- овский
+- rif
+- lern
+- ezi
+- ▁Gemeinde
+- ▁guten
+- ▁hid
+- ▁poetry
+- ▁mano
+- ▁такая
+- ▁plot
+- ▁selling
+- ▁gekommen
+- ▁Maria
+- ▁Min
+- 导致
+- ີ
+- gelegt
+- 操
+- fod
+- 就算
+- لب
+- 家人
+- 墙
+- ところで
+- 脑
+- ▁Komm
+- 版
+- なの
+- ▁yap
+- ▁proposition
+- eze
+- ▁latin
+- 喂
+- gebaut
+- cinc
+- politik
+- ▁そこで
+- aquesta
+- ▁але
+- gala
+- ▁actor
+- ggi
+- тя
+- არ
+- unu
+- 荷
+- ĩ
+- ээр
+- ▁sil
+- ▁поселок
+- ▁Pat
+- 了一些
+- uren
+- found
+- 摇
+- ент
+- ▁signed
+- くなる
+- ▁display
+- 不一样
+- やつ
+- 内部
+- ▁tema
+- 演讲
+- ▁select
+- ▁ashamed
+- 決勝
+- 户
+- ▁elder
+- 寒
+- jin
+- 发生了
+- wyn
+- स्
+- 孩子们
+- ちゃ
+- ▁rude
+- ▁именно
+- әй
+- 生气
+- 販売
+- ▁yw
+- ▁movements
+- 好き
+- ▁feared
+- 实际
+- ▁dun
+- ▁creation
+- ▁pied
+- 腕
+- point
+- жил
+- யி
+- Bri
+- ▁cad
+- genda
+- ▁kissed
+- 标
+- ▁chest
+- ▁cheeks
+- 避
+- 冒
+- 宗
+- kwi
+- ▁Tal
+- ▁cycle
+- ▁geschrieben
+- られない
+- lian
+- 映
+- lebt
+- 帰
+- ▁adopt
+- たり
+- kola
+- ▁payment
+- ான்
+- ▁hol
+- త
+- ▁bible
+- ▁achieve
+- 意味着
+- hü
+- пят
+- ▁rail
+- おい
+- ▁действительно
+- ▁Platz
+- اط
+- 卵
+- ▁materials
+- رت
+- ッと
+- ware
+- sses
+- ▁disposition
+- ▁glücklich
+- tial
+- fund
+- ▁continu
+- stimmt
+- 話を
+- kur
+- ▁attached
+- ▁mild
+- ▁locked
+- 染
+- フェ
+- ▁recover
+- ▁Qu
+- 顾
+- 办公室
+- ▁sou
+- kata
+- hak
+- 立て
+- 桥
+- ள்
+- лт
+- バイ
+- グラ
+- ▁shout
+- 一切都
+- 忍
+- ▁comments
+- ▁United
+- аться
+- үүл
+- ▁transition
+- ▁manufacture
+- 所以你
+- aine
+- 習
+- Ge
+- accord
+- ▁przy
+- ▁leurs
+- ▁jsem
+- ▁verb
+- 也可以
+- 就要
+- 你去
+- ▁brush
+- lov
+- шта
+- ▁jobs
+- 份
+- ▁appreciate
+- ガー
+- 围
+- ▁maj
+- emo
+- ▁supported
+- ▁abroad
+- ▁unu
+- ▁thấy
+- ▁sc
+- ้า
+- ▁못
+- pes
+- ▁Mag
+- ின்
+- ▁salut
+- tatu
+- ▁juga
+- ්
+- ох
+- つまり
+- 几年
+- ▁நான்
+- ▁font
+- ▁Kirche
+- ▁knee
+- работ
+- ▁Zwei
+- ▁distress
+- ▁Bau
+- 猪
+- wur
+- hof
+- ▁blessing
+- ழ
+- mera
+- ▁heavily
+- ▁الص
+- cover
+- 逆
+- 磨
+- 려
+- 签
+- 強く
+- гра
+- 含
+- ▁behavior
+- وان
+- ▁pine
+- bewusst
+- ▁neun
+- ▁bones
+- ▁それが
+- ▁armed
+- ▁quant
+- нут
+- 同志
+- るか
+- ▁refuge
+- 说过
+- ирован
+- 播
+- 见到
+- いった
+- ▁Bill
+- ▁reserve
+- ▁farmer
+- roy
+- 没有什么
+- ▁intimate
+- شن
+- சு
+- ▁essentially
+- ▁aquest
+- ▁corre
+- ▁sala
+- ▁ändern
+- 郎
+- んじゃないか
+- ▁Tages
+- ▁ここで
+- ▁сама
+- 厂
+- ▁хар
+- 错了
+- ளி
+- ▁june
+- ▁competition
+- ▁fois
+- ilen
+- リア
+- 我不会
+- ▁ruth
+- ▁opposed
+- ▁penn
+- ▁هستند
+- 拖
+- ▁antwortete
+- ▁jeune
+- إ
+- ▁repeat
+- ▁greek
+- 就说
+- ș
+- '!」'
+- ▁teachers
+- ▁sy
+- あなた
+- 看来
+- isation
+- ▁rya
+- 一段时间
+- ▁belong
+- obu
+- ാ
+- ▁traditional
+- ▁painted
+- ことも
+- чь
+- ▁Mrs
+- сов
+- ▁settlement
+- いろいろ
+- ▁کنید
+- load
+- ▁massive
+- ▁zag
+- ▁alternative
+- dá
+- aren
+- メートル
+- ▁retired
+- ▁consult
+- ▁feels
+- njye
+- 训练
+- ▁charming
+- hop
+- ▁尽管
+- wl
+- ▁sole
+- gad
+- ▁karo
+- nez
+- tec
+- 我告诉你
+- 国の
+- jon
+- ▁bio
+- крас
+- tali
+- ▁việc
+- ▁reasonable
+- 制度
+- ▁kate
+- bw
+- юсь
+- نش
+- ▁schaffen
+- 搭
+- ▁slipped
+- 気温
+- tī
+- ゴール
+- ّ
+- ▁adopted
+- 更に
+- para
+- となっています
+- ind
+- ▁etc
+- んでしょうか
+- ▁scan
+- ▁ambition
+- iques
+- ▁basket
+- rau
+- ▁bark
+- 鞋
+- させる
+- dora
+- igu
+- éc
+- ían
+- ▁supreme
+- يف
+- krat
+- mund
+- 卷
+- avo
+- 暴力
+- ▁seldom
+- hack
+- urt
+- ▁approaching
+- 幼
+- др
+- ▁murmured
+- kni
+- ▁goodness
+- ▁Так
+- zeug
+- 合わせ
+- Po
+- 腰
+- ▁Val
+- ▁consistent
+- ▁rever
+- ▁gap
+- ▁gehe
+- ▁dame
+- witt
+- ▁handelt
+- ści
+- dreh
+- ▁pages
+- يو
+- 另外
+- ě
+- nach
+- ▁unfortunate
+- headed
+- 恨
+- path
+- ましたね
+- 主人
+- ▁совет
+- rī
+- ▁aren
+- ять
+- ▁depends
+- flo
+- miss
+- ▁таки
+- ▁zeigt
+- ▁terre
+- ▁commanded
+- ▁noi
+- bell
+- ▁Tatsache
+- cycl
+- رك
+- mani
+- ்
+- ▁muito
+- duk
+- lir
+- ▁tidak
+- ▁regiment
+- izo
+- ほか
+- ▁clerk
+- ▁vào
+- ▁года
+- 雄
+- ▁bullet
+- tree
+- ▁sempre
+- hlen
+- дэг
+- erin
+- rimo
+- 银
+- ▁teams
+- ▁foi
+- ums
+- ▁shortly
+- ▁cure
+- atge
+- 恢复
+- ▁Hol
+- づ
+- ▁Test
+- ▁drinking
+- ▁released
+- tics
+- ▁eastern
+- St
+- ▁zaman
+- 作用
+- ▁versus
+- ense
+- সে
+- aught
+- ▁жив
+- 라고
+- 押
+- ▁reflected
+- イギリス
+- ▁Hund
+- கை
+- وع
+- 交易
+- ▁blank
+- ▁Situation
+- ▁жизни
+- ▁minor
+- 大変
+- 으로
+- ▁obvious
+- 环
+- 攻
+- ▁useless
+- ▁blieb
+- のほう
+- ▁slaves
+- 材
+- ▁également
+- fal
+- сын
+- ▁tras
+- овой
+- ▁wherever
+- рез
+- ▁contest
+- care
+- ▁institution
+- дав
+- гло
+- 的那
+- ▁contribution
+- 这边
+- нне
+- ▁administration
+- 互联网
+- 了啊
+- ▁donde
+- она
+- ▁Demà
+- ление
+- ▁也许
+- ▁jour
+- ▁বা
+- ес
+- 打算
+- ▁быў
+- awa
+- mere
+- रा
+- 我不是
+- platz
+- 私も
+- pè
+- geven
+- ▁ensure
+- ▁importante
+- 彼
+- ▁داشت
+- 表情
+- тру
+- ところが
+- ▁vital
+- bab
+- 給
+- nio
+- ▁nutzen
+- ją
+- 泡
+- 不得不
+- rend
+- یان
+- 柱
+- ▁glorious
+- ensi
+- 这事
+- ாக
+- abu
+- 有一天
+- ▁feast
+- 동
+- ▁olan
+- 把这个
+- ▁inhabitants
+- キロ
+- ließ
+- 资金
+- 庄
+- ▁Tas
+- area
+- 気が
+- gues
+- يق
+- шин
+- 積
+- bug
+- 練習
+- ▁stroke
+- ▁Sy
+- ▁või
+- 狂
+- 迟
+- ▁Tot
+- arse
+- ▁plum
+- 不应该
+- ▁handed
+- ▁earnest
+- 勤
+- 对此
+- 我来
+- 敌人
+- 佳
+- ▁acquired
+- lak
+- жал
+- ▁fever
+- ▁impulse
+- 任何人
+- 南部
+- ▁dove
+- ▁ihres
+- 저
+- nega
+- 前面
+- готов
+- ▁newspaper
+- ックス
+- rait
+- мар
+- 碎
+- 老公
+- 所做的
+- ▁otro
+- ▁difficulties
+- 開催
+- ▁owner
+- 批
+- bü
+- aver
+- дать
+- ▁trace
+- ▁gleichen
+- தை
+- weisen
+- aɣ
+- ▁coin
+- شه
+- 드
+- 例えば
+- agon
+- ▁print
+- ▁treffen
+- اً
+- 乗り
+- 法国
+- ذا
+- ▁hombre
+- 隐
+- 不足
+- ▁Spa
+- kı
+- ▁departure
+- 的方法
+- 老婆
+- ▁хотел
+- ▁creating
+- nten
+- 台風
+- 人も
+- ▁وجود
+- ▁Sk
+- ates
+- ▁tough
+- ▁ўсё
+- ▁unlike
+- flu
+- ▁zweite
+- 冠
+- oba
+- 以上の
+- 一件
+- sell
+- ova
+- ▁accomplished
+- ально
+- てください
+- 峰
+- pati
+- 住在
+- ▁ignorant
+- 事实上
+- ▁deeper
+- ▁travers
+- ▁venture
+- 予定
+- らず
+- 病人
+- يك
+- হা
+- ▁Wirtschaft
+- 是由
+- ワン
+- cola
+- صل
+- 妻子
+- ▁我在
+- ▁reputation
+- ▁entfernt
+- trieb
+- ▁fiscal
+- ▁combined
+- การ
+- ▁plainly
+- ▁closing
+- ▁og
+- غر
+- ▁eher
+- 押し
+- ▁فقط
+- тры
+- 多的
+- 要去
+- イメージ
+- 欧洲
+- och
+- ṣ
+- ▁reckon
+- ▁granted
+- 際
+- ể
+- ▁italy
+- 熟
+- ļ
+- 淡
+- ▁whisper
+- лай
+- ▁painting
+- ▁poi
+- ▁smoking
+- ▁понимаю
+- ёр
+- krank
+- ▁insist
+- うまく
+- ▁fairy
+- 但他
+- アップ
+- мест
+- цен
+- quis
+- まさに
+- tör
+- ▁mono
+- 試
+- 続
+- ▁orange
+- キャ
+- 你不能
+- お金
+- ▁Sohn
+- гд
+- 哪儿
+- 网站
+- 你怎么
+- amento
+- icia
+- 深刻
+- тр
+- ▁mist
+- تو
+- ▁weder
+- 以降
+- 乳
+- ▁attractive
+- 漂亮
+- 广
+- ▁sue
+- ▁咱们
+- にとって
+- 追い
+- ден
+- ▁Cla
+- ▁такие
+- 始まり
+- ках
+- حق
+- ▁Ali
+- 内心
+- ▁jemals
+- ▁suspect
+- 次の
+- ▁wollten
+- ▁hunter
+- ▁slope
+- lee
+- agenda
+- ▁spre
+- ká
+- 供
+- ▁pond
+- 你了
+- になり
+- 还能
+- ▁risks
+- ものが
+- ▁Av
+- юцца
+- ▁Because
+- ▁мой
+- usi
+- 叔
+- ▁seeking
+- ▁momento
+- ▁margins
+- 近く
+- ▁General
+- ▁epi
+- hö
+- ▁compelled
+- так
+- тон
+- mä
+- ▁وأ
+- 艺
+- ▁pound
+- ์
+- ▁admiration
+- ▁songs
+- 行動
+- ▁dein
+- 館
+- ▁ښه
+- ▁abuse
+- ▁Stück
+- ▁pope
+- ois
+- ▁Umu
+- كي
+- ▁igihe
+- になっています
+- ấ
+- ッド
+- manda
+- hush
+- ▁interessant
+- ▁hörte
+- ▁bya
+- ▁jean
+- 的手
+- ▁infinite
+- ▁entonces
+- ают
+- 感兴趣
+- ける
+- 一体
+- foot
+- дол
+- ому
+- эв
+- chas
+- ând
+- rück
+- zia
+- ▁이런
+- 难道
+- ▁adapt
+- ▁Kor
+- 心配
+- ▁regions
+- ▁eagerly
+- ▁Mensch
+- 一项
+- sab
+- 殿
+- ▁Kol
+- ▁communities
+- ▁kaum
+- odor
+- わり
+- ▁Kla
+- 而言
+- ń
+- ▁yard
+- dine
+- ино
+- ▁lunch
+- 武器
+- ▁我现在
+- ▁forehead
+- ▁Bon
+- ▁fame
+- Ж
+- ▁那是
+- ▁withdraw
+- ▁mask
+- ▁تع
+- ▁dag
+- liness
+- 佩
+- 三年
+- ▁всем
+- ▁aho
+- レン
+- 哪里
+- полит
+- 냐
+- ▁idle
+- ▁dealing
+- bern
+- 的信息
+- ▁روی
+- ご紹介
+- ками
+- geze
+- eti
+- has
+- ▁Mary
+- ▁planning
+- сь
+- 每次
+- ュ
+- 控
+- 阶段
+- ▁Kunst
+- iff
+- لى
+- liga
+- ▁جو
+- ▁remove
+- 老人
+- ▁mister
+- 準備
+- ▁islands
+- ▁erreichen
+- 多い
+- ▁characters
+- ▁Bis
+- ▁вся
+- 一致
+- 祖
+- stieg
+- ▁mom
+- ▁само
+- geri
+- boy
+- ▁Rwanda
+- ▁estat
+- 印度
+- ▁client
+- wacht
+- 收入
+- ▁Mes
+- bles
+- jä
+- ▁acting
+- ▁infrastructure
+- ▁predict
+- ổ
+- щи
+- ロン
+- cido
+- grav
+- ▁мог
+- ▁burned
+- ▁gebracht
+- ▁shell
+- ▁australia
+- cult
+- 金融
+- ▁steep
+- ▁rang
+- 出生
+- 假设
+- ▁مر
+- ▁anirà
+- сен
+- ▁まあ
+- 这是我
+- ▁되게
+- voj
+- ——
+- ▁organic
+- छ
+- 的影响
+- ▁gather
+- 对话
+- 很重要
+- ât
+- ffen
+- dauer
+- ват
+- tons
+- 跟着
+- 并不是
+- 今晚
+- ▁Sinn
+- ▁nghĩ
+- ▁Bericht
+- ▁É
+- cono
+- lose
+- 辞
+- こちらです
+- ▁일
+- 爱情
+- だな
+- 说明
+- ях
+- ▁tag
+- ▁اون
+- ▁electric
+- 昔
+- ▁commander
+- 凡
+- கு
+- ▁chin
+- det
+- 続け
+- ▁fy
+- 比如说
+- ▁thrill
+- 一千
+- 证据
+- だったら
+- ▁яны
+- ▁causa
+- ▁insult
+- ▁fou
+- час
+- 实际上是
+- 莎
+- ▁Rück
+- ▁vessels
+- ▁junge
+- ▁eens
+- ▁inclined
+- ▁opposition
+- 計
+- ▁Forschung
+- spec
+- ▁unfortunately
+- ▁towns
+- ▁الر
+- 提到
+- ▁shirt
+- ▁сделать
+- 执行
+- 野菜
+- 快速
+- hole
+- ▁இரு
+- isiert
+- ▁ella
+- ▁Richtung
+- ▁wedding
+- ▁Viele
+- ▁leadership
+- 義
+- 你觉得
+- 销
+- 逃げ
+- ▁ali
+- ▁institutions
+- nā
+- aven
+- ▁billy
+- ▁schrecklich
+- licher
+- ▁icyo
+- ब
+- ▁descend
+- 答应
+- 判
+- eck
+- ▁gihe
+- 注意到
+- 情感
+- dha
+- '35'
+- ▁undertake
+- ▁extrem
+- 拼
+- ▁Unterschied
+- won
+- ▁equipment
+- لم
+- 多く
+- år
+- lässt
+- ▁philosopher
+- ▁uttered
+- 毕业
+- 主任
+- 星期
+- 施
+- ▁sed
+- маг
+- 再见
+- ▁Yu
+- 我跟
+- 长的
+- やった
+- いか
+- 发现了
+- くん
+- ▁shared
+- 租
+- ların
+- ▁doubtless
+- aires
+- كل
+- යි
+- 가지고
+- ソン
+- 的吗
+- ▁Einige
+- ▁vrouw
+- 不够
+- ▁intense
+- fol
+- ▁glow
+- ▁permitted
+- ▁pis
+- quelle
+- ёт
+- 成长
+- طور
+- lib
+- 跟我说
+- 最後の
+- 価格
+- 逼
+- ▁settle
+- 结构
+- stände
+- κ
+- fallen
+- ▁negro
+- ▁♫
+- chin
+- ▁лучше
+- য়ে
+- schein
+- нул
+- ▁While
+- ike
+- ▁tijd
+- ▁học
+- ्या
+- ▁emotional
+- 模
+- city
+- zed
+- ▁passat
+- ▁avons
+- 래
+- tah
+- ▁человека
+- 是一
+- говор
+- ҡы
+- ▁desde
+- ▁deny
+- ▁obligation
+- ভ
+- رق
+- 異
+- stad
+- ▁providing
+- 专家
+- ▁Ag
+- zak
+- ▁Sand
+- ▁lighted
+- ▁baba
+- hall
+- ografi
+- ▁nella
+- stab
+- ▁substantial
+- иль
+- கோ
+- ▁vague
+- 一方
+- ▁cease
+- ▁Erfahrung
+- 严重
+- λ
+- 的眼睛
+- ▁ありがとうございます
+- 去找
+- ▁historical
+- олог
+- 踏
+- ▁Name
+- 意识
+- 何で
+- 将来
+- 域
+- یس
+- どういう
+- したり
+- gura
+- ▁curve
+- ▁Aufgabe
+- ▁fence
+- ▁dancing
+- ▁flung
+- ▁investigation
+- folge
+- ▁evolution
+- ▁და
+- ▁Мо
+- گە
+- ▁conserva
+- нии
+- ▁load
+- ▁nhiều
+- ▁parole
+- 尊重
+- ',"'
+- psy
+- 効果
+- ▁egypt
+- geht
+- たちが
+- ulu
+- ús
+- 表演
+- 候
+- そういった
+- ▁speaker
+- ▁East
+- ▁utiliz
+- 记忆
+- ▁wooden
+- ▁beloved
+- 一度
+- ▁allows
+- ▁pursue
+- keeper
+- 環境
+- leh
+- 胜
+- ▁phase
+- 긴
+- 更好
+- 没想到
+- ím
+- tě
+- ▁lecture
+- ▁응
+- 订
+- 犬
+- ▁vorbei
+- ▁Mü
+- ▁lately
+- nosti
+- ▁bod
+- 一定会
+- programm
+- ▁arme
+- ▁irish
+- ữ
+- 恐惧
+- 생
+- idade
+- 按照
+- 佛
+- ▁Gesetz
+- ▁rabbit
+- 钟
+- ▁wilde
+- 国民
+- west
+- ה
+- 我々
+- 慢慢
+- 刑
+- 頼
+- tun
+- 搜索
+- note
+- ▁grandfather
+- larını
+- ▁decisions
+- ▁murmur
+- ▁terra
+- ▁할
+- ▁explore
+- ▁Ven
+- していて
+- கள
+- ▁TED
+- tivo
+- 文字
+- メンバー
+- bà
+- 驚
+- ▁staring
+- 電話
+- 满足
+- 支付
+- pac
+- ▁cliff
+- erson
+- のため
+- 穿过
+- 插
+- пал
+- ▁trunk
+- ▁league
+- ▁acest
+- ▁Tan
+- orient
+- cele
+- ▁desperate
+- ▁彼は
+- tical
+- ना
+- ▁expensive
+- ڕ
+- 适
+- ▁stuck
+- ▁Stunden
+- ▁dared
+- ▁Jahrhundert
+- gres
+- ஹ
+- ▁cheerful
+- かどうか
+- oor
+- 議員
+- 更新
+- 益
+- һы
+- ▁Rest
+- 分かって
+- 创
+- ▁après
+- ▁إن
+- иде
+- Q
+- ông
+- ▁obey
+- しても
+- iad
+- tje
+- nner
+- zimmer
+- ますよね
+- ▁sistem
+- ▁workers
+- ▁거기
+- 员工
+- ▁spain
+- ▁nad
+- ▁rend
+- ▁observation
+- ▁llama
+- ▁volumes
+- 地下
+- 北朝鮮
+- ந
+- gir
+- ▁declare
+- ▁upward
+- 極
+- طر
+- ▁deserve
+- ▁klein
+- ▁defence
+- ▁increases
+- ▁simp
+- ▁tramp
+- 台湾
+- 对他
+- bord
+- ▁camera
+- ancy
+- ▁jacob
+- ▁kwe
+- ată
+- 问题是
+- 実際に
+- ▁Sche
+- welt
+- ▁Arten
+- gla
+- ▁Bürger
+- የ
+- ộ
+- ▁Worte
+- ever
+- 梦想
+- bec
+- ▁begged
+- ▁wretched
+- ▁hơn
+- ▁хотя
+- 些
+- ስ
+- ант
+- ▁sending
+- 降り
+- ▁gilt
+- lando
+- że
+- ▁India
+- ▁questa
+- grün
+- 手を
+- qi
+- ▁Mel
+- ▁assure
+- 这意味着
+- あって
+- смотр
+- iber
+- ナー
+- шка
+- ця
+- ▁entering
+- ▁prze
+- tak
+- ▁hervor
+- ages
+- ▁offset
+- اج
+- ▁今日
+- ▁cũng
+- ▁consumer
+- zzi
+- 신
+- ▁adding
+- ▁virginia
+- ▁capture
+- ▁pistol
+- ▁curl
+- ▁accounts
+- 律
+- ▁Zahl
+- rut
+- なん
+- 紧张
+- ▁pursuit
+- ▁Michael
+- ası
+- ▁Far
+- ▁Berg
+- itza
+- stück
+- mira
+- ▁besonders
+- ▁enhance
+- fulness
+- dhi
+- யாக
+- ▁چې
+- ube
+- 肩
+- ▁jews
+- ▁أنه
+- ▁ils
+- ▁swear
+- ▁erreicht
+- ▁sentiment
+- ▁reaching
+- илась
+- かり
+- ▁cents
+- 중
+- 让她
+- ▁volta
+- ▁yell
+- 今天的
+- ские
+- ▁expansion
+- ▁notre
+- ▁torture
+- ▁purposes
+- 她们
+- ▁disgust
+- ެ
+- ▁aktiv
+- ▁highlight
+- yol
+- ้
+- ▁pit
+- 所以我们
+- 力を
+- ▁magazine
+- ▁America
+- rance
+- 它会
+- 崎
+- ▁saddle
+- ▁pont
+- 来た
+- blau
+- ક
+- tama
+- ვი
+- ▁Richtig
+- ▁depart
+- dag
+- ▁只要
+- ▁Trans
+- lagen
+- ▁кан
+- になっている
+- 思って
+- 不起
+- ▁indulge
+- polis
+- 鳥
+- ▁четырнадцать
+- ▁тэр
+- ভা
+- gabe
+- ▁langsam
+- በ
+- ▁должен
+- ▁leverage
+- ▁bench
+- 摆
+- ▁distinction
+- /
+- 大雨
+- 地上
+- ने
+- куп
+- ▁biết
+- действ
+- tore
+- east
+- ▁brook
+- brach
+- issa
+- 해서
+- ▁lesen
+- ▁Erde
+- ▁bari
+- ▁quelque
+- say
+- sinn
+- ▁vend
+- 階
+- rand
+- نە
+- ▁damals
+- рак
+- ▁recovered
+- 過去
+- の大
+- ▁verd
+- ники
+- 运行
+- ▁begins
+- дум
+- ▁mbere
+- 圧
+- ▁alice
+- 范
+- ▁cutting
+- ▁rescue
+- бар
+- 奶奶
+- ▁incredible
+- 理论
+- ▁grain
+- ▁habits
+- なし
+- ▁merry
+- ▁جا
+- ったり
+- уме
+- 现代
+- نس
+- ▁whereas
+- ▁cloth
+- ▁havas
+- ää
+- 时期
+- 启
+- ▁gesture
+- ▁belonged
+- ▁Ко
+- تە
+- ▁случае
+- jet
+- 拿着
+- ▁journal
+- ▁rien
+- 暮らし
+- 活躍
+- èrent
+- 得多
+- 合わせて
+- vuga
+- ▁lodge
+- 是他
+- ▁gingen
+- தே
+- ロシア軍
+- してきた
+- يس
+- ação
+- ата
+- 一方で
+- ▁tat
+- 観
+- 一名
+- ▁بد
+- ▁irre
+- ▁خیلی
+- 過ぎ
+- ▁sowohl
+- 歌曲
+- nett
+- ▁früh
+- ▁Та
+- ▁гэж
+- 宋
+- 荒
+- 掌
+- kad
+- ▁Reise
+- ▁dread
+- ▁fog
+- 没错
+- bled
+- うわ
+- 积极
+- ▁Verbindung
+- ▁sensible
+- ▁mounted
+- ▁interes
+- mur
+- 设备
+- 音楽
+- '2000'
+- ▁никто
+- rug
+- ಾ
+- We
+- 承
+- ▁shadows
+- ▁vin
+- nung
+- 被称为
+- ▁Weißt
+- ▁arrow
+- jal
+- زه
+- ▁misery
+- 狼
+- lets
+- 的机会
+- wig
+- glich
+- 遗
+- fehl
+- ▁komm
+- ▁なるほど
+- 信任
+- ▁brachte
+- iel
+- ▁editor
+- ▁Alter
+- 他对
+- ости
+- ▁gegeben
+- над
+- جي
+- نم
+- gard
+- lja
+- 傻
+- лав
+- ť
+- 、3
+- ▁padre
+- 厳しい
+- 我不能
+- んじゃない
+- ▁observ
+- उ
+- ▁prepare
+- 珠
+- tanga
+- でございます
+- ▁bowed
+- ▁Super
+- '27'
+- iers
+- ись
+- ▁Sen
+- 不如
+- ▁jumped
+- ▁Gal
+- ▁permet
+- 仕
+- 很少
+- ރ
+- 投げ
+- 了这个
+- ▁troubled
+- ▁sufficiently
+- 我不想
+- hun
+- ▁cars
+- rzy
+- ▁그래
+- 欢
+- gelijk
+- erlei
+- ▁whence
+- рук
+- și
+- ▁Martin
+- stelle
+- ▁Van
+- ▁باید
+- 塩
+- ▁schl
+- 采取
+- ング
+- 让人
+- 为什么要
+- ▁lamb
+- 진
+- ▁fired
+- ▁Wissenschaft
+- 沿
+- 多了
+- daki
+- れない
+- ▁señor
+- dura
+- zahl
+- ▁zeer
+- ▁external
+- ▁كانت
+- mour
+- 失去
+- ▁Vereinigte
+- ▁муж
+- ર
+- cchi
+- فعل
+- ▁schreiben
+- ▁Deshalb
+- ▁falls
+- ▁disc
+- ▁drum
+- ▁нибудь
+- ▁рэ
+- 革命
+- ▁faut
+- 離
+- みると
+- ▁dalam
+- ▁Eis
+- ▁helpless
+- bbi
+- 闻
+- ▁голос
+- 的脸
+- ▁schöne
+- 电脑
+- ▁scorn
+- 没什么
+- 正解
+- spann
+- ▁Danke
+- 飞机
+- ▁runs
+- ▁includes
+- assa
+- ▁gedacht
+- ▁fins
+- verse
+- aged
+- enzi
+- bé
+- mento
+- ▁Arbeits
+- fang
+- ▁straw
+- ▁Schul
+- 訪
+- ▁experiences
+- ▁beam
+- ▁rendered
+- データ
+- ▁lesson
+- おり
+- ▁verse
+- ▁mirror
+- زا
+- ▁cheek
+- カメラ
+- ▁Gegen
+- ял
+- 部队
+- 穿着
+- ▁leads
+- ત
+- ▁technical
+- ▁conceal
+- ▁zeal
+- ▁authorities
+- ▁losing
+- arri
+- 我们是
+- ▁wagon
+- ▁Bett
+- ▁reports
+- ▁weapons
+- spel
+- ▁kick
+- ▁그게
+- ▁assumed
+- ▁qualities
+- あげ
+- 魅力
+- ▁dwell
+- マイ
+- ▁peak
+- 見せ
+- ▁jug
+- ▁endlich
+- ▁tio
+- 我没
+- ▁stretch
+- À
+- falls
+- ្
+- пі
+- nih
+- ▁aufge
+- 公開
+- аз
+- lju
+- گاه
+- ▁eve
+- 谈谈
+- მა
+- ▁namens
+- ▁করে
+- ジェ
+- وو
+- aging
+- tered
+- вай
+- ▁zit
+- とこ
+- ස්
+- iah
+- ▁discipline
+- ര
+- 미
+- Re
+- bara
+- ▁exposed
+- شت
+- ▁infant
+- ▁Ä
+- ▁email
+- ▁owe
+- 期间
+- ▁icy
+- 一句话
+- 徐
+- ょ
+- ▁yara
+- ▁Ana
+- ▁carbon
+- 行了
+- eja
+- あす
+- ▁pile
+- ▁planned
+- kita
+- ▁railway
+- ▁знаете
+- 一句
+- 製
+- üb
+- 今夜
+- ▁ahora
+- ▁naj
+- ?」
+- dou
+- şte
+- 正是
+- ▁ultimately
+- 沖縄
+- bahn
+- vine
+- 减少
+- ▁پس
+- ▁Washington
+- ▁considering
+- ▁Era
+- ▁persona
+- nimmt
+- 有时候
+- ▁bara
+- 用的
+- ▁tomorrow
+- ifa
+- 他们是
+- 看了
+- を中心に
+- front
+- ayı
+- ▁primary
+- 的心
+- 过程
+- ▁spur
+- ▁distribu
+- 要么
+- уш
+- ▁nk
+- 戦争
+- fir
+- ▁измен
+- 案件
+- ▁neighborhood
+- 意識
+- ▁drunk
+- ▁よろしくお願いします
+- 黑人
+- ▁tran
+- ▁мир
+- ▁extend
+- ▁USA
+- 神经
+- 时候
+- ▁그렇게
+- که
+- ▁raw
+- nus
+- 学会
+- mpel
+- impo
+- 军队
+- ▁Gericht
+- arra
+- ▁winds
+- ▁jungen
+- ప
+- ▁Lassen
+- ▁деньги
+- ▁sabe
+- ▁mobile
+- пло
+- ති
+- ▁cub
+- illo
+- ecek
+- ▁こちらは
+- ▁Gri
+- ി
+- ▁rằng
+- ▁Mil
+- piece
+- ▁mucho
+- 处于
+- ▁pier
+- ▁aloud
+- ▁Gold
+- ద
+- ▁Hilfe
+- قي
+- ▁americans
+- geno
+- ▁gates
+- vio
+- 早期
+- ▁separated
+- ▁நீ
+- ન
+- 出す
+- 最新
+- ▁mam
+- illes
+- ▁ведь
+- ▁avant
+- ▁заб
+- уй
+- ▁এই
+- 更好的
+- key
+- ▁kannst
+- ▁quantity
+- ▁свое
+- ▁toujours
+- ▁ages
+- ു
+- ませんでした
+- ▁opinions
+- ▁images
+- Se
+- 窗
+- setzung
+- 开放
+- 就能
+- みて
+- tritt
+- 辺
+- range
+- 運動
+- ▁wander
+- вет
+- 費
+- ▁haste
+- 正式
+- ▁sexual
+- ṭ
+- ▁Gedanken
+- ▁reli
+- それに
+- ▁100
+- ▁weird
+- ದ
+- 细胞
+- ▁همه
+- ▁portrait
+- ▁chill
+- 生活中
+- чер
+- 纽约
+- estra
+- ▁sera
+- 全く
+- anz
+- ▁слова
+- ▁County
+- もん
+- ▁det
+- وج
+- ▁deed
+- ▁claims
+- ▁delightful
+- ▁vroeg
+- ▁яе
+- ข
+- 別の
+- ▁тех
+- 면은
+- ark
+- ▁scattered
+- ▁gazed
+- rado
+- гром
+- वा
+- Any
+- обра
+- 仙
+- ality
+- ▁missed
+- ▁prin
+- 宣布
+- ▁flatter
+- 都没
+- ▁essere
+- ▁personally
+- 子里
+- 後ろ
+- 离婚
+- ▁magnificent
+- ▁miracle
+- ▁rolling
+- ▁thread
+- 述
+- ▁fuel
+- ▁altar
+- ▁قبل
+- دى
+- お母さん
+- 行く
+- ▁daughters
+- өг
+- ▁crash
+- ະ
+- ू
+- ▁challenges
+- его
+- 哪个
+- 介
+- ▁rồi
+- tron
+- akt
+- ▁ص
+- undzwanzig
+- ▁gens
+- maid
+- できます
+- plaats
+- ▁bosom
+- ロシアの
+- ▁mm
+- ▁methods
+- staan
+- 兄
+- ection
+- 握
+- ▁그때
+- ான
+- كى
+- せる
+- بع
+- colored
+- ▁ĉe
+- нда
+- mbre
+- ▁robe
+- マスク
+- ▁Mars
+- ▁concert
+- mov
+- you
+- ▁midnight
+- ▁missing
+- 分钟
+- ▁drag
+- ার
+- ▁joint
+- nost
+- つき
+- ▁Ён
+- ▁于是
+- sini
+- ▁Jag
+- ương
+- 历
+- ables
+- ▁restrain
+- ▁africa
+- ▁horizon
+- 以至于
+- خت
+- ▁reduction
+- ▁совершенно
+- ▁insisted
+- 邀请
+- 欲
+- 坦
+- bru
+- ▁Anfang
+- ▁zeggen
+- 우
+- ▁Feuer
+- freund
+- bwira
+- 减
+- ண்
+- ▁Afrika
+- ▁außer
+- 这么做
+- 宅
+- hearted
+- legt
+- 長い
+- にお
+- omen
+- ▁trou
+- ▁Brief
+- ▁anna
+- ▁vest
+- ▁你就
+- ских
+- ▁bestimmte
+- ▁sia
+- 练
+- ▁raising
+- 竹
+- 规则
+- ▁ignorance
+- cep
+- dek
+- ▁گفت
+- ▁einzu
+- 敗
+- ▁führte
+- ▁хорош
+- حل
+- ▁이렇게
+- 成本
+- 撤
+- ▁consequences
+- ▁factor
+- 旁边
+- ▁patron
+- ▁involve
+- pas
+- 风险
+- erie
+- cap
+- 宿
+- 够
+- ▁Ф
+- 一生
+- られました
+- ▁sink
+- 建造
+- ▁valor
+- ▁против
+- ▁cock
+- ▁نیست
+- 年龄
+- ティー
+- 严
+- izza
+- ▁момент
+- mäßig
+- ▁这是一个
+- ▁province
+- ▁convert
+- tina
+- cup
+- tiques
+- ▁breed
+- 干吗
+- rb
+- gefallen
+- ▁paint
+- ▁foe
+- ulation
+- 沃
+- ▁kugira
+- igh
+- っていうこと
+- ▁அவ
+- plic
+- koze
+- 亡
+- ▁prote
+- pok
+- ▁primera
+- ▁Sun
+- ▁Volk
+- ▁entra
+- ▁другой
+- ▁aufzu
+- емся
+- ▁cards
+- のよ
+- куль
+- 改善
+- ▁perché
+- 执
+- ▁Dingen
+- ▁mainly
+- wohn
+- ドル
+- ▁cet
+- ヒット
+- ವ
+- ▁folk
+- onde
+- このように
+- 劲
+- 明らかに
+- ▁trên
+- থ
+- 旗
+- 봐
+- ަ
+- ▁மா
+- ▁significantly
+- ▁chase
+- ▁مثل
+- 振り
+- 当初
+- 儿童
+- ▁nam
+- 会議
+- 剑
+- ӓ
+- ▁Richard
+- 날
+- ▁absent
+- რ
+- луч
+- 载
+- ▁هناك
+- ▁appar
+- せて
+- ▁Band
+- ▁我不
+- 後に
+- oedd
+- De
+- ▁минут
+- 没有任何
+- 距離
+- 竞争
+- ▁Ö
+- ▁Fahr
+- gaben
+- ▁15
+- sert
+- 危
+- ▁кого
+- 本人
+- ά
+- ▁Air
+- ▁Kultur
+- yal
+- ▁lust
+- 漫
+- ▁developing
+- '26'
+- ▁boots
+- あとは
+- stream
+- ▁ບໍ່
+- 弾
+- dress
+- ▁quelques
+- ▁Bewegung
+- ▁messenger
+- жен
+- ▁roads
+- ų
+- ▁menschen
+- gender
+- deck
+- ▁molto
+- mart
+- 学家
+- ▁basic
+- 也不会
+- ▁nag
+- チャ
+- を受けて
+- ի
+- 有的
+- ▁firmly
+- 提高
+- ▁одно
+- 団体
+- ender
+- ▁Bezug
+- kura
+- wort
+- ▁inclu
+- لار
+- 作家
+- 話題
+- 住民
+- bura
+- çı
+- ▁crowded
+- diği
+- ▁outcome
+- 有效
+- volle
+- ▁May
+- َّ
+- ▁honey
+- 行って
+- чет
+- boot
+- 分かる
+- یی
+- 家伙
+- boat
+- wara
+- stimmen
+- 公平
+- 発見
+- ▁Вы
+- 的情况下
+- ర
+- ▁sana
+- ▁ram
+- ▁тысяча
+- ▁especial
+- ▁hastily
+- されます
+- ▁Vielen
+- 和其他
+- 或者是
+- ▁hai
+- ▁nearest
+- ▁multitude
+- kraft
+- үүд
+- minister
+- ivi
+- ▁rất
+- ▁plays
+- ▁clay
+- 销售
+- ▁porte
+- 周围
+- wechsel
+- ууд
+- ▁erzählen
+- ▁ernst
+- ▁оно
+- ▁系
+- 市民
+- ▁leaning
+- ▁overcome
+- 长期
+- melt
+- ▁Groß
+- үр
+- ▁innerhalb
+- ▁scrap
+- cita
+- ▁returns
+- ▁april
+- 奈
+- gehalten
+- 应用程序
+- dela
+- ▁kri
+- amp
+- 滚
+- ▁grasp
+- ▁cách
+- 软
+- ▁corps
+- elde
+- ▁lincoln
+- ▁lee
+- 全て
+- ▁continually
+- ▁Rus
+- ▁خوب
+- 一步
+- 人と
+- 一九
+- ▁đầu
+- ▁lokal
+- ▁молод
+- に対し
+- ege
+- ▁disorder
+- ▁implement
+- 吐
+- 默
+- nick
+- 足够
+- 者が
+- ▁twin
+- ▁между
+- 餐
+- 下的
+- タイ
+- iment
+- 自分が
+- ▁reveal
+- ờ
+- schlagen
+- uki
+- richten
+- тельно
+- 相关
+- ▁kadar
+- ▁cosas
+- ▁chinese
+- ▁Paris
+- ▁rebel
+- 挑戦
+- ▁এক
+- ▁humor
+- ▁improving
+- һа
+- いただいて
+- ваць
+- ▁fühlen
+- ▁cerca
+- āk
+- ▁ornament
+- ▁Wahrheit
+- 带到
+- 科技
+- 找到了
+- 真的很
+- euse
+- ▁purple
+- 聞き
+- 外面
+- ▁deshalb
+- ▁amid
+- ▁Mac
+- 疯狂
+- ления
+- 颜色
+- ▁clau
+- ுக்கு
+- ▁continua
+- 职业
+- ネット
+- 張
+- کرد
+- 浪
+- 的行为
+- kleid
+- ல்ல
+- ▁september
+- ▁Schiff
+- tava
+- 软件
+- 遠
+- 也不是
+- ▁Este
+- 思う
+- ▁confirm
+- tete
+- になると
+- 乌
+- istan
+- どれ
+- äre
+- 皇上
+- ▁oath
+- 愤怒
+- دل
+- 朗
+- ▁apparent
+- ồ
+- ▁khác
+- ▁shaking
+- ▁acquainted
+- ▁citizen
+- stav
+- ▁biz
+- 陸
+- 得很
+- 权力
+- ▁lieben
+- ▁fost
+- ▁badly
+- ▁King
+- ▁mankind
+- util
+- ▁هیچ
+- ҳ
+- tele
+- sucht
+- 戻
+- Da
+- ▁centuries
+- ようです
+- нар
+- ▁urged
+- чен
+- matic
+- ▁throwing
+- ▁outlook
+- ▁loves
+- ▁religi
+- を感じ
+- OK
+- gab
+- ▁Nähe
+- ▁issued
+- zioni
+- ▁đây
+- iriza
+- gesch
+- 裁
+- aud
+- ▁Universität
+- さっき
+- lika
+- ▁studio
+- рат
+- 聴
+- чин
+- 分からない
+- ▁зада
+- ▁afterward
+- lerini
+- ▁conta
+- ましたが
+- 逊
+- 트
+- ▁있는
+- ▁Tr
+- ▁illness
+- 見え
+- ▁alter
+- ▁sounded
+- posta
+- ▁Job
+- ▁significa
+- 下降
+- ▁третий
+- nyo
+- elen
+- ▁کنیم
+- avuga
+- ▁poem
+- ▁Sur
+- 变成了
+- ख
+- 再び
+- sili
+- 早就
+- نج
+- greg
+- 絶対
+- ▁First
+- trac
+- ▁photo
+- 当年
+- greifen
+- iĝas
+- ▁wären
+- ▁caso
+- ode
+- '45'
+- 能不能
+- Me
+- らしい
+- ▁enthusiasm
+- マー
+- ▁bene
+- 警戒
+- انه
+- ừ
+- krieg
+- ř
+- 爬
+- ▁شهر
+- ▁preserve
+- ▁Kur
+- 震
+- ▁spear
+- 浅
+- ▁bwo
+- ▁moeten
+- imba
+- ▁comparison
+- したこと
+- ▁câ
+- 食べて
+- 胖
+- 拡大
+- ▁право
+- ▁kw
+- ▁الخ
+- 蓝
+- 臣
+- 什
+- ▁recorded
+- ▁rational
+- ▁estimate
+- 你知道吗
+- ▁thinks
+- 見た
+- ▁arrest
+- ssy
+- ▁trend
+- 表面
+- eke
+- ▁durant
+- ▁contain
+- ▁Dy
+- ற்க
+- ைய
+- 一件事
+- 临
+- кә
+- sicher
+- 人都
+- ▁gathering
+- ▁transaction
+- coll
+- ▁explo
+- ▁maria
+- ovan
+- ▁当我
+- ног
+- 什么事
+- ला
+- 同事
+- ▁toch
+- 議
+- ▁vorher
+- 说是
+- ▁Wat
+- ▁genuine
+- リーグ
+- ckte
+- jahr
+- ▁madam
+- ध
+- مون
+- タイム
+- ▁sig
+- scribe
+- 购买
+- premi
+- Pa
+- ▁হ্যাঁ
+- ога
+- 识
+- ▁Fern
+- 日子
+- 斜
+- 분
+- ника
+- ▁hook
+- ania
+- ▁schwa
+- ▁Zeitpunkt
+- ▁год
+- ▁meadow
+- ▁charity
+- ▁نام
+- ▁column
+- ministr
+- ▁Kraft
+- ঁ
+- 的钱
+- ▁revealed
+- 宣
+- ▁fearful
+- ▁melancholy
+- ▁daher
+- 机器
+- ▁первая
+- まずは
+- vesti
+- 把自己
+- ▁ansehen
+- ▁lugar
+- ▁Geschäft
+- 交给
+- ▁leaf
+- зо
+- 腹
+- 虎
+- 'No'
+- ქ
+- ▁compliment
+- 但这
+- ▁trot
+- ▁lands
+- ▁Fuß
+- 報告
+- ொ
+- 巡
+- iš
+- ▁manners
+- جد
+- 治療
+- 等待
+- خط
+- ▁brow
+- ▁mock
+- овая
+- tico
+- 生き
+- ▁wet
+- 怎样
+- passen
+- lur
+- ▁Yuba
+- シー
+- 不在
+- embro
+- ধ
+- ▁zonder
+- ន
+- Ü
+- ▁craft
+- nji
+- 档
+- فی
+- য
+- ▁misfortune
+- kaba
+- вес
+- ക
+- flow
+- ▁Pero
+- さんです
+- アウト
+- чым
+- 过程中
+- ppel
+- これで
+- cce
+- вае
+- ▁alike
+- 阻止
+- しましょう
+- ▁blu
+- 嘉
+- ▁senior
+- ▁stellte
+- version
+- 药物
+- ▁然后我
+- 忘
+- вала
+- ▁gloom
+- 复杂
+- 的节目
+- ▁mejor
+- ča
+- wende
+- 演员
+- ▁borne
+- センター
+- ▁спа
+- ▁ebi
+- ▁requires
+- 杜
+- ▁پا
+- ▁fac
+- 医療
+- pher
+- 第二个
+- ▁switch
+- yne
+- ▁theatre
+- 作った
+- 信号
+- 体验
+- лам
+- 你需要
+- ▁sixth
+- ▁abandoned
+- ▁tipo
+- 類
+- ▁noted
+- 规
+- ▁initial
+- 上去
+- شار
+- ▁charged
+- 宇
+- 돼
+- ▁ciudad
+- 当たり
+- 为你
+- ▁cares
+- އ
+- 活着
+- ▁encourage
+- 門
+- tures
+- 간
+- ▁Би
+- ▁referred
+- erde
+- ▁leaned
+- дом
+- ▁Blut
+- ▁шо
+- 非常感谢
+- انت
+- 画面
+- ▁marks
+- ▁flock
+- ะ
+- 以下
+- ▁công
+- ▁saß
+- ▁originally
+- ▁sowie
+- kti
+- ▁encouraged
+- cm
+- psi
+- ▁жизнь
+- ▁hunger
+- ता
+- 我今天
+- 忠
+- stel
+- уп
+- 自信
+- قى
+- ▁volen
+- acco
+- ▁acu
+- ▁яшчэ
+- ▁而是
+- lation
+- aks
+- nish
+- angan
+- 芸
+- гад
+- 할
+- ям
+- ี
+- ogni
+- ▁hoping
+- 約
+- 紙
+- 出版
+- さんも
+- ▁Hat
+- ▁likewise
+- 치
+- ▁contempt
+- ▁Seiten
+- ▁свою
+- 我我
+- ▁wanting
+- ไ
+- య
+- anze
+- こそ
+- ▁operational
+- вести
+- سر
+- ▁gaan
+- کس
+- maß
+- 民族
+- 仅仅
+- 阿姨
+- فل
+- ▁тысяч
+- rò
+- DNA
+- 工程
+- ▁stout
+- 毎日
+- ▁senses
+- 研
+- dil
+- ▁fuhr
+- aca
+- 担当
+- ▁говорить
+- ▁dispute
+- rgi
+- ▁landscape
+- tima
+- ▁الذي
+- сте
+- ▁brick
+- بول
+- jer
+- َا
+- ▁examples
+- 对我们
+- ▁Hin
+- ▁großartig
+- ▁gir
+- ▁engineering
+- ▁cau
+- ▁monk
+- mona
+- ځ
+- 上班
+- 作者
+- ▁má
+- ▁الط
+- зов
+- 伝
+- ▁chỉ
+- dern
+- ▁provisions
+- '28'
+- ales
+- ▁Bal
+- 创建
+- atory
+- ▁introduce
+- ▁назад
+- bart
+- 好きな
+- mother
+- üz
+- ▁engage
+- гон
+- лад
+- ▁champion
+- 子的
+- 食品
+- 見える
+- SNS
+- ▁inferior
+- ▁gratitude
+- 九州
+- ▁trained
+- ▁mensen
+- 可怕的
+- 闲
+- 走吧
+- stern
+- view
+- 玛
+- 开了
+- 単
+- ▁finance
+- ▁Gebäude
+- führung
+- 是这样
+- の影響で
+- すでに
+- 的关系
+- ▁personas
+- ▁bride
+- ▁occasions
+- 自分で
+- ▁priests
+- と同じ
+- 搬
+- imiento
+- сем
+- ▁То
+- 呗
+- iche
+- used
+- стаў
+- ▁chicken
+- ▁Straße
+- 好处
+- 是非常
+- hom
+- 公司的
+- away
+- ▁temperature
+- ▁sob
+- ▁darling
+- ▁knock
+- tick
+- 术
+- 甜
+- franc
+- ▁гол
+- ▁Quan
+- based
+- 不得
+- しっかりと
+- ▁artificial
+- sburg
+- メディア
+- 大きい
+- නි
+- 增
+- ▁आ
+- られている
+- өөр
+- tät
+- 很大
+- ▁todas
+- 工作的
+- aş
+- 听说
+- 所说的
+- vali
+- schied
+- ▁objective
+- سو
+- 区域
+- じゃなくて
+- avait
+- likuwa
+- 鹿
+- ▁laut
+- ではなく
+- ▁monta
+- ▁swell
+- 主席
+- ▁anderer
+- ▁ellos
+- 是吗
+- ▁determine
+- 的第一
+- ▁Familien
+- ▁threatened
+- のも
+- 灰
+- 我的意思是
+- ▁شو
+- ▁utmost
+- miz
+- ▁nacht
+- ▁arts
+- 束
+- ▁commit
+- ▁backward
+- ▁efficiency
+- ▁surrender
+- ebe
+- とともに
+- نده
+- 通知
+- ەر
+- 这些人
+- baha
+- ▁sonst
+- git
+- vale
+- ▁зам
+- iness
+- ▁Preis
+- お伝え
+- ▁imagined
+- 内で
+- ▁visual
+- 의
+- ▁Bedeutung
+- んだろう
+- まった
+- чат
+- wang
+- ▁probable
+- 他也
+- ▁conclude
+- ▁Ili
+- 財
+- 設
+- वि
+- ots
+- dium
+- eerd
+- ▁employment
+- 的女人
+- ຮ
+- 梁
+- 変わって
+- 民主
+- quent
+- 十六
+- uge
+- ▁retain
+- ச்ச
+- ▁monarch
+- ▁receiving
+- かん
+- utse
+- ▁dreißig
+- ▁Anne
+- ▁employ
+- оз
+- ▁attempted
+- ▁july
+- ▁nào
+- tief
+- すぐに
+- ēja
+- ▁draußen
+- ▁könig
+- 我们也
+- رج
+- ▁detective
+- kup
+- ▁geç
+- そのまま
+- ぶり
+- வர
+- ▁childhood
+- кал
+- ▁ciutat
+- ении
+- ▁bloom
+- かも
+- ▁wilt
+- koj
+- ▁poured
+- んの
+- ▁hinzu
+- 節
+- ager
+- 附近
+- ▁femme
+- 才是
+- cco
+- ▁ramp
+- によると
+- buka
+- ▁уг
+- ▁eki
+- ▁monument
+- ▁contains
+- 公众
+- öz
+- ▁items
+- Mo
+- поль
+- ▁kleiner
+- 从来没有
+- が入って
+- ▁instruments
+- ▁thine
+- ▁essen
+- 侧
+- habit
+- 桃
+- をお
+- ▁suchen
+- 敏
+- ▁Region
+- shin
+- تل
+- ▁Tiu
+- 尚
+- 葉
+- ators
+- наход
+- డ
+- ▁могут
+- zik
+- ステ
+- 态
+- wagen
+- ▁recognition
+- mez
+- フィ
+- ▁safely
+- ▁Jack
+- ija
+- ▁телефон
+- овка
+- ▁commitment
+- 跟你说
+- ▁wage
+- зда
+- සි
+- тын
+- 好不好
+- ▁мало
+- ▁alleen
+- ▁பு
+- ▁hasn
+- スの
+- ってきた
+- 悪い
+- ▁relatively
+- ▁Mur
+- ▁Bru
+- ▁relationships
+- ▁rat
+- るように
+- ▁Doch
+- ▁noon
+- 同样的
+- ▁integr
+- 简单的
+- ▁seal
+- гляд
+- ▁Uhr
+- 我怎么
+- ▁dining
+- ▁gebe
+- culture
+- 師
+- 他不
+- ▁kot
+- ▁huis
+- 帮你
+- 大家都
+- ▁solve
+- fur
+- ▁poll
+- ▁jamais
+- meri
+- ▁cared
+- bonne
+- 思想
+- 数量
+- 一半
+- ▁furniture
+- ▁territory
+- ▁verbunden
+- gaan
+- Imana
+- herr
+- ▁estar
+- ▁revis
+- hale
+- ▁Kü
+- ▁lucky
+- 账
+- ▁pursued
+- の皆さん
+- っていうのが
+- ▁mondo
+- 瞬間
+- スポーツ
+- dě
+- ▁ugly
+- 秦
+- 資
+- ▁ولكن
+- ▁gering
+- пас
+- 始めた
+- ▁cries
+- ありがとう
+- 裂
+- ▁amor
+- alis
+- haber
+- ë
+- 巨大
+- гер
+- たの
+- kee
+- وە
+- ▁groot
+- ▁प्र
+- цев
+- 新たに
+- 谱
+- ары
+- نفس
+- ▁ваш
+- ▁кры
+- maal
+- ▁mum
+- iter
+- ▁mig
+- bá
+- 盟
+- 向け
+- ▁betty
+- ▁никогда
+- ▁кү
+- гэ
+- どの
+- ブラ
+- 北部
+- йн
+- ▁peasant
+- gren
+- 很容易
+- 下さい
+- ไม่
+- ▁Ek
+- ▁Are
+- friend
+- ▁invisible
+- evi
+- ▁climb
+- lek
+- rika
+- tua
+- ▁terug
+- эк
+- ▁mijnheer
+- ▁decir
+- ▁Mill
+- とおり
+- ▁studi
+- ▁shield
+- 伸
+- assi
+- ▁award
+- 需
+- pira
+- 禁止
+- ▁autor
+- ▁contre
+- ▁entend
+- 给他们
+- ▁hugh
+- '!"'
+- ▁passa
+- ▁erwähnt
+- 公开
+- ▁zweiten
+- iau
+- ▁Ok
+- mitten
+- ৈ
+- 喊
+- dle
+- 来る
+- inu
+- ▁ئە
+- ▁freely
+- 汤
+- なんですよ
+- ▁nuestro
+- ▁couch
+- ▁mistaken
+- '400'
+- ▁expecting
+- 攻击
+- ▁landing
+- 你对
+- 的就是
+- ▁größte
+- ▁那就
+- 这对
+- 点击
+- ried
+- ▁bowl
+- 裏
+- და
+- kah
+- بط
+- face
+- lekt
+- ▁urbo
+- ▁solchen
+- 距离
+- вели
+- ▁drama
+- をした
+- chip
+- اص
+- mouth
+- ▁statt
+- antwort
+- kho
+- โ
+- 她是
+- ▁literary
+- ▁بیا
+- fassen
+- ▁michael
+- ▁nights
+- ▁jag
+- 都可以
+- ▁Geschichten
+- дын
+- 出来的
+- yê
+- 模型
+- ▁آیا
+- ▁пятнадцать
+- ▁belle
+- ピー
+- ▁Stra
+- бит
+- ▁merchant
+- ▁russia
+- 骂
+- ▁Amb
+- 今は
+- youtube
+- 수
+- ▁grade
+- ▁ɣer
+- ც
+- 撒
+- ▁pipeline
+- 聚
+- ▁Möglichkeiten
+- たのは
+- ▁julia
+- 容疑者は
+- 微信
+- ▁얘기
+- くなって
+- ▁umuntu
+- sprung
+- などに
+- খা
+- ▁vorstellen
+- ▁decent
+- 動か
+- もいい
+- 的力量
+- トン
+- ▁acted
+- ▁Personen
+- 恐
+- 大型
+- ▁reproach
+- エン
+- ▁nhà
+- ▁Heute
+- に対する
+- マーク
+- ▁Street
+- ▁toutes
+- ▁hinein
+- ▁двенадцать
+- даа
+- ▁ब
+- ▁Ah
+- ера
+- ▁kama
+- ▁hob
+- ▁abzu
+- 黑暗
+- ▁hated
+- ▁steal
+- ▁더
+- ▁yıl
+- ▁事实上
+- volu
+- ▁powder
+- Ch
+- klu
+- 针
+- ং
+- ▁Entscheidung
+- ▁mail
+- 话题
+- ▁otra
+- rir
+- ▁rhy
+- ▁сразу
+- 你可能
+- cija
+- там
+- dow
+- 摸
+- rite
+- ienne
+- 感染者
+- play
+- lick
+- ▁denk
+- ▁están
+- 若い
+- ▁verändert
+- チャンス
+- 红色
+- ▁bade
+- ây
+- mt
+- nych
+- 世代
+- ▁johnson
+- ▁fears
+- som
+- 什么呢
+- 去做
+- ▁prompt
+- 最終
+- ▁capabilities
+- ுவ
+- ▁Anti
+- ▁harder
+- 在那
+- ides
+- ▁examination
+- ▁invitation
+- ▁Erfolg
+- 小的
+- 定义
+- 测
+- ▁això
+- 住了
+- 赫
+- ▁авто
+- ▁apt
+- ▁Roman
+- wahr
+- hard
+- ▁irà
+- ▁toute
+- 会場
+- ▁spectacle
+- 個人
+- 德国
+- حد
+- 别的
+- けて
+- ▁näher
+- mpe
+- 用意
+- ▁کی
+- ▁bedroom
+- ▁puc
+- ▁nascut
+- 一日
+- ▁30
+- ▁meantime
+- ということは
+- ▁Charles
+- ▁shareholders
+- 明白了
+- 국
+- 故
+- ыз
+- තු
+- 十一
+- usta
+- ೆ
+- ▁parla
+- اخت
+- ▁inches
+- power
+- 的这种
+- ずに
+- berry
+- 品牌
+- 引起
+- ▁historia
+- ▁pret
+- ▁fantastic
+- 威胁
+- kru
+- 当他们
+- ؤ
+- esi
+- ▁occupation
+- rose
+- ር
+- ▁statue
+- 決定
+- ясн
+- ો
+- ▁Ton
+- ▁blij
+- uɣ
+- ogo
+- ▁favorable
+- 我从
+- ▁outline
+- atur
+- んじゃ
+- ▁বি
+- ▁conception
+- ▁comun
+- 大谷
+- меш
+- 었
+- ▁session
+- ▁fed
+- 今まで
+- ▁ridiculous
+- 妹妹
+- бр
+- ▁maig
+- フランス
+- ▁2017
+- ترین
+- ▁глаза
+- ▁её
+- ▁Computer
+- ▁vanished
+- kauf
+- ▁Unterstützung
+- ▁gukora
+- 做的事情
+- 业务
+- ▁schm
+- 러
+- ▁gotten
+- ▁pār
+- ▁rece
+- ▁zuerst
+- கா
+- 猎
+- ▁vergessen
+- ▁weet
+- ný
+- ▁Japan
+- ▁pointing
+- ▁estos
+- ▁unconscious
+- だけど
+- னா
+- ▁permission
+- هایی
+- вел
+- 一定是
+- 头发
+- ине
+- 悔
+- ▁Zum
+- ッシュ
+- 杀了
+- 様子
+- ▁blush
+- ▁sharply
+- 做得
+- ▁befindet
+- ▁cyf
+- ▁football
+- ▁alguna
+- пов
+- 泥
+- ghi
+- 十八
+- ▁treball
+- ▁inspired
+- ▁pattern
+- ▁crystal
+- 効
+- ▁cultural
+- ▁monster
+- ▁Bücher
+- ▁என்ன
+- сел
+- ▁pomp
+- ▁これで
+- 記
+- eyed
+- mati
+- トリ
+- boo
+- 庭
+- commerce
+- pru
+- ▁records
+- 取得
+- culo
+- ▁Mau
+- ようと
+- ▁Ban
+- 特殊
+- ฉัน
+- 采访
+- дра
+- ош
+- イベント
+- 发出
+- ▁prayers
+- ▁employees
+- klo
+- count
+- ▁panel
+- ▁crossing
+- 擦
+- 这是我们
+- ▁Frank
+- ▁kinda
+- ghan
+- reichen
+- ▁shone
+- có
+- ▁tutti
+- 真实
+- ▁avenue
+- ちょうど
+- kultur
+- 在我的
+- ▁iyo
+- ▁fetch
+- essen
+- ▁detect
+- ▁England
+- ▁село
+- ▁난
+- ▁reaction
+- 特定
+- 采
+- person
+- 惊讶
+- ▁strangers
+- ▁gemeinsam
+- тыр
+- ▁beard
+- ▁gleiche
+- ▁Africa
+- 周り
+- ▁сколько
+- ▁steadily
+- voca
+- 两年
+- zira
+- ▁startled
+- ▁muttered
+- 寄せ
+- 你应该
+- 策
+- ▁jud
+- 在他们
+- 闪
+- 개
+- ▁millor
+- ▁vein
+- に向けて
+- 香港
+- 播放
+- ▁accurate
+- 扬
+- 살
+- ▁Website
+- ফ
+- 覚
+- ▁whirl
+- よし
+- ope
+- ລ
+- 英雄
+- ▁我会
+- ▁gek
+- фі
+- 是怎么
+- ▁rap
+- ますか
+- ▁attacked
+- ▁endure
+- ▁tie
+- يع
+- rp
+- rash
+- ▁specifically
+- ▁collected
+- nahme
+- uous
+- න්න
+- dru
+- 稳定
+- oid
+- と思った
+- ▁каб
+- how
+- ▁以及
+- ▁anchor
+- 行き
+- ▁pia
+- ingham
+- ▁также
+- 礼物
+- ▁Typ
+- 理想
+- ▁lleva
+- ないように
+- 角色
+- uck
+- ી
+- ▁abstract
+- ▁Steve
+- Pro
+- ▁secured
+- ▁бес
+- mist
+- 生意
+- ▁consequently
+- ▁vriend
+- 的世界
+- ▁führt
+- ໄ
+- ▁railroad
+- ▁ticket
+- igis
+- ▁Ji
+- 道德
+- ▁disgrace
+- 소
+- ▁compare
+- 我必须
+- ▁wandering
+- ▁knocked
+- ▁tant
+- liter
+- ▁procure
+- ▁oedd
+- media
+- ▁offensichtlich
+- 都市
+- fed
+- ▁続いて
+- کار
+- ▁resistance
+- ▁ئى
+- ▁menos
+- ▁maken
+- 空气
+- 제
+- stock
+- 負け
+- ▁disappointed
+- fee
+- 選手の
+- ▁admire
+- ่า
+- ▁marie
+- 立即
+- 質問
+- ▁directions
+- eron
+- ▁Jean
+- ясь
+- يات
+- კ
+- 汁
+- 兴奋
+- ▁egg
+- كا
+- ▁erinnern
+- ▁Polizei
+- ய்
+- ▁È
+- 吸引
+- iere
+- ຍ
+- ▁dip
+- ▁starts
+- ▁свой
+- cir
+- ক্ষ
+- 兴趣
+- ्य
+- fekt
+- ▁Marc
+- ▁coward
+- 匹
+- ▁உள்ள
+- ▁dalla
+- とした
+- ▁brutal
+- ▁traf
+- holen
+- 肥
+- gian
+- 鲜
+- 课程
+- 会儿
+- ▁resumed
+- ຕ
+- ▁территория
+- schul
+- ▁complet
+- ▁sighed
+- 限制
+- ▁elsewhere
+- ança
+- 质量
+- ノー
+- ▁Versuch
+- образ
+- kwe
+- ▁virus
+- 天気
+- 筋
+- 的书
+- ▁timing
+- 熟悉
+- 促
+- ▁дома
+- 对了
+- ▁Damit
+- 我给你
+- ▁Bel
+- ối
+- ▁Chance
+- জা
+- ルー
+- 快点
+- これも
+- anca
+- ▁frau
+- 周辺
+- ▁pierre
+- 干嘛
+- ▁فکر
+- 教堂
+- ▁確かに
+- 香り
+- 道理
+- agit
+- 有多少
+- 指示
+- ம்ப
+- ▁patch
+- スーパー
+- ▁tonight
+- ▁Markt
+- رم
+- ▁Regel
+- migrant
+- miş
+- sió
+- 巧
+- kha
+- Ha
+- mination
+- ▁astonishment
+- ▁asia
+- má
+- kota
+- дээ
+- ▁Fehler
+- ▁Dieses
+- ▁persone
+- cant
+- ▁aquel
+- 消防
+- ▁perfection
+- cció
+- rren
+- 门口
+- fest
+- ▁viz
+- 粗
+- ▁nonsense
+- schätz
+- ができる
+- 程序
+- ▁говорил
+- ▁Clo
+- siz
+- ▁identify
+- ós
+- ▁omu
+- 祝
+- 続き
+- ▁род
+- ock
+- 表現
+- ▁slightest
+- aĵo
+- 豊
+- ▁пожалуйста
+- ски
+- ▁Lehrer
+- ▁erfolgreich
+- moor
+- பு
+- ▁players
+- ▁vì
+- ▁descended
+- qual
+- xo
+- ▁hopeless
+- ▁conducted
+- 手术
+- ▁Ay
+- ▁12
+- っちゃう
+- 生活的
+- ▁dragged
+- ▁بعض
+- нет
+- ▁inte
+- мон
+- ▁teu
+- тө
+- ▁liquid
+- ▁hip
+- find
+- moni
+- ▁State
+- Е
+- ▁shooting
+- ▁legend
+- 埋
+- ▁revenues
+- ▁unglaublich
+- abel
+- слав
+- ▁”
+- 听起来
+- 所有人
+- ▁düşün
+- 了他的
+- 残り
+- fro
+- ▁bend
+- ▁wirst
+- ▁akk
+- ▁hitherto
+- 成長
+- こうやって
+- ▁dirty
+- metri
+- 分の
+- ▁perceive
+- ▁jur
+- vra
+- خا
+- ▁năm
+- 晴れ
+- 酷
+- ▁poste
+- лер
+- lais
+- ▁sav
+- 认真
+- 毕竟
+- with
+- ▁regional
+- ▁scholar
+- 苹果
+- 改めて
+- 因素
+- டா
+- кы
+- ▁sembla
+- zustellen
+- ▁pupil
+- ifying
+- 缓
+- 手に
+- 名前
+- ▁совсем
+- 相談
+- ▁ал
+- ▁kendi
+- ▁promote
+- ▁стороны
+- 了我们
+- 发生的
+- прост
+- ▁mateix
+- ▁дума
+- ково
+- ▁Green
+- 自动
+- lima
+- もらう
+- ели
+- ▁departed
+- ▁belonging
+- ▁Star
+- وز
+- テーマ
+- யின்
+- ▁hesitated
+- ▁Om
+- 评
+- ▁Hoch
+- 正确
+- 協力
+- ikan
+- はどう
+- ▁lane
+- machen
+- reba
+- ▁ব
+- 逮捕
+- 在美国
+- ▁гру
+- osi
+- '1000'
+- 泉
+- ▁examined
+- 方は
+- ▁disco
+- ▁Ale
+- ▁terwijl
+- ▁dig
+- 避免
+- ▁leisure
+- worm
+- ▁depth
+- 卢
+- ▁worried
+- 做好
+- 出身
+- 生まれ
+- ▁قرار
+- 废
+- ám
+- zzo
+- culp
+- ▁Gebiet
+- ibya
+- ач
+- ▁scenes
+- 王子
+- gelegen
+- 基于
+- ▁partie
+- lhe
+- ▁belt
+- রি
+- änder
+- ▁compassion
+- ♪
+- 下了
+- ями
+- ▁devotion
+- ろう
+- 他の
+- ケット
+- ▁monitor
+- ▁häufig
+- 検討
+- tö
+- enta
+- ▁sido
+- 補
+- 聊天
+- 我一直
+- ▁гэтага
+- ▁называ
+- 蒸
+- ▁当你
+- shyi
+- houden
+- conc
+- ▁device
+- この日
+- 凉
+- ▁Del
+- mica
+- 剂
+- ドイツ
+- 是要
+- 隣
+- ▁narrative
+- ▁Medien
+- ▁francis
+- Р
+- ▁beaucoup
+- ▁whistle
+- 她在
+- ▁kumu
+- ▁droit
+- 我们正在
+- nahm
+- etta
+- ▁fille
+- Do
+- ▁owned
+- 通り
+- もらって
+- 写的
+- 疼
+- ל
+- ām
+- пра
+- них
+- gali
+- shya
+- ▁France
+- ▁cord
+- だし
+- rab
+- ▁estava
+- ▁folgen
+- 遺
+- 位于
+- ▁sä
+- ▁Sport
+- ▁kui
+- 页
+- の方が
+- nata
+- ▁facing
+- ▁பொ
+- もない
+- He
+- ▁inventory
+- 实验
+- uter
+- raten
+- Sa
+- ▁drank
+- قی
+- 伺
+- ▁Barcelona
+- ▁keeps
+- ▁하고
+- ▁mult
+- gni
+- 금
+- ▁astonished
+- ▁readily
+- ▁unexpected
+- ▁mug
+- どうして
+- ҙә
+- 跡
+- еб
+- ▁Hill
+- 面临
+- onia
+- ▁cinema
+- čas
+- 创业
+- ▁зап
+- rücken
+- ▁silly
+- ▁facility
+- ▁除了
+- 用于
+- ▁Monsieur
+- ばかり
+- ▁collect
+- ▁انجام
+- ▁gifts
+- ▁doctors
+- ▁emotions
+- ▁republican
+- ▁maintained
+- ▁snake
+- ▁سي
+- 身份
+- ▁ней
+- ▁models
+- ▁gracious
+- ibwa
+- leigh
+- بت
+- ▁meines
+- 警察は
+- stall
+- ▁percentage
+- خوا
+- ▁crop
+- ▁ainsi
+- 忘了
+- cide
+- ▁killing
+- ▁tool
+- 的好
+- teria
+- ▁mutual
+- 剩
+- ▁wreck
+- ▁chiefly
+- ▁values
+- ▁strengthen
+- ▁cómo
+- 같
+- 的能力
+- ▁wunderbar
+- steigen
+- ▁heir
+- ▁2018
+- кат
+- zī
+- ▁அது
+- ▁கூ
+- breite
+- ▁guter
+- ▁duck
+- 相对
+- 承诺
+- 叔叔
+- ▁bana
+- 使う
+- زار
+- 组成
+- lö
+- としても
+- kap
+- 操作
+- 物质
+- Ba
+- 入って
+- halte
+- ▁tau
+- ▁weapon
+- ▁байгаа
+- rok
+- ▁graduate
+- 心情
+- ブル
+- ▁höher
+- leu
+- するため
+- ▁vai
+- ころ
+- 五年
+- ▁landed
+- 闭
+- ▁след
+- ▁selected
+- 放心
+- 秘
+- 等着
+- nol
+- gling
+- ▁对于
+- ▁lend
+- ▁povas
+- 券
+- ▁sources
+- ▁Kampf
+- はある
+- ▁Gro
+- ▁jemanden
+- 要做
+- ▁peaceful
+- як
+- 规定
+- dolph
+- 追求
+- univers
+- 添加
+- ▁pump
+- バン
+- 特徴
+- 多年
+- easing
+- 枝
+- ▁Louis
+- ▁Rob
+- 改革
+- 一本
+- ▁climbed
+- ▁bueno
+- ▁analysis
+- важ
+- シン
+- ▁roger
+- sein
+- ▁riesige
+- organ
+- ▁plat
+- ▁seventh
+- imper
+- ика
+- ニー
+- ▁breeze
+- ▁slavery
+- モン
+- ▁Seg
+- cru
+- 这个时候
+- zte
+- nici
+- ▁grandmother
+- 这是一种
+- 最後に
+- 厉害
+- ề
+- ▁usted
+- ▁split
+- 的儿子
+- 速度
+- gers
+- ▁heavens
+- شان
+- ▁planta
+- 奇怪
+- ▁Sed
+- ▁yüz
+- дай
+- ▁saturday
+- うれしい
+- 奇怪的
+- 联邦
+- جو
+- ydi
+- ▁cyn
+- ▁зах
+- üste
+- agne
+- мол
+- ায়
+- ▁bwe
+- бира
+- toria
+- 尖
+- 好多
+- undu
+- ▁urban
+- гч
+- ыч
+- 穴
+- ▁hiding
+- ▁loan
+- ແ
+- ▁kant
+- ▁complain
+- ▁apa
+- ▁точно
+- 抜け
+- 建物
+- ▁Win
+- يب
+- 재
+- 政府は
+- ▁Professor
+- ▁naked
+- 期間
+- ▁essay
+- 翔
+- ▁propose
+- ▁circumstance
+- めた
+- 不见
+- tour
+- 使われ
+- гар
+- 我妈
+- ▁zuvor
+- 坚
+- mik
+- 财
+- ▁museum
+- ▁건
+- ▁engineer
+- 日から
+- uß
+- ▁solar
+- nın
+- ▁gesamte
+- ▁后来
+- となって
+- 写了
+- ▁agreeable
+- ▁canoe
+- ▁rug
+- ▁Hor
+- جان
+- 什么呀
+- ▁marble
+- meister
+- ەت
+- ▁dorothy
+- ▁confused
+- ноў
+- лат
+- ffi
+- née
+- 否则
+- Applause
+- ண்டு
+- ▁complicated
+- ▁guerra
+- キー
+- 你不是
+- ▁最后
+- class
+- ▁swiftly
+- lton
+- 滴
+- ▁vollständig
+- ▁будем
+- ▁gleam
+- ▁finest
+- ザー
+- ▁fiel
+- ▁gusa
+- ▁motive
+- 時の
+- ▁circ
+- timo
+- ▁こんにちは
+- rated
+- هی
+- ect
+- ▁moins
+- ▁rested
+- fau
+- dden
+- elijke
+- 鼓
+- ough
+- kub
+- ▁quart
+- ຫ
+- 菌
+- skri
+- まれ
+- ▁Juli
+- جه
+- ▁dorm
+- ▁selfish
+- 混乱
+- ▁佢
+- なんですか
+- ▁examine
+- ▁insurance
+- гай
+- 一开始
+- ▁disturb
+- cción
+- 政権
+- 伏
+- 征
+- ▁Foto
+- ▁stati
+- 讯
+- xed
+- ▁chemical
+- ром
+- ships
+- ligen
+- ▁halb
+- ▁introduction
+- ▁дальше
+- ▁dish
+- 联合
+- posa
+- 积
+- 基因
+- ይ
+- 商店
+- 粒
+- 不去
+- ▁Fran
+- 彼此
+- ▁nieder
+- ▁mache
+- 都要
+- три
+- 気になる
+- ▁основ
+- ▁achieved
+- ▁نظر
+- 込む
+- زم
+- 高的
+- Di
+- 使い
+- 某些
+- ▁Joe
+- пам
+- ▁cloak
+- 投入
+- ▁captured
+- 週
+- 就这样
+- сс
+- 所谓的
+- ▁arī
+- ▁spoil
+- nuncia
+- ▁ɣef
+- word
+- ▁materially
+- ▁thực
+- ▁öffentlichen
+- ▁яв
+- gum
+- ▁ئۇ
+- ▁quote
+- ▁indicated
+- ▁stor
+- tare
+- ▁kaufen
+- ▁google
+- ▁unterstützen
+- 对自己
+- ▁wichtige
+- ▁recovery
+- mensch
+- ▁Ehe
+- ▁lessons
+- neɣ
+- 던
+- 怎么回事
+- octubre
+- 頑張って
+- أكثر
+- ▁testing
+- އި
+- ști
+- dır
+- ▁byose
+- ▁polite
+- рова
+- imana
+- ▁exhausted
+- ▁rivers
+- を行う
+- 帝国
+- oval
+- ▁例えば
+- ▁Rom
+- 这不
+- ▁Yn
+- ▁patent
+- 動画
+- рас
+- ▁mole
+- 不必
+- cura
+- ▁anh
+- ▁überall
+- 美国人
+- ▁disappointment
+- ▁soup
+- ktu
+- ▁transformation
+- ▁Verhalten
+- 後半
+- ▁байдаг
+- 合う
+- ▁Sil
+- ぜ
+- 扎
+- trud
+- ży
+- ▁shepherd
+- ▁возможно
+- レス
+- cier
+- ▁sunshine
+- ▁Telefon
+- 愉快
+- ▁havia
+- aşı
+- のだ
+- 进行了
+- 农
+- ▁continuous
+- ▁públic
+- ▁poz
+- ▁давай
+- 满意
+- 悲伤
+- lma
+- ▁curse
+- reb
+- 空気
+- قال
+- ▁player
+- 安定
+- maker
+- 的那样
+- ▁ntabwo
+- ▁способ
+- heben
+- ▁hecho
+- wide
+- Ho
+- then
+- ありました
+- ▁pendant
+- ▁咁
+- 但它
+- 能量
+- ▁Himmel
+- 十九
+- 有些人
+- 年的
+- ▁свои
+- ▁practically
+- ость
+- ▁deserted
+- 说他
+- ▁buying
+- ▁thorn
+- ▁holds
+- 信仰
+- lings
+- ▁glimpse
+- ▁CA
+- ▁Kal
+- slav
+- ҙа
+- 思います
+- 何も
+- ँ
+- regen
+- ッチ
+- druk
+- قص
+- ▁preferred
+- 续
+- بة
+- ▁мин
+- fälle
+- ▁travail
+- ▁他是
+- ழு
+- ▁matt
+- ít
+- eği
+- 勝ち
+- মি
+- ▁ваше
+- するのは
+- ෝ
+- 出した
+- ▁bite
+- 黄色
+- مال
+- ▁preparing
+- ▁há
+- ử
+- digit
+- ▁World
+- かせ
+- ieron
+- 答
+- 遇
+- カル
+- ファ
+- بن
+- 화
+- ▁priv
+- kaza
+- ▁giờ
+- ப்பட்ட
+- ▁bé
+- ▁چند
+- kut
+- ▁darf
+- ▁دوست
+- 具有
+- fes
+- еть
+- ▁Cette
+- konstru
+- ▁thứ
+- ▁jako
+- ছিল
+- organisation
+- ▁Essen
+- ▁surprising
+- 渴望
+- 资
+- 浦
+- isieren
+- ▁байв
+- ▁Ideen
+- స
+- ▁größer
+- besitz
+- bour
+- ән
+- werfen
+- ▁Ebene
+- dah
+- 美丽
+- 监
+- ドラマ
+- じゃん
+- ▁калі
+- 独自
+- 時期
+- ▁cose
+- ▁چیزی
+- ▁joining
+- ▁confession
+- 天天
+- ▁processes
+- 보
+- ▁比如说
+- 也就是
+- ーム
+- 是我的
+- ▁Stil
+- スピード
+- 少年
+- ▁fühlte
+- 予
+- vig
+- ики
+- ▁nap
+- ▁angels
+- ▁schedule
+- rán
+- ▁boston
+- ▁oude
+- まま
+- душ
+- 腾
+- ху
+- 咖啡
+- валі
+- ▁лес
+- 刷
+- ▁suspected
+- ▁inquiry
+- ▁amusement
+- ordnung
+- ▁Klasse
+- 食べる
+- 我记得
+- 连接
+- فق
+- ▁precisely
+- ープ
+- ▁왜
+- ▁Thomas
+- 場合は
+- schlag
+- னர்
+- ▁instructions
+- turi
+- ▁passieren
+- ▁medium
+- 姓
+- ▁укра
+- にする
+- wasser
+- ▁cigar
+- ▁pessoas
+- ▁thirst
+- ▁gö
+- がこの
+- ▁norman
+- geni
+- 拜
+- rina
+- اک
+- 出现了
+- 纯
+- ffin
+- 忘れ
+- 戦い
+- 效果
+- こういった
+- kõ
+- ▁مق
+- ▁روز
+- 妙
+- ▁jones
+- ▁During
+- 碰
+- 这条
+- っちゃ
+- рон
+- pack
+- gina
+- が起き
+- ▁sins
+- gelassen
+- prä
+- ▁Wand
+- istes
+- plat
+- ▁telephone
+- conf
+- 桌
+- ▁shine
+- 不知
+- جل
+- ▁Red
+- ▁pearl
+- ▁prec
+- 家の
+- ▁Süd
+- ▁într
+- 牧
+- agu
+- お話
+- ▁sym
+- 意見
+- case
+- ▁2016
+- ▁tenía
+- ▁Netz
+- lusion
+- ▁sola
+- なくなって
+- 克斯
+- 习
+- енно
+- ▁lucy
+- 日間
+- 단
+- መ
+- 肺
+- Sch
+- 文章
+- ▁festival
+- нае
+- дам
+- ppy
+- ▁glaubt
+- ило
+- 漏
+- सा
+- は何
+- rico
+- 吃了
+- ▁première
+- 移動
+- ▁River
+- 穷
+- ▁sensitive
+- 今週
+- ▁provides
+- нуть
+- 今度は
+- 获
+- ▁attract
+- ▁durante
+- 桑
+- ▁கு
+- を作って
+- volv
+- ື
+- ▁armen
+- 現
+- abwe
+- 세
+- ých
+- 隊
+- chet
+- ▁euro
+- kula
+- ▁hoofd
+- ▁visitor
+- ▁nooit
+- ятся
+- にした
+- ▁August
+- 每年
+- ▁هست
+- ▁colors
+- ▁buri
+- ▁definitiv
+- ▁Wu
+- かない
+- eno
+- গে
+- ▁لكن
+- egi
+- ▁своей
+- ▁Jordi
+- pfen
+- ▁solicit
+- ▁goods
+- ▁composition
+- gine
+- 第二天
+- までの
+- ▁intellect
+- ▁vow
+- 现在是
+- ▁frown
+- ▁furnished
+- ▁обе
+- য়া
+- 收到
+- ▁sicherlich
+- ▁setzte
+- ▁harsh
+- 専門家
+- ▁Тэр
+- 社長
+- 的一切
+- ハー
+- 迅速
+- hend
+- られています
+- 毁
+- ന
+- 順
+- ▁lunga
+- 最初の
+- anta
+- ▁expectation
+- ▁أي
+- ью
+- ▁дээр
+- ▁মা
+- udo
+- ▁jail
+- 计算
+- ▁나도
+- お父さん
+- fä
+- '*'
+- ▁Studenten
+- ▁Gut
+- ▁你可以
+- 合同
+- рус
+- zam
+- rul
+- 帅
+- ▁Why
+- ▁moderate
+- 块钱
+- හි
+- alia
+- ▁timid
+- ▁cigarette
+- 没关系
+- ologist
+- rufen
+- legi
+- دید
+- ▁முடி
+- 敷
+- 方も
+- 刺激
+- 误
+- ▁fortunate
+- ▁crow
+- kehr
+- ーション
+- ▁nut
+- 你认为
+- frei
+- ▁mình
+- ▁existed
+- 烈
+- lij
+- agost
+- ▁partnership
+- ู
+- ▁ده
+- ▁così
+- ▁hebt
+- fire
+- ▁país
+- druck
+- Ә
+- ▁demon
+- ▁improvements
+- 这样一个
+- ▁jealous
+- 化学
+- 連絡
+- 医学
+- huit
+- ▁transfer
+- گیر
+- ▁favourite
+- 强烈
+- お伝えします
+- 卫生
+- ▁jerusalem
+- ঘ
+- ارت
+- 化的
+- ▁parallel
+- ções
+- 概念
+- 我们知道
+- anu
+- したと
+- ▁我不知道
+- 候補
+- 在他的
+- iens
+- 적
+- مى
+- ▁Jud
+- に戻
+- ▁Gleich
+- ▁dynamic
+- ▁arrested
+- 母親
+- ▁establishment
+- 何を
+- ▁père
+- ▁Regi
+- لف
+- 科学家
+- ▁xwe
+- 갔
+- ▁Fan
+- ▁television
+- bach
+- zes
+- 数学
+- கே
+- ってくる
+- ớ
+- ▁граждан
+- 下一个
+- ▁стран
+- loj
+- ▁是啊
+- ▁masters
+- лез
+- ▁какая
+- オープン
+- ▁option
+- ັ
+- ▁identity
+- ▁submit
+- ▁jusqu
+- 抑え
+- gefühl
+- ▁medio
+- bira
+- 我应该
+- 勝利
+- ĝis
+- ▁handkerchief
+- 夫妻
+- 孩子的
+- ▁vue
+- ▁sober
+- 阵
+- ▁lightly
+- clos
+- ▁laten
+- chie
+- şti
+- 那个人
+- informa
+- rough
+- ▁links
+- ▁distinguish
+- tego
+- ▁civilization
+- ▁jaar
+- ▁drugs
+- ▁convenient
+- いると
+- ನ
+- ▁например
+- dock
+- ▁disturbed
+- würdig
+- ▁deutlich
+- hri
+- destin
+- 免费
+- zina
+- ▁revenge
+- zogen
+- чка
+- 瀬
+- ▁concealed
+- rwanda
+- ▁باز
+- fahr
+- ilo
+- 한테
+- 奶
+- いるんです
+- аба
+- 得点
+- institu
+- нан
+- ▁hedge
+- ▁Every
+- 違い
+- ▁Ly
+- ▁بس
+- свя
+- 你这个
+- ▁dressing
+- 猛
+- 俗
+- ச்சி
+- ▁Ang
+- 问我
+- ▁بزرگ
+- ▁chapel
+- ходит
+- ▁kiu
+- ений
+- 我跟你说
+- 堡
+- utter
+- 躲
+- ▁ໄປ
+- ▁fabric
+- ▁frankly
+- を出
+- 漁
+- 队长
+- ▁هل
+- 的歌曲
+- 쫌
+- baar
+- 还有一个
+- ▁boast
+- ▁Pf
+- ▁bunu
+- ▁Kas
+- сол
+- rock
+- 元気
+- お願い
+- ▁Santa
+- imento
+- 抜
+- ▁чувств
+- ණ
+- 書いて
+- ▁dwelling
+- 聞いた
+- ▁Laufe
+- brechen
+- stol
+- 也在
+- раж
+- 给她
+- ▁arrangement
+- ▁Са
+- zat
+- jung
+- '29'
+- minded
+- ▁vivid
+- ▁18
+- 走到
+- 去看
+- ▁weren
+- うち
+- ▁Mir
+- 背后
+- abilir
+- ホームラン
+- ures
+- 上がる
+- nischen
+- bridge
+- lara
+- hound
+- ▁technologies
+- ▁accompany
+- დ
+- tempe
+- ▁cream
+- ▁anticipated
+- même
+- лей
+- ▁Linie
+- が行われ
+- ▁Pres
+- 言った
+- past
+- ruf
+- ▁trait
+- uche
+- 모
+- ▁scotland
+- ▁struggling
+- ▁vậy
+- posit
+- 患
+- ミサイル
+- ▁preserved
+- ▁зачем
+- ▁defeat
+- avu
+- ▁aby
+- ▁pint
+- iji
+- leute
+- known
+- 済
+- サービス
+- ▁ёсць
+- 実際
+- 的国家
+- ▁são
+- ▁copper
+- 宾
+- ▁Partei
+- ziel
+- ▁illustration
+- ими
+- politi
+- ハン
+- ▁US
+- bry
+- ▁verge
+- öd
+- 法官
+- itud
+- 番の
+- 差不多
+- 很有
+- ▁jury
+- ▁Pas
+- 知道了
+- ▁объ
+- edifici
+- 雨の
+- 类似
+- ▁stamp
+- stimul
+- 我们应该
+- ▁monte
+- ީ
+- 称之为
+- 细节
+- ote
+- ▁номер
+- ライン
+- 日中
+- 赌
+- ctive
+- 页面
+- gion
+- ▁creative
+- 始
+- ▁Venim
+- ▁beings
+- schloss
+- ▁influ
+- 它在
+- ▁гэтым
+- 传播
+- 子どもたち
+- ▁Vous
+- mış
+- ▁sino
+- fach
+- ▁Stelle
+- 対象
+- 个月
+- big
+- ▁autumn
+- chron
+- ▁давайте
+- ▁eaten
+- ▁crept
+- цэ
+- 値
+- ▁defense
+- 激しい
+- ▁echo
+- nelle
+- ▁owing
+- ▁Black
+- ▁moest
+- ▁behalf
+- ▁происходит
+- impa
+- ▁minimum
+- niz
+- ▁investiga
+- 着急
+- ▁october
+- られます
+- ▁virgin
+- ▁lean
+- ළ
+- ▁ответ
+- えた
+- ▁звони
+- いけ
+- 享受
+- äch
+- 你们的
+- onge
+- ▁Hände
+- trat
+- ▁outward
+- 这里的
+- ▁notwithstanding
+- ▁volunteer
+- ▁Smith
+- чная
+- 去吧
+- 湾
+- 帯
+- きている
+- вез
+- 制限
+- maya
+- どうですか
+- ▁clothing
+- 性格
+- ▁Baby
+- bbing
+- වා
+- 花了
+- аю
+- ▁entwickeln
+- ▁approval
+- ▁irrit
+- 他没有
+- ▁Alles
+- ▁associate
+- ▁kenne
+- ▁bolt
+- 永远不会
+- lug
+- ▁perfekt
+- gué
+- nutz
+- 想知道
+- ▁bestimmten
+- ▁nail
+- ビー
+- ыми
+- ▁hatred
+- glia
+- 保存
+- ▁answers
+- ▁hefyd
+- 不懂
+- 催
+- aeth
+- ▁math
+- 的一种
+- ounce
+- eco
+- қа
+- ĝa
+- regel
+- うん
+- ▁betrachten
+- பா
+- nyama
+- 起き
+- 玛丽
+- ▁Fort
+- 願
+- ▁Chúng
+- wald
+- moto
+- ▁sustain
+- 명
+- ▁founded
+- ▁bringt
+- 工资
+- undi
+- Emp
+- ▁isso
+- ▁pris
+- と一緒に
+- eaux
+- ủ
+- aĵoj
+- ても
+- 海洋
+- ▁passes
+- otto
+- ▁Gesundheit
+- կ
+- 先週
+- unga
+- зал
+- ▁immortal
+- ▁covering
+- ▁wash
+- ích
+- bula
+- казать
+- wak
+- issen
+- 担
+- 泽
+- nico
+- ▁hyn
+- ▁react
+- ▁мор
+- mol
+- ▁definite
+- ▁decrease
+- ▁scarce
+- ▁beating
+- print
+- 強化
+- ▁Wind
+- egu
+- ▁справ
+- леп
+- gga
+- ▁nhưng
+- ▁poc
+- 是最
+- ங்கள
+- ▁compete
+- かかる
+- ▁territori
+- hana
+- 続く
+- ground
+- 経
+- 就是我
+- 買
+- وری
+- 邦
+- ミリ
+- ▁Gib
+- nă
+- ▁segui
+- 种族
+- gung
+- ▁formerly
+- ▁gün
+- ents
+- யும்
+- кие
+- ▁Kindern
+- ▁farewell
+- ▁cotton
+- пон
+- ▁mechanical
+- ▁enabled
+- थ
+- wesen
+- ▁prejudice
+- 과
+- ▁Organisation
+- کان
+- ▁beasts
+- 敌
+- ▁breathing
+- ▁đề
+- ▁dues
+- ▁grandes
+- ▁letting
+- wegen
+- ▁provision
+- ▁fishing
+- ▁ongoing
+- ŭ
+- 地位
+- ▁그러
+- 和平
+- ▁länger
+- ▁canada
+- 过的
+- ▁alma
+- ▁Pal
+- cular
+- வெ
+- ▁elegant
+- 司令
+- かね
+- ▁interessiert
+- ▁feu
+- ▁violet
+- маш
+- ▁Landes
+- ▁gazing
+- gestalt
+- 一个小
+- нат
+- klar
+- ló
+- ▁gallant
+- pä
+- ▁tutto
+- ▁În
+- kore
+- 正确的
+- lever
+- ▁restrict
+- ▁Lösung
+- ▁applications
+- ▁witch
+- ительно
+- yong
+- 会在
+- تج
+- 彼女
+- stärk
+- たちは
+- гре
+- ▁Spi
+- ▁tribes
+- η
+- six
+- sitz
+- komen
+- rê
+- ▁geboren
+- ▁niets
+- ▁prominent
+- けどね
+- 诉
+- 添
+- 发生了什么
+- tamente
+- ▁unver
+- 韩
+- ỗ
+- ▁viņa
+- heure
+- ያ
+- 兹
+- るんだ
+- 巨
+- дж
+- ඉ
+- ři
+- ▁upstairs
+- ▁entertain
+- цию
+- ▁indicate
+- kab
+- ▁rarely
+- 株
+- ▁imperial
+- pression
+- дем
+- 应用
+- 形で
+- crib
+- ▁replace
+- ▁beaten
+- 阴
+- ▁আছে
+- 平时
+- 立刻
+- 一群
+- ▁Außerdem
+- джа
+- ▁seltsam
+- ▁ته
+- 유
+- いわゆる
+- দা
+- ▁dimension
+- ▁Bob
+- 岁的
+- 思い出
+- open
+- 孤独
+- ▁disaster
+- களில்
+- baba
+- ▁touching
+- をかけ
+- ▁tribe
+- kati
+- 航空
+- ▁Nation
+- ▁maxim
+- 留在
+- 방
+- となり
+- 手が
+- ▁ähnlich
+- ēt
+- ▁strict
+- いきたい
+- により
+- гән
+- ▁wives
+- ▁политик
+- クラ
+- どちら
+- ▁Herausforderung
+- 这一切
+- ▁sailed
+- lash
+- ▁hanno
+- тка
+- 月の
+- ▁Vorstellung
+- sible
+- участ
+- 倾
+- namen
+- हरू
+- 込まれ
+- daw
+- 揚げ
+- っていく
+- nsen
+- ▁collar
+- ނ
+- hill
+- ▁petite
+- alität
+- ▁grab
+- ▁Kam
+- ▁egal
+- 绿
+- 事业
+- ores
+- ▁bewegen
+- besch
+- ▁Titel
+- 了她
+- きょうの
+- 这两个
+- 危険
+- labor
+- ▁funds
+- 十三
+- 갈
+- ▁jaw
+- ▁senator
+- ▁doorway
+- ationen
+- ્
+- зар
+- litz
+- dó
+- ▁определ
+- ▁あれ
+- ▁accent
+- yana
+- ▁wrath
+- Li
+- ▁moses
+- arna
+- ▁Hay
+- ええ
+- ▁interfere
+- ▁Text
+- ▁можна
+- 臭
+- なんですけれども
+- embra
+- zó
+- лич
+- 描
+- 发表
+- ぁ
+- itor
+- ▁cheese
+- ▁stato
+- 돈
+- ▁Mitarbeiter
+- ▁label
+- কার
+- ▁spun
+- ▁propos
+- ▁speaks
+- 在家
+- 赞
+- ór
+- ▁finds
+- 市で
+- 眠
+- ነ
+- flor
+- ニア
+- ▁мяне
+- ▁اگر
+- ▁operator
+- tention
+- ▁definition
+- 見事
+- вра
+- ▁büyük
+- tribut
+- порт
+- 非常非常
+- вяз
+- rov
+- ▁walter
+- oti
+- ▁marched
+- ▁restored
+- ells
+- ▁trug
+- 好啊
+- onda
+- ▁riches
+- ▁Hoffnung
+- 太太
+- ுடன்
+- lob
+- ▁belongs
+- ▁hans
+- 非常重要
+- ▁présent
+- 分子
+- 某个
+- 讨厌
+- ▁attracted
+- しまう
+- ▁encounter
+- nnes
+- zzle
+- レイ
+- ▁attain
+- ▁вами
+- ▁mutter
+- 开车
+- 바
+- ▁funeral
+- щу
+- اره
+- ▁shiver
+- 创新
+- ▁стал
+- treiben
+- 保险
+- ▁lightning
+- ▁complaint
+- ▁african
+- 重大
+- 你跟
+- ▁같은
+- چى
+- 对他们
+- ▁quel
+- ▁sama
+- ▁обо
+- 遊
+- aves
+- ▁manufacturing
+- 汗
+- 教会
+- 做一个
+- ▁wid
+- мент
+- ▁november
+- ▁месяц
+- ▁elbow
+- ▁tank
+- ほぼ
+- лем
+- ति
+- NHK
+- кин
+- ▁Bilder
+- ▁zeker
+- стве
+- 一个非常
+- ▁curt
+- ტ
+- ▁asset
+- 破坏
+- anye
+- ▁differences
+- ▁нашей
+- clin
+- ウクライナの
+- 行われた
+- ▁Shi
+- dors
+- ▁democratic
+- ▁Obwohl
+- صف
+- هر
+- ▁feeble
+- leş
+- 这种情况
+- ssin
+- を受けた
+- kai
+- ▁robust
+- ▁besondere
+- ▁elected
+- ▁Modell
+- ▁seconds
+- lauf
+- ダイ
+- ګ
+- セット
+- 時は
+- Applaus
+- 白人
+- ▁свобод
+- 病毒
+- 损
+- ජ
+- ▁Mat
+- ▁executive
+- ▁onto
+- lag
+- ▁важно
+- ან
+- ▁concerns
+- ء
+- ebbe
+- ▁Tru
+- ▁seed
+- 推荐
+- ▁Staat
+- 祭
+- ▁cleared
+- მო
+- eixen
+- 让它
+- သ
+- wiesen
+- Ro
+- train
+- ▁acknowledge
+- 取れ
+- 風が
+- 依然
+- ▁indifferent
+- 人数
+- ▁generate
+- ▁frost
+- itaj
+- ▁reminded
+- ▁romance
+- 隠
+- ▁ireland
+- ▁Ха
+- 最多
+- ▁layer
+- 柳
+- 就有
+- ▁caution
+- ▁bezeichnet
+- ▁denied
+- سە
+- ▁profile
+- 的位置
+- tato
+- ▁politische
+- ▁Dra
+- 高度
+- ちゃった
+- ▁Bay
+- 成立
+- ▁stephen
+- laş
+- 変わり
+- neb
+- ält
+- oza
+- coloured
+- рок
+- ▁independence
+- ▁traveller
+- 接着
+- utu
+- 许多人
+- ▁leather
+- 这项
+- 进步
+- ▁exists
+- kia
+- 进一步
+- ړ
+- ▁Char
+- ▁English
+- cional
+- maakt
+- ▁zá
+- ▁wereld
+- ▁дэ
+- 他们都
+- ▁Sicht
+- 出来了
+- وك
+- ▁christians
+- 種類
+- gone
+- ▁говорят
+- ▁danach
+- ▁positions
+- 问你
+- यो
+- ▁daylight
+- wall
+- えない
+- 不住
+- ▁echt
+- 脳
+- 搜
+- ▁consists
+- děl
+- 人在
+- ▁Adam
+- umva
+- ▁formula
+- anzi
+- ▁тысячи
+- ▁casual
+- ины
+- ▁eighth
+- plica
+- guer
+- 世界的
+- ▁استفاده
+- ▁German
+- ▁بیشتر
+- ▁Ŝi
+- ează
+- 反対
+- ▁determination
+- 时刻
+- iller
+- 히
+- 不愿意
+- ▁republic
+- respon
+- alen
+- liste
+- ▁outer
+- ▁echte
+- 对待
+- 私たち
+- ▁которых
+- って言って
+- 你先
+- ่
+- ▁aveva
+- nym
+- acions
+- ▁然后他
+- ▁democracy
+- 填
+- ▁гэты
+- ▁Bruder
+- ▁characteristic
+- ▁Kris
+- ▁neuro
+- politic
+- ▁visitors
+- ໂ
+- ▁gale
+- tır
+- ▁protestant
+- ▁befinden
+- ▁одиннадцать
+- ▁Schi
+- いまして
+- öhn
+- 연
+- そういうこと
+- arbeiten
+- лас
+- ▁porter
+- 做到
+- cun
+- нер
+- nette
+- atan
+- ▁zweitausend
+- ల
+- ట
+- öö
+- ▁relevant
+- 鼓励
+- 很棒
+- fford
+- ▁dios
+- ▁Tak
+- 个小时
+- 我发现
+- ▁Ul
+- rating
+- ▁creek
+- ▁aggressive
+- ▁трэба
+- それでも
+- 手里
+- へと
+- بح
+- нік
+- 一个问题
+- 公主
+- 见面
+- ▁saa
+- 始め
+- 崩
+- ivo
+- Mi
+- ▁extensive
+- ▁спасибо
+- ▁properties
+- ▁Asia
+- ▁Gran
+- 序
+- 애
+- posing
+- 发生的事情
+- 想起
+- 十七
+- ▁Anwendung
+- лө
+- ▁Member
+- トー
+- 孔
+- ▁наши
+- ▁span
+- ▁swung
+- ▁perception
+- हा
+- ▁ox
+- 劝
+- ▁laying
+- ▁carpet
+- wch
+- vē
+- はもう
+- 赚
+- ▁Demo
+- mala
+- daten
+- მე
+- quar
+- ▁تمام
+- anima
+- ▁radical
+- ▁власти
+- anı
+- ▁Ils
+- 証
+- '00'
+- hub
+- ало
+- の姿
+- るのが
+- раў
+- 読
+- '%'
+- ваў
+- 监狱
+- すいません
+- ▁execute
+- bred
+- ▁profess
+- ▁Fenster
+- 不上
+- exc
+- ▁torment
+- 互相
+- ▁unge
+- 着我
+- ரிய
+- の一
+- ▁sci
+- 設置
+- versa
+- ▁größten
+- vê
+- ▁graceful
+- popul
+- 夺
+- ಸ
+- 採
+- とにかく
+- name
+- ▁Partner
+- あるいは
+- ▁nuestra
+- ▁impressed
+- ức
+- 对不对
+- IT
+- 方便
+- pek
+- 一周
+- ਾ
+- ම්
+- ▁California
+- ▁assembly
+- ▁bran
+- жда
+- 可以在
+- ▁lei
+- 在她
+- 有趣的
+- ▁divide
+- 直播
+- nemen
+- の話
+- 求め
+- 彼の
+- ▁trends
+- ▁veröffentlicht
+- илось
+- 助け
+- க்கப்பட்ட
+- ▁دې
+- ▁rapport
+- ▁tones
+- nego
+- 怎么说
+- ົ
+- 也很
+- grat
+- träge
+- ▁großartige
+- ▁cape
+- ▁successfully
+- سى
+- علم
+- 撮
+- jah
+- 安妮
+- ▁whit
+- inin
+- ▁Funk
+- 他自己
+- ▁funding
+- jär
+- 女朋友
+- ホテル
+- ▁poly
+- ▁awkward
+- iyorum
+- wich
+- lish
+- ▁Jung
+- cata
+- جم
+- ▁Then
+- bby
+- embe
+- ▁Kein
+- 防止
+- 访问
+- ▁blanket
+- 鉄
+- ▁andrew
+- ▁grounds
+- ▁چا
+- ▁Israel
+- ▁pap
+- 出てくる
+- ▁convention
+- றி
+- えっ
+- ▁Wieder
+- 充满
+- ▁Australia
+- fik
+- аем
+- ▁schlechte
+- ▁Tier
+- ▁febrer
+- 屈
+- 肖
+- 課題
+- idas
+- ▁milli
+- ▁parish
+- borough
+- 進め
+- ▁Zeug
+- عب
+- En
+- ▁exceedingly
+- 柔
+- 砂
+- 郑
+- ▁таксама
+- leb
+- 最低
+- 当て
+- ▁japan
+- 你把
+- ächtig
+- sort
+- ▁shudder
+- ▁profitable
+- 如果他们
+- ▁放一下
+- クロ
+- ände
+- 他们说
+- ▁Durant
+- 很高兴
+- pare
+- artig
+- schlossen
+- ▁ora
+- ▁அந்த
+- cà
+- 掉了
+- fè
+- ▁canal
+- ▁administra
+- ▁bears
+- 出现在
+- ▁shriek
+- ▁знал
+- ebilir
+- ▁schrieb
+- contra
+- ▁roar
+- 間に
+- 液
+- 相互
+- ▁package
+- でお
+- ▁Maar
+- 식
+- けども
+- ▁romantic
+- ▁tales
+- ▁gelernt
+- ▁invent
+- 大体
+- ▁anyhow
+- を作る
+- stoß
+- ▁podcast
+- ▁Пра
+- ლი
+- 奔
+- lp
+- män
+- кар
+- mesi
+- 的观点
+- gani
+- Mail
+- ▁mwa
+- サッカー
+- ▁alien
+- tiu
+- 是真的
+- 計画
+- りの
+- 診
+- 医疗
+- 明确
+- ▁parece
+- десят
+- иш
+- 哇
+- ▁meta
+- CO
+- lec
+- ▁但是我们
+- argent
+- 炸
+- 我先
+- ▁studying
+- shaped
+- 的样子
+- ▁anymore
+- ▁attendant
+- 可怕
+- ▁erfahren
+- nade
+- ▁zona
+- ▁Dame
+- ▁containing
+- ▁Patienten
+- ▁болон
+- 統
+- ר
+- ▁ჰო
+- ▁соглас
+- মে
+- 目に
+- ▁gyda
+- ▁endeavour
+- 那时
+- 発射
+- бур
+- ▁cael
+- จะ
+- 言われて
+- ▁meg
+- 観光
+- ▁envelope
+- ▁говорю
+- ▁snap
+- ▁truck
+- 电子
+- ຊ
+- ▁endeavor
+- ▁stole
+- ▁beheld
+- 焼
+- 这也是
+- energie
+- ▁acid
+- ▁communicate
+- ▁Хо
+- 估计
+- 怎么会
+- зем
+- මා
+- ▁Australien
+- fish
+- 嫁
+- igheid
+- odi
+- 区别
+- нев
+- liko
+- tale
+- ▁Camp
+- 文学
+- krit
+- hielt
+- ▁tube
+- ▁карт
+- igne
+- ді
+- 我再
+- 的身体
+- obo
+- 告诉我们
+- ▁dhateng
+- ▁invention
+- ări
+- ▁அவர
+- ▁Herrn
+- าย
+- ▁cunning
+- folk
+- をしている
+- ▁Fra
+- ▁ignor
+- ▁Denken
+- 付き
+- ▁слуша
+- 実験
+- patri
+- ▁Ganz
+- ▁kurze
+- ▁saving
+- 東北
+- 韦
+- るのか
+- ▁baz
+- ▁Freude
+- ▁километр
+- 責任
+- ▁restless
+- ском
+- ▁Ty
+- ▁факт
+- ▁whip
+- 收集
+- となります
+- スタン
+- ▁folly
+- ▁wool
+- ▁uses
+- મ
+- ▁Unsere
+- 全員
+- стат
+- ▁regards
+- ▁Flu
+- El
+- ▁recommend
+- るような
+- tering
+- ству
+- ీ
+- cope
+- 每天都
+- harmoni
+- ▁کرده
+- ▁appointment
+- 下雨
+- 广泛
+- ehr
+- für
+- 向き
+- cell
+- юр
+- വ
+- 设置
+- 这只是
+- ▁erzählt
+- publik
+- ▁nämlich
+- 经理
+- ▁senate
+- ▁perpetual
+- 的研究
+- 昼
+- ற்று
+- ▁Grand
+- 和我们
+- ものです
+- まって
+- hra
+- ▁laura
+- 可能性がある
+- ▁Cho
+- ▁mama
+- ▁Cam
+- chair
+- ▁chart
+- ▁опять
+- ▁injury
+- scheid
+- ambo
+- ▁gospel
+- projekt
+- cı
+- ▁elephant
+- ▁наверное
+- nisse
+- ▁dislike
+- ▁stomach
+- ▁amendment
+- 応援
+- ▁barrier
+- ▁upset
+- 独特
+- 白色
+- ▁dedi
+- 無理
+- ▁extract
+- 時間が
+- 小说
+- ▁pains
+- 就是你
+- 観測
+- ▁所以说
+- 粘
+- 選手は
+- yin
+- ▁этому
+- ▁lass
+- 湯
+- 指挥
+- ▁illumin
+- rush
+- 進んで
+- দের
+- élé
+- ▁helps
+- fia
+- ▁Bildung
+- ▁Jeder
+- 亏
+- 億円
+- ご覧ください
+- gress
+- ▁romans
+- ▁Haben
+- トラ
+- 国家的
+- mato
+- hut
+- 支え
+- ▁tudi
+- ▁disappear
+- ына
+- ▁Gesundheits
+- ▁ነው
+- ▁Beziehung
+- cam
+- ▁Drei
+- 扣
+- ▁celebrated
+- ▁coisa
+- ▁convey
+- ▁aussehen
+- weis
+- ▁Hon
+- 止め
+- 碗
+- ▁acquisitions
+- 关闭
+- ▁Ба
+- 超级
+- ▁tables
+- ໃ
+- ▁ndetse
+- 一面
+- ▁واقع
+- ีย
+- ibility
+- 載
+- ilor
+- 信じ
+- وف
+- раг
+- grade
+- 上の
+- ▁Viņš
+- ӧ
+- 手段
+- бил
+- ▁damp
+- ▁Alex
+- ▁taxes
+- ▁chez
+- ▁50
+- ▁zumindest
+- がいい
+- 盤
+- brau
+- 歴史
+- kopf
+- ▁wordt
+- ▁ເອີ
+- 두
+- 차
+- เป็น
+- 茨
+- ▁cable
+- ▁grows
+- ▁constitute
+- ▁یې
+- ▁interessante
+- てしまう
+- 備
+- ▁despre
+- ▁promo
+- 回目
+- 那个时候
+- ▁imbere
+- 探索
+- phon
+- 途中
+- ▁machines
+- ▁Handel
+- ▁salvation
+- ▁parted
+- ▁punto
+- алі
+- ▁moja
+- 壮
+- halb
+- ▁corporate
+- 者的
+- ▁flank
+- 勇
+- ▁relieved
+- ますよ
+- ена
+- drückt
+- Н
+- ▁churches
+- めちゃくちゃ
+- 自民党
+- 进了
+- 仁
+- ▁swim
+- ▁loĝ
+- ▁الآن
+- hnen
+- 也不能
+- ▁eben
+- とっても
+- ټ
+- ▁getroffen
+- ▁stake
+- kou
+- ▁Fisch
+- schap
+- 太好了
+- zwi
+- ▁vive
+- 如果他
+- 最重要的
+- γ
+- ますが
+- 軍事
+- ▁woe
+- ウェ
+- 息子
+- ▁Steuer
+- 肯定是
+- ▁tay
+- ▁boss
+- граф
+- rons
+- lasse
+- 灭
+- க்கும்
+- lini
+- 体験
+- ▁delivery
+- ▁temptation
+- 忘记
+- ウル
+- 奴
+- ▁thành
+- ▁په
+- ▁writers
+- ▁assistant
+- ▁Tam
+- кө
+- 我爸
+- ▁boom
+- чный
+- だね
+- اپ
+- weil
+- uwa
+- ▁facilities
+- ▁monkey
+- 私たちは
+- صور
+- ▁тому
+- સ
+- 真相
+- 这就
+- 削
+- ▁Sim
+- ▁squire
+- ▁britain
+- 败
+- ۰
+- しかった
+- ちなみに
+- ستان
+- ▁trifle
+- dist
+- ▁options
+- empat
+- 敲
+- ▁Monat
+- 会说
+- стой
+- しようと
+- ▁langen
+- ப்பி
+- ▁Az
+- 焦
+- ますので
+- ছ
+- sters
+- 糟糕
+- erei
+- ▁subtle
+- 胎
+- に行く
+- ▁яна
+- ▁времени
+- スタジオ
+- ▁сообщ
+- 普遍
+- yim
+- pē
+- ▁sitzen
+- プーチン大統領
+- 你没有
+- —
+- ▁Monate
+- ▁dive
+- ▁Fre
+- 各地
+- ▁bias
+- ml
+- ▁bị
+- ▁gardens
+- ▁ceremony
+- ▁roots
+- غل
+- geschichte
+- ▁Ĉi
+- ▁minu
+- ▁имеет
+- 来年
+- iaeth
+- 小さな
+- 饿
+- ▁можа
+- ▁Hall
+- 帽子
+- کش
+- ▁pense
+- ▁Tisch
+- эконом
+- ufu
+- の世界
+- ▁Leistung
+- ▁steer
+- сю
+- hul
+- ーク
+- ですので
+- 爷爷
+- вон
+- ▁wax
+- 违
+- дь
+- 一个月
+- ▁isabel
+- 向你
+- ె
+- lari
+- ▁peril
+- メッセージ
+- すぎ
+- 것
+- ▁общем
+- ▁Ken
+- ▁natives
+- 邪
+- ▁Anda
+- nice
+- 结合
+- 把她
+- chang
+- кро
+- ▁Kunden
+- 首都
+- 地点
+- ▁launched
+- ▁cran
+- вен
+- фер
+- 開始
+- ப்பா
+- ▁Dum
+- 的意思
+- ▁dragon
+- ▁silently
+- なければ
+- 荣
+- ▁Junge
+- ▁searching
+- 这将
+- ▁erklären
+- ▁scratch
+- ▁ehrlich
+- emba
+- ▁Liste
+- 応
+- 까지
+- ▁fitted
+- とする
+- GAAP
+- ▁limits
+- რი
+- を使った
+- 배
+- ovi
+- eight
+- раш
+- ▁marine
+- ▁Tar
+- ▁Many
+- 壊
+- 년
+- ▁troubles
+- ▁nuclear
+- ▁lodging
+- خص
+- ▁termina
+- 多分
+- Weiß
+- 有所
+- acak
+- wise
+- prof
+- num
+- 英语
+- abana
+- 胜利
+- cada
+- autobús
+- ▁residence
+- ولا
+- ▁zeg
+- ▁Jim
+- 抵
+- ▁Bö
+- ▁Hauptstadt
+- 京都
+- iling
+- 我把
+- ▁appetite
+- ▁oameni
+- ▁cardinal
+- häng
+- 地元
+- ▁doubtful
+- reiz
+- リスク
+- 来て
+- 慌
+- ғ
+- しょう
+- ▁bother
+- 算是
+- ▁reverse
+- ▁dearest
+- ของ
+- ▁занима
+- ▁giả
+- 不像
+- ▁newspapers
+- ▁boil
+- のうち
+- μ
+- ▁мужчин
+- 一声
+- ār
+- ▁fury
+- かわいい
+- ầ
+- api
+- ố
+- ở
+- ▁Aufmerksamkeit
+- shore
+- 堆
+- ▁центр
+- 議論
+- ได้
+- ですし
+- 如今
+- ▁passions
+- ▁limbs
+- ▁producer
+- verein
+- ▁مج
+- ▁glen
+- ▁natur
+- rade
+- ▁verrückt
+- 版本
+- 回転
+- lica
+- ▁hinder
+- ▁hastened
+- ▁anticipate
+- 大事な
+- の問題
+- ▁zwölf
+- 先に
+- ▁그니까
+- ▁erstaunlich
+- ▁assault
+- 发生在
+- ▁vede
+- ▁bust
+- ▁binnen
+- ▁lloc
+- は今
+- tré
+- といった
+- court
+- ▁exquisite
+- wunde
+- ның
+- ▁Lee
+- ▁bundle
+- duct
+- 高速
+- ▁nas
+- mac
+- ▁достаточно
+- 体育
+- チャー
+- さあ
+- ▁desires
+- ▁Geschäfts
+- gana
+- 辛苦
+- 略
+- 交谈
+- 的那种
+- bí
+- レベル
+- ▁brass
+- 说得
+- 무
+- dala
+- 歯
+- ▁shalt
+- kuba
+- スタ
+- ▁взгляд
+- 对象
+- ▁clinical
+- 目を
+- 殖
+- industrie
+- 戈
+- 的主要
+- ▁Sel
+- ԥ
+- enne
+- ▁ralph
+- ▁resting
+- ась
+- aran
+- body
+- ▁esteem
+- ▁hail
+- ඩ
+- ▁вместе
+- 仔细
+- ▁rip
+- ▁Aquesta
+- ▁Welche
+- Bo
+- されていた
+- ▁multe
+- 锅
+- 否
+- ▁наша
+- ▁export
+- ▁vigorous
+- ▁Phil
+- мын
+- ▁scandal
+- 炮
+- カレー
+- ▁Gla
+- ▁بازی
+- تان
+- ▁prevented
+- ▁tranquil
+- ▁tobacco
+- ▁thence
+- 집
+- ▁tire
+- ▁schwarze
+- リング
+- ▁embrace
+- ▁schu
+- ほら
+- ▁rural
+- ҙы
+- ▁skirt
+- ▁каждый
+- るという
+- ▁Bitte
+- 仍
+- ຄ
+- ▁wrap
+- vara
+- ەوە
+- 者は
+- воз
+- нал
+- ▁offers
+- έ
+- 恐怖
+- 安心
+- ▁geweest
+- ▁dumb
+- ▁municipal
+- фон
+- 上げて
+- tē
+- 小组
+- ▁Carl
+- ▁golf
+- ▁Position
+- ▁después
+- 绑
+- '2019'
+- ▁scientists
+- ▁但是你
+- 경
+- ▁دیگر
+- ▁excess
+- লো
+- erweise
+- pence
+- ▁prey
+- 把它们
+- قة
+- そこで
+- ▁fünfzig
+- пле
+- 微笑
+- ▁cabinet
+- ▁accused
+- muzi
+- 確保
+- ▁решил
+- ▁Büro
+- 的最
+- كر
+- ▁formation
+- ▁princes
+- ですけれども
+- still
+- әр
+- ▁баш
+- ▁entertainment
+- ິ
+- 奉
+- '800'
+- ▁собственно
+- 放下
+- ということを
+- bird
+- システム
+- ток
+- ▁Tochter
+- 被告
+- ▁uncertain
+- ▁нельзя
+- ▁شخص
+- ほん
+- arma
+- ▁insight
+- ▁ruler
+- ▁altijd
+- ▁habla
+- ären
+- ▁Here
+- zunehmen
+- ▁passionate
+- ▁nunca
+- ▁typically
+- 童
+- 와
+- ▁supplies
+- räum
+- 温暖
+- anche
+- ▁lustig
+- stop
+- ▁ascend
+- かし
+- ▁Stunde
+- ▁начина
+- وار
+- றை
+- ර්
+- ▁marquis
+- ▁Quin
+- 伙伴
+- ▁إذا
+- ▁Pop
+- 惠
+- кую
+- のない
+- anne
+- していきます
+- 友達
+- ▁retire
+- 来源
+- almente
+- frau
+- ▁ўжо
+- mates
+- няя
+- нич
+- ▁seda
+- 消失
+- ısı
+- ▁wrapped
+- வில்
+- ▁política
+- 行われ
+- 見えて
+- ▁mau
+- lager
+- ▁při
+- 上来
+- lett
+- ލ
+- dang
+- ▁bez
+- енный
+- 哲学
+- ены
+- جز
+- ▁barrel
+- 相反
+- ▁neighbourhood
+- パス
+- zeichne
+- esten
+- メダル
+- ▁Kosten
+- ▁relativ
+- коп
+- ▁älter
+- 编辑
+- الي
+- gründe
+- голов
+- 体を
+- ஷ
+- 有机会
+- ▁nobles
+- 这件事情
+- が発生
+- ということなんです
+- 告诉他
+- 往往
+- ▁دارم
+- ũ
+- yah
+- 五个
+- న
+- ▁argue
+- ශ
+- ピン
+- imos
+- ▁trusted
+- ▁прямо
+- рог
+- ▁depuis
+- idades
+- Ч
+- ▁যে
+- ではありません
+- ていて
+- 億
+- ▁unpleasant
+- smith
+- わない
+- ▁latest
+- рг
+- んじゃないかな
+- nische
+- ▁pulling
+- を持つ
+- puesto
+- 乘
+- 酒店
+- ▁такого
+- ▁Gelegenheit
+- rats
+- ▁svoj
+- ▁viva
+- 这就是为什么
+- 杂
+- ▁مورد
+- পর
+- ▁pushing
+- ▁depths
+- しよう
+- ヨ
+- Ne
+- gles
+- ▁guards
+- 均
+- ▁hadden
+- ▁maison
+- ▁sketch
+- вых
+- 宝宝
+- 弁
+- ▁surgeon
+- ▁aŭ
+- ▁siempre
+- sea
+- 除非
+- ▁далее
+- quí
+- 感觉到
+- capacit
+- ▁advantages
+- ▁chant
+- umwa
+- bä
+- 森林
+- ▁foul
+- مار
+- 矿
+- 跨
+- ▁staying
+- 热情
+- мысл
+- ▁paw
+- ▁Straßen
+- ▁forecast
+- ▁собира
+- اری
+- 互
+- obli
+- енные
+- ப்பட
+- ▁fum
+- ▁cá
+- ▁distinctly
+- ада
+- wari
+- 飲
+- 一部分
+- 市内
+- 怎么能
+- gge
+- ▁عنوان
+- 出演
+- ▁sterben
+- ▁reproduc
+- ▁register
+- ▁Mitte
+- tiga
+- bron
+- ▁Software
+- ▁knights
+- たちの
+- ▁место
+- ▁menschliche
+- ising
+- ▁repent
+- ▁mechanism
+- taka
+- ▁haut
+- 需求
+- ▁stopping
+- ợ
+- ▁breathe
+- 濃
+- 不容易
+- stick
+- sagen
+- ▁guarantee
+- 这些东西
+- 詰め
+- 缝
+- ▁sabi
+- 所在
+- ランナー
+- ▁представи
+- ▁является
+- 特别是
+- شي
+- Đ
+- ▁després
+- 辺り
+- 也要
+- ▁образом
+- 听着
+- ▁earnestly
+- ▁Tim
+- воль
+- 就没有
+- ост
+- ▁oleh
+- 何が
+- ▁donne
+- فة
+- trägt
+- ▁brains
+- 幸运
+- ▁deadly
+- станов
+- valu
+- ürü
+- いえ
+- ▁skull
+- gebiet
+- Mar
+- 敵
+- 攻め
+- ▁spielt
+- source
+- baren
+- ▁Bevölkerung
+- 都能
+- ▁stirred
+- ▁commonly
+- ▁pang
+- ▁Before
+- 角度
+- ▁tener
+- ▁winning
+- ▁öffentliche
+- 截
+- نة
+- ▁parce
+- dler
+- させた
+- уль
+- ▁Größe
+- ▁nei
+- ▁conservative
+- ▁туда
+- というもの
+- 用の
+- ▁boxes
+- ▁reception
+- ▁Şi
+- ▁timber
+- ▁Druck
+- 発生
+- мир
+- 適
+- 弃
+- ▁Sud
+- 解決
+- க்கி
+- ▁baga
+- からも
+- وح
+- ▁comprend
+- 在你的
+- 工人
+- 划
+- ложен
+- 原则
+- 史上
+- klä
+- ե
+- markt
+- 荡
+- kari
+- walt
+- オン
+- ▁neues
+- ▁Ergebnisse
+- ▁courts
+- нг
+- 버
+- shot
+- ең
+- ▁blew
+- haltung
+- ▁Herzen
+- грамм
+- ▁participate
+- ▁competi
+- ちょ
+- 알
+- ▁gerek
+- ▁президент
+- ▁esti
+- ண்ண
+- 공
+- 교
+- ▁fatigue
+- ▁fatto
+- ▁aquella
+- altra
+- ▁blossom
+- 恵
+- 鮮
+- 瑟
+- ▁jewel
+- ▁Rock
+- 帝
+- ось
+- ▁remainder
+- ▁راه
+- 特別
+- ▁technique
+- ▁pillow
+- opa
+- 많
+- ▁category
+- ▁таким
+- ませ
+- ▁monday
+- それぞれ
+- pio
+- läuft
+- ▁zunächst
+- ▁Second
+- volg
+- motiv
+- ▁nerves
+- ▁programme
+- omu
+- tious
+- ▁Bell
+- ▁augen
+- дон
+- 一把
+- инский
+- ▁ஆஹ்
+- ▁omdat
+- ▁dutch
+- سته
+- ▁profitability
+- affe
+- ▁standards
+- 付近
+- 俄罗斯
+- kreis
+- ▁gelang
+- овых
+- кор
+- ▁procession
+- ơ
+- 将其
+- 他有
+- 爱的
+- 很有趣
+- ▁Omu
+- кія
+- ▁summit
+- ▁fünfzehn
+- چه
+- 生活在
+- コー
+- 情况下
+- ▁gesamten
+- 海岸
+- 非常好
+- 家长
+- écri
+- ▁murderer
+- ▁Он
+- ▁مست
+- ▁accelerate
+- itas
+- cía
+- ▁diejenigen
+- ▁contracts
+- ▁commun
+- 高中
+- hte
+- ▁delivering
+- たっぷり
+- kala
+- ▁popula
+- 公园
+- ▁domain
+- ▁journalist
+- ▁repar
+- crit
+- ための
+- ▁Auswirkungen
+- ▁былі
+- ▁Papier
+- 所以他们
+- lē
+- பார்
+- 外交
+- ▁Bundes
+- 我们家
+- そんなに
+- ▁stolen
+- 低い
+- 附
+- zaba
+- иа
+- ▁egyptian
+- ା
+- 달
+- 脏
+- ▁uwo
+- lă
+- 会不会
+- ▁Web
+- ▁olduğu
+- дор
+- regul
+- 時点で
+- 必ず
+- ță
+- wissenschaftlich
+- ▁scripture
+- mper
+- を示
+- ▁alterna
+- kke
+- حر
+- shuri
+- ▁format
+- amerikanische
+- ння
+- ▁serving
+- ▁doll
+- ▁resemble
+- ▁slain
+- 走进
+- 美丽的
+- 展開
+- ▁wilderness
+- ▁creo
+- kind
+- ▁Stein
+- ▁mussten
+- gramm
+- ▁абсолютно
+- ని
+- film
+- 不好意思
+- ▁sailor
+- 我看到
+- 笑声
+- 嘅
+- 火车
+- 便宜
+- gid
+- うまい
+- 也就是说
+- 中継
+- ▁cats
+- ango
+- wehr
+- чил
+- ▁flashed
+- 小孩
+- 으면
+- ▁occupy
+- 不对
+- ▁victims
+- rink
+- ▁shan
+- ▁dife
+- ▁должно
+- あんまり
+- чай
+- ▁وقتی
+- 谈话
+- 珍
+- 庁
+- コース
+- rechte
+- ▁компани
+- 后的
+- ▁fem
+- ▁utili
+- struktur
+- 付出
+- mount
+- ▁humour
+- ții
+- ▁respectable
+- を迎え
+- бри
+- ▁бог
+- myaka
+- mast
+- ▁accomplish
+- оро
+- 什么意思
+- ▁allowing
+- ▁streams
+- 扫
+- mming
+- ▁Pferd
+- ▁aya
+- ގެ
+- ▁Canada
+- chief
+- ▁erhielt
+- ▁scared
+- ▁burnt
+- ▁gloomy
+- ▁выбор
+- lha
+- ▁さて
+- нулся
+- ▁Link
+- ımı
+- ベル
+- ிருந்த
+- 仪
+- 睡眠
+- ▁God
+- ▁Vorteil
+- ▁Stellen
+- ístic
+- を作り
+- ▁Reich
+- 重点
+- 氷
+- つい
+- лийн
+- 早速
+- ▁số
+- ▁стол
+- ▁Haar
+- ゾ
+- 説
+- ▁equity
+- ▁grove
+- ▁زندگی
+- お客さん
+- 选举
+- 妮
+- फ
+- ывать
+- 開かれ
+- kamer
+- хай
+- ▁adjusted
+- парт
+- ▁insanlar
+- 垃圾
+- ան
+- だが
+- 您可以
+- mena
+- ▁suitable
+- ދ
+- fähig
+- bikorwa
+- 上升
+- ▁needle
+- ▁hawk
+- มี
+- ベスト
+- ▁prend
+- ▁clima
+- poko
+- 地元の
+- ▁jos
+- قابل
+- ▁prudent
+- 什么是
+- ▁agony
+- ▁inevitable
+- losen
+- ▁таких
+- ▁wages
+- ▁otros
+- date
+- 双方
+- 诊
+- ▁cluster
+- 桂
+- 杉
+- インド
+- ▁meaningful
+- iet
+- ▁fruits
+- 継
+- ▁status
+- 的男人
+- ▁trage
+- ▁polly
+- zehn
+- でしたが
+- ▁самое
+- amour
+- 和她
+- ufer
+- ▁transparent
+- rier
+- tero
+- スク
+- を入れ
+- ▁offence
+- ▁இருந்த
+- ライブ
+- ▁washed
+- ▁obscure
+- esso
+- 規模
+- ▁resident
+- ▁resolve
+- ▁betrachtet
+- ▁foarte
+- 幻
+- 正好
+- ▁damn
+- zugehen
+- 那儿
+- ஓ
+- წ
+- tian
+- 層
+- ▁assembled
+- 中国人
+- ▁chairs
+- 夸
+- ▁сур
+- 解除
+- ▁Dit
+- 終わり
+- To
+- ▁menyang
+- tanz
+- ▁officials
+- ▁libro
+- ▁Glauben
+- ibu
+- ▁prim
+- гы
+- ▁Most
+- 构建
+- ▁entweder
+- ▁bauen
+- 是关于
+- 動物
+- ▁Mund
+- ambu
+- かけて
+- rop
+- ▁generations
+- ▁Joan
+- ▁dash
+- ▁masses
+- ▁mögen
+- بات
+- oud
+- 込んだ
+- ▁brows
+- 演奏
+- 一気に
+- 防衛
+- cula
+- ▁Interesse
+- еры
+- 方針
+- 和他的
+- mog
+- ຈ
+- 輝
+- ▁disguise
+- ▁varia
+- ▁leaped
+- नि
+- ▁leer
+- шь
+- zieren
+- ļa
+- ına
+- rique
+- 彼女は
+- ளை
+- 雑
+- amendement
+- 这本书
+- ▁confined
+- 不久
+- 将在
+- ▁nephew
+- imit
+- бан
+- 缘
+- gado
+- 我们今天
+- ereza
+- 墓
+- ▁domina
+- 买了
+- 体制
+- 稳
+- صر
+- 一万
+- 競
+- 浴
+- 的结果
+- 時から
+- पा
+- ▁hopefully
+- ▁còn
+- 什么都
+- ▁poe
+- ▁fastened
+- ▁brands
+- ▁ấy
+- 烂
+- ▁florence
+- ▁legislation
+- gari
+- ▁высо
+- barra
+- ▁Plu
+- aciones
+- dna
+- 一首
+- 賀
+- ▁december
+- ▁valid
+- cine
+- ▁erinnert
+- pole
+- familie
+- 琳
+- strat
+- ▁humans
+- gā
+- ▁kinder
+- ▁scri
+- ampl
+- ▁blast
+- 琴
+- 曹
+- ▁dispar
+- 绿色
+- 挣
+- ▁продолжа
+- кру
+- ▁steamer
+- ▁arguments
+- ▁toda
+- を超える
+- ▁kullan
+- tli
+- ▁Zustand
+- ▁suspicious
+- 脉
+- ▁obstacle
+- 競技
+- 椅子
+- ▁Maß
+- ▁temporary
+- 的任何
+- ▁persuaded
+- 如果你想
+- 逐
+- ▁cambia
+- ウン
+- ό
+- ▁abbiamo
+- ▁vary
+- ▁понятно
+- ぐらいの
+- ものの
+- 辈
+- ▁commenced
+- ▁Schrei
+- ligt
+- 主题
+- 都知道
+- schule
+- ▁separa
+- web
+- ▁Kaj
+- 飞行
+- thon
+- ▁nnyo
+- ump
+- ▁Service
+- ▁irgendeine
+- eus
+- 件事
+- 这么大
+- 的老
+- ▁protected
+- ▁effectively
+- 分かります
+- dür
+- 主要是
+- кус
+- ▁enorme
+- ष
+- エネルギー
+- ▁sunset
+- нэ
+- wah
+- ▁призна
+- кино
+- ▁deinen
+- ▁costume
+- くない
+- کو
+- ▁homo
+- ▁کردن
+- tò
+- ума
+- もありました
+- いきましょう
+- ▁seats
+- ▁Dort
+- ▁mars
+- ▁supporting
+- ▁replaced
+- ▁hammer
+- ▁Two
+- ▁interven
+- sley
+- ▁машин
+- 把这
+- ▁genauso
+- مۇ
+- нская
+- ▁furnish
+- ▁units
+- 的爱
+- ▁strangely
+- 먹
+- steuer
+- 受伤
+- 関係者
+- 是这样的
+- 失去了
+- ▁اینجا
+- 分け
+- ທ
+- ▁globe
+- 市場
+- 播客
+- иг
+- 资本
+- ▁contents
+- hanga
+- zicht
+- ▁großes
+- 还会
+- ▁jove
+- кер
+- ▁saints
+- 丰富
+- ▁Vol
+- ▁fazer
+- ▁mou
+- دون
+- ▁Ntabwo
+- ▁Pass
+- ▁fires
+- ▁devices
+- ▁tienen
+- нымі
+- ▁mademoiselle
+- ガス
+- taba
+- ▁achter
+- 鬼子
+- 还真
+- 買い
+- ▁rays
+- ▁Cent
+- ▁buryo
+- ▁cual
+- ▁amigo
+- ▁murdered
+- んでいる
+- ▁vulgar
+- elte
+- ▁seas
+- ▁либо
+- пуска
+- ▁aceste
+- 申请
+- そば
+- ▁Brit
+- ▁bitten
+- 的历史
+- ▁Natürlich
+- ▁Texas
+- 炭
+- ▁Schwester
+- ▁блок
+- ▁neighbors
+- ாள்
+- 一時
+- ▁двух
+- rita
+- ▁effet
+- ▁часть
+- מ
+- ▁speziell
+- ▁групп
+- われ
+- 亲自
+- جی
+- ashobora
+- ▁armies
+- tuk
+- ▁succession
+- 上昇
+- ▁Minister
+- 一些人
+- ৃ
+- 究
+- ▁wing
+- いう
+- 的呀
+- anthrop
+- の一部
+- ▁fountain
+- 兼
+- ▁minut
+- りとか
+- 初の
+- 把他们
+- ว่า
+- ▁Design
+- ▁familia
+- 神秘
+- lui
+- ▁testimony
+- 教えて
+- ▁abruptly
+- ▁Une
+- ▁decidi
+- riet
+- 类型的
+- ▁meetings
+- 받
+- 월
+- рай
+- space
+- iadau
+- ▁pall
+- ▁antwoordde
+- 職員
+- مز
+- каў
+- think
+- ▁insbesondere
+- ▁solitary
+- ヨーロッパ
+- ▁responded
+- 中国的
+- ▁puzzled
+- 逆に
+- ▁Mul
+- 赢得
+- గ
+- сөн
+- 不是我
+- ▁sauce
+- ついて
+- ▁diferente
+- life
+- 民主党
+- ▁pronounced
+- までに
+- 公民
+- ▁hohe
+- ▁diverse
+- ▁namely
+- ▁weigh
+- pian
+- 这张
+- ▁thereby
+- வீ
+- ▁tragedy
+- が出て
+- ▁但是他
+- 谷歌
+- іцца
+- ök
+- nika
+- ▁chứ
+- 剪
+- ▁offices
+- 你们俩
+- ▁channels
+- zette
+- sozi
+- ▁twi
+- 公園
+- gize
+- rata
+- 養
+- ▁scott
+- ▁grip
+- ▁longing
+- ▁texas
+- ▁diamond
+- 事業
+- 这场
+- ▁Kä
+- bou
+- 想过
+- ுள்ள
+- 一遍
+- 尝
+- schutz
+- ▁clara
+- ▁уу
+- вяр
+- 善良
+- ▁memories
+- üyor
+- ▁producing
+- ▁sadly
+- ▁wink
+- 生存
+- 现在的
+- 协议
+- ▁scent
+- ىنى
+- hine
+- ▁veces
+- ▁pří
+- ▁Ng
+- 购
+- ▁Werk
+- ▁cog
+- 行政
+- ワールドカップ
+- ▁Schmerz
+- ▁milit
+- liği
+- ▁neglect
+- pier
+- 黑色
+- ▁என்
+- ▁functions
+- ▁harvest
+- 시간
+- ▁floating
+- が今
+- ▁metro
+- lingen
+- яць
+- ▁partially
+- ▁indication
+- ▁Rei
+- スタッフ
+- ▁Sub
+- 予報
+- ▁guilt
+- gesehen
+- 的学生
+- ▁sergeant
+- mari
+- ады
+- Wa
+- 平等
+- 房间里
+- 않
+- ▁documents
+- ▁cavalry
+- ▁பல
+- fashioned
+- ▁sets
+- бой
+- ▁nehme
+- த்தின்
+- ▁پیش
+- क्
+- 广播
+- ае
+- 动作
+- ▁integration
+- 石头
+- ▁результат
+- ▁amas
+- ▁brig
+- দে
+- 饼
+- ▁fiction
+- 題
+- ▁кстати
+- ▁shu
+- ▁பிர
+- しまって
+- flex
+- eorang
+- made
+- рал
+- shop
+- 領
+- scheiden
+- ĉa
+- 旁
+- 舍
+- ї
+- 不是说
+- truc
+- schiff
+- ▁Scott
+- ія
+- ▁roses
+- ▁但是在
+- 胆
+- ▁Ра
+- داری
+- 我们想
+- Ab
+- ▁trabajo
+- に関する
+- ▁defect
+- ▁Gil
+- ერ
+- ▁wandered
+- 会談
+- ckle
+- 勝負
+- Au
+- meli
+- ▁bose
+- 飲み
+- šo
+- 植
+- ▁애들
+- 所说
+- vian
+- ▁لقد
+- 作って
+- ▁satisfy
+- ffle
+- ▁Arzt
+- ▁wann
+- ▁deeds
+- ▁myth
+- ▁Nummer
+- ▁criticism
+- ▁Kro
+- 故意
+- 单位
+- ▁맞아
+- ستا
+- ▁guardian
+- ▁indignation
+- Ver
+- ▁recollection
+- 会見
+- ▁whoever
+- break
+- 気温が
+- সি
+- ▁gleichzeitig
+- 陛下
+- ▁delicious
+- 开始了
+- 读者
+- beeld
+- 起了
+- ▁illegal
+- 地面
+- ▁yeni
+- ▁enjoyment
+- ▁outrage
+- 并在
+- ▁flee
+- 屏幕
+- больш
+- ▁interact
+- ▁momentum
+- ▁hano
+- ▁inspiration
+- liches
+- ලා
+- iser
+- ▁glasses
+- 加上
+- 暑
+- орон
+- 耐
+- ▁dramatic
+- ▁равно
+- hla
+- gó
+- 协
+- 障碍
+- 倉
+- ▁sunk
+- ▁шестнадцать
+- ▁tendency
+- ▁Island
+- するなど
+- 预测
+- よい
+- ▁unbedingt
+- ▁bleibt
+- ▁шмат
+- ▁disp
+- angira
+- arca
+- ▁crimes
+- onne
+- ▁loyal
+- ▁apostle
+- ▁betray
+- оны
+- ▁inward
+- ilia
+- 烦
+- 的日子
+- ▁사람
+- lg
+- 見る
+- rinner
+- 沉默
+- ▁seize
+- ஞ்ச
+- iler
+- ▁правильно
+- なあ
+- ▁بسیار
+- 媳妇
+- ▁говоря
+- 主意
+- 물
+- ▁ruhig
+- ▁Turn
+- 不出
+- сом
+- ▁Nel
+- 紫
+- unddreißig
+- を務め
+- unta
+- ▁Autor
+- ▁diana
+- ▁alexander
+- ▁telegraph
+- ▁ساخت
+- нен
+- ▁amen
+- 焦虑
+- ▁Artikel
+- аар
+- حة
+- ▁seves
+- ▁пост
+- һә
+- 天下
+- Un
+- 明显
+- ▁لدي
+- лаг
+- ▁Rand
+- ▁personality
+- hart
+- abandi
+- っていました
+- zek
+- ▁arrangements
+- بد
+- ▁sermon
+- が多く
+- 的照片
+- ▁ruined
+- ahu
+- なと
+- 智慧
+- ัง
+- ▁теле
+- tou
+- ▁geheel
+- ению
+- ▁своих
+- ▁virtues
+- ▁widely
+- ▁mingled
+- ▁Kir
+- 我们没有
+- ccio
+- ▁Club
+- 我们不
+- езд
+- してた
+- ▁rider
+- teilung
+- كو
+- 一片
+- ▁женщин
+- سان
+- 重复
+- 生日
+- ▁approved
+- 绕
+- 機能
+- ishi
+- ograph
+- 讲述
+- 危机
+- 広がって
+- 获得了
+- رە
+- 意见
+- 妇
+- ▁loaded
+- cation
+- いいですか
+- ▁fran
+- услов
+- ▁pastor
+- 统
+- 提供了
+- 再開
+- ▁puff
+- ն
+- 障害
+- ▁financi
+- 给大家
+- kund
+- ▁tune
+- arro
+- ▁없어
+- ▁levi
+- lade
+- 今も
+- ▁yea
+- ▁serpent
+- ▁endless
+- いいですね
+- いき
+- ▁Where
+- ▁Beste
+- riz
+- 离开了
+- 径
+- ränk
+- ному
+- 伸び
+- ▁Fähigkeit
+- ▁Schlaf
+- ▁duchess
+- 你和
+- ▁لأن
+- 就这么
+- ▁executed
+- ▁heels
+- ▁chicago
+- 성
+- ▁vader
+- zahlen
+- ▁periods
+- called
+- just
+- ▁тринадцать
+- さい
+- dron
+- 改变了
+- रि
+- ▁vera
+- ▁sharing
+- inta
+- ▁Feld
+- 負
+- ▁ihe
+- кров
+- dog
+- ▁чт
+- pool
+- ▁dreamed
+- ▁heilig
+- ▁slender
+- 在家里
+- ▁uncon
+- മ
+- ▁그리고
+- ▁Kandi
+- Union
+- 英里
+- ▁chances
+- ryo
+- 场景
+- 遥
+- 链接
+- 移民
+- 十四
+- ▁sac
+- fact
+- ▁selten
+- ▁Arab
+- 谈到
+- ▁feather
+- ▁inch
+- 尽可能
+- 都被
+- 日常
+- ▁attorney
+- human
+- ▁helpful
+- 劳动
+- 经营
+- ▁mă
+- ▁Cro
+- ▁languages
+- ▁Esperanto
+- лев
+- ▁sustainable
+- タン
+- 生産
+- 纵
+- うちの
+- ▁promotion
+- lateral
+- stamm
+- making
+- 華
+- ▁risen
+- geist
+- ▁vehicle
+- 大量的
+- сна
+- ▁careless
+- ваю
+- 村さん
+- 一刻
+- 育て
+- ▁Ча
+- 取り組み
+- ▁passengers
+- 雇
+- း
+- ▁chimney
+- ▁엄청
+- ▁abundant
+- ใน
+- 调整
+- 場で
+- moc
+- ▁sao
+- ▁Generation
+- 男女
+- ▁provin
+- ▁funktionieren
+- ▁limp
+- ▁calmly
+- gesellschaft
+- ▁Ram
+- ▁quello
+- ▁tv
+- ▁agents
+- 、4
+- あら
+- 和他们
+- ▁беларус
+- ▁Bezirk
+- どっち
+- ▁villages
+- ▁intervals
+- ▁cooking
+- ▁printed
+- ▁revelation
+- ▁Gene
+- 也就
+- ǧ
+- 今シーズン
+- 女王
+- ▁mwen
+- 我们对
+- стер
+- 上で
+- 链
+- ▁politique
+- ▁sunlight
+- ▁differently
+- god
+- 보고
+- 暑さ
+- 对她
+- ▁dost
+- 图像
+- ▁poden
+- igno
+- çu
+- ▁holes
+- かれ
+- 多久
+- ▁consumers
+- ▁sweat
+- bac
+- 喝酒
+- ▁کم
+- ▁movies
+- ▁одной
+- τ
+- テン
+- arian
+- ▁verdad
+- قول
+- 司机
+- 承担
+- ▁exceed
+- низ
+- 今日も
+- 措施
+- ▁clan
+- uso
+- ▁Last
+- 回应
+- ▁nowhere
+- ência
+- ▁islam
+- ɣa
+- gende
+- بان
+- 報
+- quir
+- ▁Lehr
+- Ц
+- ▁nerve
+- ▁january
+- 批判
+- 很大的
+- 男朋友
+- 通信
+- کت
+- 栏
+- いきました
+- ▁путин
+- 町の
+- 混合
+- ▁absorbed
+- wami
+- もともと
+- èn
+- 有趣
+- ▁hinunter
+- state
+- ▁fathers
+- ▁sweep
+- 情報を
+- ▁alcohol
+- bonye
+- ggy
+- රි
+- fitting
+- isierung
+- aq
+- ▁exposure
+- 紹介
+- iker
+- indu
+- ▁friday
+- цыя
+- brü
+- ▁faculty
+- ▁convict
+- ▁tad
+- 誰か
+- نظر
+- 新型コロナ
+- 업
+- ို
+- 遺体
+- ▁addressing
+- 什么的
+- たって
+- ▁veu
+- ▁presents
+- ▁depression
+- ハイ
+- ▁Suche
+- 财产
+- ▁Ту
+- 前半
+- ▁joan
+- を引き
+- стрел
+- 田中
+- ▁user
+- ▁constitu
+- ▁Publikum
+- ▁Tour
+- ەکە
+- 機関
+- ▁muốn
+- овал
+- ▁moonlight
+- ▁welke
+- まだまだ
+- ▁которой
+- ▁participa
+- 彻底
+- ▁discussions
+- ▁году
+- idée
+- 醉
+- 出して
+- ▁Nachrichten
+- fordert
+- ▁proces
+- 高さ
+- ▁Esto
+- 예
+- 切って
+- ▁Inhalt
+- ▁fortunately
+- ево
+- ndre
+- ▁قو
+- arı
+- 你都
+- فه
+- 披露
+- market
+- ▁сами
+- tano
+- 了一下
+- 平衡
+- '150'
+- gekommen
+- 多大
+- ተ
+- ▁lieber
+- zas
+- ▁demands
+- 多少钱
+- べき
+- 飛び
+- шә
+- ▁stare
+- 他们有
+- abantu
+- 하지
+- ▁hı
+- 魂
+- ▁supplied
+- 豚
+- шен
+- ▁preparation
+- bing
+- ▁heavenly
+- 規
+- ▁nhất
+- selt
+- ▁გა
+- ▁gig
+- ▁можем
+- 銃
+- ضا
+- ▁raus
+- ▁Ergebnis
+- ▁forests
+- ▁erstellen
+- ▁Fle
+- ஸ
+- ▁behaviour
+- 年轻的
+- lder
+- энд
+- тел
+- யான
+- ▁Please
+- ▁travelling
+- 多么
+- volution
+- 더라고
+- ▁größere
+- を変え
+- ▁procedure
+- 遭
+- ▁Material
+- ▁campus
+- を行って
+- 級
+- руу
+- roch
+- ▁Mä
+- 現在の
+- 你已经
+- чын
+- ▁erwartet
+- ▁requirements
+- 就不会
+- 専門
+- médi
+- 转向
+- Lachen
+- 未来的
+- ▁commune
+- ▁Fluss
+- 孝
+- tega
+- ▁Zimmer
+- 我认为这
+- ▁Cra
+- レース
+- 状况
+- ▁جدا
+- 一大
+- 不是你
+- зор
+- ▁choses
+- ▁Tie
+- ▁representative
+- ▁pent
+- ▁knot
+- 有点像
+- െ
+- ▁slim
+- ▁Sho
+- っか
+- чных
+- 法案
+- ța
+- ▁barbari
+- ▁Sto
+- ▁promptly
+- камі
+- изм
+- oog
+- ▁verme
+- ▁adorn
+- やめ
+- ▁commence
+- ▁Hör
+- schritt
+- emt
+- ところに
+- ▁vz
+- ▁тип
+- 债
+- леч
+- がん
+- ▁добра
+- abend
+- excel
+- ▁aceast
+- のもの
+- 你真的
+- ▁invented
+- 矛盾
+- ▁neutral
+- と思うんですけど
+- erna
+- rf
+- zima
+- 规模
+- ワー
+- 喺
+- ▁currency
+- ▁eagle
+- lom
+- magn
+- ▁claimed
+- ▁pockets
+- 점
+- etto
+- 的目标
+- ▁হয়
+- ▁blade
+- ▁laŭ
+- евская
+- 重要な
+- dī
+- ▁кабинет
+- ▁zéro
+- ▁francs
+- 大多数人
+- вил
+- ▁identified
+- 你可
+- ▁logic
+- 你给我
+- تس
+- axe
+- ▁achievement
+- ▁contribute
+- ▁Ober
+- 反応
+- 宝贝
+- もあり
+- ▁itu
+- 基地
+- ▁admiral
+- nydd
+- shyira
+- 中学
+- drin
+- teen
+- こんなに
+- رات
+- 脸上
+- ში
+- ▁colleagues
+- 同情
+- ▁random
+- 覆
+- 挤
+- 声が
+- ▁bewegt
+- 结束了
+- 安静
+- 영
+- ▁weiteren
+- ▁okul
+- ▁начальник
+- erfolg
+- ▁grava
+- ▁pove
+- ▁Pod
+- 的数据
+- ▁typical
+- クリ
+- schaffen
+- 也好
+- ▁wrought
+- জন
+- kommt
+- だったり
+- dust
+- terra
+- 講
+- ▁bushes
+- ▁muß
+- ユ
+- ▁хэл
+- ▁waist
+- 我一
+- 跌
+- 工业
+- ▁dave
+- ئو
+- 回头
+- ▁Radio
+- '600'
+- ▁confirmed
+- ということが
+- 外国
+- 来讲
+- ža
+- chal
+- ▁mouse
+- 仲間
+- ▁alert
+- ላ
+- ▁Council
+- 肌
+- ▁beginnt
+- 铺
+- ▁Day
+- ▁hired
+- 全国の
+- だい
+- 伟
+- ук
+- ▁condemned
+- いよいよ
+- ▁stap
+- ▁música
+- ithi
+- ▁conform
+- performing
+- ▁incredibly
+- ▁cresc
+- ▁lingvo
+- 扭
+- 胃
+- ▁thumb
+- ▁honestly
+- isk
+- ffel
+- ▁specimen
+- glass
+- はこちら
+- ▁spark
+- munt
+- 西方
+- ▁grupo
+- 你不会
+- ▁söyle
+- histoire
+- વ
+- 随便
+- rama
+- ಕ
+- なきゃいけない
+- ▁circuit
+- の声
+- ▁Rue
+- 练习
+- ▁prayed
+- 更大的
+- 納
+- برا
+- ほかの
+- 회
+- ▁entdeckt
+- ▁команд
+- фор
+- 这也
+- ▁nasıl
+- nehmer
+- 相关的
+- 原谅
+- рыв
+- 送到
+- ▁deutsche
+- IS
+- ▁purse
+- 上有
+- ▁soziale
+- ▁rhe
+- 代码
+- ует
+- 輪
+- ▁Hallo
+- 比我
+- iest
+- かかって
+- ▁tid
+- posició
+- ▁daring
+- してくれ
+- sprach
+- 東部
+- 毎
+- ▁agency
+- poj
+- 知って
+- gamba
+- ▁carr
+- 指导
+- ▁yeux
+- 飛
+- ▁followers
+- 贾
+- ▁obedience
+- ▁spät
+- ▁mischief
+- 報道
+- 川さん
+- action
+- 迎
+- ▁resort
+- ▁deals
+- ▁warned
+- 摔
+- 丑
+- ▁Einzel
+- லாம்
+- ほどの
+- ▁mob
+- ああ
+- もんね
+- jed
+- 하면
+- precia
+- 百万
+- ▁represents
+- российск
+- ▁entitled
+- ▁derived
+- ▁lace
+- ▁resource
+- ▁certainty
+- ▁gott
+- 又是
+- ▁gallop
+- ジュ
+- iwa
+- 電気
+- 这就是我
+- руч
+- ▁dolan
+- npr
+- Il
+- ▁coarse
+- ▁процесс
+- みます
+- ▁tenderness
+- ▁charges
+- 野さん
+- ▁emerge
+- umber
+- 喊道
+- 检
+- ▁satu
+- ▁sahen
+- є
+- ▁notwendig
+- தோ
+- ▁flames
+- 士兵
+- ▁Seele
+- خبر
+- ▁Nhưng
+- ▁från
+- ▁کښې
+- ▁아니야
+- ▁affectionate
+- ▁curtain
+- ▁detailed
+- やら
+- 我叫
+- 一系列
+- ▁пара
+- oce
+- ▁exempl
+- 范围
+- がいる
+- sions
+- ▁artists
+- ària
+- 凭
+- ▁divorce
+- ▁хоёр
+- ▁corridor
+- ▁lofty
+- 最高気温
+- roma
+- 陷入
+- 開け
+- ▁nivel
+- 集团
+- waka
+- ב
+- ▁Feind
+- 损失
+- ▁орган
+- mite
+- polo
+- 内の
+- ▁Zugang
+- 使って
+- ▁fein
+- ▁numéro
+- ▁allí
+- ற்ப
+- ▁pressing
+- afrika
+- ষ
+- dores
+- 可以看到
+- ▁roughly
+- ▁стара
+- 技能
+- ▁savings
+- ▁farmers
+- ▁theater
+- glück
+- magnet
+- ▁când
+- ▁barely
+- ▁briefly
+- вялі
+- போது
+- ôn
+- 児
+- احت
+- と呼ばれる
+- cloth
+- ▁aka
+- 和一个
+- ▁painter
+- ▁challenging
+- 初めての
+- ▁borrow
+- 完美
+- ▁pillar
+- ί
+- ரோ
+- ▁Frei
+- 军事
+- ▁advertising
+- ▁solitude
+- ことができる
+- やってる
+- 力が
+- ▁bargain
+- のではないか
+- 不少
+- ジャンプ
+- ბა
+- ビル
+- жо
+- ▁sees
+- يح
+- ▁normalerweise
+- bih
+- ▁před
+- ▁telefon
+- ▁injured
+- ▁marsh
+- uburyo
+- ▁правда
+- 就把
+- ▁газ
+- 瞧
+- ▁moves
+- के
+- ▁writes
+- ▁advocate
+- なんですよね
+- ▁sebagai
+- ▁wherefore
+- ▁hoop
+- maga
+- ▁sphere
+- 来月
+- guin
+- 咬
+- ▁underlying
+- ▁olduğunu
+- にいる
+- 主动
+- ▁plusieurs
+- みんなで
+- ً
+- nwa
+- 燃
+- lijke
+- 私人
+- ▁Studie
+- 例子
+- ূ
+- 牢
+- ▁happily
+- 따
+- ciò
+- ▁идет
+- ▁парти
+- を入れて
+- ▁ranks
+- īt
+- ctic
+- 就是这样
+- ▁blows
+- 有用
+- ▁colon
+- 吵
+- 找你
+- ▁pensar
+- кім
+- 彼は
+- 迹
+- ▁organized
+- 柄
+- дог
+- ▁countess
+- niu
+- liest
+- ▁construct
+- ో
+- ▁newly
+- ▁Kontrolle
+- ▁சிற
+- ▁nahe
+- ▁тү
+- стей
+- 指摘
+- ▁announcement
+- ▁tenant
+- ▁mexico
+- おはようございます
+- vuit
+- 突破
+- ▁видел
+- terie
+- ▁energ
+- houd
+- 봤
+- ▁harbor
+- տ
+- ▁verlieren
+- avaient
+- aquest
+- ▁Dal
+- 优势
+- ▁deja
+- ember
+- 讨
+- 竟然
+- ▁slecht
+- initiative
+- ehrt
+- るんです
+- ativ
+- ▁Studien
+- letter
+- 观点
+- 到底是
+- ▁stond
+- ▁penetrate
+- ▁dividend
+- ▁investors
+- 警方
+- ▁иван
+- 面白い
+- っちゃった
+- ▁বল
+- ▁revolver
+- ateur
+- руш
+- misch
+- ▁vengeance
+- シュ
+- quatre
+- ▁wheat
+- shaw
+- ▁nawe
+- trop
+- cret
+- があると
+- 指出
+- рә
+- ▁simon
+- 创作
+- ▁thither
+- ивать
+- ▁beschäftigt
+- ▁brethren
+- igita
+- ▁Nie
+- 这个词
+- ▁Brown
+- قط
+- ribu
+- ▁bitterly
+- ▁sofa
+- ▁component
+- ▁ventured
+- これだけ
+- bora
+- 很久
+- 았
+- ▁Religion
+- 缺乏
+- ▁Pin
+- 咱俩
+- ▁panic
+- 很明显
+- zorg
+- 的父亲
+- ▁survive
+- 一辈子
+- ▁gravely
+- ▁Tagen
+- ▁counted
+- ▁positively
+- 召
+- дым
+- ූ
+- ▁dacă
+- 関連
+- ினார்
+- 沿着
+- ▁bisher
+- はありません
+- だろ
+- ▁elf
+- ް
+- 共产党
+- ▁clutch
+- anten
+- 咲
+- ▁casi
+- یل
+- ▁conversion
+- 知道的
+- ▁warten
+- ьте
+- ▁پای
+- feuer
+- 瓜
+- ▁بۆ
+- ▁tenir
+- ▁Ну
+- 不明白
+- ▁napoleon
+- の映像
+- 两种
+- 自从
+- 幸せ
+- hib
+- jwe
+- ▁podemos
+- ▁pirate
+- ▁bedeuten
+- лова
+- 我得
+- 切れ
+- orden
+- つ目
+- ▁weeping
+- をつけて
+- ছি
+- 衰
+- ▁интересно
+- ▁tradu
+- ▁сме
+- 互动
+- ស
+- 행
+- がい
+- ▁versuchte
+- 恒
+- 르
+- ▁Anzahl
+- 彼ら
+- ▁practices
+- ▁мед
+- menti
+- ▁weiterhin
+- ▁restore
+- ologia
+- ▁computers
+- ▁algun
+- 用来
+- 干净
+- school
+- аза
+- ▁vivo
+- ▁Grundlage
+- ukan
+- なくても
+- ▁estado
+- ▁attempts
+- ▁Ĉu
+- ▁sample
+- bund
+- ▁analog
+- ▁celle
+- ▁salva
+- ▁providence
+- すばらしい
+- ▁House
+- ▁trim
+- ▁erkannt
+- ▁rushing
+- 旋
+- ゼ
+- ▁bulk
+- はおよそ
+- 世界中
+- ▁findet
+- cultura
+- زو
+- ▁instances
+- зан
+- ගේ
+- であり
+- енных
+- ▁santa
+- ▁jewish
+- gele
+- ▁gusta
+- ▁condemn
+- '5000'
+- 司法
+- führer
+- ▁sais
+- үл
+- 東京オリンピック
+- に入り
+- 诚
+- ධ
+- ▁nouvelle
+- ▁invite
+- ovat
+- ▁treaty
+- ▁navy
+- urteil
+- minate
+- іх
+- 襲
+- ▁mexican
+- ggle
+- اي
+- ▁germans
+- ▁aqui
+- 在我们的
+- ▁canvas
+- ▁selection
+- ijo
+- 送り
+- ▁blaze
+- lığı
+- ▁gesund
+- gora
+- ▁آب
+- ▁Who
+- 知り
+- nız
+- ▁ôl
+- ▁suck
+- ▁displayed
+- えば
+- ▁undoubtedly
+- 的过程
+- طة
+- ▁Así
+- ▁warrant
+- ▁ursprünglich
+- ▁کردم
+- пен
+- 婴儿
+- 蛇
+- ▁هغه
+- ▁illustrate
+- 事実
+- ▁bureau
+- fata
+- なか
+- ▁Richter
+- mbwa
+- 费用
+- 那就
+- ▁Teile
+- ▁Daniel
+- ▁nên
+- ▁Ад
+- ▁deploy
+- ký
+- Č
+- 純
+- ▁entscheiden
+- ▁چیز
+- ▁Reaktion
+- ▁можете
+- währ
+- アイ
+- က
+- ▁acquire
+- лив
+- ▁caesar
+- 有事
+- ské
+- ▁doth
+- ▁tête
+- 蓝色
+- ▁conversations
+- ▁wept
+- 入れる
+- ▁politischen
+- ▁jerk
+- ▁elaborate
+- 然后你
+- ▁весь
+- ▁ladder
+- に行って
+- 分开
+- ▁advancing
+- macht
+- ▁které
+- 電力
+- Wi
+- дыр
+- ▁hamwe
+- とう
+- ▁lasted
+- タイミング
+- ▁afite
+- 报纸
+- 事態
+- ▁Ker
+- 漂
+- ▁employer
+- ▁discourse
+- ▁cannon
+- ▁Kommission
+- yat
+- shim
+- ▁которое
+- valo
+- даў
+- istoj
+- breng
+- ▁prevail
+- ニング
+- ▁julie
+- nine
+- ljen
+- вым
+- ▁pension
+- 縮
+- ▁сум
+- ▁kleines
+- ▁publish
+- lden
+- ▁alex
+- ▁niece
+- 閉
+- に来て
+- ▁simplicity
+- もらった
+- 吃的
+- ▁idol
+- ester
+- ਰ
+- はね
+- ээс
+- 针对
+- 阳光
+- たく
+- 書き
+- ▁brood
+- ୍
+- が続く
+- craft
+- 专门
+- 調整
+- んですけども
+- 批评
+- ▁dependent
+- 沟通
+- 全面
+- 两千
+- 掌握
+- ▁других
+- kere
+- 으니까
+- 基金
+- iska
+- ▁kang
+- empresa
+- ▁daniel
+- ▁четыреста
+- 着你
+- 반
+- utilitza
+- 会被
+- ▁riu
+- 主張
+- نية
+- philosoph
+- ▁munsi
+- ▁помог
+- طل
+- ▁measured
+- 稍微
+- ▁وقت
+- 第四
+- ▁flows
+- ▁hogy
+- 都很
+- ▁Act
+- までは
+- 朋友们
+- ▁trials
+- ▁schlug
+- 符合
+- 目標
+- ▁shower
+- シーン
+- チョ
+- нные
+- 分手
+- ▁kent
+- ެއް
+- ින්
+- ▁владимир
+- ▁Angelegenheit
+- ▁ascertain
+- ▁gaat
+- ▁الث
+- oja
+- ▁birthday
+- 続いては
+- ▁robin
+- 緊張
+- ▁Fol
+- ▁сделал
+- ▁amused
+- plex
+- 刑事
+- 我还是
+- 储
+- ▁Zahlen
+- やす
+- ▁frozen
+- ごろ
+- このまま
+- ▁sailors
+- ▁önce
+- lite
+- ▁eerste
+- 补充
+- 走向
+- ▁».
+- 現地
+- 愚蠢
+- Lo
+- gă
+- artagnan
+- ▁suite
+- іў
+- 有两个
+- 机器人
+- ▁relate
+- ▁fisher
+- 财富
+- ▁vanity
+- 尊
+- 凶
+- ɛa
+- 谋
+- umba
+- きています
+- ▁Wild
+- ▁magistrate
+- ▁مرد
+- Ich
+- مب
+- appel
+- 姿勢
+- 疲
+- પ
+- yez
+- 工程师
+- 乎
+- მი
+- ▁exemple
+- ▁Even
+- ▁ample
+- 混ぜ
+- ▁Poli
+- ▁Ari
+- 皇帝
+- ▁schr
+- ▁picking
+- 庆
+- ▁quella
+- يز
+- 保護
+- 結局
+- блі
+- ▁Cur
+- 窗户
+- ическая
+- মান
+- '31'
+- 你来
+- ▁семнадцать
+- ▁scout
+- ▁Lebensmittel
+- 收听
+- Qu
+- 上次
+- قوم
+- கால
+- جميع
+- wür
+- ▁Good
+- 建设
+- 世界の
+- δ
+- ▁raison
+- ▁Mike
+- нный
+- шо
+- idio
+- るため
+- ვა
+- fällt
+- ▁declined
+- geza
+- 如果您
+- 大声
+- すべて
+- いるのは
+- 赢
+- ▁obeyed
+- ▁depending
+- ▁colony
+- 側に
+- стью
+- ក
+- 워
+- osten
+- ▁flourish
+- 是如此
+- を奪
+- ▁كنت
+- ▁swing
+- 脾气
+- ▁garments
+- 前进
+- brid
+- ▁Și
+- ▁sarah
+- 鼻
+- ▁Ny
+- Ja
+- ▁swallow
+- ▁entreat
+- pran
+- ▁angesehen
+- 警告
+- ▁experiments
+- 坏了
+- 援
+- kunst
+- ▁correspond
+- ▁быстро
+- 首相
+- ziert
+- ですからね
+- ▁judges
+- ▁passe
+- ▁frontier
+- 玩儿
+- ▁байх
+- の影響
+- gericht
+- zeichen
+- bari
+- 两位
+- ▁colours
+- 唯一的
+- 実現
+- 鳴
+- 寻求
+- 評価
+- ▁impre
+- ▁aspects
+- ▁zeigte
+- ▁compte
+- ▁secrets
+- ması
+- 句话
+- ifik
+- ň
+- 策略
+- ▁Henry
+- 一口
+- ▁Бо
+- ប
+- ං
+- ▁chính
+- gier
+- বার
+- нию
+- ケース
+- ▁fanny
+- ▁curs
+- 烤
+- wedd
+- bian
+- ▁twist
+- ▁mère
+- ▁почти
+- сим
+- 日本は
+- ▁idiot
+- ▁hoog
+- лиз
+- ▁comprehend
+- ▁Wald
+- 可怜的
+- 引用
+- 统治
+- тик
+- ▁들어
+- ▁lawn
+- きっかけ
+- práv
+- らし
+- なと思って
+- ▁ceiling
+- ijwe
+- ▁그럼
+- ▁susan
+- 側の
+- のところ
+- ▁проект
+- だという
+- 汇
+- ▁Freiheit
+- 좋
+- ▁تت
+- ▁trembled
+- ▁Bil
+- 祈祷
+- 樹
+- estima
+- cze
+- ▁goat
+- ße
+- ことです
+- ▁sai
+- 俊
+- ▁сюда
+- ▁Rose
+- ▁install
+- 威尔
+- bett
+- ▁Indian
+- ▁luxury
+- 聖
+- 芝
+- 投稿
+- ▁Dun
+- ▁повер
+- などと
+- dec
+- ▁emma
+- ▁uncertainties
+- lok
+- っしゃ
+- ▁signor
+- 频道
+- 딱
+- ▁ئۆ
+- ▁servir
+- 復
+- 仅
+- ብ
+- 委員会
+- agira
+- ாய்
+- 我们能
+- ▁ministers
+- を加え
+- 使用的
+- paga
+- ▁urge
+- ▁phenomena
+- 翼
+- ahan
+- 指導
+- bruch
+- 标志
+- abord
+- osos
+- ▁essence
+- ए
+- atori
+- ▁dried
+- fica
+- ▁ratio
+- いますが
+- ▁Lang
+- 半年
+- 裁判
+- ▁drops
+- ▁вось
+- ▁Änderung
+- 芬
+- ▁Beginn
+- ▁emily
+- ▁дур
+- ▁investing
+- ديد
+- әк
+- 운
+- ▁skills
+- ▁potentially
+- ▁herunter
+- リード
+- ▁revolt
+- ▁soit
+- ▁Bor
+- メン
+- 权利
+- ▁vertical
+- ▁believing
+- хі
+- 自治体
+- ▁считаю
+- 金メダル
+- ▁groan
+- ることができ
+- 演出
+- களுக்கு
+- үн
+- rog
+- kubwa
+- веч
+- ifer
+- ▁emphasis
+- 卓
+- getragen
+- ▁ليس
+- ▁pacific
+- なくなる
+- ▁bisa
+- 扇
+- ▁Jen
+- ▁страны
+- ▁alarmed
+- 署
+- ▁Sorgen
+- 係
+- ▁كما
+- ▁باشد
+- ▁wheels
+- 最後
+- 共和党
+- ▁tho
+- ива
+- 鼠
+- 半分
+- yum
+- 分かりました
+- ugi
+- はその
+- フレ
+- ▁September
+- ▁вечер
+- と言って
+- takse
+- ▁voller
+- šu
+- éra
+- 涉
+- banga
+- ▁bass
+- ▁Diskussion
+- eres
+- station
+- ▁forming
+- ▁attacks
+- ▁renew
+- ▁centro
+- ggs
+- ყ
+- ▁diminish
+- ▁faded
+- ▁instruction
+- ルド
+- сал
+- ▁wach
+- ▁sozialen
+- 的部分
+- 増加
+- 相似
+- 净
+- 泣
+- 在这种情况下
+- 味道
+- bourg
+- を確認
+- 叫我
+- ▁steve
+- 师父
+- kumi
+- ▁Meer
+- 一代
+- ▁Städte
+- ▁erwarten
+- 小子
+- ▁folded
+- ▁Mut
+- fähr
+- اخ
+- ▁sleeve
+- ড
+- ▁subsequent
+- ▁bessere
+- トロ
+- rong
+- ▁лично
+- сту
+- ест
+- ▁weer
+- ▁giới
+- ▁назва
+- ▁sara
+- ▁acknowledged
+- bier
+- ▁Boston
+- ▁مد
+- 孤
+- 姨
+- ▁Risiko
+- غان
+- 犯人
+- ▁transmit
+- ▁regularly
+- ▁власть
+- 做到这一点
+- 確かに
+- спор
+- ▁finde
+- грам
+- 可以说
+- ▁highway
+- 非洲
+- ▁crush
+- ▁tett
+- 横浜
+- 月份
+- ▁gelesen
+- なくなった
+- ▁whale
+- fällig
+- ▁ripe
+- ▁wherein
+- 이랑
+- ダメ
+- 勉強
+- ▁nang
+- ▁Mitglied
+- ▁famille
+- 繰り返し
+- カード
+- 智能
+- dain
+- ехал
+- 罗马
+- 複
+- 訳
+- '99'
+- 实在
+- 楽しい
+- 이야
+- 発言
+- 命运
+- aktiv
+- ▁dense
+- ▁Up
+- ▁reside
+- 大家好
+- ▁dug
+- ▁genetic
+- 方の
+- の間
+- ▁primitive
+- ▁condu
+- 听听
+- ▁dropping
+- 進出
+- या
+- まい
+- тек
+- ▁lowest
+- ▁хүн
+- ▁Kann
+- ▁trova
+- 妇女
+- шир
+- ▁sandy
+- ▁Això
+- ▁которую
+- ▁gossip
+- ▁Ме
+- 随时
+- ▁adventures
+- ▁jsme
+- 对手
+- 特征
+- ミス
+- ▁repose
+- ▁Gruppen
+- ▁jackson
+- ▁cage
+- ▁Gewalt
+- ▁Armee
+- систем
+- ▁rigid
+- ంట
+- ▁trink
+- 大卫
+- မ
+- ▁tumble
+- сә
+- 到现在
+- 我们如何
+- ▁summon
+- ▁lên
+- gesetz
+- クション
+- そこから
+- サイ
+- 관
+- ▁그랬
+- 帮忙
+- ▁lantern
+- ▁Klein
+- 果たして
+- fusion
+- hild
+- wirtschaftliche
+- 去世
+- цо
+- present
+- ▁users
+- ▁pelo
+- сид
+- ▁sağ
+- ▁Schuld
+- 老爷
+- 出席
+- ▁license
+- 臨
+- ину
+- ▁neglected
+- ▁Stu
+- දි
+- ▁maka
+- ▁Zusammen
+- ▁harp
+- mill
+- 是从
+- скай
+- 彼得
+- 度过
+- schte
+- 爽
+- 漫画
+- 杀死
+- 保留
+- عرف
+- ▁emergency
+- 他已经
+- ▁plead
+- するという
+- ▁destiny
+- 解放
+- 午
+- まさか
+- そんなこと
+- 滞
+- pell
+- ▁presume
+- ތ
+- ▁Chicago
+- ический
+- ▁crois
+- ▁plastic
+- linda
+- ▁consist
+- ▁jolly
+- uke
+- 董事会
+- يى
+- ▁dedicated
+- ຂ
+- ▁проста
+- 体の
+- ფ
+- 交代
+- いただく
+- ▁genoeg
+- 边缘
+- ▁ninth
+- ▁flushed
+- 他想
+- ▁impatient
+- hst
+- гын
+- ▁agua
+- ▁basin
+- ▁prosperity
+- 今朝
+- posto
+- 罚
+- ▁lease
+- ические
+- ▁idee
+- ▁shouting
+- ▁soixante
+- ▁oogen
+- リスト
+- ▁frog
+- 两次
+- ونه
+- 区の
+- trouw
+- ▁consisted
+- 郊
+- ▁Ibi
+- ▁Standard
+- ▁partir
+- 声明
+- čč
+- ▁cordial
+- ▁brute
+- ▁rond
+- ほしい
+- 几个月
+- ▁denkt
+- 消费
+- road
+- kül
+- ደ
+- א
+- ▁знаешь
+- ▁ruins
+- زان
+- ▁segments
+- lise
+- あなたの
+- 着他
+- 还得
+- 思った
+- 俺は
+- ▁incorpora
+- ウィ
+- ▁принима
+- 其中一个
+- ▁induced
+- ▁lively
+- кән
+- ▁recruit
+- ▁viola
+- 雨が
+- ▁động
+- ▁peep
+- fter
+- ▁Sommer
+- 是很
+- を探
+- 让自己
+- ▁occasional
+- лах
+- ▁выход
+- kud
+- స్
+- ▁верх
+- ▁nouveau
+- ▁amerikanischen
+- ▁ross
+- ▁взял
+- ▁questioned
+- ▁Begriff
+- ▁anfangen
+- imwe
+- вис
+- ▁eby
+- ▁irregular
+- inya
+- ▁April
+- мах
+- 物を
+- hafte
+- 男が
+- ▁sites
+- 事を
+- ▁Jahres
+- ▁realm
+- 纹
+- ▁erinnere
+- kosten
+- 成熟
+- 仗
+- ِي
+- 捨て
+- մ
+- ತ
+- 입
+- はまだ
+- ▁luego
+- ▁dunkle
+- 三百
+- ▁Pot
+- کاری
+- ▁deaf
+- 物の
+- 其实是
+- iam
+- ▁impacted
+- цый
+- 生徒
+- ▁thankful
+- melde
+- '36'
+- ▁situated
+- 食材
+- barkeit
+- рем
+- 给你们
+- 为此
+- 他妈
+- ču
+- 冲突
+- ▁confessed
+- альный
+- 墨
+- ▁친구
+- 固定
+- 主义者
+- ▁spy
+- 驾驶
+- ises
+- ▁exit
+- lep
+- しまった
+- ர்கள்
+- ▁discern
+- ▁preparations
+- 群体
+- ▁menschlichen
+- zieht
+- ▁institute
+- 其他的
+- ▁neighbor
+- gte
+- 他们可以
+- hängen
+- гаар
+- رک
+- zier
+- 呐
+- পা
+- ▁jewels
+- 誰も
+- jj
+- ▁terme
+- жив
+- 眉
+- ansah
+- 带来了
+- iumenge
+- handlung
+- 电台
+- 少ない
+- ▁Farbe
+- 浪费
+- ▁Jugend
+- ▁чуть
+- 行為
+- ард
+- neuf
+- ガン
+- lch
+- mö
+- ukuri
+- ▁gewinnen
+- 得了
+- remo
+- 弥
+- ▁tại
+- ▁Mai
+- 当局
+- link
+- 黄金
+- тели
+- 的家庭
+- ▁suburb
+- ▁captive
+- ▁summoned
+- 堪
+- 葬
+- 不说
+- ▁bump
+- ▁camel
+- まとめ
+- ▁checked
+- 一回
+- ത
+- イル
+- ▁ebenfalls
+- gust
+- gic
+- ▁luna
+- ▁austria
+- ▁exhibit
+- ög
+- ▁francisco
+- únic
+- seven
+- オミクロン株
+- 过去了
+- 石油
+- racy
+- 飯
+- ▁Wil
+- ▁lado
+- ▁flies
+- беж
+- ỏ
+- ▁شب
+- idir
+- ▁variation
+- 聞く
+- ▁doit
+- 平静
+- ▁Hälfte
+- ▁aujourd
+- woman
+- ▁мои
+- ▁stran
+- ▁convince
+- ▁prendre
+- lī
+- ▁geschickt
+- ▁perfume
+- 头上
+- ▁конце
+- 里克
+- ▁turk
+- ▁divert
+- 没人
+- حي
+- 郭
+- ▁benutzt
+- ですかね
+- nier
+- Rh
+- 适应
+- ▁trebuie
+- ▁сай
+- 前から
+- ▁pledge
+- が必要です
+- 冷静
+- ▁fist
+- 咯
+- ▁Gefahr
+- 懸念
+- ▁брат
+- 予選
+- cock
+- つけて
+- fond
+- ▁кому
+- ▁cô
+- 为自己
+- ▁täglich
+- チェック
+- 强大
+- ела
+- さが
+- uzu
+- 拉斯
+- ▁banda
+- 遍
+- ▁Länder
+- ັນ
+- களின்
+- ▁guessed
+- ソース
+- hoff
+- 掛け
+- 在哪儿
+- වෙ
+- 拔
+- ▁threshold
+- ▁advised
+- ▁Behandlung
+- тора
+- ▁chaque
+- ▁varied
+- kay
+- カン
+- lā
+- ▁jerry
+- kunft
+- ип
+- ▁publication
+- ▁Gründe
+- 艺术家
+- لە
+- ▁Vergangenheit
+- 冇
+- 深い
+- ốc
+- ▁pueden
+- ▁sentiments
+- 西日本
+- ▁readers
+- ▁Element
+- ▁төр
+- ▁gebruik
+- lisa
+- ▁factory
+- ▁kau
+- ▁Mitglieder
+- 将是
+- ▁вдруг
+- 工場
+- 风格
+- ации
+- ▁interna
+- ▁recalled
+- ▁scream
+- twitter
+- 解説
+- xes
+- 専
+- ▁Stern
+- ▁няма
+- ▁visitar
+- ▁owners
+- 0,000
+- ▁gefragt
+- さすが
+- ▁خواهم
+- ▁shoe
+- ▁dashed
+- ▁zelfs
+- bola
+- ӑ
+- ▁Schlüssel
+- ວ່າ
+- ĝe
+- 高级
+- ▁persuade
+- думал
+- ということですね
+- ▁celui
+- 聞
+- री
+- 他人
+- ▁Ini
+- নের
+- ▁pře
+- ▁estan
+- ▁recognised
+- 伟大
+- からです
+- 知道你
+- ▁reeds
+- கின்றன
+- 陣
+- zugeben
+- ▁downward
+- 総理
+- ▁plen
+- 商量
+- 杆
+- ▁nosotros
+- ▁vater
+- ▁Ва
+- ▁Ell
+- 唱歌
+- ▁права
+- 学者
+- ▁recollect
+- ▁representation
+- ▁Gefängnis
+- િ
+- ▁Erinnerung
+- 他被
+- ▁арт
+- ẽ
+- ać
+- ▁moss
+- luft
+- 置いて
+- ▁Akt
+- سف
+- 心脏
+- ▁babies
+- tekereza
+- 食事
+- කා
+- торы
+- блюд
+- のみ
+- 分から
+- رى
+- ▁drain
+- aidd
+- ▁Abantu
+- ģi
+- 行った
+- 詳しい
+- ▁opge
+- ▁deiner
+- メニュー
+- mula
+- 得意
+- ▁struggled
+- 窓
+- 偶
+- ▁Besuch
+- ங்கு
+- ▁enfants
+- ▁Bern
+- ▁perish
+- 千葉県
+- ▁від
+- 困惑
+- ▁آخر
+- ▁Fue
+- க்கா
+- できて
+- kond
+- ▁სა
+- ▁christianity
+- くれ
+- ▁nka
+- ڪ
+- ▁segon
+- lique
+- 强大的
+- っぱ
+- ▁gains
+- 那时候
+- 카
+- '&'
+- ▁rechts
+- TV
+- 日本代表
+- 他们没有
+- lice
+- ▁наших
+- ▁Beruf
+- 兆
+- лес
+- ▁shrink
+- ▁Albert
+- ayi
+- ▁forbid
+- ▁miteinander
+- 奸
+- 靴
+- uloj
+- 州的
+- ▁ashes
+- クト
+- дов
+- ▁refresh
+- тат
+- ▁freundlich
+- lista
+- ▁rings
+- ▁fancied
+- aktion
+- ▁diseases
+- ▁bekam
+- 東京の
+- ▁quien
+- 上下
+- 我又
+- しょ
+- ▁Як
+- ▁biblioteca
+- ▁jemandem
+- 事物
+- ckey
+- суд
+- ▁overhead
+- どうも
+- جمع
+- ▁Parti
+- ▁pill
+- قدم
+- ▁Insel
+- るから
+- ▁sobald
+- ▁Kap
+- උ
+- ▁aga
+- ▁என்ற
+- ▁Ни
+- いている
+- 险
+- ▁Berlin
+- ▁victoria
+- rato
+- simil
+- сек
+- ▁categories
+- ▁Quelle
+- ▁Lake
+- seb
+- dama
+- ▁восемнадцать
+- antoj
+- が高い
+- ▁лишь
+- ▁año
+- bring
+- ▁그렇
+- нні
+- ▁дня
+- ற்
+- 当時の
+- ▁renewed
+- corn
+- ▁goals
+- 小学
+- 两天
+- ▁Gesetze
+- ▁estamos
+- どうやって
+- ▁livre
+- 案子
+- landa
+- провод
+- ម
+- 別に
+- ▁красно
+- 七十
+- ごはん
+- атор
+- ▁yonder
+- 听众
+- acre
+- ▁uncomfortable
+- sätze
+- ▁programa
+- bê
+- ▁observations
+- ▁muse
+- кли
+- 大幅
+- 他把
+- ▁beef
+- ঠ
+- ▁dialogue
+- ▁marvel
+- 请你
+- ▁awakened
+- ▁Iran
+- automat
+- ▁чалавек
+- வில்லை
+- 实施
+- ▁contributed
+- ▁richtige
+- 它们的
+- みました
+- ▁сказала
+- ▁кем
+- ▁escrit
+- ▁hierher
+- 他现在
+- ▁indifference
+- யே
+- 亡くなった
+- 方面的
+- 数は
+- 川の
+- ಮ
+- ▁Objekt
+- シーズン
+- ▁teatre
+- 制定
+- 钢
+- ▁Ressourcen
+- 箭
+- ▁augment
+- 担任
+- 丰
+- ▁escort
+- iris
+- 月から
+- ница
+- ▁wenige
+- пут
+- ız
+- ▁footsteps
+- ▁Umwelt
+- شى
+- ▁mixture
+- 呼びかけ
+- 南方
+- ▁erg
+- 岸田総理
+- ି
+- cem
+- ▁tests
+- неш
+- mura
+- gré
+- ▁менее
+- há
+- ▁pepper
+- ங்கி
+- 代の
+- ▁разговор
+- yamba
+- 等你
+- ▁indignant
+- 했어
+- ▁энэ
+- ▁contented
+- ▁speci
+- ▁Staats
+- ție
+- その時
+- ුව
+- ▁alliance
+- 上学
+- おっ
+- 前線
+- చ
+- ▁schönen
+- 分かった
+- ация
+- 緑
+- 勇敢
+- ▁thoughtful
+- жан
+- ▁trouve
+- сно
+- ▁Rechts
+- ର
+- ▁february
+- 一辆
+- 举行
+- ères
+- ▁minha
+- toxic
+- 牵
+- 释放
+- ▁gallery
+- кажу
+- ▁bestand
+- ▁fils
+- ▁Charakter
+- ▁Sinne
+- sache
+- ▁Gast
+- 在这些
+- ▁sper
+- ヌ
+- ái
+- ▁Wood
+- 六十
+- ▁simul
+- しかない
+- 仮
+- 医師
+- ▁Florida
+- ▁distributed
+- ▁flour
+- oxid
+- ன்ன
+- の部分
+- 出る
+- ▁saat
+- ▁giá
+- ▁reporter
+- 就没
+- ▁amy
+- ▁profund
+- ▁International
+- ▁framework
+- ვე
+- zir
+- ложил
+- 加速
+- しまいました
+- ▁Ты
+- lion
+- 伪
+- 杀人
+- 准确
+- 値上げ
+- ▁Organ
+- することで
+- ▁funciona
+- нак
+- どうしても
+- 夕方
+- ▁вс
+- 为我们
+- ▁gezeigt
+- ponent
+- ワクチン接種
+- ▁velvet
+- ▁attraction
+- と思いますね
+- 巻
+- 購入
+- ▁Spaß
+- ▁här
+- ▁Baum
+- 必要な
+- ▁instruct
+- ▁lordship
+- 知らない
+- ▁arbeitet
+- 稍
+- 怖い
+- ▁Ehre
+- 실
+- 覆盖
+- ▁ладно
+- ▁absorb
+- ▁tread
+- 踏み
+- cate
+- 誘
+- 空港
+- ▁patriot
+- িয়া
+- trust
+- ▁Beth
+- ▁transactions
+- ▁tạo
+- firma
+- ▁Gö
+- 实践
+- teg
+- とり
+- ▁claro
+- もので
+- かというと
+- 打了
+- 我需要
+- ▁admired
+- ▁bronze
+- duction
+- Bu
+- ▁одного
+- yla
+- Ş
+- 的女儿
+- جن
+- ▁стала
+- ▁conceived
+- цей
+- ▁akan
+- biologi
+- ▁palabra
+- ▁stolz
+- デン
+- wärts
+- ▁jsou
+- ▁twilight
+- ▁Brazil
+- ▁verbracht
+- reka
+- ▁понимаете
+- つける
+- ppi
+- ▁calcul
+- general
+- ▁igual
+- ▁wealthy
+- usia
+- оо
+- ▁harmony
+- ▁Code
+- 協
+- ▁принят
+- 强调
+- пала
+- ▁symptoms
+- ▁trading
+- ▁chef
+- ▁answering
+- 降低
+- 著
+- ▁expedi
+- 这个地方
+- ▁swimming
+- ▁froh
+- ▁roused
+- 車が
+- フォ
+- ▁chatter
+- brun
+- Univers
+- ▁earliest
+- hita
+- ▁max
+- ▁awoke
+- 農
+- ladi
+- ▁scor
+- 垂
+- 教练
+- vollen
+- ▁уверен
+- 儿啊
+- pfel
+- Ko
+- ▁bate
+- 慎
+- 夫婦
+- ▁tunnel
+- ▁visto
+- あん
+- ▁uncertainty
+- 有限
+- ▁conquer
+- ▁decades
+- ▁Lü
+- ▁Mach
+- ▁Sehen
+- чыць
+- 差し
+- ▁warriors
+- なさい
+- 四年
+- под
+- illon
+- 狠
+- 感激
+- vik
+- ▁decade
+- ▁lime
+- 的确
+- ▁Elizabeth
+- 선
+- ボー
+- ▁plunged
+- 实验室
+- ▁Nachricht
+- ühr
+- 泊
+- ▁Viņa
+- ▁managing
+- ▁seul
+- heden
+- ▁Höhe
+- ▁robber
+- ▁testament
+- bereich
+- medic
+- ipun
+- トラック
+- cott
+- antic
+- ބ
+- ▁собой
+- 发挥
+- мін
+- ▁Mond
+- ▁carro
+- どうした
+- aco
+- ていきます
+- ▁Després
+- 凝
+- 计算机
+- ▁Podcast
+- ▁получа
+- ▁fais
+- ▁mould
+- ▁полу
+- паль
+- 想像
+- ▁Schutz
+- step
+- 相比
+- いち
+- けん
+- ną
+- బ
+- ーター
+- ▁principi
+- 二零
+- mail
+- ▁hire
+- ▁legte
+- ▁degli
+- ▁плохо
+- کا
+- 会長
+- ランキング
+- хүү
+- 盐
+- ▁limb
+- ▁mineral
+- 最喜欢的
+- 取った
+- ▁proven
+- mack
+- ▁Themen
+- ▁democrat
+- 才会
+- 通報
+- new
+- ▁satan
+- ▁yaptı
+- ▁jene
+- koll
+- შ
+- ▁retro
+- ▁عم
+- '2015'
+- 払
+- ▁Ces
+- ▁innocence
+- 間違い
+- ▁array
+- ▁yid
+- ▁déjà
+- ▁şekilde
+- 始まった
+- あなたは
+- ▁electro
+- 邮
+- 文明
+- 入院
+- داد
+- 当她
+- ▁upright
+- ▁jours
+- ポン
+- ▁Cel
+- すこと
+- ▁ekonomi
+- ▁suicide
+- きっと
+- كت
+- ▁corruption
+- ▁tue
+- ▁Karte
+- 危機
+- мас
+- ▁الل
+- っこ
+- ▁apprehension
+- 磁
+- worthy
+- ▁constrain
+- lwa
+- ▁jacket
+- problem
+- ▁anxiously
+- ▁hohen
+- ▁কর
+- angi
+- ▁subscribe
+- ▁проблема
+- ▁цяпер
+- 篇
+- 好看
+- ▁lösen
+- ▁parson
+- গা
+- ש
+- 廊
+- 天然
+- ▁своего
+- 联盟
+- üü
+- ▁cob
+- 赖
+- ▁Sä
+- ▁rattle
+- тин
+- おく
+- ޭ
+- 华盛顿
+- messen
+- ▁göster
+- ▁muchos
+- зон
+- 拆
+- ▁tracks
+- 不确定
+- ▁المن
+- ランド
+- ▁niba
+- ไป
+- ▁resulted
+- 这个国家
+- もあった
+- leta
+- ▁multaj
+- ▁classic
+- 梨
+- ކު
+- empre
+- վ
+- ▁matthew
+- ▁strategies
+- ▁Rad
+- セン
+- ねぇ
+- 屋さん
+- 旨
+- ▁amazement
+- ږ
+- ▁같이
+- mbang
+- ▁Max
+- 曜
+- ▁terribly
+- schütt
+- ▁адзін
+- と発表しました
+- ▁Gottes
+- 拿到
+- 構
+- クイズ
+- ▁hideous
+- ▁مهم
+- ▁каза
+- 타
+- tzt
+- ▁çalış
+- 来吧
+- 平和
+- ▁twe
+- ▁mule
+- クリーム
+- 成绩
+- ▁vrij
+- というのも
+- ▁weißen
+- ▁cathedral
+- ▁viņš
+- ▁pobre
+- 构
+- てくれ
+- issement
+- শি
+- ▁tenth
+- ▁pretended
+- 朵
+- дт
+- ▁thief
+- ▁planted
+- ▁persist
+- 喷
+- ▁долго
+- ▁zufällig
+- ddu
+- 的精神
+- たった
+- анд
+- 钻
+- 郡
+- ▁завтра
+- ▁eminent
+- ▁losses
+- ▁drie
+- 棋
+- annya
+- 楚
+- ▁причин
+- ▁flick
+- ▁bekommt
+- ▁pode
+- ▁unterstützt
+- ▁Tiere
+- 的内容
+- が出る
+- ▁какое
+- ▁Ball
+- するために
+- ▁mwaka
+- 回目の
+- ▁visits
+- hne
+- ▁australian
+- ▁oldest
+- ▁variable
+- ssystem
+- ▁rely
+- ▁стоит
+- 客人
+- altro
+- 亡くな
+- ▁dispers
+- 被害者
+- bber
+- ▁Kerl
+- 裤
+- 等我
+- ▁Mikro
+- ▁Barr
+- овый
+- ▁occup
+- ች
+- くり
+- 我确实
+- ையும்
+- ▁situas
+- 洁
+- finanz
+- ▁gewe
+- 徳
+- داخل
+- ▁threatening
+- 失望
+- ▁Beweise
+- ろうと
+- ▁holl
+- 准备好
+- ▁bells
+- 啲
+- ▁shakespeare
+- 拳
+- скую
+- kasi
+- 推动
+- ▁Schlag
+- ým
+- ▁oncle
+- ▁dorthin
+- ▁assert
+- ಲ
+- 培训
+- ▁unwilling
+- 位の
+- ▁bills
+- ▁drivers
+- ▁instru
+- 弟弟
+- 各国
+- tip
+- ▁avail
+- kade
+- 瞎
+- 公子
+- 历史上
+- エリア
+- ▁tierra
+- ▁старо
+- 皆
+- ▁headquarters
+- 翻译
+- 組織
+- ▁Feder
+- ood
+- экс
+- ▁videos
+- 为我
+- ception
+- 官员
+- 審
+- ▁자기
+- ▁Kollegen
+- imbu
+- nywa
+- ▁raven
+- ▁sultan
+- ffy
+- guha
+- 阻
+- шым
+- рек
+- ▁Chan
+- 夏天
+- 対戦
+- ▁derzeit
+- けば
+- 自分たち
+- ▁einzigen
+- '2020'
+- 籍
+- ▁pluck
+- ▁Allgemeinen
+- ▁Einfluss
+- 为什么不
+- ▁environmental
+- сць
+- ▁separation
+- siniz
+- ▁Fal
+- 娶
+- ▁මේ
+- ▁induce
+- ▁ebenso
+- ▁donner
+- ▁снова
+- orde
+- 打席
+- 概
+- 收拾
+- ▁Finger
+- ▁Schwarz
+- やすく
+- ▁linen
+- ▁filling
+- 贡献
+- 震惊
+- ▁Präsidenten
+- ▁proceeding
+- 地图
+- champ
+- issabte
+- ▁быць
+- 带走
+- зав
+- ▁kämpfen
+- 捜索
+- ▁policies
+- 演技
+- лап
+- 思いました
+- ▁egy
+- ▁плат
+- 分配
+- 驱
+- 耳朵
+- 降る
+- ▁sally
+- ▁gotta
+- 氏は
+- どういうこと
+- عتقد
+- 繁
+- バッター
+- ▁decree
+- をかけて
+- ▁deinem
+- ▁beggar
+- ジョン
+- tino
+- ▁gehören
+- hwa
+- 授
+- łu
+- ▁strongest
+- ▁recommendation
+- ▁nchi
+- zil
+- 了个
+- ▁prepara
+- 电子邮件
+- 知事
+- ▁trabaja
+- arak
+- 覚えて
+- 前回
+- ▁habitual
+- cky
+- 耶
+- ▁செய்த
+- cina
+- łem
+- ▁pencil
+- ▁ĝin
+- そもそも
+- 였
+- ▁triumphant
+- tiği
+- ▁visiting
+- ▁recording
+- 变得更
+- 牺牲
+- ▁hesitation
+- ▁erschien
+- 千万
+- kende
+- rrington
+- ùng
+- ▁conquest
+- ▁openly
+- ▁sali
+- ▁herauszufinden
+- ▁особенно
+- бод
+- せい
+- ▁sulla
+- ņu
+- 弯
+- ▁tug
+- ▁européenne
+- ▁gewisse
+- čin
+- ением
+- てた
+- ▁promises
+- дат
+- ırı
+- ▁buiten
+- ızı
+- ▁inflict
+- mädchen
+- ▁ermöglicht
+- ▁maintenance
+- ▁curiously
+- خواه
+- ▁hình
+- ▁третья
+- ▁кир
+- 战略
+- лон
+- ▁rocky
+- 来週
+- straße
+- 问他
+- ▁mächtig
+- 炉
+- 纪念
+- ́
+- ホントに
+- antwoord
+- 完成了
+- コーナー
+- ▁musician
+- ▁Chinese
+- キュ
+- 信心
+- ▁mansion
+- 奏
+- ąc
+- 娱乐
+- ▁extension
+- ▁decay
+- ▁arbeitete
+- ▁dood
+- 很好的
+- やん
+- ▁hoy
+- 근
+- ▁anguish
+- 僕が
+- が見
+- ▁electricity
+- espera
+- ▁cove
+- ▁feathers
+- lif
+- 说的话
+- なります
+- ▁hilft
+- entes
+- 有问题
+- ▁reuni
+- 安慰
+- ирует
+- diplom
+- ję
+- deki
+- anja
+- 找不到
+- цвет
+- brand
+- 闘
+- 有着
+- 居民
+- නා
+- したもの
+- troph
+- ▁purely
+- ▁troop
+- ずつ
+- рост
+- ▁nicholas
+- шей
+- ▁moeder
+- ショ
+- ▁Sonnen
+- ▁aucun
+- してくれる
+- 物語
+- 乾
+- முறை
+- ▁landlord
+- ▁interval
+- 一套
+- unternehmen
+- 怨
+- 宏
+- 部长
+- ▁менән
+- ▁lagi
+- ▁historian
+- ්‍ර
+- ▁erect
+- ▁waved
+- кнул
+- ▁plains
+- мал
+- rce
+- tuvo
+- 残って
+- 泳
+- 图书馆
+- ▁inclination
+- kamp
+- ▁такі
+- 出发
+- にくい
+- ▁verde
+- 盯着
+- ▁jimmy
+- ▁messages
+- ▁liebt
+- 点儿
+- ▁lifting
+- пры
+- ▁productivity
+- پل
+- креп
+- ▁witnesses
+- яз
+- るんですけど
+- 师傅
+- année
+- ▁Mehr
+- ▁equivalent
+- ▁regi
+- ën
+- زي
+- てくれる
+- 突き
+- ▁wütend
+- ▁translation
+- 回忆
+- ▁هنا
+- 導
+- ▁partit
+- ▁maintenant
+- cela
+- ▁außerhalb
+- のことを
+- ▁turkey
+- ▁які
+- 有任何
+- ▁disciples
+- 終
+- なと思います
+- 梯
+- kole
+- ▁Künstler
+- ▁konzentrieren
+- ▁crimson
+- енко
+- 私たちの
+- ▁Sonne
+- 符
+- ▁garrison
+- യ
+- ▁furious
+- 金属
+- 出于
+- ssé
+- ▁durchgeführt
+- ▁victor
+- dele
+- ▁packed
+- 漂亮的
+- ▁madre
+- aggio
+- үүр
+- ついた
+- 在外面
+- 长得
+- ▁cần
+- ▁blowing
+- දු
+- ▁Mari
+- ▁strictly
+- ющих
+- lessness
+- ▁Prä
+- ajo
+- 资料
+- ▁Lord
+- くれて
+- 肚子
+- ▁wounds
+- ▁nếu
+- チーズ
+- carna
+- ホール
+- 들이
+- ску
+- だんだん
+- 克里斯
+- ▁бүр
+- ▁demonstrate
+- haga
+- 寻
+- 追加
+- рей
+- 头脑
+- 立つ
+- 芳
+- जा
+- ▁Oku
+- 西部
+- 期望
+- ▁saber
+- れている
+- 柏
+- ▁девятнадцать
+- 거
+- инг
+- うちに
+- ▁nhận
+- ▁sommes
+- なのに
+- ▁betrayed
+- Ç
+- つもり
+- るんですよ
+- त्
+- ▁herein
+- りたい
+- nium
+- 人気の
+- 现象
+- 抬
+- 的公司
+- ნა
+- hali
+- ▁admirable
+- 髪
+- 这里有
+- ▁ultimate
+- ▁devas
+- つけた
+- 募
+- ▁warmth
+- ▁reject
+- 贸易
+- 警視庁
+- 簡単に
+- 感受
+- ▁organizations
+- ▁realmente
+- ▁schwarzen
+- ▁Glo
+- '75'
+- ▁unjust
+- 居然
+- 的例子
+- ▁Jungen
+- ▁Què
+- ования
+- ▁retra
+- 一些东西
+- ishwa
+- ▁dingen
+- ▁nime
+- ▁què
+- цов
+- 取って
+- komeza
+- 躺在
+- ▁gravity
+- ▁Таму
+- ▁englishman
+- ▁artistic
+- ▁housing
+- 在于
+- ▁allan
+- 温柔
+- amerika
+- 的时刻
+- fahrt
+- ▁schätze
+- ▁Opfer
+- 伦敦
+- ▁greeted
+- ▁ranch
+- ▁tú
+- ▁exile
+- 小时候
+- 董
+- ▁illusion
+- ▁Finanz
+- ▁willst
+- ▁Come
+- ▁geheim
+- ▁weep
+- ▁Waffen
+- bó
+- vā
+- 昏
+- ▁underneath
+- dığı
+- ▁Jane
+- ▁вопросы
+- ▁dopo
+- 杰克
+- 主義
+- gesprochen
+- ▁groote
+- ▁история
+- 受け入れ
+- 負担
+- ▁масс
+- awo
+- ▁cứ
+- ▁advise
+- 涨
+- 笑顔
+- plu
+- ▁আর
+- 兽
+- ▁stroll
+- 哪些
+- ▁conceive
+- なった
+- ▁nachdenken
+- ▁precede
+- 先輩
+- graben
+- ▁Einer
+- ▁Boot
+- バック
+- ілі
+- дова
+- скі
+- 无论是
+- eira
+- ▁comedy
+- 你得
+- ▁Però
+- 入れて
+- роў
+- ▁narra
+- ography
+- 味が
+- 帰って
+- 拍摄
+- 评估
+- ▁inquire
+- 类型
+- گان
+- ▁bestehen
+- ▁nacional
+- ▁fremd
+- 天空
+- ▁opponent
+- gebildet
+- ものは
+- escent
+- ▁blown
+- ▁unterschiedlich
+- ▁substitute
+- 我只
+- ▁withdrew
+- ▁Kri
+- 込め
+- ▁From
+- ▁explosion
+- fragen
+- ▁exclude
+- 女性が
+- papier
+- ▁heiße
+- ▁praktisch
+- ▁mga
+- ▁foster
+- 滋
+- ▁Earth
+- ▁troviĝas
+- 教室
+- 到时候
+- けが
+- ▁frightful
+- ▁nationale
+- 在那个
+- ланд
+- ▁firing
+- ора
+- たん
+- 存在的
+- 镜
+- trakt
+- 術
+- اث
+- ▁那我们
+- ▁scarlet
+- ▁exhaust
+- хар
+- ▁circum
+- ▁ancora
+- 皮肤
+- 替え
+- 充满了
+- づけ
+- 放心吧
+- 你看看
+- ▁ажил
+- ▁yourselves
+- 串
+- 做一些
+- ▁enlighten
+- تاب
+- ▁vierzig
+- 统计
+- ▁harris
+- kanye
+- ثر
+- 申
+- ▁moan
+- ▁impressive
+- ▁Tele
+- кур
+- цыі
+- 导演
+- 你说的
+- いなかった
+- ставить
+- ▁blend
+- ▁wretch
+- ▁descent
+- seid
+- ▁stages
+- 剤
+- ▁정도
+- 我知道你
+- ▁Holz
+- ▁boven
+- ルール
+- ▁هە
+- ▁adjust
+- 続けて
+- 海上
+- ▁Argument
+- chau
+- 勇气
+- 洪
+- 見ると
+- ▁motionless
+- が必要
+- ▁Schulen
+- ▁Spitze
+- ▁tenia
+- 明け
+- 遅
+- 投手
+- chain
+- 跟踪
+- ▁Zeiten
+- ▁mọi
+- 待って
+- haupt
+- daj
+- 的妻子
+- ▁urgent
+- tack
+- ▁publi
+- 桶
+- 盆
+- 夹
+- ▁crushed
+- これまでに
+- ▁oriental
+- てくる
+- ▁reverence
+- 在过去的
+- ▁heroes
+- arlo
+- ▁toilet
+- しゃべ
+- クラス
+- ünü
+- paro
+- することが
+- mela
+- ▁jealousy
+- fant
+- 挡
+- ▁significance
+- 系列
+- 灾
+- ▁gym
+- завод
+- 気に
+- ▁tipp
+- ▁barbara
+- ▁Actualment
+- ▁Today
+- 態
+- ющие
+- ▁discussing
+- 能源
+- ▁granda
+- ▁спросил
+- onic
+- ือ
+- ▁Neben
+- さない
+- アー
+- ▁아이
+- ▁hizo
+- いたします
+- 外国人
+- มา
+- klop
+- 反映
+- wirken
+- すと
+- フォー
+- ьян
+- そのあと
+- ▁кара
+- ▁jeff
+- 死者
+- ሰ
+- ▁violently
+- 読み
+- giving
+- ▁하나
+- чик
+- 之下
+- 回り
+- 조
+- တ
+- umbi
+- ▁convent
+- ▁altered
+- 成为一个
+- ▁conven
+- 으
+- ▁affirm
+- ▁المح
+- ▁Kern
+- angka
+- ஃப
+- 听见
+- 的经历
+- 燕
+- 赔
+- 」「
+- ▁pav
+- ▁starke
+- ▁tiger
+- 且
+- 的发展
+- pois
+- ▁busca
+- らが
+- ыт
+- ความ
+- ologische
+- ▁другие
+- の上に
+- ▁comrades
+- ゅう
+- ত্র
+- 充分
+- tez
+- ▁petition
+- 人群
+- руб
+- tafel
+- 到来
+- ▁sailing
+- ఇ
+- ▁милли
+- 超え
+- ▁Hü
+- ▁четвёртый
+- 取决于
+- yobozi
+- لق
+- 写作
+- strahl
+- るために
+- ▁edition
+- ▁possibilities
+- 社交
+- ▁attribute
+- ▁enjoying
+- サイド
+- ured
+- 手指
+- 叛
+- istische
+- ▁tự
+- 法庭
+- ▁бизнес
+- 开玩笑
+- ▁academic
+- ▁позвол
+- 贷款
+- 为他
+- ▁Vall
+- 比例
+- princip
+- ەی
+- ▁increasingly
+- 爆炸
+- ▁heartily
+- ▁japanese
+- êr
+- 巻き
+- enfant
+- ından
+- ▁collabora
+- undvierzig
+- 尘
+- ▁whither
+- 不怕
+- ką
+- mēr
+- ▁그러면
+- ▁diversos
+- ▁medal
+- ifies
+- стары
+- valent
+- ▁ungefähr
+- 쪽
+- ▁dagegen
+- 高齢者
+- ▁তিনি
+- ড়
+- 記念
+- larda
+- 态度
+- ▁leisten
+- ▁связан
+- ▁assez
+- ▁stove
+- 白い
+- ებ
+- 痴
+- ▁mới
+- 殺害
+- 創
+- ▁align
+- 特别的
+- ildi
+- ▁Karriere
+- ▁laat
+- richtung
+- 真っ
+- adores
+- 疫情
+- といえば
+- 맞
+- ▁proceedings
+- 不一定
+- ▁throng
+- дает
+- 美国的
+- ▁genom
+- 企
+- 泪
+- ▁greeks
+- ▁compound
+- ▁reporting
+- escola
+- どおり
+- ▁butler
+- ▁groupe
+- ట్
+- дад
+- めっちゃ
+- 年前に
+- three
+- ҵ
+- 闷
+- 詳しく
+- 著名的
+- ▁envy
+- 一眼
+- もらえ
+- ▁punish
+- يرة
+- 骄傲
+- デザイン
+- ▁menjadi
+- 悩
+- を得
+- দু
+- 進む
+- 好事
+- 一座
+- ▁cargo
+- 你再
+- 聞こえ
+- 작
+- ▁ваши
+- でしたね
+- ▁platforms
+- ▁clair
+- るんですね
+- ▁personne
+- 现在我们
+- において
+- ▁madness
+- 본
+- ▁Wesentlichen
+- ▁konuş
+- ▁откуда
+- tius
+- енная
+- вит
+- يج
+- ▁shops
+- zep
+- ží
+- ▁Botschaft
+- ▁devant
+- ▁abraham
+- ▁respective
+- 马克
+- شو
+- здоров
+- ▁motives
+- ības
+- ▁encouraging
+- 辣
+- 還
+- ▁möglicherweise
+- ▁Realität
+- 国外
+- 尼亚
+- larında
+- 挺好的
+- ▁거지
+- ▁Ding
+- exist
+- 再生
+- ހ
+- 揭
+- 高橋
+- gression
+- ▁zusätzliche
+- ▁karşı
+- 做的是
+- ▁beobachtet
+- õi
+- ▁Genau
+- үүн
+- ▁anim
+- 徹底
+- キン
+- 下げ
+- ▁şu
+- 失敗
+- ▁Hab
+- பட
+- ▁Quel
+- rono
+- ▁Cer
+- ▁تلك
+- の動き
+- 感覚
+- 続ける
+- mania
+- bala
+- 给自己
+- લ
+- ▁Strom
+- ▁Што
+- 精彩
+- ▁jew
+- ▁Entscheidungen
+- ▁customs
+- ▁dismay
+- 大学的
+- 视为
+- なお
+- ஐ
+- ▁lachen
+- ドラ
+- ото
+- ▁porch
+- 寄り
+- ▁consul
+- ▁dusk
+- 全世界
+- ▁diverses
+- ພ
+- ▁excellence
+- 她会
+- رن
+- ▁concentra
+- 議会
+- ▁Over
+- ழ்
+- 我们看到
+- ▁mach
+- ▁Regen
+- 万一
+- ▁Пер
+- すべき
+- ▁Frieden
+- ▁шаг
+- oph
+- ▁viene
+- 若者
+- ▁hostile
+- 拿走
+- 注意力
+- į
+- spiegel
+- ていない
+- 中国の
+- dige
+- 万美元
+- 发明
+- dorf
+- ▁россия
+- tesse
+- зве
+- шло
+- ▁скоро
+- ▁dentro
+- ▁ngày
+- してきました
+- ▁metge
+- ▁echter
+- 聚集
+- 劫
+- ٹ
+- ▁Führung
+- ɛe
+- ▁villain
+- ▁deceived
+- ▁eenige
+- 的变化
+- つつ
+- sieg
+- ▁traurig
+- ▁membres
+- ận
+- もう一度
+- 站起来
+- ありがとうございます
+- tòria
+- ▁trente
+- ▁Imp
+- čen
+- ▁protested
+- 试着
+- ▁welfare
+- 福岡
+- ▁oui
+- гр
+- 病気
+- 某人
+- roning
+- 中には
+- ▁encountered
+- 浩
+- 真实的
+- ▁vile
+- ▁battery
+- 提升
+- ου
+- ира
+- 摘
+- ▁Amerikaner
+- ր
+- ▁ambassador
+- に入る
+- ▁بش
+- 这个故事
+- ▁tengo
+- ▁fragments
+- 舌
+- タイプ
+- ิน
+- ▁relatives
+- 提前
+- ▁secretly
+- 晴
+- 我们俩
+- rührt
+- ▁Center
+- ▁martyr
+- 邻居
+- ▁süß
+- ▁exceptional
+- ほうがいい
+- 魏
+- 嫁给
+- 床上
+- 这真的
+- ダンス
+- park
+- ▁eigenes
+- ▁naught
+- ▁yapı
+- 道歉
+- ▁catching
+- 这份
+- ▁verstanden
+- 透明
+- тся
+- 的生命
+- ▁wichtiger
+- 沮丧
+- '65'
+- ▁montre
+- ▁tarafından
+- 依赖
+- ▁aufgrund
+- ホーム
+- ▁swarm
+- 全力
+- 終わ
+- ▁earthly
+- 捉え
+- ပ
+- ▁expanding
+- 懒
+- ▁Uganda
+- 辆
+- ▁morris
+- 唉
+- 有钱
+- gero
+- ▁només
+- 孟
+- ▁Nó
+- ▁cơ
+- できるよう
+- ▁haunt
+- かって
+- しまいます
+- ▁satisfactory
+- 你为什么
+- 差异
+- てきました
+- ▁unmöglich
+- ▁давно
+- ▁Wohn
+- 做过
+- 厨房
+- ışı
+- 到这里
+- ▁compensation
+- ▁teh
+- ▁aroused
+- kulu
+- cava
+- ▁verses
+- ▁каким
+- ▁deliberately
+- ▁zullen
+- 爱你
+- lje
+- ▁carl
+- ▁계속
+- 消费者
+- 日本で
+- schluss
+- lıyor
+- 広島
+- ▁أنا
+- 常见
+- ▁Joseph
+- ▁muscles
+- ▁tarde
+- 向上
+- ▁проблем
+- ัน
+- ▁cao
+- 瞬间
+- onym
+- バラ
+- ▁Journal
+- 緊急
+- ▁Winter
+- ▁моя
+- 更有
+- ▁بخش
+- 的房子
+- ceea
+- mişti
+- ライト
+- ▁calculated
+- ▁général
+- 茂
+- 심
+- ▁cuatro
+- 生まれた
+- ▁travelled
+- 成就
+- かつて
+- ▁joc
+- ▁parting
+- ▁hört
+- ▁sống
+- stup
+- meze
+- щен
+- ▁адна
+- ским
+- ▁attending
+- ▁crest
+- 把握
+- ▁cellar
+- ▁الناس
+- beck
+- 男性が
+- 好吃
+- னு
+- ▁justified
+- ້າ
+- рв
+- 动力
+- ▁Straf
+- どのように
+- लाई
+- ▁woke
+- ▁fascinating
+- صح
+- ▁vrai
+- に住む
+- cite
+- ▁Fest
+- 狱
+- ▁tumult
+- かく
+- けない
+- ්‍ය
+- ▁дух
+- ▁holland
+- 仇
+- ▁inspector
+- ▁Dol
+- 大小
+- 誉
+- ▁Jahrzehnt
+- みましょう
+- ▁Gall
+- ▁hiện
+- 停下来
+- 居住
+- wandel
+- мел
+- މ
+- 悠
+- ▁slaughter
+- 肝
+- の前に
+- 粉丝
+- ▁swamp
+- を集め
+- ▁befand
+- ▁пятьсот
+- ▁Edward
+- 加油
+- ▁femmes
+- geschlagen
+- ▁wilson
+- ▁failing
+- ▁Früh
+- ▁manuscript
+- めの
+- ▁000
+- 情报
+- quadr
+- ▁observing
+- ถ
+- ▁tommy
+- 看过
+- するよう
+- ▁twas
+- ▁Labor
+- ▁jahre
+- ▁cruelty
+- ▁flowing
+- писа
+- ▁Other
+- ▁trenta
+- ▁corrupt
+- 如果说
+- 厅
+- 계
+- igihugu
+- ▁ກະ
+- ▁causing
+- шча
+- 少爷
+- ▁severely
+- ▁dünya
+- leiter
+- ティング
+- အ
+- yel
+- ических
+- ▁nineteenth
+- ማ
+- 成功的
+- 的看法
+- ▁Traum
+- ▁Gui
+- ▁edit
+- ▁exterior
+- 思维
+- held
+- 色の
+- ▁sincere
+- لىق
+- ▁sever
+- ▁considerably
+- 有一点
+- logi
+- ▁också
+- 团体
+- ходят
+- ▁colonies
+- berries
+- kunde
+- ▁oldu
+- determin
+- ▁poets
+- 与此同时
+- 貴
+- ▁incremental
+- bereit
+- ▁lowered
+- ໍ
+- 激动
+- ▁Gas
+- 富士
+- зер
+- ▁Sar
+- なるほど
+- ▁cooper
+- قب
+- 轻松
+- ographic
+- ▁uneasy
+- ▁ancestors
+- ște
+- core
+- ▁colli
+- ▁okw
+- ının
+- 損
+- 経営
+- 変わる
+- modul
+- valuation
+- 说的是
+- தன்
+- 状態で
+- bû
+- ▁مخ
+- lò
+- ▁должна
+- ▁confine
+- насці
+- ▁devour
+- ▁shaft
+- ▁purchased
+- mwi
+- 物种
+- 僕の
+- 찮
+- 阿尔
+- ▁paradise
+- ▁glare
+- альная
+- евич
+- ▁rejected
+- ▁стать
+- シェ
+- ▁Glas
+- 艰难
+- ▁lily
+- ▁flora
+- 之一是
+- ▁aliaj
+- ▁starb
+- 働き
+- ▁Seit
+- '.000'
+- ▁Zweck
+- ▁Fähigkeiten
+- 災害
+- なこと
+- 迈
+- ▁commissioner
+- フリー
+- 피
+- ▁verändern
+- 冒险
+- adel
+- ▁begonnen
+- 小学校
+- 后果
+- '32'
+- ubuzima
+- 的身份
+- ▁allah
+- ▁screw
+- 早い
+- ▁Strateg
+- ách
+- ▁fram
+- igte
+- ▁заяв
+- ▁dritte
+- പ
+- ▁iemand
+- хә
+- 两人
+- рым
+- jja
+- ▁vot
+- われて
+- ▁nuevo
+- ▁erklärt
+- せず
+- އް
+- ▁zufrieden
+- 歩いて
+- ▁declaration
+- AI
+- 的文章
+- ▁مردم
+- おう
+- technologie
+- ▁книг
+- rani
+- 保守
+- ▁confer
+- ▁questi
+- affi
+- ▁mijne
+- ▁hither
+- 这些事情
+- ▁twentieth
+- ▁White
+- ▁hue
+- jih
+- garde
+- დე
+- ▁greeting
+- ebwa
+- ▁Sachen
+- ▁gladly
+- 稿
+- ▁encuentra
+- ウイルス
+- خی
+- سة
+- ▁sour
+- ▁equi
+- ▁tempest
+- ▁провер
+- rava
+- द्
+- 伙
+- تها
+- ▁psycho
+- ▁blocks
+- ▁Öffentlichkeit
+- 暂时
+- ▁shopping
+- 他要
+- ▁disagreeable
+- ▁dominion
+- 井さん
+- வாக
+- ▁marketplace
+- ▁compromise
+- 浜
+- いただきたい
+- 作る
+- 위
+- YouTube
+- ziehung
+- ▁terrorist
+- 随后
+- ▁seltsame
+- 血液
+- وش
+- ▁tragic
+- ▁danced
+- ▁excessive
+- мыш
+- fari
+- sail
+- ็
+- 是一名
+- ▁gaf
+- ▁invested
+- 深入
+- 奴隶
+- ੀ
+- ▁destined
+- ▁böse
+- 夕
+- ается
+- 人士
+- ▁gulf
+- そうで
+- 很快就
+- ▁eldest
+- ▁Angeles
+- 従
+- வின்
+- 我非常
+- prüf
+- 車の
+- 纪
+- ▁placing
+- 営業
+- ▁lovers
+- ▁tornar
+- leistung
+- 東海
+- 苦し
+- kuza
+- ▁然后呢
+- vuze
+- 虐待
+- فن
+- trans
+- ▁membre
+- したということです
+- maschine
+- ▁Ash
+- laba
+- ▁Stre
+- そうか
+- ையில்
+- 恐怕
+- ▁footprint
+- ▁gush
+- kiza
+- گیری
+- 公式
+- ▁Ruhe
+- ▁imperfect
+- ▁posted
+- prop
+- 的目的
+- ▁prospects
+- ▁Version
+- このような
+- ކ
+- ո
+- கிற
+- ▁arrange
+- ▁мяс
+- ▁transportation
+- 装置
+- ▁Dü
+- itel
+- ļu
+- คุณ
+- 你喜欢
+- ▁basa
+- vro
+- ▁নাই
+- ▁Euro
+- praw
+- ▁militar
+- ▁Fel
+- теп
+- ▁twisted
+- 筹
+- ▁fosse
+- ▁юу
+- ▁Fred
+- ▁harbour
+- slov
+- pov
+- ▁கட
+- ອງ
+- 你没
+- лэг
+- ▁недо
+- ▁wishing
+- 上げる
+- 扱
+- ▁retorted
+- 疑い
+- ▁އެ
+- ivity
+- Ra
+- cado
+- 不管是
+- 无论如何
+- 即将
+- ▁aange
+- ாத
+- ▁lifetime
+- вання
+- 哎
+- kräfte
+- irira
+- ▁defeated
+- დი
+- ▁llega
+- ▁слово
+- ▁آل
+- ▁Kurs
+- ▁lebte
+- ▁Med
+- گرد
+- 制裁
+- 다가
+- 歇
+- ▁broader
+- quant
+- ▁slate
+- ▁запад
+- を終え
+- ▁crawl
+- алт
+- нула
+- ▁рыб
+- ▁verstehe
+- зм
+- 严重的
+- ▁mereka
+- 要不然
+- ありまして
+- ▁Central
+- ▁amazon
+- ▁دارند
+- ниц
+- ▁mater
+- ▁regulatory
+- ▁Verfügung
+- 丘
+- 已经被
+- ▁offerings
+- ▁Tio
+- லே
+- 一路
+- ёл
+- ర్
+- ▁gevoel
+- お天気
+- ▁chorus
+- ▁shallow
+- 祥
+- ▁дерев
+- 的其他
+- amaga
+- ▁trata
+- іст
+- Reg
+- ▁Monaten
+- ▁schwach
+- ▁Bio
+- rew
+- ▁cinco
+- ▁Vertrauen
+- オンライン
+- 辩论
+- 说到
+- ▁dön
+- 宣言
+- ってきました
+- ▁roi
+- ないか
+- 下载
+- 这将是
+- ω
+- ▁phenomenon
+- اغ
+- 売り
+- кас
+- ▁foam
+- ▁tako
+- ▁trumpet
+- kür
+- 野球
+- 行われる
+- 旦
+- ķ
+- ▁platja
+- خانه
+- 打电话给
+- ▁bloody
+- かける
+- ▁attentive
+- ▁reckless
+- 倒是
+- のほうが
+- ▁publicly
+- 教师
+- пля
+- 護
+- 不相信
+- 山口
+- ▁gefährlich
+- ▁johnny
+- ▁punct
+- فهم
+- 形象
+- ▁lump
+- 神奇
+- 不是吗
+- 赚钱
+- 有意义
+- জি
+- ▁йо
+- ▁bonds
+- 겠
+- 糟糕的
+- ▁shaken
+- ປ
+- estudi
+- ▁prose
+- ▁chains
+- 伝わ
+- cac
+- ▁wist
+- قت
+- ▁spectator
+- 庫
+- ▁comparatively
+- ▁zulk
+- ▁notable
+- ▁Пры
+- cida
+- ▁sustained
+- volk
+- ▁сторон
+- ▁Kongress
+- いたのは
+- ần
+- ▁pow
+- ▁waving
+- ▁breathed
+- 供应
+- ▁höchst
+- ▁получил
+- ▁juda
+- 僕も
+- ▁Manchmal
+- ▁banner
+- үз
+- ▁corpse
+- 心中
+- ▁einiger
+- 确实是
+- bericht
+- ▁Orte
+- ▁Angel
+- 服装
+- 柴
+- ▁Аднак
+- 三天
+- ▁Umgebung
+- ▁Flor
+- ▁corte
+- ▁migra
+- ▁expanded
+- ▁positioned
+- 始まる
+- ▁تش
+- ▁Бу
+- 线索
+- чал
+- ▁Big
+- 的父母
+- 融
+- ▁philadelphia
+- ▁concrete
+- ▁hacia
+- いるので
+- ▁niemals
+- gyn
+- ▁schneller
+- ▁compris
+- ることで
+- 有一次
+- 凤
+- ▁tribunal
+- ▁engineers
+- таб
+- drücke
+- ▁voran
+- ▁dumm
+- 싶
+- вшись
+- ▁anstatt
+- ront
+- なんとか
+- garten
+- 恐らく
+- коло
+- '34'
+- 等于
+- acqua
+- 觉得自己
+- ▁manier
+- ▁voz
+- ことが分かりました
+- だけでなく
+- ▁بالا
+- ▁большой
+- 的第一个
+- 누
+- 整体
+- яў
+- 这个世界
+- ▁dishes
+- たくさんの
+- ▁över
+- ▁shocked
+- 经历了
+- 栄
+- ▁pleasing
+- ابت
+- 物理
+- koresha
+- ▁ojos
+- ▁Parliament
+- いらっしゃる
+- பதி
+- 一杯
+- 標
+- 难以
+- ▁dien
+- ▁buena
+- ▁якія
+- の間に
+- ▁naval
+- ▁teori
+- 责
+- ▁plato
+- өрө
+- kera
+- ▁searched
+- ▁remembering
+- ▁ceva
+- ▁interpretation
+- ▁месте
+- ▁deemed
+- ▁positiv
+- azioni
+- ▁exhibition
+- ▁davant
+- ▁dome
+- 少女
+- geschäft
+- 难过
+- sanga
+- ▁loudly
+- ▁sinner
+- werp
+- 马上就
+- rima
+- ▁থেকে
+- 玻璃
+- رح
+- ▁begannen
+- ▁cherish
+- ▁bail
+- ่อ
+- ▁Ara
+- потреб
+- ආ
+- ເ
+- 보다
+- preis
+- ▁McC
+- ▁creep
+- ▁behandelt
+- フル
+- ▁fees
+- اك
+- 打破
+- ستر
+- лены
+- ▁drowned
+- 卑
+- ▁unabhängig
+- ৰি
+- きれい
+- ▁mô
+- 知道我
+- treffen
+- ▁People
+- 你今天
+- 还可以
+- 最初に
+- 新型コロナウイルス
+- ▁announce
+- ▁ocup
+- ▁решение
+- 请求
+- issant
+- trek
+- ▁bout
+- ▁Great
+- 自ら
+- ▁نوع
+- ▁ribbon
+- おいしそう
+- ペン
+- ▁ibintu
+- ziri
+- нский
+- るんですか
+- ▁திரு
+- ancia
+- ▁exploit
+- 厳
+- ▁contemplate
+- كە
+- 工厂
+- foc
+- ▁inquiries
+- ▁observer
+- ▁entry
+- cora
+- িত
+- ивает
+- ҙе
+- hield
+- equip
+- を与え
+- 相同的
+- ▁practic
+- ▁smallest
+- 年度
+- ▁unity
+- 値段
+- 就像我
+- ৰা
+- 一篇
+- ▁reducing
+- ▁مص
+- ▁wrist
+- 小さい
+- 遇到了
+- ▁injustice
+- ච
+- 就是我们
+- ▁gasped
+- 之中
+- ▁zouden
+- home
+- 它们是
+- ▁مختلف
+- 蔵
+- ▁Madame
+- ▁manche
+- 婆
+- intel
+- ようになった
+- ▁দেখ
+- ▁controlled
+- ▁fazla
+- ▁coral
+- remos
+- ▁bending
+- 瞒
+- おっしゃって
+- ▁Não
+- ▁provinces
+- ▁peng
+- ▁мала
+- 你可以看到
+- ▁florida
+- ▁Estas
+- ▁disclose
+- 说服
+- 错过
+- 浸
+- 認
+- ត
+- ▁flexibility
+- ▁entertained
+- 実施
+- ▁diamonds
+- あんた
+- 这首歌
+- ▁frac
+- 生きて
+- '46'
+- ግ
+- の名前
+- ▁thorough
+- ▁multipli
+- ким
+- ▁dahin
+- 尸体
+- ▁seeds
+- 遅れ
+- asso
+- boj
+- bě
+- grand
+- ▁Mini
+- 因为她
+- යක්
+- 仔
+- 岳
+- ▁Lass
+- liku
+- ием
+- 花园
+- ▁besoin
+- ▁anos
+- 诸
+- ▁oogenblik
+- ニューヨーク
+- ▁può
+- ▁British
+- 提到的
+- इ
+- 出口
+- ▁legacy
+- ▁fraud
+- ▁corporation
+- 見えます
+- ัก
+- 歌手
+- بخ
+- ј
+- 藤さん
+- ▁있었
+- 找我
+- ▁Hinter
+- 並
+- ▁vraiment
+- ▁bann
+- வ்
+- টে
+- デビュー
+- ▁consolation
+- 欣赏
+- 変わった
+- 摆脱
+- 我当时
+- 这句话
+- ▁loans
+- ▁verkaufen
+- 飾
+- ▁Sex
+- 飲食店
+- валь
+- ▁alexandr
+- ▁нормально
+- 所谓
+- ▁вещи
+- 余り
+- υ
+- ▁drill
+- ▁preceding
+- 斗争
+- 我们从
+- ▁Stand
+- 的母亲
+- länder
+- imiz
+- 一些事情
+- ▁zweifel
+- ▁quanto
+- ▁salon
+- 穆
+- ▁misschien
+- ▁Motor
+- ▁prev
+- qü
+- ebla
+- 指定
+- szcz
+- ▁ambitious
+- 面包
+- uganda
+- 雾
+- 営
+- بور
+- рин
+- ▁tribu
+- 元素
+- ▁alongside
+- 哀
+- 했는데
+- ▁negli
+- ▁أنها
+- 第五
+- ▁Rede
+- ▁Nova
+- ซ
+- 始めて
+- ▁fuck
+- にとっては
+- 问问
+- ▁tenemos
+- 狭
+- το
+- 人たちが
+- ▁kya
+- ▁Ing
+- ▁हो
+- こうして
+- obten
+- ▁العالم
+- 我才
+- ▁antaŭ
+- ▁yielded
+- ▁treu
+- 说说
+- ында
+- slag
+- ▁bella
+- ▁руки
+- 편
+- 劇
+- ▁krijg
+- '700'
+- 输入
+- ▁erzählte
+- ▁quasi
+- ▁hommes
+- 何度も
+- lý
+- 逐渐
+- ▁банк
+- ▁komt
+- 咩
+- насць
+- ▁dürfen
+- ▁прад
+- ▁இட
+- ங்களை
+- noma
+- რე
+- かれた
+- ыць
+- 邪恶
+- ▁ĉiuj
+- 怎么做
+- 地域の
+- ▁cough
+- ▁многие
+- 膜
+- دام
+- ▁calculation
+- ▁студ
+- 农民
+- に向かって
+- ▁machinery
+- fühl
+- ▁jonge
+- 这就是我们
+- 活用
+- 长时间
+- Ө
+- モデル
+- ▁mandat
+- ▁neighbours
+- 狙
+- వా
+- рот
+- ▁clue
+- ▁mentre
+- алга
+- شون
+- 这样的人
+- ▁fanden
+- ▁weekend
+- ▁heroic
+- ▁Kamera
+- ▁recommended
+- ▁Komp
+- ▁geschlossen
+- ▁Hintergrund
+- ▁normally
+- ▁viu
+- 感染拡大
+- ▁كۆ
+- eqq
+- 代理
+- ▁زمان
+- بدأ
+- alde
+- ▁түүний
+- ▁elector
+- 看一下
+- ▁thor
+- لند
+- ▁побед
+- うえで
+- 竜
+- ▁проблемы
+- ▁стране
+- aggi
+- چە
+- ▁selle
+- ストレート
+- Ү
+- 燃料
+- ▁constructed
+- ▁henri
+- stricken
+- ▁прошло
+- ▁misma
+- шке
+- рош
+- èl
+- ▁ehemalige
+- 亿美元
+- orna
+- тая
+- న్న
+- ということになります
+- 青年
+- gezeichnet
+- ረ
+- Ingenieur
+- ショット
+- ▁tricks
+- 我可
+- ßen
+- ▁glowing
+- 노
+- ▁hymn
+- ▁olive
+- ▁towers
+- 莲
+- sponsor
+- 自体
+- энне
+- фо
+- ▁feeding
+- ▁nueva
+- 下一
+- ▁کردند
+- makers
+- biro
+- 合法
+- 在线
+- andra
+- ▁spraw
+- 再来
+- პ
+- 培养
+- 乔治
+- いただきました
+- ▁знает
+- ▁mason
+- gence
+- umbu
+- どころ
+- ▁داشته
+- போ
+- ▁downstairs
+- ▁بم
+- 出た
+- 是一个非常
+- ▁judged
+- ល
+- muka
+- ▁ashore
+- ▁Ausbildung
+- opera
+- hour
+- بوو
+- 祸
+- ▁collapse
+- 的角色
+- 2%
+- 挨
+- ŵ
+- 脂
+- vallen
+- 急に
+- ▁campo
+- fili
+- ბ
+- тары
+- mien
+- ศ
+- ▁intercourse
+- ผ
+- ▁Rahmen
+- пита
+- ▁gast
+- 好奇
+- 轨
+- 乐队
+- ование
+- 没有办法
+- 脑子
+- ▁traveling
+- ural
+- 笑话
+- ▁polític
+- どこに
+- 这些是
+- 면서
+- ▁Wy
+- ически
+- 话说
+- jó
+- 過ごし
+- ической
+- 鈴木
+- に入れ
+- jährige
+- kurs
+- ▁formidable
+- ▁pinch
+- ▁assigned
+- ▁Können
+- ▁verdienen
+- уют
+- werte
+- ▁fluid
+- ▁پێ
+- брос
+- ▁avoided
+- чих
+- ▁memiliki
+- バランス
+- ▁kell
+- ▁Anem
+- ▁richtigen
+- сси
+- ▁amazed
+- برد
+- بال
+- ▁Quant
+- ▁могли
+- вест
+- ▁supplement
+- ▁Werkzeug
+- 暴露
+- unch
+- ▁terrace
+- voor
+- 戏剧
+- 大好き
+- ète
+- 姜
+- ▁어떻게
+- ▁Figur
+- raba
+- ▁sina
+- 最佳
+- 廷
+- 八年
+- ▁Rücken
+- 大夫
+- lustra
+- ▁flush
+- ▁difícil
+- ▁rejoined
+- ▁Oni
+- رز
+- ▁reinforce
+- 女的
+- ▁patterns
+- ありますね
+- avais
+- ▁ceux
+- çar
+- 膨
+- ▁triste
+- 場面
+- ちゃって
+- луу
+- шиг
+- கூட
+- 成分
+- ▁senza
+- ▁опас
+- ▁negoci
+- flamm
+- wirtschaft
+- もそう
+- 五百
+- 标签
+- ▁Auge
+- woord
+- を守る
+- 坑
+- アジア
+- ▁것도
+- ▁vaccin
+- 隐藏
+- ▁côté
+- теля
+- 复杂的
+- bö
+- ▁shells
+- 크
+- 履
+- それだけ
+- prise
+- control
+- zwei
+- ▁parlament
+- Italia
+- 邓
+- ▁alto
+- ▁chuck
+- していない
+- ならない
+- ▁yaşa
+- ให้
+- альна
+- шёл
+- ▁Obama
+- ▁Road
+- ▁exclamation
+- ▁tới
+- شكل
+- 给予
+- 有利
+- 現実
+- 跟我们
+- 世界各地
+- astro
+- ▁More
+- ▁Vergleich
+- だということです
+- 晃
+- 爆发
+- 父親
+- elimina
+- ▁completion
+- ▁kostenlos
+- ▁wussten
+- 年以上
+- 这么多年
+- たかった
+- ▁acute
+- таў
+- ▁conquered
+- ▁benshi
+- ▁Sau
+- ▁torch
+- ▁мире
+- ▁maintaining
+- ▁spider
+- child
+- ▁baker
+- ▁тийм
+- 鍋
+- ▁dasselbe
+- ▁Best
+- ▁offend
+- ▁язык
+- ▁insects
+- 踢
+- ਸ
+- ئی
+- لَ
+- ▁agitation
+- Alb
+- ▁Bord
+- ▁göre
+- ▁Quer
+- ▁attach
+- ▁sinking
+- 这可能
+- 呼び
+- 朕
+- 彭
+- ▁reluctant
+- ியா
+- zukommen
+- ике
+- ▁좋아
+- 国防
+- 掘
+- ▁Museum
+- ▁saba
+- ▁Zug
+- ▁mußte
+- лена
+- ▁لن
+- ▁độ
+- 老鼠
+- ▁öffnen
+- ▁vorne
+- ▁batter
+- できません
+- Vi
+- 资格
+- ▁hazard
+- gata
+- ▁nancy
+- ▁гос
+- ▁economi
+- 太平洋
+- んじゃないですか
+- ▁выгляд
+- ▁الأمر
+- ▁marvellous
+- 西班牙
+- ходить
+- ▁Party
+- ▁stack
+- دىن
+- 但我们
+- ▁muchas
+- ▁ҡу
+- čí
+- ▁halbe
+- 葡萄
+- ▁Pil
+- ▁dwelt
+- ▁havis
+- ▁adoption
+- ▁différent
+- 各种各样的
+- 당
+- 会社の
+- gebrochen
+- ség
+- 的消息
+- 勢い
+- ின
+- ▁participation
+- ▁fühlt
+- 灵魂
+- 叹
+- дея
+- 指控
+- ▁möglichen
+- schlaf
+- icio
+- lili
+- ▁aufgenommen
+- 各地で
+- ▁logical
+- 目が
+- ▁котором
+- ▁competitors
+- 待ち
+- 配合
+- ▁لأ
+- ▁adjustment
+- 足球
+- ▁muti
+- ▁germ
+- ▁이거
+- の情報
+- labora
+- っち
+- 古代
+- пат
+- ▁cai
+- 自杀
+- 打击
+- ▁charlie
+- лага
+- くれた
+- ▁север
+- kenntnis
+- 可爱
+- 公里
+- 看起来像
+- ▁virtual
+- ▁guitar
+- streit
+- gawa
+- ▁Such
+- 1%
+- ▁weiße
+- ▁ivan
+- ▁خلال
+- ▁Wohl
+- ▁leise
+- ansi
+- 逻辑
+- 하게
+- ▁oxford
+- ште
+- ▁revel
+- ▁sahip
+- bintu
+- ▁tien
+- ▁trauma
+- ▁trước
+- ▁honourable
+- ▁probability
+- Est
+- ▁Mak
+- gereza
+- 矢
+- ▁lớn
+- 回復
+- ▁brisk
+- rende
+- プレ
+- emis
+- 我开始
+- ▁potent
+- 客气
+- 几次
+- 还记得
+- това
+- ▁mevrouw
+- ▁eighteenth
+- ывает
+- ▁fing
+- wohl
+- こちらは
+- 是对的
+- iện
+- jana
+- 得更
+- 连续
+- ▁süre
+- ائل
+- தாக
+- 在里面
+- 走り
+- ▁ardent
+- 的事儿
+- buk
+- ▁chuyện
+- wuchs
+- زى
+- ▁allgemeine
+- 肠
+- 吻
+- 抱歉
+- 得太
+- ゼロ
+- 这个东西
+- ▁March
+- 听说过
+- ▁زمین
+- ▁fragment
+- ▁بودند
+- 舒服
+- ジャー
+- öpfe
+- ▁dieselbe
+- 要請
+- ▁reasoning
+- modell
+- لات
+- xxam
+- 斯坦
+- 的天气
+- ▁خل
+- ▁cùng
+- introdu
+- 有名
+- Й
+- 稀
+- meni
+- ▁Proto
+- 这是你
+- vocation
+- 大丈夫です
+- ▁плане
+- なもの
+- ▁Erfahrungen
+- しましたが
+- 賃
+- ▁welcher
+- ▁riep
+- ▁legisla
+- けた
+- ▁мной
+- hong
+- ▁você
+- ▁baseball
+- ▁slap
+- objet
+- ▁Nda
+- ▁شيء
+- ಯ
+- ijas
+- vēl
+- ĝo
+- mada
+- ▁mystic
+- EC
+- 課
+- ▁experts
+- 杂志
+- 昭和
+- 因为这
+- ▁yose
+- ▁preference
+- ▁Flug
+- 簡単
+- ▁impatience
+- 쓰
+- プレゼント
+- หน
+- ▁ولی
+- ▁slay
+- ▁så
+- 今後の
+- ▁числе
+- ▁ຢູ່
+- ▁хотите
+- ▁никаких
+- ▁நட
+- lette
+- mong
+- していると
+- ▁več
+- ▁dismissed
+- ▁Wissenschaftler
+- ▁liquor
+- ▁pursuing
+- を目指す
+- glaub
+- бро
+- ▁buff
+- 下班
+- ▁ilk
+- ▁Untersuchung
+- ▁Tradition
+- ▁linked
+- ▁knit
+- ▁successor
+- linie
+- ▁Matt
+- ▁количество
+- ▁French
+- センチ
+- நேர
+- ário
+- ▁insect
+- aigua
+- qq
+- アフリカ
+- ރު
+- キング
+- の一つ
+- ▁converted
+- ▁vault
+- wain
+- schel
+- samkeit
+- ỉ
+- ▁personnes
+- ▁staircase
+- 咨询
+- ▁slumber
+- ▁Со
+- corr
+- schicht
+- ▁clasped
+- sigur
+- ▁concur
+- 姉
+- ▁hẽe
+- ▁pueblo
+- ▁Cat
+- 任何事情
+- ▁جهان
+- 去哪儿
+- нных
+- marin
+- kaya
+- ▁Todes
+- ләр
+- ▁Gan
+- ੇ
+- ▁routine
+- 竞选
+- 如果是
+- 生病
+- ▁punished
+- ▁libre
+- قات
+- ▁bamb
+- ▁demonstration
+- ▁retained
+- ▁nhìn
+- ▁엄마
+- ▁Worten
+- kapa
+- ල්
+- ▁siege
+- ▁üç
+- を伝え
+- 女生
+- ▁schützen
+- ▁família
+- 严格
+- ▁singer
+- 青春
+- ▁Besitz
+- ▁poems
+- しております
+- 考试
+- わら
+- 女の子
+- バル
+- ▁Merc
+- ▁scope
+- なきゃ
+- 不是一个
+- ▁loyalty
+- 躺
+- 研究所
+- ▁juffrouw
+- 英尺
+- ▁verkauft
+- груз
+- ▁jongen
+- 贝尔
+- ▁أع
+- ▁pai
+- 读书
+- 现在已经
+- 问道
+- 很长
+- щих
+- esca
+- ckel
+- ▁thanked
+- ▁Produktion
+- ▁Milliarden
+- 子供たち
+- ▁bodily
+- gada
+- 鉄道
+- گل
+- 显
+- ▁Both
+- ▁carrier
+- fér
+- aime
+- 的许多
+- arrêt
+- profit
+- ▁breathless
+- いたら
+- 妖
+- が一番
+- ▁verbessern
+- 瘦
+- ▁mall
+- ないので
+- ▁traces
+- ▁timp
+- 后悔
+- téri
+- 向前
+- یز
+- 範囲
+- ▁dealt
+- 乖
+- ▁desirable
+- 去看看
+- 考える
+- ▁erster
+- лик
+- ▁рассказыва
+- サイト
+- ıldı
+- клон
+- 即使是
+- ▁Home
+- ngingo
+- 際に
+- ▁abode
+- してます
+- ▁всю
+- ▁près
+- 興味
+- 街道
+- wè
+- ški
+- ▁precaution
+- 芽
+- ▁원래
+- 解决方案
+- ▁이러
+- 届け
+- ▁collective
+- ▁pious
+- kina
+- ▁Struktur
+- tata
+- 든
+- ▁trotzdem
+- AR
+- ▁offensive
+- おき
+- Tech
+- ▁Ал
+- 最后一个
+- ▁Dorf
+- ▁Deutschland
+- ちゃんの
+- してほしい
+- ▁streng
+- வும்
+- ▁horrid
+- ▁Kontakt
+- ▁molly
+- 牧师
+- sprache
+- ▁Haushalt
+- 昌
+- ▁Fünf
+- ▁regain
+- ▁Ländern
+- 考えた
+- 一起去
+- ህ
+- ▁terrified
+- ▁learnt
+- ▁witnessed
+- ▁trov
+- ▁keiner
+- ▁Beziehungen
+- 把我们
+- زل
+- ▁amafaranga
+- 起来了
+- ▁franchise
+- ▁abundance
+- ▁atlantic
+- ▁airport
+- كس
+- せない
+- kong
+- ▁conclu
+- 的态度
+- 的音乐
+- ▁Sind
+- 蜂
+- ▁nữa
+- たんですけど
+- 回报
+- ுடைய
+- ▁domini
+- ▁shillings
+- ▁encara
+- ▁entgegen
+- ţă
+- виз
+- ▁обще
+- ަށް
+- ▁Verwaltung
+- ▁شروع
+- ▁Aktivität
+- 癌症
+- yandi
+- ▁seulement
+- 得好
+- esprit
+- yaga
+- 想办法
+- ▁Francisco
+- の予想
+- ▁Wein
+- 晶
+- ït
+- تنا
+- ▁serie
+- ▁characteristics
+- ▁mesmo
+- ▁Schulter
+- 阔
+- ▁کے
+- laki
+- nood
+- 的状态
+- sett
+- フト
+- ▁Virginia
+- メーカー
+- ▁acum
+- ▁Vila
+- muş
+- кана
+- カラ
+- ▁tract
+- ▁шар
+- fordern
+- スマホ
+- 季節
+- ▁داده
+- ново
+- 減少
+- 任何东西
+- ▁части
+- ები
+- යේ
+- へん
+- ▁consolid
+- 惩罚
+- ▁Krebs
+- ▁pregunta
+- ▁дараа
+- ▁barri
+- ▁кроме
+- ▁поле
+- 受欢迎
+- коў
+- lux
+- 柜
+- iek
+- 店舗
+- itari
+- 参考
+- भा
+- ▁договор
+- ▁recess
+- atura
+- 识别
+- ▁bieten
+- ▁என
+- 換
+- ▁Fortschritt
+- ▁trotz
+- ▁youngest
+- कार
+- 对对对
+- க்கிற
+- 跑了
+- 予約
+- 颗
+- ▁lawyers
+- ▁своим
+- ▁Nya
+- 嫂子
+- ▁mining
+- ▁submitted
+- ▁кил
+- ▁guided
+- 女性の
+- 안
+- 迁
+- ทํา
+- ▁bắt
+- ওয়া
+- 温泉
+- नी
+- ▁bike
+- ▁tossed
+- ஸ்ட
+- ▁Brand
+- ▁ثم
+- ▁Ти
+- 纠
+- ▁சரி
+- 었어
+- ▁emerged
+- ▁versuche
+- これまでの
+- 包含
+- ▁offended
+- ▁già
+- ▁passer
+- 您说
+- 锦
+- klin
+- ▁rechten
+- 地球上
+- тара
+- ▁machten
+- 下次
+- ▁privat
+- 疾
+- ను
+- ▁slice
+- தற்கு
+- ▁destination
+- てしまった
+- дали
+- 你可能会
+- ▁comprehensive
+- ワイ
+- 数が
+- τα
+- amiento
+- рать
+- ▁Theorie
+- らせ
+- Music
+- ▁columns
+- ▁зрения
+- 坊
+- ▁incapable
+- 내
+- 一根
+- ▁Jun
+- ▁guerre
+- ▁prudence
+- ▁spielte
+- жим
+- kiwa
+- කි
+- ▁relax
+- ifiziert
+- ▁Slide
+- ▁errand
+- ▁drawer
+- 年生
+- 落とし
+- てない
+- ▁reserved
+- ▁мира
+- 惹
+- 鶏
+- ▁suffice
+- ▁premium
+- ▁handful
+- été
+- ▁олон
+- であれば
+- party
+- ▁истории
+- 看待
+- ▁работы
+- ▁اینکه
+- ▁borders
+- 最大の
+- енным
+- 終了
+- čno
+- ▁winding
+- 加拿大
+- あんな
+- ▁Johnson
+- ってください
+- beera
+- ▁dreaming
+- ▁tropical
+- 方案
+- ویل
+- ▁georgia
+- සා
+- ▁있고
+- ▁amidst
+- 扯
+- เขา
+- ▁emerging
+- ▁Roger
+- ▁projet
+- ستی
+- ▁Gel
+- ▁drap
+- ▁spit
+- hund
+- мак
+- 议员
+- 际
+- zusetzen
+- ピッチャー
+- 意大利
+- விட
+- رض
+- ▁rép
+- ▁хө
+- ▁Long
+- 带来的
+- ▁слож
+- 扮演
+- нк
+- ▁அறி
+- ▁converse
+- 超越
+- 引き続き
+- JR
+- 大手
+- fowl
+- pata
+- ▁goddess
+- 妃
+- ▁commend
+- ディー
+- рис
+- ▁Hotel
+- ラスト
+- ním
+- rän
+- gah
+- 多个
+- 教え
+- 佐藤
+- ▁boldly
+- 悩み
+- ▁которого
+- 自転車
+- ちゃんが
+- 核心
+- vacu
+- ▁resent
+- ▁último
+- 的大脑
+- 发言
+- cule
+- ▁wählen
+- ―
+- 辱
+- 강
+- ruka
+- 傾向
+- еду
+- ▁reicht
+- ▁répondit
+- дин
+- 绳
+- 雕
+- 你知道我
+- 工作人员
+- ▁boiling
+- ▁моск
+- 顺利
+- ▁polic
+- terior
+- ▁sect
+- 荣誉
+- хэн
+- わせ
+- irdi
+- itatea
+- ▁draft
+- 的衣服
+- تور
+- 3%
+- 堵
+- யோ
+- ාව
+- να
+- raub
+- 悟
+- 侯
+- ▁entschieden
+- ▁Zeichen
+- 感謝
+- ▁Planeten
+- ▁fueron
+- 就知道
+- ▁loi
+- iteit
+- ▁gestern
+- ▁otras
+- ▁Öl
+- ▁neighbour
+- ▁standpoint
+- сар
+- 完全に
+- 不舒服
+- VTR
+- ▁spaniards
+- ▁Sozial
+- ▁dirt
+- ▁hwn
+- 颜
+- 晨
+- abanya
+- рий
+- ▁chiar
+- 斤
+- ▁getötet
+- ▁unnecessary
+- bizi
+- brew
+- ▁پار
+- Plattform
+- ▁gewonnen
+- 規制
+- 发送
+- を使う
+- ▁priority
+- ▁Geheimnis
+- ▁suspended
+- 同時に
+- geteilt
+- ▁willingly
+- ▁seien
+- ▁Unser
+- 不动
+- ▁Fund
+- AS
+- ▁nama
+- ކަ
+- 感动
+- ţii
+- american
+- 你们都
+- ▁justify
+- ▁steward
+- ▁Kim
+- 舅
+- мур
+- 袭击
+- اشت
+- ▁реально
+- 正直
+- 没问题
+- līdz
+- 最後は
+- 阶
+- 남
+- ▁bản
+- ▁이게
+- ▁இரண்டு
+- 歴
+- ▁нэр
+- ▁Wol
+- ▁clung
+- 大使
+- ▁Flo
+- ▁Industrie
+- udge
+- ▁massa
+- ▁此外
+- ▁clergy
+- のもと
+- 免疫
+- ▁работает
+- 相撲
+- ▁থাক
+- ▁candidate
+- かなと思います
+- 跟大家
+- ordre
+- hala
+- ▁canadian
+- 小朋友
+- aña
+- ▁coordina
+- teeka
+- plin
+- といって
+- 抗议
+- 羞
+- ង
+- ▁spreading
+- ▁sadece
+- ▁einzelne
+- 的任务
+- ▁Tā
+- ▁sinh
+- ▁allies
+- ▁behave
+- ▁wildly
+- ▁generated
+- パラリンピック
+- 这段
+- 明星
+- سبب
+- 時間を
+- ▁strand
+- presi
+- 준
+- 抱怨
+- 首脳
+- ▁Market
+- ▁Harris
+- ▁Unit
+- atrice
+- boden
+- 清晰
+- ماذا
+- 大学の
+- ţe
+- விட்ட
+- ▁innovative
+- ▁database
+- 肯定会
+- ▁capa
+- ▁actua
+- morph
+- ▁einzelnen
+- bali
+- ▁juan
+- いだ
+- 忍受
+- ▁creator
+- ▁Denk
+- urwa
+- ս
+- ▁artillery
+- މަ
+- ▁توسط
+- ▁gilbert
+- んや
+- stürzt
+- ▁telah
+- あい
+- ▁situé
+- 笔记
+- ってくれ
+- ضحك
+- ▁anthony
+- wissenschaft
+- しゃ
+- ▁tante
+- 一点点
+- ▁большая
+- menya
+- ӗ
+- ရ
+- 规划
+- ▁vilaĝo
+- 合って
+- ▁preacher
+- ▁ноль
+- ▁verder
+- 从事
+- ▁spreche
+- ▁potatoes
+- ▁corpo
+- था
+- ▁faz
+- ਨ
+- ▁geliebt
+- 放松
+- 的产品
+- ▁Angriff
+- ▁thật
+- ▁irgendwo
+- curi
+- 高校生
+- ▁petr
+- ▁Their
+- проч
+- ニュー
+- ▁spared
+- ފ
+- 这个节目
+- 戒
+- ▁Nachde
+- 千葉
+- ibyo
+- τη
+- 就好
+- ▁fortnight
+- 周末
+- 見ていきます
+- 我想知道
+- хам
+- ▁consumption
+- ▁আৰু
+- ▁sympathetic
+- ▁Konzept
+- んだよね
+- ▁Geräusch
+- んだって
+- ▁subsequently
+- ▁Russia
+- ▁کسی
+- ことがある
+- ▁afforded
+- ự
+- まえ
+- pfa
+- flug
+- ▁queste
+- ▁progressive
+- 似的
+- mita
+- кет
+- ▁sentir
+- 桌子
+- ▁хочет
+- ▁Rasse
+- ▁Fällen
+- anın
+- があるんです
+- 築
+- рван
+- と共に
+- ▁Dick
+- マジ
+- 此时
+- ▁Spo
+- ▁bonnet
+- ҥ
+- āju
+- 扩大
+- 这两
+- 体重
+- ෙන්
+- шил
+- مَ
+- 也能
+- tiere
+- 但在
+- 唤
+- ▁çocuk
+- ▁heutigen
+- 冬天
+- ▁Put
+- 彼らは
+- ▁guarded
+- ▁نشان
+- 向我
+- قام
+- stained
+- ▁estoy
+- 靠近
+- ▁protein
+- ப்பட்டது
+- 争论
+- ați
+- ่น
+- ▁buchstäblich
+- ▁усё
+- ▁Lor
+- тура
+- ▁pensi
+- まん
+- وض
+- ▁evangeli
+- 是非
+- полага
+- 原始
+- عي
+- ிலும்
+- мян
+- 不信
+- afi
+- ድ
+- 拓
+- ふだん
+- 消失了
+- ▁يتم
+- ▁amusing
+- ▁punch
+- ▁när
+- 听到了
+- 资产
+- ▁Kath
+- 哭了
+- ▁tremble
+- ▁leiden
+- ánh
+- 有助于
+- ▁prosper
+- itse
+- ▁عليه
+- рев
+- சை
+- ▁Haut
+- ▁внимание
+- 详细
+- ▁있잖아
+- 投資
+- ▁garanti
+- град
+- ▁가서
+- ▁другом
+- ごと
+- 新鲜
+- ▁faintly
+- ▁Amazon
+- ▁Ан
+- 耀
+- ▁Erklärung
+- IC
+- 反而
+- quarter
+- ▁来一首
+- 性别
+- 罩
+- ▁charlotte
+- ▁attained
+- ▁детей
+- ▁profond
+- ▁نیز
+- ▁Яны
+- 但是如果
+- ▁Männern
+- ▁поскольку
+- 配置
+- gespielt
+- ▁eran
+- ▁Fur
+- ▁algunos
+- ▁unua
+- ▁отношения
+- 菊
+- 姐妹
+- 選手権
+- 护理
+- ▁strive
+- wright
+- 誓
+- ▁halted
+- Argent
+- ▁поводу
+- ▁morir
+- ▁Gewinn
+- ▁seguir
+- ▁requested
+- '37'
+- ▁Alice
+- 我跟你
+- ▁schlimm
+- ▁генерал
+- 挖
+- χ
+- ▁кооператив
+- 蓄
+- 自動車
+- ▁merchants
+- ▁그래도
+- autor
+- ▁Те
+- ▁wichtigsten
+- 考虑到
+- нә
+- ▁Ray
+- 埼玉県
+- ź
+- 咪
+- ▁удар
+- 広がり
+- 叫什么
+- 或许
+- ûn
+- ▁November
+- geworfen
+- ধা
+- ▁assuming
+- ▁Außen
+- ▁Debatte
+- ▁nhân
+- 利亚
+- ▁muslim
+- гээ
+- Michel
+- ▁serait
+- 山さん
+- ன்று
+- bris
+- ▁telegram
+- 笨
+- 陶
+- عرض
+- '198'
+- 調べています
+- 会发生什么
+- ▁smash
+- ▁петр
+- 盲
+- ෂ
+- ▁вроде
+- ▁sólo
+- ▁Times
+- 信用
+- 술
+- 파
+- 訪問
+- ▁mı
+- komst
+- ඳ
+- ▁александр
+- ▁happier
+- 毫无
+- ▁Roma
+- ▁kalt
+- niveau
+- қә
+- kurikira
+- 終わった
+- твар
+- poro
+- аваць
+- 你就是
+- ▁deserved
+- īja
+- '33'
+- ¡
+- 陰
+- ▁speculation
+- 警察官
+- 大きさ
+- 八十
+- visa
+- ▁rey
+- 十万
+- seks
+- 也得
+- 去年の
+- ▁Andrew
+- 我刚才
+- 这可能是
+- tausch
+- 继
+- ▁conspicuous
+- ▁remembrance
+- gebung
+- ▁скажу
+- 旅游
+- ▁jedno
+- schreiben
+- 相处
+- ▁respectful
+- ▁plague
+- すべての
+- 不幸
+- чна
+- ென்
+- kuta
+- ▁vehicles
+- stell
+- 每周
+- 之处
+- 给您
+- ацион
+- ▁Anna
+- わけですね
+- म्
+- ▁tempted
+- おととい
+- ▁جای
+- 而已
+- 赤ちゃん
+- covid
+- page
+- ระ
+- ▁fig
+- ッツ
+- 年前の
+- ▁theories
+- 究竟
+- −
+- wanga
+- prost
+- ▁pony
+- ▁heures
+- ▁samuel
+- seitig
+- 就跟
+- ▁дуу
+- どうなる
+- かかり
+- 整天
+- もう一つ
+- 倾向于
+- 溪
+- ▁prevailed
+- ▁یاد
+- ▁climbing
+- пот
+- ▁sõ
+- führt
+- чо
+- ▁investigate
+- sobola
+- jski
+- 忧
+- दि
+- ▁mike
+- ▁Song
+- wish
+- ▁Tä
+- 形状
+- စ
+- asyon
+- ▁roast
+- haired
+- ▁vais
+- 記者
+- тим
+- ▁luke
+- ඒ
+- ặ
+- ▁confederate
+- 올
+- ▁masa
+- credi
+- ەکان
+- Work
+- udio
+- ▁disagree
+- 向他
+- 傘
+- kämpft
+- ▁довольно
+- Смех
+- ▁vraag
+- drag
+- ▁proto
+- ▁sûr
+- 的建议
+- ▁seeming
+- حمل
+- 耍
+- ▁whereupon
+- ▁carries
+- 抵抗
+- の結果
+- 陵
+- ▁Zwischen
+- 매
+- 菅
+- 博物馆
+- 六个
+- werken
+- すぎる
+- ▁darunter
+- ▁intervention
+- を示しました
+- ▁implementation
+- 共有
+- cji
+- ▁embargo
+- 줄
+- ▁negotiations
+- ▁torrent
+- rilla
+- 都已经
+- 更重要
+- 水分
+- డు
+- ▁شوند
+- ▁null
+- خارج
+- ▁usage
+- をしていた
+- 扶
+- 문
+- 산
+- 列表
+- ▁suited
+- улы
+- ▁выступ
+- それも
+- 袖
+- 次は
+- 始めました
+- ▁approximately
+- を続けて
+- 这座
+- ێن
+- ▁compass
+- ций
+- ▁quaranta
+- ▁tym
+- ▁bibli
+- مات
+- ▁بىر
+- klad
+- ировал
+- 你不要
+- ▁необходимо
+- ▁promising
+- ▁Meilen
+- lege
+- loge
+- ▁figured
+- 申し
+- ときは
+- ▁equality
+- 之类的
+- ▁erhob
+- ケーキ
+- 認識
+- ▁reconcile
+- ▁yabo
+- ▁debat
+- コード
+- haya
+- 那我就
+- 惨
+- 昆
+- ▁слишком
+- お互い
+- 交渉
+- ▁Daher
+- ▁plea
+- ▁habia
+- 会让
+- чны
+- 孕
+- 笑了
+- 喜马拉雅
+- 訓練
+- ▁говорили
+- ▁места
+- 編
+- рма
+- 組み
+- 就开始
+- त्र
+- ▁تخ
+- вший
+- ▁Grenzen
+- 我以为
+- 注册
+- 伝統
+- ரே
+- ▁Sendung
+- ▁souvent
+- 绝望
+- ▁gesicht
+- гьы
+- かけた
+- ▁franz
+- 砸
+- ▁ఆఁ
+- ▁pauvre
+- ▁rejoice
+- 本周
+- lamp
+- נ
+- ▁cuộc
+- ▁donald
+- 行きます
+- 这是一个非常
+- тав
+- 上がった
+- 就是一个
+- istischen
+- ή
+- кина
+- に行った
+- ▁Nick
+- నా
+- ▁clearing
+- ▁tinha
+- 에는
+- タル
+- 乃
+- zeichnung
+- ▁товарищество
+- 桜
+- 卒業
+- يه
+- ▁luz
+- 왔
+- ▁qo
+- ladı
+- baum
+- 完美的
+- 疗
+- いるという
+- ▁verdient
+- 오
+- ▁слыш
+- ище
+- ▁tact
+- middel
+- 迎えた
+- 玲
+- お二人
+- ▁curtains
+- alimenta
+- ▁tooth
+- のですが
+- ▁foresee
+- Risas
+- àtic
+- コメント
+- 产业
+- ▁Lager
+- gewalt
+- んでしょう
+- حص
+- ▁fühle
+- ▁respected
+- 宣传
+- 行吗
+- たこと
+- चा
+- bürger
+- ▁alchemist
+- ▁Jones
+- чака
+- ▁шест
+- ľ
+- 俱乐部
+- ▁денег
+- 繰り
+- karte
+- 黎
+- 比你
+- ▁shrill
+- піс
+- ▁accordance
+- තා
+- ▁بشكل
+- 营销
+- ▁halo
+- ▁washing
+- ▁heath
+- ワーク
+- ▁mature
+- ▁Wohnung
+- ▁representatives
+- information
+- ھا
+- material
+- ▁rosy
+- ▁regulate
+- ▁cinquanta
+- ▁veteran
+- dov
+- 只需要
+- ▁hull
+- ▁但是他们
+- 跟她
+- 产生了
+- 時には
+- 一颗
+- ▁riot
+- 这并不
+- ▁pagan
+- ▁veux
+- ▁evolu
+- luck
+- ▁será
+- ▁übrig
+- ▁resta
+- 负担
+- ▁başka
+- ▁пример
+- ▁parle
+- ন্ত
+- estro
+- ▁Regeln
+- vió
+- 尿
+- ුණ
+- 素晴らしい
+- ▁தொ
+- pí
+- phra
+- ▁efter
+- 糸
+- ု
+- 楽しめる
+- ▁persian
+- ▁Fil
+- ▁возник
+- 了下来
+- ▁willkommen
+- 饱
+- 糟
+- 畑
+- ▁gigantic
+- ▁buffalo
+- ▁meditation
+- ユー
+- ▁solemnly
+- よろしくお願いします
+- ▁explica
+- 溜
+- ਦ
+- ▁wünschte
+- 備え
+- lige
+- ubwi
+- виду
+- 我不认为
+- очка
+- خان
+- ancien
+- ▁supposing
+- ▁gloves
+- ▁lahko
+- ▁работу
+- 访
+- sional
+- 所以我想
+- ከ
+- ገ
+- และ
+- シリーズ
+- てくれた
+- ▁смог
+- ▁Lied
+- の方に
+- ▁confront
+- ▁சில
+- ▁parler
+- ▁agnes
+- 坎
+- 撕
+- ▁medizinische
+- 少数
+- Man
+- 正义
+- 微博
+- ıcı
+- ▁jungle
+- طب
+- ▁vicar
+- ▁Grad
+- ▁thereof
+- 像你
+- lach
+- '38'
+- ▁Verständnis
+- ▁энерг
+- ギャ
+- ▁blog
+- 一个新的
+- ▁samo
+- IP
+- ▁junior
+- ▁kümmern
+- ▁commis
+- burger
+- 機会
+- ▁хол
+- ブリ
+- ▁توان
+- อง
+- 沼
+- ▁calendar
+- ▁angefangen
+- ▁கொ
+- 这时
+- 敏感
+- ▁tlie
+- というのを
+- ▁fürchte
+- excursió
+- ራ
+- ツアー
+- 天堂
+- ▁oficial
+- 達成
+- bouw
+- ▁forbidden
+- 全ての
+- 価値
+- ▁bruce
+- ジー
+- 胶
+- ▁Lernen
+- 我有一个
+- 野党
+- обр
+- あさって
+- いるのか
+- ▁nützlich
+- ▁partial
+- ▁tempt
+- يون
+- シュート
+- ▁dachten
+- 平安
+- ▁ruled
+- ▁integrated
+- 新型コロナの
+- ▁grind
+- vina
+- ▁сильно
+- ▁месца
+- んだから
+- 可以用
+- ▁dernier
+- сет
+- cato
+- ▁голову
+- raient
+- ▁ayuda
+- くなった
+- 開幕
+- baye
+- ▁гри
+- 所以它
+- 及时
+- プラス
+- ნი
+- 驾
+- ▁mortgage
+- ▁awesome
+- 怀孕
+- ▁нават
+- дела
+- ▁float
+- ▁Revolution
+- 一看
+- 違います
+- ▁садовое
+- 槽
+- ▁invasion
+- ▁особо
+- ▁splash
+- 相手の
+- عاد
+- ▁dla
+- 国連
+- ▁contend
+- ▁bubble
+- ਹ
+- 只会
+- ▁tính
+- ▁martha
+- ▁چرا
+- 你不知道
+- risti
+- ▁хто
+- の天気
+- paka
+- ▁المت
+- 建て
+- коммерческое
+- ▁tersebut
+- 細胞
+- ▁snapped
+- حدود
+- 进化
+- natured
+- 围绕
+- toren
+- 臓
+- 稼
+- 我们认为
+- 新型コロナウイルスの
+- ▁adequate
+- 监控
+- ▁precipita
+- ▁lorsqu
+- venue
+- 股票
+- ったこと
+- ▁inspire
+- ▁verließ
+- ▁Га
+- ▁sensor
+- 空中
+- ठ
+- ਕ
+- ▁Aspekt
+- ▁intens
+- শা
+- 見つかった
+- ▁remedy
+- みたいに
+- ったのは
+- ▁greatness
+- ▁comunica
+- гул
+- ఆ
+- かさ
+- っていうか
+- 楽しんで
+- 立法
+- ▁müsste
+- licit
+- BC
+- ▁دهد
+- ▁Ansicht
+- ▁sozusagen
+- ▁awhile
+- த்துக்
+- ▁Apple
+- ▁obliga
+- ▁dolor
+- іль
+- ுள்ளது
+- ▁Vari
+- 批准
+- ▁components
+- ▁Latin
+- ▁rebellion
+- 预算
+- ▁genial
+- ক্ত
+- 爆発
+- ▁Mannes
+- tawa
+- ▁சொல்ல
+- 炒
+- ▁thursday
+- මු
+- гур
+- ▁gasp
+- ▁focusing
+- ▁June
+- 拿出
+- ▁Bald
+- ▁естественно
+- ▁настоящ
+- 棍
+- 記憶
+- ▁chariot
+- ▁comrade
+- ▁лицо
+- ▁rachel
+- メント
+- 九点
+- ▁altri
+- 有更多的
+- ▁nuit
+- 㗎
+- エル
+- ▁resolute
+- ensemble
+- চি
+- 就得
+- 的项目
+- ருக்கு
+- ▁nannte
+- ̣
+- ▁Verantwortung
+- 所以这是
+- 将会
+- ▁lemon
+- gänge
+- 涉及
+- ョ
+- kili
+- eṭṭ
+- ▁Antrag
+- ▁blanche
+- されていました
+- ▁Gru
+- 戻って
+- ▁تنها
+- ことによって
+- ▁Town
+- ▁leider
+- mali
+- ▁yep
+- 你妈
+- 塗
+- 梅雨
+- ▁добро
+- 的速度
+- ▁которым
+- 芸人
+- 選手たち
+- kord
+- bald
+- ▁Bol
+- plant
+- agh
+- 伸ばし
+- кала
+- 我们不能
+- 欣
+- ▁preach
+- ▁breit
+- ▁இருக்க
+- ▁செய்ய
+- ▁pilgrim
+- ▁voix
+- ▁alguns
+- ▁veya
+- 请问
+- заў
+- どこか
+- ического
+- ती
+- rauch
+- 的概念
+- PR
+- щие
+- ranye
+- ▁één
+- ▁fé
+- 政治家
+- drop
+- 取消
+- とみられる
+- ▁accomplishment
+- ▁vegeta
+- ▁camin
+- アプリ
+- ▁assumptions
+- mütig
+- ailleurs
+- ▁Во
+- wadde
+- prozess
+- ▁resulting
+- жды
+- こん
+- открыт
+- асці
+- ފަ
+- ensor
+- zaam
+- ▁actively
+- ksa
+- ▁Ці
+- juk
+- эш
+- körper
+- インタビュー
+- 来来来
+- yama
+- чным
+- issait
+- ▁uncommon
+- ▁despatch
+- 但他们
+- ▁klas
+- ▁Valley
+- ▁часто
+- ▁tế
+- 奪
+- ▁Open
+- န
+- ▁façon
+- なんでしょうか
+- ▁gag
+- ▁Start
+- ▁отдел
+- 说法
+- ▁добры
+- いこう
+- grä
+- ▁louise
+- ▁gewisser
+- reiche
+- 崩れ
+- ▁fulfilled
+- 措置
+- 这话
+- ▁gedaan
+- いたい
+- ▁wail
+- diye
+- 近づ
+- ▁тобой
+- ▁sheriff
+- 限定
+- 校长
+- 運営
+- ▁gyfer
+- adju
+- そうですよね
+- 柯
+- ▁occurrence
+- ട
+- 这一次
+- ▁folgte
+- 过得
+- 的专辑
+- 夜晚
+- 事務所
+- 惜
+- ▁aunque
+- 等到
+- ボールを
+- 夏の
+- ▁falta
+- ▁frühen
+- ▁daraus
+- ▁saith
+- ▁Mira
+- ▁sanft
+- 尴尬
+- んですかね
+- ▁religiöse
+- ▁nennt
+- дель
+- ▁شک
+- ▁childish
+- ▁мнение
+- ▁stalk
+- ▁خوش
+- Һ
+- ▁bởi
+- 見込みです
+- ▁kişi
+- ▁acaba
+- ▁Live
+- ruff
+- ▁approve
+- ေ
+- 賛
+- züge
+- রু
+- 工夫
+- 铜
+- ▁همچنین
+- 형
+- ▁reflecting
+- 入れた
+- 也不知道
+- 醒来
+- related
+- аас
+- ▁filter
+- ▁sunny
+- ▁cuenta
+- 蛮
+- 熬
+- ▁catherine
+- ▁respekt
+- 尽快
+- ▁vertraut
+- ▁cautious
+- 겠다
+- kami
+- ▁États
+- ▁warrior
+- ▁circular
+- ▁orang
+- チェ
+- ▁specially
+- قان
+- ▁doom
+- 裔
+- ▁kubi
+- ▁raid
+- わず
+- 你又
+- ▁erklärte
+- われた
+- hurst
+- ▁strait
+- 弄清楚
+- ▁класс
+- 它可以
+- ありますか
+- 拐
+- 체
+- バイデン大統領
+- 揺れ
+- gibt
+- reise
+- いただいた
+- ▁dump
+- ▁meek
+- ▁hayi
+- proof
+- fassung
+- ▁Stock
+- 坡
+- 煤
+- 遗憾
+- ▁Experiment
+- トマト
+- 协会
+- ▁contribu
+- 东北
+- ▁aynı
+- sız
+- ▁geschaffen
+- ▁secondary
+- ▁pitt
+- auto
+- гуля
+- mies
+- lohn
+- 離れた
+- сот
+- 所有人都
+- スペイン
+- 維持
+- ▁Tema
+- ▁আমি
+- 日まで
+- 銀行
+- 整理
+- 原子
+- 殿下
+- вернул
+- ▁goede
+- ▁relates
+- 近くの
+- ▁клуб
+- ▁douglas
+- 仪式
+- ▁paz
+- ▁implica
+- কি
+- korera
+- 试试
+- ▁translated
+- ▁тое
+- 单独
+- ியது
+- ▁algorithm
+- 몇
+- ▁случай
+- ▁anterior
+- 俳優
+- ▁comic
+- ▁candidates
+- もうちょっと
+- 一緒
+- ość
+- ▁nước
+- ▁thay
+- ▁policeman
+- 足を
+- 測
+- ▁combine
+- ார்கள்
+- ▁Pour
+- ▁gaining
+- 更容易
+- ▁bloß
+- ▁Punkte
+- જ
+- ▁chocolate
+- ည
+- 労働
+- それぞれの
+- ▁robbed
+- ▁fourteenth
+- 专辑
+- 物価
+- 北方
+- یده
+- ▁commanding
+- বু
+- ▁joyous
+- ▁drehte
+- ▁товарищ
+- ▁eingesetzt
+- ▁sullen
+- 五千
+- 姑
+- ▁девушк
+- 我以前
+- ▁elderly
+- を取って
+- けさ
+- 日目
+- 织
+- ▁elektr
+- ▁hearty
+- ▁trigger
+- トル
+- ▁waarom
+- ▁zač
+- หล
+- ▁schaute
+- ▁starve
+- erade
+- ▁yapma
+- ▁precise
+- ▁kinderen
+- ▁حول
+- rolle
+- ▁productive
+- мова
+- ▁bằng
+- 酬
+- ▁შე
+- ulin
+- ▁regulations
+- 直後
+- ▁насколько
+- 祝福
+- 结论
+- ▁Kanada
+- ▁philosophical
+- ▁slash
+- dynami
+- 他是一个
+- くなります
+- ▁shelf
+- '95'
+- ますけど
+- ▁deprived
+- ▁stray
+- 出门
+- 小屋
+- そうそう
+- вался
+- ▁mujer
+- ու
+- 泄
+- ▁моей
+- 窃
+- ▁überrascht
+- ▁diğer
+- ▁حتى
+- ▁loin
+- 雨雲が
+- ▁importantly
+- ▁fiery
+- ▁contradict
+- ▁Яна
+- ▁balloon
+- ając
+- ▁einfacher
+- ▁bestowed
+- 島の
+- город
+- ▁moyen
+- 浪漫
+- ▁தலை
+- ▁fortress
+- ▁lebih
+- schild
+- nimi
+- coli
+- というような
+- ▁sprak
+- ▁fright
+- ▁benutzen
+- 吞
+- ▁Schwierigkeiten
+- ▁estão
+- ▁vide
+- ▁frederick
+- 延長
+- 的城市
+- ināt
+- 貌
+- ▁Hey
+- 雰囲気
+- ▁aggr
+- ▁wiederholt
+- າຍ
+- ▁musket
+- ▁swan
+- TO
+- ▁Gang
+- cala
+- ێک
+- look
+- ▁الْ
+- 社員
+- ▁wales
+- ▁problème
+- こちらも
+- zusehen
+- ▁پو
+- sichtig
+- ▁electronic
+- гээр
+- 把你的
+- ▁server
+- ▁exertion
+- 屏
+- ല
+- ▁ejemplo
+- ▁invariably
+- fugi
+- 外部
+- gruppe
+- mord
+- 為
+- ந்தி
+- ▁señora
+- двига
+- ▁exert
+- 諸
+- ▁كبير
+- ▁feedback
+- ▁Terra
+- 大切な
+- ющий
+- そこまで
+- чар
+- ▁Var
+- 食べた
+- ▁glancing
+- 薪
+- 还有什么
+- ▁compact
+- 审查
+- 原さん
+- ▁хоть
+- ▁أحد
+- mén
+- 絶対に
+- ▁merge
+- ってきて
+- ▁swallowed
+- ▁Wachstum
+- ▁Technik
+- مدينة
+- ▁fossil
+- ▁organiza
+- 把这些
+- 引っ
+- を見つけ
+- 阁
+- てほしい
+- нө
+- ▁impart
+- いますね
+- ▁numero
+- あふれ
+- 立場
+- kutu
+- ▁correspondence
+- 增加了
+- ▁ragged
+- А
+- ▁hindi
+- ▁dyn
+- ▁successive
+- がこちら
+- ▁verdammt
+- 严肃
+- ▁barren
+- ▁prairie
+- లు
+- ▁Ora
+- ▁понять
+- ▁vamos
+- ▁conventional
+- に参加
+- ▁vegetables
+- ▁стали
+- ▁تغییر
+- 希腊
+- ▁kelkaj
+- 算了
+- '48'
+- ацыі
+- ▁kitty
+- 踊
+- ▁Wahrscheinlich
+- meza
+- ▁swore
+- ▁cheerfully
+- zego
+- ▁storage
+- تك
+- ▁bezahlt
+- muz
+- 五月
+- jol
+- дей
+- oğu
+- icamente
+- ▁hinauf
+- син
+- ▁prova
+- ▁letzter
+- ▁temperament
+- ▁rencontre
+- ▁influenced
+- ラーメン
+- ▁некалькі
+- ▁parlour
+- ▁Mariya
+- fläche
+- ▁Meister
+- 提出了
+- を求め
+- 有意思
+- няў
+- ▁polar
+- ▁Verwendung
+- ▁стен
+- ەم
+- 的作品
+- ▁verstand
+- 所以如果你
+- ▁mieux
+- ▁요즘
+- ▁diversity
+- бак
+- 恐れ
+- 案例
+- 洲
+- ▁tamam
+- ▁salary
+- ▁возможность
+- 它将
+- kret
+- 氧
+- 龍
+- ▁industries
+- chá
+- ▁Vier
+- ▁swinging
+- ▁bacon
+- пись
+- ▁edith
+- 诶
+- ັດ
+- ▁знаком
+- фу
+- ою
+- ừng
+- ▁impress
+- 立って
+- ões
+- хад
+- 锐
+- ▁resemblance
+- 罢
+- 拯救
+- 我也是
+- ▁tightly
+- けがを
+- лева
+- 货币
+- ▁beautifully
+- యి
+- ▁Histori
+- ▁Harry
+- 訴え
+- 集団
+- даль
+- 短信
+- 可以通过
+- лаш
+- 苗
+- ▁summary
+- 프
+- ▁первых
+- 企画
+- oxy
+- っつ
+- ▁celebra
+- 准备好了
+- ▁обрат
+- 大胆
+- tuig
+- θ
+- ▁mabel
+- бла
+- метр
+- 50%
+- ▁fugitive
+- ▁babe
+- だよね
+- 名单
+- ▁haunted
+- 做到了
+- lerde
+- white
+- ▁technische
+- 意味で
+- 脂肪
+- ▁dissolve
+- zze
+- ▁일단
+- ▁loath
+- ▁intensity
+- 提案
+- 提交
+- ▁steckt
+- 耶稣
+- ▁orbit
+- ▁русски
+- ▁fiercely
+- 年後
+- 只是一个
+- Ex
+- ▁sakj
+- 查看
+- ▁آر
+- ▁Ost
+- ▁stump
+- ▁enthalten
+- ▁bitterness
+- 会合
+- 他又
+- ▁hesitate
+- 神奈川県
+- 键
+- كار
+- 鹰
+- ▁درباره
+- жиг
+- ▁cabeza
+- lada
+- ▁bidding
+- тыш
+- ▁cautiously
+- 主播
+- Gelächter
+- ▁erlaubt
+- ▁reconciliation
+- 崩溃
+- وارد
+- pion
+- ▁fapt
+- 協議
+- ▁gemeinsame
+- 키
+- ▁دارید
+- ▁Qual
+- 東日本
+- 三千
+- ▁میں
+- ▁venus
+- ▁undertaking
+- 所以我认为
+- 克拉
+- ▁câu
+- われる
+- гэр
+- aurait
+- ▁никак
+- かわい
+- 别说
+- hilfe
+- ▁writ
+- beug
+- ▁sombre
+- ptic
+- 如果有
+- NG
+- limi
+- 的事实
+- ▁cảm
+- rott
+- ▁район
+- ▁desolate
+- ▁picturesque
+- বো
+- 晕
+- ذهب
+- ることができる
+- ▁weinig
+- ▁siya
+- ▁stain
+- 惑
+- ▁pourquoi
+- ▁строительный
+- ▁doğru
+- ▁Institution
+- 送给
+- ▁Indien
+- 干啥
+- design
+- ゆっくり
+- 近い
+- ▁Sicher
+- ▁раньше
+- 용
+- ▁effektiv
+- NATO
+- ▁incentive
+- ▁assisted
+- 什么东西
+- ▁stirring
+- ▁corresponding
+- हि
+- 复制
+- 窝
+- ▁Füße
+- ▁tất
+- ències
+- ▁Ati
+- 钥匙
+- ▁mysteries
+- ▁bietet
+- ▁gegenseitig
+- ▁எழு
+- ▁tej
+- 庆祝
+- เล
+- ▁நகர
+- るもの
+- மீ
+- نز
+- sinde
+- サル
+- になってる
+- 上空
+- 言われた
+- ▁Julia
+- ۋا
+- ▁grundlegende
+- ▁comte
+- pida
+- ▁speedily
+- 我们希望
+- respect
+- 普通の
+- ▁thoughtfully
+- ▁Är
+- 宜
+- sorge
+- ▁betrifft
+- ▁dụng
+- ար
+- 你如何
+- 地址
+- ▁يكون
+- лета
+- 都说
+- 赏
+- 強調
+- ▁Sab
+- лос
+- tijd
+- ▁bezahlen
+- ▁Serie
+- イタリア
+- ▁flexible
+- ▁isolated
+- ▁angrily
+- ゲスト
+- 被认为是
+- 人工
+- tauchen
+- ▁точки
+- 張り
+- ▁Right
+- borg
+- ligi
+- about
+- 匠
+- さんから
+- ▁passive
+- ▁delayed
+- かす
+- ダン
+- জে
+- 想着
+- ▁vacant
+- 宴
+- ▁congregation
+- 分かり
+- خور
+- ▁juice
+- ستخدم
+- るべき
+- 子どもの
+- ▁poble
+- 三振
+- けれど
+- 緩和
+- ▁pigeon
+- сны
+- ▁большое
+- ▁Westen
+- әһе
+- сни
+- ▁ensuite
+- ▁agencies
+- ▁remorse
+- ▁смотрите
+- الات
+- 工事
+- 터
+- ▁отвеча
+- igkeiten
+- ▁Universitat
+- bwiye
+- という感じ
+- 통
+- ▁identific
+- 男性は
+- 冻
+- ▁quá
+- 日曜日
+- ▁সেই
+- ミン
+- ído
+- ▁parcel
+- ▁êtes
+- ▁ringing
+- 癌
+- ▁baixa
+- ▁благо
+- するか
+- 组合
+- ▁predomin
+- reisen
+- hnung
+- ▁கரு
+- 贪
+- ▁Häuser
+- ▁nowadays
+- ▁Victoria
+- ثلاث
+- imecres
+- ▁ошиб
+- 保健
+- ▁goose
+- ▁nieuwe
+- ▁consented
+- 匪
+- ▁geloof
+- ▁Würde
+- ▁pup
+- 集まって
+- ▁parade
+- これらの
+- 周围的
+- 美好的
+- bildung
+- ▁plank
+- ▁blink
+- ▁behalten
+- ▁сказали
+- ▁каких
+- ▁දෙ
+- 佑
+- 呂
+- ▁bestimmt
+- ▁بده
+- ţa
+- 但她
+- 寿
+- ▁ўз
+- iera
+- ▁shrewd
+- ▁рядом
+- меж
+- 毎年
+- ▁persistent
+- ▁addict
+- ▁mogelijk
+- ▁Christian
+- ▁disconnect
+- ▁dadurch
+- gisha
+- ▁politicians
+- ▁audio
+- 让人们
+- ▁Harvard
+- ▁Niemand
+- fí
+- ড়া
+- ▁digest
+- 同一个
+- 契約
+- ▁corona
+- kku
+- 広い
+- 以色列
+- ▁Kreuz
+- 陪伴
+- ゃ
+- ▁Gü
+- ▁uitge
+- ▁seasonal
+- ▁sanction
+- Th
+- යා
+- vär
+- 幻想
+- ▁Kate
+- 麦克
+- 榜
+- ▁exchanged
+- ▁verfolgt
+- メジャー
+- ▁jardin
+- ▁Kritik
+- ▁Islam
+- 心态
+- 侵
+- ▁console
+- ▁الان
+- ▁fulfil
+- 多数
+- fiel
+- ▁voraus
+- conscious
+- sohn
+- 부터
+- 時半
+- ▁canon
+- ▁zahlreiche
+- 盒子
+- মু
+- länge
+- pila
+- ▁линия
+- 偶然
+- ▁whatsoever
+- 转变
+- ▁eleventh
+- 暗示
+- 心理学
+- ▁albert
+- 表现出
+- 蜜
+- ▁pennsylvania
+- 새
+- らん
+- たくない
+- tev
+- плы
+- 扑
+- ▁taun
+- 手中
+- ▁onward
+- імі
+- 空間
+- ▁thời
+- ▁евро
+- ▁tuesday
+- ▁scotch
+- ▁смысле
+- しているということです
+- ▁redeem
+- čni
+- 評
+- лык
+- ▁Divendres
+- ประ
+- ▁tested
+- ▁passar
+- 偶尔
+- ▁Group
+- ▁знаем
+- 二十四
+- 想定
+- ▁Satz
+- 淘
+- 摧毁
+- ▁dapat
+- ▁helmet
+- ▁historic
+- 特定的
+- ▁youthful
+- ▁relieve
+- ▁chris
+- ▁argued
+- 十几
+- ▁eerst
+- ▁документ
+- ▁poate
+- мор
+- 退休
+- 你有没有
+- 帰り
+- ▁මෙ
+- হু
+- 鎖
+- 長く
+- ▁deliberate
+- казаць
+- 違
+- ▁shed
+- ographie
+- 舒
+- ▁jaun
+- ▁wanneer
+- ▁ببین
+- 燃烧
+- taking
+- ▁Schau
+- 发现自己
+- 出てきた
+- comm
+- وجه
+- ▁contro
+- ▁angenehm
+- চ্ছ
+- ▁awaiting
+- عَ
+- ▁watu
+- ▁overwhelming
+- 宠
+- ▁peuvent
+- ▁erlebt
+- が好き
+- 仕事を
+- 体系
+- vidi
+- ▁relating
+- ▁sweetness
+- ▁rhai
+- ▁круг
+- カリ
+- ▁هستم
+- ▁disput
+- 発達
+- schnitt
+- ▁Bü
+- ▁recognise
+- ▁tracta
+- 只是为了
+- 通過
+- ではなくて
+- 始まって
+- ▁discharge
+- 不出来
+- ▁inventor
+- ▁bog
+- ▁français
+- 贤
+- 锻炼
+- びっくり
+- аваны
+- 公正
+- ゅ
+- ▁страна
+- 安装
+- ▁kwenye
+- ніка
+- urage
+- 在网上
+- ▁drown
+- 年ぶりの
+- terri
+- zellen
+- ת
+- セル
+- いかがですか
+- 授業
+- ▁morgan
+- ▁miller
+- ほうが
+- ▁neuer
+- 论文
+- ▁wunder
+- ▁Fach
+- ▁skies
+- 标题
+- ▁Sitz
+- zional
+- خواهی
+- 見てみましょう
+- ▁transformed
+- 没办法
+- ▁melted
+- ▁encontrar
+- ▁genre
+- 擅长
+- ▁Middle
+- 说实话
+- ática
+- ności
+- グル
+- ▁cultivated
+- ▁pioneer
+- 基準
+- 生态
+- ▁இல்ல
+- ▁neden
+- スープ
+- ▁Questo
+- ▁pry
+- 食感
+- ▁Wall
+- ▁regulation
+- 锋
+- 我认为这是
+- 近くに
+- gemeen
+- ▁hospitality
+- ▁erected
+- ▁Have
+- ▁interaction
+- greif
+- ಂ
+- ▁tissue
+- ▁таго
+- ▁элек
+- 不要再
+- ▁Brun
+- ▁gill
+- っちゃって
+- what
+- 変え
+- ▁adapted
+- 铃
+- マンション
+- جار
+- ▁frighten
+- ▁Gw
+- ▁Top
+- ▁juist
+- ▁Jeff
+- ▁wrth
+- ▁frisch
+- 好听
+- ▁penalty
+- ▁څه
+- ▁marion
+- ▁Front
+- ▁Arch
+- ▁dicho
+- ော
+- ▁optimistic
+- ▁Zusammenhang
+- әл
+- 铁路
+- ▁accounting
+- ▁vocal
+- оби
+- ▁representing
+- ▁cancel
+- ĕ
+- 转移
+- రి
+- 七年
+- ▁kubera
+- 饮食
+- ▁unfold
+- 无聊
+- ▁Stephen
+- ▁bestow
+- ▁регион
+- ▁terrific
+- 监管
+- 場所に
+- وک
+- ▁matin
+- ▁conveyed
+- ▁склада
+- ペース
+- ▁adhere
+- ついに
+- And
+- ▁underground
+- tuvi
+- ▁apenas
+- ▁allowance
+- ▁противо
+- Rires
+- ▁нашего
+- ▁نیاز
+- ▁begon
+- ▁harold
+- ことから
+- 定期
+- 返し
+- ベース
+- ▁кост
+- 测量
+- ▁deceive
+- ダウン
+- 部署
+- 应对
+- 找个
+- 命名
+- ▁bucket
+- ▁traitor
+- ඇ
+- ʻ
+- ވަ
+- gá
+- ▁அமை
+- ▁Zweifel
+- овым
+- 殺人
+- 落下
+- ▁Tell
+- ▁стало
+- 跪
+- ▁Lieblings
+- ▁earthquake
+- రు
+- рик
+- 処
+- ▁pavement
+- 我真的很
+- 離れ
+- ▁mora
+- 開いて
+- 十个
+- ww
+- indre
+- eerde
+- 次に
+- 否定
+- ▁violin
+- ▁numa
+- ドリ
+- 加州
+- ましたけど
+- ▁моему
+- vēr
+- ▁enfin
+- いける
+- ▁plunder
+- grado
+- ジャン
+- ▁podía
+- били
+- ▁mourn
+- ▁circula
+- ▁screamed
+- ▁eternity
+- ▁privi
+- ほう
+- ▁되는
+- 正面
+- ▁üle
+- ombra
+- ▁வரு
+- pane
+- åg
+- ▁entlang
+- 团长
+- 驻
+- マル
+- 残る
+- ných
+- 思いを
+- ▁Jackson
+- ҭ
+- も多い
+- ▁එක
+- を行い
+- achtig
+- 的证据
+- ▁geschehen
+- 慮
+- ▁gentil
+- ▁найти
+- わし
+- 分裂
+- impi
+- 悦
+- ▁định
+- 弁護士
+- 陽性
+- zubringen
+- 早点
+- ▁autonom
+- 手術
+- 種目
+- 我说的
+- ޮ
+- 审判
+- 年纪
+- ▁其实我
+- ▁gris
+- ▁показал
+- ▁tourist
+- быт
+- 机场
+- ▁interrupt
+- 守り
+- 筋肉
+- ▁manière
+- ユニ
+- 作戦
+- ▁proced
+- 響
+- ▁morality
+- ▁Wunder
+- igeze
+- ▁awfully
+- ▁gesti
+- ▁Allerdings
+- ▁학교
+- ▁dwarf
+- 上市
+- ▁futbol
+- кен
+- ▁construir
+- 疯
+- 大雪
+- ▁physics
+- ▁beteiligt
+- オーストラリア
+- ▁hearth
+- эння
+- ▁cavern
+- бат
+- евский
+- 路线
+- іі
+- ▁Donc
+- ና
+- ▁appreciation
+- ▁trousers
+- 基础设施
+- ▁gewissen
+- 城镇
+- 你看到
+- ▁организаци
+- 悬
+- 抹
+- ▁hinzufügen
+- ▁brillant
+- ▁Donald
+- ご飯
+- ▁dominant
+- рах
+- 伙计
+- meɣ
+- ▁benim
+- 鉴
+- выше
+- 郁
+- 捷
+- クリスマス
+- ▁ມັນ
+- ▁اسم
+- その中で
+- 看不见
+- ывают
+- らせて
+- 饮
+- йын
+- かぶ
+- 掌声
+- ори
+- ▁puzzle
+- ្រ
+- 队伍
+- ▁carved
+- ది
+- ▁admission
+- слуш
+- 鼻子
+- ouvri
+- pta
+- ▁yani
+- ▁extending
+- ப்படுகிறது
+- தில்
+- 官方
+- шчы
+- ▁пойм
+- 皆さんに
+- रे
+- ▁minim
+- ▁forgetting
+- rseits
+- طي
+- 日々
+- unterschiedliche
+- teko
+- bewegung
+- ▁coffin
+- ▁Information
+- 挥
+- いきたいと思います
+- ▁arribar
+- kubernetes
+- ▁nachzudenken
+- '55'
+- 狙い
+- '3000'
+- ▁ultra
+- лым
+- 業者
+- ▁تحت
+- ▁menace
+- ▁Instrument
+- いかない
+- ▁Boy
+- گا
+- ▁Gehen
+- ەیە
+- ▁capability
+- 양
+- dział
+- වේ
+- 经典
+- の仕事
+- abri
+- 都不知道
+- 的家人
+- луг
+- 游泳
+- ▁owl
+- 仍然是
+- ▁attachment
+- 今から
+- ▁Apo
+- ▁наступ
+- Oh
+- 所以当
+- ▁тихо
+- ▁casting
+- 然后在
+- 常に
+- ▁laatste
+- weza
+- тир
+- マス
+- ▁bộ
+- ▁luther
+- ▁عام
+- ▁скажем
+- ▁меньше
+- 溶
+- ▁делает
+- ▁Hinsicht
+- ▁ambient
+- ▁бя
+- 一号
+- illion
+- ▁Zweiten
+- ajja
+- ▁подобн
+- ▁sylvia
+- 縁
+- ごめん
+- ▁trở
+- ▁crude
+- すみません
+- fanya
+- 灾难
+- นี้
+- ▁dritten
+- ▁transferred
+- ▁freight
+- ▁chemin
+- ▁milieu
+- 他认为
+- を発表
+- ▁cruise
+- ▁deepest
+- нап
+- леш
+- ▁Ecke
+- ▁Kreis
+- 要不要
+- öß
+- ▁Minute
+- ▁процент
+- 广州
+- ▁sublime
+- ▁получается
+- ▁Brad
+- 然后是
+- ▁allgemein
+- 农村
+- ▁болно
+- ▁Eigenschaft
+- 比較
+- ▁grandi
+- drängt
+- ۋ
+- ▁அதிக
+- ▁Cri
+- 場所で
+- سك
+- ▁çek
+- щий
+- 纽
+- 冯
+- ▁живот
+- ▁مې
+- 便利
+- krimin
+- 当然是
+- ▁kicked
+- MS
+- 友好
+- ▁Krankenhaus
+- ▁Trotz
+- じゃないか
+- 成果
+- 剩下的
+- ▁Vision
+- ▁verdade
+- ▁pê
+- ▁Түүний
+- ▁پیدا
+- '='
+- ▁пусть
+- ▁Privat
+- マネ
+- ▁Quar
+- 物体
+- ▁ນີ້
+- ませんか
+- 单词
+- 触れ
+- 山上
+- ママ
+- koloni
+- bogen
+- ▁đổi
+- ▁Tanzania
+- 在我看来
+- 太陽
+- нец
+- 生长
+- ▁Dev
+- ▁scrutin
+- نَ
+- ▁девятьсот
+- llah
+- の裏
+- trum
+- に加え
+- ▁fulfill
+- ▁regardless
+- ▁surpass
+- ▁வந்த
+- ▁seixanta
+- ▁inflation
+- ▁augenblick
+- орд
+- ▁infinitely
+- ▁ufite
+- 时光
+- ▁Lauf
+- 精力
+- ēr
+- が確認され
+- மெ
+- හා
+- ▁பேச
+- schmerz
+- unterricht
+- reɣ
+- 创造了
+- ▁veure
+- ▁sechzig
+- steller
+- ▁retour
+- ▁untersucht
+- に入った
+- 方がいい
+- ▁constitutional
+- 号码
+- ▁educational
+- มัน
+- ▁Those
+- ▁Quando
+- ▁herbert
+- 以便
+- ▁années
+- ▁acres
+- 帆
+- ▁maître
+- ▁langer
+- 罗斯
+- ▁syria
+- office
+- ▁nebo
+- ▁контрол
+- ▁everyday
+- ▁ähnliche
+- ▁diferents
+- නවා
+- 密码
+- ▁Blue
+- ▁broadcast
+- žu
+- дают
+- أن
+- pik
+- 月亮
+- stimmung
+- ▁Krä
+- ▁satellite
+- ▁weltweit
+- ▁figli
+- ▁tailor
+- くれる
+- 腐
+- ሽ
+- の写真
+- ▁كه
+- 変更
+- ▁negroes
+- ▁همین
+- ▁Acht
+- strom
+- مەن
+- 勾
+- 妨
+- 一夜
+- 働いて
+- ▁lachte
+- ▁persisted
+- perto
+- larga
+- படி
+- ▁Wirkung
+- ▁gezicht
+- ▁فرا
+- geladen
+- овые
+- vana
+- ▁bemerkte
+- ▁beliebt
+- ▁гаражно
+- 싸
+- ないといけない
+- ▁provoca
+- ▁Ausdruck
+- 两者
+- ▁canvi
+- ▁complexion
+- гран
+- ▁vedere
+- DA
+- ӡ
+- häuser
+- とかも
+- 留下来
+- ▁چون
+- Atlanti
+- schwer
+- 姿を
+- 死去
+- ▁bewildered
+- ▁refrain
+- ▁verfügbar
+- ▁dije
+- ▁praying
+- ▁fuera
+- 代价
+- ▁Selle
+- হে
+- ▁hyd
+- 全身
+- يين
+- ▁немного
+- ▁أخرى
+- ▁Qualität
+- 护士
+- 年齢
+- ▁tension
+- Prince
+- ième
+- шет
+- 谦
+- ẩ
+- ucu
+- ▁ними
+- ছা
+- ▁Ky
+- 知道他
+- 蔡
+- 拦
+- instagram
+- ovitch
+- ▁orphan
+- телей
+- ▁hablar
+- ▁subsist
+- ▁grape
+- アナ
+- давать
+- фан
+- ▁välja
+- を使い
+- 汽
+- 另一种
+- нікі
+- 排除
+- ▁scientist
+- ভি
+- лыг
+- ▁tribute
+- таш
+- ▁lind
+- 低気圧
+- ▁ohio
+- ▁universit
+- 简直
+- zaj
+- ▁Letzte
+- țe
+- abaye
+- ▁nke
+- 週末
+- 仏
+- 主持人
+- ▁먹고
+- ▁recherch
+- ▁wohin
+- ▁pulse
+- ▁lazy
+- ▁impatiently
+- ▁invalid
+- ىدى
+- を巡って
+- 富有
+- ▁identical
+- 霸
+- 刊
+- responsabil
+- 違反
+- ▁futuro
+- ▁داریم
+- ителей
+- ▁Zeitung
+- سون
+- 置き
+- ▁pete
+- 基督教
+- щения
+- 時代の
+- 上げた
+- ▁военкомат
+- 赤い
+- 安全保障
+- ▁descri
+- ▁Cle
+- ▁Kapitän
+- ▁regelmäßig
+- ▁verfolgen
+- 脅
+- 高騰
+- ▁Einrichtung
+- itarian
+- 大规模
+- 是个好
+- âr
+- ▁Brust
+- ▁herausfinden
+- ▁privacy
+- ▁важны
+- ▁Out
+- mitglied
+- ▁dedicat
+- 廃
+- ▁signature
+- ▁பகுதி
+- ▁arriva
+- ▁quiero
+- この時間
+- abba
+- ▁Nehmen
+- ▁enlarge
+- ாங்க
+- ▁Lady
+- 趁
+- ▁greece
+- ▁surgery
+- ▁dunkel
+- dokument
+- ▁quiere
+- ってしまう
+- 雨が降り
+- ▁hause
+- ▁discretion
+- 的风险
+- ほんとに
+- ▁Universum
+- ▁Foo
+- kê
+- ▁Pala
+- ▁exclusive
+- крат
+- ▁assumption
+- ▁sev
+- を持ち
+- ▁exhibited
+- ▁такую
+- ▁hỏi
+- ▁produziert
+- ▁jersey
+- ▁Schönheit
+- 额外的
+- ▁Vorstand
+- ▁sotto
+- ▁eagerness
+- funa
+- ▁unseen
+- ▁jumping
+- ▁legitimate
+- ▁physically
+- ▁hafi
+- ▁poetic
+- iyordu
+- 唱的歌
+- 崇
+- ▁rubbed
+- ▁humility
+- 启动
+- ▁accessible
+- ▁hoarse
+- ▁nơi
+- ▁bois
+- үй
+- حب
+- cari
+- ▁Fleisch
+- ▁Для
+- singa
+- 犹豫
+- ▁Bedingungen
+- ▁englische
+- ▁sneer
+- を挙げ
+- ▁sentimental
+- 运营
+- ▁jaroj
+- 覧
+- 我一直在
+- 潘
+- ▁ingredient
+- 真ん中
+- ▁Alexander
+- ▁brigade
+- ▁oxygen
+- 相同
+- 过度
+- 校园
+- 何だ
+- ключа
+- شە
+- تُ
+- ▁Rang
+- 远离
+- 相次いで
+- ▁byinshi
+- ▁tätig
+- umugore
+- ▁dictate
+- ▁nhau
+- ția
+- 船长
+- ▁năng
+- 侠
+- 誤
+- となっている
+- 但是如果你
+- ▁hyper
+- 该怎么办
+- 大学生
+- ればいい
+- 这很
+- тоў
+- ság
+- ▁fifteenth
+- ▁đâu
+- 길
+- ▁известно
+- ▁glacier
+- einig
+- сво
+- 那一刻
+- に乗って
+- ẻ
+- 罐
+- ▁arriving
+- ▁který
+- ▁закры
+- べる
+- 革
+- ▁retirement
+- ▁marching
+- ှ
+- ▁annoyed
+- ▁üblich
+- yerek
+- говарива
+- kampf
+- දා
+- 提议
+- ▁Western
+- なぁ
+- ▁kubona
+- ▁کجا
+- gezi
+- 疏
+- 进展
+- ▁города
+- ▁hurriedly
+- ▁kiuj
+- 年ぶりに
+- ▁appearing
+- NA
+- ▁elfu
+- 鸡蛋
+- 涛
+- ▁bernard
+- ▁manual
+- бед
+- ▁banc
+- 美しい
+- schneiden
+- لعب
+- ▁Maßnahmen
+- ▁superstition
+- speicher
+- ▁embark
+- 観客
+- koko
+- ▁manipula
+- お話を
+- ▁joyful
+- হি
+- ▁incessant
+- ▁courtesy
+- ҳа
+- ▁fain
+- imbi
+- '['
+- 为他们
+- ▁determina
+- 本部
+- 民間
+- もらいたい
+- 裸
+- أت
+- 做了什么
+- лыш
+- kiye
+- bleib
+- ▁şeyler
+- 売れ
+- ▁Germany
+- 我很高兴
+- ありますよね
+- 呵
+- ▁вокруг
+- 专注于
+- ▁sequence
+- 予算
+- ▁indirect
+- ▁bilden
+- lässig
+- 合适
+- ຽ
+- ჩ
+- You
+- 所属
+- ▁tries
+- ▁speeches
+- カット
+- ▁trivial
+- ▁posterior
+- 見通し
+- ▁rhan
+- оруж
+- 去过
+- ҩ
+- ▁monstrous
+- 遊び
+- ▁hàng
+- ▁союз
+- 都不会
+- 見られます
+- ▁ernsthaft
+- ▁Tatsächlich
+- ▁мозг
+- しん
+- beho
+- horse
+- ▁breach
+- ファイ
+- ▁když
+- ▁muscle
+- ▁comparable
+- ▁grasped
+- 誕生
+- ▁Nta
+- ▁zyn
+- ▁Они
+- 有几个
+- 碑
+- Ю
+- ▁Firefox
+- ▁carolina
+- 控え
+- ່າ
+- 布莱
+- gruppen
+- 澤
+- ▁круп
+- 働く
+- نق
+- 볼
+- ▁yake
+- ▁tiek
+- 冠军
+- পে
+- baho
+- ▁Streit
+- 晋
+- ▁Brü
+- ▁направлен
+- 维尔
+- fär
+- presa
+- についても
+- ▁tõ
+- ▁administrative
+- ▁communicated
+- ubah
+- реп
+- க்கிறது
+- 老大
+- 選ばれ
+- ▁благодар
+- ▁desperately
+- corre
+- šķ
+- ▁sweeping
+- 怎么可能
+- ggie
+- ▁recur
+- ▁உட
+- ▁floated
+- 推出
+- прэ
+- 在上面
+- 織
+- ◆
+- ▁twelfth
+- 腺
+- 女孩子
+- ▁counting
+- ▁hurrying
+- 如果没有
+- ▁Abschluss
+- ▁stealing
+- ▁Deck
+- 稲
+- ▁Bevor
+- कि
+- みてください
+- ▁کمی
+- ▁Verkehr
+- ▁Martí
+- ▁assess
+- 逮捕された
+- ▁حرف
+- gekomen
+- 天才
+- 类似的
+- ▁Online
+- ď
+- ▁እንደ
+- 订阅
+- 渋谷
+- ▁už
+- サイズ
+- 支配
+- чку
+- ▁advertise
+- ▁wanjye
+- urile
+- 丧
+- ▁aussieht
+- ▁общественн
+- ▁Like
+- ▁kirk
+- ▁tennis
+- ▁журналист
+- ▁сторону
+- ▁دنیا
+- 온
+- လ
+- ▁главное
+- ▁fiz
+- funktion
+- lå
+- ▁fotografi
+- шат
+- 対策を
+- かれて
+- 缠
+- ወ
+- ▁вполне
+- ▁Flugzeug
+- ▁trường
+- grond
+- ▁echoed
+- ▁phát
+- だいぶ
+- ▁Sara
+- ▁conversa
+- ▁stimme
+- ▁Quina
+- 指数
+- 祖父
+- wahl
+- ▁klug
+- righteous
+- рап
+- ▁mujeres
+- 的痛苦
+- 六年
+- 全都
+- ▁clasp
+- 浏览
+- топ
+- ▁horizontal
+- 坚定
+- ▁wengi
+- ▁bored
+- 局长
+- ▁travaille
+- த்திற்கு
+- fluss
+- 岗
+- 現状
+- ▁但是当
+- ▁Operation
+- فور
+- ▁junto
+- 你到底
+- ▁convenience
+- 抑郁
+- ▁survived
+- ▁образова
+- 公寓
+- लि
+- ▁mesure
+- ització
+- パラ
+- ▁mourning
+- ▁dart
+- ▁Grün
+- ސ
+- ਿ
+- ▁wednesday
+- 外出
+- ▁hasty
+- 完整的
+- ▁compre
+- வரை
+- 议会
+- ఏ
+- 會
+- ▁farklı
+- kracht
+- 你可以在
+- カップ
+- 无数
+- ▁Gute
+- гийг
+- ▁llarg
+- ▁elevated
+- ▁elsie
+- ▁Kommunikation
+- あたり
+- impine
+- 特殊的
+- ழி
+- тере
+- ▁そのため
+- нием
+- ▁armor
+- ▁Weiter
+- 在中国
+- ško
+- É
+- යට
+- ▁thyself
+- 沸
+- ▁achtzehn
+- ▁extinct
+- 太多了
+- ▁dresses
+- ▁Prime
+- ▁carrera
+- ுகிறது
+- 不认识
+- ▁expenditure
+- ▁innumerable
+- 気象庁
+- 公安
+- grown
+- 不开
+- ▁infantry
+- ▁abide
+- に合わせて
+- '39'
+- 疲れ
+- ▁schuld
+- スキー
+- ▁jog
+- وضع
+- ▁européen
+- ▁seria
+- 合格
+- いら
+- ာ
+- 伐
+- ▁உம்ம்
+- రా
+- 无关
+- ▁diffus
+- দিন
+- 的语言
+- ▁Sarah
+- 书中
+- storm
+- 受害者
+- 仕組み
+- ▁banquet
+- 早餐
+- кос
+- 亿元
+- 10%
+- 供給
+- アイデア
+- технолог
+- ▁diwujudke
+- ่ง
+- ▁energetic
+- ▁bunny
+- 从来不
+- burgh
+- ிருக்கிற
+- ▁glittering
+- 嘘
+- 8%
+- 留言
+- 二十五
+- CD
+- さえ
+- ▁geändert
+- ▁увидел
+- ▁roedd
+- 你这么
+- BA
+- ▁spray
+- ことができます
+- 有能力
+- alina
+- ▁disposal
+- ▁système
+- ▁dreary
+- ▁quarante
+- ▁بأن
+- ▁vient
+- 第三个
+- ▁vacation
+- ▁Ohne
+- 币
+- 倡
+- ▁unnatural
+- 결
+- 少なくとも
+- ▁Another
+- みんなが
+- ▁fuss
+- ▁خواهد
+- ▁patro
+- ▁personage
+- ▁жыв
+- ▁ddim
+- альным
+- ▁scheinen
+- ▁probabil
+- ▁Bewusstsein
+- ▁Militär
+- 裙
+- ▁কথা
+- ▁Sagte
+- ▁whe
+- pola
+- ▁missus
+- 我觉得我
+- ▁quaint
+- viel
+- ▁telescope
+- 涵
+- ▁тупик
+- 在某种程度上
+- 言うと
+- 传递
+- 的灵魂
+- दा
+- ▁töten
+- 事务
+- ▁demonstrated
+- 教训
+- maro
+- ▁figura
+- しろ
+- 痕
+- ▁предлага
+- ▁metrics
+- ண்டி
+- ▁Iyo
+- を前に
+- 4%
+- ▁marian
+- hagen
+- ▁ради
+- цеп
+- ▁слушай
+- ▁демократ
+- ▁ситуация
+- ▁Drogen
+- ▁dose
+- افت
+- ▁اول
+- ▁suppress
+- bril
+- ▁beschlossen
+- 涂
+- 謎
+- ಪ
+- badde
+- ▁있는데
+- ▁drunken
+- ▁seule
+- ▁afge
+- 囲
+- 爸妈
+- 棚
+- 峡
+- ▁bacteria
+- ▁erscheinen
+- が続き
+- ț
+- قار
+- хоз
+- 紧急
+- 好朋友
+- рада
+- ▁examining
+- 同性恋
+- 韩国
+- 很高
+- ▁спец
+- ヘル
+- ере
+- 一歩
+- batu
+- ▁automobile
+- 執
+- を含む
+- ▁gerecht
+- ▁welcomed
+- ▁agricultural
+- っていうのを
+- ▁tudo
+- 相信我
+- 喜欢你
+- ▁woorden
+- 彻
+- তু
+- ارات
+- 这都是
+- 模糊
+- 逆転
+- ダブル
+- ▁мае
+- ショー
+- ▁dirigi
+- 检测
+- 哪怕
+- 我父亲
+- ▁eric
+- із
+- ▁parc
+- dodd
+- 饰
+- រ
+- ▁December
+- 겠지
+- ▁yali
+- ▁gordon
+- ▁Stamm
+- बा
+- ▁ikintu
+- مین
+- ச்சு
+- 美好
+- ▁interrog
+- Aplausos
+- EM
+- 警報
+- ▁Treffen
+- ▁Tau
+- ▁rive
+- を食べ
+- ▁sử
+- ▁останови
+- ▁oliver
+- onian
+- ▁Ez
+- stuk
+- 逛
+- ▁внутри
+- ▁decisive
+- ▁unglücklich
+- の様子
+- 広がる
+- 这个项目
+- ▁Klima
+- ▁primero
+- jang
+- ▁Betrieb
+- ▁sealed
+- ▁insane
+- ۔
+- ▁gewoon
+- 为您
+- ▁tenderly
+- ▁самый
+- ▁oppose
+- ▁ответил
+- eremoni
+- саб
+- ▁correctly
+- FIL
+- าน
+- ▁어디
+- ▁realidad
+- behörde
+- gerufen
+- уха
+- 房子里
+- ▁sujet
+- ▁classroom
+- ▁hübsch
+- ▁sekolah
+- ▁گروه
+- ▁Gli
+- 足以
+- ▁Carolina
+- yorum
+- ▁epidemi
+- áz
+- liwa
+- ボン
+- ▁feminine
+- ▁które
+- ▁primo
+- ▁verdi
+- кот
+- ▁graham
+- mele
+- 你们两个
+- िक
+- 感受到
+- 无人
+- 될
+- ▁knelt
+- ▁tìm
+- ▁hebrew
+- ▁Meter
+- กล
+- ▁clothed
+- 使える
+- 陽
+- amazi
+- Kanal
+- ▁бага
+- 心灵
+- 逸
+- ▁klingt
+- ▁видео
+- лоб
+- 撤退
+- 景色
+- nect
+- ▁vorsichtig
+- ಗ
+- ▁bachelor
+- ▁хочешь
+- ▁شرکت
+- 見ていきましょう
+- もあって
+- 6%
+- ▁Mah
+- ▁Су
+- லோ
+- வேண்டும்
+- 浓
+- ▁carpenter
+- ▁shrub
+- ▁nourish
+- 始める
+- ようになって
+- しまして
+- ▁awareness
+- 闯
+- ▁невозможно
+- ▁나오
+- boro
+- ▁strife
+- ▁dismiss
+- 综合
+- িয়ে
+- witz
+- حدث
+- pür
+- 图片
+- 놀
+- 鍵
+- ▁subdued
+- spoonful
+- ▁Finally
+- 看不到
+- 运动员
+- کات
+- ▁rector
+- 的角度
+- ▁faculties
+- 麺
+- 円安
+- 緊急事態宣言
+- ▁molecule
+- ▁plough
+- ▁вчера
+- ▁tiến
+- Pierre
+- مند
+- 市では
+- ▁Ми
+- bika
+- ▁obstinate
+- ▁saviour
+- 庞
+- pflicht
+- 大海
+- ▁habitat
+- 这是我的
+- urira
+- 識
+- ですとか
+- hani
+- ▁candid
+- 屁
+- ▁ممکن
+- ▁এবং
+- ▁ລະ
+- tsinda
+- شته
+- 农业
+- 不同意
+- ▁Februar
+- లో
+- ிலிருந்து
+- 太子
+- kleri
+- ▁яму
+- ▁draught
+- 前提
+- ▁envia
+- ▁também
+- ▁crust
+- ▁Калі
+- ▁Verfahren
+- ▁manifestation
+- ▁verbringen
+- ▁честно
+- 做饭
+- 今天晚上
+- ▁видите
+- kò
+- ▁marshal
+- ▁Bäume
+- ▁astonishing
+- ▁conjecture
+- ▁embarrassment
+- 畅
+- 眼泪
+- koli
+- альные
+- ▁concentrated
+- 岩石
+- ười
+- ▁amiable
+- ▁hiyo
+- ধ্য
+- 自行车
+- ▁societies
+- 舰
+- ▁አዎ
+- ▁realise
+- ▁pouvoir
+- mico
+- に移
+- 就行了
+- 合适的
+- ▁setanta
+- fenster
+- ▁dismal
+- ▁Wunsch
+- ▁wolves
+- lil
+- 建設
+- ▁rwose
+- annu
+- ▁aircraft
+- イチ
+- 拥抱
+- 電車
+- ▁sixteenth
+- ▁Kurz
+- Episode
+- ▁فیلم
+- ību
+- 並んで
+- 三点
+- cò
+- ▁yella
+- знача
+- ▁حا
+- この辺り
+- альных
+- いくと
+- ▁Album
+- ▁radiant
+- ▁nossa
+- 层面
+- хим
+- ▁diversaj
+- ▁imitate
+- 谋杀
+- ▁resentment
+- ▁shark
+- 总结
+- 進化
+- 运作
+- ▁assertion
+- ູ
+- ▁Praxis
+- にあります
+- ▁confidential
+- uğu
+- wego
+- ▁Ро
+- ▁embraced
+- nutri
+- 与其
+- ▁buzz
+- ▁camino
+- يط
+- 视觉
+- achtet
+- 良い
+- ▁скорее
+- 珀
+- っていうふうに
+- 继承
+- 你刚才
+- ▁صورت
+- ▁países
+- имся
+- 是一样的
+- тэн
+- ▁гэтай
+- ▁находится
+- ▁voilà
+- ▁startling
+- 我想要
+- 维持
+- ▁tray
+- රු
+- ▁повтор
+- 的主题
+- contin
+- 的人来说
+- ▁demanda
+- 旧統一教会
+- ▁inspection
+- 先ほどの
+- ▁Senat
+- ▁breadth
+- sehn
+- ального
+- 明明
+- ▁respira
+- جنوب
+- 落在
+- 耻
+- 試験
+- ▁explaining
+- 确
+- ▁perdu
+- ▁gorgeous
+- 陌生人
+- ▁Absicht
+- ولد
+- 遇见
+- шан
+- 高級
+- ▁parlor
+- хлоп
+- ษ
+- ▁Dijous
+- ▁schlafen
+- 改造
+- ▁miriam
+- ▁Té
+- ▁shipping
+- そこは
+- äck
+- ステージ
+- 十分钟
+- 太郎
+- ▁pasture
+- ワード
+- ▁nicely
+- 増えて
+- ▁beobachten
+- 网上
+- ▁fröhlich
+- ▁އަ
+- தொடர்
+- モー
+- льны
+- ▁стоял
+- дзень
+- ▁газет
+- 名为
+- を認め
+- 但这是
+- 决心
+- origine
+- 느
+- Gelach
+- foje
+- ▁charter
+- 这些问题
+- ▁vinden
+- ▁Nieder
+- yumba
+- பர
+- が出た
+- ▁everlasting
+- ▁trouver
+- zí
+- 成年人
+- ▁keer
+- ▁Anwalt
+- kaka
+- ▁nosso
+- ▁asserted
+- овали
+- AD
+- 接到
+- କ
+- ĥ
+- 疫苗
+- 这一切都
+- 理性
+- ▁polished
+- ▁flavor
+- смотрел
+- 産業
+- 你爸
+- ▁பய
+- 逗
+- ▁Unión
+- ▁rumor
+- そうですか
+- 我就不
+- ▁attributed
+- ツー
+- kları
+- ▁passenger
+- ▁bliss
+- 喽
+- ሁ
+- ▁verheiratet
+- ▁mindestens
+- ▁پنج
+- ▁highness
+- verfahren
+- تَ
+- 利润
+- ▁funkci
+- ▁readiness
+- ▁escala
+- ▁канал
+- ▁система
+- ராக
+- 抛
+- ▁Demokratie
+- いただきたいと思います
+- arbre
+- 在此
+- ▁vind
+- 我们有一个
+- ▁واحد
+- ▁gosh
+- 拨
+- ▁depressed
+- ṛṛ
+- ▁unlock
+- の前で
+- 俺が
+- ▁거의
+- 小麦
+- ▁whereby
+- ▁tint
+- mizi
+- ▁stretching
+- 趋势
+- 名古屋
+- ภ
+- ấu
+- ▁putea
+- 试验
+- ▁clergyman
+- 劣
+- équipe
+- খে
+- 決まり
+- ший
+- ▁motiva
+- ▁Atem
+- ▁ئەم
+- 遮
+- 천
+- 煙
+- ▁Bildschirm
+- パパ
+- 回到了
+- ▁await
+- ▁voted
+- 构成
+- 滅
+- ▁Hari
+- ▁righteousness
+- ▁женщина
+- 白天
+- ▁gerald
+- 旋转
+- 我想说
+- ▁surf
+- 酱
+- ▁Tief
+- দ্
+- ▁உள்ளது
+- 尸
+- ứng
+- boek
+- brook
+- ▁Erwachsene
+- 激励
+- щик
+- ▁acceptable
+- ▁món
+- ▁enthusiastic
+- ▁upside
+- ่ว
+- umuryango
+- 収穫
+- 欺负
+- ▁engaging
+- ▁Reform
+- ▁chop
+- үнд
+- 的需求
+- 很简单
+- loch
+- ▁EBITDA
+- ނު
+- 龄
+- ▁Spring
+- ▁logo
+- ▁engel
+- ▁fry
+- 私たちが
+- ▁stammt
+- 焦点
+- ▁Mexico
+- ▁kerja
+- 都内の
+- For
+- 大厅
+- ▁noranta
+- einheit
+- ちゃんは
+- 我们开始
+- ▁Option
+- ▁사람이
+- 足够的
+- んだね
+- SA
+- ई
+- ▁gewöhnlich
+- 炒め
+- ▁squeeze
+- 技巧
+- ▁tyrant
+- ▁کمک
+- ▁animated
+- ▁coverage
+- 獲得
+- ▁кап
+- ▁Schrift
+- ауы
+- 見えない
+- ▁그치
+- カフェ
+- が流れ
+- 亲密
+- ▁purity
+- 等一下
+- 就好了
+- ▁foremost
+- ▁niedrige
+- కు
+- halen
+- ட்டை
+- zünd
+- ▁olur
+- ▁eliminate
+- pez
+- ▁xem
+- 重视
+- ▁konzentriert
+- ▁extinguish
+- ▁cursed
+- ▁louder
+- ▁disdain
+- strich
+- の中から
+- フライ
+- Krista
+- ▁الأول
+- ▁plantation
+- ▁eindeutig
+- ▁einzigartig
+- 弦
+- 訪れた
+- рю
+- 盒
+- ▁hauptsächlich
+- ▁Тэрээр
+- 保障
+- ▁সং
+- ▁Norden
+- ▁stamped
+- ехать
+- verkehr
+- wszy
+- ▁pleaded
+- 忌
+- 熱中症
+- empi
+- арга
+- 一分钟
+- ▁kadın
+- адзе
+- 我妈妈
+- živ
+- 袭
+- ▁меч
+- ▁квартал
+- ▁Garten
+- ▁frantic
+- ▁Hij
+- ▁iraq
+- がついて
+- ▁ئې
+- 凌
+- ▁squirrel
+- ▁denomina
+- ▁imposed
+- цэн
+- 他们已经
+- ▁Theater
+- ▁yiwen
+- ▁Pap
+- 央
+- ▁tüm
+- ▁depended
+- 亦
+- ▁gerçek
+- ▁berühmte
+- hize
+- ▁Veränderungen
+- ▁unworthy
+- feind
+- дох
+- Instagram
+- ▁Benutzer
+- ▁dernière
+- ▁Biz
+- ▁زیر
+- ießen
+- ▁attempting
+- ▁процентов
+- が見つか
+- 接受了
+- ▁связи
+- ▁слышал
+- 你有什么
+- யூ
+- ცი
+- نے
+- ▁nearby
+- 踩
+- ▁nhw
+- ▁درست
+- شعر
+- ▁waiter
+- 容器
+- серьезн
+- 前後
+- schijn
+- ӧр
+- environ
+- ▁tiuj
+- 壳
+- ▁افراد
+- ▁Muster
+- 会发生
+- ▁아무
+- ली
+- से
+- はどんな
+- eeuw
+- сидел
+- 娃
+- 睁
+- sammlung
+- குறி
+- ▁shit
+- пыт
+- 营养
+- просил
+- रो
+- chè
+- ▁complained
+- ▁Szene
+- оф
+- ▁imitation
+- цам
+- runner
+- 保罗
+- ▁بدون
+- ▁Mol
+- 痛み
+- iĝo
+- ▁tackle
+- скіх
+- ▁mehreren
+- vance
+- ▁öğren
+- 本质上
+- ボード
+- 隔离
+- ▁encouragement
+- 政党
+- ▁үйл
+- 文本
+- konna
+- weiß
+- 辜
+- 担忧
+- third
+- อน
+- бач
+- ▁devote
+- おいしく
+- ▁January
+- ▁consumed
+- ▁pyramid
+- ▁verursacht
+- ▁banking
+- コロナ禍で
+- 決まって
+- ▁realised
+- ▁Syn
+- 向こう
+- 轩
+- ▁международн
+- ▁orchestra
+- 账户
+- ▁أيضا
+- ▁привет
+- ▁digo
+- ▁мере
+- ▁reaches
+- ▁nhiên
+- ▁aktuelle
+- stow
+- あそこ
+- 况
+- 紀
+- ǧǧ
+- ▁retten
+- ▁stability
+- ▁mož
+- ▁يُ
+- ▁waking
+- ▁Cada
+- 됐
+- 车站
+- 你想要
+- 並み
+- ەڵ
+- ▁collaboration
+- 歳で
+- 더라
+- ਤ
+- ▁skeleton
+- ▁Church
+- ▁duration
+- clav
+- ▁другое
+- ▁frère
+- ▁Institute
+- ▁quina
+- இன்
+- rijk
+- 最新的歌
+- 我刚刚
+- актив
+- ▁phần
+- ▁wandte
+- ▁Power
+- ▁région
+- ▁வந்து
+- ▁cierto
+- 基本上是
+- ▁Back
+- が生まれ
+- ▁whispering
+- ▁страх
+- ▁copies
+- ▁rouge
+- ▁убийств
+- ▁eloquent
+- ▁Hong
+- ▁kuva
+- ющим
+- Net
+- 晚了
+- ▁nostra
+- 我们一起
+- 然后他们
+- движ
+- 的反应
+- 这样的话
+- 考え方
+- 舒适
+- あるんですけど
+- லில்
+- 意味着什么
+- locat
+- ▁boiled
+- 不明
+- ▁fashionable
+- ▁proprietor
+- 複数の
+- 框架
+- ▁beantworten
+- ▁segundo
+- 導入
+- альной
+- おなか
+- ▁cinquante
+- 札
+- ▁guinea
+- ▁gabriel
+- 的作用
+- 買って
+- わかる
+- ▁geschah
+- aquell
+- ▁fino
+- درا
+- সু
+- ▁Händen
+- kamera
+- ▁expertise
+- ▁scenario
+- ▁حيث
+- 次々と
+- 連携
+- 打って
+- 我心里
+- 嫂
+- を出して
+- 诈
+- 太棒了
+- を目指し
+- 扩展
+- ▁concentration
+- 修复
+- ▁boca
+- fay
+- 机制
+- లే
+- ▁дверь
+- 底下
+- 侦
+- 栗
+- ▁Interview
+- ▁agitated
+- 沃尔
+- みる
+- ▁curios
+- āka
+- ▁solely
+- хгүй
+- 발
+- ▁Gedicht
+- ▁klicken
+- 到处
+- 不管你
+- 提示
+- 覚え
+- zah
+- ▁ayrı
+- 捡
+- ▁warfare
+- ▁colonial
+- ▁аан
+- ▁superintendent
+- ▁放一首
+- 一顿
+- 这些都是
+- ▁répond
+- 小学生
+- でもいい
+- 가지구
+- ▁lawrence
+- ایت
+- legung
+- sexual
+- 詰ま
+- アリ
+- 敦
+- 验
+- மைய
+- してしまう
+- 谨慎
+- ▁difficile
+- ▁конкрет
+- ▁describing
+- jährigen
+- ▁bastante
+- 诚实
+- ▁exercised
+- furi
+- ▁facile
+- ี่
+- ▁celebrate
+- Risate
+- 金曜日
+- ▁Mission
+- ▁спас
+- umunsi
+- 对他的
+- zelve
+- 星星
+- ▁erano
+- ▁Osten
+- ▁gael
+- 慧
+- ▁ප්‍ර
+- 朋友圈
+- ▁tusschen
+- yita
+- 今度
+- 周年
+- 聚会
+- дні
+- 漠
+- 耗
+- 毫无疑问
+- 轰
+- ▁скажите
+- ▁sandwich
+- ▁chambre
+- ▁traveled
+- ▁decidedly
+- ▁حالا
+- ▁despise
+- nesi
+- ŝa
+- ▁Schauen
+- 水果
+- ▁herald
+- 婚礼
+- ▁overwhelmed
+- ▁heathen
+- 一堆
+- netz
+- ▁mississippi
+- 土曜日
+- ▁internacional
+- あすの
+- hydr
+- 跳舞
+- ▁enthält
+- ▁reel
+- 按钮
+- 有效的
+- mayı
+- ▁resigned
+- ▁quantities
+- ▁submission
+- ваецца
+- ▁stärker
+- 倒れ
+- estructura
+- ▁Gegensatz
+- ▁llibre
+- 走过
+- ▁reminder
+- وں
+- щение
+- Wow
+- ▁içinde
+- 还想
+- ▁contrived
+- ▁ignore
+- 我也不
+- 져
+- ▁gregory
+- ▁hejuru
+- ▁korea
+- ▁realiza
+- 娘娘
+- ▁cooked
+- ▁Stimmen
+- ▁satisf
+- 脆弱
+- ▁vaguely
+- 顶部
+- 误会
+- 無料
+- 受到了
+- 的时候呢
+- 把手
+- ▁subsidi
+- ▁necesita
+- 假装
+- ▁persecu
+- ▁dintre
+- ▁endured
+- lender
+- étais
+- ▁litter
+- 愧
+- なのです
+- 我们发现
+- ▁Tamen
+- ▁hopeful
+- check
+- 声称
+- 意义上
+- 僕ら
+- 各个
+- ▁vertrauen
+- පා
+- ▁khu
+- ▁pathetic
+- 不清楚
+- 側は
+- 一个很好的
+- 六月
+- kker
+- ▁hả
+- ▁harness
+- ▁vanish
+- geblieben
+- 塚
+- ▁orchard
+- 争议
+- තර
+- 的核心
+- ▁bland
+- ▁heftig
+- ▁terrorism
+- 并非
+- நிலை
+- 中午
+- 衆
+- ▁arnold
+- 貨
+- ▁gehabt
+- 7%
+- فكر
+- 演唱的歌
+- 分布
+- 戦闘
+- ▁spake
+- ▁negativ
+- ▁kimwe
+- 畏
+- 眼前
+- ▁بۇ
+- ▁sorgen
+- 圣经
+- 只要你
+- ▁flowed
+- 当成
+- ▁önemli
+- ▁zurückkehren
+- 砲
+- ▁Soldaten
+- はっきり
+- energia
+- 言われ
+- が大きく
+- ▁plein
+- éré
+- 彼女の
+- ▁kasuta
+- 战士
+- ▁repeating
+- invest
+- ▁virtuous
+- ▁laboratory
+- ネタ
+- руп
+- ად
+- 贼
+- 殴
+- ▁буй
+- ▁இர
+- Video
+- ىلى
+- fydd
+- 几乎没有
+- 山田
+- гуль
+- ▁Pel
+- 任命
+- ▁discontent
+- ▁ຊິ
+- ▁winner
+- 可爱的
+- ▁verbreitet
+- 星球
+- ▁ааа
+- 耐心
+- ▁overtake
+- 災
+- ▁подумал
+- ▁hände
+- 打扰
+- வெளி
+- ▁wonderfully
+- ▁staatliche
+- 的房间
+- に来た
+- ▁nursing
+- ▁dunia
+- 笼
+- ▁chưa
+- ▁Verlust
+- Domin
+- kommst
+- монт
+- 逃跑
+- 估
+- 你还是
+- ▁stepping
+- 驱动
+- ▁сначала
+- しばらく
+- 衝撃
+- ▁blake
+- ▁Frankreich
+- ▁exploration
+- と思うんですが
+- ▁clinic
+- 違って
+- 寒さ
+- ▁какую
+- ▁soort
+- meid
+- ▁slack
+- ▁күр
+- ց
+- ▁Ozean
+- ić
+- 退出
+- ▁historically
+- ▁punika
+- 画像
+- ▁weighed
+- 本地
+- ấp
+- 济
+- ▁pouvait
+- ヒント
+- 레
+- ▁vuitanta
+- ▁verantwortlich
+- ▁neunzehn
+- hampton
+- alist
+- ▁склад
+- 時ごろ
+- Comp
+- '47'
+- ▁Cafodd
+- amategeko
+- ▁mnie
+- ওঁ
+- ▁öffnete
+- デー
+- あなたが
+- ▁enorm
+- 明治
+- 今大会
+- '`'
+- ▁archbishop
+- ▁иногда
+- ▁прежде
+- ▁göz
+- ▁entrar
+- ▁знать
+- числ
+- 奇妙
+- ▁kuvuga
+- ▁grub
+- ▁rubber
+- 坐着
+- に関しては
+- ▁preaching
+- işle
+- ▁psychological
+- ▁survivor
+- ▁Oberfläche
+- 这个想法
+- 脖子
+- 观看
+- ▁Verstand
+- ▁кож
+- ▁нужны
+- ▁diff
+- integra
+- ▁Hinweis
+- ▁maakte
+- 我希望你
+- 循环
+- ▁Simon
+- 中止
+- ▁crucial
+- ▁праца
+- ▁thông
+- дук
+- ▁ownership
+- ந்தார்
+- ▁Paar
+- ▁carta
+- 大変な
+- ▁hayat
+- ់
+- ▁ситуации
+- 欺骗
+- ▁regime
+- まるで
+- مُ
+- وص
+- 一个女人
+- 資金
+- ം
+- ▁kompliziert
+- ▁Geburt
+- 衝突
+- ▁Play
+- ▁Ärzte
+- ▁snatched
+- خنده
+- īgi
+- vè
+- ▁küçük
+- ▁navig
+- 両親
+- تحدث
+- ▁attire
+- ▁Pause
+- 連れて
+- ▁Einsatz
+- kî
+- ▁Free
+- 认识到
+- லு
+- 扩
+- 男生
+- 仓
+- ▁Laura
+- ▁magnitude
+- ацыя
+- torial
+- ▁tercer
+- 岸田
+- テスト
+- بى
+- état
+- ▁máy
+- schläge
+- 激烈
+- ▁같은데
+- ▁kep
+- 这就像
+- পি
+- φ
+- 肩膀
+- ▁psychology
+- ▁wünschen
+- ▁wiped
+- を求める
+- ظهر
+- ▁streak
+- ▁Aktion
+- لې
+- ▁vấn
+- ▁owed
+- ▁assessment
+- ▁solomon
+- 魔法
+- ▁digging
+- ▁Kleidung
+- ▁волн
+- піса
+- 직
+- ښ
+- 叠
+- 垫
+- ▁provoke
+- ▁disturbance
+- 范围内
+- 教学
+- ▁труб
+- ▁Deutsch
+- ێت
+- ▁করা
+- ▁Englisch
+- fluent
+- ▁какого
+- 陪你
+- sulta
+- ▁applause
+- ▁тэгээд
+- ას
+- ▁mmhm
+- ▁rascal
+- тәр
+- ▁honesty
+- ▁Phase
+- Risos
+- ▁постоянно
+- 申請
+- ັກ
+- ▁hängt
+- ▁affecting
+- ▁Veränderung
+- ▁опыт
+- હ
+- ▁اليوم
+- ▁друж
+- 碰到
+- ▁Bühne
+- ▁порядк
+- ҭа
+- 新聞
+- ▁Stimm
+- ▁Süden
+- مام
+- 演説
+- 世界上最
+- leitung
+- 遣
+- 北日本
+- 감
+- 碳
+- ▁collecting
+- っぽ
+- ができて
+- やりたい
+- ▁vardı
+- ▁reap
+- 背中
+- ▁Farb
+- 焼け
+- ないよ
+- 摄
+- 하다
+- 외
+- ▁appealed
+- ッキー
+- ▁prick
+- 上手
+- お届け
+- kono
+- ▁мэдэ
+- 顾问
+- 很清楚
+- 斑
+- ▁интернет
+- 刃
+- һын
+- ▁نور
+- awal
+- ▁flashing
+- ▁Since
+- 大师
+- ▁macro
+- เก
+- ▁unconsciously
+- ▁passé
+- ▁dreaded
+- ძ
+- ▁cependant
+- 초
+- ▁Там
+- ▁бело
+- ▁personnel
+- 犯行
+- ▁priorities
+- шев
+- ▁grote
+- neuron
+- ▁treating
+- âme
+- ▁Bach
+- 有许多
+- සු
+- ▁Pic
+- ▁Ereignisse
+- 感染対策
+- ▁discoveries
+- spoiled
+- lith
+- 喝了
+- ታ
+- ▁Geschlecht
+- ▁excursion
+- 保険
+- ▁fonction
+- 場に
+- maktadır
+- ▁judicial
+- 安倍元総理
+- ▁Sobald
+- lardan
+- кажет
+- ディング
+- ▁harmless
+- 这么一个
+- ▁ministry
+- ▁ຫັ້ນ
+- ▁cotxe
+- 嘉宾
+- を取る
+- ىسى
+- lomb
+- という話
+- ▁siaj
+- ambul
+- ▁referring
+- 小小的
+- ▁utterance
+- 角落
+- ▁kehrte
+- パリ
+- 捐
+- ▁provincial
+- 想出
+- ▁benefici
+- ▁Aku
+- ▁исслед
+- ▁Kong
+- ıp
+- 激情
+- 周期
+- ▁Ruf
+- 坐下
+- ▁ewig
+- ▁African
+- ▁çünkü
+- ▁academy
+- ▁Maybe
+- 无限
+- čči
+- ▁connu
+- 这孩子
+- ▁banker
+- 村上
+- ▁authentic
+- 寂
+- 郷
+- ▁kết
+- ヴィ
+- ▁luncheon
+- ▁알아
+- 冷凍
+- ▁resume
+- андр
+- ▁refined
+- clus
+- 这样的一个
+- ாளர்
+- 鲍
+- ▁nobility
+- ▁Ул
+- ໍາ
+- を重ね
+- ▁嗯嗯
+- 談
+- ▁awaited
+- فَ
+- ▁Aussage
+- ღ
+- ▁vermute
+- 涌
+- హ
+- 月曜日
+- ▁tốt
+- esque
+- ▁جدید
+- 選んだ
+- ▁Wann
+- 身后
+- アイス
+- カ月
+- ▁beschreiben
+- 诱
+- ▁chwarae
+- 附近的
+- inclou
+- 什么样
+- 請
+- කු
+- ल्
+- ▁initially
+- 期节目
+- ▁roaring
+- ▁Ар
+- ▁другим
+- ▁oven
+- 必须要
+- 20%
+- ▁saloon
+- ▁врач
+- せん
+- ▁lewis
+- ▁vicious
+- ▁Zentrum
+- ▁brazil
+- 的权利
+- 美味
+- 不许
+- ▁பயன்படுத்த
+- ▁вульф
+- ▁видимо
+- 先発
+- きれいに
+- ▁ужас
+- ▁obtaining
+- ▁Aufnahme
+- ▁podría
+- ▁sadness
+- 拿出来
+- 必要がある
+- лыҡ
+- 作业
+- 采用
+- ▁vya
+- ▁feat
+- ▁Although
+- なければならない
+- 大規模な
+- 推进
+- 我只想
+- ▁durum
+- 还有其他
+- どうする
+- œuvre
+- ▁eloquence
+- ▁бөгөөд
+- 栋
+- 鏡
+- būt
+- günstig
+- 开会
+- ▁Grab
+- ▁அவர்கள்
+- ロシアが
+- voja
+- ▁felicit
+- ▁attendance
+- 否认
+- ぇ
+- ლა
+- わずか
+- ▁hunted
+- ようやく
+- 爵士
+- ހަ
+- 然而
+- ▁mournful
+- 長期
+- wamu
+- 期限
+- 询问
+- ▁установ
+- ▁rainbow
+- 騒
+- ▁diligent
+- ▁vergangen
+- ▁asylum
+- ▁yanjye
+- koht
+- ▁comfortably
+- gerät
+- 凶手
+- ▁armour
+- pari
+- ▁dignified
+- ▁cuerpo
+- ▁Prinzip
+- associa
+- つま
+- ▁excite
+- ▁matches
+- ▁brauche
+- ▁كې
+- ddling
+- ▁baggage
+- ▁outstanding
+- 秒钟
+- ▁vierzehn
+- 克服
+- ▁blazing
+- பின்
+- 下がって
+- ▁руку
+- jící
+- わたし
+- න්නේ
+- ▁consegui
+- 他们正在
+- ▁qualified
+- ▁tradi
+- 统一
+- 人が死亡
+- buah
+- ▁защит
+- ▁donkey
+- 新幹線
+- 欲しい
+- クル
+- ifuza
+- miento
+- ▁lumber
+- jye
+- ▁meteor
+- ▁gradual
+- களும்
+- ▁crowned
+- quisition
+- ショート
+- 糊
+- ही
+- チャレンジ
+- னும்
+- ▁एक
+- ▁sydney
+- ▁Taylor
+- 克莱
+- 拿起
+- リーダー
+- 兔子
+- 其他地方
+- ▁Graf
+- ▁ужо
+- ël
+- өнгө
+- ▁fünfundzwanzig
+- ▁estimated
+- 一趟
+- ▁quoted
+- த்தா
+- spruch
+- 激しく
+- ▁Был
+- čil
+- トラブル
+- пустил
+- ▁seventeenth
+- ▁thirteenth
+- ▁compelling
+- ▁seemingly
+- ▁نگاه
+- 更好地
+- ▁bunk
+- のまま
+- nev
+- 借り
+- ▁кре
+- krist
+- 七月
+- ▁Atom
+- spezifisch
+- 一直是
+- Orient
+- йтесь
+- ▁meilleur
+- '120'
+- ▁habt
+- ▁liking
+- ▁legislature
+- どのような
+- 一块儿
+- いくつか
+- ▁romp
+- ▁trench
+- ▁сильн
+- 凑
+- 鸣
+- 登録
+- ▁frail
+- 获取
+- ▁Main
+- いること
+- 広く
+- プリン
+- 親子
+- 段階
+- 豆腐
+- 很长时间
+- ▁Castell
+- 真理
+- ▁треб
+- ▁endeavoured
+- Institut
+- 萌
+- Infrastruktur
+- 堀
+- 诉讼
+- ▁прекрасно
+- ってほしい
+- キャン
+- 徐々に
+- ▁يجب
+- ▁vall
+- ზ
+- eisen
+- ▁Gewicht
+- ▁العمل
+- ▁contemporary
+- 戦略
+- ▁wesentlich
+- కి
+- 予定です
+- っけ
+- ▁bello
+- ▁socialist
+- ということなんですね
+- ▁cement
+- ▁그걸
+- ▁shifting
+- ▁Sup
+- ▁дене
+- ▁governed
+- schäm
+- ▁jurisdiction
+- ▁giorno
+- ▁triple
+- funk
+- ▁sneak
+- 院长
+- 食料
+- чныя
+- vira
+- ▁spoon
+- сця
+- ▁silenci
+- 出售
+- 几十年
+- geordnet
+- ▁Philosophie
+- ▁congratulate
+- リズム
+- ▁scheen
+- ▁пятый
+- ▁tidings
+- ▁podem
+- 北陸
+- 大姐
+- ▁hệ
+- ▁Lind
+- accus
+- иров
+- 另
+- 付いて
+- ▁vegades
+- ▁কৰি
+- ▁inherent
+- räumen
+- ▁Vorschlag
+- ▁Dabei
+- ッジ
+- ▁Interessen
+- 得很好
+- неп
+- ▁keel
+- 一阵
+- ▁treasury
+- ▁sculpture
+- ▁vật
+- ▁Beach
+- できなかった
+- ちょっと待って
+- 巨人
+- ▁Verein
+- 乗せ
+- ▁Ple
+- 冲动
+- 休み
+- syll
+- ත්ත
+- ോ
+- ▁غیر
+- ロック
+- чки
+- 回答说
+- うまみ
+- ලි
+- thorpe
+- 其中一些
+- をつか
+- ▁outdoor
+- に向けた
+- ▁amuse
+- К
+- ▁bantu
+- uwen
+- 辅
+- lulu
+- নী
+- ativa
+- 網
+- 詹姆斯
+- 懂得
+- 太空
+- 四处
+- いずれも
+- ▁Vertrag
+- ▁verbinden
+- 脊
+- ఎ
+- ▁обычно
+- ▁dispatch
+- 切った
+- と指摘
+- ▁komplett
+- 火事
+- ▁migration
+- kintu
+- 働
+- 札幌
+- ▁касается
+- 告诉他们
+- ▁ingenious
+- だとか
+- عادة
+- 金钱
+- ضع
+- 般的
+- ▁fling
+- 正如你
+- ▁durchführen
+- コーヒー
+- スケート
+- 的意思是
+- 很多事情
+- ▁crisp
+- ▁grunt
+- ▁kutoka
+- ES
+- 爵
+- ▁dự
+- ▁catastrophe
+- ▁tưởng
+- ▁disregard
+- ▁gezegd
+- ▁Kredit
+- 強さ
+- ▁بودم
+- OS
+- beruf
+- ▁handling
+- ▁pesca
+- 根本不
+- ▁Champ
+- ▁девят
+- 晒
+- ▁evaluate
+- ▁Dewi
+- ▁ответствен
+- щего
+- 认知
+- ovna
+- ▁paragraph
+- 決めて
+- ▁roared
+- ▁Quindi
+- 毫不
+- 름
+- ▁Sitzung
+- கிறது
+- ▁willow
+- 欢迎来到
+- 船上
+- 维护
+- ▁tanta
+- ▁Europäischen
+- ▁Molt
+- শ্
+- tiene
+- 種類の
+- を超えて
+- ▁terminal
+- ▁consulted
+- ▁portal
+- алды
+- ▁المع
+- assassinat
+- 証明
+- 标记
+- يار
+- 一起工作
+- ▁ئەو
+- ▁Pflanzen
+- ▁embarrassed
+- kawa
+- ▁serene
+- ▁streaming
+- ▁jason
+- ▁stride
+- ▁лиш
+- 多次
+- භ
+- ▁almighty
+- ▁hurricane
+- ▁Science
+- ▁سره
+- fashe
+- quadra
+- 常常
+- 国民の
+- ▁dạ
+- põ
+- ▁discount
+- を出す
+- 到目前为止
+- fähigkeit
+- 谈判
+- スパイ
+- ▁patiently
+- ▁lautet
+- ▁болох
+- 惊人的
+- fleur
+- ▁prosperous
+- 委屈
+- ică
+- ▁Brian
+- 两百
+- ▁ভাল
+- ▁brandy
+- 实在是
+- àng
+- ▁produit
+- палі
+- 客厅
+- を示す
+- ▁gezwungen
+- исто
+- stairs
+- ▁lapse
+- 镜头
+- 般
+- ▁Està
+- operative
+- ▁Veranstaltung
+- 二人
+- දී
+- ▁moll
+- ▁courteous
+- رَ
+- ▁Oberst
+- 谁呀
+- 車両
+- анг
+- ▁série
+- ▁behandeln
+- பட்ட
+- とかそういう
+- лок
+- ▁کړ
+- ▁Austr
+- 看起来很
+- 酒吧
+- ▁বলে
+- пок
+- インターネット
+- 天皇
+- ▁estaban
+- зур
+- 极端
+- ▁distrust
+- بری
+- 去哪里
+- ▁сельсовет
+- ▁نحن
+- ▁erhöht
+- ▁mặt
+- 軒
+- ▁Abendessen
+- ▁accusation
+- 领导人
+- льная
+- ▁görün
+- 年生の
+- 伊藤
+- 支払
+- 肌肉
+- ▁zitten
+- したうえで
+- chimp
+- 恋爱
+- ▁ег
+- ▁зелен
+- 动机
+- ▁Konflikt
+- 군
+- ▁sabbath
+- '".""'
+- ▁kapit
+- セント
+- ▁polish
+- ▁simultaneously
+- की
+- ▁kugeza
+- 相当于
+- ▁годы
+- علوم
+- итесь
+- ▁expressing
+- іла
+- 美女
+- umugabo
+- قطع
+- ▁schickte
+- ▁реша
+- ▁Fälle
+- ெய்
+- migr
+- 己
+- ální
+- ▁людям
+- ▁некоторые
+- バーディー
+- fungu
+- ▁executing
+- بحث
+- schieben
+- 引发
+- 质疑
+- ▁Gründen
+- grenzen
+- ▁persecution
+- ▁Zitat
+- 土砂災害
+- ▁besorgt
+- ▁unfair
+- 到这儿
+- 인데
+- عات
+- ▁ребята
+- ouvre
+- 第二次
+- рыс
+- ших
+- ▁тус
+- しまい
+- 思われ
+- 鹅
+- ▁Katze
+- ▁strengthened
+- kamu
+- ▁Royal
+- ombro
+- ▁folgt
+- ▁destaca
+- ▁همان
+- خورد
+- 倦
+- スポット
+- rechnung
+- ▁inheritance
+- ىدۇ
+- ▁பிற
+- ▁Kevin
+- ▁behaved
+- 数百
+- 明显的
+- roving
+- ▁наконец
+- 稚
+- 구나
+- 犹太人
+- 确认
+- ▁தீ
+- لِ
+- 二十年
+- 为何
+- ходил
+- жат
+- märk
+- ұ
+- ▁الوقت
+- ▁depois
+- ▁цар
+- ▁despised
+- 商人
+- ొ
+- ▁besuchen
+- 詐欺
+- スタイル
+- ▁überzeugt
+- ここの
+- wili
+- ▁Life
+- 特别好
+- ▁ziet
+- ラム
+- を続ける
+- rinda
+- 호
+- kanya
+- ▁chairman
+- 木さん
+- トイレ
+- крыва
+- ▁stooped
+- ▁Schatten
+- ▁kuwi
+- 一本书
+- ebene
+- вец
+- ▁занят
+- ▁garment
+- ▁Danach
+- ▁ăn
+- ▁Как
+- が続いています
+- 遂
+- क्ष
+- ▁acceptance
+- ▁körperlich
+- 義務
+- 该怎么
+- ائي
+- ▁krista
+- ителя
+- ымі
+- ▁Neuro
+- ▁semble
+- を進めて
+- 見られる
+- 遭受
+- ▁Einheit
+- ▁ditch
+- ▁Sydney
+- ▁üzerinde
+- 葛
+- ▁российской
+- ▁шестьсот
+- ▁advertisement
+- ▁lucruri
+- 伯特
+- 小林
+- ▁bamu
+- հ
+- ▁Pakistan
+- ▁hành
+- わからない
+- ▁historische
+- kiem
+- ▁nghe
+- かつ
+- 约会
+- もらえる
+- tsiooni
+- ▁preached
+- 울
+- 胳膊
+- ▁exaggerate
+- ▁subscription
+- ▁sworn
+- ▁evolved
+- 认可
+- verhalten
+- ▁hạ
+- нде
+- ▁путина
+- ▁accumulate
+- いくら
+- ▁Dž
+- 先頭
+- ▁berlin
+- 僅か
+- ▁Knie
+- 数据库
+- ▁жест
+- ▁زیادی
+- ▁envoy
+- ▁самого
+- väg
+- 录音
+- جعل
+- タイトル
+- ▁protecting
+- brian
+- 緩
+- 勘
+- もしかしたら
+- ▁shrugged
+- willig
+- ▁иә
+- නය
+- 三个月
+- 時代に
+- 音频
+- 来看看
+- මි
+- ▁vicinity
+- ▁choosing
+- ▁வழி
+- 奇迹
+- ▁Schreiben
+- лээ
+- croft
+- ▁kontrol
+- бот
+- ▁izmanto
+- ▁좋은
+- пуск
+- parent
+- ▁container
+- 疯了
+- 尼斯
+- 確認された
+- ▁thicket
+- ▁верн
+- ▁dankbar
+- 并将
+- 大陆
+- ▁comercial
+- ▁vorhanden
+- jev
+- 对面
+- ▁tiefer
+- ▁cama
+- 不能再
+- だと思う
+- 則
+- 鸭
+- ▁soutenir
+- 幺
+- ▁Schlacht
+- ▁chaos
+- ▁meisje
+- を行った
+- ▁conjunt
+- ▁Ansatz
+- 皇后
+- 环境中
+- 님
+- 世紀
+- も含めて
+- '98'
+- ობ
+- त्य
+- ▁derecho
+- ▁davis
+- ▁Шу
+- お酒
+- ▁verhindern
+- ▁komunumo
+- ▁mochte
+- こだわり
+- 设施
+- 天使
+- ▁schade
+- позиц
+- 9%
+- пай
+- ▁richmond
+- ▁посмотрел
+- eusement
+- meester
+- ỹ
+- োৱা
+- 久しぶり
+- 候选人
+- ங்களில்
+- ন্দ
+- ▁Zusammenarbeit
+- 少しずつ
+- ▁Tausende
+- 东方
+- いただき
+- aquestes
+- ▁gefällt
+- гада
+- 薇
+- ▁mathematics
+- ▁처음
+- lago
+- amara
+- zentrum
+- ▁forgiveness
+- ▁Each
+- ▁அதன்
+- ▁Deci
+- 他にも
+- ▁mirth
+- 建立了
+- Ɣ
+- сор
+- ▁Wörter
+- Bahn
+- ▁bajo
+- ▁carne
+- 可怜
+- thought
+- größe
+- ガラス
+- ▁screaming
+- ▁ändert
+- 培
+- 给了我
+- અ
+- ▁thôi
+- ▁indicator
+- ▁alfred
+- ▁کنی
+- ▁gezien
+- جلس
+- ▁disciplined
+- ▁housekeeper
+- 体操
+- 还挺
+- ▁submarine
+- いきなり
+- كرة
+- ▁erneut
+- ▁Medizin
+- 资助
+- vå
+- ▁lucru
+- ▁разные
+- ▁oars
+- ливо
+- 可惜
+- ▁Oxford
+- Kampagne
+- ▁generosity
+- ▁무슨
+- ญ
+- 縦
+- 偽
+- ível
+- パワー
+- ▁moviment
+- 尽量
+- ▁universitat
+- ME
+- ễ
+- ▁нравится
+- ▁понимаешь
+- ▁böyle
+- 学术
+- こんにちは
+- ▁kvar
+- 摊
+- 棵
+- ▁hereafter
+- Franc
+- ▁prussia
+- ஷ்
+- 天哪
+- ▁Hügel
+- 最新の
+- ▁Korea
+- ▁мөн
+- ▁হয়ে
+- ▁kaiser
+- ático
+- ▁mooi
+- 晩
+- ึ
+- ▁cambio
+- holm
+- ственно
+- ▁implied
+- ▁év
+- 我个人
+- ▁jenny
+- 就是因为
+- Ĉ
+- を通じて
+- ▁değiş
+- ▁borrowed
+- 是否有
+- ▁Tout
+- ▁festgestellt
+- きょうも
+- ▁تک
+- 短暂
+- ▁franklin
+- ▁piled
+- 还是要
+- 自主
+- یەک
+- ▁contempor
+- 狩
+- 불
+- نظام
+- 亲戚
+- 活跃
+- 真心
+- 会让你
+- ▁کشور
+- ování
+- ▁stesso
+- бег
+- 临时
+- 八点
+- いろいろな
+- घ
+- 螺
+- க்கை
+- ▁consultation
+- ▁Wut
+- ▁Personal
+- ▁quedar
+- ▁こうした中
+- ▁Ку
+- ▁prolonged
+- ▁folgenden
+- ▁temporal
+- ▁bleef
+- ăng
+- バター
+- ▁Saya
+- ▁detected
+- ▁Про
+- ▁translate
+- 信念
+- асць
+- ないんです
+- ▁کوچک
+- 両方
+- ▁contrari
+- ▁அத
+- ோம்
+- お肉
+- ņa
+- ▁Einstellung
+- したいと思います
+- ▁встреча
+- wyl
+- 侵攻
+- 我刚
+- ۱
+- 房地产
+- ষ্ট
+- ▁ஆனால்
+- 收藏
+- مثل
+- ▁Philip
+- ▁reliable
+- ▁conspiracy
+- ▁indispensable
+- 日本海側
+- の歴史
+- ▁কোন
+- ▁superiority
+- 子弹
+- 的意见
+- ▁conqueror
+- 帖
+- 迎え
+- ▁одоо
+- 優しい
+- ическое
+- wyth
+- ▁одну
+- wirkung
+- ▁financing
+- ▁ascended
+- كتب
+- 八月
+- ▁zoals
+- ヶ
+- ▁கட்ட
+- ▁petty
+- ▁cricket
+- すぎて
+- 得起
+- cross
+- 加强
+- crypt
+- ▁jünger
+- ▁ຫວາ
+- ▁considér
+- ▁Studio
+- вд
+- 另外一个
+- ▁kennedy
+- ▁castell
+- าม
+- ▁abrupt
+- ▁buli
+- identifi
+- ▁disadvantage
+- ▁නො
+- ▁arasında
+- ража
+- ▁хотят
+- ▁apron
+- ▁damned
+- 不在乎
+- ▁horace
+- 帮助我们
+- communica
+- жин
+- 貸
+- たちに
+- ▁complement
+- ▁والم
+- джи
+- ▁Rick
+- கிறார்
+- ▁maximum
+- อา
+- ▁ҡара
+- ▁lebendig
+- ▁счита
+- 毫
+- ▁mechanic
+- ෑ
+- ナンバー
+- 餐厅
+- 援助
+- ▁khá
+- ▁creu
+- apport
+- ▁continual
+- 了多少
+- ところです
+- 但我认为
+- ▁Villa
+- ▁reagieren
+- ▁нічога
+- 筒
+- 贫困
+- ▁puerta
+- ▁pathway
+- 效率
+- 津波
+- ▁Europ
+- ▁бесп
+- ▁счет
+- 对抗
+- 生物学
+- writer
+- 認め
+- ▁extravagant
+- ▁umbrella
+- ▁jullie
+- ▁distressed
+- ▁precisa
+- 称为
+- ▁honorable
+- ూ
+- 伊斯兰
+- 尊敬
+- ▁clinging
+- ▁бала
+- льных
+- pā
+- ▁civilized
+- 出てきて
+- BI
+- ▁apparatus
+- ▁затем
+- にわたって
+- 道具
+- ▁Grenze
+- ▁велико
+- печат
+- ▁babi
+- ▁blunt
+- ▁محل
+- 漆
+- ছো
+- ▁vegetable
+- regierung
+- かき
+- ▁ocasi
+- ▁lacking
+- 颤抖
+- ▁thereupon
+- 另一方面
+- 最後まで
+- düğü
+- 七点
+- basha
+- bikora
+- 共享
+- 存储
+- ▁clark
+- 是什么意思
+- ▁schoon
+- ▁Nahrung
+- ▁Elektro
+- ▁yapıyor
+- ことば
+- kibi
+- ▁Tony
+- hér
+- 粮
+- 起床
+- :“
+- Râsete
+- 萧
+- ハウス
+- partei
+- 分别
+- ▁principalment
+- 戴着
+- ▁پرو
+- occupa
+- 部落
+- ▁favourable
+- ▁expose
+- 売り上げ
+- ▁Marie
+- 怪我
+- ▁практически
+- ▁별로
+- 偷偷
+- ▁complexity
+- eût
+- vamo
+- ▁automatic
+- mysl
+- ремен
+- dimensional
+- прям
+- ▁Beweis
+- 犠牲
+- нең
+- anomena
+- строй
+- ▁طريق
+- の間で
+- ▁ethel
+- 締め
+- 只有一个
+- 分散
+- ▁alright
+- プラ
+- ▁approaches
+- ြ
+- 汪
+- овского
+- человеческ
+- ượ
+- 発売
+- ▁quindi
+- คน
+- ▁diplomat
+- ▁mulher
+- 人才
+- ▁scold
+- 灰色
+- 寸
+- 叙
+- ▁covenant
+- ▁Mind
+- ▁Four
+- 气候
+- ▁kennt
+- сер
+- ▁pew
+- guye
+- валася
+- ▁instructed
+- ▁இல்லை
+- 地看着
+- 国葬
+- ▁газар
+- 掩
+- 筆
+- 艾伦
+- 飛ば
+- ID
+- ▁substitu
+- tracht
+- 名称
+- だと思って
+- ▁mientras
+- 相手に
+- ▁Jason
+- appropri
+- ▁höre
+- 捜
+- ▁தனது
+- ▁مشکل
+- بند
+- 犹太
+- ジョ
+- ▁Dienste
+- 武装
+- ydı
+- ▁இருந்தது
+- ▁праз
+- gemacht
+- ▁feder
+- 炊
+- 合理的
+- leuchtet
+- ▁Bereit
+- ▁taylor
+- そうと
+- ивают
+- 惊喜
+- 知道吗
+- ▁constance
+- あげる
+- ворот
+- 台上
+- plau
+- 剥
+- 古老的
+- 也知道
+- ▁strategi
+- ▁amateur
+- ▁mettre
+- 日军
+- んでしょうね
+- ゥ
+- ▁orleans
+- 说出来
+- 眼里
+- ▁blunder
+- あいつ
+- 一个小时
+- ▁moist
+- ▁teatr
+- 以一种
+- ▁разве
+- 欺
+- ▁vernünftig
+- 疼痛
+- রের
+- ▁Kohle
+- géni
+- ▁oyun
+- ▁healing
+- brä
+- father
+- 王国
+- 伸出
+- 就不能
+- 火山
+- ▁пару
+- 最后一次
+- ▁Kö
+- 巾
+- abaturage
+- ▁defiance
+- ▁москвы
+- 観光客
+- 够了
+- ▁olw
+- ▁açık
+- ▁primi
+- czas
+- ▁المس
+- ▁blev
+- ▁sauber
+- ▁voting
+- ▁complicat
+- ณ
+- ▁través
+- ▁optimize
+- ▁melodi
+- ▁lavoro
+- ▁подожд
+- ▁войны
+- するのが
+- ▁diminu
+- と呼ばれ
+- ▁самых
+- ▁bijna
+- ▁bildet
+- つながり
+- 棉
+- روس
+- 始终
+- ▁yacht
+- ▁packet
+- šā
+- しているんです
+- ▁Wid
+- ▁hose
+- istisch
+- ▁prezent
+- ▁missionary
+- ▁commonplace
+- 駆け
+- プロジェクト
+- ▁circus
+- クラブ
+- ▁customary
+- ▁exclusively
+- 鑑
+- 枠
+- 吵架
+- ▁peine
+- 一起来
+- 時まで
+- いいね
+- ▁mathematical
+- 珍しい
+- ▁иначе
+- ▁depriv
+- ▁venice
+- ▁sitzt
+- 留给
+- ▁Court
+- ▁zooals
+- ぷ
+- ▁versteckt
+- ▁stata
+- ▁billig
+- TA
+- shima
+- 树林
+- ▁iawn
+- ▁plac
+- ১
+- ▁memorial
+- 在做什么
+- ▁thường
+- ▁ladyship
+- world
+- 危険な
+- ▁До
+- の中でも
+- ▁mostrar
+- 昨晚
+- ▁appreciated
+- ▁جنگ
+- ▁bluff
+- 庙
+- ▁emphasize
+- ▁renown
+- 沟
+- 陸上
+- 一点也不
+- lê
+- сия
+- 椅
+- œ
+- 函
+- ▁admiring
+- ▁sacrament
+- 财务
+- 节奏
+- 礼貌
+- 广场
+- ▁implore
+- ицы
+- マリ
+- 这个事情
+- いいのか
+- があるので
+- 年级
+- kiko
+- ▁exam
+- っていない
+- ▁diameter
+- ▁Palm
+- бә
+- 起诉
+- ▁ہو
+- 大好きな
+- ▁cetera
+- ▁पर
+- もう少し
+- 瘾
+- 涙
+- вания
+- ▁overflow
+- ▁ожида
+- 临床
+- ▁сябе
+- männer
+- ▁contradiction
+- 吊
+- ▁사람들
+- ▁ساعت
+- ▁العديد
+- ▁никакого
+- 的思想
+- ▁obstant
+- andika
+- ▁legion
+- ▁cultiv
+- ▁arriba
+- ▁przed
+- võt
+- 行う
+- ય
+- ▁allerdings
+- ogene
+- schalten
+- demokrat
+- ▁traced
+- ▁считает
+- ▁produc
+- 春天
+- ▁burada
+- 赶快
+- င်
+- ં
+- ゼレンスキー大統領
+- ▁случилось
+- ▁состав
+- Ҡ
+- ▁bemerkt
+- 原本
+- 現金
+- Gerät
+- のようなもの
+- енә
+- ▁Pur
+- ▁kreativ
+- ▁behauptet
+- ▁للم
+- ▁новый
+- ▁hardware
+- свет
+- ければ
+- 贫
+- 誰が
+- ▁marque
+- ▁stuart
+- を見ると
+- ▁Menschheit
+- 深深
+- очку
+- ব্য
+- ▁roam
+- ▁kujya
+- 二百
+- 行不行
+- 慣れ
+- ▁savu
+- 原発
+- ▁hakkında
+- 规矩
+- ▁stubborn
+- ▁полно
+- ▁übrigens
+- ▁offenbar
+- ▁tipus
+- ▁strained
+- madı
+- ドン
+- 朝から
+- ロボット
+- ▁verletzt
+- 的说法
+- ண்ட்
+- 尤
+- 我听说
+- 救助
+- 体調
+- ▁cooperation
+- 做了一个
+- ▁junger
+- 一点儿
+- ▁dusty
+- 开枪
+- ▁Angebot
+- 珊
+- ▁Тэд
+- 义务
+- නු
+- interest
+- 血管
+- ▁trouva
+- වැ
+- истов
+- ▁ҡал
+- ģ
+- ▁vulnerable
+- ▁receipt
+- 洗濯
+- تعلم
+- 厕所
+- ▁conductor
+- ▁schreibt
+- ▁Verbrechen
+- ▁замечательн
+- ▁adviser
+- ▁hostess
+- 挙
+- ע
+- ▁cylinder
+- ▁امروز
+- ▁treason
+- ▁Sever
+- ıyla
+- ▁Vogel
+- ▁wertvoll
+- 书记
+- 跃
+- ▁gravel
+- ▁preliminary
+- ▁bảo
+- 証拠
+- ▁solved
+- ▁будто
+- わよ
+- 果然
+- вацца
+- ことになりました
+- 媒
+- یں
+- ▁accuracy
+- ▁commodity
+- ▁District
+- بيع
+- ĵ
+- ▁implemented
+- 三月
+- バレ
+- ▁краін
+- цией
+- 能看到
+- 或其他
+- 嗨
+- അ
+- ▁belangrijk
+- 舟
+- 포
+- 償
+- ▁komplexe
+- ▁basketball
+- ▁Sekunden
+- ▁noisy
+- ▁interruption
+- 说完
+- ケア
+- illus
+- ▁compliance
+- ▁اتفاق
+- ▁psalm
+- ▁electrical
+- ენ
+- ▁vragen
+- ▁shun
+- 逮捕されました
+- ▁severity
+- 之内
+- 么
+- half
+- 找出
+- ٍ
+- ▁موقع
+- ▁Signal
+- 我问你
+- ▁pobl
+- цяг
+- 契
+- 贯
+- ▁făcut
+- ▁Đây
+- 阴影
+- 南京
+- ▁pouvez
+- ▁Spieler
+- евой
+- kipe
+- тап
+- 花钱
+- ▁doktor
+- ▁вперед
+- ▁обязательно
+- ▁صحبت
+- iyoruz
+- 製品
+- ஞ
+- 抬起
+- 合意
+- ▁quả
+- ▁coch
+- ковский
+- 儀
+- する方針
+- ▁fringe
+- geschrieben
+- が起きた
+- 価
+- ▁государство
+- buye
+- ▁внутрен
+- 疑いが持たれています
+- ▁мама
+- угл
+- ாவது
+- になれ
+- ▁salad
+- 什么都不
+- ▁ghastly
+- 匆忙
+- 忽视
+- ▁universities
+- ▁Handlung
+- cull
+- ▁maggie
+- ▁Papa
+- ̀
+- 旺
+- ▁zerstört
+- ▁vapor
+- ▁bafite
+- 欲望
+- ▁sicherzustellen
+- ▁Voll
+- টো
+- ▁материал
+- ▁gemein
+- ▁sorrowful
+- 诗歌
+- ibindi
+- 保安
+- ▁đấy
+- 不管怎样
+- ▁automatically
+- まっている
+- ムー
+- ▁Shu
+- 怎么着
+- 苏联
+- ▁Jersey
+- ▁произошло
+- ▁Bạn
+- ▁Viertel
+- exclusi
+- 售
+- 唯
+- 取代
+- ▁handeln
+- ▁blur
+- 相机
+- 种植
+- ▁hark
+- 污
+- ▁псих
+- ▁ritual
+- ▁потеря
+- 你放心
+- ▁rejoiced
+- طلب
+- ▁visage
+- ぶつ
+- operation
+- ▁камен
+- ▁conseil
+- ▁liable
+- 蚊
+- эргэ
+- ▁यस
+- работал
+- üßt
+- ランク
+- ▁occhi
+- ▁Мин
+- ▁beendet
+- ▁kitten
+- ▁зуб
+- ▁Kenya
+- ▁ikibazo
+- ▁أيضًا
+- デジタル
+- ▁abbey
+- 会觉得
+- ично
+- ிருக்க
+- を通して
+- 那不是
+- жыць
+- 通行
+- ▁longue
+- ▁Heimat
+- ▁intrigue
+- قدر
+- бен
+- ▁joven
+- bücher
+- 山本
+- ▁priorit
+- 承受
+- 结束时
+- wezi
+- ▁regal
+- ▁emit
+- ▁анти
+- 判決
+- ኝ
+- ▁eyebrows
+- ▁bicycle
+- ▁çıkar
+- дина
+- みそ
+- பர்
+- 争取
+- 个性
+- 五分钟
+- ก็
+- ▁смотри
+- kontroll
+- 밖에
+- ▁exalted
+- 消え
+- ▁gebeten
+- ER
+- ▁прибыть
+- 弾道ミサイル
+- ▁решения
+- ▁அவள்
+- 火星
+- 怖
+- 预期
+- 衡
+- störung
+- усь
+- グルメ
+- ▁minority
+- ▁neighbouring
+- ▁Leidenschaft
+- ▁oppressed
+- 你会看到
+- gène
+- ▁tutor
+- 陷
+- ▁äußerst
+- 嘞
+- ▁començar
+- objekt
+- ▁proverb
+- கம்
+- anstalt
+- ▁holiness
+- ▁шүү
+- ▁государства
+- ▁мус
+- ▁pregunt
+- 砍
+- ▁überzeugen
+- ▁다른
+- слух
+- ▁twig
+- зад
+- analyse
+- そうした
+- 战场
+- 网友
+- ლ
+- Ḥ
+- ▁njegov
+- ▁repeatedly
+- ▁اولین
+- ▁любой
+- ▁archer
+- ▁اور
+- 積み
+- étude
+- 我总是
+- ▁zweihundert
+- ▁komis
+- 抚
+- ごとに
+- ▁resurrection
+- 杠
+- ▁Mehrheit
+- ▁squadron
+- frica
+- 合理
+- 農家
+- sensibil
+- ទ
+- ▁структур
+- ▁бүтэ
+- 引き上げ
+- ▁Innen
+- нікаў
+- 加工
+- ▁производ
+- 单身
+- 呈现
+- ▁Schnee
+- ▁برنامه
+- ▁gleaming
+- 応え
+- ▁creeping
+- ▁ligne
+- ▁працу
+- ▁corri
+- เส
+- 現れ
+- ▁перш
+- очки
+- ▁genutzt
+- 皆様
+- 総裁
+- ▁robinson
+- ▁часов
+- нена
+- ▁kettle
+- 诗人
+- شاهد
+- äß
+- ▁muerte
+- ▁fake
+- ▁société
+- ▁микрорайон
+- 밥
+- ブラジル
+- ▁desirous
+- 一个男人
+- ▁spectrum
+- springen
+- 孙子
+- ▁सम
+- りません
+- 不太好
+- stāv
+- ▁prediction
+- したところ
+- ▁coil
+- ▁Fast
+- ゴン
+- コーチ
+- ▁proclaimed
+- ▁jugador
+- ▁توانم
+- ஊ
+- 体内
+- ▁interference
+- 四百
+- が増えて
+- 打つ
+- wickelt
+- ▁Мне
+- ▁kvin
+- дро
+- ▁crab
+- ▁Literatur
+- ▁shawl
+- ▁outfit
+- 倾听
+- 不良
+- ▁Chef
+- zuhalten
+- 普通に
+- ▁そんな中
+- ▁groaned
+- の方は
+- 艇
+- ▁Petr
+- ▁Wach
+- bole
+- еньки
+- 难受
+- 発信
+- 及其
+- ▁scatter
+- ▁drifted
+- ▁rejoicing
+- ▁друзья
+- ▁karakter
+- 是一件
+- assemble
+- を進める
+- でしょうね
+- ▁Befehl
+- 直前
+- జ
+- 亚马逊
+- 社交媒体
+- miseks
+- 设计师
+- ▁majoria
+- 很可能
+- 也许是
+- испол
+- 飘
+- ø
+- أصبح
+- ▁orthodox
+- ▁thieves
+- 鹿児島
+- 难以置信
+- 明天早上
+- jū
+- национальн
+- 热爱
+- ுகிறார்
+- ▁piety
+- あれは
+- 农场
+- AC
+- ▁مرا
+- ▁eleanor
+- ▁prêt
+- ▁کتاب
+- 大谷選手
+- 行事
+- ▁allusion
+- ▁yavuze
+- ▁trá
+- ▁суу
+- ▁gọi
+- 雀
+- மோ
+- 不然
+- かしら
+- 优秀的
+- blé
+- リアル
+- gewinn
+- 导致了
+- 辩
+- 旬
+- リオ
+- iddwa
+- ブランド
+- 内閣
+- ▁وقد
+- 册
+- ▁concession
+- 選択
+- ▁Cambridge
+- 込
+- لەر
+- ▁ĉefe
+- ▁dragging
+- ▁sparkling
+- ▁consistently
+- 吹き
+- ▁пакуль
+- holz
+- ドア
+- 做得很好
+- attend
+- ▁benjamin
+- ▁shifted
+- ▁Spur
+- ваюць
+- ▁hynny
+- ▁elevation
+- 軽く
+- ▁trình
+- ボタン
+- ganya
+- операци
+- ▁رسید
+- 又不是
+- ▁frenchman
+- 看着我
+- ▁suppressed
+- kijk
+- ▁perquè
+- ▁জন্য
+- ▁remarkably
+- aĝo
+- ▁ernest
+- 军官
+- ▁download
+- mette
+- ▁Microsoft
+- 沖
+- 勧
+- archiv
+- سرع
+- 一支
+- ひとつ
+- ▁цаг
+- dessus
+- 当前
+- 释
+- wendung
+- されたのは
+- 意志
+- に近い
+- 这是什么
+- スペース
+- ▁ruling
+- كۈ
+- genomen
+- ▁malheur
+- سلام
+- ▁выборы
+- 県内
+- ▁australische
+- ▁untersuchen
+- 鎌倉
+- 促进
+- ▁Geschenk
+- 诊断
+- ▁jeanne
+- waż
+- ▁groom
+- を受ける
+- ▁lettre
+- ▁adjacent
+- 砖
+- 挟
+- эль
+- ▁presque
+- 远远
+- 地理
+- 的感受
+- ▁Eric
+- читыва
+- concili
+- ▁basil
+- 配信
+- ▁desenvolup
+- 桐
+- 縫
+- 跑到
+- 同じように
+- ▁freuen
+- 諦め
+- 雨雲
+- ▁selben
+- لج
+- 三次
+- 平方
+- ▁vaig
+- ▁Social
+- カナダ
+- ▁москве
+- 定位
+- すり
+- ▁getrennt
+- bbling
+- ▁syr
+- ▁integrity
+- UN
+- پە
+- エース
+- ▁Verbraucher
+- 舎
+- ▁caracter
+- 見てみますと
+- விய
+- 听过
+- 谁能
+- 深度
+- 术语
+- と思うので
+- 消除
+- 撑
+- ▁şimdi
+- ▁savoir
+- 代わりに
+- حصل
+- ▁Medikamente
+- やっと
+- ຫຼ
+- を獲得
+- ▁pitiful
+- ▁легко
+- ▁besiege
+- 有趣的是
+- 集合
+- generation
+- ▁publisher
+- жым
+- ಡ
+- 穆斯林
+- ▁declaring
+- ビジネス
+- ▁elkaar
+- ▁visibility
+- 争い
+- ▁scary
+- 慢点
+- ▁leiten
+- って言った
+- 我也不知道
+- ড়ি
+- ▁westward
+- ▁repress
+- ▁fehlt
+- ृ
+- ▁installed
+- ▁сожалению
+- ▁언니
+- 雇佣
+- ▁repos
+- ▁парк
+- ▁accuse
+- سپ
+- みたいな感じ
+- 飛行
+- 阿里
+- ▁demonstra
+- ▁ridicule
+- ▁மிகவும்
+- 脑袋
+- ▁Company
+- চে
+- ▁Senator
+- AT
+- ▁veranda
+- 征服
+- 布里
+- يَ
+- 丈
+- ▁சேர்
+- 崇拜
+- ivamente
+- ▁Water
+- ▁glimmer
+- していること
+- II
+- 衛
+- 喜剧
+- 手紙
+- ▁집에
+- ējā
+- ▁Block
+- ▁väl
+- undneunzig
+- 詞
+- ▁слов
+- ▁Kalifornien
+- ει
+- haza
+- 趣
+- ▁Basis
+- ▁Cela
+- قۇ
+- 动议
+- 是多么
+- やろう
+- ▁neighboring
+- ▁Hast
+- алда
+- вание
+- どこまで
+- ▁lavora
+- ▁erstellt
+- ▁кеше
+- ▁Perspektive
+- ▁cualquier
+- ▁chemistry
+- ліз
+- ▁inherited
+- もち
+- ▁surge
+- 消費
+- ώ
+- ▁erforderlich
+- 須
+- ▁обнаруж
+- ▁descending
+- avons
+- mbri
+- ▁televisi
+- ▁europäischen
+- ▁Está
+- ▁rocket
+- druž
+- ခ
+- ▁맨날
+- 従業員
+- ▁среди
+- غو
+- 穿上
+- ▁phụ
+- 任何一个
+- ぱ
+- вших
+- ▁wizard
+- レシピ
+- 谁知道
+- ▁batteries
+- historie
+- ▁laquelle
+- ▁svět
+- тыя
+- 질
+- ▁hợp
+- ▁tiếp
+- ▁نزدیک
+- スイーツ
+- ▁vermeiden
+- ▁striving
+- ▁aufmerksam
+- 上课
+- 機械
+- ▁empower
+- 決して
+- 开门
+- ▁Gerechtigkeit
+- ▁힘들
+- 療養
+- ▁Three
+- existent
+- abhängig
+- ▁чём
+- ▁peggy
+- пресс
+- ވެ
+- ▁systematic
+- ▁než
+- ▁simplement
+- ▁verwirrt
+- せっかく
+- されてる
+- 赞助
+- 预防
+- ▁två
+- 赋予
+- ▁существует
+- ▁utility
+- ▁Wesen
+- soever
+- ▁Alors
+- 業務
+- 爪
+- ▁beschrieben
+- ▁supernatural
+- ▁فإن
+- ▁analyze
+- ▁rezult
+- ອນ
+- 流动
+- ▁listeners
+- ▁futur
+- Liber
+- 残疾
+- 厌
+- 험
+- PA
+- 輸入
+- يص
+- ▁automatisch
+- 欧米
+- ▁Belgi
+- 引导
+- わけですよね
+- ▁영화
+- ▁skilful
+- ▁называется
+- ēl
+- دۇ
+- sparen
+- 歌词
+- ▁Schließlich
+- 芒
+- 分かんない
+- ப்படுத்த
+- ▁Republic
+- 公表
+- ীয়
+- 几分钟
+- 晚餐
+- 桌子上
+- schlüssel
+- скія
+- ▁Цяпер
+- ▁музыка
+- ▁Russland
+- ▁найд
+- ▁merciful
+- 迹象
+- ▁printing
+- ▁growled
+- ▁речь
+- 気分
+- 可能性があります
+- ▁plunge
+- ▁பெண்
+- ▁melhor
+- ání
+- ▁Fä
+- ▁transit
+- ▁гэсэн
+- ▁Gesamt
+- ándose
+- artisan
+- ▁Clark
+- кле
+- What
+- 一个巨大的
+- ▁Dreh
+- raad
+- 给出
+- ▁perished
+- ▁overthrow
+- ▁mantle
+- rides
+- studie
+- 为什么会
+- ▁lingered
+- 完整
+- ▁scroll
+- 底部
+- ▁Treppe
+- provi
+- 前往
+- 福利
+- 钢琴
+- صنع
+- ▁подход
+- だったので
+- 遗产
+- ▁infirm
+- පු
+- するのか
+- ▁Geschwindigkeit
+- ▁contemplation
+- ▁loneliness
+- 芯
+- 토
+- 据说
+- acca
+- நா
+- ▁koran
+- ▁schimb
+- ңә
+- ▁Тут
+- ▁plug
+- ▁torre
+- decorat
+- preservation
+- ސް
+- ▁трудно
+- を巡る
+- ائية
+- 黒い
+- ▁Mario
+- ▁handled
+- ▁biological
+- 场所
+- ▁تأ
+- 哲
+- ▁brightness
+- យ
+- ▁October
+- ▁adjoining
+- ▁irgendwann
+- 虹
+- ▁schafft
+- 问一下
+- ▁ethics
+- 微妙
+- 総合
+- reißen
+- ▁показыва
+- ▁sparkle
+- ▁faci
+- ▁Brasil
+- marsch
+- ▁entrepreneur
+- ucci
+- 出会い
+- плеч
+- 목
+- ▁Schauspieler
+- ▁largo
+- 评价
+- ▁sniff
+- ▁recibi
+- こんな感じ
+- 村庄
+- губ
+- ▁bölge
+- ▁چطور
+- ▁avenge
+- 派遣
+- ▁đồng
+- ▁новая
+- ▁achtzig
+- 齢
+- ▁practised
+- 老百姓
+- ▁plaça
+- があれば
+- 座位
+- 安倍
+- 车辆
+- 良く
+- ▁друга
+- ▁дежур
+- ौ
+- 自動
+- ▁وهو
+- ▁fué
+- legation
+- あした
+- ▁pip
+- 디
+- ▁intimacy
+- ▁пункт
+- ▁ecosystem
+- 网站上
+- ложить
+- 凭什么
+- ▁aldı
+- ▁wambaye
+- フィー
+- 発電
+- ▁regent
+- になってくる
+- 飛行機
+- stunden
+- ▁Emma
+- 回事
+- 比尔
+- 颠
+- ▁Gouverneur
+- ▁delicacy
+- ▁обсужда
+- 絞
+- ▁removal
+- 歩く
+- getrieben
+- ▁basta
+- ▁ясно
+- ковская
+- ▁sasa
+- ベン
+- 职位
+- 盈
+- ▁furchtbar
+- 致力于
+- 繁荣
+- 整備
+- ▁übernehmen
+- ▁අප
+- われわれ
+- 小型
+- 深夜
+- ▁Manche
+- 児童
+- 仕掛け
+- ▁একটি
+- ▁Bush
+- ҡан
+- ▁alleged
+- 走路
+- を紹介
+- ▁promet
+- େ
+- ▁دختر
+- ホームページ
+- 約束
+- யு
+- を起こし
+- larına
+- prinz
+- ▁بأ
+- 奋
+- 四月
+- 你会发现
+- 福島
+- 墨西哥
+- ▁latitude
+- ▁እን
+- 我就想
+- ▁horseback
+- ரெ
+- рская
+- っていきます
+- 哟
+- ▁psychische
+- ▁dainty
+- ▁inquiring
+- ▁başladı
+- ▁خاص
+- 하니까
+- 好听的
+- ▁chiama
+- ▁knocking
+- ▁carelessly
+- 达成
+- ▁понят
+- ▁precept
+- േ
+- 债务
+- ▁schlimmer
+- 最重要的是
+- 姥
+- 枯
+- 見た目
+- 转身
+- ħ
+- 破壊
+- ▁Wichtig
+- 典型的
+- ▁lawful
+- ▁caravan
+- 来找我
+- ▁самым
+- rühm
+- 凍
+- 描いた
+- ޅ
+- 新規感染者
+- 依頼
+- 不算
+- ▁forsake
+- 密切
+- schieß
+- ▁semana
+- kuti
+- ীর
+- ▁geschafft
+- ▁président
+- ▁socrates
+- 頑張り
+- ▁malice
+- က်
+- ▁Million
+- ▁revolutionary
+- моў
+- ▁tavern
+- 島さん
+- чала
+- ▁Sco
+- څ
+- ▁Griff
+- の様子を
+- ▁fantastisch
+- ▁максим
+- ▁verlangen
+- ▁verdict
+- キャンプ
+- を抱え
+- 時間帯
+- ▁너가
+- ื
+- ペア
+- ▁шоссе
+- 男の子
+- ▁Muslim
+- 抑
+- ▁Dazu
+- моло
+- 搁
+- 秩序
+- ▁Schluss
+- берег
+- ▁რომ
+- ▁поднял
+- ▁athlete
+- 慢慢地
+- pharma
+- ▁bobby
+- entreprise
+- すき
+- ▁könne
+- ▁realizing
+- 交换
+- ▁metaphor
+- ▁Investor
+- ્ય
+- ▁nadie
+- たいと思います
+- ▁stitch
+- ▁dimly
+- คร
+- 即便
+- 一応
+- ▁pedra
+- ▁interface
+- ▁قىل
+- ància
+- 把它放在
+- アーティスト
+- ▁wußte
+- spitze
+- 很喜欢
+- って思って
+- 艘
+- კა
+- を訴え
+- ▁Umugabo
+- ▁shattered
+- garuka
+- 回复
+- saison
+- 友人
+- biza
+- ▁resign
+- ▁renewal
+- ছেন
+- を止め
+- ▁Dach
+- 半島
+- ▁removing
+- 是什么样子
+- 有人说
+- ビア
+- 会話
+- 学位
+- ▁racing
+- 哨
+- ▁секрет
+- ▁pubblic
+- скры
+- ▁아직
+- geschnitten
+- angwa
+- 价值观
+- czą
+- 有这样的
+- ウム
+- باب
+- өс
+- ホント
+- ▁cynnwys
+- ▁restructuring
+- 共和国
+- 亚洲
+- ▁metod
+- ▁نفر
+- ▁thích
+- ビール
+- zieh
+- 業界
+- dringen
+- niedrig
+- と見られる
+- ▁qualche
+- 失礼
+- ฟ
+- Ž
+- ▁зүйл
+- ▁measurement
+- фарм
+- เร
+- ਲ
+- ▁гораздо
+- 鹏
+- ▁ہے
+- sabye
+- īga
+- ходзіць
+- öffentlich
+- 暑い
+- ▁roland
+- ▁tariff
+- 皆さんも
+- ▁我想听
+- న్
+- 練
+- 冤
+- 阿拉伯
+- 幻灯片
+- ▁massacre
+- 봤어
+- ▁Beine
+- سوف
+- ▁kritisch
+- ▁frock
+- ▁разных
+- ▁Mama
+- സ
+- 拾
+- 録
+- ▁Đó
+- ▁Betracht
+- 同伴
+- 使命
+- ▁consisting
+- бло
+- ▁daddy
+- ▁matrimoni
+- プログラム
+- 明智
+- 真诚
+- ▁rotten
+- ▁convertir
+- ▁смерт
+- 墙上
+- 服用
+- appelle
+- ▁twain
+- ▁Dunkelheit
+- ▁Identität
+- ▁pharaoh
+- ▁structural
+- 겨
+- ธ
+- سط
+- ▁будуць
+- 多年来
+- やってみ
+- ▁Arthur
+- 发行
+- 童年
+- 忘记了
+- ▁whim
+- æ
+- ▁என்பது
+- ▁quivering
+- 先制
+- 依靠
+- 那天晚上
+- тычна
+- 兔
+- kārt
+- stift
+- 感染者数
+- ▁алло
+- ▁влия
+- 嫌疑人
+- ▁olympi
+- ▁помню
+- ▁توانید
+- ▁keenly
+- ▁Pflege
+- กับ
+- ▁около
+- 広げ
+- bido
+- ▁Später
+- アナウンサー
+- 린
+- ছিলেন
+- ટ
+- ▁supplier
+- ▁geistige
+- 解散
+- ▁нашем
+- 深く
+- わかった
+- Direct
+- писать
+- ▁ўсе
+- ▁stimulate
+- 六点
+- 稽
+- おすすめ
+- 拝
+- әү
+- 埃及
+- ▁avea
+- ▁quoth
+- ▁принял
+- simila
+- ▁posible
+- 추
+- ▁città
+- 收获
+- ▁Pflicht
+- ▁Sehr
+- ▁constable
+- gaciro
+- 通道
+- ▁jasper
+- 된
+- ۇن
+- ▁Avenue
+- ▁hurled
+- ▁چهار
+- ıdır
+- ▁пасля
+- сцю
+- ▁falsehood
+- 好消息
+- ▁Golf
+- 斯顿
+- ▁boundary
+- 恰
+- ৌ
+- β
+- ▁beberapa
+- 銭
+- uɣal
+- ▁حو
+- ▁stripped
+- ałem
+- சூ
+- ▁Kommentare
+- ▁countless
+- გი
+- 下がり
+- għ
+- ▁있다
+- 祈
+- ▁obedient
+- ▁precedent
+- ▁dialect
+- ště
+- を目指して
+- ▁charley
+- веж
+- に警戒
+- どうなって
+- 玄
+- 얘
+- ગ
+- ▁Innovation
+- ▁venerable
+- ▁Schaden
+- గా
+- ▁deployment
+- ▁discharged
+- ▁bribe
+- ▁choked
+- เด
+- ницы
+- ▁Бер
+- ▁shareholder
+- ▁irresistible
+- 색
+- ▁ertragen
+- ▁دانش
+- 猜测
+- håll
+- ▁skr
+- ▁начала
+- jú
+- حاول
+- ិ
+- ▁شدند
+- してくれた
+- ▁kombin
+- درس
+- ▁cuanto
+- ▁fakt
+- ▁loaf
+- 후
+- 予測
+- 治愈
+- 细菌
+- escence
+- ▁Diana
+- 辰
+- ▁ermöglichen
+- ▁области
+- ▁apprehend
+- ▁sincerity
+- ▁Marine
+- ▁conduc
+- ▁глаз
+- मि
+- 字母
+- 午前中
+- 不止
+- ▁atrodas
+- ▁встрет
+- ▁coneix
+- リップ
+- europäische
+- träger
+- 日期
+- ▁splendour
+- 準決勝
+- ▁Kauf
+- ▁equipped
+- 伊朗
+- ▁Verfassung
+- ▁racial
+- ▁wistful
+- يست
+- اقتصاد
+- ▁begrijp
+- ▁überprüfen
+- 挣扎
+- ▁вижу
+- 聊聊
+- ▁greet
+- 躁
+- ק
+- 创伤
+- ▁ведаю
+- 旅程
+- ▁llegar
+- етесь
+- ▁mbili
+- 寒い
+- ▁calor
+- ▁conoce
+- ▁worte
+- undsiebzig
+- ▁stumbled
+- 剣
+- ▁займа
+- 楼梯
+- 市长
+- 低下
+- ▁вспомни
+- ▁holmes
+- 未知
+- ことになります
+- ډ
+- 辨
+- ▁contemptuous
+- '......'
+- ▁darted
+- zustand
+- ▁грани
+- ビデオ
+- ▁soothe
+- 짜
+- 创始人
+- ▁imprisonment
+- ▁intensely
+- 在乎
+- leɣ
+- traction
+- ificació
+- fellow
+- ంది
+- foli
+- 対決
+- بِ
+- 长官
+- 머
+- ▁Ankaŭ
+- 纯粹
+- ▁unmittelbar
+- ▁Ursache
+- овское
+- ▁granite
+- ▁avem
+- 一生中
+- گەن
+- анс
+- ▁epic
+- ▁virtually
+- ▁tylko
+- を防ぐ
+- ▁podia
+- ▁snatch
+- 替代
+- 費用
+- 购物
+- 組み合わせ
+- 長崎
+- ▁لذا
+- 더
+- ▁واقعا
+- ▁maior
+- ▁ieder
+- をはじめ
+- 点钟
+- ელ
+- ▁Kontext
+- ▁Verbesserung
+- サポート
+- geleitet
+- ތަ
+- ▁wickedness
+- ▁kugirango
+- 装饰
+- ▁azul
+- コロナ禍
+- 集体
+- ▁Null
+- Europe
+- 幹部
+- ▁Umfrage
+- 澄
+- স্থা
+- ▁cafe
+- 展开
+- пак
+- ▁приходит
+- 携
+- 教えてくれ
+- 晚安
+- 夫妇
+- εί
+- 如果不是
+- 谈过
+- ▁controversy
+- ▁nyingi
+- ▁lần
+- まとめて
+- につながる
+- ようになりました
+- ▁beeinflusst
+- ▁Italien
+- ▁classical
+- スリー
+- bilidad
+- нув
+- ピーク
+- ▁erleben
+- と述べ
+- ▁humid
+- 海军
+- brennen
+- ▁henceforth
+- ▁گرفته
+- 栄養
+- йшоў
+- ▁famine
+- 之所以
+- ▁improvis
+- жә
+- ▁المست
+- ▁burial
+- ів
+- ешься
+- 冷たい
+- 实话
+- ▁Fou
+- ▁przez
+- ▁Mathematik
+- ▁furnace
+- ▁ອື
+- 舞蹈
+- ▁Abteilung
+- ḥem
+- ▁Fair
+- ▁avut
+- ▁dringend
+- ▁Lincoln
+- ▁вариант
+- ▁bemerkenswert
+- 困扰
+- ంద
+- ▁fertile
+- 另一边
+- ▁sangat
+- 基金会
+- 注文
+- між
+- ▁Sagen
+- 告诉她
+- ಹ
+- ▁instinctively
+- อย่าง
+- 恳求
+- 製造
+- ▁gratify
+- ぼく
+- ▁grit
+- ▁Anderson
+- ▁turtle
+- ▁unusually
+- 赢了
+- 会导致
+- ▁Karl
+- ▁Wetter
+- gültig
+- ▁römische
+- 摄影
+- 吃完
+- ▁declara
+- '250'
+- 团结
+- 每当
+- 知ってる
+- 酵
+- ▁Kapital
+- 职业生涯
+- 重症化
+- вернуть
+- ambaye
+- 洪水
+- observa
+- ွ
+- スペシャル
+- ▁equation
+- 恭喜
+- ▁инде
+- 宪法
+- ▁northwest
+- ▁Müll
+- ▁oyster
+- ▁devons
+- 几年前
+- ந்தது
+- ▁Verteidigung
+- ミー
+- ▁Details
+- ▁gewann
+- 蛋糕
+- ▁Kleid
+- つながって
+- ▁combina
+- 被迫
+- ▁geldi
+- ▁confronted
+- 僵
+- 季节
+- ▁그건
+- ▁soothing
+- ത്ത
+- ▁хэрэг
+- 牛肉
+- ▁papel
+- ▁Meeres
+- ▁Fox
+- ▁Darüber
+- 偏见
+- メール
+- お茶
+- 卡尔
+- MA
+- Tool
+- 扮
+- ▁crise
+- ▁efficiencies
+- ▁participants
+- ▁refusal
+- ▁알바
+- ņēm
+- ▁여기
+- BM
+- école
+- ▁upgrade
+- ▁superb
+- ते
+- 言わ
+- ▁черт
+- ▁господин
+- ▁fireplace
+- ▁Campus
+- ▁Hollywood
+- ▁experiencing
+- 震度
+- ▁никого
+- ▁системы
+- 可靠
+- klima
+- 帽
+- 誕生日
+- ▁видим
+- ブルー
+- 惯
+- ▁biology
+- ▁annoyance
+- गा
+- 回去吧
+- に入れて
+- vogel
+- ▁современн
+- ▁Wolf
+- சோ
+- 失踪
+- ▁spill
+- 埃尔
+- 这让我
+- 大众
+- チュ
+- ▁ignored
+- 变得更加
+- ▁beforehand
+- ై
+- ▁anticipation
+- ▁imprisoned
+- 伴侣
+- トランプ
+- ▁ilgili
+- ▁பண்ண
+- ▁maggior
+- ▁hydro
+- ▁unexpectedly
+- ▁opportun
+- ▁jî
+- 肢
+- ባ
+- 孫
+- ▁entscheidend
+- ▁விளையாட
+- ▁salud
+- 英語
+- ▁смысл
+- কো
+- ▁fui
+- ▁pike
+- こんなこと
+- 分野
+- 艳
+- ը
+- ▁staggered
+- ▁League
+- னால்
+- 不幸的是
+- Datei
+- mdash
+- ▁cedar
+- 部隊
+- おうち
+- ▁biraz
+- 慰
+- 拥
+- Community
+- ▁gouvernement
+- 暮らす
+- ▁drog
+- ▁இசை
+- 打印
+- ▁turkish
+- 过程当中
+- ▁кел
+- М
+- 这是关于
+- ▁barber
+- ▁kinh
+- ▁bezeichnen
+- 松本
+- ▁subordinate
+- 嘲笑
+- まれた
+- 包围
+- 非法
+- 買い物
+- Ɛ
+- ▁pequeño
+- 忽略
+- 猛烈
+- kundig
+- ▁бич
+- ▁stockings
+- 終わって
+- бежал
+- 王爷
+- าร
+- ▁அல்லது
+- ▁moore
+- 跟你们
+- ▁인제
+- ▁Kiel
+- ▁lúc
+- ▁apology
+- ロシア側
+- ▁eĉ
+- が出ています
+- 措
+- 昂
+- ແລ້ວ
+- ▁phantom
+- ▁població
+- 吉尔
+- わかって
+- getreten
+- ▁exceeding
+- ▁Management
+- ▁Şimdi
+- 虚拟
+- 这段时间
+- ▁communion
+- っきり
+- 植え
+- 这个过程
+- ુ
+- お伝えしました
+- ▁встреч
+- ▁besuchte
+- ৰে
+- したのが
+- が発表され
+- 胀
+- ▁remnant
+- したのです
+- нис
+- mıştır
+- ▁شدن
+- ▁colleague
+- 抑制
+- 润
+- ▁президента
+- 環
+- 伞
+- ▁tecnologia
+- ▁последние
+- ▁restoration
+- あらゆる
+- まいります
+- ▁qualcosa
+- fleck
+- ▁بیمار
+- ▁vegetation
+- ▁distracted
+- ▁hamlet
+- თი
+- schneid
+- satisfied
+- నే
+- கொள்ள
+- bwenge
+- ▁எனக்கு
+- 玫瑰
+- なければいけない
+- だからこそ
+- 継続
+- ▁aufgewachsen
+- ▁explicit
+- ული
+- ▁nightmare
+- komeje
+- 书籍
+- 려고
+- burton
+- bär
+- ▁chama
+- girl
+- பிடி
+- 深圳
+- ▁Küche
+- 实力
+- govor
+- 努
+- ▁собственн
+- ▁або
+- 俄
+- ▁affliction
+- ▁chancellor
+- ▁suivant
+- ▁Beide
+- 輸
+- 电池
+- стоян
+- ▁babylon
+- ▁Ça
+- こともある
+- ▁kız
+- ▁scoundrel
+- ▁vorbereitet
+- ▁apologize
+- 折磨
+- ▁pierced
+- ساعد
+- ▁protector
+- ▁lydia
+- ▁connais
+- ▁actress
+- 患有
+- ▁tromp
+- ▁rejoin
+- ▁Kenn
+- ▁quién
+- 蕾
+- 격
+- わかりました
+- を含め
+- 反馈
+- ▁grandeur
+- ▁maud
+- ▁Pfund
+- 几周
+- 格雷
+- しません
+- ivität
+- ▁brace
+- ▁trọng
+- 루
+- tempo
+- گذاری
+- ▁পরি
+- liegt
+- ▁Bang
+- 婷
+- ▁Vietnam
+- ▁cœur
+- ▁doppelt
+- へえ
+- 言ってる
+- ▁już
+- 收到了
+- 幽
+- ▁nötig
+- ▁четвёртая
+- 민
+- ים
+- 介護
+- ▁людзі
+- گران
+- ங்
+- 家具
+- 動いて
+- ▁isaac
+- ▁першы
+- সব
+- RO
+- 坐下来
+- ▁Investition
+- ▁verzweifelt
+- ▁Maschinen
+- ▁솔직히
+- origen
+- だけではなく
+- ▁خب
+- 遭遇
+- ▁crave
+- 更快
+- ▁effi
+- 大爷
+- 黙
+- ▁Canadian
+- ▁aufgeregt
+- 绅士
+- pathie
+- 布朗
+- ▁devient
+- 返回
+- ▁ooit
+- 优秀
+- ▁Protest
+- ▁predecessor
+- 預
+- 티
+- ▁Stärke
+- ▁dirige
+- ▁sáng
+- ることができます
+- ▁бывает
+- ▁faisait
+- يقة
+- 所以如果
+- undfünfzig
+- 尔顿
+- 彦
+- built
+- ้น
+- держать
+- ▁хамт
+- ▁prodig
+- යෙන්
+- ια
+- 椒
+- ▁tyranny
+- ▁않아
+- ▁evolve
+- ▁proprio
+- ▁없는
+- ▁bombard
+- ▁Ohio
+- ырға
+- 역
+- gespräch
+- ▁хамгийн
+- ▁мистер
+- 困難
+- ▁Thu
+- ほかにも
+- therapie
+- ▁revolu
+- バイク
+- ▁finanzielle
+- 辩护
+- ▁scrub
+- ▁judging
+- ▁freue
+- ▁крем
+- wash
+- 来到这里
+- 逃走
+- ▁última
+- ▁انسان
+- ▁Lä
+- ▁müde
+- 加盟
+- ணை
+- 西安
+- 土著
+- ▁ministre
+- 役割
+- ▁geholfen
+- ▁hết
+- ▁Madrid
+- ▁Stuhl
+- 疑問
+- 昨天晚上
+- 我的朋友
+- 跑步
+- ▁баб
+- corp
+- گشت
+- ▁knapp
+- 要素
+- Restaurant
+- ▁kürzlich
+- ▁voluntary
+- ▁член
+- ▁angst
+- ▁ubwa
+- ▁wartete
+- ▁inhabited
+- 分ほど
+- 汤姆
+- ▁трав
+- と見られ
+- 初め
+- গ্র
+- ตร
+- ▁پسر
+- ▁woher
+- koop
+- technolog
+- stelling
+- 巢
+- ▁Michigan
+- ▁hamilton
+- 浑
+- iPhone
+- ▁gekauft
+- ▁아닌
+- ▁девочк
+- ▁министр
+- озер
+- ▁boundaries
+- ▁exploring
+- シャン
+- фар
+- ▁repel
+- バンド
+- ▁volont
+- ▁позвони
+- ▁employee
+- ▁trobar
+- ▁paddle
+- 黛
+- ▁обраща
+- ▁identi
+- ▁Einkommen
+- ▁radiation
+- راض
+- 动手
+- ▁chú
+- stehenden
+- 递
+- ▁mcc
+- 收看
+- ▁Clinton
+- ▁Vorsitzende
+- 运输
+- '900'
+- ▁sincerely
+- ▁Küste
+- matur
+- 取る
+- 던데
+- ▁specialist
+- ケン
+- 搬到
+- ▁voet
+- zulassen
+- ▁ankoraŭ
+- ▁grinned
+- ▁municipi
+- ▁zweimal
+- ▁үҙе
+- 抗議
+- ▁gorge
+- ▁имею
+- ▁Weltkrieg
+- ируют
+- ▁Patri
+- ▁settlers
+- ▁بچه
+- 傅
+- قليل
+- દ
+- ▁Dimarts
+- ▁오늘
+- ▁공부
+- ▁Вось
+- ▁crawled
+- ▁suspend
+- ▁daudz
+- 申し上げ
+- ▁durfte
+- ▁brake
+- チン
+- ぽ
+- ▁Master
+- ▁certificate
+- ▁страшно
+- ▁statute
+- ▁Kaiser
+- ▁Beau
+- 有名な
+- ҟ
+- ẹ
+- ▁profi
+- ▁popularity
+- 饶
+- ▁repetition
+- ▁sechzehn
+- effizient
+- 差距
+- ▁cobert
+- 突出
+- 选手
+- ▁bleeding
+- ▁рабіць
+- فرد
+- 做些什么
+- ▁patrol
+- 升级
+- 九月
+- ▁однако
+- 味わい
+- 微信公众号
+- 끼
+- ▁얼마
+- ▁фильм
+- ▁drilling
+- ström
+- ▁মই
+- 話し合
+- ▁plateau
+- ▁komuni
+- 今天早上
+- ▁tumor
+- stritten
+- 折り
+- 当たり前
+- 時刻
+- ▁inevitably
+- ▁kontraŭ
+- ▁liệu
+- 巣
+- 迷惑
+- ▁fascinated
+- ючы
+- 発展
+- 解答
+- ▁Doktor
+- 지만
+- ▁wolle
+- ▁gerçekten
+- hagi
+- brett
+- 寨
+- ▁انتخاب
+- ▁battalion
+- ▁Európ
+- 岸田総理大臣
+- ▁средств
+- 巴黎
+- 站着
+- 有很大的
+- 越多
+- ▁sigui
+- ▁жар
+- 认为这是
+- ▁Schön
+- ってきます
+- ▁natuurlijk
+- église
+- 贈
+- ރި
+- ▁specio
+- ▁yankee
+- 言われる
+- 報じ
+- ▁autour
+- ▁Estados
+- 也是如此
+- сцен
+- IM
+- ▁Pont
+- வர்கள்
+- ▁Notiz
+- 感知
+- 如果你愿意
+- ỡ
+- ▁어떤
+- グリーン
+- ▁специальн
+- ▁2010
+- ▁delegate
+- ▁Depression
+- ▁wallace
+- 你能不能
+- 神社
+- 赞成
+- ībā
+- trekken
+- ▁Stone
+- ▁cielo
+- 僧
+- ▁Haufen
+- بەر
+- ▁nasty
+- ▁placid
+- ▁abbot
+- ▁имел
+- 这么好
+- ▁erfordert
+- ▁отец
+- ▁bên
+- ▁trifling
+- ▁angemessen
+- 慈善
+- ▁legislative
+- كَ
+- ṭṭ
+- 針
+- ▁Verkauf
+- ▁Ле
+- 江戸
+- 嶋
+- ථ
+- パフォーマンス
+- 猩
+- ▁взять
+- 尽く
+- 辉
+- ▁медведев
+- ▁transcend
+- ங்களுக்கு
+- 局面
+- ▁glove
+- ▁приехал
+- ▁violation
+- 昨年
+- 脇
+- ▁Health
+- ক্র
+- 企业家
+- ҙар
+- 住房
+- lendi
+- ▁mound
+- ▁gestorben
+- ▁ungewöhnlich
+- ▁mouvement
+- ベー
+- 無事
+- 防御
+- ▁elementary
+- ▁kneeling
+- を広げ
+- öffne
+- 七个
+- важа
+- ▁Pul
+- ▁далеко
+- 在这一点上
+- ▁reconstruct
+- ▁одном
+- 废话
+- рина
+- ▁opini
+- 静岡県
+- ▁стане
+- 指标
+- 狐狸
+- ד
+- ▁Ariko
+- ▁Global
+- ▁pretence
+- 轨道
+- ▁magnetic
+- ▁gravit
+- CM
+- 楽しく
+- ▁столько
+- ▁refusing
+- きちんと
+- 污染
+- ▁demokratische
+- ▁принципе
+- 布拉
+- ▁Dennoch
+- 确切
+- ▁ivory
+- ▁Bauern
+- ▁Zucker
+- やろ
+- ▁frente
+- ▁сельск
+- ▁petrol
+- 影响力
+- 差点
+- 海底
+- antrag
+- ▁Bundesstaat
+- яўля
+- ▁роман
+- রো
+- ▁probablement
+- ▁siebzig
+- ▁antonio
+- guna
+- キック
+- ▁strove
+- röst
+- ないこと
+- ▁hiç
+- litten
+- ▁начинает
+- ▁Führer
+- ▁introducing
+- ▁miglior
+- ですもんね
+- ▁apollo
+- ▁relaxed
+- ▁Го
+- sinzi
+- kunga
+- 手臂
+- ▁домой
+- ▁glitter
+- 老太太
+- ▁dodge
+- ▁бюджет
+- ▁Fakten
+- گار
+- activitat
+- ▁parecía
+- ▁cradle
+- ▁дуб
+- добав
+- خوان
+- ▁viņu
+- prej
+- 模仿
+- ▁bazı
+- коммун
+- 写道
+- ▁treachery
+- ▁vị
+- ▁Looking
+- espècie
+- ▁거기서
+- కా
+- ના
+- 性质
+- LA
+- 毅
+- ▁праблем
+- ▁exclaim
+- ▁aufhören
+- 异常
+- 到最后
+- ▁courtyard
+- 勢力
+- ▁prophecy
+- ▁recipe
+- ▁doomed
+- 优雅
+- 迈克尔
+- ▁Ды
+- ▁furiously
+- ▁sicherstellen
+- 尾巴
+- tempered
+- 这样的事情
+- ▁thức
+- 抱着
+- ۋە
+- ▁американск
+- пэўн
+- ▁hội
+- ▁Jordan
+- 人工智能
+- ▁trenches
+- レーン
+- ზე
+- ▁bridle
+- ▁suspense
+- ▁Schriftsteller
+- 匹配
+- ▁binding
+- プリ
+- ▁heutige
+- 感動
+- ▁depict
+- きれいな
+- dolf
+- ▁Direktor
+- ▁benötigt
+- 등
+- ▁missouri
+- ▁paradox
+- ▁warehouse
+- ▁Johann
+- forschung
+- やったら
+- いかに
+- 发射
+- ▁compel
+- ▁massachusetts
+- ▁وهذا
+- ▁conosc
+- ▁entschlossen
+- ▁gaunt
+- 仕上げ
+- 徴
+- ơi
+- ▁дети
+- ▁dikontrak
+- ▁néixer
+- ▁долларов
+- 塑造
+- ▁uplift
+- لىرى
+- 教徒
+- 任何地方
+- ▁камер
+- chamber
+- ▁marilla
+- ▁Stell
+- ▁haughty
+- ▁sledge
+- ▁facilit
+- ▁каш
+- 百五十
+- 在那儿
+- sicherheit
+- 案内
+- 久保
+- ັ້ນ
+- වල
+- ▁monastery
+- ▁Überzeugung
+- ▁crooked
+- ▁эксперт
+- ▁intolerable
+- 掃除
+- ▁холод
+- 弗雷
+- вшие
+- ҟа
+- فعال
+- 我第一次
+- 大楼
+- kirche
+- ព
+- ▁ammunition
+- ▁applaud
+- давал
+- ▁medicina
+- ▁schooner
+- ▁Christi
+- ▁orienta
+- 一体何
+- っぽい
+- 顺便说一句
+- ▁ເດ
+- ▁necessita
+- 代替
+- 浸水
+- 服从
+- ▁ethical
+- 苍
+- 言います
+- ▁многих
+- ▁وإ
+- حاضر
+- ▁говорите
+- ▁emphatic
+- 聞きました
+- 困境
+- ▁سے
+- 拠点
+- 不记得
+- いらっしゃい
+- 有什么事
+- arrell
+- 楠
+- ▁survival
+- ▁өмнө
+- 碰巧
+- lož
+- 抜き
+- ▁outbreak
+- streich
+- 任何其他
+- ▁держа
+- ▁plaster
+- 崔
+- ▁Если
+- キャプテン
+- 來
+- 皿
+- ▁хотелось
+- 步骤
+- ▁черн
+- ▁hagati
+- ▁surround
+- ▁Twa
+- ▁அதை
+- ▁Nachmittag
+- ▁baptism
+- ▁секунд
+- ごめんなさい
+- 决策
+- ▁reasonably
+- 介意
+- ▁eky
+- という状況
+- ▁anfing
+- 食べ物
+- ▁banyak
+- ▁injuries
+- ނަ
+- 失礼します
+- 病例
+- 甘い
+- тверд
+- ▁Fremde
+- ▁الذين
+- keneye
+- ▁zaidi
+- ▁ravine
+- ▁accommodate
+- 朴
+- ▁biscuit
+- 衆議院
+- ▁victorious
+- أخذ
+- ▁Großteil
+- そうなんですね
+- ▁augustus
+- ▁вызыва
+- 初戦
+- 能找到
+- ▁நீங்கள்
+- 二零零
+- mówi
+- 举起
+- 服务器
+- freiheit
+- structure
+- 神経
+- 妥
+- 信頼
+- ゴルフ
+- 经历过
+- 默默
+- ▁Creek
+- ▁aṭas
+- ▁Guerra
+- 宮崎
+- ▁siguiente
+- 兵器
+- ▁replica
+- 赔偿
+- ▁hiểu
+- 過去最多
+- 臂
+- ▁resol
+- ▁panting
+- жер
+- 时尚
+- मु
+- Qué
+- 涼
+- ▁illustrious
+- ▁indefinite
+- 厄
+- ▁bedeckt
+- ▁shrine
+- 潜水
+- ▁exig
+- ▁حتی
+- дзіць
+- ▁спин
+- 竞
+- рист
+- と比べて
+- 케
+- ▁preocupa
+- ▁preĝejo
+- ▁vẫn
+- ▁behaupten
+- яўляецца
+- ▁notamment
+- 運転手
+- ▁weariness
+- ▁rimwe
+- 吉田
+- そっか
+- ▁flint
+- 衷
+- 豹
+- ▁زۆر
+- ▁заметил
+- gypt
+- ▁Milch
+- 大人気
+- 很多时候
+- рожд
+- ▁второго
+- 卸
+- 祝你
+- мель
+- कु
+- 被告知
+- ▁correspondent
+- ▁propaga
+- 读到
+- 作战
+- 燃え
+- ▁우리가
+- ▁passionately
+- ▁För
+- ▁хор
+- 甚
+- 頂いて
+- ք
+- ▁bệnh
+- ▁offspring
+- ▁Ancak
+- トレーニング
+- この時期
+- 買う
+- 因为这是
+- 乗客
+- 强迫
+- 市長
+- ▁researchers
+- が行われました
+- freude
+- ▁гэтыя
+- ▁scenery
+- ignit
+- енько
+- 物品
+- 紅
+- ▁Original
+- 찍
+- hypno
+- ▁режим
+- ▁ahubwo
+- honneur
+- 行星
+- ▁imaginary
+- winkel
+- өгө
+- ▁ваша
+- ▁tâm
+- ޑ
+- ▁Président
+- 見てください
+- 奖励
+- ▁giống
+- ▁حيا
+- ▁clatter
+- ▁circulation
+- 調理
+- ございます
+- ▁பாட
+- 从哪里
+- 很酷
+- 对我说
+- ▁Urteil
+- ▁Entdeckung
+- ▁proclamation
+- 查询
+- ▁wireless
+- なと思いました
+- ▁deixar
+- ▁거는
+- مجموعة
+- rühren
+- 协调
+- 活発
+- schuh
+- რო
+- 른
+- そろそろ
+- 支撑
+- ▁아니면
+- 有足够的
+- 品质
+- になりそうです
+- لدى
+- 裕
+- ▁grammar
+- ▁lượng
+- ▁преступлени
+- 牛奶
+- ▁đường
+- ▁만나
+- ▁ricevis
+- ▁außerdem
+- ▁wholesale
+- 列車
+- ▁jupiter
+- 和我一起
+- ▁acabar
+- 液体
+- ▁있지
+- cyaha
+- 碎片
+- ▁crater
+- 十月
+- impuls
+- したあと
+- ▁elektron
+- 分ごろ
+- ある程度
+- 跟他说
+- ▁titre
+- ▁своими
+- ▁acuerdo
+- නම්
+- ▁бүх
+- いませんでした
+- 話す
+- 大切に
+- 認められ
+- ▁хотели
+- 放置
+- illard
+- Mobil
+- 그
+- ニュースをお伝えします
+- 监督
+- ētā
+- aardig
+- ▁discrimination
+- 延伸
+- รา
+- 流量
+- ▁considerat
+- 었는데
+- ▁pronto
+- 贷
+- 素材
+- ▁алексей
+- ▁caroline
+- 屁股
+- 辞职
+- 占据
+- 我不得不
+- ಗಳ
+- 很开心
+- eater
+- ▁Ahnung
+- ▁secular
+- 理念
+- 貴重な
+- ▁Abschnitt
+- ▁hiring
+- 寒気
+- ▁vigor
+- ▁fick
+- ▁decorated
+- ดี
+- 跟随
+- ▁español
+- ▁помочь
+- ▁entsprechend
+- 인가
+- ächte
+- ▁Zehn
+- ▁quinze
+- 双手
+- ▁đô
+- yorsunuz
+- 共通
+- ▁tutta
+- 仰
+- ▁sentido
+- ▁accommodation
+- ▁frequency
+- 友谊
+- ▁Nigeria
+- 邮件
+- ▁публи
+- ámos
+- 就业
+- सि
+- ▁fiddle
+- ▁أول
+- ▁northward
+- 很奇怪
+- 这就是你
+- ▁бүл
+- 机关
+- 愁
+- ▁Tränen
+- ▁airplane
+- சிய
+- ▁moralische
+- ▁දැ
+- ▁luôn
+- spetta
+- ▁fiend
+- 干预
+- ▁potenc
+- 勃
+- ޓ
+- ▁transparency
+- ▁hypothesis
+- 守備
+- 作为一名
+- ▁damsel
+- 勝手に
+- ▁fancies
+- ▁bấ
+- RE
+- ▁cruz
+- 不允许
+- お昼
+- ▁запис
+- працоў
+- ハハハ
+- 한데
+- ▁realization
+- 随机
+- 分解
+- köz
+- ▁характер
+- 轴
+- urukundo
+- ▁surtout
+- 印刷
+- 我从来没有
+- 现金
+- オフ
+- ▁chị
+- ▁lascia
+- tropic
+- ▁rwego
+- ▁Carol
+- ött
+- と思っています
+- فريق
+- 弓
+- ▁recupera
+- レストラン
+- މު
+- 宿泊
+- ▁abolish
+- jumu
+- ▁قم
+- ▁diventa
+- ▁chronicle
+- 师兄
+- ▁умер
+- 研讨会
+- 嘴唇
+- 一首歌
+- որ
+- ▁allocation
+- にならない
+- ▁existiert
+- ▁homeward
+- gewicht
+- 马车
+- ▁beneficial
+- ▁Hunderte
+- ▁Thor
+- ▁различ
+- धा
+- 写信
+- undsechzig
+- ಟ
+- 随着时间的推移
+- ▁полностью
+- ▁çoğu
+- 駄目
+- シャツ
+- kogu
+- ▁mwana
+- てくれました
+- прыг
+- ▁prolong
+- なんですけども
+- 起源
+- ▁Matthew
+- 限り
+- ▁repentance
+- ▁hermano
+- ▁dinero
+- ▁oscar
+- исты
+- 氛
+- ▁securing
+- ▁ukuthi
+- ▁derjenige
+- ▁Beitrag
+- 上午
+- 난
+- ▁gibb
+- ▁Evrop
+- コントロール
+- ▁Records
+- 牙齿
+- ▁Ні
+- ▁ғой
+- ▁jimmie
+- ultima
+- ▁Earl
+- ▁complac
+- 相遇
+- 拘
+- ▁verlangt
+- ēji
+- 玩具
+- 出発
+- 框
+- ▁deceased
+- ▁причем
+- ▁geöffnet
+- ▁melody
+- ▁få
+- ▁있을
+- ▁مرة
+- important
+- 投资者
+- ▁southward
+- ▁உள்ளன
+- 航行
+- 借口
+- ční
+- េ
+- ▁erheblich
+- 視聴者
+- ▁heiraten
+- 就是为了
+- ▁neunzig
+- 復帰
+- 回顾
+- ▁dagger
+- 言いました
+- ▁feverish
+- 尖叫
+- ▁Hass
+- ▁fearless
+- ▁programming
+- yonna
+- ▁extremity
+- ▁avere
+- ▁minimal
+- гээд
+- こない
+- 났
+- ▁wondrous
+- ▁دیگه
+- すごいね
+- losigkeit
+- продукт
+- ▁unaware
+- ▁factories
+- კი
+- 起作用
+- ▁millionaire
+- লের
+- черед
+- 躍
+- 钉
+- ▁varieties
+- ▁mauvais
+- ▁vairāk
+- ▁booth
+- ▁dónde
+- ۇق
+- service
+- 最早
+- ▁unkind
+- 이나
+- fisch
+- ▁adverse
+- ▁узнал
+- ▁가고
+- ▁athos
+- ח
+- 乾燥
+- روب
+- न्छ
+- ▁darkened
+- ▁applies
+- 亨利
+- ▁приказ
+- って言う
+- ▁முன்
+- વા
+- ▁spontaneous
+- ▁দিয়ে
+- schrecken
+- stancia
+- ల్
+- ▁갔다
+- gefüllt
+- 普段
+- ▁speck
+- なんだろう
+- ▁Вот
+- 营地
+- டெ
+- gelaufen
+- 邻
+- ▁Acest
+- 掲げ
+- ▁corporal
+- おかげで
+- 呼んで
+- নো
+- ▁Fernsehen
+- ▁galley
+- 天主教
+- ▁precision
+- ▁uneasiness
+- ▁фотограф
+- ▁pēc
+- 遵守
+- ▁Again
+- ▁kontrollieren
+- ▁olabilir
+- ▁luce
+- ▁rubbing
+- bwy
+- ▁decoration
+- ▁repay
+- ▁kullanıl
+- ▁immigration
+- 毯
+- 蒋
+- ▁volatility
+- ▁похоже
+- 쪼
+- ▁grievous
+- つなげ
+- マリウポリ
+- 一个地方
+- 進み
+- 反复
+- ▁tiên
+- ▁Waffe
+- ▁высоко
+- はありませんでした
+- ▁expressive
+- 増えている
+- 皱
+- 돌
+- ▁نبود
+- ▁Dutzend
+- と思うんですね
+- 对你来说
+- ▁geography
+- steigt
+- ▁variant
+- 助理
+- ▁Vögel
+- ▁احساس
+- ▁Klin
+- ▁residential
+- Imperi
+- ▁modified
+- ▁solange
+- ▁filings
+- αν
+- وظ
+- ▁splendor
+- ▁Kaffee
+- әлә
+- eficient
+- 豊かな
+- ▁معا
+- ▁старш
+- ▁сложно
+- ▁хва
+- หา
+- πο
+- ▁Burg
+- ▁федеральн
+- ▁лож
+- ▁prostitu
+- 伤心
+- ▁Howard
+- опо
+- まらない
+- ▁vieux
+- хватил
+- 巫
+- 吕
+- ▁đúng
+- ▁nightingale
+- 选项
+- 同士
+- に到着
+- 設定
+- ▁postpone
+- ▁нужен
+- ▁крут
+- 绝不
+- ▁robbery
+- ▁Му
+- ▁snarl
+- ▁cél
+- வான
+- ▁Anthony
+- ▁Krankheiten
+- ▁reappear
+- あるんですね
+- 清醒
+- тянул
+- 违反
+- ▁život
+- ക്ക
+- 琼
+- 防犯カメラ
+- Effekt
+- ▁elastic
+- horaho
+- 思え
+- мәй
+- ▁troy
+- てみよう
+- سطح
+- kombe
+- ▁Tanz
+- ▁wipe
+- 漢
+- ▁cherche
+- 粮食
+- ▁držav
+- 模様
+- 知識
+- ▁trả
+- ▁mold
+- ɣur
+- ▁softened
+- 絡
+- 袁
+- ▁기억
+- 忠诚
+- 预计
+- ▁descendants
+- 結び
+- 别忘了
+- 还有一些
+- ▁machst
+- อบ
+- ▁modification
+- 挽
+- 刮
+- 不断地
+- klā
+- ذكر
+- 这两天
+- ▁Philipp
+- 主持
+- ▁sıra
+- 上涨
+- kreuz
+- ▁அவன்
+- трымліва
+- ▁ເອົາ
+- ▁sprinkle
+- ▁hesitating
+- 目撃
+- 資料
+- ▁chinesische
+- ▁transmission
+- ▁trebui
+- 风景
+- cloud
+- かかった
+- 疫
+- ってくれる
+- 傾
+- ebɣa
+- ▁pregnant
+- ▁memorable
+- ▁Unterricht
+- ▁majestic
+- ▁Transport
+- ▁abyss
+- ▁voce
+- 挺好
+- ▁Station
+- 主人公
+- لىك
+- ganira
+- ▁geeignet
+- ▁kentucky
+- ▁telèfon
+- 茨城県
+- 酢
+- よろしくお願いいたします
+- ▁begeistert
+- にもかかわらず
+- ▁profesor
+- 清理
+- ▁사실
+- ▁rumour
+- ▁Forscher
+- ▁cupboard
+- 見つけた
+- 千万别
+- と思ったら
+- ▁нашу
+- ▁хэсэг
+- 翁
+- キャラクター
+- ▁ນະ
+- ▁охран
+- ordinate
+- 考えると
+- ▁gelten
+- ▁chalk
+- ▁пути
+- ▁competent
+- 赠
+- ▁cecilia
+- ▁спокойно
+- ▁exempt
+- 苦労
+- ▁cambridge
+- 美洲
+- ゆっくりと
+- ▁краіны
+- 減ら
+- 下一步
+- ▁cripple
+- ▁sunrise
+- 没法
+- ▁vincent
+- かなと思って
+- 毕
+- ரும்
+- 平常
+- 祖先
+- ▁நெ
+- lood
+- 喘
+- ಬ
+- ラジオ
+- منطقة
+- ▁civilian
+- 快递
+- ัด
+- 仆人
+- ▁liquidity
+- ▁Onkel
+- 地铁
+- ▁thiết
+- 参观
+- 来自于
+- وست
+- ▁jelly
+- 爸爸妈妈
+- stunde
+- 見ている
+- 'ON'
+- ▁Termin
+- ដ
+- 嘱
+- ▁hudson
+- ▁நிற
+- ▁fraction
+- গু
+- দ্ধ
+- 媳妇儿
+- 近づいて
+- ứ
+- 支出
+- すてきな
+- 贺
+- ▁ceremonies
+- ▁поддержк
+- ▁безопасности
+- ▁말이야
+- ▁regió
+- ▁obstruct
+- ▁mercat
+- 转移到
+- 领先
+- 美容
+- даецца
+- 活力
+- ី
+- ▁shrank
+- ▁mañana
+- այ
+- ▁têm
+- NN
+- 広い範囲で
+- 乐意
+- intensive
+- 教団
+- 番号
+- ▁galv
+- 약
+- ▁لذلك
+- ▁ஆகும்
+- ▁дахь
+- ▁полковник
+- ▁নিয়ে
+- 谈恋爱
+- ▁nursery
+- ▁flaming
+- ▁Darwin
+- ▁شکل
+- 短期
+- 挫
+- ▁Georgia
+- 霊
+- ▁negotiate
+- ▁gahunda
+- ▁fuerza
+- ▁Kapitel
+- ▁puritan
+- 尊严
+- ▁এখন
+- ▁இருக்கும்
+- ブロック
+- 撒谎
+- লাম
+- ▁noticing
+- ▁rebuke
+- ▁vexed
+- 年目
+- ĝoj
+- вэр
+- 生きる
+- ▁выступа
+- 赴
+- ▁очевидно
+- gång
+- 明らか
+- liegenden
+- 各自
+- зву
+- 해가지고
+- ับ
+- ▁Yani
+- ▁матери
+- ▁сделали
+- ▁آنجا
+- ▁Zunächst
+- ▁Пасля
+- ▁싶어
+- ▁наук
+- ▁جوان
+- ▁homoj
+- 毛病
+- 几百
+- 重量
+- ޔ
+- ▁Lächeln
+- ▁vijf
+- ▁imperative
+- 财政
+- писыва
+- 曲げ
+- なのかな
+- ハリ
+- ▁Landschaft
+- дорож
+- ēju
+- につなが
+- ▁betroffen
+- 貧
+- ▁یافت
+- 修改
+- ▁Porque
+- 懸
+- แล้ว
+- ▁einschließlich
+- ▁jüngste
+- ▁übertragen
+- ▁государстве
+- ▁Мар
+- ▁Señor
+- եր
+- 国民党
+- 目指
+- 可见
+- 闺女
+- 些什么
+- ▁resignation
+- weichen
+- ▁rusty
+- ているということです
+- 指南
+- 祖母
+- 侍
+- ▁منطقه
+- ▁பின்னர்
+- ވާ
+- ▁utilization
+- ▁nhỏ
+- 野蛮
+- ▁Beck
+- 我们确实
+- ▁hannah
+- 飲んで
+- diğini
+- ▁зараз
+- 虽
+- 全国各地
+- rigg
+- ▁düş
+- 督
+- ▁Sala
+- 併
+- époque
+- ▁malgranda
+- ▁proclaim
+- pferd
+- ▁Anzeige
+- ▁yardım
+- jyanye
+- ▁gait
+- ନ
+- ▁Bemühungen
+- 洒
+- 翠
+- ▁싶은
+- 哭泣
+- technik
+- 清洁
+- ▁Fac
+- temperatur
+- 光明
+- ติ
+- ▁Lippen
+- 僚
+- ばっかり
+- ▁Roboter
+- ாட்சி
+- mahanga
+- ▁dreizehn
+- 站住
+- ▁Funktionen
+- 自衛隊
+- 花费
+- 公布
+- ▁implant
+- ▁murray
+- 深处
+- ▁partake
+- цаў
+- 琪
+- ▁excellency
+- ▁insignificant
+- ご存じ
+- ▁especie
+- ▁deputy
+- দেশ
+- ъезд
+- 物質
+- ▁verteilt
+- ▁spinning
+- spirited
+- ▁fairies
+- ▁Bydd
+- ống
+- ▁exerc
+- ワイン
+- yitibwa
+- いらっしゃいます
+- 喜び
+- やってきた
+- ▁humil
+- ▁kumenya
+- ическим
+- 作られた
+- 八百
+- えっと
+- ▁alumni
+- ▁отделения
+- ▁kulturelle
+- ▁headache
+- 呼ばれ
+- ▁zamanda
+- ▁ekster
+- වන
+- થ
+- ใจ
+- ▁així
+- gebunden
+- 行列
+- ▁прошу
+- ▁oblige
+- ▁тады
+- amenye
+- 判决
+- 是时候
+- プレッシャー
+- ▁Terror
+- ▁jordan
+- ▁погиб
+- ība
+- ▁ведае
+- 時過ぎ
+- ▁Pep
+- 转换
+- を持っている
+- ▁snug
+- ▁долг
+- bourne
+- という意味
+- 尺
+- ▁Schicksal
+- ▁hóa
+- ポーランド
+- ހު
+- agrada
+- ▁utilisé
+- どれだけ
+- ▁желез
+- ▁może
+- ▁oppression
+- 因为它们
+- 自豪
+- 融合
+- schossen
+- դ
+- ▁grâce
+- ▁посмотреть
+- 枕
+- ▁foliage
+- むしろ
+- 损害
+- குதி
+- ▁பிரி
+- dığını
+- ▁restaur
+- 婆婆
+- ▁inizia
+- ▁Leiter
+- 拠
+- 分离
+- 侄
+- ▁geoffrey
+- 깐
+- 遵循
+- ▁begrenzt
+- riff
+- ▁хозяйство
+- ▁হবে
+- ющей
+- ▁Public
+- ıyı
+- お子さん
+- Psycho
+- ▁Glen
+- 昨夜
+- 様々な
+- نامه
+- ▁twinkle
+- strafe
+- ▁healthcare
+- 算法
+- ▁worthless
+- 遠く
+- 乐观
+- ▁vardır
+- 著名
+- 象征
+- ▁مادر
+- 显得
+- ▁worauf
+- ▁چو
+- '2011'
+- puesta
+- 締
+- ▁மிக
+- āli
+- ▁아빠
+- ▁fathom
+- ▁новые
+- ▁valiant
+- ▁fanatic
+- 很快就会
+- бросил
+- ▁gusto
+- ▁procura
+- 傲
+- 喻
+- 火曜日
+- ▁scruple
+- ▁unbekannt
+- ▁몰라
+- ▁chronic
+- ▁முன்ன
+- ▁ziek
+- 誇
+- ▁folosi
+- juri
+- ▁Anruf
+- 恼
+- ▁оружие
+- ▁eccentric
+- ▁inconvenience
+- ▁luxurious
+- ▁đưa
+- ▁benannt
+- ▁puedo
+- ▁defensive
+- 同じような
+- ▁звоните
+- たりとか
+- 艰
+- bearer
+- 灌
+- ▁enclosed
+- ▁lizzie
+- verhältnis
+- ▁አይ
+- ▁pickwick
+- ▁Armut
+- ▁nelson
+- ▁allemaal
+- kumva
+- ▁klub
+- ▁commencement
+- ▁discord
+- alphabet
+- ▁nördlich
+- ポーズ
+- 驳
+- ▁đối
+- ▁чинь
+- ‍
+- ▁südlich
+- 郎さん
+- ክ
+- 項
+- ▁erfüllen
+- ▁Что
+- ▁головой
+- 嘴里
+- ować
+- ▁hinweg
+- 拉丁
+- ▁самой
+- を求めて
+- 食べられる
+- に当たる
+- прашива
+- シンプル
+- ▁sarebbe
+- 职责
+- 模拟
+- 国境
+- ▁다시
+- ▁titan
+- テロ
+- 藤井
+- builder
+- ▁Massachusetts
+- ▁gäbe
+- ▁먹어
+- ▁сосед
+- ▁heritage
+- 早晨
+- ▁rappel
+- ণে
+- ▁ehren
+- ▁politika
+- ▁facilitate
+- 卫星
+- ▁lächeln
+- ▁erhöhen
+- 严厉
+- おしゃれ
+- ▁Pacific
+- 康复
+- 暴行
+- と思うんですよね
+- ▁prostrate
+- 胡子
+- 这时候
+- ஃப்
+- ▁antagonist
+- ▁фед
+- 权威
+- 眼镜
+- ▁Wang
+- ▁депутат
+- ▁существо
+- ▁hubiera
+- ლო
+- ▁olacak
+- 孤立
+- ▁affront
+- 予防
+- ▁Susan
+- klagen
+- ▁parrot
+- 日常生活
+- ▁měl
+- ▁لطفا
+- 茫
+- ▁موضوع
+- 栽培
+- ▁Board
+- ▁Northern
+- しょうゆ
+- 市にある
+- ▁prosecution
+- ▁можешь
+- アニメ
+- 边界
+- dependence
+- американ
+- 埋め
+- alytic
+- ▁animation
+- ▁وكان
+- 農業
+- 尻
+- จาก
+- ラウンド
+- ▁magician
+- состоя
+- ▁freak
+- 再一次
+- ▁лидер
+- ▁داره
+- 子育て
+- ▁verbal
+- ▁benötigen
+- 끔
+- பெயர்
+- 貼
+- アイドル
+- fleisch
+- ▁Point
+- ▁پیر
+- ▁Branche
+- 計算
+- ▁burglar
+- খন
+- 速い
+- ▁furent
+- 悪化
+- ▁wholesome
+- 普及
+- ▁gaily
+- 秘书
+- Produzent
+- 悼
+- ▁enforcement
+- ות
+- 场合
+- 侵入
+- ▁nommé
+- ▁아니라
+- ▁oggi
+- ▁fiber
+- 偉
+- ▁perceiving
+- ▁dinosaur
+- チャンピオン
+- موسيق
+- සේ
+- yinza
+- ▁들어가
+- killer
+- ▁plump
+- 进攻
+- いったん
+- 婦
+- ▁HIV
+- ▁haciendo
+- ▁немножко
+- ▁оппозици
+- ▁thereafter
+- богат
+- سازی
+- 会出现
+- ▁écrit
+- ▁disappearance
+- ▁хаце
+- 百姓
+- ▁وهي
+- говорил
+- ▁prakti
+- ต้อง
+- ▁nerv
+- ▁Kelly
+- ▁Ausnahme
+- 動く
+- σε
+- ▁reverend
+- ホン
+- ▁угодно
+- 抄
+- ▁магчыма
+- ▁எல்லா
+- ▁Erstens
+- ▁crag
+- ▁машина
+- ▁forthwith
+- 携带
+- වත්
+- ▁earnestness
+- ▁interposed
+- ▁представлен
+- ▁trẻ
+- 記事
+- hati
+- ▁stieß
+- ▁sponge
+- ೇ
+- ▁Columbia
+- ▁Großbritannien
+- ▁федерации
+- ничтож
+- ▁offense
+- Bomb
+- 吉姆
+- ێکی
+- ▁estudio
+- ▁darwin
+- ▁viên
+- ຸ
+- ▁возвраща
+- смеш
+- 裁判所
+- 吾
+- ▁완전
+- 成千上万
+- ▁abilities
+- 関係者によりますと
+- 别动
+- 30%
+- 武汉
+- ▁craig
+- ▁economist
+- わけじゃない
+- ▁ülke
+- ▁fung
+- ▁cyose
+- ▁herausgefunden
+- ▁допустим
+- 脑海中
+- ▁맛있
+- ▁دقیق
+- ▁Truppen
+- 連勝
+- ▁perilous
+- 骨头
+- ▁هنوز
+- カウント
+- ▁unangenehm
+- ▁exhort
+- ▁heavier
+- රා
+- 流浪
+- 我爱你
+- 你也可以
+- ▁kijken
+- 処分
+- '2012'
+- ▁Walter
+- ▁reflex
+- 関心
+- ▁Teufel
+- ▁congratulations
+- ▁Dilluns
+- 鶴
+- CEO
+- ▁Tippett
+- ▁achieving
+- ▁Business
+- roost
+- 永久
+- замен
+- ▁Clara
+- このうち
+- 身材
+- ▁junta
+- 輸出
+- ▁دیگری
+- ▁vendor
+- が出ている
+- ▁сёння
+- 幽默
+- ▁Francis
+- ▁regula
+- পুর
+- 兵士
+- ▁Normal
+- sponde
+- たらいい
+- 段階で
+- ▁composer
+- ▁Junior
+- ▁leonard
+- されていて
+- ▁Eindruck
+- solució
+- ▁southwest
+- ▁equipo
+- ▁Metall
+- ▁voters
+- வன்
+- ▁mosquito
+- ▁irgendetwas
+- ▁següent
+- ▁loại
+- င်း
+- 現役
+- alisierung
+- 穿越
+- ▁fervent
+- 描いて
+- 电视台
+- nachricht
+- 主流
+- 广东
+- waardig
+- 필
+- ▁Toronto
+- ▁alteration
+- ▁diligence
+- 閉じ
+- との関係
+- государств
+- ▁Wilson
+- στ
+- シア
+- мовы
+- ▁curr
+- тып
+- 主演
+- ▁neugierig
+- ▁элемент
+- ▁vibration
+- お弁当
+- 甜蜜
+- ▁nikola
+- ▁chacun
+- 登记
+- ▁flirt
+- ▁rapidity
+- ▁pourrait
+- ▁ومن
+- быстр
+- avion
+- ব্
+- 几十
+- ▁людзей
+- ▁Geschmack
+- 構造
+- 日連続で
+- が必要な
+- 続けた
+- כ
+- ▁sequential
+- ▁whistling
+- 垣
+- ▁Gestalt
+- ▁그래가지고
+- 交換
+- ▁compose
+- праў
+- ▁estudiant
+- 憧れ
+- ▁infernal
+- ▁věc
+- ▁navigation
+- 选民
+- ▁recap
+- 享
+- ▁америк
+- ▁Hungar
+- 天赋
+- ▁emerald
+- ▁पनि
+- ▁شامل
+- ▁Motiv
+- ▁aufregend
+- 此刻
+- ▁generating
+- وى
+- ▁вялікі
+- ▁оказыва
+- myśl
+- ▁fácil
+- ▁treacherous
+- 湘
+- そっち
+- ▁harriet
+- 雷雨
+- 一瞬
+- ▁pouco
+- 特別な
+- 智力
+- ගෙන
+- ▁hunne
+- 绩
+- ▁Emotionen
+- ಅ
+- に基づ
+- 威廉
+- ▁beseech
+- ▁dramatically
+- 落ち着
+- 非難
+- 見通しです
+- ▁срок
+- んですけれど
+- ulira
+- プラン
+- 搅
+- ▁advisor
+- ▁knives
+- acağım
+- ▁Ла
+- ▁警察によりますと
+- 几个小时
+- 是正确的
+- ▁schrie
+- ரின்
+- 改正
+- ▁lyric
+- ▁могла
+- ବ
+- ▁penetration
+- ▁Nächste
+- ▁být
+- atrix
+- ساز
+- 你为什么不
+- ▁konata
+- ётся
+- たぶん
+- ०
+- ▁superficial
+- ▁unreasonable
+- ▁điểm
+- ▁grotesque
+- ▁coroner
+- ▁Beschreibung
+- ▁다음
+- ▁refreshment
+- 昭
+- 传达
+- ▁надеюсь
+- ំ
+- ▁Ontario
+- ▁divinity
+- ▁vehement
+- ▁settling
+- 保育
+- ▁лучш
+- ▁bekomme
+- っていう感じ
+- ▁Witz
+- 歩き
+- ضاء
+- diagnose
+- 沿岸
+- 衡量
+- げん
+- ▁நல்ல
+- 改进
+- ிடம்
+- ▁большие
+- ▁Vä
+- ▁Stress
+- ▁транс
+- ▁dauern
+- platte
+- ఁ
+- १
+- ▁humiliation
+- ▁بیرون
+- ▁Könnte
+- 軍事侵攻
+- ことにしています
+- लो
+- れん
+- டோ
+- ждения
+- ▁dickens
+- 江湖
+- ▁ansieht
+- ▁insgesamt
+- ▁вещь
+- ▁دنبال
+- 寒冷
+- ▁lobby
+- ハム
+- 年ぶり
+- 死刑
+- 在接下来的
+- 绘
+- פ
+- ▁thú
+- ▁millones
+- ▁Presse
+- コート
+- エイ
+- 右边
+- entrada
+- liselt
+- ▁Engagement
+- 芋
+- ▁worüber
+- ▁regretted
+- 首席
+- そうなんだ
+- ▁costru
+- を決める
+- fuß
+- റ
+- ▁Margaret
+- 亭
+- 参议员
+- ▁Nutzen
+- sluit
+- étend
+- gambi
+- ▁workshop
+- ▁Sprach
+- schleunig
+- બ
+- ▁además
+- ▁золот
+- 聞かれ
+- ▁endowed
+- ▁strode
+- крыты
+- 하면은
+- '4000'
+- ▁kommun
+- ত্ত
+- 总理
+- ▁миллионов
+- ▁escaping
+- န်
+- ウクライナ侵攻
+- ▁முதல்
+- ▁Provinz
+- ▁Questa
+- 有哪些
+- ▁occupant
+- ▁rugged
+- 調べに対し
+- ▁البر
+- ▁Gedanke
+- 我只是想
+- 篮
+- 贩
+- ঝ
+- ▁arguing
+- ▁хлеб
+- ▁Certain
+- を巡り
+- ▁پشت
+- 生き物
+- ▁parola
+- ▁수도
+- 主教
+- 包装
+- 遗传
+- เธอ
+- 举办
+- 陥
+- 艦
+- ▁shabby
+- 透露
+- скага
+- ▁picnic
+- ▁construcció
+- 占领
+- ▁activist
+- işim
+- であること
+- ▁davvero
+- ▁crític
+- 珍惜
+- ▁çalışma
+- ▁другого
+- ▁rogue
+- ▁geliyor
+- ্ট
+- ▁todavía
+- ▁ډېر
+- ▁गर्न
+- おかしい
+- ▁действия
+- こういうこと
+- ին
+- ▁примерно
+- ▁Greg
+- られていた
+- 猴子
+- ▁adieu
+- ▁готовы
+- ▁akzeptieren
+- 纱
+- ▁Bewertung
+- おそれがあります
+- ▁вопросов
+- ▁hybrid
+- ▁único
+- ▁کردیم
+- ▁reprit
+- escena
+- ▁Ҡа
+- ▁hoặc
+- ▁nostrils
+- ▁champagne
+- やり方
+- ▁smote
+- 圧倒
+- ならば
+- ▁babiri
+- セカンド
+- 才知道
+- 連覇
+- ▁идти
+- ▁imposing
+- Book
+- waarde
+- yczn
+- らっしゃる
+- ▁hoorde
+- ▁verbessert
+- ▁zooveel
+- 形容
+- もらいました
+- ▁inviting
+- іўся
+- ▁volcano
+- 新潟県
+- ▁eastward
+- froid
+- ніз
+- mauer
+- lösung
+- ▁undertook
+- 伤口
+- ивается
+- ▁زمانی
+- කට
+- イヤー
+- 興
+- ▁militärische
+- マイナス
+- 这部电影
+- ▁trifft
+- ▁доктор
+- ▁analytics
+- ્ર
+- 领袖
+- ▁notorious
+- ▁piercing
+- ▁película
+- ▁compara
+- ▁молча
+- ▁commentary
+- ▁Morris
+- станци
+- チュー
+- ▁conscientious
+- 坠
+- ▁Stoff
+- غۇ
+- 对我们来说
+- серд
+- 受け止め
+- ▁Brüder
+- 皇家
+- 提起
+- 検証
+- ގ
+- rechnet
+- ழை
+- ▁yakın
+- ▁alluded
+- ▁Parlement
+- ▁ripple
+- ▁trocken
+- вуча
+- ▁Saison
+- といわれて
+- ▁Hugh
+- 議長
+- 敞
+- શ
+- ▁erscheint
+- ▁خاطر
+- ▁нужна
+- sighted
+- 做不到
+- ふた
+- 向かい
+- 分类
+- 見たこと
+- ▁comparative
+- 翅膀
+- 丛
+- ▁necklace
+- 交往
+- 建築
+- 大伙
+- ▁станция
+- ▁geraten
+- ▁Gebet
+- ▁peuple
+- ▁weibliche
+- 重症者
+- ▁Vortrag
+- ▁Раз
+- گذار
+- ▁acerca
+- பை
+- してしまった
+- どうでしょうか
+- ▁জান
+- ▁disclosure
+- ▁geschieht
+- ▁здравствуйте
+- ▁apprentice
+- ▁Blumen
+- シングル
+- ▁одним
+- 入侵
+- ▁näi
+- に関して
+- ▁wakati
+- ▁качестве
+- ғыҙ
+- ▁blickte
+- ▁anecdote
+- ایل
+- ▁secund
+- ▁سەر
+- على
+- ▁devout
+- 整整
+- 现在正在
+- dzē
+- 脆
+- 羡慕
+- ▁Houston
+- ▁Erwartung
+- ぴったり
+- ▁genießen
+- ▁شاید
+- ▁nombreux
+- ▁Bruce
+- ▁genus
+- 两周
+- パット
+- ▁лежа
+- gefahren
+- ▁آس
+- อยู่
+- ▁Pennsylvania
+- ▁છે
+- 佐々木
+- durchschnittlich
+- 机械
+- 晴れて
+- ▁humbly
+- ▁afecta
+- ▁тракт
+- 屋根
+- ▁করেন
+- 知らせ
+- ▁diagram
+- ▁evitar
+- くなっている
+- ▁intuition
+- ▁jonathan
+- blätter
+- ▁default
+- ▁measuring
+- зван
+- ▁어제
+- ▁protocol
+- últim
+- してくる
+- 処理
+- ეს
+- 发誓
+- ▁Mount
+- autant
+- 如果有人
+- ▁миров
+- 障
+- ▁анализ
+- ▁அழை
+- 如果你能
+- 停電
+- 的角度来看
+- வரும்
+- 친
+- 戚
+- ▁Präsentation
+- ▁Festival
+- 伊丽莎白
+- ▁Geräte
+- ▁иметь
+- ▁cherry
+- ▁Vergnügen
+- ▁بىلەن
+- ▁때문에
+- 淹
+- ヴ
+- ▁Portugal
+- ▁Crist
+- 尤其
+- ▁Major
+- の様子です
+- 趣味
+- gesteld
+- 糊涂
+- 色んな
+- ▁اصلی
+- ▁Ausschuss
+- వు
+- ▁pluraj
+- 室内
+- ировали
+- 構え
+- runde
+- 棄
+- ▁گوش
+- ▁михаил
+- 医疗保健
+- ▁waarop
+- を決めました
+- 体现
+- ▁voiture
+- ▁Ufer
+- ▁Route
+- もらいます
+- ▁tác
+- میر
+- 拒
+- を果たし
+- ▁nachgedacht
+- 페
+- ▁комитет
+- ▁ມີ
+- ▁داستان
+- 减肥
+- геҙ
+- ジン
+- 左边
+- ▁Sorge
+- ▁чаго
+- ▁incense
+- ▁العام
+- 旭
+- ▁đại
+- ▁بنابراین
+- ▁смотреть
+- ▁دلیل
+- キム
+- 말
+- ▁investor
+- 塞尔
+- 小伙子
+- 屎
+- ▁Jennifer
+- ും
+- ▁уровне
+- ▁homage
+- ▁видели
+- 正しい
+- ▁laboratori
+- に住んで
+- ▁illinois
+- ▁tiền
+- サラダ
+- ▁boughs
+- ▁russell
+- ▁sagst
+- 警備
+- zuziehen
+- 甘み
+- ▁sinister
+- ரீ
+- ような感じ
+- ریک
+- 姚
+- ▁Wähler
+- ▁columbia
+- ▁ekzistas
+- ▁perplexed
+- ▁братьев
+- 渔
+- ▁grill
+- ▁exalt
+- kontakt
+- ▁feit
+- ▁governess
+- ▁kurya
+- ▁Kindheit
+- ▁sichtbar
+- پتۇ
+- เรา
+- ▁hump
+- ▁پول
+- spread
+- 愈
+- ▁clumsy
+- ▁plutôt
+- 취
+- ▁Lewis
+- 関東地方
+- ▁каманд
+- 哲学家
+- ▁написал
+- ▁jazz
+- そのとき
+- ▁tiel
+- よろしく
+- ▁Sekunde
+- پوش
+- 甲子園
+- ▁Widerstand
+- jüdische
+- ▁pretext
+- ▁début
+- ▁Standort
+- ▁половин
+- ▁shovel
+- йшлі
+- С
+- ▁dennoch
+- schwäch
+- 毒品
+- 救命
+- ▁tiam
+- ▁forbear
+- ▁convincing
+- ▁miraculous
+- ▁поговорить
+- ▁mugihe
+- intérieur
+- 睡着了
+- ▁여자
+- 100%
+- 하기
+- 修理
+- ruḥ
+- 翌日
+- ▁siebzehn
+- ▁waistcoat
+- 继续前进
+- ▁cuán
+- ▁urging
+- 給付
+- mıştı
+- 茹
+- វ
+- ピンチ
+- 80%
+- 缓慢
+- ▁Ли
+- 保密
+- 镜子
+- ▁felix
+- あると思います
+- つらい
+- ▁خودش
+- ▁detachment
+- ▁prescription
+- 貢
+- 都道府県
+- ▁cavalier
+- もしかして
+- 尼克
+- ▁petersburg
+- ▁zunehmend
+- đ
+- layan
+- 哄
+- ō
+- ▁zukünftige
+- ▁declining
+- ▁extern
+- சொல்
+- 积累
+- っていうのも
+- europa
+- स्त
+- ▁starving
+- 祭り
+- 呼吁
+- ٌ
+- ▁corazón
+- ▁сталин
+- ▁eugene
+- ▁participating
+- 做生意
+- ▁condens
+- 描かれ
+- łow
+- できれば
+- ▁zacht
+- 删除
+- 適用
+- ▁скажи
+- ▁Definition
+- intérêt
+- 滥
+- ▁Một
+- schloß
+- ▁প্রতি
+- 取り組んで
+- ▁tolerate
+- ▁điện
+- ▁auction
+- ნე
+- ▁Gefangene
+- zungu
+- ▁hieß
+- 董事长
+- ▁calamity
+- ▁precipice
+- ▁ایجاد
+- ▁hoffnung
+- ▁nuovo
+- 囚犯
+- ▁două
+- タイムリー
+- 主管
+- ▁Rw
+- ▁حسنا
+- ▁meditate
+- ▁Fakat
+- メダルを獲得
+- دعو
+- 博客
+- ▁schweigen
+- ▁cemetery
+- ▁lloyd
+- 審査
+- 啤酒
+- 成績
+- nął
+- ▁rook
+- ▁Association
+- ▁Perezida
+- ▁baltimore
+- ▁endurance
+- 洛杉矶
+- 消耗
+- 物理学
+- ющее
+- стреля
+- 很多人都
+- 揺
+- ច
+- 这辈子
+- ▁Knochen
+- ▁και
+- ▁видно
+- ▁Wirklichkeit
+- ▁چشم
+- ソフト
+- ப்போ
+- 弟子
+- 1,000
+- 干什么呀
+- գ
+- ▁অব
+- ▁unreal
+- ▁Kristo
+- gewandt
+- ▁হয়েছে
+- сроч
+- ▁volgende
+- ▁gelukkig
+- 하면서
+- ▁زبان
+- ▁arbitr
+- ምን
+- ▁Davis
+- 書いた
+- ергә
+- ▁marcus
+- を持った
+- 尉
+- ▁hilfreich
+- ▁вместо
+- ▁څنګه
+- ▁irgend
+- ▁gambling
+- ▁উপ
+- ▁возможности
+- просить
+- ▁unterscheiden
+- ▁feststellen
+- ゴー
+- 食堂
+- ▁gelebt
+- んでしょ
+- знания
+- ▁estudiar
+- زده
+- ▁норм
+- жир
+- ▁shameful
+- 열
+- ▁течение
+- ▁stammered
+- 阴谋
+- уулах
+- ▁ransom
+- kapital
+- ▁franco
+- 奈良
+- 顺便
+- ▁slipping
+- ığın
+- ალ
+- ▁wichtigste
+- 料金
+- 坛
+- ▁малая
+- 属下的一个
+- 谐
+- 박
+- いかがでしょうか
+- ▁ماشین
+- 読んで
+- ▁шестой
+- Tabelle
+- ۲
+- ▁algún
+- ▁unanimous
+- ▁thống
+- ▁skupin
+- 暮
+- やめて
+- 曇り
+- писал
+- 驶
+- ▁fidelity
+- ▁pouvons
+- ▁mondiale
+- 速報
+- ▁Überleben
+- 離れて
+- lardı
+- ▁quitted
+- ぴ
+- 販
+- ▁느낌
+- ▁воздух
+- ▁patriarch
+- 沙漠
+- ▁развива
+- глядзе
+- 優先
+- ▁Má
+- தான
+- 你怎么知道
+- ▁dispense
+- 変えて
+- ီ
+- 鸦
+- ▁Eigentum
+- ▁discouraged
+- 這
+- こういうふうに
+- 阶级
+- せば
+- ▁leitet
+- theorie
+- ▁cultivation
+- leihen
+- eceğim
+- 巨大な
+- ▁Lektion
+- มาก
+- らせる
+- торгов
+- ▁Empfehlung
+- ▁celestial
+- ▁occidental
+- алтай
+- ▁athletic
+- 桁
+- affaire
+- ものすごい
+- ▁civilisation
+- ▁اتاق
+- 这几天
+- ▁Europäische
+- 注定
+- 该地区
+- 析
+- 掏
+- ▁Mitgliedstaaten
+- ▁recognizing
+- 력
+- ▁казалось
+- ▁Sturm
+- パーク
+- рабатыва
+- ▁военно
+- ▁sentinel
+- どうしよう
+- ▁spike
+- 良心
+- временно
+- ຖ
+- 甄
+- 抛弃
+- ▁получить
+- ▁abgeschlossen
+- 伍德
+- 残念
+- ▁collector
+- ▁микро
+- ▁joshua
+- ▁период
+- ့
+- ▁பெரிய
+- Source
+- ющего
+- くなっています
+- ▁astronomi
+- 汇报
+- 復活
+- țele
+- デモ
+- хир
+- 仙台
+- 囊
+- 舱
+- ▁coincidence
+- ▁compromis
+- メイン
+- inspiring
+- ▁politeness
+- 碧
+- ▁வழங்க
+- 扉
+- ▁pudding
+- ▁baptist
+- ▁Vull
+- ▁epoch
+- ▁combusti
+- entwicklung
+- ▁дней
+- 负面
+- 帰国
+- ▁байгуул
+- ▁Nachbarn
+- ▁대학
+- espér
+- ▁Disney
+- спя
+- ডি
+- なじみ
+- ▁Bedürfnisse
+- 极其
+- ▁بسیاری
+- ▁zurückkommen
+- うどん
+- 悄悄
+- ▁зохио
+- メダリスト
+- ▁kesk
+- ▁possono
+- 棕色
+- 総理大臣
+- වෙන
+- nummer
+- 異なる
+- 城堡
+- ғына
+- ▁relaciona
+- ▁hobby
+- ▁людьми
+- mektedir
+- ▁caballero
+- ▁του
+- ▁ہے۔
+- ▁адрес
+- ▁никакой
+- ▁باشه
+- ▁durchaus
+- ▁außen
+- ▁politician
+- höchste
+- を行いました
+- عامل
+- ણ
+- ቀ
+- যোগ
+- 六个月
+- ▁sophia
+- endroit
+- どうでしょう
+- ێر
+- ファー
+- је
+- فضل
+- 感染症
+- 让我们看看
+- 屋顶
+- 飛車
+- ▁ذات
+- drž
+- 泼
+- asanzwe
+- ား
+- 厚生労働省
+- ▁dungeon
+- ▁جيد
+- 押さえ
+- ▁vollkommen
+- క్క
+- zwingen
+- ведения
+- ▁Mühe
+- ▁seneng
+- нести
+- 幼儿园
+- 磅
+- 腔
+- 烦恼
+- ▁Fahrzeug
+- 眼神
+- ▁чисто
+- ▁далей
+- を迎える
+- ▁sexuelle
+- オリジナル
+- 马丁
+- ▁aufbauen
+- ausschuss
+- における
+- 周囲
+- 狮子
+- できるだけ
+- gegriffen
+- ▁langue
+- ウクライナ軍
+- ▁herzlich
+- ▁suffi
+- ▁İki
+- ▁gehst
+- 苦しい
+- توانیم
+- 塑料
+- ▁chơi
+- ▁khó
+- ▁არა
+- ▁самые
+- ▁tedious
+- 感染状況
+- ước
+- ▁întreb
+- 每一次
+- 岭
+- ▁Vậy
+- ▁discomfort
+- ▁настолько
+- 捐赠
+- ▁капитан
+- konsum
+- ▁رجل
+- ポリ
+- ходили
+- ▁다니
+- ▁économique
+- 敗れ
+- génér
+- ▁Cross
+- ルート
+- lumina
+- 吸收
+- ▁косм
+- 假期
+- klapp
+- 验证
+- ▁Fond
+- ▁bizim
+- ▁portuguese
+- ▁rubbish
+- 복
+- oubli
+- 干吗呀
+- ▁хот
+- 运气
+- тыўна
+- ▁இப்ப
+- ▁kanggo
+- ெட்
+- 寡
+- ▁sanctuary
+- 써
+- よろしい
+- 邸
+- ▁tradicional
+- ▁bandage
+- ▁ukrain
+- 渡辺
+- ミニ
+- というわけで
+- ▁mängi
+- ちゃいます
+- ▁рух
+- abash
+- ▁Gedächtnis
+- ▁pièce
+- 医療機関
+- ▁immune
+- 火災
+- ▁forlorn
+- genossen
+- ▁хоча
+- räsentiert
+- ▁horribly
+- ▁безусловно
+- ▁кӱ
+- ▁семьсот
+- ▁sweetheart
+- ▁مؤ
+- 解雇
+- 涉及到
+- ▁воды
+- 況
+- 塘
+- ▁harvard
+- 罰
+- ▁Speicher
+- ▁benedict
+- ▁fellowship
+- 在这方面
+- 英文
+- ▁pronounce
+- ţ
+- ▁proprie
+- ▁болсон
+- リュ
+- ▁celebration
+- ▁Güte
+- 正在进行
+- 蔽
+- 청
+- 膝盖
+- ▁радио
+- ▁rustic
+- ▁общество
+- ▁pulpit
+- ▁Fußball
+- ▁Josep
+- cliffe
+- ▁தெரி
+- 現代
+- podobn
+- ▁fascination
+- ▁który
+- ▁devait
+- ▁دهند
+- 再現
+- ▁geographical
+- 变革
+- ▁мисс
+- 史密斯
+- ニック
+- ▁Egypt
+- строить
+- 精神病
+- ▁모르겠
+- ▁русско
+- gegenwärtig
+- 둘
+- 씩
+- ▁ĉirkaŭ
+- 喫
+- एको
+- 神话
+- ▁titul
+- ▁Träume
+- ▁আস
+- 厳しく
+- ▁dazzling
+- ▁erwiderte
+- ▁Überraschung
+- ▁gedanken
+- 增强
+- 基督
+- ાર
+- ▁luggage
+- ▁이번에
+- täuscht
+- ড়ে
+- ▁mwiza
+- әҙер
+- щим
+- ▁marvelous
+- 入り口
+- wahrscheinlich
+- 述べ
+- ▁velmi
+- ർ
+- ▁Londres
+- ▁تولید
+- ▁Sonntag
+- ▁hôm
+- 腐败
+- 갖고
+- 承認
+- 考验
+- ▁Chu
+- ▁aisle
+- ▁beauties
+- ambigu
+- ぼう
+- ▁hippo
+- 霉
+- ▁overlooked
+- ▁Takže
+- ▁moisture
+- తా
+- ▁hoàn
+- ってみよう
+- 太平洋側
+- ▁cultivate
+- ▁wobei
+- ▁ecclesiastical
+- édé
+- 爱尔兰
+- ▁пространств
+- おばあちゃん
+- ▁Training
+- పో
+- 餐馆
+- ▁dripping
+- geschenk
+- ▁auditor
+- ▁unequal
+- amatu
+- 白宫
+- mutima
+- ▁fisherman
+- 疆
+- ないですか
+- ▁drake
+- 判定
+- ▁disseny
+- kungu
+- 買った
+- ▁troublesome
+- Blanc
+- ▁доступ
+- 证实
+- ▁mẹ
+- ▁மனித
+- ▁Vermögen
+- 급
+- 引擎
+- づらい
+- Unis
+- ▁Anspruch
+- 房屋
+- 引退
+- ▁борис
+- માં
+- ▁Kategorie
+- ▁зусім
+- 잡
+- 娅
+- ▁Spanish
+- ▁thanksgiving
+- లా
+- ▁хороший
+- ▁honom
+- 隐私
+- ▁flip
+- ▁occurring
+- ▁Ereignis
+- wheel
+- ウォー
+- 生み出
+- だと思うんです
+- quote
+- ▁Despite
+- まいりましょう
+- tumye
+- ▁humorous
+- 女優
+- ▁Voraus
+- ってしまった
+- ▁Köpfe
+- 最終的に
+- ▁rhyme
+- ▁clump
+- ▁видеть
+- ▁luckily
+- ▁رنگ
+- ▁malaria
+- 生涯
+- патрэб
+- ప్ప
+- ▁Foundation
+- ▁secrecy
+- 谜
+- 纤
+- ŝanĝ
+- ▁Provi
+- ▁дүр
+- ご覧いただ
+- беҙ
+- ▁audit
+- ▁spice
+- ▁süd
+- கல
+- 吸引力
+- 色々
+- ▁drüben
+- ▁schüttelte
+- ▁completamente
+- 決断
+- ▁думать
+- 励
+- 塑
+- ▁novelty
+- 龟
+- ▁Flügel
+- ▁diferencia
+- ▁cristian
+- ▁urmă
+- ▁jesuit
+- ▁fördern
+- َلَ
+- と述べました
+- 晴れる
+- を発表しました
+- 言える
+- 陌生
+- ▁medication
+- 季度
+- ▁lächelte
+- 綱
+- ున్నా
+- gratul
+- ▁моего
+- سې
+- ▁اين
+- 兄弟们
+- ▁ئەوە
+- 卿
+- 荒谬
+- 弘
+- ▁제일
+- ▁будешь
+- 适用于
+- 食べたい
+- 40%
+- ▁Ре
+- 场比赛
+- ▁xả
+- ▁kostet
+- ▁bố
+- Werbung
+- ▁Academy
+- ▁mohammed
+- ▁آمریکا
+- 唐纳德
+- 罗伯特
+- ▁помощью
+- 自殺
+- ▁Vì
+- ▁Rauch
+- 剧院
+- ▁curate
+- ▁Moore
+- を越え
+- 一件事情
+- ▁алма
+- ▁distint
+- ▁absolv
+- ▁deceit
+- ▁propriety
+- 怪物
+- 根本就不
+- 度目の
+- ▁Entfernung
+- ▁Après
+- ▁signifas
+- 日ざし
+- ▁Pläne
+- ▁Nerven
+- ▁güzel
+- ビッグ
+- ▁hydrogen
+- ▁winzige
+- farben
+- 鞭
+- ▁thompson
+- ঙ্গ
+- ▁griechische
+- ▁хөгж
+- 観察
+- sloten
+- ▁chunk
+- ▁installation
+- クリア
+- 汚
+- 빠
+- ុ
+- ▁elapsed
+- ▁Government
+- 毁灭
+- ▁rapture
+- しさを
+- ▁Symbol
+- 不思議な
+- gestiegen
+- ▁practise
+- ▁athens
+- ▁recreation
+- պ
+- 猿
+- ▁موجود
+- 現象
+- ▁крайней
+- ▁morbid
+- 感冒
+- ждение
+- 引入
+- ▁делают
+- ▁moustache
+- 洗澡
+- ▁되는데
+- 選び
+- flipp
+- ならではの
+- ▁chemist
+- ▁rustle
+- ▁Student
+- ղ
+- ▁longtemps
+- ▁verschwunden
+- 讽刺
+- ▁sturdy
+- 游客
+- ▁arising
+- ▁irritated
+- ダム
+- ▁твой
+- عالم
+- ▁heroine
+- 見たい
+- 税收
+- オレンジ
+- ▁üret
+- 耕
+- 恥ずかし
+- ▁왜냐면
+- ▁sidewalk
+- ▁никому
+- ▁protože
+- ▁involving
+- ▁alguma
+- krebs
+- бель
+- ▁geplant
+- ść
+- нюю
+- torium
+- ▁abnormal
+- ▁condescend
+- 滤
+- 参议院
+- 別れ
+- ▁Cousin
+- ▁kansas
+- ▁contributing
+- ▁ஆன்
+- ▁волос
+- 带领
+- ▁constantinople
+- ▁inasmuch
+- ում
+- ▁Ähnlich
+- 分かりません
+- ▁spiral
+- ▁tradici
+- 追いかけ
+- 睇
+- 水曜日
+- ވެސް
+- ▁reassure
+- ▁Melbourne
+- ▁gegründet
+- ▁водо
+- 北京オリンピック
+- ▁voyez
+- 介入
+- 获胜
+- ▁سمت
+- 虑
+- 溢
+- ▁která
+- verwandt
+- ▁බව
+- ピンク
+- میل
+- ▁Kommentar
+- 漬け
+- lassung
+- エビ
+- 嫉妒
+- ▁Klimawandel
+- ▁солдат
+- 指责
+- ▁основан
+- ▁implies
+- 排名
+- 飼
+- ▁Rechnung
+- ▁monotonous
+- ▁spēlē
+- ▁medieval
+- ▁rhetoric
+- ▁detached
+- ▁Mountain
+- ▁isolation
+- ▁войск
+- 解决这个问题
+- ▁хочется
+- articula
+- ▁umutima
+- ▁curb
+- ▁liefern
+- 伯格
+- ▁Multi
+- ▁russische
+- 矮
+- ▁Zuhause
+- ▁malcolm
+- ▁nevoie
+- läufig
+- ▁shelves
+- 落ち着いて
+- ▁смерти
+- ▁feudal
+- 澤さん
+- あえて
+- klassi
+- ▁constituent
+- ▁другая
+- ▁rabbi
+- ライス
+- ▁Sklaven
+- ▁الواقع
+- まもなく
+- ▁dictator
+- 丫头
+- 荷兰
+- ▁சீ
+- ▁salmon
+- ▁flores
+- っていうところ
+- 初期
+- 卧
+- ▁hypothe
+- ▁entfernen
+- ▁insolent
+- 不合适
+- ▁মানে
+- ρα
+- ▁Dacă
+- ▁Montag
+- ▁чуж
+- рожа
+- 準
+- ▁Folie
+- 慈
+- 抖
+- ▁Colorado
+- ▁Nutzung
+- ▁cognitive
+- ▁după
+- مدرسة
+- 残忍
+- 요일
+- sexuell
+- ▁القر
+- 尋
+- ▁kemudian
+- ▁persuasion
+- 狄
+- 줘
+- お客様
+- ▁jefferson
+- ▁pivot
+- ▁машины
+- نقل
+- ▁استخدام
+- ބަ
+- ▁whipped
+- ▁geleden
+- ▁escucha
+- ▁Could
+- ▁остров
+- ▁cavall
+- quê
+- 须
+- ជ
+- ▁versteht
+- 拭
+- interesse
+- ▁eksp
+- ▁disappoint
+- ▁även
+- 钩
+- katholisch
+- ▁کودک
+- 紫色
+- ▁cố
+- 代の男性
+- Time
+- 收益
+- ▁чад
+- ▁elevat
+- ▁hacía
+- ▁Alkohol
+- 蔑
+- 褒
+- 从那以后
+- ▁извините
+- ▁тело
+- ▁spirituelle
+- யார்
+- ▁sünd
+- 设法
+- ▁geographic
+- 獲
+- முக
+- ▁impamvu
+- ▁хэдэн
+- ▁Physik
+- ▁umfasst
+- 反抗
+- ▁concede
+- ▁обща
+- ▁zwang
+- 取引
+- ministerium
+- 蹲
+- ০
+- ▁Prüfung
+- ▁embedded
+- ▁possiamo
+- タクシー
+- ▁пятая
+- 出てきました
+- ▁discrimina
+- ェ
+- 能做到
+- ▁clenched
+- ▁pequeña
+- ▁tennessee
+- 哮
+- 煎
+- かっこいい
+- ▁请播放
+- ▁așa
+- ▁Jemand
+- gehoben
+- befehl
+- ▁mildred
+- ▁disrupt
+- mètre
+- コスト
+- ▁нашим
+- ใ
+- 膝
+- ▁ausgewählt
+- ▁nécessaire
+- ▁eclipse
+- 换句话说
+- 丼
+- 誠
+- جنس
+- ▁daisy
+- ▁dzīv
+- மும்
+- 納得
+- kräftig
+- ▁merupakan
+- コラボ
+- 記者会見
+- 收购
+- ▁crusade
+- 金額
+- höhe
+- ایە
+- ▁hína
+- ▁snuff
+- ▁социал
+- möglichkeit
+- ijoro
+- ▁clamor
+- 順位
+- ▁husk
+- ▁Flüchtling
+- 奥巴马
+- ▁duidelijk
+- ▁குறை
+- twenty
+- ▁discut
+- ▁пройд
+- お待ち
+- 发动机
+- 警官
+- ତ
+- ▁aufzubauen
+- 这个样子
+- 넘
+- ▁сергей
+- 愛知県
+- ▁shroud
+- コンビニ
+- 行方不明
+- ▁люблю
+- ▁rusange
+- ▁твои
+- となりそうです
+- 訴
+- ▁augustine
+- чёт
+- ものすごく
+- ▁Society
+- 动态
+- ▁vierundzwanzig
+- ▁ocurr
+- ▁mitigate
+- konomi
+- ▁testify
+- ▁чист
+- ▁имеют
+- ▁destructive
+- ▁историю
+- ▁мальчик
+- ▁alcuni
+- 渠道
+- 砂糖
+- ▁besagt
+- علام
+- waffen
+- 原則
+- ▁agafar
+- ▁supposition
+- ▁diplomatic
+- ▁marshall
+- fünfhundert
+- ▁pilgrimage
+- 今のところ
+- ālā
+- ▁legitim
+- ித்தார்
+- ▁мысли
+- ሚ
+- ▁securities
+- 辖
+- ▁bestätigt
+- umukobwa
+- ▁Therefore
+- 軟
+- ▁problém
+- 並ぶ
+- ҙың
+- ▁Họ
+- 寄付
+- ченко
+- gerissen
+- ▁hâ
+- ▁компьютер
+- ▁முக்கிய
+- こんなふうに
+- 媛
+- 놓
+- ▁vigour
+- ▁apparition
+- ▁imbaraga
+- ▁gratified
+- symmetri
+- ▁clash
+- ▁milestone
+- hundertfünf
+- ▁Doctor
+- ლებ
+- ▁camí
+- عيد
+- 鸿
+- ▁disponible
+- ▁butterfly
+- ▁teenager
+- ▁carlyle
+- ▁Mereka
+- ▁хэв
+- ступил
+- とみられています
+- entendre
+- 棘
+- ▁enabling
+- ▁faltered
+- ▁bizarre
+- ▁nodding
+- präg
+- 重症
+- ▁cartoon
+- ▁hareket
+- とみられます
+- ▁Republik
+- பாடு
+- ▁width
+- ▁риск
+- ▁Runde
+- ẫ
+- ڈ
+- ▁zugänglich
+- ျ
+- 詩
+- مجتمع
+- ゲット
+- 散步
+- dämm
+- ▁eigenlijk
+- 菓子
+- 飛んで
+- 级别
+- 厌恶
+- 齿
+- ▁ingenuity
+- ▁последний
+- ▁руководств
+- ▁NASA
+- ▁muffled
+- ▁theoretical
+- ▁நிறுவ
+- 女孩儿
+- ▁shilling
+- ▁ethnic
+- ▁сделан
+- 合わせた
+- క్
+- счастлив
+- 茅
+- ▁Indonesia
+- ▁مدرسه
+- ▁정말
+- ▁Fehl
+- 豪華
+- ▁chestnut
+- 伙计们
+- が止ま
+- 新宿
+- ▁பெரு
+- 了解更多
+- ▁серьезно
+- 績
+- 面试
+- 结局
+- pflege
+- Á
+- թ
+- ▁crouched
+- フルーツ
+- 최
+- ▁espacio
+- خصوص
+- ▁eindelijk
+- 悲剧
+- zauber
+- physiologi
+- ▁подготов
+- ▁лепш
+- 一个例子
+- 远处
+- 合計
+- చ్చ
+- 赎
+- ຢ
+- ▁ecstasy
+- ▁نمایش
+- ӹн
+- 実態
+- rechnen
+- менять
+- ▁новых
+- 竟
+- ▁landlady
+- 繊
+- ▁gutanga
+- ▁ominous
+- ▁voulu
+- 証言
+- 演示
+- ▁junk
+- ▁integrate
+- 隠れ
+- ▁Außer
+- ▁Entschuldigung
+- ▁Kommissar
+- 扫描
+- ▁hoofs
+- wäss
+- اگه
+- ▁лица
+- 嘿
+- ▁merchandise
+- ▁செயல்
+- ▁elkander
+- ▁spüren
+- ▁clown
+- ▁важна
+- ▁calculate
+- 収入
+- 続けている
+- 灵活
+- ▁தனி
+- ▁fügte
+- ▁взаимо
+- ▁갑자기
+- ▁Republikaner
+- ▁infidel
+- ▁desolation
+- ▁längst
+- 坚强
+- ▁cicl
+- 行きたい
+- lehrer
+- ▁literal
+- 韓
+- 非常有趣
+- ▁marcel
+- ▁Intelligenz
+- 冈
+- ▁пошли
+- 气候变化
+- が必要だ
+- ▁странно
+- με
+- 种族主义
+- の疑いで
+- ▁yahweh
+- 斥
+- 至关重要
+- ▁Kämpfe
+- ▁detained
+- ▁هنر
+- ▁sovint
+- ▁syllable
+- ▁mittlere
+- schalt
+- aufnahme
+- トルコ
+- ▁цели
+- ▁judith
+- ▁spacious
+- 海滩
+- کۆ
+- ▁yazı
+- ▁भए
+- ▁Minnesota
+- ▁использовать
+- ▁languid
+- ▁آورد
+- ▁reiterate
+- ▁Patrick
+- ▁убива
+- ▁توجه
+- Europa
+- ▁تواند
+- 崎さん
+- ▁Richtlinie
+- ▁kibazo
+- ▁potenziell
+- ▁deferred
+- ▁பிறகு
+- ނެ
+- ▁usurp
+- 羽毛
+- schwor
+- نوشت
+- ▁appoint
+- ▁sancho
+- ▁குழந்தை
+- ▁Үүний
+- ▁línea
+- ▁Studium
+- ▁Ireland
+- ▁Modern
+- 病床
+- льныя
+- ▁кровь
+- 査
+- 心疼
+- 렸
+- ▁يوجد
+- owski
+- ▁konkret
+- ▁பற்றி
+- ▁categori
+- ▁نقش
+- дзь
+- 炼
+- ▁நிகழ்
+- ▁indicating
+- ▁Gegenteil
+- ▁Emily
+- ▁война
+- 行われている
+- ▁presidential
+- ▁Little
+- கொள்
+- 肤
+- ▁Existenz
+- 拜访
+- ▁antony
+- ▁Samuel
+- 見つかり
+- ▁může
+- 垒
+- 慷慨
+- ▁Ernährung
+- ▁displeasure
+- ッグ
+- 捉
+- ▁говорим
+- ▁değiştir
+- 必然
+- ▁condicion
+- ▁welsh
+- 拜拜
+- 失业
+- ▁sparrow
+- アピール
+- ▁sociedad
+- ދަ
+- ދު
+- ▁사람들이
+- ▁mercado
+- 見つけ
+- 文書
+- ▁Auftrag
+- ▁Annahme
+- ▁Coast
+- ներ
+- 霜
+- ▁boyunca
+- 時半ごろ
+- ▁asteroid
+- ▁коротк
+- ▁قانون
+- ிற்கு
+- nshingano
+- ▁musí
+- 边境
+- ▁riddle
+- 伍
+- 謝
+- ▁Illinois
+- ▁دانشگاه
+- schließt
+- ▁पु
+- イラン
+- ▁ĉu
+- ミング
+- ジョー
+- ▁comunidad
+- ▁companionship
+- 轻易
+- qqim
+- ▁portray
+- ▁үнэ
+- ёшь
+- 好奇心
+- になりたい
+- 虾
+- bitekerezo
+- ▁Pēc
+- ▁antiquity
+- ▁científic
+- 淋
+- アルバム
+- ▁acontece
+- おととし
+- 師匠
+- gemeinschaft
+- 看一看
+- きょう午前
+- 抜いて
+- 谢谢大家
+- ▁bırak
+- versorgung
+- ▁Konferenz
+- GPS
+- ▁혼자
+- ▁gracias
+- ▁athenian
+- ▁persoon
+- ▁inaugur
+- ▁менш
+- 群众
+- ▁northeast
+- ▁vorgeschlagen
+- 雨が降って
+- ▁patriotism
+- 档案
+- luğu
+- ໃຫ້
+- ▁defiant
+- ▁malicious
+- ▁Kansas
+- ▁chaplain
+- 残酷
+- ▁bertram
+- 交付
+- 消化
+- ▁Felsen
+- ▁bathroom
+- おじさん
+- ▁байр
+- ▁palju
+- ▁Alpha
+- ▁Katastrophe
+- ▁Respekt
+- ▁обязан
+- ツイッター
+- ▁oyna
+- ▁غذا
+- ▁أفضل
+- ▁لدينا
+- filtr
+- 层次
+- 主导
+- 속
+- ጥ
+- ▁sophisticated
+- ▁буквально
+- ▁façana
+- 侮辱
+- ▁дороги
+- 前两天
+- 将棋
+- ▁너는
+- ريخ
+- ුරු
+- 労
+- 戴安娜
+- 逝
+- ▁سیستم
+- ̃
+- 剥夺
+- ▁ensued
+- 这是一件
+- schwimm
+- 是不可能的
+- 辞め
+- ▁pueda
+- església
+- süchtig
+- ▁birçok
+- 出租车
+- 귀
+- 麗
+- 적으로
+- 注射
+- schränkung
+- ▁snare
+- ▁skate
+- ▁retard
+- 一生懸命
+- ▁ängstlich
+- ▁vegada
+- ডা
+- あるんですか
+- 起訴
+- räte
+- 徽
+- ▁Posteriorment
+- ▁nyinshi
+- ▁trabalho
+- 코
+- ▁చె
+- ▁Больш
+- ロンドン
+- ▁verwandelt
+- ▁bagian
+- 設計
+- zugreifen
+- ▁பழ
+- 语音
+- ▁naples
+- 世の中
+- 括
+- 嚟
+- ఫ
+- 挖掘
+- 世帯
+- ▁scottish
+- 見直し
+- ركز
+- 現れた
+- ▁Stich
+- ▁refinement
+- ▁keith
+- 老虎
+- бега
+- ▁temporarily
+- コンピュータ
+- 腸
+- 昆虫
+- アート
+- '1.5'
+- 維
+- ▁straightforward
+- ঐ
+- 凌晨
+- 繁殖
+- ধি
+- beamte
+- ▁руках
+- ▁задерж
+- ▁остава
+- 積極的に
+- 匆
+- ▁Interessant
+- ▁законопроект
+- liśmy
+- ▁cantonada
+- ▁مرکز
+- 杭州
+- ▁verurteilt
+- 噴火
+- ▁адказ
+- ▁Nacional
+- expliqu
+- ▁rebuild
+- people
+- ıyordu
+- 눈
+- 설
+- キッチン
+- ▁لطفاً
+- ▁decât
+- ▁일본
+- ▁prodigious
+- ▁Jacob
+- 散歩
+- 传奇
+- ▁klassische
+- ▁существу
+- ▁форма
+- 釣り
+- ▁подпис
+- 风暴
+- ▁Opera
+- ▁институт
+- ическом
+- ▁michigan
+- バッグ
+- ▁clinton
+- それこそ
+- весел
+- 商業
+- ハンド
+- ipps
+- ▁spouse
+- ▁trustee
+- ▁площад
+- ▁uzata
+- صاب
+- geheimnis
+- 披
+- ച
+- ▁sorgfältig
+- 현
+- แต่
+- ▁discreet
+- chirurg
+- といわれる
+- ുന്ന
+- ▁возраст
+- ▁birkaç
+- schirm
+- 环节
+- ▁intact
+- ▁Então
+- طبق
+- 亨
+- forderung
+- 階段
+- 教导
+- auftrag
+- kümme
+- 所需的
+- ▁Jimmy
+- ▁kümmert
+- 대로
+- ▁aquellos
+- እ
+- ▁susceptible
+- 痕迹
+- ▁fuerte
+- トレー
+- ▁invece
+- ложения
+- 静岡
+- kündigt
+- ▁hoffentlich
+- ▁audible
+- 학년
+- ▁Finanzierung
+- າມ
+- ▁simbol
+- rätt
+- ாலும்
+- ▁سخت
+- ▁ĉefa
+- ▁veröffentlichen
+- ▁медицин
+- ▁دوباره
+- アスリート
+- ▁건데
+- ordination
+- あぁ
+- ▁utawa
+- 判明
+- マウンド
+- 木曜日
+- PCR
+- ▁produzieren
+- ▁tactics
+- 可能性もある
+- ிங்
+- هدف
+- artista
+- 違った
+- 弊
+- ▁bijzonder
+- ▁nghệ
+- ▁boulder
+- 逃避
+- 减轻
+- 唉呀
+- ▁Einfach
+- ▁Hütte
+- ▁Feli
+- ▁Charlie
+- 反発
+- ▁navigate
+- 極めて
+- ▁дожд
+- ▁забыл
+- ▁bourgeois
+- ▁steadfast
+- 졌
+- ሆ
+- ▁voulez
+- ▁силы
+- فروش
+- ▁Chief
+- 想一想
+- າກ
+- 항
+- 芸術
+- ▁सु
+- ▁implicit
+- ▁duncan
+- ▁واحدة
+- ▁humming
+- muster
+- 装备
+- ▁membership
+- کشید
+- ▁bequem
+- ▁vieille
+- ▁begleitet
+- ▁empfind
+- ▁果啲
+- ▁impulsive
+- ▁அரச
+- ▁позвонил
+- ▁düster
+- ▁bunlar
+- ▁Offizier
+- ▁دغه
+- 贫穷
+- ▁করতে
+- 多长时间
+- 赞美
+- ▁boost
+- باد
+- ▁успел
+- 没见过
+- 変わらない
+- 肾
+- ▁industrious
+- ▁конфликт
+- ▁беларускай
+- ▁carrière
+- ▁zgod
+- ▁renounce
+- 股份
+- глянул
+- faktor
+- 臨時
+- 拜托
+- building
+- ▁demselben
+- ▁Spiegel
+- ▁enchanted
+- ▁그럴
+- 抚养
+- ▁아예
+- ▁conserve
+- 姐夫
+- ▁erwähnen
+- ▁influential
+- ▁первого
+- ▁кандидат
+- vermögen
+- ▁penitent
+- 受益
+- ▁wiederholen
+- atangiye
+- க்காக
+- 面积
+- ▁aconsegui
+- ▁columbus
+- ▁verpflichtet
+- 貫
+- ▁tournament
+- 令人惊讶
+- ▁hinterlassen
+- ▁servicio
+- ▁словно
+- 地板上
+- ىرى
+- シート
+- キーウ
+- 训
+- ਜ
+- ▁Ayrıca
+- ▁bình
+- ▁resembling
+- ▁birlikte
+- ▁আমার
+- ▁vienna
+- ▁retiring
+- ▁Yagize
+- 贵族
+- ▁mnoh
+- 宿舍
+- 成为一名
+- 投降
+- ▁Zahn
+- ▁Ня
+- عيش
+- ▁fritz
+- крыл
+- ▁execu
+- 乏
+- 瓷
+- ס
+- ▁gobierno
+- ▁westminster
+- ▁Усё
+- 책
+- ▁temporada
+- 隙
+- 昇
+- ಾಗ
+- ▁мужик
+- ლე
+- լ
+- ▁llavors
+- シュー
+- 教師
+- ランチ
+- ▁тэг
+- сурс
+- 早已
+- ▁Bridge
+- ▁geleistet
+- ▁mỗi
+- 水準
+- ▁되지
+- ▁triangle
+- ▁fuori
+- 玛丽拉
+- изова
+- ▁Dichter
+- 異常
+- 炸弹
+- ▁elevator
+- ▁gateway
+- ルーム
+- 观念
+- ▁signify
+- ▁distraction
+- 推广
+- equilibri
+- ▁wunderschön
+- 哑
+- 弗兰克
+- 谣
+- ▁مكان
+- 疑惑
+- ▁Bürgermeister
+- ▁beetle
+- রাজ
+- ことはない
+- verbrauch
+- 風景
+- ▁kaldı
+- zusammenarbeiten
+- ▁appelé
+- kandida
+- ▁compost
+- ▁заўсёды
+- ▁степени
+- ▁erstaunt
+- ▁tödlich
+- 对他来说
+- ▁seguro
+- ッカー
+- 怜
+- を対象に
+- τι
+- झ
+- ▁requiring
+- indirimbo
+- ▁gufata
+- förmig
+- ▁thrice
+- ▁piteous
+- espai
+- 百六十
+- 背叛
+- ▁برخی
+- チェン
+- ▁Prinz
+- 暴風
+- ळ
+- ▁развития
+- ▁хүрэ
+- ▁Firma
+- 报仇
+- ▁chuckled
+- ▁sacrifi
+- されていない
+- お疲れさま
+- ▁Experten
+- ▁республик
+- ▁peninsula
+- 乗用車
+- ▁좋아하
+- ▁parliamentary
+- ල්ල
+- ாமல்
+- 简短
+- ▁forfeit
+- ङ
+- پذیر
+- 畳
+- 冷蔵庫
+- ▁rôle
+- بناء
+- ▁Summe
+- ▁любим
+- ▁spars
+- ▁konkur
+- ើ
+- ែ
+- ੋ
+- ▁Així
+- ▁lấy
+- ▁않았
+- 汀
+- డి
+- 打败
+- ▁unendlich
+- ▁гости
+- ▁сабе
+- ▁tehnolo
+- بێت
+- ▁posibil
+- 揮
+- 逢
+- ▁chuyển
+- 眞
+- ▁Kennedy
+- ▁miliard
+- ▁эфир
+- ọ́
+- ▁метод
+- なりません
+- schäf
+- ▁роль
+- 这项工作
+- ېرى
+- 虐
+- 恭
+- ▁Ukraine
+- ▁gratification
+- ▁सं
+- ěl
+- 另一件事
+- ▁teilweise
+- 新潟
+- 並べ
+- こいつ
+- ġ
+- ▁কিছু
+- 태
+- ▁perchance
+- グッズ
+- ▁transplant
+- ▁impartial
+- 入ってる
+- 小さく
+- んねん
+- 的一件事是
+- ▁lehnte
+- ▁distingu
+- ▁metropolitan
+- 처럼
+- ▁gegessen
+- 呈
+- ▁trouvé
+- ▁recurring
+- お菓子
+- ▁ຫຍັງ
+- ホワイト
+- 담
+- 兜
+- આ
+- 阪
+- 塌
+- 锡
+- ढ
+- २
+- 扛
+- ỳ
+- 雌
+- 忽
+- 偿
+- И
+- 捧
+- 釈
+- 滨
+- ሄ
+- 娇
+- ូ
+- 铭
+- 滩
+- ャ
+- ύ
+- ޯ
+- 斌
+- 절
+- 종
+- 託
+- ޫ
+- 缶
+- 崖
+- ദ
+- 潰
+- 緊
+- ɗ
+- 蔓
+- 仑
+- ஈ
+- 브
+- ৎ
+- 厦
+- 扰
+- អ
+- 벌
+- 증
+- ২
+- ਵ
+- 骚
+- 吨
+- 歓
+- 竖
+- 址
+- 瞥
+- ള
+- 渋
+- 挪
+- 暇
+- 掛
+- յ
+- 铅
+- 钓
+- 橡
+- 拡
+- 狐
+- 줬
+- 출
+- ٽ
+- এ
+- 柿
+- 络
+- 乙
+- ቃ
+- 幾
+- 亜
+- 嗅
+- 咕
+- 喔
+- 畜
+- 茄
+- 글
+- ጠ
+- Б
+- 學
+- 플
+- 勉
+- 咸
+- 锤
+- ð
+- 건
+- 능
+- ఉ
+- 歧
+- 叨
+- ҿ
+- 烫
+- 坤
+- 芦
+- $
+- 監
+- គ
+- 践
+- 침
+- 蟹
+- 唠
+- 합
+- ຜ
+- 堤
+- 肘
+- 宪
+- 임
+- 梳
+- 霞
+- 薛
+- 병
+- 순
+- 攀
+- 驴
+- 灣
+- 甩
+- 邊
+- 妄
+- 론
+- 軸
+- 메
+- 脈
+- 튼
+- 蓉
+- 赐
+- ਮ
+- 葱
+- 魅
+- 盼
+- ံ
+- 秃
+- 甭
+- է
+- 란
+- ぺ
+- ጣ
+- ৱ
+- 窟
+- 靖
+- 颈
+- 壶
+- 꾸
+- 앞
+- ኛ
+- 浆
+- ঙ
+- 仆
+- Д
+- 叉
+- 亀
+- 猴
+- 茎
+- 삼
+- 嬉
+- 澳
+- 颁
+- 笛
+- 커
+- 稻
+- 엔
+- 筛
+- 魁
+- 閣
+- 渣
+- 蹴
+- 搏
+- ੰ
+- 肆
+- ں
+- ਆ
+- 蛙
+- 磊
+- ସ
+- 點
+- 汰
+- 棟
+- 陕
+- 憎
+- 绣
+- 歪
+- 頻
+- 趟
+- 岂
+- О
+- ވ
+- 胞
+- ష
+- 표
+- 噴
+- 堕
+- 圏
+- 咳
+- 淑
+- 庸
+- 疚
+- 嘎
+- ג
+- 滝
+- 譲
+- 炫
+- ପ
+- 蓬
+- 绵
+- 谎
+- ዳ
+- 碍
+- 巷
+- 驰
+- ኔ
+- ጋ
+- ξ
+- 평
+- ű
+- ါ
+- 詳
+- 圭
+- 顽
+- 蹦
+- 枉
+- 頑
+- 慣
+- 鋭
+- 鲸
+- 栖
+- 姫
+- 嫩
+- 큰
+- 渊
+- 掠
+- ಳ
+- ਂ
+- 衔
+- 憋
+- 癖
+- 顧
+- 확
+- 얼
+- 宰
+- 厨
+- ડ
+- բ
+- ဆ
+- 颖
+- ੱ
+- 碌
+- 駐
+- 覇
+- 禅
+- 肿
+- 乞
+- 們
+- 듣
+- ਪ
+- 匿
+- 梗
+- 幹
+- ޖ
+- 俘
+- 접
+- 彪
+- 醋
+- 阅
+- 紋
+- ሉ
+- ୁ
+- 國
+- 袍
+- 癒
+- °
+- 夷
+- 腻
+- Î
+- 愣
+- 堅
+- 賢
+- 沾
+- 贬
+- 绸
+- 먼
+- 맛
+- 灿
+- 끝
+- 吟
+- ണ
+- 駆
+- 镖
+- 활
+- 妆
+- 硕
+- ჯ
+- 硅
+- 몰
+- 纷
+- 彫
+- 渗
+- 陋
+- 賄
+- 陀
+- 셨
+- –
+- ჰ
+- 奎
+- 杏
+- ڑ
+- ମ
+- 译
+- ט
+- 盾
+- 盔
+- 贞
+- 溺
+- 坪
+- 잠
+- 육
+- 쁘
+- 竭
+- 佢
+- 辽
+- 袜
+- 栈
+- ሎ
+- ፈ
+- ೊ
+- Í
+- ፍ
+- 詹
+- 怎
+- 仿
+- 婶
+- 循
+- 백
+- 馅
+- 橙
+- 徹
+- 鍛
+- ሱ
+- 儒
+- 恕
+- 耸
+- ೂ
+- 淳
+- 翌
+- 餌
+- 庇
+- 捞
+- 斐
+- 膏
+- ۵
+- 럽
+- 隅
+- 啪
+- 辐
+- 熙
+- ඔ
+- 紛
+- 捏
+- 떨
+- 손
+- 駒
+- 愚
+- 椎
+- 朽
+- ̉
+- 踪
+- ዚ
+- 帘
+- 嘟
+- 颊
+- 惧
+- ኮ
+- 藩
+- 筑
+- 廉
+- એ
+- ሳ
+- 葵
+- 慕
+- 泛
+- 窒
+- ધ
+- 랬
+- 昧
+- 족
+- 屿
+- 죽
+- 팔
+- ယ
+- 粥
+- ዋ
+- 餅
+- 討
+- 廣
+- 붙
+- 씨
+- 犹
+- 叮
+- 萩
+- 歹
+- 咽
+- 湧
+- 侣
+- 蝶
+- 捆
+- В
+- ڻ
+- శ
+- ખ
+- 懲
+- ؛
+- 哩
+- 尹
+- 웃
+- ỷ
+- 망
+- 즈
+- 贱
+- 瞅
+- 斎
+- П
+- ሪ
+- 졸
+- 貯
+- 郵
+- 频
+- 囚
+- 鲨
+- 凳
+- 缸
+- 짝
+- ۀ
+- ɓ
+- 뭘
+- 홍
+- 옷
+- 쳐
+- 참
+- 痒
+- 혼
+- 眨
+- 揉
+- Ғ
+- 테
+- ծ
+- 咒
+- 绒
+- 厘
+- 변
+- 십
+- 厢
+- 琐
+- ਼
+- ẳ
+- 缴
+- 驼
+- ዲ
+- 嵐
+- 礁
+- 粹
+- 독
+- 俳
+- 妞
+- 쉬
+- 毙
+- ዛ
+- 岐
+- 闇
+- 肮
+- ـ
+- 債
+- 盯
+- 銅
+- 卦
+- 莹
+- ။
+- ቅ
+- 樱
+- 笠
+- 薯
+- 醇
+- 栓
+- ़
+- 崛
+- ካ
+- 刹
+- 奨
+- 迅
+- њ
+- җ
+- 릴
+- 투
+- 谭
+- 俯
+- 帳
+- 帐
+- 瑶
+- 떡
+- ោ
+- 溝
+- ೀ
+- 谍
+- 습
+- 噩
+- ҷ
+- ੁ
+- 淀
+- 勺
+- 簿
+- 曝
+- ւ
+- 饺
+- 棺
+- 斩
+- ።
+- 貝
+- 鼎
+- Ё
+- 뒤
+- Ŝ
+- 별
+- ိ
+- 殻
+- 舗
+- 婿
+- 韵
+- ੈ
+- 충
+- 酔
+- 繋
+- 剛
+- 愤
+- 韧
+- 웠
+- ಣ
+- ฐ
+- 哺
+- 哼
+- ඟ
+- 嗡
+- 緒
+- 姬
+- 慶
+- 匈
+- 懐
+- ഷ
+- 挠
+- 氓
+- இ
+- 桩
+- ۍ
+- ៅ
+- 纲
+- 妾
+- 軌
+- 渴
+- 聘
+- ם
+- 衍
+- 랐
+- ŷ
+- 奄
+- 酋
+- ူ
+- ହ
+- 睹
+- 拌
+- ຶ
+- 绪
+- 瘫
+- 濱
+- 隧
+- 瞄
+- ൂ
+- 耿
+- ൻ
+- 吱
+- 喇
+- ഗ
+- շ
+- 嘻
+- ಷ
+- ಚ
+- 霧
+- 넣
+- 怠
+- 杖
+- 錦
+- 屠
+- 빼
+- 鞠
+- 眺
+- 藻
+- 栅
+- 矛
+- 冊
+- ထ
+- 聆
+- ശ
+- 怡
+- 宵
+- 악
+- Ы
+- 嚷
+- 패
+- ァ
+- 勋
+- 언
+- 慨
+- 赦
+- 萍
+- ռ
+- 막
+- አ
+- ζ
+- 豁
+- 휴
+- 肚
+- 捣
+- ٿ
+- 셔
+- 昨
+- 皓
+- 喃
+- എ
+- 喉
+- 貿
+- 趴
+- 迭
+- 謀
+- 孩
+- ቤ
+- 薦
+- 粛
+- 哒
+- ീ
+- 趾
+- 酿
+- 섯
+- 痘
+- 茜
+- 蚀
+- 菇
+- எ
+- 躬
+- 誌
+- 佬
+- ¿
+- 俵
+- ċ
+- 꺼
+- 渐
+- 卒
+- 鸽
+- 發
+- ళ
+- ហ
+- 亩
+- 唇
+- 壇
+- ಶ
+- 蝇
+- ચ
+- 蜀
+- 隻
+- 俱
+- 泌
+- 剿
+- 磕
+- ቸ
+- அ
+- 衬
+- 처
+- 垮
+- 琉
+- 墅
+- 쯤
+- 畀
+- 険
+- 扁
+- 橱
+- 창
+- 咁
+- 婉
+- 특
+- 沒
+- 倫
+- 噜
+- 钦
+- 傍
+- ӷ
+- 嗓
+- 枫
+- 답
+- ਗ
+- 茸
+- 兑
+- 攒
+- צ
+- 惰
+- 림
+- 숙
+- 榄
+- 氢
+- 喧
+- 览
+- 澜
+- 羅
+- 忏
+- 憩
+- 拱
+- 轿
+- Ú
+- 슬
+- 倩
+- ਅ
+- 剔
+- 距
+- 禀
+- 揍
+- ޒ
+- 嵌
+- 瞬
+- 憲
+- 蹭
+- 凹
+- 판
+- ィ
+- ९
+- 凸
+- 橘
+- 苔
+- 蕉
+- 浇
+- ৯
+- 獄
+- 穂
+- „
+- 贸
+- ʼ
+- 읽
+- 聪
+- 료
+- 옆
+- 乜
+- 골
+- 對
+- 谨
+- 斧
+- 켜
+- 縄
+- 菱
+- '#'
+- 吼
+- 허
+- 嗽
+- 蹄
+- 拇
+- ಜ
+- 該
+- 텐
+- 북
+- 髓
+- ਬ
+- 浄
+- 荘
+- Ô
+- 죠
+- ះ
+- 栃
+- 몸
+- 瓣
+- 莓
+- 굴
+- 塾
+- 遵
+- ן
+- ۹
+- 甸
+- 娟
+- 蠢
+- 함
+- 獣
+- 缅
+- ೋ
+- 틀
+- 각
+- 啫
+- 屯
+- 經
+- Г
+- 餃
+- ੂ
+- 療
+- 绍
+- 槛
+- ჭ
+- ధ
+- 겼
+- 曦
+- 涅
+- 涡
+- 鄙
+- 霖
+- 麟
+- ભ
+- 冥
+- 谅
+- 희
+- 饲
+- 潇
+- 飽
+- 骆
+- 哉
+- ఒ
+- ઓ
+- 萎
+- 膚
+- 斉
+- 皂
+- 屑
+- 悯
+- 衫
+- 鉢
+- 축
+- ሞ
+- 며
+- 법
+- 芙
+- 疤
+- 帜
+- 罕
+- 蝠
+- +
+- 향
+- 寇
+- 丫
+- 힘
+- 朋
+- 拙
+- 荆
+- ஆ
+- 늦
+- 황
+- 撇
+- 택
+- 戳
+- 랜
+- 撰
+- 璃
+- 釜
+- 혀
+- 盏
+- ォ
+- 択
+- 沛
+- 臀
+- 莽
+- ሀ
+- 酮
+- 록
+- 诵
+- 绊
+- 婪
+- ૂ
+- 硫
+- 품
+- 碁
+- 郝
+- 匀
+- 颇
+- 聋
+- 賠
+- 删
+- 阐
+- ណ
+- 妊
+- Х
+- 辟
+- 丞
+- 牡
+- 석
+- 익
+- 噬
+- 拟
+- 瞪
+- 刈
+- 坝
+- 嵩
+- ۳
+- 욕
+- 详
+- ሩ
+- 檬
+- 媚
+- 虏
+- 粪
+- 닐
+- ҽ
+- 蜗
+- 惦
+- ൾ
+- 鐘
+- 淮
+- զ
+- 베
+- 铸
+- 綿
+- 缉
+- 蘑
+- 垄
+- 粤
+- 슷
+- 풀
+- 맨
+- 骸
+- ٬
+- 唬
+- 绞
+- 블
+- 婴
+- ധ
+- 蒲
+- 넌
+- ួ
+- У
+- 综
+- 塊
+- 择
+- խ
+- 膳
+- 蒜
+- 蝙
+- ਇ
+- 嚼
+- 榴
+- 曽
+- 때
+- 枢
+- 於
+- 偵
+- 涩
+- ആ
+- 诡
+- 鳄
+- 矩
+- 溯
+- 贿
+- 검
+- 쓸
+- 칠
+- 厉
+- 責
+- 靓
+- 송
+- 炖
+- 疹
+- 肃
+- 咧
+- ଲ
+- 旱
+- ඹ
+- 魄
+- 哋
+- ○
+- 션
+- 꼬
+- ञ
+- 婢
+- 캐
+- 烘
+- 苛
+- ջ
+- ខ
+- ਟ
+- 肇
+- 栽
+- 熄
+- 馨
+- ળ
+- 紗
+- 채
+- 환
+- 窄
+- 禄
+- 탈
+- 권
+- 腥
+- 噌
+- 祐
+- 妓
+- ୟ
+- 訓
+- 淫
+- 懦
+- 昊
+- 磯
+- 糕
+- 贡
+- 篷
+- ሺ
+- 捍
+- ങ
+- 광
+- 铲
+- ۴
+- 墟
+- 粧
+- 娥
+- ৫
+- ဖ
+- 祁
+- 忆
+- 啸
+- 〈
+- ෘ
+- 懈
+- ಇ
+- 拧
+- 凿
+- 톡
+- չ
+- 녀
+- ڤ
+- 김
+- 汐
+- 糧
+- 谓
+- 瀑
+- ޝ
+- 寓
+- ਣ
+- 扒
+- 衅
+- ఈ
+- ജ
+- ۾
+- 剃
+- 腫
+- 勿
+- ញ
+- 奕
+- 깨
+- 卜
+- 꼭
+- ඕ
+- 砕
+- 澡
+- 嚣
+- 閥
+- Ĝ
+- 邱
+- 簡
+- 飙
+- ৩
+- 芭
+- 떠
+- 庶
+- ×
+- 蛛
+- 麼
+- औ
+- 째
+- 철
+- 锯
+- ぃ
+- 鉱
+- 嗰
+- 鹤
+- 혹
+- 嘶
+- 舆
+- 穀
+- 冗
+- 诫
+- 恤
+- 箸
+- 鎮
+- ז
+- ᱟ
+- 巩
+- 彬
+- 嬢
+- 瘟
+- 诀
+- 埔
+- 砰
+- 舔
+- ષ
+- 밤
+- 詰
+- 顕
+- ዘ
+- 煽
+- 綾
+- 窥
+- 絆
+- 움
+- 宛
+- 撼
+- ಧ
+- 咔
+- 與
+- 陳
+- 芹
+- 陡
+- 掷
+- 廓
+- 逻
+- ബ
+- 唆
+- 寧
+- 怯
+- 鹦
+- 裹
+- 翰
+- ቶ
+- ෞ
+- 벽
+- 奮
+- 너
+- 齋
+- 승
+- ዎ
+- ኩ
+- 뽑
+- ഇ
+- 讳
+- ភ
+- 拽
+- ଦ
+- 못
+- 彤
+- 當
+- 퍼
+- 缪
+- 唾
+- 渦
+- 跤
+- ዝ
+- 掐
+- 峭
+- 梭
+- ށ
+- 兮
+- 窑
+- 應
+- 屉
+- ঢ
+- 伽
+- џ
+- 菩
+- ฮ
+- 潤
+- 辑
+- ៉
+- 께
+- 땐
+- 鈴
+- 暂
+- 廖
+- ॉ
+- ഒ
+- 屡
+- 峻
+- 鹉
+- 熏
+- 鞍
+- 呻
+- 雯
+- 곳
+- 搂
+- 蜡
+- ฉ
+- ৪
+- 깔
+- 說
+- 헤
+- 缆
+- 涯
+- ធ
+- 掀
+- 孽
+- ഹ
+- 奠
+- 련
+- 짐
+- 樊
+- 粋
+- 衙
+- ҕ
+- 녁
+- 叭
+- ڊ
+- 멀
+- 큼
+- 詠
+- 劈
+- 높
+- আ
+- 蕨
+- 掃
+- భ
+- 驯
+- 篠
+- 뜨
+- 痪
+- 窮
+- 騎
+- ຟ
+- ሮ
+- Ř
+- 촌
+- 歉
+- ణ
+- ฝ
+- 잔
+- 沦
+- 绽
+- Τ
+- 樣
+- 暢
+- 폰
+- 窍
+- 條
+- 봉
+- 泵
+- ൽ
+- 빨
+- 암
+- 纺
+- 寛
+- 悉
+- 潭
+- ୋ
+- 焚
+- 畔
+- 嘲
+- 捂
+- 껴
+- 輔
+- 棕
+- 饥
+- ថ
+- 籠
+- ဲ
+- 佣
+- ଏ
+- ڏ
+- 泻
+- 柠
+- ഞ
+- 류
+- ፣
+- ƙ
+- ۽
+- 钞
+- 呜
+- 葫
+- ។
+- 嶽
+- ቢ
+- 濁
+- 仨
+- ӣ
+- 飓
+- ഭ
+- 沐
+- ಆ
+- 邀
+- 赁
+- 冕
+- 哗
+- 禽
+- 酶
+- 侃
+- 锈
+- 呪
+- 熔
+- 揚
+- 奢
+- Э
+- 趋
+- 嗦
+- ផ
+- 衝
+- 窖
+- 阀
+- 扳
+- 摂
+- 응
+- ඊ
+- 쉽
+- 筷
+- 妍
+- ओ
+- 斬
+- 肋
+- 클
+- 胺
+- 亵
+- 叽
+- 咎
+- 桨
+- ሥ
+- 앉
+- 潔
+- ዬ
+- 錯
+- 극
+- 宙
+- 禾
+- ৮
+- 궁
+- 넷
+- 丐
+- 睦
+- ଣ
+- 끄
+- 딩
+- ફ
+- 赋
+- 蘭
+- ۸
+- 논
+- 徊
+- 徘
+- Ш
+- 琦
+- 빌
+- 颤
+- 颂
+- 착
+- 氨
+- ։
+- 썼
+- 擅
+- ‐
+- ぉ
+- 蓮
+- ৬
+- 쿠
+- Κ
+- 釣
+- 捅
+- Â
+- 拢
+- 鵬
+- 叱
+- ३
+- •
+- 嚎
+- 싼
+- ഴ
+- 苑
+- 놨
+- 啓
+- 冨
+- 嘢
+- 鷲
+- ଇ
+- 킨
+- 邵
+- 狮
+- 燥
+- ޕ
+- ಭ
+- 犀
+- 馒
+- 癫
+- 沫
+- 늘
+- 瑜
+- љ
+- 禧
+- 딸
+- 瘤
+- 咖
+- 抒
+- 棠
+- ५
+- ቱ
+- 噂
+- 舵
+- 곤
+- 凄
+- 抠
+- Ņ
+- ሌ
+- 拷
+- 蓋
+- 寮
+- 斋
+- ४
+- ഉ
+- 걱
+- 呦
+- ሬ
+- 俭
+- 蚕
+- 揽
+- 컴
+- 舶
+- 맥
+- 焕
+- 倭
+- 睿
+- 瞳
+- 鹊
+- ಎ
+- ಥ
+- 辻
+- ☎
+- 铛
+- 弔
+- 락
+- 兩
+- 钙
+- 崽
+- Π
+- 钝
+- 낫
+- 밀
+- 銘
+- 縛
+- 殊
+- 藍
+- 俞
+- 惩
+- 刁
+- 튜
+- 褪
+- 脾
+- 谤
+- 跋
+- ८
+- 쌤
+- 획
+- Ф
+- 呕
+- 雁
+- 躯
+- 迦
+- 恳
+- 啱
+- 梵
+- 關
+- 孵
+- 晏
+- 鴨
+- ਉ
+- 棱
+- 疎
+- 恍
+- 匂
+- 咀
+- 濒
+- 渲
+- 绷
+- 냈
+- 淆
+- 胰
+- ሻ
+- 實
+- 팀
+- 坟
+- 龚
+- 쁜
+- 谊
+- 溃
+- ћ
+- 髦
+- 诽
+- 拎
+- 醤
+- 曇
+- 阎
+- 찾
+- 娩
+- ଥ
+- ሲ
+- 渺
+- 骼
+- 蘇
+- ६
+- 啃
+- 揣
+- 椰
+- 嗑
+- 弧
+- 懵
+- 찌
+- 笘
+- 辙
+- 渎
+- 苟
+- ۷
+- ቆ
+- 빵
+- ଗ
+- 汹
+- ۶
+- 檐
+- 喵
+- 騰
+- 窦
+- 歼
+- 葩
+- 범
+- Ș
+- 억
+- 隘
+- 襟
+- 逍
+- 攞
+- 櫻
+- ቻ
+- 锚
+- 赃
+- 喀
+- 璐
+- 蔼
+- ဘ
+- 區
+- 範
+- 绰
+- 沪
+- 碟
+- 沥
+- ଟ
+- 딴
+- 쩌
+- ኧ
+- ७
+- ਖ
+- 胚
+- 篱
+- 幡
+- 嶺
+- 験
+- 掺
+- 蚁
+- ৭
+- ਚ
+- 篤
+- 暫
+- 聡
+- 鷹
+- 茉
+- 氟
+- 叩
+- 擁
+- 掰
+- 嗎
+- 乍
+- Ṭ
+- 溅
+- 煲
+- Ở
+- உ
+- 핑
+- 旷
+- 烬
+- 睫
+- 굉
+- 탄
+- 沧
+- 跷
+- 剰
+- 런
+- ๊
+- ઉ
+- 岔
+- 枣
+- 渕
+- 扼
+- 咏
+- 佟
+- Ā
+- 悚
+- 灶
+- ẵ
+- 靳
+- Ο
+- 醸
+- 褐
+- 愉
+- 媳
+- 筝
+- 觅
+- 彰
+- 逮
+- 逞
+- 矫
+- 殷
+- 센
+- 盎
+- ঃ
+- 巅
+- 聽
+- 澈
+- 亞
+- 姗
+- 赂
+- 鑫
+- 聂
+- 闸
+- Ä
+- 퇴
+- 螃
+- 冉
+- 簧
+- 疟
+- 갑
+- З
+- 倔
+- 绎
+- 翅
+- 搵
+- 僻
+- 眷
+- 藓
+- ଅ
+- ኑ
+- Ǧ
+- ௌ
+- ଯ
+- 萝
+- İ
+- 芥
+- ٻ
+- 壕
+- 謡
+- 媽
+- 긍
+- ձ
+- 褶
+- 鉛
+- 첫
+- 惶
+- 笹
+- 徙
+- 搾
+- Ҳ
+- 걍
+- 膀
+- 揪
+- ໊
+- 逾
+- 왕
+- ഡ
+- 렌
+- 씬
+- 悖
+- 痩
+- 裡
+- 揃
+- 挚
+- 娄
+- 眩
+- 氣
+- ਈ
+- ጀ
+- 膛
+- ଜ
+- ੍
+- ፋ
+- 甥
+- 驸
+- 荧
+- 雞
+- 灸
+- 焉
+- 琅
+- 煌
+- 唧
+- 戮
+- 뛰
+- 쩔
+- 팅
+- 춰
+- 骏
+- 凛
+- 볶
+- 欄
+- 镶
+- 墜
+- 猶
+- 械
+- 酌
+- Α
+- 듯
+- 瑛
+- 滕
+- 챙
+- 镯
+- 蠕
+- 丙
+- 灼
+- 訟
+- 鄂
+- 馈
+- 柬
+- 黏
+- 轶
+- 豫
+- 氯
+- 幫
+- 喚
+- 惚
+- ઈ
+- 渠
+- 镀
+- ୀ
+- 漱
+- Ý
+- ӯ
+- 皈
+- 咗
+- 莺
+- ឹ
+- 柵
+- 碱
+- ጊ
+- 徑
+- 隋
+- 啰
+- ሊ
+- 祠
+- 층
+- 诏
+- 텔
+- 찬
+- 箇
+- 钧
+- 뉴
+- 體
+- 痫
+- 麒
+- փ
+- ᱩ
+- 녔
+- 搓
+- 柚
+- 竿
+- 轄
+- अ
+- ಒ
+- 榨
+- 립
+- 蛤
+- 惭
+- 炳
+- 飢
+- ဟ
+- ቺ
+- 帧
+- 鳞
+- 곱
+- 헬
+- ଆ
+- 锣
+- 倪
+- 鸥
+- 錠
+- ڳ
+- ဝ
+- ๆ
+- 견
+- 琶
+- 饵
+- 將
+- 怂
+- 蔭
+- 蛾
+- 榆
+- 鎌
+- 揾
+- 탕
+- ዐ
+- 倘
+- 秉
+- 賭
+- 菠
+- 瑙
+- 囱
+- 훨
+- 缕
+- ఓ
+- 撩
+- 篡
+- Ī
+- 缀
+- 짓
+- 踝
+- 辫
+- Σ
+- ၏
+- 邢
+- 氮
+- 존
+- 탁
+- 硝
+- 萱
+- 拴
+- 烛
+- ޗ
+- 铝
+- 锥
+- 쇼
+- 暦
+- 焊
+- 릭
+- 腎
+- 瘩
+- 淼
+- Μ
+- 乒
+- 牲
+- 疮
+- 蓓
+- 酪
+- 霄
+- ޢ
+- ኪ
+- 搐
+- 츠
+- 绅
+- ៃ
+- 빡
+- 粟
+- 遏
+- 颅
+- ጉ
+- 링
+- 퓨
+- 磷
+- 耽
+- 祟
+- 瘍
+- 염
+- 謙
+- 霾
+- 臻
+- 猟
+- 诞
+- ៊
+- 蔷
+- 賊
+- 訂
+- 贅
+- Щ
+- 晾
+- Õ
+- ଶ
+- 霆
+- 喪
+- 剖
+- 嗒
+- 樋
+- 疙
+- 靶
+- 纬
+- ਡ
+- 값
+- 娴
+- 貞
+- 擒
+- ൊ
+- 跚
+- 厩
+- 烯
+- 炽
+- आ
+- 駄
+- 侑
+- 袱
+- 픈
+- 玮
+- 朔
+- ኖ
+- ዜ
+- 滔
+- 겁
+- 맘
+- 漬
+- 俣
+- 羔
+- 囤
+- 烷
+- 嗣
+- 峠
+- ǎ
+- 晗
+- 乓
+- 蹒
+- 髄
+- ڙ
+- 睾
+- 兢
+- 咐
+- 铮
+- ೈ
+- 睐
+- 蹊
+- 깝
+- 翘
+- 螂
+- 椭
+- ઇ
+- 鈍
+- ္
+- 參
+- 眯
+- 秩
+- 敛
+- ଡ
+- 鋼
+- 榈
+- 號
+- 掂
+- 磐
+- 萬
+- 裴
+- 阮
+- 蔚
+- 雍
+- 悍
+- 젠
+- ޙ
+- 駿
+- 拂
+- 腊
+- 寞
+- 穹
+- 悴
+- 憔
+- 琥
+- 琵
+- 밑
+- 恥
+- ဂ
+- 垢
+- 茬
+- 坨
+- 遛
+- 涕
+- 掲
+- 咚
+- 覺
+- 嘈
+- 峥
+- 꼈
+- 蝉
+- 麵
+- ٠
+- 땜
+- 祀
+- 譜
+- ൈ
+- ᱱ
+- 량
+- 酥
+- 橇
+- 靡
+- 槐
+- Ấ
+- 錢
+- 槻
+- 暧
+- 侥
+- ଧ
+- 척
+- 턴
+- 吠
+- 甫
+- 壌
+- 갖
+- 鳳
+- ኦ
+- 놔
+- 쓴
+- 寅
+- 麓
+- Ö
+- ഥ
+- 驭
+- ቦ
+- ጅ
+- 낙
+- 鹃
+- Ա
+- 屌
+- 壤
+- Å
+- 낮
+- ဒ
+- 隶
+- 讓
+- 豌
+- 憾
+- 噢
+- 闫
+- ฤ
+- ժ
+- ٔ
+- 岬
+- 萤
+- ӳ
+- ճ
+- ଭ
+- 匣
+- 효
+- 傑
+- 완
+- 瑾
+- 荫
+- ଷ
+- ጨ
+- 액
+- 諭
+- છ
+- 痊
+- ऊ
+- 羹
+- 漩
+- 砾
+- 岚
+- 裳
+- 규
+- థ
+- ψ
+- 콘
+- 惟
+- 렵
+- 짧
+- 钏
+- 窜
+- 臼
+- 闽
+- Ó
+- 內
+- 诬
+- 樂
+- 煞
+- 蝎
+- 弛
+- 從
+- 닭
+- င
+- 缔
+- 岌
+- 怼
+- Ò
+- 巍
+- 곡
+- 啼
+- 桓
+- 谂
+- 賂
+- 鳍
+- 摧
+- ޤ
+- 킹
+- 冀
+- 彗
+- 铐
+- 銚
+- Ṣ
+- 깜
+- 烙
+- 蜱
+- 梶
+- 胴
+- 莞
+- Ọ
+- Ӱ
+- 呱
+- «
+- 狸
+- 瑕
+- 痰
+- 纫
+- 臆
+- ዙ
+- 紊
+- 肴
+- 迂
+- 隼
+- 稜
+- 꿀
+- 茁
+- 淌
+- ഫ
+- 꽤
+- 믄
+- 赣
+- ๋
+- 洽
+- 농
+- 霹
+- 倚
+- 晦
+- 踹
+- ᱠ
+- 덕
+- 嗖
+- 濡
+- 猬
+- 낼
+- 덜
+- Ε
+- 撬
+- 呛
+- þ
+- ጂ
+- 珑
+- 瑠
+- 虜
+- 훈
+- 晤
+- 舘
+- 嗜
+- 渓
+- 瓢
+- 眶
+- 쳤
+- 瞰
+- 腌
+- 勲
+- 蕴
+- 韶
+- 뜻
+- 懿
+- 蕊
+- ቁ
+- 黃
+- 陨
+- 鳌
+- 匕
+- 籽
+- 냄
+- 帶
+- 稠
+- 餓
+- 裾
+- 梧
+- ̇
+- ኬ
+- Ḍ
+- Я
+- ়
+- 닌
+- 荻
+- ਧ
+- 烨
+- 瘸
+- 蜿
+- 齊
+- 團
+- 姻
+- ឡ
+- 瞻
+- ฎ
+- 捎
+- ሜ
+- 嚇
+- ଙ
+- 殉
+- ቼ
+- 厥
+- ਫ
+- ጎ
+- 酗
+- 哔
+- 刨
+- 侨
+- 痹
+- 섭
+- 웬
+- 骡
+- 汝
+- 蜒
+- 桦
+- 嘣
+- 冶
+- 峙
+- 樽
+- 變
+- 湛
+- 雳
+- 呸
+- 數
+- 檀
+- 辍
+- 笋
+- ៀ
+- 崭
+- 꿈
+- 藉
+- ຝ
+- 悄
+- օ
+- ଳ
+- ಉ
+- 롱
+- 쌓
+- ॅ
+- 엽
+- 총
+- 寥
+- ՝
+- 롤
+- 沌
+- 允
+- 咄
+- 撂
+- ኋ
+- 苇
+- 賣
+- 엄
+- 恙
+- 碾
+- 蜷
+- 닥
+- 觀
+- 겹
+- 茵
+- 蹑
+- 吝
+- 璇
+- ໋
+- ័
+- 墩
+- 猥
+- 널
+- 잤
+- 玷
+- 薫
+- 蹬
+- 弈
+- ፊ
+- 俐
+- 頬
+- 秤
+- ଛ
+- ၊
+- 쎄
+- 獅
+- 橫
+- ך
+- 吏
+- 抨
+- 鯉
+- 腱
+- ඛ
+- 缰
+- 秽
+- 렇
+- 锄
+- 呉
+- ሙ
+- 찰
+- 苯
+- 람
+- 摁
+- 幣
+- 圍
+- 俏
+- 撵
+- Ś
+- 혜
+- 炕
+- ଚ
+- 猝
+- 쭉
+- 穏
+- 珂
+- ਭ
+- 멋
+- 喳
+- 狡
+- 嘀
+- 韬
+- 瞭
+- 惺
+- ಖ
+- 끌
+- 쌍
+- 讽
+- 缎
+- 庐
+- 拣
+- 槟
+- 侶
+- 嘆
+- 뻔
+- 單
+- 處
+- 殆
+- 噛
+- 庚
+- 뻐
+- 즐
+- 梓
+- 램
+- 빈
+- 雏
+- 諮
+- 嚏
+- 득
+- 蜥
+- 뿌
+- 鳩
+- 谕
+- 匮
+- 婊
+- 匙
+- 炬
+- 싱
+- 楓
+- 畸
+- 頓
+- ஒ
+- 噪
+- 犁
+- 懊
+- 谚
+- 畴
+- <
+- ቡ
+- 讐
+- 鮭
+- 꿔
+- 隷
+- 梢
+- 鯛
+- 坷
+- 晟
+- 簸
+- 疵
+- 閲
+- 낸
+- 컨
+- 흥
+- 眸
+- 侏
+- 臺
+- '>'
+- 짤
+- 胱
+- Γ
+- 沁
+- 墳
+- 襄
+- 蛎
+- 빙
+- 矣
+- 湊
+- ಈ
+- 믿
+- 잉
+- ෛ
+- 蟑
+- 畠
+- 셋
+- 槿
+- 嬛
+- 狈
+- 羚
+- 隈
+- 泞
+- 荐
+- 吭
+- 剁
+- ٘
+- 炙
+- 蟆
+- 牽
+- 翡
+- 튀
+- ଁ
+- ᱤ
+- 嗷
+- 绮
+- 簇
+- 娱
+- 幌
+- 犊
+- 漓
+- ቴ
+- 牟
+- 昕
+- 념
+- 诋
+- 俸
+- 翩
+- 钥
+- 淤
+- 玻
+- 绥
+- 칼
+- 澎
+- ٫
+- 攘
+- 溉
+- 瀚
+- 庵
+- ઘ
+- 塀
+- 茧
+- 哆
+- 롯
+- 挿
+- 殡
+- 荃
+- 釉
+- 窯
+- 伶
+- 爺
+- 瞑
+- ڌ
+- 厮
+- 汶
+- 풍
+- 팬
+- 兒
+- 絹
+- 薩
+- 祷
+- 慑
+- 憂
+- ฏ
+- 嫉
+- 輩
+- 拯
+- 淇
+- 馋
+- ᱭ
+- 啄
+- 猖
+- 絮
+- Δ
+- 穗
+- 녹
+- 맡
+- 켓
+- Ē
+- 辗
+- ఖ
+- 嫣
+- Η
+- ዊ
+- 轉
+- 惮
+- 傀
+- 儡
+- 啬
+- 颓
+- 渭
+- 筐
+- ଉ
+- ዱ
+- 샤
+- ሸ
+- ឺ
+- 鞘
+- 驿
+- 푸
+- 桅
+- 浙
+- 俾
+- 叼
+- ቂ
+- 愕
+- 聲
+- 넓
+- 赘
+- 蛊
+- 儲
+- 岖
+- 죄
+- È
+- 蛍
+- 朦
+- ዕ
+- 貂
+- 뀌
+- 폭
+- 權
+- 渉
+- 跺
+- 酝
+- ጭ
+- 焙
+- 渇
+- 筱
+- ඝ
+- 撸
+- 洼
+- 脐
+- 曖
+- 옮
+- ኳ
+- 翟
+- 쟁
+- 蘸
+- 稔
+- 紳
+- 喱
+- 蚤
+- 雹
+- 鞅
+- 鲤
+- ☆
+- 妬
+- 蹈
+- 捋
+- 矜
+- ђ
+- 털
+- 潼
+- ኒ
+- ዶ
+- 袒
+- 멘
+- 閑
+- 陌
+- অ
+- 玺
+- 羁
+- 卤
+- 渝
+- 戎
+- ෆ
+- ჟ
+- 坍
+- 邑
+- 昵
+- 忒
+- 砌
+- 磋
+- 汲
+- 웨
+- 젤
+- 珈
+- 綻
+- 镰
+- 暁
+- 敖
+- 槌
+- 踱
+- 堺
+- ᱮ
+- 摞
+- ቹ
+- 钳
+- 倶
+- 诛
+- 寐
+- 邹
+- 缭
+- 솔
+- 撲
+- 醍
+- 樓
+- ः
+- 胧
+- 讪
+- 醐
+- ݨ
+- ඨ
+- 蜴
+- 毗
+- 虞
+- 떼
+- 롭
+- 聯
+- 钠
+- 峯
+- 璋
+- 賓
+- 淵
+- ～
+- 蜚
+- ޚ
+- 湃
+- ̍
+- ಫ
+- 亢
+- 腑
+- 숨
+- ౌ
+- 枷
+- 檜
+- 谬
+- 岑
+- 袄
+- 铀
+- Ì
+- 鲑
+- \
+- 杵
+- 韭
+- 콜
+- 烹
+- ඬ
+- ኸ
+- ਥ
+- 킬
+- ඥ
+- 熨
+- 擂
+- 赡
+- 滿
+- ሷ
+- 摯
+- 诠
+- 匡
+- 谟
+- 瞩
+- 腓
+- 黯
+- ڼ
+- 帚
+- 뿐
+- 엑
+- 흐
+- ଂ
+- 顷
+- ឆ
+- 潦
+- 탔
+- ኘ
+- 앙
+- 箔
+- 똑
+- 匾
+- 징
+- ሴ
+- ጃ
+- 邝
+- 딜
+- ဉ
+- 漪
+- 胁
+- 羲
+- 耙
+- 馀
+- 訊
+- 氪
+- 瘠
+- ١
+- 닝
+- 젊
+- 흔
+- 邉
+- 骤
+- 슨
+- 骰
+- 滇
+- 繕
+- 辦
+- ஃ
+- 抉
+- 渍
+- 彝
+- 曰
+- 顎
+- 랩
+- 슈
+- 씻
+- 냉
+- 율
+- 縣
+- 询
+- 럴
+- Մ
+- 亥
+- 樟
+- 樫
+- 畫
+- 홀
+- 尧
+- 骄
+- 饷
+- 걘
+- 얻
+- 浊
+- 령
+- 씀
+- ぅ
+- 鐵
+- 蔗
+- 遼
+- 谛
+- 杭
+- 毋
+- ၀
+- 땡
+- ሠ
+- 憤
+- 涟
+- 芜
+- 鸳
+- 콩
+- 椿
+- 怦
+- 鸯
+- ઝ
+- ၎
+- ऐ
+- 擬
+- 酯
+- 灘
+- 钮
+- 痢
+- 醫
+- 餸
+- 꽃
+- 凱
+- 忡
+- 酎
+- 냥
+- 吮
+- 弩
+- 續
+- 荔
+- 狳
+- 產
+- ጆ
+- 犰
+- 骇
+- ଖ
+- 噼
+- 紡
+- 줌
+- 짱
+- 趕
+- 峨
+- 컬
+- 粑
+- 攸
+- ŋ
+- 嗱
+- ኤ
+- 槙
+- Ҫ
+- ዓ
+- 몬
+- 拗
+- 營
+- 帥
+- 曙
+- ጫ
+- 锻
+- 浚
+- 幂
+- 俑
+- 笃
+- 鬱
+- 廿
+- 祺
+- 夠
+- 杞
+- 钾
+- ݙ
+- 舷
+- 購
+- 瘀
+- 萃
+- 蜓
+- 즌
+- 筏
+- 涤
+- 曳
+- ฆ
+- 랄
+- 蒼
+- 忑
+- ៍
+- 胭
+- ቲ
+- ቷ
+- 嬷
+- 釧
+- 鳗
+- 忐
+- 鼹
+- 緻
+- 轼
+- 벤
+- 傳
+- 悸
+- 렀
+- 訣
+- 暉
+- 锌
+- 哧
+- 娲
+- 禹
+- 窘
+- 蜻
+- Ł
+- 嚓
+- 눠
+- 덴
+- 똥
+- 춤
+- 唄
+- 諾
+- 楔
+- 薗
+- 錮
+- 楷
+- 遡
+- ڀ
+- 폴
+- 鱿
+- 嗝
+- 델
+- ഈ
+- 뚫
+- 켰
+- 柑
+- 啡
+- 漕
+- 凪
+- 蹋
+- ഏ
+- ─
+- 煩
+- 객
+- 專
+- 篝
+- Ż
+- 깊
+- 댓
+- 웹
+- ੜ
+- 迢
+- 钰
+- 芊
+- 薰
+- ੌ
+- ᱢ
+- 孜
+- Ķ
+- 渤
+- 绯
+- 碇
+- 諏
+- Ĵ
+- 帷
+- 涸
+- 蟋
+- 芮
+- 邋
+- 懇
+- 擎
+- 馍
+- 掖
+- 鯨
+- ዩ
+- 燒
+- 钵
+- ૃ
+- 瞌
+- ഖ
+- 沽
+- 蟀
+- ќ
+- 锂
+- 荚
+- 鼬
+- 덟
+- 딘
+- 쌀
+- 蟒
+- 嫦
+- 藝
+- 姊
+- 률
+- ఐ
+- 漉
+- 誕
+- 寫
+- 憬
+- 萦
+- 균
+- 낀
+- 압
+- 옥
+- 卉
+- 焰
+- 偎
+- 殃
+- 肛
+- 噔
+- 鹭
+- ٺ
+- 壺
+- 蝦
+- 멍
+- 윤
+- 컵
+- 픽
+- Ê
+- 榊
+- 總
+- 灏
+- 噺
+- 紺
+- 捨
+- 厕
+- 盪
+- 脯
+- 驹
+- ᱜ
+- 崴
+- 榛
+- 邯
+- 鬟
+- Հ
+- ऱ
+- 섬
+- 왠
+- 竺
+- 勅
+- 栾
+- 譬
+- 琛
+- 蕃
+- 홉
+- 骁
+- 栩
+- 攥
+- 俪
+- 刽
+- 笙
+- 遢
+- 삶
+- 걷
+- 눌
+- Š
+- 菁
+- 輕
+- 妳
+- 汾
+- 磺
+- 醺
+- ሶ
+- ፕ
+- ᱚ
+- 댄
+- 셀
+- 템
+- 憨
+- 錬
+- 喙
+- 氦
+- 땅
+- 릉
+- 肽
+- 颯
+- 黔
+- 锵
+- 릿
+- 밍
+- 绛
+- ヂ
+- 嫖
+- 炜
+- 瘪
+- 臊
+- 벗
+- 봄
+- 쟤
+- 폐
+- 蒿
+- 엘
+- 詣
+- 羨
+- 캠
+- 孰
+- 铬
+- 恺
+- 恢
+- 佰
+- 蚱
+- 渥
+- 纂
+- 纶
+- 벅
+- 叡
+- 捶
+- 綺
+- 眾
+- 憧
+- ऑ
+- ڄ
+- 昴
+- 誹
+- 謗
+- 棣
+- 汕
+- 沏
+- ᱫ
+- 빛
+- 馁
+- 鵜
+- 涮
+- ঈ
+- ଼
+- 앤
+- 玟
+- 芷
+- 妒
+- 柊
+- 琊
+- 竣
+- 斟
+- 騙
+- Έ
+- ዮ
+- 瀧
+- 艮
+- ဗ
+- 糙
+- 闵
+- ጓ
+- 腳
+- 蜕
+- 韻
+- 뮤
+- ඵ
+- 낭
+- 룸
+- 蝗
+- 蹂
+- ਏ
+- 腮
+- 럭
+- 쁠
+- ඡ
+- 诧
+- 镁
+- 鄉
+- 鍾
+- 窿
+- 蚪
+- 蝌
+- 疡
+- ዴ
+- 砥
+- 뷰
+- 엉
+- 皙
+- 撅
+- 犟
+- 娯
+- 掴
+- 둥
+- 헐
+- 혈
+- 砺
+- ਯ
+- 麹
+- ฬ
+- 籔
+- 럼
+- 捻
+- 벨
+- 홈
+- 嫡
+- 睛
+- ぢ
+- 遷
+- 伎
+- ᱾
+- 哽
+- ឈ
+- 堰
+- 磡
+- 焱
+- 翎
+- 矗
+- 翊
+- Ν
+- ೃ
+- ၁
+- 腋
+- 舊
+- 豬
+- 꼐
+- 썰
+- 펜
+- 坞
+- ஏ
+- 糯
+- ޞ
+- 倣
+- 凋
+- 엠
+- 헌
+- 흘
+- 諜
+- ヅ
+- 頸
+- ሂ
+- ፎ
+- 꼼
+- 둔
+- 삭
+- ♫
+- 貓
+- 踵
+- 惋
+- 價
+- 歡
+- 昱
+- 浣
+- 讹
+- 喆
+- 擔
+- ף
+- 龛
+- 艶
+- 苣
+- 涓
+- 躏
+- 窪
+- 屹
+- 恬
+- 裘
+- 糗
+- 绚
+- 錄
+- 霓
+- 噱
+- 껄
+- 槍
+- 蹩
+- '~'
+- 孢
+- 춘
+- 힐
+- 햄
+- 夯
+- 潟
+- 漾
+- 偃
+- 咣
+- 癞
+- 訃
+- 딨
+- Ն
+- 찜
+- 촬
+- 춥
+- 罠
+- 麸
+- 獭
+- 镳
+- 铠
+- ጡ
+- 吒
+- 삐
+- 잃
+- 찐
+- 틱
+- 핀
+- 皐
+- 闺
+- 榎
+- 娼
+- ୂ
+- ឋ
+- 阂
+- 몽
+- Ļ
+- Ս
+- ృ
+- 詫
+- 铎
+- 僅
+- 菓
+- ៏
+- 卯
+- 噎
+- 榕
+- 馄
+- 樵
+- 랙
+- ᱛ
+- 괴
+- 낳
+- 힌
+- 疱
+- 舜
+- 祛
+- 箍
+- 劉
+- 枸
+- 盹
+- 觑
+- ൃ
+- 戯
+- 驗
+- 례
+- 핫
+- Җ
+- Қ
+- 醛
+- 嘩
+- 慵
+- 섞
+- 큐
+- 팩
+- ઠ
+- ឌ
+- 圖
+- 藕
+- 蚓
+- 呑
+- 喩
+- 곧
+- 닫
+- 밝
+- 앨
+- 毡
+- 跛
+- 秧
+- 嗮
+- 褛
+- 讷
+- 툰
+- '@'
+- 虱
+- 蟾
+- 壱
+- 晖
+- 碜
+- 嫔
+- Ă
+- Ģ
+- ٢
+- 鋳
+- 협
+- 颚
+- 踮
+- 酰
+- 鳟
+- 禍
+- 阜
+- 挛
+- 箕
+- ᱧ
+- 腩
+- 꽂
+- 쏘
+- 탑
+- 碘
+- 잖
+- 彷
+- 鸠
+- 缇
+- 绢
+- ฑ
+- 猕
+- 밴
+- 퀴
+- 戟
+- 巳
+- 隕
+- 啤
+- 擢
+- 睽
+- 辘
+- 痉
+- ፖ
+- 熠
+- 鞑
+- 饨
+- ሐ
+- 褴
+- 阈
+- 묻
+- 烏
+- 酚
+- 惣
+- ຣ
+- 阑
+- 犒
+- 碉
+- 龈
+- 靼
+- ਘ
+- 蛰
+- 阙
+- 놈
+- 땠
+- 鬃
+- 또
+- 哝
+- Ồ
+- 낄
+- 싹
+- 첨
+- 蔻
+- 鑼
+- 毀
+- 痣
+- 釘
+- 偕
+- 湍
+- 燎
+- 钗
+- 咫
+- 덩
+- 箋
+- 罹
+- 孚
+- 宓
+- ඤ
+- ቋ
+- 燈
+- 왼
+- 꿨
+- 흰
+- ၍
+- 庾
+- 盡
+- 躇
+- 鎧
+- 幢
+- 浒
+- 胤
+- 荤
+- 鼾
+- 孀
+- 蹿
+- 裝
+- 獾
+- 皖
+- 蛀
+- 蚯
+- 颐
+- 좌
+- 棲
+- ^
+- 獗
+- 睑
+- 苞
+- 鳕
+- 腦
+- 뒷
+- Ј
+- 楞
+- ଠ
+- ኞ
+- ዞ
+- 납
+- 떴
+- 쩍
+- 斷
+- 憶
+- 泸
+- 勸
+- 铰
+- 洱
+- 緯
+- 겸
+- ૌ
+- 粽
+- 咤
+- 宕
+- 쿨
+- ֆ
+- ڃ
+- ‟
+- 鞦
+- 빤
+- 뼈
+- 슴
+- 엇
+- 칸
+- 莘
+- 挝
+- 韆
+- 夭
+- 腆
+- Β
+- 뚱
+- 칭
+- 閃
+- 蜍
+- 謹
+- 迄
+- 坯
+- 傣
+- 谏
+- 祯
+- 羯
+- 鹑
+- ሃ
+- Բ
+- ഓ
+- 寶
+- 戲
+- 轧
+- 拮
+- 椛
+- 佃
+- 蠣
+- 礎
+- 囧
+- 丕
+- 摒
+- 榻
+- ヱ
+- 扈
+- 薙
+- 렁
+- ‧
+- 啷
+- 舉
+- 侬
+- 邃
+- 垛
+- 怔
+- 闰
+- ፓ
+- 혔
+- ⁄
+- 氰
+- 끗
+- 덮
+- 證
+- 虻
+- 俚
+- 壑
+- 瞿
+- 藥
+- 衩
+- ѳ
+- 喎
+- 醜
+- 갠
+- 뒀
+- 멜
+- 컸
+- 핸
+- 뚝
+- 晌
+- අ
+- ೌ
+- 扪
+- '|'
+- 荨
+- 깎
+- 짬
+- 雙
+- 髅
+- 圃
+- 蓟
+- 鹫
+- 榮
+- 绫
+- 藐
+- 贻
+- ኢ
+- 亂
+- 禮
+- 賜
+- 샵
+- 측
+- ฒ
+- 胥
+- 骷
+- 앴
+- 褂
+- 較
+- 恃
+- 垦
+- 麝
+- 诲
+- 뜬
+- 갚
+- 턱
+- ፅ
+- 媲
+- 馏
+- 駕
+- 톤
+- 닮
+- 壽
+- 剽
+- 篓
+- 馊
+- 鹂
+- 圓
+- 壹
+- 잊
+- Ե
+- ٩
+- 亘
+- 缽
+- 翱
+- 貪
+- 얀
+- 펴
+- ঔ
+- 罂
+- 鳖
+- 黝
+- 汞
+- 痺
+- 佯
+- 稷
+- 恪
+- 彌
+- 砦
+- ጪ
+- 爭
+- 찔
+- 痞
+- 喰
+- 狒
+- 戬
+- 簪
+- 憑
+- ‒
+- 嗌
+- 嗤
+- 囉
+- 諒
+- 뚜
+- 팡
+- 罵
+- 楊
+- 噶
+- 凜
+- ≪
+- 〜
+- 匐
+- 芈
+- 鹌
+- 诃
+- 鳃
+- 】
+- 涎
+- 腼
+- Є
+- ጤ
+- 泷
+- 漸
+- 蚌
+- 끈
+- 땀
+- 띠
+- 爾
+- 脓
+- 髋
+- ૈ
+- 廳
+- 虔
+- 缜
+- 楢
+- 褥
+- 暄
+- 禺
+- 怅
+- ፒ
+- 阉
+- 끓
+- 렉
+- 벼
+- 뽀
+- 틴
+- 팠
+- Ӹ
+- 褚
+- 啜
+- 雖
+- ୱ
+- 【
+- 槲
+- ଫ
+- 雉
+- 馥
+- 烊
+- ୃ
+- 摹
+- 羟
+- 騨
+- 琢
+- 钊
+- ൺ
+- ٥
+- ඈ
+- 꼴
+- 댔
+- 豐
+- 廻
+- 沂
+- 啾
+- 埼
+- 徨
+- 剐
+- 唏
+- 轲
+- 牒
+- ဏ
+- 븐
+- 샘
+- 據
+- 峪
+- 錫
+- 밖
+- 漳
+- 炅
+- 荼
+- ၂
+- 玖
+- 絢
+- 镍
+- 沓
+- 뺐
+- 칙
+- 桝
+- 啕
+- 紐
+- 讀
+- 凰
+- 麥
+- 吆
+- 淞
+- 瓮
+- 耦
+- 佘
+- 喋
+- 孺
+- 螨
+- ៈ
+- 苫
+- 闩
+- Θ
+- 醬
+- 뷔
+- 篆
+- 鉾
+- 蔬
+- 桔
+- 锢
+- 讣
+- 辕
+- 鸵
+- 꽁
+- Ԑ
+- 赈
+- 遁
+- 隨
+- 掸
+- 暨
+- 陇
+- 宦
+- 戛
+- 睬
+- ፡
+- 嘚
+- 歲
+- 汛
+- 렛
+- 씹
+- 잇
+- 嚴
+- 檢
+- 흑
+- 麋
+- 佼
+- 랫
+- 鮎
+- 脖
+- 歷
+- 勞
+- 楂
+- 蚂
+- Կ
+- 兀
+- ሏ
+- 舛
+- 녕
+- 뇌
+- 릇
+- 짠
+- 힙
+- 冴
+- 蕎
+- 崗
+- 膿
+- 侮
+- 茛
+- 攪
+- 逅
+- 匍
+- 邂
+- 瘁
+- Ի
+- ڇ
+- ឃ
+- △
+- ঋ
+- ਝ
+- 曬
+- 멈
+- 묵
+- 컷
+- 혁
+- 휘
+- 깃
+- 椋
+- ヵ
+- 潺
+- 螳
+- 鳅
+- ಘ
+- ဇ
+- 迸
+- 拄
+- 浃
+- 钛
+- 낚
+- 얇
+- 헷
+- 흡
+- ਛ
+- 搀
+- 렴
+- 黜
+- 闆
+- 惬
+- 帛
+- 戰
+- ৷
+- ଘ
+- €
+- 蜢
+- 鲈
+- 遐
+- 뽕
+- 캡
+- 砧
+- 碼
+- 邨
+- ឱ
+- 溥
+- 咿
+- 锹
+- 砚
+- 熹
+- Ả
+- ‚
+- 珺
+- 紬
+- Ẹ
+- 凧
+- 戊
+- 곰
+- 畢
+- 鱗
+- 苹
+- 笞
+- 箫
+- Դ
+- 瀕
+- Ū
+- ፉ
+- 谩
+- 겉
+- 쭈
+- 탐
+- Ό
+- 廟
+- 怆
+- 絕
+- 꽉
+- 鬣
+- 塵
+- 羡
+- Λ
+- 댕
+- 뜯
+- ጽ
+- 稱
+- 覗
+- 꾼
+- 劃
+- 卻
+- 栞
+- 雛
+- 跆
+- 抿
+- 잘
+- 啧
+- 俨
+- 汴
+- 賽
+- 叟
+- Ҙ
+- ଓ
+- 渾
+- 糾
+- 긁
+- "\x93"
+- Ġ
+- ǹ
+- ٰ
+- ឿ
+- 泯
+- 泾
+- 묘
+- 쪘
+- 쫓
+- 婧
+- 籁
+- 淄
+- 痨
+- 弑
+- 忱
+- 淨
+- 縞
+- 酣
+- 繼
+- Ι
+- ѓ
+- ଞ
+- 壞
+- 겪
+- 烽
+- 묶
+- 썹
+- 풋
+- 宍
+- 苷
+- 靜
+- 羌
+- 矯
+- Χ
+- 啮
+- 胯
+- 賺
+- 吖
+- 냅
+- 붕
+- 쉴
+- ૉ
+- 獨
+- 덤
+- 詮
+- 臃
+- 焖
+- 獒
+- 紹
+- 諗
+- 岱
+- 璨
+- 讃
+- 둬
+- 璧
+- 痔
+- 冽
+- 舀
+- 弼
+- 徇
+- 綠
+- 繭
+- 镣
+- 驮
+- ऋ
+- 搡
+- 搪
+- 砷
+- 닷
+- 닿
+- 땄
+- 욱
+- 웅
+- 찝
+- ဦ
+- 噗
+- 醚
+- 谑
+- 紘
+- 炷
+- 枭
+- 罔
+- 蚣
+- 颌
+- 璞
+- 닦
+- ĺ
+- ሕ
+- 馴
+- 긋
+- 랍
+- 咨
+- 粱
+- ໆ
+- 蛭
+- 骋
+- 囡
+- ץ
+- 劵
+- 濟
+- 纣
+- 荟
+- 蛟
+- 뀐
+- 밟
+- Պ
+- 膠
+- ៗ
+- ൗ
+- 狛
+- 琰
+- 畝
+- 觐
+- 擀
+- 蒔
+- 蚜
+- 귄
+- 빅
+- 쫄
+- 촉
+- 螈
+- 蚝
+- 郸
+- 飴
+- 鸞
+- 娓
+- 皎
+- 眈
+- 捺
+- 砒
+- 糞
+- 茱
+- 裆
+- 갤
+- 뺏
+- 쭐
+- 쿄
+- 팟
+- 赝
+- 犸
+- 蜊
+- 惆
+- 瘴
+- 笆
+- 讥
+- 钨
+- 驷
+- 宸
+- 戾
+- 賑
+- ڍ
+- ሯ
+- ጦ
+- 랭
+- 룩
+- 뺄
+- 샐
+- 숫
+- 팝
+- Ո
+- 桟
+- 罄
+- 槃
+- 捗
+- 轱
+- 濕
+- 谙
+- ሦ
+- 榔
+- 溫
+- 璀
+- 诣
+- 늙
+- 썩
+- 쫙
+- 톱
+- 爛
+- 铂
+- 奧
+- 鹈
+- 赳
+- 膵
+- 渚
+- 缈
+- 耘
+- 唰
+- 綴
+- 豺
+- 龊
+- 龌
+- ٣
+- ඌ
+- 犷
+- 葆
+- 颞
+- 馳
+- 릎
+- 숭
+- 쌩
+- 썸
+- 祇
+- 險
+- 蝓
+- 鴻
+- 熾
+- 蛞
+- 沱
+- 潍
+- 堇
+- ਐ
+- 尬
+- 粕
+- 辄
+- 껍
+- 넨
+- 룹
+- 샌
+- 쌌
+- 좁
+- 핵
+- Գ
+- 榉
+- 臥
+- 鹕
+- 蟻
+- 咝
+- 缥
+- ∞
+- 刍
+- 惘
+- 牦
+- 絲
+- ጁ
+- 屬
+- 讧
+- 돋
+- 맙
+- ౦
+- ᱞ
+- 啟
+- 뺀
+- 埗
+- 悶
+- 毂
+- 贋
+- 顆
+- 鲟
+- 孬
+- 昙
+- 薮
+- 壓
+- 狀
+- 缤
+- 藜
+- 鹽
+- 麩
+- 갓
+- 멤
+- 믹
+- 뱅
+- 붓
+- 윗
+- 쩐
+- 췄
+- 斓
+- 莴
+- 펙
+- ヲ
+- 釋
+- 袅
+- 躊
+- 阚
+- 囔
+- 铆
+- Ț
+- ሑ
+- ጌ
+- 兎
+- 戍
+- 涝
+- 诟
+- 铿
+- 쨌
+- 쩨
+- Ĥ
+- Ӧ
+- ፀ
+- 冑
+- 廠
+- 앱
+- 秆
+- 糠
+- 鮫
+- 桧
+- 垩
+- 耷
+- 镐
+- 엊
+- 夙
+- 宥
+- 濠
+- 繍
+- ઢ
+- 佗
+- 戶
+- 皑
+- 蝾
+- 쫍
+- 莆
+- 饉
+- 懷
+- එ
+- 啁
+- 赊
+- 鸪
+- 挞
+- 鹧
+- ඓ
+- 蔫
+- 뜩
+- 띄
+- 략
+- 뤄
+- 맵
+- 폼
+- 甬
+- 烩
+- 碴
+- 钍
+- 掇
+- 閒
+- 奘
+- 羿
+- 趙
+- Ō
+- Ά
+- ഘ
+- ၉
+- ቪ
+- 弋
+- 斡
+- 衲
+- 辿
+- 쥐
+- 쪄
+- 珞
+- 瓯
+- 壬
+- 鬧
+- 忖
+- ¥
+- ဓ
+- 肓
+- 薅
+- 靈
+- 靛
+- ఘ
+- ಏ
+- 轭
+- 윙
+- ÿ
+- 亟
+- 呲
+- 咻
+- 硒
+- 絨
+- 魯
+- 껌
+- 덥
+- 듬
+- 빽
+- 飒
+- 迥
+- Վ
+- 彙
+- 즘
+- ϊ
+- 蜈
+- 嗲
+- 觎
+- 轟
+- 귤
+- 낯
+- 쌈
+- 찢
+- 쾌
+- 팍
+- ဥ
+- 揄
+- 煦
+- 熵
+- 淺
+- 玫
+- 哐
+- 藪
+- Ñ
+- Φ
+- Ԥ
+- ٤
+- 埠
+- 拈
+- 炯
+- 굽
+- 뻤
+- 뿔
+- 셜
+- Į
+- Թ
+- 珏
+- 疽
+- 缨
+- 揶
+- 囃
+- 梱
+- 餡
+- 鰹
+- 燻
+- Ħ
+- 撥
+- 爻
+- 脫
+- 雫
+- 霁
+- ኙ
+- 擇
+- 踌
+- 鬓
+- 겐
+- 쩡
+- 텀
+- 텍
+- ਓ
+- 뭉
+- 햇
+- 鯖
+- 髭
+- 傭
+- 蛐
+- 鰻
+- 遜
+- ഠ
+- ›
+- 怄
+- 裟
+- 啩
+- 恣
+- 斛
+- 檻
+- 谔
+- 踞
+- 닉
+- 맹
+- 잼
+- 詐
+- 櫃
+- 浏
+- 綬
+- 攫
+- 冢
+- 綜
+- 芃
+- 歆
+- 殇
+- 鄭
+- 鲱
+- 黨
+- ፃ
+- ᱨ
+- Ụ
+- ‹
+- 滓
+- 濤
+- 갱
+- 앗
+- 쵸
+- 캔
+- 抡
+- 輿
+- 擊
+- 苓
+- 歎
+- 幔
+- 愫
+- 萼
+- 檸
+- 嵇
+- 薏
+- 蘿
+- Ӓ
+- ፌ
+- 딪
+- Ք
+- ሔ
+- 憐
+- 洸
+- 燧
+- 珩
+- 껏
+- 셉
+- 즉
+- 펌
+- 珐
+- 瘙
+- 粵
+- 膽
+- 荏
+- 镊
+- 劾
+- 妲
+- 缮
+- 靚
+- 鳏
+- Ţ
+- ఠ
+- ቄ
+- 昀
+- 椀
+- 袤
+- 遽
+- 흠
+- Џ
+- 仃
+- 婵
+- 诩
+- 鍼
+- 쉐
+- 탱
+- 奚
+- 饯
+- 桢
+- 樺
+- 锭
+- 蚬
+- 跻
+- ޠ
+- 膺
+- ԑ
+- ޏ
+- ဤ
+- 哂
+- 攰
+- 葦
+- 錐
+- "\x94"
+- ỵ
+- 靭
+- 뺑
+- 엥
+- 탠
+- 帼
+- 〇
+- 姪
+- 晰
+- 摺
+- ឬ
+- 恻
+- 惡
+- 蹉
+- Ã
+- ၅
+- 궈
+- 멸
+- 빴
+- 쇠
+- 푼
+- Տ
+- 巌
+- 掳
+- 撚
+- 膈
+- 蛹
+- 胛
+- 舫
+- 缚
+- 舐
+- 鲫
+- 荞
+- 豉
+- 叻
+- 骛
+- 龋
+- 糜
+- 둑
+- 뢰
+- 륵
+- 얹
+- 煉
+- 痤
+- 蝽
+- 煨
+- 膦
+- 嚿
+- 蛆
+- 蟲
+- 钴
+- 顯
+- '}'
+- ᱴ
+- 歸
+- 湄
+- Ő
+- Լ
+- Ռ
+- ٦
+- 銷
+- 깥
+- 뱃
+- 嶙
+- 鄱
+- 錣
+- 烃
+- 謁
+- 炀
+- 潢
+- 煜
+- 娑
+- 痿
+- ၆
+- ቨ
+- 跎
+- 꼽
+- 댁
+- 밭
+- 섹
+- 숏
+- 쎈
+- 쥬
+- 츄
+- 칵
+- 콤
+- ૅ
+- 瑰
+- 襷
+- 楕
+- 瞠
+- 酉
+- 鹬
+- 叵
+- 婕
+- 庖
+- 觊
+- 謂
+- 闊
+- ٨
+- 偌
+- 挎
+- 锏
+- 룰
+- 뭣
+- 봇
+- 빔
+- 융
+- 찼
+- 퉁
+- 헛
+- 汎
+- 蟠
+- ゞ
+- 箏
+- 峋
+- 堑
+- 痍
+- 纥
+- 勵
+- 粼
+- 钚
+- 퀄
+- ᱥ
+- 窩
+- 螯
+- 髻
+- 鱲
+- ′
+- 巽
+- 谄
+- 꼰
+- 뇨
+- 뜰
+- 볍
+- 빚
+- 툭
+- 펑
+- ਊ
+- 涣
+- 淬
+- 稅
+- 魇
+- 姣
+- 疣
+- 胫
+- 酊
+- 檗
+- 猾
+- 跄
+- 踉
+- 瓒
+- 骊
+- 鬆
+- ґ
+- 咦
+- 肱
+- 臉
+- 鞄
+- 넉
+- 빗
+- 뻥
+- 얄
+- 엎
+- 칩
+- 훔
+- Ə
+- ઊ
+- 泱
+- 狞
+- 윈
+- 欽
+- 涧
+- 匝
+- 籌
+- 鲶
+- 嘗
+- 鰂
+- 鹩
+- 濮
+- 姦
+- 恿
+- 袈
+- £
+- 撐
+- 曉
+- 聰
+- 蔥
+- 郴
+- ʿ
+- ໌
+- ၈
+- ጄ
+- ឥ
+- 戌
+- 蕙
+- 蠻
+- 贮
+- 铵
+- 깄
+- 앵
+- 혐
+- 檎
+- 緋
+- 桉
+- 骞
+- 坳
+- 箴
+- 桀
+- 鏑
+- 瀛
+- 礴
+- 芪
+- ঊ
+- 倖
+- 谵
+- '{'
+- Þ
+- ၄
+- 喹
+- 燔
+- 芍
+- 詢
+- 遙
+- 깡
+- 뀔
+- 넥
+- 젝
+- 핏
+- 횟
+- Օ
+- ઞ
+- 洩
+- 颧
+- 燭
+- 뭐
+- ঞ
+- 繰
+- 呷
+- 鲻
+- ங
+- 鳝
+- 鹪
+- 偻
+- 珥
+- 铡
+- ಞ
+- 戀
+- 狰
+- 璜
+- 纭
+- 蝈
+- ሟ
+- ‽
+- 敝
+- 砝
+- 삿
+- 샷
+- 쏠
+- 쿼
+- 揸
+- 锰
+- 掟
+- 葭
+- 鸚
+- 謳
+- 罷
+- 湮
+- 蜃
+- Œ
+- ಠ
+- 诙
+- ਠ
+- ಐ
+- 厭
+- 咛
+- 掻
+- 揿
+- 纏
+- 荀
+- 菏
+- 蓦
+- 袂
+- 镫
+- 뱀
+- 뱉
+- 짰
+- 嘔
+- 歐
+- 礒
+- 葡
+- 鋸
+- 匯
+- 滉
+- 倹
+- ៌
+- 擺
+- 痱
+- 篑
+- 遲
+- 鹳
+- 嬴
+- 惴
+- 鰺
+- ၃
+- 彈
+- 滟
+- 簾
+- 苋
+- ٧
+- ਢ
+- ၌
+- ★
+- ゑ
+- 窈
+- 葺
+- 谪
+- 랗
+- 맣
+- 틈
+- 耆
+- 苅
+- 铉
+- 鹋
+- 妩
+- 塹
+- 牆
+- 獐
+- Չ
+- ଵ
+- 怵
+- 獻
+- 硌
+- 磴
+- ୌ
+- 绌
+- 굿
+- 껀
+- 띵
+- 뻘
+- 슐
+- 쩜
+- 툴
+- 椽
+- 濂
+- 焗
+- 裱
+- 챔
+- 穩
+- 茗
+- 橿
+- 镑
+- 脷
+- 錆
+- 寰
+- 阡
+- 忻
+- 矾
+- 镭
+- 骜
+- 駛
+- 詔
+- 냬
+- 뉘
+- 듀
+- 횡
+- ȋ
+- ቧ
+- ᱷ
+- 厲
+- 屐
+- 徵
+- 谗
+- 貰
+- 렬
+- 쿵
+- 唸
+- 嚥
+- 塙
+- 澪
+- 櫓
+- 燦
+- అ
+- І
+- 噚
+- 榷
+- 孪
+- 揀
+- 擞
+- 纜
+- Ø
+- Ρ
+- ⠀
+- 伢
+- 缄
+- 翳
+- 蛳
+- 諫
+- 谥
+- 賦
+- 雜
+- 鮑
+- Ζ
+- Ї
+- Ֆ
+- ٪
+- 丟
+- 獎
+- 秸
+- 郦
+- 隱
+- 깼
+- 므
+- 팽
+- 푹
+- 仄
+- 廈
+- 杳
+- 涞
+- 淅
+- 袴
+- 繳
+- 撫
+- 嘤
+- 笺
+- 髮
+- 穣
+- 纾
+- 鲭
+- ኃ
+- 垠
+- 牠
+- 睨
+- ၇
+- ጻ
+- 佞
+- 櫛
+- 虛
+- 鲷
+- 갇
+- 똘
+- 쇄
+- 쑥
+- 젖
+- 탭
+- 훌
+- 淚
+- 贏
+- 镂
+- 阖
+- 夥
+- 滲
+- 簽
+- 菰
+- 鹹
+- 绗
+- 牝
+- 霏
+- 逵
+- 鹜
+- 鸸
+- 噏
+- 忤
+- 瞓
+- 啖
+- Ω
+- ՛
+- ޟ
+- ሹ
+- ᱵ
+- 殼
+- 礫
+- 簌
+- 蛔
+- 蛯
+- 铤
+- 곽
+- 귈
+- 됩
+- 벳
+- 숱
+- 숲
+- 첩
+- 탓
+- 틸
+- 펼
+- Ҷ
+- ጸ
+- 鋪
+- 쁨
+- 饽
+- 忿
+- 濾
+- 畿
+- 澁
+- 穢
+- 矶
+- 绉
+- 嚢
+- 犄
+- 筵
+- 茴
+- 莅
+- Ջ
+- ଢ
+- 嗔
+- 诓
+- 隍
+- 셰
+- ŕ
+- ఛ
+- 悌
+- 槇
+- 蘋
+- 鵝
+- 녜
+- 맻
+- 벙
+- 섰
+- 썬
+- 쏟
+- 퀘
+- 흉
+- Ď
+- Ь
+- 寵
+- 廢
+- 祓
+- 麽
+- 蹼
+- 鋒
+- 槭
+- 懋
+- 窕
+- ̂
+- 嵜
+- 掣
+- 灞
+- 襯
+- 邈
+- 锱
+- 魷
+- Շ
+- ޘ
+- ቫ
+- 峽
+- 碓
+- 謠
+- 遴
+- 颦
+- 멕
+- 킥
+- ዥ
+- ឧ
+- 叢
+- 鍮
+- 黐
+- 갸
+- 뎅
+- 옵
+- 훠
+- 覽
+- 暹
+- 甾
+- 疸
+- 鴎
+- 缛
+- 烁
+- 猁
+- 猞
+- ឯ
+- 佝
+- 欖
+- 瀝
+- 罡
+- 萸
+- 藿
+- ̄
+- 咂
+- 螢
+- 铢
+- 頚
+- 饗
+- 鸢
+- ઑ
+- Ừ
+- 嗬
+- 嘥
+- 嚕
+- 爲
+- 纨
+- 겟
+- 굔
+- 냠
+- 콕
+- 텝
+- 훅
+- Ң
+- 悻
+- 潛
+- 苺
+- 藁
+- 鶯
+- 黍
+- 넛
+- 濫
+- 鄰
+- 蠅
+- 襖
+- 枇
+- 盧
+- 廚
+- 褓
+- 賤
+- 幄
+- 栀
+- 盃
+- 鑊
+- 珅
+- 绾
+- 蔦
+- 遨
+- Ҩ
+- ۂ
+- 嬲
+- 繫
+- 膊
+- 룬
+- 맺
+- 옴
+- 쭤
+- 팁
+- 폈
+- 瑪
+- 癸
+- 삽
+- 첼
+- 锑
+- 龜
+- 嵯
+- 氚
+- 蓼
+- ຼ
+- 珪
+- 揖
+- 瑚
+- 膻
+- 霎
+- 飕
+- Ή
+- Խ
+- 峦
+- 廁
+- 蘆
+- 衢
+- 蹟
+- 锉
+- ሼ
+- 啵
+- 恆
+- 焯
+- 诿
+- 깅
+- 뛸
+- 륨
+- 밸
+- 쉰
+- 텅
+- 壯
+- 夾
+- 峒
+- 蝋
+- 餘
+- 勁
+- 哏
+- 檄
+- 赭
+- 毓
+- 燗
+- 鮨
+- 榭
+- 啶
+- 梆
+- 嵘
+- 轢
+- 嗟
+- ጮ
+- 嬰
+- 捱
+- 蹶
+- 꺾
+- 텨
+- Ӑ
+- ኗ
+- ዤ
+- ዷ
+- ፤
+- 冚
+- 搖
+- 楣
+- 浔
+- 瞟
+- 诨
+- 骥
+- 닙
+- 젓
+- 펐
+- 繞
+- 鹸
+- 瀞
+- 燮
+- 苜
+- 湎
+- 靱
+- 閻
+- 杷
+- 臧
+- 噻
+- 囫
+- 溴
+- 阄
+- Ҵ
+- 簑
+- 薔
+- 蠔
+- 頁
+- ፐ
+- 〔
+- 檔
+- 閱
+- 겄
+- 괄
+- 굶
+- 귓
+- 깠
+- 꽈
+- 넜
+- 럿
+- 옹
+- 욜
+- 쬐
+- 漑
+- 潸
+- 唷
+- 笈
+- 鱈
+- 蓿
+- 剌
+- 酩
+- 佇
+- 唑
+- 嶼
+- 钒
+- 胝
+- 胼
+- 蛱
+- 圩
+- 礙
+- 趨
+- Υ
+- 櫂
+- 玥
+- 瑄
+- 绺
+- 蔔
+- 鸬
+- 鹚
+- Æ
+- ୈ
+- 墾
+- 搶
+- 盅
+- 绡
+- 蚩
+- 閪
+- "\x9E"
+- ­
+- ̈
+- ۖ
+- ଝ
+- ሾ
+- ዪ
+- 琏
+- 굵
+- 눅
+- 늬
+- 됨
+- 붉
+- 샴
+- 씌
+- 얜
+- 옳
+- 욘
+- 傥
+- 蜣
+- 迴
+- 鱧
+- 唢
+- 殒
+- 菫
+- 沣
+- 爐
+- 泗
+- 揆
+- 靉
+- 倏
+- 疥
+- 卞
+- 噉
+- 囵
+- 殘
+- 氙
+- 腚
+- 銮
+- ᱣ
+- 慘
+- 搔
+- 犠
+- 盥
+- 綫
+- 蒐
+- 褲
+- 訝
+- 辊
+- ©
+- ®
+- ቭ
+- 〕
+- 吡
+- 啉
+- 痂
+- 觞
+- 貘
+- 鷺
+- 눕
+- 늑
+- 늫
+- 렷
+- 벚
+- 뻗
+- 얌
+- 얗
+- 왓
+- 짖
+- Ư
+- 唁
+- 圀
+- 鐔
+- 륜
+- 뻑
+- 쓱
+- 왤
+- 滘
+- 濑
+- 岷
+- 疋
+- 蓑
+- 譚
+- 铯
+- 毘
+- 諦
+- 襁
+- 讴
+- 鄞
+- 緣
+- 膘
+- 禰
+- 泮
+- 璎
+- 莊
+- 蔺
+- 裨
+- 陂
+- 馗
+- ڦ
+- ಃ
+- ጩ
+- 渫
+- 溧
+- 獠
+- 祢
+- 诌
+- 赅
+- 괌
+- 렘
+- 렙
+- ኚ
+- Ủ
+- 兇
+- 彿
+- 荠
+- 谀
+- 댈
+- 룡
+- 륙
+- 및
+- 뿜
+- 셈
+- 읍
+- 찡
+- 毬
+- 辋
+- 箩
+- 饅
+- 拚
+- 紆
+- 葚
+- 儺
+- 籾
+- 菀
+- 呤
+- 煸
+- 琲
+- 胍
+- 玳
+- 谴
+- 镉
+- 俎
+- 洵
+- 锲
+- 颉
+- 僱
+- 柘
+- 栎
+- 疝
+- 萢
+- 鑽
+- 骶
+- 〆
+- 儘
+- 汩
+- 腭
+- Փ
+- 佈
+- 掮
+- 梏
+- 歙
+- 毽
+- 涿
+- 矬
+- 엮
+- Û
+- Ѐ
+- ዣ
+- ᱡ
+- ᱦ
+- 喬
+- 嫲
+- 嬌
+- 懶
+- 筲
+- 糅
+- 辭
+- 霭
+- 낑
+- 뎌
+- 뛴
+- 봅
+- 솜
+- 엣
+- 왁
+- 찹
+- 칫
+- 懼
+- 禎
+- ဌ
+- 娆
+- 鬚
+- 荥
+- 笏
+- 嶌
+- 癣
+- 攣
+- 鍬
+- 嘌
+- 捌
+- 孑
+- 淦
+- 瑁
+- 硼
+- 擾
+- 泓
+- 閾
+- 楯
+- 蝕
+- ゐ
+- 哙
+- 姝
+- 孖
+- 盂
+- 胄
+- Ɓ
+- ଃ
+- ഊ
+- ዌ
+- ፆ
+- 倜
+- 卅
+- 卍
+- 柩
+- 鲅
+- 갛
+- 껑
+- 껜
+- 륭
+- 뭇
+- 슝
+- 싯
+- 쏴
+- 잎
+- 콧
+- 팜
+- 펀
+- ඞ
+- 婀
+- 톨
+- 洙
+- 硤
+- 梠
+- 锆
+- 筧
+- 鵤
+- 菖
+- 邬
+- 軋
+- 栉
+- 忪
+- 桎
+- 筠
+- 脍
+- 锃
+- 佔
+- 儆
+- 掬
+- 旮
+- 荊
+- ᱹ
+- 隽
+- 饴
+- 훑
+- Ċ
+- Է
+- ቬ
+- ጧ
+- 奂
+- 滷
+- 癮
+- 蝼
+- 슥
+- 쏙
+- 왈
+- 팥
+- 핥
+- Ը
+- 乸
+- 擤
+- 縱
+- 铍
+- 멧
+- Ў
+- 嘹
+- 埴
+- 悅
+- 欅
+- 谒
+- 鴛
+- ឲ
+- 嘭
+- 箪
+- 鴦
+- 恫
+- 覃
+- 穎
+- 郜
+- 韋
+- 僭
+- 痙
+- 邁
+- 哌
+- 疇
+- 惇
+- 侗
+- 箒
+- 埂
+- 讶
+- 邺
+- 鲹
+- 稞
+- 蒡
+- 賴
+- Զ
+- 绶
+- 贖
+- 铱
+- Ҭ
+- 亳
+- 坩
+- 柒
+- 纰
+- 觸
+- Ɗ
+- ፏ
+- ⋯
+- 劑
+- 擴
+- 殲
+- 溏
+- 茯
+- 깍
+- 붐
+- 뻣
+- 샹
+- 줍
+- 쯔
+- 펠
+- 敕
+- 깬
+- 꾹
+- 뮬
+- 빰
+- 숍
+- 즙
+- 쭝
+- 쾅
+- 퀸
+- 킵
+- 펭
+- 헝
+- 俥
+- 滂
+- 瘘
+- 캘
+- 嗪
+- 衞
+- 睢
+- 铣
+- 韮
+- 翦
+- 娠
+- 旛
+- 翫
+- 蕈
+- 譯
+- 吽
+- 囿
+- ͘
+- 讚
+- 钹
+- Ƙ
+- ǔ
+- Ұ
+- 阕
+- 阱
+- 麿
+- 쫑
+- 쿡
+- ဧ
+- ሣ
+- 嬤
+- 懺
+- 晁
+- 殓
+- 滾
+- 苻
+- 钯
+- 饬
+- 껐
+- 꿋
+- 덧
+- 뵈
+- 엌
+- 잌
+- Ë
+- 沅
+- 瑩
+- 撻
+- 笥
+- 尕
+- 簀
+- 竈
+- 縷
+- 鮪
+- 糀
+- 謄
+- 侘
+- 鰭
+- 氩
+- 籐
+- 舖
+- 皺
+- 伫
+- 弭
+- 欸
+- 泙
+- Ύ
+- 祚
+- 缢
+- 聒
+- 诰
+- 鄢
+- 鴫
+- 鹄
+- 齡
+- Ξ
+- ಓ
+- ᱰ
+- 垓
+- 燉
+- 筈
+- 贊
+- 鬥
+- Ĩ
+- Ź
+- ഐ
+- ඖ
+- ᱲ
+- 侩
+- 惱
+- 杓
+- 涪
+- 漿
+- 炆
+- 觥
+- 颍
+- 餵
+- 麾
+- 겜
+- 괘
+- 꿍
+- 돕
+- 돗
+- 딲
+- 릏
+- 맑
+- 뵙
+- 솥
+- 앚
+- 쫀
+- 쭘
+- 펄
+- Ũ
+- Ӗ
+- ┐
+- 榧
+- 癜
+- 缱
+- 饕
+- 驅
+- 鲇
+- 젯
+- 퍽
+- 갯
+- 憫
+- 爍
+- 鄧
+- 萘
+- 廼
+- 吁
+- 茲
+- 鉤
+- 埚
+- 墉
+- 慳
+- 泔
+- 猢
+- 瑭
+- 旌
+- 孱
+- 屆
+- 弐
+- 珉
+- 祗
+- 薑
+- Ŵ
+- ʽ
+- Њ
+- ѐ
+- ഛ
+- ቾ
+- Ị
+- 唞
+- 囁
+- 搽
+- 旯
+- 绻
+- 腧
+- 膣
+- 謊
+- 谆
+- 谌
+- 蹤
+- 钽
+- 靂
+- 뚤
+- 몫
+- 삥
+- 웜
+- 줏
+- 쩰
+- 츤
+- 캉
+- 큘
+- 팸
+- 펍
+- 펫
+- 헨
+- 崑
+- 瑤
+- 痧
+- 轍
+- 顛
+- 飚
+- 鵑
+- 龅
+- 롬
+- 잣
+- 횐
+- ಛ
+- ඍ
+- 滯
+- 薹
+- 譴
+- 桿
+- 氤
+- 蹙
+- ÷
+- 鵯
+- 臘
+- 鲼
+- 甑
+- 鯵
+- 嬬
+- 婺
+- 杈
+- 鬶
+- 鲠
+- 鳶
+- 嬸
+- 骅
+- ઋ
+- 啐
+- 嗫
+- 尴
+- 徉
+- 抻
+- 煬
+- 瓊
+- 祜
+- 虢
+- 鈿
+- 俶
+- 倌
+- 撳
+- 棗
+- 樸
+- 珲
+- 癬
+- 笪
+- 錶
+- 삘
+- 씽
+- ኡ
+- ፔ
+- 剱
+- 彎
+- 獰
+- 甕
+- 綁
+- 腍
+- 芡
+- 薈
+- 蜆
+- 逑
+- 겔
+- 곈
+- 뤘
+- 뾰
+- 옇
+- 윽
+- 잭
+- 텃
+- 텼
+- 픔
+- 兖
+- 勳
+- 擰
+- 朧
+- 桤
+- 睏
+- 迩
+- 흙
+- 舩
+- 訛
+- 馮
+- 撈
+- 攬
+- 祎
+- 饒
+- 儚
+- ឍ
+- 纈
+- 纐
+- 莳
+- 氘
+- 鑓
+- 葳
+- 莪
+- 儂
+- 繇
+- 苒
+- 恸
+- 舢
+- 刎
+- 徜
+- 桠
+- 繹
+- 芫
+- 杢
+- 榫
+- 氲
+- 睜
+- 箜
+- 篌
+- 貅
+- 閖
+- ᱼ
+- 屙
+- 敘
+- 暈
+- 梾
+- 籤
+- 謬
+- 낌
+- 씁
+- 씸
+- Ĕ
+- Ŭ
+- Ӡ
+- ቿ
+- 傢
+- 呟
+- 悭
+- 溟
+- 璽
+- 瓴
+- 绔
+- 芩
+- 貔
+- 酞
+- 釀
+- 鍊
+- 鲃
+- 깁
+- 돔
+- 둡
+- 랴
+- 썪
+- 짭
+- 짼
+- 퀵
+- 폿
+- 홋
+- ఞ
+- 潞
+- 癲
+- 鋲
+- 캄
+- 퐁
+- 瀾
+- 訶
+- 贄
+- ಢ
+- 羰
+- 羸
+- 麂
+- 
+- 剋
+- 滁
+- 瑳
+- 谶
+- 荸
+- 좀
+- 碲
+- 楳
+- 鲳
+- 煕
+- 戇
+- 溲
+- 膑
+- ޣ
+- 壷
+- 擠
+- 聿
+- 伉
+- 滦
+- 睥
+- 繩
+- 脘
+- 荽
+- 崙
+- 攤
+- 柾
+- 砀
+- 籮
+- 蠡
+- 谧
+- ഔ
+- 挲
+- 晔
+- 琮
+- 瘋
+- 藨
+- 钣
+- Ơ
+- Ҟ
+- ಔ
+- ዑ
+- ጵ
+- Ứ
+- ‫
+- ∈
+- 凇
+- 懑
+- 掙
+- 揼
+- 涜
+- 炔
+- 繪
+- 腟
+- 芾
+- 錘
+- 頒
+- 驟
+- 鹞
+- 꽝
+- 넬
+- 눴
+- 뜸
+- 밉
+- 뵀
+- 삔
+- 샜
+- 셌
+- 쑤
+- 엿
+- 콸
+- Ҥ
+- Ҽ
+- ဿ
+- 捩
+- 禿
+- 竇
+- 譽
+- 郫
+- 둠
+- 뒹
+- 렐
+- 맴
+- 뽈
+- 첸
+- 拋
+- 淙
+- 盞
+- 丶
+- 寬
+- 獵
+- 窰
+- 舳
+- 註
+- ઃ
+- 诅
+- 閤
+- 鴇
+- 嘧
+- 慄
+- 攝
+- 蝨
+- 鰯
+- 貶
+- 臾
+- 笕
+- 頷
+- 镌
+- 竪
+- 噤
+- 诘
+- 锗
+- 閘
+- 嗚
+- 壩
+- 撺
+- 晷
+- 桡
+- 棹
+- 耧
+- 趸
+- →
+- 妁
+- 牯
+- 瓿
+- 笄
+- 蛄
+- 豈
+- 铖
+- 骐
+- 鷉
+- 鸾
+- 屍
+- 楸
+- 踽
+- 锒
+- 鲲
+- 섀
+- 켄
+- Ľ
+- ӊ
+- ጇ
+- ጬ
+- ◎
+- 嚨
+- 姒
+- 蒺
+- 蝸
+- 輻
+- 鸨
+- 齟
+- 깰
+- 끽
+- 낵
+- 눔
+- 닳
+- 밋
+- 밲
+- 벡
+- 뺨
+- 쉘
+- 슉
+- 쌔
+- 짚
+- 촛
+- 춧
+- 캣
+- 캥
+- 튕
+- 휙
+- ਔ
+- ఢ
+- ೧
+- ᱯ
+- 偲
+- 劍
+- 枱
+- 膩
+- 艷
+- 菸
+- 詛
+- 豇
+- 낡
+- 됬
+- 둣
+- 쌰
+- 撷
+- 贰
+- 躙
+- 辯
+- 遑
+- Յ
+- 楮
+- 誼
+- 瞞
+- 祿
+- 绦
+- 廬
+- 皋
+- 妝
+- 鸮
+- 떤
+- 圳
+- 捲
+- 陝
+- 獺
+- 媾
+- 魍
+- 鼩
+- 鴈
+- ゝ
+- 狲
+- 釐
+- 铄
+- 沮
+- 蘼
+- 邕
+- 钎
+- 靥
+- 鞣
+- ♥
+- 姘
+- 娣
+- 稹
+- 胳
+- 郅
+- 阆
+- 颱
+- 餮
+- ־
+- 儋
+- 厝
+- 肅
+- 誊
+- 騷
+- 숟
+- "\x92"
+- "\x97"
+- Ճ
+- ՞
+- ኜ
+- ዢ
+- ዦ
+- Ố
+- 嘍
+- 揩
+- 曆
+- 栢
+- 潋
+- 箬
+- 糍
+- 遞
+- 髡
+- 鰐
+- 鲡
+- 鲢
+- 齬
+- 걀
+- 꺄
+- 꿉
+- 뗄
+- 빕
+- 얽
+- 읏
+- 잰
+- 쟀
+- 컥
+- 킷
+- 햐
+- 흩
+- ඣ
+- ፑ
+- ៣
+- 寳
+- 淖
+- 灑
+- 錨
+- 駁
+- 랠
+- 썽
+- 웰
+- 젬
+- 탬
+- 툼
+- 핍
+- 檳
+- 籃
+- 僑
+- 橼
+- 脹
+- 銛
+- 钜
+- 盱
+- 籲
+- 陲
+- 颔
+- 勖
+- 蓖
+- 郢
+- 檯
+- 粂
+- 獏
+- 燿
+- 祕
+- 鯊
+- 枡
+- 惕
+- 兌
+- 钼
+- 鞆
+- ●
+- 綦
+- 蕗
+- 埜
+- 焘
+- 劭
+- 愎
+- 橈
+- 鎚
+- 锷
+- 鞏
+- 齁
+- 龇
+- ဍ
+- ♂
+- 侪
+- 窠
+- 肪
+- 蜇
+- 逹
+- 邛
+- 卟
+- 撓
+- 燐
+- 纖
+- 郓
+- 闱
+- 餞
+- 鹓
+- 俬
+- 呔
+- 澧
+- 燴
+- 犍
+- 羧
+- 葶
+- 谲
+- ¬
+- Ć
+- ː
+- ̓
+- ڱ
+- ۚ
+- ඪ
+- ፁ
+- ᱪ
+- Ẩ
+- 呎
+- 哣
+- 噃
+- 搣
+- 淒
+- 苕
+- 萋
+- 襪
+- 訇
+- 諄
+- 謢
+- 邳
+- 鉈
+- 鴉
+- 鸻
+- 갭
+- 곶
+- 넹
+- 뗐
+- 룽
+- 맷
+- 샾
+- 쐈
+- 쨍
+- 챠
+- 컹
+- 튠
+- 푠
+- ˮ
+- Ί
+- Ҿ
+- ޛ
+- ዟ
+- 悗
+- 耒
+- 躾
+- 鏟
+- 閂
+- 곁
+- 맏
+- 뮌
+- 찻
+- 睪
+- 礦
+- 筺
+- 艱
+- 賬
+- 镕
+- 蕕
+- 炝
+- 聾
+- 逡
+- ฌ
+- 瀋
+- 詭
+- 鲣
+- 侉
+- 埙
+- 慟
+- 蜉
+- 钡
+- 輯
+- 諧
+- 吳
+- 鐙
+- 陛
+- 撹
+- 苧
+- 劔
+- 濛
+- 齧
+- Ժ
+- 剝
+- 吩
+- 徕
+- 镓
+- ൌ
+- 佻
+- 嚅
+- 岘
+- 窨
+- 跬
+- 銜
+- 骢
+- 魉
+- 鰓
+- 㩒
+- 嬗
+- 旎
+- 旖
+- 氫
+- 洄
+- 牺
+- 篙
+- 舂
+- 闌
+- 飄
+- Ծ
+- ḿ
+- 仞
+- 嘯
+- 噓
+- 囹
+- 圄
+- 岿
+- 恁
+- 揦
+- 殚
+- 沆
+- 簕
+- 莠
+- 莼
+- 萊
+- 鰆
+- ɔ
+- ≡
+- 砣
+- 辇
+- 궐
+- 뽐
+- "\x84"
+- "\x9A"
+- §
+- Ť
+- Ձ
+- ۓ
+- ጢ
+- ፂ
+- ፄ
+- Ổ
+- Ờ
+- 僆
+- 棂
+- 獸
+- 瓏
+- 瘢
+- 聳
+- 荛
+- 衿
+- 遒
+- 鑰
+- 镬
+- 깽
+- 꺠
+- 끙
+- 늪
+- 댐
+- 딥
+- 딧
+- 랖
+- 룻
+- 릅
+- 봔
+- 봬
+- 뼛
+- 섣
+- 쉼
+- 싣
+- 쎘
+- 얏
+- 윌
+- 쥴
+- 쯧
+- 챈
+- 챌
+- 촥
+- 톰
+- 핬
+- 圪
+- 屜
+- 捽
+- 珮
+- 碛
+- 臟
+- 螫
+- 輋
+- 醪
+- 骠
+- 똠
+- 맸
+- 샬
+- 沔
+- 禊
+- 襦
+- 趄
+- 邇
+- 剜
+- 筍
+- 緞
+- 虧
+- 趔
+- 铩
+- 埒
+- 憚
+- 扦
+- 罟
+- 囮
+- 賈
+- 匁
+- 吲
+- 哚
+- 雎
+- 鄒
+- 飫
+- 縊
+- 讼
+- 濯
+- 竊
+- 铕
+- 蓣
+- ๅ
+- 潑
+- 瀉
+- Ґ
+- ಊ
+- 屓
+- 碣
+- 粳
+- 苈
+- 蕤
+- 誨
+- 跖
+- 騾
+- 魑
+- ⸺
+- 倅
+- 幇
+- 廂
+- 柺
+- 楡
+- 瀨
+- 狆
+- 籏
+- 籬
+- 跣
+- 锶
+- 镆
+- 韌
+- ΐ
+- 佚
+- 汜
+- 牍
+- 牴
+- 癱
+- 蟬
+- 顏
+- ϋ
+- ዉ
+- ▪
+- 佥
+- 刿
+- 噙
+- 孭
+- 洮
+- 漲
+- 猷
+- 瓤
+- 疊
+- 癡
+- 矍
+- 硯
+- 稃
+- 讫
+- 贔
+- 꾀
+- 떵
+- 맬
+- 빳
+- 챕
+- "\x91"
+- "\x96"
+- ƴ
+- ؔ
+- ઔ
+- ෲ
+- ኣ
+- ១
+- 啞
+- 夘
+- 嵋
+- 徬
+- 慾
+- 斂
+- 渌
+- 滌
+- 漯
+- 燶
+- 砵
+- 稟
+- 笤
+- 蓆
+- 螞
+- 覓
+- 諺
+- 诳
+- 谡
+- 踎
+- 躉
+- 鈔
+- 铋
+- 陞
+- 顼
+- 麪
+- 갬
+- 궜
+- 궤
+- 꿇
+- 덱
+- 떳
+- 띨
+- 룐
+- 몹
+- 삑
+- 슁
+- 쏭
+- 앰
+- 욤
+- 웩
+- 잦
+- 죙
+- 챘
+- 첵
+- 촘
+- 쿤
+- 킴
+- 텁
+- 퓸
+- 훼
+- ઍ
+- 冧
+- 勐
+- 擸
+- 晉
+- 汨
+- 燊
+- 瘡
+- 癇
+- 眀
+- 鎹
+- 铷
+- 鯰
+- 鱒
+- 딤
+- 앓
+- 얍
+- 윷
+- 쟈
+- 팎
+- 囗
+- 諌
+- 鋤
+- 壆
+- 嶄
+- 碕
+- 綵
+- 闢
+- 鳐
+- 秣
+- 簗
+- 蕪
+- 氹
+- 兪
+- 恹
+- 鈕
+- 钇
+- 蝣
+- 軀
+- 畲
+- 埕
+- 潅
+- 瞼
+- 褄
+- 詈
+- 邏
+- 颶
+- 莟
+- 艙
+- 碩
+- 筅
+- 钐
+- 寤
+- 徭
+- 銑
+- 朊
+- 楝
+- 澩
+- 竽
+- 褌
+- 覲
+- 铑
+- 鵠
+- 捭
+- 哞
+- 墊
+- 忾
+- 杼
+- 玑
+- 砭
+- 芨
+- 菡
+- 锇
+- 埓
+- 壅
+- 峅
+- 崧
+- 徠
+- 悱
+- 瑧
+- 脩
+- 邙
+- 铨
+- 鱸
+- 鸩
+- ẅ
+- 壢
+- 泫
+- 爰
+- 箝
+- 耄
+- 耋
+- ▲
+- 垭
+- 巒
+- 捯
+- 撿
+- 攋
+- 梼
+- 璟
+- 疖
+- 痦
+- 紥
+- 缬
+- 腈
+- 菝
+- 逓
+- 铒
+- 鬢
+- 魃
+- 뎁
+- ѝ
+- 垚
+- 擲
+- 眦
+- 絃
+- 舄
+- 蘅
+- 隸
+- 髯
+- 鲛
+- 鹱
+- 뎠
+- 딛
+- 몄
+- 쨈
+- 휠
+- 휩
+- ȃ
+- Ց
+- ۃ
+- ॠ
+- ḅ
+- ヮ
+- 勻
+- 卌
+- 吋
+- 噸
+- 囝
+- 奀
+- 戆
+- 揈
+- 揞
+- 搲
+- 攏
+- 昶
+- 暅
+- 椶
+- 榲
+- 橞
+- 涠
+- 珙
+- 琬
+- 磚
+- 粝
+- 糰
+- 緬
+- 罫
+- 羈
+- 葜
+- 蒹
+- 蕁
+- 薷
+- 蠱
+- 襴
+- 轸
+- 邰
+- 鏈
+- 鬘
+- 龐
+- 긱
+- 꺽
+- 늠
+- 뎀
+- 딕
+- 띡
+- 뵐
+- 셧
+- 슌
+- 웍
+- 윳
+- 짙
+- 쫒
+- 텄
+- 헉
+- 헹
+- 훗
+- Ώ
+- ♯
+- 刪
+- 妯
+- 廪
+- 瀟
+- 犧
+- 畦
+- 癪
+- 矽
+- 禪
+- 腴
+- 袢
+- 鉦
+- 鏝
+- 뽁
+- 켈
+- 휜
+- 沭
+- 漣
+- 磔
+- 蕩
+- ଐ
+- 仟
+- 壟
+- 妪
+- 淝
+- 紓
+- 苴
+- 莜
+- 隴
+- 饌
+- 駭
+- 鹘
+- 黢
+- ઐ
+- 浛
+- ㄟ
+- 剷
+- 圻
+- 澹
+- 砻
+- 肄
+- 崂
+- 痾
+- 稗
+- 褻
+- 迨
+- 镧
+- 霰
+- 顰
+- 輦
+- 輛
+- 焔
+- 篭
+- 踐
+- 坻
+- 왜
+- 殯
+- 靄
+- 琨
+- 闖
+- 騭
+- 蝿
+- 頤
+- 厠
+- 夲
+- 嫪
+- 玘
+- 蘊
+- 黚
+- 黧
+- 媞
+- 咥
+- 嘜
+- 嵊
+- 椴
+- 濞
+- 緘
+- 藺
+- 蝮
+- 醴
+- 鉉
+- 鹗
+- 贲
+- ഃ
+- ဠ
+- 偈
+- 唪
+- 嗳
+- 姶
+- 嫻
+- 孥
+- 崁
+- 彧
+- 徂
+- 枞
+- 狽
+- 皲
+- 紮
+- 缦
+- 莒
+- 裃
+- ṃ
+- 穫
+- ゚
+- 喑
+- 摟
+- 擋
+- 氡
+- 篾
+- 絣
+- 绐
+- 聩
+- 蚶
+- 螟
+- 襞
+- 賁
+- 踟
+- 蹰
+- 鈷
+- 镛
+- 闾
+- 髂
+- 鲆
+- 齒
+- 쌋
+- 췌
+- ॊ
+- ఔ
+- ဩ
+- ኅ
+- ※
+- 俅
+- 唖
+- 尷
+- 洇
+- 澆
+- 绀
+- 蕭
+- 틋
+- ኻ
+- 佷
+- 侈
+- 뉜
+- 쉑
+- 쑈
+- "\x8A"
+- "\x9D"
+- Ğ
+- ǐ
+- ಝ
+- ೯
+- ඃ
+- ฯ
+- ဈ
+- ሒ
+- ኼ
+- ጴ
+- ៖
+- ៦
+- ᱬ
+- ᱶ
+- ᱸ
+- □
+- 䁅
+- 呯
+- 唈
+- 唳
+- 喐
+- 嗄
+- 噹
+- 娌
+- 娛
+- 寢
+- 嶂
+- 恽
+- 慚
+- 懞
+- 懣
+- 抌
+- 攜
+- 曚
+- 枥
+- 柽
+- 樑
+- 樞
+- 樾
+- 牀
+- 狍
+- 稙
+- 繚
+- 舸
+- 芎
+- 衾
+- 訕
+- 豢
+- 躝
+- 轎
+- 酐
+- 鎏
+- 鏊
+- 钆
+- 钪
+- 钶
+- 雋
+- 饋
+- 鬠
+- 鸫
+- 龠
+- 갰
+- 겅
+- 곗
+- 곪
+- 굼
+- 낱
+- 냇
+- 넒
+- 닛
+- 댑
+- 덨
+- 듦
+- 땔
+- 떄
+- 뗀
+- 똔
+- 롷
+- 롹
+- 묽
+- 볐
+- 빢
+- 뼌
+- 뽂
+- 샛
+- 샥
+- 솟
+- 숄
+- 숑
+- 슛
+- 쐬
+- 쑨
+- 쓕
+- 앳
+- 얕
+- 옅
+- 웁
+- 윰
+- 쟨
+- 젼
+- 짹
+- 쫘
+- 쭌
+- 챗
+- 펩
+- 푯
+- 핌
+- 𢱕
+- 櫚
+- 煅
+- 甌
+- 莢
+- 驕
+- 髌
+- 랏
+- 쏜
+- 옐
+- 핼
+- ஔ
+- 僖
+- 恊
+- 滙
+- 澍
+- 癢
+- 粿
+- 翹
+- 蔀
+- 蛸
+- 躓
+- 鏃
+- 飩
+- 髀
+- 吶
+- 垃
+- 巉
+- 巔
+- 怩
+- 搗
+- 楦
+- 琚
+- 篁
+- 脲
+- 誡
+- 阊
+- 鬻
+- 鸱
+- 叺
+- 湟
+- 頗
+- 魟
+- 狢
+- 畐
+- 畷
+- 椹
+- 諍
+- 醮
+- 鐸
+- 釗
+- 镗
+- 锴
+- ゙
+- 戕
+- 稈
+- 纒
+- 亓
+- 庹
+- 氖
+- 祉
+- 鉗
+- 础
+- 嚯
+- 堉
+- 桫
+- 椤
+- 楋
+- 瀣
+- 珧
+- 礬
+- 舾
+- 邾
+- 鵺
+- 鼱
+- 滢
+- 臍
+- 堝
+- 弍
+- 晞
+- 椁
+- 濺
+- 睄
+- 礇
+- 笫
+- 蠟
+- 鎂
+- 閨
+- 怏
+- 慷
+- 瀏
+- 綸
+- 罎
+- 閩
+- ጐ
+- ឫ
+- 伧
+- 僂
+- 冼
+- 夔
+- 媠
+- 嵴
+- 幟
+- 畊
+- 磬
+- 窺
+- 簷
+- 胿
+- 臬
+- 蚵
+- 蹚
+- 鋏
+- 鏖
+- 霑
+- 騮
+- 鲔
+- 끅
+- ̋
+- Љ
+- ѕ
+- ២
+- 佶
+- 唻
+- 抾
+- 柞
+- 澱
+- 錙
+- 늄
+- 뒨
+- 믈
+- 콰
+- ±
+- ¶
+- ʾ
+- ̊
+- ଈ
+- ଔ
+- ஶ
+- ኀ
+- ጳ
+- ጼ
+- ጾ
+- ጿ
+- ፥
+- ឮ
+- ០
+- ៤
+- ៨
+- ᱝ
+- 㓤
+- 仝
+- 倧
+- 刄
+- 厍
+- 咇
+- 唥
+- 喏
+- 嗞
+- 囍
+- 圜
+- 埞
+- 塬
+- 塱
+- 墀
+- 墮
+- 壸
+- 婭
+- 岫
+- 崃
+- 崋
+- 弶
+- 愜
+- 憊
+- 挈
+- 揜
+- 摮
+- 擘
+- 擱
+- 昐
+- 枘
+- 枳
+- 椚
+- 槁
+- 樒
+- 櫸
+- 淸
+- 溘
+- 溼
+- 燙
+- 痈
+- 硎
+- 篩
+- 簒
+- 縝
+- 縻
+- 纔
+- 荪
+- 葯
+- 蚴
+- 蛏
+- 蛻
+- 蝰
+- 蠹
+- 裇
+- 裥
+- 誅
+- 豎
+- 貲
+- 踭
+- 踴
+- 蹌
+- 蹣
+- 鑿
+- 铼
+- 锺
+- 镲
+- 颙
+- 駈
+- 駱
+- 鮓
+- 鮟
+- 鯇
+- 鰈
+- 鰜
+- 鱇
+- 鲀
+- 鵲
+- 괭
+- 굘
+- 긌
+- 깟
+- 깻
+- 꼿
+- 넙
+- 뉸
+- 뗘
+- 뚠
+- 띃
+- 렜
+- 룔
+- 멓
+- 멱
+- 뭄
+- 뺌
+- 뿅
+- 뿍
+- 숯
+- 슘
+- 쎌
+- 얠
+- 옌
+- 잴
+- 쩝
+- 쳇
+- 췻
+- 츰
+- 캇
+- 켔
+- 퀭
+- 킁
+- 탤
+- 튄
+- 팰
+- 혓
+- 홧
+- 훤
+- 휑
+- 힉
+- ȇ
+- ˎ
+- 躼
+- 곯
+- 덫
+- 햅
+- Ě
+- 狹
+- 睚
+- 觜
+- 겡
+- 셴
+- 쌜
+- ಋ
+- 壘
+- 孛
+- 忟
+- 旻
+- 榑
+- 煥
+- 狎
+- 眇
+- 罘
+- 胪
+- 脛
+- 舨
+- 镒
+- 餼
+- 馐
+- 촤
+- 歃
+- 禛
+- ♭
+- 垌
+- 尭
+- 晝
+- 楹
+- 滄
+- 砜
+- 菟
+- 蒨
+- 藷
+- 鏢
+- 鐡
+- 頌
+- 馕
+- 鰲
+- 鳉
+- 豕
+- 蜘
+- ㄧ
+- 嵬
+- 忸
+- 暝
+- 盜
+- 螣
+- 謐
+- 嘬
+- 圾
+- 洟
+- 舁
+- 醗
+- 铟
+- 颏
+- 黌
+- 栂
+- 瘻
+- 瞋
+- 窣
+- 窸
+- 絋
+- 鶫
+- 銕
+- √
+- 蚺
+- 蹕
+- 窶
+- 牻
+- 擯
+- 愼
+- 榾
+- 癩
+- 筜
+- 筼
+- 聶
+- 蟄
+- 鍔
+- 頽
+- 黠
+- 梛
+- 莨
+- 骈
+- 鸊
+- 倬
+- 唛
+- 嗶
+- 嘰
+- 嚐
+- 媪
+- 徛
+- 憺
+- 揠
+- 甦
+- 翃
+- 肭
+- 芗
+- 茔
+- 萁
+- 葎
+- 蓠
+- 蛉
+- 蜮
+- 贶
+- 趖
+- 轡
+- 釁
+- 鈉
+- 隹
+- 餉
+- 饪
+- 騅
+- 橹
+- 篳
+- 蟯
+- Ъ
+- ឦ
+- ∠
+- 〝
+- ヰ
+- 俠
+- 冪
+- 埤
+- 墘
+- 嬅
+- 峤
+- 巿
+- 扻
+- 搦
+- 攔
+- 昰
+- 枋
+- 槊
+- 渑
+- 燵
+- 猊
+- 簋
+- 肼
+- 臚
+- 艄
+- 茆
+- 茼
+- 菘
+- 菪
+- 諷
+- 譟
+- 躑
+- 輓
+- 郯
+- 郾
+- 鄄
+- 鋆
+- 铳
+- 锨
+- 閏
+- 颪
+- 馭
+- 髒
+- 鱔
+- ǰ
+- ॲ
+- ୫
+- ឪ
+- 愠
+- 歛
+- 皚
+- 硲
+- 稣
+- 蛚
+- 輾
+- 馩
+- 꽌
+- 롸
+- 밧
+- 뱄
+- "\x80"
+- "\x99"
+- Ï
+- ǒ
+- ȏ
+- ̌
+- ̔
+- ̟
+- Ӯ
+- ٗ
+- ۗ
+- ޡ
+- ऍ
+- ॆ
+- ॔
+- ૢ
+- ୪
+- ೦
+- ೨
+- ೩
+- ෳ
+- ቮ
+- ጹ
+- ḉ
+- ḫ
+- ṇ
+- ṉ
+- ẃ
+- Ạ
+- Ầ
+- Ậ
+- Ệ
+- Ự
+- ⁠
+- ∙
+- ⊙
+- ◯
+- ⸻
+- 〟
+- 㧎
+- 䒏
+- 䒐
+- 佮
+- 俟
+- 倯
+- 倻
+- 偢
+- 僥
+- 儍
+- 凊
+- 匱
+- 叁
+- 嗐
+- 嘏
+- 噁
+- 囪
+- 埵
+- 堯
+- 奓
+- 姍
+- 娉
+- 尣
+- 弸
+- 怍
+- 悽
+- 挜
+- 挹
+- 揗
+- 摈
+- 斃
+- 昉
+- 曵
+- 梣
+- 棧
+- 楫
+- 橐
+- 欒
+- 殭
+- 殳
+- 洎
+- 浐
+- 涷
+- 玭
+- 瑋
+- 璁
+- 璈
+- 甙
+- 畈
+- 瘓
+- 眙
+- 硃
+- 碚
+- 磧
+- 竅
+- 筥
+- 篦
+- 粲
+- 糒
+- 繻
+- 罅
+- 胗
+- 舺
+- 艋
+- 艤
+- 艪
+- 艸
+- 茕
+- 荜
+- 莵
+- 菉
+- 菔
+- 萠
+- 蓺
+- 蔣
+- 蘂
+- 蟥
+- 覯
+- 訐
+- 訥
+- 詡
+- 誣
+- 誦
+- 誻
+- 謨
+- 譖
+- 豂
+- 赧
+- 趌
+- 趺
+- 躅
+- 軚
+- 輘
+- 輷
+- 迾
+- 鄣
+- 醌
+- 鎅
+- 鎔
+- 鎝
+- 鏨
+- 鑄
+- 鑲
+- 钋
+- 闕
+- 陉
+- 頰
+- 餋
+- 餒
+- 餛
+- 馯
+- 骘
+- 鬍
+- 鬨
+- 魎
+- 鲮
+- 鲿
+- 鳫
+- 鵞
+- 鵡
+- 鶉
+- 鹛
+- 鼆
+- 鼐
+- 걜
+- 겋
+- 곌
+- 굥
+- 귐
+- 꽐
+- 꽥
+- 꿩
+- 끍
+- 냔
+- 냘
+- 냡
+- 넝
+- 넴
+- 놉
+- 놋
+- 놘
+- 뇽
+- 뉩
+- 늉
+- 댜
+- 듈
+- 땋
+- 떰
+- 띤
+- 맽
+- 멩
+- 몀
+- 믕
+- 볕
+- 뵌
+- 빻
+- 뿡
+- 샨
+- 숀
+- 숴
+- 슾
+- 쌕
+- 쌨
+- 썅
+- 썜
+- 쎅
+- 쏼
+- 쒀
+- 씰
+- 옉
+- 옫
+- 웟
+- 읊
+- 읜
+- 좆
+- 짯
+- 쨋
+- 쨰
+- 쩄
+- 쮸
+- 촐
+- 캬
+- 켁
+- 켐
+- 켸
+- 콥
+- 쿱
+- 퉤
+- 튬
+- 팹
+- 퐉
+- 푤
+- 퓰
+- 픕
+- 휀
+- 𢳂
+- ʺ
+- 꾜
+- 꿰
+- 쇳
+- 잽
+- Ǹ
+- ೫
+- ኟ
+- ṅ
+- †
+- 凈
+- 垪
+- 拵
+- 曷
+- 梘
+- 漚
+- 絚
+- 鯪
+- 넚
+- 쌉
+- 엡
+- ೮
+- 䢢
+- 佤
+- 凫
+- 煖
+- 獪
+- 瘆
+- 癆
+- 秬
+- 糋
+- 躹
+- 迳
+- 钌
+- 陜
+- 韪
+- 驛
+- 髙
+- 鹼
+- 갉
+- 뷴
+- 寃
+- 刳
+- 劏
+- 嚡
+- 杮
+- 槎
+- 槤
+- 樅
+- 沤
+- 炴
+- 煊
+- 熒
+- 珎
+- 璉
+- 痼
+- 簔
+- 苄
+- 苳
+- 菴
+- 蘚
+- 虬
+- 詆
+- 赍
+- 跶
+- 铪
+- 闔
+- 顱
+- 颢
+- 飜
+- 骹
+- 伜
+- 柝
+- 聟
+- 釵
+- 崆
+- 畋
+- 笸
+- 膾
+- 蕲
+- 靫
+- 頴
+- 髷
+- 鵙
+- 鸷
+- 錺
+- 蒟
+- 愴
+- 腘
+- 鯱
+- 샀
+- 끊
+- 굳
+- 拶
+- Ẓ
+- 橄
+- Ṛ
+- 惫
+- 咆
+- 옛
+- ఊ
+- 싫
+- 밌
+- 雰
+- 괜
+- 떻
+- 뭔
+- 伥
+- 飪
+- 鬪
+- 鼈
+- 걔
+- 椙
+- 蒎
+- 锿
+- 鲵
+- 亰
+- 啻
+- 囂
+- 峁
+- 弖
+- 憮
+- 桴
+- 瓘
+- 瘧
+- 秭
+- 簓
+- 薖
+- 蝴
+- 谯
+- 趿
+- 镨
+- 闳
+- 馑
+- 骺
+- 鱚
+- 鲐
+- ఋ
+- 咙
+- Ÿ
+- ഋ
+- ഢ
+- ဋ
+- 丨
+- 丱
+- 仱
+- 傈
+- 傩
+- 僳
+- 劊
+- 嗆
+- 嗹
+- 嘁
+- 岈
+- 嵖
+- 巖
+- 庠
+- 廸
+- 戔
+- 扂
+- 拏
+- 挾
+- 掗
+- 摰
+- 撣
+- 攑
+- 敍
+- 旃
+- 旒
+- 栻
+- 槓
+- 歕
+- 歨
+- 殂
+- 泠
+- 渀
+- 潲
+- 潷
+- 澀
+- 瓩
+- 甓
+- 疃
+- 癀
+- 癔
+- 竦
+- 筊
+- 篥
+- 籟
+- 籼
+- 糬
+- 緹
+- 縉
+- 縢
+- 膕
+- 臈
+- 臙
+- 艏
+- 苡
+- 莩
+- 蓊
+- 薀
+- 薜
+- 蘓
+- 裈
+- 褧
+- 覕
+- 諡
+- 謇
+- 诒
+- 诤
+- 贽
+- 軼
+- 迤
+- 逶
+- 邡
+- 醅
+- 釼
+- 錚
+- 鍚
+- 鐃
+- 鐐
+- 钺
+- 铇
+- 锕
+- 镎
+- 靑
+- 顫
+- 髖
+- 鰊
+- 鳎
+- 鳔
+- 鷽
+- 鼙
+- ኰ
+- 䖙
+- 俢
+- 儉
+- 啍
+- 埲
+- 屄
+- 戥
+- 掞
+- 枦
+- 爿
+- 笳
+- 綢
+- 繄
+- 翕
+- 芘
+- 蛣
+- 豔
+- 蹐
+- 鉀
+- 闞
+- 鵪
+- 鸝
+- 됭
+- 싰
+- 옭
+- 좔
+- ̧
+- ☉
+- 떫
+- 뱌
+- 벘
+- 씉
+- 엷
+- 읐
+- Ƴ
+- "\x81"
+- "\x8D"
+- "\x8E"
+- "\x98"
+- ¦
+- Ð
+- Ù
+- Ŋ
+- ƒ
+- Ǥ
+- ǫ
+- Ʉ
+- ˊ
+- ˋ
+- ̕
+- ̱
+- ̲
+- ̶
+- Ψ
+- Ђ
+- Ћ
+- Ќ
+- Ѹ
+- Ӌ
+- Ӳ
+- Ղ
+- Ր
+- ַ
+- ּ
+- ؓ
+- ٓ
+- ऩ
+- ॐ
+- ৗ
+- ઼
+- ૧
+- ૮
+- ୗ
+- ୨
+- ୯
+- ఝ
+- ఱ
+- ೪
+- ഝ
+- ෴
+- ་
+- ན
+- ሓ
+- ሿ
+- ቐ
+- ቓ
+- ቯ
+- ዒ
+- ዖ
+- ዡ
+- ጯ
+- ፗ
+- ៎
+- ៥
+- ៧
+- ḏ
+- ḡ
+- ḩ
+- Ḫ
+- ṁ
+- ẞ
+- Ắ
+- Ể
+- Ỉ
+- Ỗ
+- Ộ
+- Ớ
+- Ὑ
+- ῖ
+- ‰
+- ₹
+- ⃗
+- ↑
+- ↓
+- ∇
+- ∼
+- ≈
+- ▼
+- ☺
+- ✅
+- Ⱅ
+- Ⱎ
+- ゎ
+- ゔ
+- 㶶
+- 䁯
+- 䆀
+- 䱽
+- 䴕
+- 亍
+- 佉
+- 侷
+- 傃
+- 傉
+- 傱
+- 僉
+- 僊
+- 僮
+- 凩
+- 刋
+- 剎
+- 劖
+- 劼
+- 勰
+- 勼
+- 匏
+- 厴
+- 厶
+- 叅
+- 吿
+- 呓
+- 咹
+- 哓
+- 唂
+- 唎
+- 喟
+- 喲
+- 喼
+- 嗇
+- 嗿
+- 嚀
+- 嚒
+- 囑
+- 圷
+- 坜
+- 坵
+- 坼
+- 埆
+- 埭
+- 塢
+- 塭
+- 墁
+- 奬
+- 妗
+- 妠
+- 妡
+- 妣
+- 妤
+- 姹
+- 娗
+- 嫐
+- 嬶
+- 尓
+- 尙
+- 屢
+- 屣
+- 岀
+- 峄
+- 峇
+- 崞
+- 崮
+- 帙
+- 廍
+- 弌
+- 弢
+- 彥
+- 彳
+- 悒
+- 惝
+- 愔
+- 愷
+- 愾
+- 慤
+- 戞
+- 戽
+- 扴
+- 抆
+- 抔
+- 抦
+- 拃
+- 捹
+- 掁
+- 掕
+- 掼
+- 摙
+- 摳
+- 摵
+- 摷
+- 擏
+- 擷
+- 斝
+- 旳
+- 昃
+- 晡
+- 晧
+- 暸
+- 杙
+- 杣
+- 杻
+- 枧
+- 枰
+- 柸
+- 棔
+- 椏
+- 椟
+- 榙
+- 樁
+- 樗
+- 檪
+- 櫈
+- 殁
+- 殄
+- 毵
+- 氾
+- 汱
+- 洐
+- 洹
+- 淥
+- 淪
+- 湫
+- 溋
+- 溷
+- 滸
+- 潴
+- 澗
+- 澶
+- 濉
+- 瀍
+- 烝
+- 煆
+- 熈
+- 燜
+- 爨
+- 牾
+- 猗
+- 玜
+- 玠
+- 珒
+- 珣
+- 瑨
+- 瓠
+- 疔
+- 疠
+- 疴
+- 睞
+- 硚
+- 硨
+- 磲
+- 礑
+- 祆
+- 禳
+- 穐
+- 笮
+- 筌
+- 筿
+- 箆
+- 箓
+- 篋
+- 簟
+- 簫
+- 糉
+- 絅
+- 綖
+- 綯
+- 綷
+- 綽
+- 緁
+- 緲
+- 縒
+- 縹
+- 繆
+- 繏
+- 缑
+- 缙
+- 罨
+- 羂
+- 羣
+- 羶
+- 翙
+- 聼
+- 肟
+- 艀
+- 艹
+- 芰
+- 芻
+- 苎
+- 茏
+- 茖
+- 茭
+- 莚
+- 莸
+- 莾
+- 萜
+- 萡
+- 蒽
+- 蓀
+- 蓁
+- 蓥
+- 蕘
+- 蕞
+- 薟
+- 藳
+- 蛲
+- 蜑
+- 蜞
+- 蝥
+- 螋
+- 蟇
+- 蟳
+- 衒
+- 衮
+- 袛
+- 袰
+- 裄
+- 裎
+- 褦
+- 褫
+- 襤
+- 觔
+- 觚
+- 詬
+- 諚
+- 諤
+- 謔
+- 譫
+- 讒
+- 诮
+- 谘
+- 谳
+- 貉
+- 貮
+- 赀
+- 赓
+- 赟
+- 踅
+- 蹺
+- 輊
+- 輟
+- 輳
+- 轾
+- 辎
+- 辶
+- 迍
+- 郃
+- 郗
+- 郛
+- 郧
+- 鄯
+- 醯
+- 釆
+- 釿
+- 鈪
+- 鉅
+- 鉎
+- 銥
+- 銳
+- 鎭
+- 鐇
+- 鐶
+- 鑛
+- 鑢
+- 钕
+- 钤
+- 钫
+- 钬
+- 钿
+- 铙
+- 铧
+- 铽
+- 锝
+- 锟
+- 镝
+- 镡
+- 闼
+- 隗
+- 雺
+- 霈
+- 韃
+- 韜
+- 韫
+- 餚
+- 餾
+- 饑
+- 馔
+- 駝
+- 驺
+- 驽
+- 骝
+- 髧
+- 鮒
+- 鮖
+- 鯣
+- 鰌
+- 鰒
+- 鱉
+- 鱷
+- 鲗
+- 鲩
+- 鲾
+- 鳀
+- 鳊
+- 鳚
+- 鳜
+- 鳢
+- 鳰
+- 鴿
+- 鹀
+- 鹟
+- 麭
+- 黩
+- 鼢
+- 鼷
+- 齴
+- 龢
+- ꞌ
+- 갼
+- 겝
+- 겻
+- 곘
+- 괍
+- 괏
+- 궂
+- 귯
+- 꺅
+- 꺤
+- 껒
+- 꽨
+- 꽷
+- 꾿
+- 뀨
+- 끕
+- 낏
+- 넵
+- 녘
+- 놥
+- 눟
+- 늗
+- 늣
+- 닪
+- 닸
+- 덷
+- 뎄
+- 뎡
+- 돚
+- 됑
+- 듭
+- 듶
+- 딫
+- 땍
+- 땟
+- 떈
+- 떱
+- 뗏
+- 똫
+- 뙤
+- 뚸
+- 뛌
+- 뜹
+- 띌
+- 띔
+- 럤
+- 롄
+- 뤃
+- 뤼
+- 맀
+- 먀
+- 먙
+- 멨
+- 묀
+- 뭍
+- 뭥
+- 뭬
+- 뭰
+- 믁
+- 믐
+- 밈
+- 밎
+- 뱁
+- 뱡
+- 벋
+- 벛
+- 볻
+- 봥
+- 뵤
+- 붜
+- 뷘
+- 뺘
+- 뻬
+- 뼘
+- 뼜
+- 뿕
+- 뿟
+- 쁩
+- 쁸
+- 삣
+- 삧
+- 삯
+- 셤
+- 셥
+- 셸
+- 솝
+- 솨
+- 쇽
+- 쉪
+- 쉭
+- 쌘
+- 쎼
+- 쏸
+- 쐐
+- 쐴
+- 쑬
+- 쒯
+- 씃
+- 앎
+- 앏
+- 앝
+- 얉
+- 얋
+- 얐
+- 얬
+- 옙
+- 옜
+- 왝
+- 왯
+- 윅
+- 읒
+- 잿
+- 쟝
+- 젭
+- 젱
+- 졀
+- 좃
+- 좇
+- 좐
+- 죈
+- 죵
+- 줜
+- 줴
+- 쥔
+- 쥘
+- 짢
+- 쩟
+- ํ
+- ଋ
+- 닯
+- 뜀
+- 룟
+- 뭡
+- 쌂
+- Ѓ
+- ઁ
+- ః
+- 摅
+- 欉
+- 洶
+- 牘
+- 篪
+- 繃
+- 艻
+- 跩
+- 鷯
+- 뫼
+- 왐
+- ఙ
+- 仂
+- 俍
+- 埖
+- 堊
+- 尅
+- 悾
+- 掅
+- 摎
+- 柢
+- 樨
+- 橂
+- 歁
+- 殕
+- 狯
+- 珜
+- 珰
+- 瑆
+- 畭
+- 穰
+- 箨
+- 缳
+- 罌
+- 耨
+- 臌
+- 苁
+- 萄
+- 蠓
+- 蠖
+- 蠶
+- 褸
+- 襠
+- 觴
+- 謖
+- 酺
+- 铊
+- 駟
+- 鰍
+- 黾
+- ♡
+- 僢
+- 剉
+- 噘
+- 圞
+- 圹
+- 岨
+- 弉
+- 昺
+- 橢
+- 濶
+- 焮
+- 禩
+- 秡
+- 秫
+- 笊
+- 茌
+- 蕖
+- 藹
+- 螅
+- 袆
+- 郏
+- 鉋
+- 銹
+- 铌
+- 駸
+- 墻
+- 悋
+- 暱
+- 櫨
+- 浬
+- 筬
+- 糺
+- 紑
+- 誂
+- 賎
+- 跹
+- 蹁
+- 蹠
+- 鰤
+- ằ
+- 蒻
+- 鞨
+- ắ
+- 笵
+- <sos>
+- <eos>
+- <sop>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: null
+    zero_infinity: true
+    brctc_risk_strategy: exp
+    brctc_group_strategy: end
+    brctc_risk_factor: 0.0
+use_preprocessor: true
+token_type: bpe
+bpemodel: data/token_list/bpe_unigram50000/bpe.model
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+short_noise_thres: 0.5
+frontend: default
+frontend_conf:
+    n_fft: 512
+    win_length: 400
+    hop_length: 160
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: false
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 27
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_ratio_range:
+    - 0.0
+    - 0.05
+    num_time_mask: 4
+normalize: global_mvn
+normalize_conf:
+    stats_file: exp/s2t_stats_raw_bpe50000/train/feats_stats.npz
+model: espnet
+model_conf:
+    ctc_weight: 0.3
+    lsm_weight: 0.1
+    length_normalized_loss: false
+    sym_na: <na>
+preencoder: null
+preencoder_conf: {}
+encoder: e_branchformer
+encoder_conf:
+    output_size: 768
+    attention_heads: 12
+    attention_layer_type: selfattn
+    pos_enc_layer_type: abs_pos
+    rel_pos_type: latest
+    attention_qk_norm: false
+    use_flash_attn: true
+    cgmlp_linear_units: 3072
+    cgmlp_conv_kernel: 31
+    use_linear_after_conv: false
+    gate_activation: identity
+    num_blocks: 9
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.1
+    input_layer: conv2d
+    layer_drop_rate: 0.0
+    linear_units: 3072
+    positionwise_layer_type: linear
+    use_ffn: true
+    macaron_ffn: true
+    merge_conv_kernel: 31
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 12
+    linear_units: 3072
+    num_blocks: 9
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.1
+    src_attention_dropout_rate: 0.1
+    self_attention_qk_norm: false
+    src_attention_qk_norm: false
+    self_attention_use_flash_attn: true
+    src_attention_use_flash_attn: true
+preprocessor: s2t
+preprocessor_conf:
+    text_prev_name: text_prev
+    text_ctc_name: text_ctc
+    fs: 16000
+    na_symbol: <na>
+    speech_length: 30
+    speech_resolution: 0.02
+    speech_init_silence: 30
+    text_prev_apply_prob: 0.5
+    time_apply_prob: 0.5
+    notime_symbol: <notimestamps>
+    first_time_symbol: <0.00>
+    last_time_symbol: <30.00>
+required:
+- output_dir
+- token_list
+version: '202310'
+distributed: true
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/acc.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/acc.png
new file mode 100644
index 0000000000000000000000000000000000000000..e414c2e3bc190a12206068bb3c0f9a6a468993d7
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/acc.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/backward_time.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/backward_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..63b79499286820d0136d5e8bdc2a5066d3d6756e
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/backward_time.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer.png
new file mode 100644
index 0000000000000000000000000000000000000000..9c366bb98c5c9bc09c20d6ee2625199c34fa3451
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer_ctc.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer_ctc.png
new file mode 100644
index 0000000000000000000000000000000000000000..38b79337308f9302f46e33db2fe4b0062dc22a34
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/cer_ctc.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/clip.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/clip.png
new file mode 100644
index 0000000000000000000000000000000000000000..789ed9ef23eeab1febcc37f3f4bab15d9acd8bce
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/clip.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/forward_time.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/forward_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..d54f64ddc161aea1784926d76008eddbcf5cc565
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/forward_time.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/gpu_max_cached_mem_GB.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/gpu_max_cached_mem_GB.png
new file mode 100644
index 0000000000000000000000000000000000000000..fc526667db6c56d0bb8af60c04c16ebc1036ee90
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/gpu_max_cached_mem_GB.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/grad_norm.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/grad_norm.png
new file mode 100644
index 0000000000000000000000000000000000000000..5d22c63d117b3ed3a11ca0566bfbfd1112ea5de5
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/grad_norm.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/iter_time.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/iter_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..936f3b5d62082379f5b1fbbca4df5c44a5dc4924
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/iter_time.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..2b84dc1b8e30670013e279f832ecadfa0098a825
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_att.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_att.png
new file mode 100644
index 0000000000000000000000000000000000000000..4432400662ab46d985167c6dbf4a652f7ecab95d
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_att.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_ctc.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_ctc.png
new file mode 100644
index 0000000000000000000000000000000000000000..c15d6cd4534616cde52fdecc4f70373d580f9e66
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_ctc.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_scale.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_scale.png
new file mode 100644
index 0000000000000000000000000000000000000000..e3a15585bf5417ddec89d579bf915165642612ba
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/loss_scale.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim0_lr0.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim0_lr0.png
new file mode 100644
index 0000000000000000000000000000000000000000..81674e3a65b9cfdd1da4edb6fae59580a585c1ce
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim0_lr0.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim_step_time.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim_step_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..45976239e4b0ac6924495b5afb8a1d8edb9ba8be
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/optim_step_time.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/train_time.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/train_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..88651ec4ca8996db83d84d50808aaec11fd01906
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/train_time.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/wer.png b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/wer.png
new file mode 100644
index 0000000000000000000000000000000000000000..5d502027ac02512e2500bd444fc27eca5b11c974
Binary files /dev/null and b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/images/wer.png differ
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.1.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.1.log
new file mode 100644
index 0000000000000000000000000000000000000000..19fe3d561c74fd64648b4eb1d1fb67546f036c4e
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.1.log
@@ -0,0 +1,3053 @@
+# Running on gpub011.delta.ncsa.illinois.edu
+# Started at Thu Feb 8 08:43:30 CST 2024
+# SLURMD_NODENAME=gpub011
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2955233
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1707576192
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2955233
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[011,016,034,076]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1707403392
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[011,016,034,076]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2467443
+# SLURM_TOPOLOGY_ADDR=ss00.ss09.gpub011
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_3e0c3ae9-034f-440b-9923-b21bb4cc8764 
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_3e0c3ae9-034f-440b-9923-b21bb4cc8764
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_3e0c3ae9-034f-440b-9923-b21bb4cc8764
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_3e0c3ae9-034f-440b-9923-b21bb4cc8764
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_3e0c3ae9-034f-440b-9923-b21bb4cc8764
+[gpub011:0/16] 2024-02-08 08:48:38,376 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub011:0/16] 2024-02-08 08:48:38,559 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub011:0/16] 2024-02-08 08:48:38,600 (s2t:464) INFO: Vocabulary size: 50002
+[gpub011:0/16] 2024-02-08 08:48:49,773 (abs_task:1271) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub011:0/16] 2024-02-08 08:48:49,779 (abs_task:1272) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub011:0/16] 2024-02-08 08:48:49,779 (abs_task:1275) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub011:0/16] 2024-02-08 08:48:49,779 (abs_task:1276) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub011:0/16] 2024-02-08 08:48:49,789 (abs_task:1285) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub011:0/16] 2024-02-08 08:48:55,873 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 08:48:56,871 (abs_task:1663) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa14c9dfe80>)
+[gpub011:0/16] 2024-02-08 08:48:56,871 (abs_task:1664) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub011:0/16] 2024-02-08 08:48:56,888 (abs_task:1665) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 08:49:11,517 (trainer:168) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub011:2467529:2467529 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2467529:2467529 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2467529:2467529 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2467529:2467529 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub011:0/16] 2024-02-08 08:49:17,259 (trainer:302) INFO: 34/45epoch started
+[gpub011:0/16] 2024-02-08 08:49:17,301 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-08 08:49:35,783 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 08:49:39,462 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9f4d7b3ee0>)
+[gpub011:0/16] 2024-02-08 08:49:39,462 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-08 08:49:39,465 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub034:4171618:4171618 [3] NCCL INFO cudaDriverVersion 12020
+gpub034:4171618:4171618 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.134<0>
+gpub034:4171618:4171618 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub034:4171618:4171618 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub034:4171618:4171676 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub034:4171618:4171676 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub034:4171618:4171676 [3] NCCL INFO Using network AWS Libfabric
+gpub034:4171618:4171676 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub034:4171618:4171676 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub034:4171618:4171676 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub034:4171618:4171676 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub034:4171618:4171676 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub034:4171618:4171676 [3] NCCL INFO Connected all rings
+gpub034:4171618:4171676 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub034:4171618:4171676 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub034:4171618:4171676 [3] NCCL INFO Connected all trees
+gpub034:4171618:4171676 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub034:4171618:4171676 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub034:4171618:4171676 [3] NCCL INFO comm 0x559d9e42e9a0 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub076:3117064:3117064 [3] NCCL INFO cudaDriverVersion 12020
+gpub076:3117064:3117064 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3117064:3117064 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3117064:3117064 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3117064:3117119 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3117064:3117119 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3117064:3117119 [3] NCCL INFO Using network AWS Libfabric
+gpub076:3117064:3117119 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub076:3117064:3117119 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3117064:3117119 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub076:3117064:3117119 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub076:3117064:3117119 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub011:2467532:2467532 [3] NCCL INFO cudaDriverVersion 12020
+gpub011:2467532:2467532 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2467532:2467532 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2467532:2467532 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2467532:2467589 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2467532:2467589 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2467532:2467589 [3] NCCL INFO Using network AWS Libfabric
+gpub011:2467532:2467589 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub011:2467532:2467589 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2467532:2467589 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub011:2467532:2467589 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub011:2467532:2467589 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958313:2958313 [2] NCCL INFO cudaDriverVersion 12020
+gpub016:2958313:2958313 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.116<0>
+gpub016:2958313:2958313 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub016:2958313:2958313 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub016:2958313:2958365 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub016:2958313:2958365 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub016:2958313:2958365 [2] NCCL INFO Using network AWS Libfabric
+gpub016:2958313:2958365 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub016:2958313:2958365 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub016:2958313:2958365 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub016:2958313:2958365 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub016:2958313:2958365 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub034:4171616:4171616 [1] NCCL INFO cudaDriverVersion 12020
+gpub034:4171616:4171616 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.134<0>
+gpub034:4171616:4171616 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub034:4171616:4171616 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub034:4171616:4171678 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub034:4171616:4171678 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub034:4171616:4171678 [1] NCCL INFO Using network AWS Libfabric
+gpub034:4171616:4171678 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub034:4171616:4171678 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub034:4171616:4171678 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub034:4171616:4171678 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub034:4171616:4171678 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub076:3117064:3117119 [3] NCCL INFO Connected all rings
+gpub076:3117064:3117119 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub076:3117064:3117119 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub076:3117064:3117119 [3] NCCL INFO Connected all trees
+gpub076:3117064:3117119 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3117064:3117119 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3117064:3117119 [3] NCCL INFO comm 0x55a55825a5f0 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub011:2467532:2467589 [3] NCCL INFO Connected all rings
+gpub011:2467532:2467589 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub011:2467532:2467589 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub011:2467532:2467589 [3] NCCL INFO Connected all trees
+gpub011:2467532:2467589 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2467532:2467589 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2467532:2467589 [3] NCCL INFO comm 0x55acc3bb5c50 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub016:2958313:2958365 [2] NCCL INFO Connected all rings
+gpub016:2958313:2958365 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub016:2958313:2958365 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub016:2958313:2958365 [2] NCCL INFO Connected all trees
+gpub016:2958313:2958365 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub016:2958313:2958365 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub016:2958313:2958365 [2] NCCL INFO comm 0x559baeb5c040 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub034:4171616:4171678 [1] NCCL INFO Connected all rings
+gpub034:4171616:4171678 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/AWS Libfabric/1
+gpub034:4171616:4171678 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub034:4171616:4171678 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub034:4171616:4171678 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub034:4171616:4171678 [1] NCCL INFO Connected all trees
+gpub034:4171616:4171678 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub034:4171616:4171678 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub034:4171616:4171678 [1] NCCL INFO comm 0x55da68fdf000 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub076:3117062:3117062 [1] NCCL INFO cudaDriverVersion 12020
+gpub076:3117062:3117062 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3117062:3117062 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3117062:3117062 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3117062:3117118 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3117062:3117118 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3117062:3117118 [1] NCCL INFO Using network AWS Libfabric
+gpub076:3117062:3117118 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub076:3117062:3117118 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3117062:3117118 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub076:3117062:3117118 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub076:3117062:3117118 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub034:4171615:4171615 [0] NCCL INFO cudaDriverVersion 12020
+gpub034:4171615:4171615 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.134<0>
+gpub034:4171615:4171615 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub034:4171615:4171615 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub034:4171615:4171679 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub034:4171615:4171679 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub034:4171615:4171679 [0] NCCL INFO Using network AWS Libfabric
+gpub034:4171615:4171679 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub034:4171615:4171679 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub034:4171615:4171679 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3117062:3117118 [1] NCCL INFO Connected all rings
+gpub076:3117062:3117118 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub076:3117062:3117118 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub076:3117062:3117118 [1] NCCL INFO Connected all trees
+gpub076:3117062:3117118 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3117062:3117118 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3117062:3117118 [1] NCCL INFO comm 0x560de3e8e1c0 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub034:4171615:4171679 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub034:4171615:4171679 [0] NCCL INFO Connected all rings
+gpub034:4171615:4171679 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO Connected all trees
+gpub034:4171615:4171679 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3117061:3117061 [0] NCCL INFO cudaDriverVersion 12020
+gpub076:3117061:3117061 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3117061:3117061 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3117061:3117061 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3117061:3117120 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3117061:3117120 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3117061:3117120 [0] NCCL INFO Using network AWS Libfabric
+gpub076:3117061:3117120 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub076:3117061:3117120 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3117061:3117120 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub076:3117061:3117120 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub034:4171615:4171679 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub034:4171615:4171679 [0] NCCL INFO comm 0x55dc8dbe4970 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub076:3117061:3117120 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3117061:3117120 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub076:3117061:3117120 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub076:3117061:3117120 [0] NCCL INFO Connected all rings
+gpub076:3117061:3117120 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3117061:3117120 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3117061:3117120 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub076:3117061:3117120 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub076:3117061:3117120 [0] NCCL INFO Connected all trees
+gpub076:3117061:3117120 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3117061:3117120 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub034:4171617:4171617 [2] NCCL INFO cudaDriverVersion 12020
+gpub034:4171617:4171617 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.134<0>
+gpub034:4171617:4171617 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub034:4171617:4171617 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub034:4171617:4171677 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub034:4171617:4171677 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub034:4171617:4171677 [2] NCCL INFO Using network AWS Libfabric
+gpub034:4171617:4171677 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub034:4171617:4171677 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub034:4171617:4171677 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub034:4171617:4171677 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub034:4171617:4171677 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub076:3117061:3117120 [0] NCCL INFO comm 0x561ec5e990b0 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub034:4171617:4171677 [2] NCCL INFO Connected all rings
+gpub034:4171617:4171677 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub034:4171617:4171677 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub034:4171617:4171677 [2] NCCL INFO Connected all trees
+gpub034:4171617:4171677 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub034:4171617:4171677 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub034:4171617:4171677 [2] NCCL INFO comm 0x558a7a00f580 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub011:2467529:2467588 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2467529:2467588 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2467529:2467588 [0] NCCL INFO Using network AWS Libfabric
+gpub011:2467529:2467588 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub011:2467529:2467588 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2467529:2467588 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub011:2467529:2467588 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub011:2467529:2467588 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub011:2467529:2467588 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub011:2467529:2467588 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub011:2467529:2467588 [0] NCCL INFO Connected all rings
+gpub011:2467529:2467588 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2467529:2467588 [0] NCCL INFO Connected all trees
+gpub011:2467529:2467588 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2467529:2467588 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2467529:2467588 [0] NCCL INFO comm 0x55ab67717e30 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub076:3117063:3117063 [2] NCCL INFO cudaDriverVersion 12020
+gpub076:3117063:3117063 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3117063:3117063 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3117063:3117063 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3117063:3117121 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3117063:3117121 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3117063:3117121 [2] NCCL INFO Using network AWS Libfabric
+gpub076:3117063:3117121 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub076:3117063:3117121 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3117063:3117121 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub076:3117063:3117121 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub076:3117063:3117121 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub076:3117063:3117121 [2] NCCL INFO Connected all rings
+gpub076:3117063:3117121 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub076:3117063:3117121 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub076:3117063:3117121 [2] NCCL INFO Connected all trees
+gpub076:3117063:3117121 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3117063:3117121 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3117063:3117121 [2] NCCL INFO comm 0x55b047547ea0 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub011:2467530:2467530 [1] NCCL INFO cudaDriverVersion 12020
+gpub011:2467530:2467530 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2467530:2467530 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2467530:2467530 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2467530:2467590 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2467530:2467590 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2467530:2467590 [1] NCCL INFO Using network AWS Libfabric
+gpub011:2467530:2467590 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub011:2467530:2467590 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2467530:2467590 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub011:2467530:2467590 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub011:2467530:2467590 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub011:2467530:2467590 [1] NCCL INFO Connected all rings
+gpub011:2467530:2467590 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub011:2467530:2467590 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub011:2467530:2467590 [1] NCCL INFO Connected all trees
+gpub011:2467530:2467590 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2467530:2467590 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2467530:2467590 [1] NCCL INFO comm 0x563b04c01990 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub016:2958314:2958314 [3] NCCL INFO cudaDriverVersion 12020
+gpub016:2958314:2958314 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.116<0>
+gpub016:2958314:2958314 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub016:2958314:2958314 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub016:2958314:2958366 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub016:2958314:2958366 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub016:2958314:2958366 [3] NCCL INFO Using network AWS Libfabric
+gpub016:2958314:2958366 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub016:2958314:2958366 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub016:2958314:2958366 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub016:2958314:2958366 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958314:2958366 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958314:2958366 [3] NCCL INFO Connected all rings
+gpub016:2958314:2958366 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub016:2958314:2958366 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub016:2958314:2958366 [3] NCCL INFO Connected all trees
+gpub016:2958314:2958366 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub016:2958314:2958366 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub016:2958314:2958366 [3] NCCL INFO comm 0x558a91ead080 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub016:2958311:2958311 [0] NCCL INFO cudaDriverVersion 12020
+gpub016:2958311:2958311 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.116<0>
+gpub016:2958311:2958311 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub016:2958311:2958311 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub016:2958311:2958367 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub016:2958311:2958367 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub016:2958311:2958367 [0] NCCL INFO Using network AWS Libfabric
+gpub016:2958311:2958367 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub016:2958311:2958367 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub016:2958311:2958367 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub016:2958311:2958367 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2467531:2467531 [2] NCCL INFO cudaDriverVersion 12020
+gpub011:2467531:2467531 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2467531:2467531 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2467531:2467531 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2467531:2467591 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2467531:2467591 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2467531:2467591 [2] NCCL INFO Using network AWS Libfabric
+gpub011:2467531:2467591 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub011:2467531:2467591 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2467531:2467591 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub011:2467531:2467591 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub011:2467531:2467591 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub016:2958311:2958367 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub016:2958311:2958367 [0] NCCL INFO Connected all rings
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958311:2958367 [0] NCCL INFO Connected all trees
+gpub016:2958311:2958367 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2467531:2467591 [2] NCCL INFO Connected all rings
+gpub011:2467531:2467591 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub011:2467531:2467591 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub011:2467531:2467591 [2] NCCL INFO Connected all trees
+gpub011:2467531:2467591 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2467531:2467591 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2467531:2467591 [2] NCCL INFO comm 0x560f9aadaa00 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub016:2958311:2958367 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub016:2958311:2958367 [0] NCCL INFO comm 0x55ec6c5ae8f0 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub016:2958312:2958312 [1] NCCL INFO cudaDriverVersion 12020
+gpub016:2958312:2958312 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.116<0>
+gpub016:2958312:2958312 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub016:2958312:2958312 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub016:2958312:2958368 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub016:2958312:2958368 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub016:2958312:2958368 [1] NCCL INFO Using network AWS Libfabric
+gpub016:2958312:2958368 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub016:2958312:2958368 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub016:2958312:2958368 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub016:2958312:2958368 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub016:2958312:2958368 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub016:2958312:2958368 [1] NCCL INFO Connected all rings
+gpub016:2958312:2958368 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub016:2958312:2958368 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/AWS Libfabric/1
+gpub016:2958312:2958368 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub016:2958312:2958368 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub016:2958312:2958368 [1] NCCL INFO Connected all trees
+gpub016:2958312:2958368 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub016:2958312:2958368 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub016:2958312:2958368 [1] NCCL INFO comm 0x564a10a37d30 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+[gpub011:0/16] 2024-02-08 08:58:18,546 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub011:0/16] 2024-02-08 09:00:56,555 (trainer:762) INFO: 34epoch:train:1-100batch: iter_time=3.917, forward_time=0.424, loss_ctc=57.807, loss_att=56.924, acc=0.726, loss=57.189, backward_time=0.314, grad_norm=51.142, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.741e-04, train_time=6.992
+[gpub011:0/16] 2024-02-08 09:03:32,919 (trainer:762) INFO: 34epoch:train:101-200batch: iter_time=9.181e-05, forward_time=0.287, loss_ctc=51.602, loss_att=45.692, acc=0.748, loss=47.465, backward_time=0.294, grad_norm=42.536, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.741e-04, train_time=1.564
+[gpub011:0/16] 2024-02-08 09:05:54,706 (trainer:762) INFO: 34epoch:train:201-300batch: iter_time=9.684e-05, forward_time=0.284, loss_ctc=39.152, loss_att=35.811, acc=0.763, loss=36.813, backward_time=0.290, grad_norm=36.878, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.741e-04, train_time=1.418
+[gpub011:0/16] 2024-02-08 09:07:51,285 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 09:08:37,090 (trainer:762) INFO: 34epoch:train:301-400batch: iter_time=9.825e-05, forward_time=0.419, loss_ctc=49.050, loss_att=50.994, acc=0.742, loss=50.410, backward_time=0.369, grad_norm=43.273, clip=100.000, loss_scale=8.759e+33, optim_step_time=0.098, optim0_lr0=1.740e-04, train_time=1.624
+[gpub011:0/16] 2024-02-08 09:10:49,222 (trainer:762) INFO: 34epoch:train:401-500batch: iter_time=1.022e-04, forward_time=0.287, loss_ctc=46.960, loss_att=43.412, acc=0.746, loss=44.477, backward_time=0.293, grad_norm=45.787, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.740e-04, train_time=1.321
+[gpub011:0/16] 2024-02-08 09:13:19,134 (trainer:762) INFO: 34epoch:train:501-600batch: iter_time=8.744e-05, forward_time=0.314, loss_ctc=46.638, loss_att=47.204, acc=0.752, loss=47.034, backward_time=0.295, grad_norm=43.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.740e-04, train_time=1.499
+[gpub011:0/16] 2024-02-08 09:15:48,856 (trainer:762) INFO: 34epoch:train:601-700batch: iter_time=9.323e-05, forward_time=0.431, loss_ctc=45.248, loss_att=46.841, acc=0.763, loss=46.363, backward_time=0.351, grad_norm=38.819, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.740e-04, train_time=1.497
+[gpub011:0/16] 2024-02-08 09:18:31,739 (trainer:762) INFO: 34epoch:train:701-800batch: iter_time=9.563e-05, forward_time=0.285, loss_ctc=37.408, loss_att=35.732, acc=0.783, loss=36.235, backward_time=0.289, grad_norm=33.343, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.740e-04, train_time=1.628
+[gpub011:0/16] 2024-02-08 09:20:57,807 (trainer:762) INFO: 34epoch:train:801-900batch: iter_time=2.923e-04, forward_time=0.392, loss_ctc=47.334, loss_att=42.192, acc=0.755, loss=43.735, backward_time=0.325, grad_norm=43.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.739e-04, train_time=1.460
+[gpub011:0/16] 2024-02-08 09:23:23,746 (trainer:762) INFO: 34epoch:train:901-1000batch: iter_time=1.050e-04, forward_time=0.290, loss_ctc=52.460, loss_att=51.230, acc=0.747, loss=51.599, backward_time=0.295, grad_norm=44.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.739e-04, train_time=1.460
+[gpub011:0/16] 2024-02-08 09:25:39,568 (trainer:762) INFO: 34epoch:train:1001-1100batch: iter_time=8.523e-05, forward_time=0.288, loss_ctc=47.100, loss_att=40.059, acc=0.764, loss=42.171, backward_time=0.294, grad_norm=44.761, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.739e-04, train_time=1.358
+[gpub011:0/16] 2024-02-08 09:28:38,456 (trainer:762) INFO: 34epoch:train:1101-1200batch: iter_time=9.256e-05, forward_time=0.421, loss_ctc=46.862, loss_att=40.942, acc=0.750, loss=42.718, backward_time=0.319, grad_norm=40.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.739e-04, train_time=1.788
+[gpub011:0/16] 2024-02-08 09:29:58,768 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-08 09:30:17,698 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 09:30:21,260 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065b2cdf0>)
+[gpub011:0/16] 2024-02-08 09:30:21,261 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-08 09:30:21,264 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 09:37:11,389 (trainer:762) INFO: 34epoch:train:1201-1300batch: iter_time=3.692, forward_time=0.289, loss_ctc=48.461, loss_att=49.164, acc=0.748, loss=48.953, backward_time=0.295, grad_norm=42.217, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.739e-04, train_time=5.130
+[gpub011:0/16] 2024-02-08 09:39:54,821 (trainer:762) INFO: 34epoch:train:1301-1400batch: iter_time=8.706e-05, forward_time=0.436, loss_ctc=52.462, loss_att=50.271, acc=0.758, loss=50.928, backward_time=0.333, grad_norm=44.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.739e-04, train_time=1.634
+[gpub011:0/16] 2024-02-08 09:41:43,885 (trainer:762) INFO: 34epoch:train:1401-1500batch: iter_time=8.012e-05, forward_time=0.290, loss_ctc=43.703, loss_att=39.824, acc=0.759, loss=40.988, backward_time=0.298, grad_norm=37.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.738e-04, train_time=1.090
+[gpub011:0/16] 2024-02-08 09:44:01,141 (trainer:762) INFO: 34epoch:train:1501-1600batch: iter_time=8.511e-05, forward_time=0.289, loss_ctc=42.698, loss_att=38.514, acc=0.767, loss=39.769, backward_time=0.293, grad_norm=37.721, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.738e-04, train_time=1.373
+[gpub011:0/16] 2024-02-08 09:46:22,786 (trainer:762) INFO: 34epoch:train:1601-1700batch: iter_time=7.998e-05, forward_time=0.292, loss_ctc=50.032, loss_att=51.786, acc=0.753, loss=51.259, backward_time=0.297, grad_norm=40.504, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.738e-04, train_time=1.416
+[gpub011:0/16] 2024-02-08 09:48:51,487 (trainer:762) INFO: 34epoch:train:1701-1800batch: iter_time=8.743e-05, forward_time=0.397, loss_ctc=46.769, loss_att=49.145, acc=0.739, loss=48.432, backward_time=0.353, grad_norm=42.038, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.738e-04, train_time=1.486
+[gpub011:0/16] 2024-02-08 09:50:53,490 (trainer:762) INFO: 34epoch:train:1801-1900batch: iter_time=8.805e-05, forward_time=0.292, loss_ctc=42.995, loss_att=43.106, acc=0.788, loss=43.073, backward_time=0.296, grad_norm=34.260, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.738e-04, train_time=1.220
+[gpub011:0/16] 2024-02-08 09:53:05,960 (trainer:762) INFO: 34epoch:train:1901-2000batch: iter_time=1.031e-04, forward_time=0.289, loss_ctc=41.566, loss_att=38.414, acc=0.790, loss=39.360, backward_time=0.295, grad_norm=33.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.738e-04, train_time=1.325
+[gpub011:0/16] 2024-02-08 09:55:17,518 (trainer:762) INFO: 34epoch:train:2001-2100batch: iter_time=8.695e-05, forward_time=0.288, loss_ctc=42.028, loss_att=40.652, acc=0.766, loss=41.065, backward_time=0.294, grad_norm=39.714, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.737e-04, train_time=1.315
+[gpub011:0/16] 2024-02-08 09:57:34,567 (trainer:762) INFO: 34epoch:train:2101-2200batch: iter_time=9.133e-05, forward_time=0.371, loss_ctc=48.683, loss_att=49.803, acc=0.752, loss=49.467, backward_time=0.400, grad_norm=38.810, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.737e-04, train_time=1.370
+[gpub011:0/16] 2024-02-08 09:59:45,494 (trainer:762) INFO: 34epoch:train:2201-2300batch: iter_time=9.174e-05, forward_time=0.294, loss_ctc=49.423, loss_att=46.353, acc=0.758, loss=47.274, backward_time=0.298, grad_norm=47.292, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.737e-04, train_time=1.308
+[gpub011:0/16] 2024-02-08 10:02:11,547 (trainer:762) INFO: 34epoch:train:2301-2400batch: iter_time=8.302e-05, forward_time=0.308, loss_ctc=43.031, loss_att=37.884, acc=0.775, loss=39.428, backward_time=0.293, grad_norm=37.875, clip=100.000, loss_scale=6.802e+33, optim_step_time=0.093, optim0_lr0=1.737e-04, train_time=1.462
+[gpub011:0/16] 2024-02-08 10:04:05,926 (trainer:762) INFO: 34epoch:train:2401-2500batch: iter_time=8.898e-05, forward_time=0.290, loss_ctc=44.679, loss_att=40.293, acc=0.767, loss=41.608, backward_time=0.297, grad_norm=37.737, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.737e-04, train_time=1.144
+[gpub011:0/16] 2024-02-08 10:04:25,954 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-08 10:04:44,796 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 10:04:48,376 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9f4d5a0ca0>)
+[gpub011:0/16] 2024-02-08 10:04:48,376 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-08 10:04:48,379 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 10:12:42,460 (trainer:762) INFO: 34epoch:train:2501-2600batch: iter_time=3.682, forward_time=0.538, loss_ctc=56.430, loss_att=56.109, acc=0.736, loss=56.205, backward_time=0.345, grad_norm=46.739, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.114, optim0_lr0=1.737e-04, train_time=5.165
+[gpub011:0/16] 2024-02-08 10:14:54,586 (trainer:762) INFO: 34epoch:train:2601-2700batch: iter_time=9.379e-05, forward_time=0.419, loss_ctc=50.591, loss_att=45.636, acc=0.759, loss=47.123, backward_time=0.351, grad_norm=41.519, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.736e-04, train_time=1.321
+[gpub011:0/16] 2024-02-08 10:17:25,934 (trainer:762) INFO: 34epoch:train:2701-2800batch: iter_time=5.675e-04, forward_time=0.325, loss_ctc=38.399, loss_att=35.093, acc=0.772, loss=36.085, backward_time=0.293, grad_norm=34.804, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.736e-04, train_time=1.513
+[gpub011:0/16] 2024-02-08 10:19:35,840 (trainer:762) INFO: 34epoch:train:2801-2900batch: iter_time=3.012e-04, forward_time=0.384, loss_ctc=47.733, loss_att=50.620, acc=0.752, loss=49.754, backward_time=0.369, grad_norm=40.597, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.736e-04, train_time=1.300
+[gpub011:0/16] 2024-02-08 10:21:53,702 (trainer:762) INFO: 34epoch:train:2901-3000batch: iter_time=1.011e-04, forward_time=0.446, loss_ctc=46.138, loss_att=42.707, acc=0.759, loss=43.736, backward_time=0.326, grad_norm=40.255, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.736e-04, train_time=1.377
+[gpub011:0/16] 2024-02-08 10:24:24,744 (trainer:762) INFO: 34epoch:train:3001-3100batch: iter_time=3.336e-04, forward_time=0.437, loss_ctc=45.841, loss_att=48.496, acc=0.759, loss=47.700, backward_time=0.359, grad_norm=42.261, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=1.736e-04, train_time=1.511
+[gpub011:0/16] 2024-02-08 10:26:54,811 (trainer:762) INFO: 34epoch:train:3101-3200batch: iter_time=4.855e-04, forward_time=0.369, loss_ctc=44.925, loss_att=45.815, acc=0.784, loss=45.548, backward_time=0.377, grad_norm=33.805, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.735e-04, train_time=1.500
+[gpub011:0/16] 2024-02-08 10:29:25,773 (trainer:762) INFO: 34epoch:train:3201-3300batch: iter_time=5.409e-04, forward_time=0.399, loss_ctc=36.835, loss_att=34.365, acc=0.795, loss=35.106, backward_time=0.344, grad_norm=31.396, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.735e-04, train_time=1.510
+[gpub011:0/16] 2024-02-08 10:31:32,216 (trainer:762) INFO: 34epoch:train:3301-3400batch: iter_time=0.002, forward_time=0.314, loss_ctc=45.509, loss_att=40.570, acc=0.768, loss=42.052, backward_time=0.315, grad_norm=40.013, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.735e-04, train_time=1.264
+[gpub011:0/16] 2024-02-08 10:34:03,401 (trainer:762) INFO: 34epoch:train:3401-3500batch: iter_time=9.146e-05, forward_time=0.412, loss_ctc=50.974, loss_att=53.973, acc=0.745, loss=53.074, backward_time=0.324, grad_norm=43.913, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.735e-04, train_time=1.512
+[gpub011:0/16] 2024-02-08 10:36:44,937 (trainer:762) INFO: 34epoch:train:3501-3600batch: iter_time=2.214e-04, forward_time=0.462, loss_ctc=44.514, loss_att=39.680, acc=0.772, loss=41.130, backward_time=0.319, grad_norm=41.659, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.735e-04, train_time=1.614
+[gpub011:0/16] 2024-02-08 10:39:23,796 (trainer:762) INFO: 34epoch:train:3601-3700batch: iter_time=3.231e-04, forward_time=0.397, loss_ctc=45.871, loss_att=39.620, acc=0.765, loss=41.495, backward_time=0.361, grad_norm=39.120, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.735e-04, train_time=1.589
+[gpub011:0/16] 2024-02-08 10:40:44,279 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-08 10:41:03,552 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 10:41:07,139 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065855450>)
+[gpub011:0/16] 2024-02-08 10:41:07,139 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-08 10:41:07,142 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 10:47:47,384 (trainer:762) INFO: 34epoch:train:3701-3800batch: iter_time=3.222, forward_time=0.347, loss_ctc=47.634, loss_att=48.722, acc=0.752, loss=48.396, backward_time=0.303, grad_norm=41.910, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.734e-04, train_time=5.036
+[gpub011:0/16] 2024-02-08 10:50:06,148 (trainer:762) INFO: 34epoch:train:3801-3900batch: iter_time=8.706e-05, forward_time=0.376, loss_ctc=51.261, loss_att=48.072, acc=0.756, loss=49.028, backward_time=0.314, grad_norm=43.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.734e-04, train_time=1.387
+[gpub011:0/16] 2024-02-08 10:52:18,694 (trainer:762) INFO: 34epoch:train:3901-4000batch: iter_time=8.142e-05, forward_time=0.287, loss_ctc=43.236, loss_att=38.822, acc=0.757, loss=40.146, backward_time=0.293, grad_norm=37.535, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.734e-04, train_time=1.325
+[gpub011:0/16] 2024-02-08 10:54:40,022 (trainer:762) INFO: 34epoch:train:4001-4100batch: iter_time=2.328e-04, forward_time=0.385, loss_ctc=42.319, loss_att=38.158, acc=0.762, loss=39.407, backward_time=0.348, grad_norm=39.771, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.734e-04, train_time=1.414
+[gpub011:0/16] 2024-02-08 10:56:40,982 (trainer:762) INFO: 34epoch:train:4101-4200batch: iter_time=0.002, forward_time=0.300, loss_ctc=49.602, loss_att=51.052, acc=0.750, loss=50.617, backward_time=0.314, grad_norm=41.893, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.734e-04, train_time=1.209
+[gpub011:0/16] 2024-02-08 10:59:09,432 (trainer:762) INFO: 34epoch:train:4201-4300batch: iter_time=8.331e-05, forward_time=0.397, loss_ctc=46.178, loss_att=47.826, acc=0.735, loss=47.331, backward_time=0.334, grad_norm=42.283, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.734e-04, train_time=1.484
+[gpub011:0/16] 2024-02-08 11:00:34,606 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 11:01:35,599 (trainer:762) INFO: 34epoch:train:4301-4400batch: iter_time=8.578e-05, forward_time=0.291, loss_ctc=42.309, loss_att=42.707, acc=0.783, loss=42.587, backward_time=0.297, grad_norm=38.222, clip=100.000, loss_scale=8.129e+33, optim_step_time=0.093, optim0_lr0=1.733e-04, train_time=1.461
+[gpub011:0/16] 2024-02-08 11:03:50,045 (trainer:762) INFO: 34epoch:train:4401-4500batch: iter_time=8.064e-05, forward_time=0.424, loss_ctc=40.880, loss_att=39.462, acc=0.775, loss=39.887, backward_time=0.344, grad_norm=33.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.733e-04, train_time=1.344
+[gpub011:0/16] 2024-02-08 11:05:56,212 (trainer:762) INFO: 34epoch:train:4501-4600batch: iter_time=8.465e-05, forward_time=0.289, loss_ctc=41.082, loss_att=40.719, acc=0.763, loss=40.828, backward_time=0.292, grad_norm=39.343, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.733e-04, train_time=1.262
+[gpub011:0/16] 2024-02-08 11:08:33,779 (trainer:762) INFO: 34epoch:train:4601-4700batch: iter_time=8.608e-05, forward_time=0.418, loss_ctc=48.412, loss_att=47.644, acc=0.752, loss=47.875, backward_time=0.361, grad_norm=39.982, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.733e-04, train_time=1.575
+[gpub011:0/16] 2024-02-08 11:10:41,592 (trainer:762) INFO: 34epoch:train:4701-4800batch: iter_time=8.388e-05, forward_time=0.292, loss_ctc=48.737, loss_att=44.119, acc=0.763, loss=45.504, backward_time=0.297, grad_norm=44.341, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.733e-04, train_time=1.278
+[gpub011:0/16] 2024-02-08 11:13:06,819 (trainer:762) INFO: 34epoch:train:4801-4900batch: iter_time=1.603e-04, forward_time=0.411, loss_ctc=42.184, loss_att=37.075, acc=0.771, loss=38.608, backward_time=0.373, grad_norm=37.015, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.733e-04, train_time=1.452
+[gpub011:0/16] 2024-02-08 11:15:07,131 (trainer:762) INFO: 34epoch:train:4901-5000batch: iter_time=8.033e-05, forward_time=0.311, loss_ctc=43.675, loss_att=40.627, acc=0.758, loss=41.542, backward_time=0.295, grad_norm=38.135, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.732e-04, train_time=1.203
+[gpub011:0/16] 2024-02-08 11:15:27,160 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-08 11:15:46,134 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 11:15:49,571 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9f3a26b3a0>)
+[gpub011:0/16] 2024-02-08 11:15:49,571 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-08 11:15:49,575 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 11:23:07,393 (trainer:762) INFO: 34epoch:train:5001-5100batch: iter_time=3.567, forward_time=0.292, loss_ctc=55.587, loss_att=56.381, acc=0.737, loss=56.143, backward_time=0.302, grad_norm=48.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.732e-04, train_time=4.802
+[gpub011:0/16] 2024-02-08 11:25:34,099 (trainer:762) INFO: 34epoch:train:5101-5200batch: iter_time=8.508e-05, forward_time=0.398, loss_ctc=49.834, loss_att=46.035, acc=0.758, loss=47.175, backward_time=0.335, grad_norm=39.137, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.732e-04, train_time=1.467
+[gpub011:0/16] 2024-02-08 11:27:29,245 (trainer:762) INFO: 34epoch:train:5201-5300batch: iter_time=8.233e-05, forward_time=0.289, loss_ctc=38.152, loss_att=35.159, acc=0.775, loss=36.057, backward_time=0.294, grad_norm=36.394, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.732e-04, train_time=1.151
+[gpub011:0/16] 2024-02-08 11:29:48,889 (trainer:762) INFO: 34epoch:train:5301-5400batch: iter_time=8.371e-05, forward_time=0.293, loss_ctc=47.360, loss_att=50.177, acc=0.754, loss=49.332, backward_time=0.299, grad_norm=41.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.732e-04, train_time=1.396
+[gpub011:0/16] 2024-02-08 11:32:07,144 (trainer:762) INFO: 34epoch:train:5401-5500batch: iter_time=8.074e-05, forward_time=0.394, loss_ctc=44.930, loss_att=42.614, acc=0.760, loss=43.309, backward_time=0.347, grad_norm=77.991, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.731e-04, train_time=1.382
+[gpub011:0/16] 2024-02-08 11:34:14,156 (trainer:762) INFO: 34epoch:train:5501-5600batch: iter_time=8.069e-05, forward_time=0.291, loss_ctc=44.842, loss_att=47.532, acc=0.763, loss=46.725, backward_time=0.296, grad_norm=39.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.731e-04, train_time=1.270
+[gpub011:0/16] 2024-02-08 11:36:27,876 (trainer:762) INFO: 34epoch:train:5601-5700batch: iter_time=8.124e-05, forward_time=0.319, loss_ctc=44.405, loss_att=45.570, acc=0.785, loss=45.220, backward_time=0.299, grad_norm=34.933, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.731e-04, train_time=1.337
+[gpub011:0/16] 2024-02-08 11:38:50,937 (trainer:762) INFO: 34epoch:train:5701-5800batch: iter_time=1.974e-04, forward_time=0.425, loss_ctc=36.976, loss_att=33.729, acc=0.799, loss=34.703, backward_time=0.345, grad_norm=31.330, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.731e-04, train_time=1.430
+[gpub011:0/16] 2024-02-08 11:41:03,768 (trainer:762) INFO: 34epoch:train:5801-5900batch: iter_time=8.250e-05, forward_time=0.290, loss_ctc=44.686, loss_att=41.360, acc=0.765, loss=42.358, backward_time=0.292, grad_norm=41.467, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.731e-04, train_time=1.328
+[gpub011:0/16] 2024-02-08 11:42:58,127 (trainer:762) INFO: 34epoch:train:5901-6000batch: iter_time=8.212e-05, forward_time=0.294, loss_ctc=50.012, loss_att=53.433, acc=0.746, loss=52.406, backward_time=0.300, grad_norm=44.386, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.731e-04, train_time=1.144
+[gpub011:0/16] 2024-02-08 11:45:18,664 (trainer:762) INFO: 34epoch:train:6001-6100batch: iter_time=2.193e-04, forward_time=0.343, loss_ctc=44.025, loss_att=39.716, acc=0.773, loss=41.009, backward_time=0.315, grad_norm=41.838, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.730e-04, train_time=1.405
+[gpub011:0/16] 2024-02-08 11:48:01,276 (trainer:762) INFO: 34epoch:train:6101-6200batch: iter_time=8.846e-05, forward_time=0.354, loss_ctc=45.594, loss_att=39.269, acc=0.766, loss=41.166, backward_time=0.327, grad_norm=39.862, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.730e-04, train_time=1.626
+[gpub011:0/16] 2024-02-08 11:49:16,011 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-08 11:49:35,383 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 11:49:38,794 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a55a04280>)
+[gpub011:0/16] 2024-02-08 11:49:38,795 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-08 11:49:38,800 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 11:56:03,397 (trainer:762) INFO: 34epoch:train:6201-6300batch: iter_time=3.647, forward_time=0.291, loss_ctc=47.672, loss_att=48.438, acc=0.751, loss=48.208, backward_time=0.297, grad_norm=41.568, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.730e-04, train_time=4.821
+[gpub011:0/16] 2024-02-08 11:58:09,375 (trainer:762) INFO: 34epoch:train:6301-6400batch: iter_time=8.840e-05, forward_time=0.289, loss_ctc=51.608, loss_att=47.766, acc=0.760, loss=48.919, backward_time=0.296, grad_norm=45.558, clip=100.000, loss_scale=7.425e+33, optim_step_time=0.094, optim0_lr0=1.730e-04, train_time=1.259
+[gpub011:0/16] 2024-02-08 11:59:32,883 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 12:00:40,172 (trainer:762) INFO: 34epoch:train:6401-6500batch: iter_time=8.737e-05, forward_time=0.475, loss_ctc=43.661, loss_att=39.120, acc=0.758, loss=40.482, backward_time=0.316, grad_norm=38.425, clip=100.000, loss_scale=7.815e+33, optim_step_time=0.104, optim0_lr0=1.730e-04, train_time=1.507
+[gpub011:0/16] 2024-02-08 12:02:38,295 (trainer:762) INFO: 34epoch:train:6501-6600batch: iter_time=7.987e-05, forward_time=0.289, loss_ctc=42.138, loss_att=38.016, acc=0.765, loss=39.252, backward_time=0.295, grad_norm=39.276, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.730e-04, train_time=1.182
+[gpub011:0/16] 2024-02-08 12:04:53,224 (trainer:762) INFO: 34epoch:train:6601-6700batch: iter_time=8.032e-05, forward_time=0.292, loss_ctc=49.094, loss_att=50.921, acc=0.749, loss=50.373, backward_time=0.297, grad_norm=42.384, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.729e-04, train_time=1.349
+[gpub011:0/16] 2024-02-08 12:06:45,762 (trainer:762) INFO: 34epoch:train:6701-6800batch: iter_time=8.909e-05, forward_time=0.291, loss_ctc=45.737, loss_att=47.131, acc=0.738, loss=46.713, backward_time=0.295, grad_norm=43.931, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.729e-04, train_time=1.126
+[gpub011:0/16] 2024-02-08 12:09:19,975 (trainer:762) INFO: 34epoch:train:6801-6900batch: iter_time=9.106e-05, forward_time=0.397, loss_ctc=42.379, loss_att=42.540, acc=0.787, loss=42.492, backward_time=0.339, grad_norm=43.953, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.729e-04, train_time=1.541
+[gpub011:0/16] 2024-02-08 12:11:25,464 (trainer:762) INFO: 34epoch:train:6901-7000batch: iter_time=8.795e-05, forward_time=0.289, loss_ctc=40.820, loss_att=39.107, acc=0.778, loss=39.621, backward_time=0.295, grad_norm=35.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.729e-04, train_time=1.256
+[gpub011:0/16] 2024-02-08 12:13:38,372 (trainer:762) INFO: 34epoch:train:7001-7100batch: iter_time=8.593e-05, forward_time=0.288, loss_ctc=41.002, loss_att=40.998, acc=0.764, loss=40.999, backward_time=0.294, grad_norm=40.055, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.729e-04, train_time=1.328
+[gpub011:0/16] 2024-02-08 12:15:51,865 (trainer:762) INFO: 34epoch:train:7101-7200batch: iter_time=8.582e-05, forward_time=0.312, loss_ctc=48.269, loss_att=47.565, acc=0.751, loss=47.776, backward_time=0.299, grad_norm=38.525, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.729e-04, train_time=1.336
+[gpub011:0/16] 2024-02-08 12:18:05,841 (trainer:762) INFO: 34epoch:train:7201-7300batch: iter_time=8.468e-05, forward_time=0.411, loss_ctc=48.403, loss_att=42.956, acc=0.767, loss=44.590, backward_time=0.319, grad_norm=46.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.728e-04, train_time=1.338
+[gpub011:0/16] 2024-02-08 12:20:29,168 (trainer:762) INFO: 34epoch:train:7301-7400batch: iter_time=8.179e-05, forward_time=0.289, loss_ctc=42.530, loss_att=37.179, acc=0.772, loss=38.784, backward_time=0.293, grad_norm=37.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.728e-04, train_time=1.434
+[gpub011:0/16] 2024-02-08 12:22:26,405 (trainer:762) INFO: 34epoch:train:7401-7500batch: iter_time=8.242e-05, forward_time=0.289, loss_ctc=43.891, loss_att=40.427, acc=0.759, loss=41.466, backward_time=0.295, grad_norm=39.675, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.728e-04, train_time=1.171
+[gpub011:0/16] 2024-02-08 12:22:46,433 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-08 12:23:05,741 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 12:23:09,284 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a865f1300>)
+[gpub011:0/16] 2024-02-08 12:23:09,284 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-08 12:23:09,287 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 12:30:37,832 (trainer:762) INFO: 34epoch:train:7501-7600batch: iter_time=3.605, forward_time=0.403, loss_ctc=54.920, loss_att=55.708, acc=0.734, loss=55.472, backward_time=0.316, grad_norm=50.251, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.728e-04, train_time=4.915
+[gpub011:0/16] 2024-02-08 12:32:56,990 (trainer:762) INFO: 34epoch:train:7601-7700batch: iter_time=7.951e-05, forward_time=0.289, loss_ctc=50.122, loss_att=43.133, acc=0.758, loss=45.229, backward_time=0.294, grad_norm=40.466, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.728e-04, train_time=1.392
+[gpub011:0/16] 2024-02-08 12:34:52,146 (trainer:762) INFO: 34epoch:train:7701-7800batch: iter_time=8.233e-05, forward_time=0.287, loss_ctc=37.860, loss_att=34.350, acc=0.771, loss=35.403, backward_time=0.292, grad_norm=35.140, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.728e-04, train_time=1.151
+[gpub011:0/16] 2024-02-08 12:37:18,160 (trainer:762) INFO: 34epoch:train:7801-7900batch: iter_time=9.180e-05, forward_time=0.395, loss_ctc=46.622, loss_att=49.043, acc=0.751, loss=48.317, backward_time=0.361, grad_norm=42.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.727e-04, train_time=1.459
+[gpub011:0/16] 2024-02-08 12:39:18,478 (trainer:762) INFO: 34epoch:train:7901-8000batch: iter_time=8.026e-05, forward_time=0.290, loss_ctc=45.261, loss_att=42.184, acc=0.753, loss=43.107, backward_time=0.295, grad_norm=40.197, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.727e-04, train_time=1.204
+[gpub011:0/16] 2024-02-08 12:41:31,814 (trainer:762) INFO: 34epoch:train:8001-8100batch: iter_time=8.189e-05, forward_time=0.290, loss_ctc=45.173, loss_att=46.200, acc=0.760, loss=45.892, backward_time=0.295, grad_norm=41.039, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.727e-04, train_time=1.333
+[gpub011:0/16] 2024-02-08 12:43:33,659 (trainer:762) INFO: 34epoch:train:8101-8200batch: iter_time=8.882e-05, forward_time=0.295, loss_ctc=43.891, loss_att=46.248, acc=0.768, loss=45.541, backward_time=0.306, grad_norm=35.134, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.727e-04, train_time=1.219
+[gpub011:0/16] 2024-02-08 12:46:03,069 (trainer:762) INFO: 34epoch:train:8201-8300batch: iter_time=8.123e-05, forward_time=0.455, loss_ctc=36.305, loss_att=34.317, acc=0.793, loss=34.913, backward_time=0.316, grad_norm=31.965, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.727e-04, train_time=1.492
+[gpub011:0/16] 2024-02-08 12:48:20,731 (trainer:762) INFO: 34epoch:train:8301-8400batch: iter_time=8.828e-05, forward_time=0.290, loss_ctc=44.979, loss_att=41.215, acc=0.760, loss=42.344, backward_time=0.294, grad_norm=41.358, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.726e-04, train_time=1.378
+[gpub011:0/16] 2024-02-08 12:50:14,570 (trainer:762) INFO: 34epoch:train:8401-8500batch: iter_time=8.443e-05, forward_time=0.293, loss_ctc=50.671, loss_att=49.349, acc=0.756, loss=49.746, backward_time=0.301, grad_norm=43.866, clip=100.000, loss_scale=7.737e+33, optim_step_time=0.093, optim0_lr0=1.726e-04, train_time=1.138
+[gpub011:0/16] 2024-02-08 12:52:50,998 (trainer:762) INFO: 34epoch:train:8501-8600batch: iter_time=8.642e-05, forward_time=0.374, loss_ctc=43.771, loss_att=39.454, acc=0.769, loss=40.749, backward_time=0.336, grad_norm=42.040, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.726e-04, train_time=1.564
+[gpub011:0/16] 2024-02-08 12:55:07,777 (trainer:762) INFO: 34epoch:train:8601-8700batch: iter_time=8.426e-05, forward_time=0.353, loss_ctc=45.172, loss_att=39.213, acc=0.756, loss=41.001, backward_time=0.311, grad_norm=39.664, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.726e-04, train_time=1.366
+[gpub011:0/16] 2024-02-08 12:56:24,268 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-08 12:56:43,624 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 12:56:47,208 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a855ac0d0>)
+[gpub011:0/16] 2024-02-08 12:56:47,208 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-08 12:56:47,216 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 13:04:07,914 (trainer:762) INFO: 34epoch:train:8701-8800batch: iter_time=4.111, forward_time=0.396, loss_ctc=47.119, loss_att=47.813, acc=0.755, loss=47.605, backward_time=0.320, grad_norm=39.391, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.726e-04, train_time=5.403
+[gpub011:0/16] 2024-02-08 13:06:13,758 (trainer:762) INFO: 34epoch:train:8801-8900batch: iter_time=8.161e-05, forward_time=0.292, loss_ctc=50.589, loss_att=49.360, acc=0.764, loss=49.729, backward_time=0.295, grad_norm=45.335, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.726e-04, train_time=1.258
+[gpub011:0/16] 2024-02-08 13:08:28,265 (trainer:762) INFO: 34epoch:train:8901-9000batch: iter_time=8.013e-05, forward_time=0.289, loss_ctc=43.084, loss_att=39.633, acc=0.761, loss=40.668, backward_time=0.293, grad_norm=34.655, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.725e-04, train_time=1.344
+[gpub011:0/16] 2024-02-08 13:11:01,825 (trainer:762) INFO: 34epoch:train:9001-9100batch: iter_time=8.156e-05, forward_time=0.472, loss_ctc=41.984, loss_att=37.884, acc=0.774, loss=39.114, backward_time=0.356, grad_norm=38.814, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.725e-04, train_time=1.537
+[gpub011:0/16] 2024-02-08 13:13:31,145 (trainer:762) INFO: 34epoch:train:9101-9200batch: iter_time=8.418e-05, forward_time=0.300, loss_ctc=48.961, loss_att=51.279, acc=0.758, loss=50.583, backward_time=0.298, grad_norm=41.331, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.725e-04, train_time=1.493
+[gpub011:0/16] 2024-02-08 13:15:43,205 (trainer:762) INFO: 34epoch:train:9201-9300batch: iter_time=8.131e-05, forward_time=0.288, loss_ctc=45.416, loss_att=49.206, acc=0.740, loss=48.069, backward_time=0.294, grad_norm=44.645, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.725e-04, train_time=1.319
+[gpub011:0/16] 2024-02-08 13:16:45,060 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 13:18:08,613 (trainer:762) INFO: 34epoch:train:9301-9400batch: iter_time=8.272e-05, forward_time=0.373, loss_ctc=42.103, loss_att=42.667, acc=0.793, loss=42.498, backward_time=0.376, grad_norm=37.564, clip=100.000, loss_scale=7.133e+33, optim_step_time=0.099, optim0_lr0=1.725e-04, train_time=1.455
+[gpub011:0/16] 2024-02-08 13:20:39,104 (trainer:762) INFO: 34epoch:train:9401-9500batch: iter_time=8.229e-05, forward_time=0.290, loss_ctc=40.751, loss_att=37.910, acc=0.796, loss=38.762, backward_time=0.300, grad_norm=33.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.725e-04, train_time=1.505
+[gpub011:0/16] 2024-02-08 13:22:45,459 (trainer:762) INFO: 34epoch:train:9501-9600batch: iter_time=7.820e-05, forward_time=0.315, loss_ctc=40.687, loss_att=39.985, acc=0.773, loss=40.195, backward_time=0.293, grad_norm=38.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.724e-04, train_time=1.263
+[gpub011:0/16] 2024-02-08 13:25:14,210 (trainer:762) INFO: 34epoch:train:9601-9700batch: iter_time=8.172e-05, forward_time=0.446, loss_ctc=48.165, loss_att=49.612, acc=0.755, loss=49.178, backward_time=0.326, grad_norm=41.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.724e-04, train_time=1.487
+[gpub011:0/16] 2024-02-08 13:27:36,182 (trainer:762) INFO: 34epoch:train:9701-9800batch: iter_time=8.453e-05, forward_time=0.292, loss_ctc=47.715, loss_att=46.102, acc=0.761, loss=46.586, backward_time=0.300, grad_norm=61.500, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.724e-04, train_time=1.420
+[gpub011:0/16] 2024-02-08 13:29:56,705 (trainer:762) INFO: 34epoch:train:9801-9900batch: iter_time=8.318e-05, forward_time=0.289, loss_ctc=41.907, loss_att=37.019, acc=0.780, loss=38.486, backward_time=0.295, grad_norm=36.041, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.724e-04, train_time=1.405
+[gpub011:0/16] 2024-02-08 13:32:07,700 (trainer:762) INFO: 34epoch:train:9901-10000batch: iter_time=8.204e-05, forward_time=0.410, loss_ctc=43.771, loss_att=40.146, acc=0.769, loss=41.234, backward_time=0.314, grad_norm=38.050, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.724e-04, train_time=1.310
+[gpub011:0/16] 2024-02-08 13:32:27,729 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-08 13:32:46,873 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 13:32:50,463 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fedadf7c0>)
+[gpub011:0/16] 2024-02-08 13:32:50,463 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-08 13:32:50,466 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 13:40:20,880 (trainer:762) INFO: 34epoch:train:10001-10100batch: iter_time=3.610, forward_time=0.297, loss_ctc=54.128, loss_att=54.992, acc=0.739, loss=54.733, backward_time=0.298, grad_norm=49.218, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.724e-04, train_time=4.931
+[gpub011:0/16] 2024-02-08 13:42:25,755 (trainer:762) INFO: 34epoch:train:10101-10200batch: iter_time=8.427e-05, forward_time=0.291, loss_ctc=49.717, loss_att=45.107, acc=0.762, loss=46.490, backward_time=0.297, grad_norm=40.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.723e-04, train_time=1.249
+[gpub011:0/16] 2024-02-08 13:44:50,838 (trainer:762) INFO: 34epoch:train:10201-10300batch: iter_time=9.126e-05, forward_time=0.399, loss_ctc=37.502, loss_att=34.396, acc=0.778, loss=35.328, backward_time=0.336, grad_norm=35.053, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.723e-04, train_time=1.450
+[gpub011:0/16] 2024-02-08 13:47:01,911 (trainer:762) INFO: 34epoch:train:10301-10400batch: iter_time=1.022e-04, forward_time=0.302, loss_ctc=46.535, loss_att=49.578, acc=0.756, loss=48.665, backward_time=0.299, grad_norm=40.688, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.723e-04, train_time=1.310
+[gpub011:0/16] 2024-02-08 13:48:57,467 (trainer:762) INFO: 34epoch:train:10401-10500batch: iter_time=8.622e-05, forward_time=0.313, loss_ctc=45.222, loss_att=42.366, acc=0.761, loss=43.223, backward_time=0.297, grad_norm=40.045, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.723e-04, train_time=1.155
+[gpub011:0/16] 2024-02-08 13:51:24,796 (trainer:762) INFO: 34epoch:train:10501-10600batch: iter_time=8.837e-05, forward_time=0.304, loss_ctc=45.001, loss_att=47.601, acc=0.764, loss=46.821, backward_time=0.313, grad_norm=41.786, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.723e-04, train_time=1.473
+[gpub011:0/16] 2024-02-08 13:53:50,045 (trainer:762) INFO: 34epoch:train:10601-10700batch: iter_time=8.837e-05, forward_time=0.396, loss_ctc=44.051, loss_att=45.033, acc=0.789, loss=44.738, backward_time=0.316, grad_norm=34.939, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.723e-04, train_time=1.452
+[gpub011:0/16] 2024-02-08 13:55:42,060 (trainer:762) INFO: 34epoch:train:10701-10800batch: iter_time=9.164e-05, forward_time=0.289, loss_ctc=36.511, loss_att=34.299, acc=0.798, loss=34.963, backward_time=0.294, grad_norm=32.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.722e-04, train_time=1.120
+[gpub011:0/16] 2024-02-08 13:58:23,321 (trainer:762) INFO: 34epoch:train:10801-10900batch: iter_time=9.180e-05, forward_time=0.289, loss_ctc=44.186, loss_att=40.270, acc=0.771, loss=41.445, backward_time=0.293, grad_norm=39.241, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.722e-04, train_time=1.612
+[gpub011:0/16] 2024-02-08 14:00:43,176 (trainer:762) INFO: 34epoch:train:10901-11000batch: iter_time=8.566e-05, forward_time=0.364, loss_ctc=50.329, loss_att=53.192, acc=0.748, loss=52.333, backward_time=0.340, grad_norm=44.407, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.722e-04, train_time=1.399
+[gpub011:0/16] 2024-02-08 14:03:45,518 (trainer:762) INFO: 34epoch:train:11001-11100batch: iter_time=0.075, forward_time=0.290, loss_ctc=43.879, loss_att=39.213, acc=0.775, loss=40.613, backward_time=0.298, grad_norm=40.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.722e-04, train_time=1.822
+[gpub011:0/16] 2024-02-08 14:05:53,764 (trainer:762) INFO: 34epoch:train:11101-11200batch: iter_time=9.086e-05, forward_time=0.292, loss_ctc=44.895, loss_att=38.957, acc=0.769, loss=40.739, backward_time=0.294, grad_norm=39.276, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.722e-04, train_time=1.283
+[gpub011:0/16] 2024-02-08 14:07:31,611 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-08 14:07:51,578 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 14:07:55,382 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8f9e157d60>)
+[gpub011:0/16] 2024-02-08 14:07:55,382 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-08 14:07:55,385 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 14:14:34,160 (trainer:762) INFO: 34epoch:train:11201-11300batch: iter_time=3.761, forward_time=0.373, loss_ctc=46.600, loss_att=46.699, acc=0.760, loss=46.669, backward_time=0.311, grad_norm=37.770, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.722e-04, train_time=5.204
+[gpub011:0/16] 2024-02-08 14:16:22,477 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 14:16:30,150 (trainer:762) INFO: 34epoch:train:11301-11400batch: iter_time=7.890e-05, forward_time=0.293, loss_ctc=50.368, loss_att=47.943, acc=0.766, loss=48.671, backward_time=0.296, grad_norm=44.266, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.093, optim0_lr0=1.721e-04, train_time=1.159
+[gpub011:0/16] 2024-02-08 14:18:37,889 (trainer:762) INFO: 34epoch:train:11401-11500batch: iter_time=8.058e-05, forward_time=0.289, loss_ctc=43.023, loss_att=39.534, acc=0.761, loss=40.581, backward_time=0.293, grad_norm=36.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.721e-04, train_time=1.277
+[gpub011:0/16] 2024-02-08 14:21:19,385 (trainer:762) INFO: 34epoch:train:11501-11600batch: iter_time=4.459e-04, forward_time=0.381, loss_ctc=41.958, loss_att=37.329, acc=0.775, loss=38.717, backward_time=0.341, grad_norm=37.450, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.721e-04, train_time=1.614
+[gpub011:0/16] 2024-02-08 14:23:24,228 (trainer:762) INFO: 34epoch:train:11601-11700batch: iter_time=8.122e-05, forward_time=0.292, loss_ctc=49.020, loss_att=50.653, acc=0.759, loss=50.163, backward_time=0.299, grad_norm=39.591, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.721e-04, train_time=1.249
+[gpub011:0/16] 2024-02-08 14:25:45,933 (trainer:762) INFO: 34epoch:train:11701-11800batch: iter_time=8.115e-05, forward_time=0.289, loss_ctc=45.060, loss_att=48.626, acc=0.743, loss=47.556, backward_time=0.293, grad_norm=41.161, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.721e-04, train_time=1.416
+[gpub011:0/16] 2024-02-08 14:27:52,638 (trainer:762) INFO: 34epoch:train:11801-11900batch: iter_time=8.272e-05, forward_time=0.293, loss_ctc=41.960, loss_att=42.336, acc=0.793, loss=42.223, backward_time=0.299, grad_norm=37.516, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.721e-04, train_time=1.267
+[gpub011:0/16] 2024-02-08 14:30:04,718 (trainer:762) INFO: 34epoch:train:11901-12000batch: iter_time=8.013e-05, forward_time=0.386, loss_ctc=40.395, loss_att=37.472, acc=0.797, loss=38.349, backward_time=0.322, grad_norm=33.040, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.720e-04, train_time=1.319
+[gpub011:0/16] 2024-02-08 14:32:20,776 (trainer:762) INFO: 34epoch:train:12001-12100batch: iter_time=8.022e-05, forward_time=0.289, loss_ctc=40.373, loss_att=39.811, acc=0.773, loss=39.979, backward_time=0.292, grad_norm=39.927, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.720e-04, train_time=1.362
+[gpub011:0/16] 2024-02-08 14:33:43,825 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 14:34:16,248 (trainer:762) INFO: 34epoch:train:12101-12200batch: iter_time=8.041e-05, forward_time=0.293, loss_ctc=48.024, loss_att=49.126, acc=0.755, loss=48.795, backward_time=0.298, grad_norm=39.669, clip=100.000, loss_scale=4.484e+33, optim_step_time=0.093, optim0_lr0=1.720e-04, train_time=1.154
+[gpub011:0/16] 2024-02-08 14:36:43,854 (trainer:762) INFO: 34epoch:train:12201-12300batch: iter_time=8.022e-05, forward_time=0.403, loss_ctc=47.858, loss_att=45.654, acc=0.762, loss=46.315, backward_time=0.317, grad_norm=45.957, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.720e-04, train_time=1.476
+[gpub011:0/16] 2024-02-08 14:38:46,576 (trainer:762) INFO: 34epoch:train:12301-12400batch: iter_time=8.037e-05, forward_time=0.289, loss_ctc=41.823, loss_att=36.946, acc=0.781, loss=38.409, backward_time=0.296, grad_norm=38.021, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.720e-04, train_time=1.226
+[gpub011:0/16] 2024-02-08 14:40:45,108 (trainer:762) INFO: 34epoch:train:12401-12500batch: iter_time=7.929e-05, forward_time=0.289, loss_ctc=43.733, loss_att=40.115, acc=0.770, loss=41.200, backward_time=0.295, grad_norm=39.057, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.719e-04, train_time=1.186
+[gpub011:0/16] 2024-02-08 14:41:05,137 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-08 14:41:24,632 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 14:41:28,183 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fcfcffcd0>)
+[gpub011:0/16] 2024-02-08 14:41:28,183 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-08 14:41:28,186 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 14:48:23,696 (trainer:762) INFO: 34epoch:train:12501-12600batch: iter_time=3.173, forward_time=0.383, loss_ctc=54.116, loss_att=56.584, acc=0.734, loss=55.844, backward_time=0.347, grad_norm=49.111, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.719e-04, train_time=4.586
+[gpub011:0/16] 2024-02-08 14:50:34,704 (trainer:762) INFO: 34epoch:train:12601-12700batch: iter_time=8.263e-05, forward_time=0.290, loss_ctc=49.487, loss_att=44.902, acc=0.754, loss=46.277, backward_time=0.294, grad_norm=39.475, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.719e-04, train_time=1.309
+[gpub011:0/16] 2024-02-08 14:53:01,377 (trainer:762) INFO: 34epoch:train:12701-12800batch: iter_time=8.161e-05, forward_time=0.286, loss_ctc=37.968, loss_att=34.868, acc=0.770, loss=35.798, backward_time=0.290, grad_norm=34.859, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.719e-04, train_time=1.467
+[gpub011:0/16] 2024-02-08 14:55:13,904 (trainer:762) INFO: 34epoch:train:12801-12900batch: iter_time=1.678e-04, forward_time=0.390, loss_ctc=46.662, loss_att=49.466, acc=0.750, loss=48.625, backward_time=0.338, grad_norm=41.372, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.719e-04, train_time=1.325
+[gpub011:0/16] 2024-02-08 14:57:33,754 (trainer:762) INFO: 34epoch:train:12901-13000batch: iter_time=8.368e-05, forward_time=0.288, loss_ctc=44.807, loss_att=42.359, acc=0.755, loss=43.093, backward_time=0.293, grad_norm=40.918, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.719e-04, train_time=1.397
+[gpub011:0/16] 2024-02-08 14:59:43,402 (trainer:762) INFO: 34epoch:train:13001-13100batch: iter_time=8.286e-05, forward_time=0.290, loss_ctc=44.878, loss_att=46.352, acc=0.761, loss=45.910, backward_time=0.296, grad_norm=40.584, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.718e-04, train_time=1.298
+[gpub011:0/16] 2024-02-08 15:02:20,603 (trainer:762) INFO: 34epoch:train:13101-13200batch: iter_time=8.345e-05, forward_time=0.403, loss_ctc=43.919, loss_att=46.072, acc=0.769, loss=45.426, backward_time=0.319, grad_norm=37.121, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.718e-04, train_time=1.572
+[gpub011:0/16] 2024-02-08 15:04:09,515 (trainer:762) INFO: 34epoch:train:13201-13300batch: iter_time=8.286e-05, forward_time=0.288, loss_ctc=36.974, loss_att=35.030, acc=0.790, loss=35.614, backward_time=0.293, grad_norm=32.753, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.718e-04, train_time=1.088
+[gpub011:0/16] 2024-02-08 15:06:28,111 (trainer:762) INFO: 34epoch:train:13301-13400batch: iter_time=8.108e-05, forward_time=0.296, loss_ctc=44.316, loss_att=40.885, acc=0.764, loss=41.914, backward_time=0.295, grad_norm=41.463, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.718e-04, train_time=1.387
+[gpub011:0/16] 2024-02-08 15:08:53,332 (trainer:762) INFO: 34epoch:train:13401-13500batch: iter_time=8.241e-05, forward_time=0.377, loss_ctc=50.027, loss_att=49.823, acc=0.754, loss=49.884, backward_time=0.340, grad_norm=43.189, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.718e-04, train_time=1.452
+[gpub011:0/16] 2024-02-08 15:10:47,119 (trainer:762) INFO: 34epoch:train:13501-13600batch: iter_time=8.497e-05, forward_time=0.289, loss_ctc=43.551, loss_att=38.563, acc=0.774, loss=40.060, backward_time=0.295, grad_norm=41.093, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.718e-04, train_time=1.137
+[gpub011:0/16] 2024-02-08 15:13:15,978 (trainer:762) INFO: 34epoch:train:13601-13700batch: iter_time=8.153e-05, forward_time=0.288, loss_ctc=45.023, loss_att=39.442, acc=0.756, loss=41.116, backward_time=0.292, grad_norm=40.864, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.717e-04, train_time=1.488
+[gpub011:0/16] 2024-02-08 15:14:49,565 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-08 15:15:08,998 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 15:15:12,531 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99d5932380>)
+[gpub011:0/16] 2024-02-08 15:15:12,531 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-08 15:15:12,536 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 15:21:21,934 (trainer:762) INFO: 34epoch:train:13701-13800batch: iter_time=3.446, forward_time=0.444, loss_ctc=46.843, loss_att=47.520, acc=0.753, loss=47.317, backward_time=0.324, grad_norm=40.184, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.717e-04, train_time=4.859
+[gpub011:0/16] 2024-02-08 15:23:39,517 (trainer:762) INFO: 34epoch:train:13801-13900batch: iter_time=7.559e-05, forward_time=0.289, loss_ctc=50.395, loss_att=46.170, acc=0.763, loss=47.438, backward_time=0.295, grad_norm=43.832, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.717e-04, train_time=1.376
+[gpub011:0/16] 2024-02-08 15:25:52,825 (trainer:762) INFO: 34epoch:train:13901-14000batch: iter_time=8.005e-05, forward_time=0.286, loss_ctc=43.325, loss_att=38.823, acc=0.759, loss=40.174, backward_time=0.296, grad_norm=35.443, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.717e-04, train_time=1.332
+[gpub011:0/16] 2024-02-08 15:28:00,782 (trainer:762) INFO: 34epoch:train:14001-14100batch: iter_time=8.083e-05, forward_time=0.419, loss_ctc=41.458, loss_att=37.113, acc=0.767, loss=38.416, backward_time=0.330, grad_norm=36.362, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.717e-04, train_time=1.280
+[gpub011:0/16] 2024-02-08 15:30:19,896 (trainer:762) INFO: 34epoch:train:14101-14200batch: iter_time=8.268e-05, forward_time=0.292, loss_ctc=48.679, loss_att=49.798, acc=0.753, loss=49.462, backward_time=0.297, grad_norm=42.026, clip=100.000, loss_scale=3.297e+33, optim_step_time=0.093, optim0_lr0=1.717e-04, train_time=1.391
+[gpub011:0/16] 2024-02-08 15:32:25,020 (trainer:762) INFO: 34epoch:train:14201-14300batch: iter_time=8.123e-05, forward_time=0.289, loss_ctc=45.352, loss_att=47.549, acc=0.738, loss=46.890, backward_time=0.295, grad_norm=42.753, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.716e-04, train_time=1.251
+[gpub011:0/16] 2024-02-08 15:34:34,700 (trainer:762) INFO: 34epoch:train:14301-14400batch: iter_time=7.951e-05, forward_time=0.405, loss_ctc=41.833, loss_att=42.378, acc=0.787, loss=42.215, backward_time=0.341, grad_norm=37.272, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.716e-04, train_time=1.296
+[gpub011:0/16] 2024-02-08 15:36:51,247 (trainer:762) INFO: 34epoch:train:14401-14500batch: iter_time=8.359e-05, forward_time=0.289, loss_ctc=41.011, loss_att=38.698, acc=0.780, loss=39.392, backward_time=0.294, grad_norm=35.438, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.716e-04, train_time=1.365
+[gpub011:0/16] 2024-02-08 15:38:58,010 (trainer:762) INFO: 34epoch:train:14501-14600batch: iter_time=8.340e-05, forward_time=0.288, loss_ctc=39.848, loss_att=40.033, acc=0.768, loss=39.978, backward_time=0.294, grad_norm=39.966, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.716e-04, train_time=1.268
+[gpub011:0/16] 2024-02-08 15:40:59,507 (trainer:762) INFO: 34epoch:train:14601-14700batch: iter_time=8.194e-05, forward_time=0.339, loss_ctc=47.812, loss_att=47.056, acc=0.755, loss=47.283, backward_time=0.301, grad_norm=39.182, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.716e-04, train_time=1.215
+[gpub011:0/16] 2024-02-08 15:43:19,589 (trainer:762) INFO: 34epoch:train:14701-14800batch: iter_time=8.567e-05, forward_time=0.338, loss_ctc=47.953, loss_att=43.489, acc=0.767, loss=44.828, backward_time=0.317, grad_norm=44.785, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.716e-04, train_time=1.401
+[gpub011:0/16] 2024-02-08 15:45:27,954 (trainer:762) INFO: 34epoch:train:14801-14900batch: iter_time=8.139e-05, forward_time=0.291, loss_ctc=41.749, loss_att=36.820, acc=0.775, loss=38.299, backward_time=0.296, grad_norm=37.113, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.715e-04, train_time=1.283
+[gpub011:0/16] 2024-02-08 15:47:40,200 (trainer:762) INFO: 34epoch:train:14901-15000batch: iter_time=2.338e-04, forward_time=0.319, loss_ctc=43.271, loss_att=40.169, acc=0.761, loss=41.100, backward_time=0.300, grad_norm=37.762, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.715e-04, train_time=1.322
+[gpub011:0/16] 2024-02-08 16:25:55,421 (trainer:361) INFO: 34epoch results: [train] iter_time=0.290, forward_time=0.336, loss_ctc=45.396, loss_att=43.729, acc=0.763, loss=44.229, backward_time=0.312, grad_norm=40.439, clip=100.000, loss_scale=6.019e+33, optim_step_time=0.096, optim0_lr0=1.728e-04, train_time=1.673, time=6 hours, 58 minutes and 46.82 seconds, total_count=540000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=35.899, cer_ctc=0.183, loss_att=37.243, acc=0.701, cer=0.285, wer=0.989, loss=36.840, time=37 minutes and 50.89 seconds, total_count=168156, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-08 16:26:14,011 (trainer:416) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub011:0/16] 2024-02-08 16:26:14,072 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/29epoch.pth
+[gpub011:0/16] 2024-02-08 16:26:14,072 (trainer:290) INFO: 35/45epoch started. Estimated time to finish: 3 days, 11 hours and 46 minutes
+[gpub011:0/16] 2024-02-08 16:26:14,083 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-08 16:26:32,006 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 16:26:35,447 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065a0fd30>)
+[gpub011:0/16] 2024-02-08 16:26:35,447 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-08 16:26:35,451 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 16:33:14,995 (trainer:762) INFO: 35epoch:train:1-100batch: iter_time=2.937, forward_time=0.374, loss_ctc=52.199, loss_att=54.027, acc=0.750, loss=53.479, backward_time=0.308, grad_norm=45.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.715e-04, train_time=4.208
+[gpub011:0/16] 2024-02-08 16:35:21,607 (trainer:762) INFO: 35epoch:train:101-200batch: iter_time=8.114e-05, forward_time=0.330, loss_ctc=47.202, loss_att=42.071, acc=0.775, loss=43.610, backward_time=0.341, grad_norm=43.197, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.715e-04, train_time=1.267
+[gpub011:0/16] 2024-02-08 16:37:24,935 (trainer:762) INFO: 35epoch:train:201-300batch: iter_time=8.363e-05, forward_time=0.333, loss_ctc=46.495, loss_att=50.000, acc=0.745, loss=48.949, backward_time=0.319, grad_norm=40.618, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.715e-04, train_time=1.233
+[gpub011:0/16] 2024-02-08 16:39:53,081 (trainer:762) INFO: 35epoch:train:301-400batch: iter_time=8.480e-05, forward_time=0.317, loss_ctc=43.986, loss_att=42.491, acc=0.767, loss=42.940, backward_time=0.308, grad_norm=38.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.715e-04, train_time=1.481
+[gpub011:0/16] 2024-02-08 16:42:00,008 (trainer:762) INFO: 35epoch:train:401-500batch: iter_time=2.459e-04, forward_time=0.345, loss_ctc=56.322, loss_att=55.749, acc=0.755, loss=55.921, backward_time=0.331, grad_norm=47.730, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.714e-04, train_time=1.269
+[gpub011:0/16] 2024-02-08 16:43:53,356 (trainer:762) INFO: 35epoch:train:501-600batch: iter_time=8.159e-05, forward_time=0.296, loss_ctc=46.042, loss_att=47.758, acc=0.760, loss=47.243, backward_time=0.301, grad_norm=41.242, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.714e-04, train_time=1.133
+[gpub011:0/16] 2024-02-08 16:46:20,788 (trainer:762) INFO: 35epoch:train:601-700batch: iter_time=8.567e-04, forward_time=0.350, loss_ctc=44.055, loss_att=41.378, acc=0.782, loss=42.181, backward_time=0.329, grad_norm=36.641, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.714e-04, train_time=1.474
+[gpub011:0/16] 2024-02-08 16:48:30,531 (trainer:762) INFO: 35epoch:train:701-800batch: iter_time=8.222e-05, forward_time=0.303, loss_ctc=46.808, loss_att=50.666, acc=0.744, loss=49.509, backward_time=0.303, grad_norm=44.429, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.714e-04, train_time=1.298
+[gpub011:0/16] 2024-02-08 16:50:27,119 (trainer:762) INFO: 35epoch:train:801-900batch: iter_time=7.913e-05, forward_time=0.339, loss_ctc=39.056, loss_att=41.609, acc=0.737, loss=40.843, backward_time=0.305, grad_norm=36.849, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.714e-04, train_time=1.166
+[gpub011:0/16] 2024-02-08 16:52:37,160 (trainer:762) INFO: 35epoch:train:901-1000batch: iter_time=3.822e-04, forward_time=0.369, loss_ctc=46.698, loss_att=45.589, acc=0.763, loss=45.922, backward_time=0.319, grad_norm=37.717, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.714e-04, train_time=1.300
+[gpub011:0/16] 2024-02-08 16:54:48,544 (trainer:762) INFO: 35epoch:train:1001-1100batch: iter_time=8.385e-05, forward_time=0.297, loss_ctc=43.837, loss_att=41.572, acc=0.754, loss=42.252, backward_time=0.297, grad_norm=44.640, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.713e-04, train_time=1.314
+[gpub011:0/16] 2024-02-08 16:57:08,340 (trainer:762) INFO: 35epoch:train:1101-1200batch: iter_time=4.537e-04, forward_time=0.375, loss_ctc=48.861, loss_att=46.300, acc=0.760, loss=47.069, backward_time=0.340, grad_norm=43.203, clip=100.000, loss_scale=6.594e+33, optim_step_time=0.099, optim0_lr0=1.713e-04, train_time=1.398
+[gpub011:0/16] 2024-02-08 16:58:24,745 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-08 16:58:43,866 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 16:58:47,492 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065b68cd0>)
+[gpub011:0/16] 2024-02-08 16:58:47,492 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-08 16:58:47,495 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 17:05:12,167 (trainer:762) INFO: 35epoch:train:1201-1300batch: iter_time=3.559, forward_time=0.298, loss_ctc=48.965, loss_att=52.243, acc=0.759, loss=51.260, backward_time=0.301, grad_norm=41.055, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.713e-04, train_time=4.838
+[gpub011:0/16] 2024-02-08 17:07:10,316 (trainer:762) INFO: 35epoch:train:1301-1400batch: iter_time=8.335e-05, forward_time=0.297, loss_ctc=49.422, loss_att=48.766, acc=0.764, loss=48.963, backward_time=0.301, grad_norm=39.116, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.713e-04, train_time=1.181
+[gpub011:0/16] 2024-02-08 17:09:26,491 (trainer:762) INFO: 35epoch:train:1401-1500batch: iter_time=8.500e-05, forward_time=0.400, loss_ctc=46.273, loss_att=37.764, acc=0.770, loss=40.317, backward_time=0.331, grad_norm=95.939, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.713e-04, train_time=1.361
+[gpub011:0/16] 2024-02-08 17:11:46,362 (trainer:762) INFO: 35epoch:train:1501-1600batch: iter_time=8.315e-05, forward_time=0.332, loss_ctc=45.324, loss_att=50.452, acc=0.757, loss=48.914, backward_time=0.303, grad_norm=38.388, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.713e-04, train_time=1.397
+[gpub011:0/16] 2024-02-08 17:13:45,747 (trainer:762) INFO: 35epoch:train:1601-1700batch: iter_time=8.328e-05, forward_time=0.292, loss_ctc=48.989, loss_att=51.962, acc=0.748, loss=51.070, backward_time=0.299, grad_norm=44.229, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.712e-04, train_time=1.195
+[gpub011:0/16] 2024-02-08 17:15:57,765 (trainer:762) INFO: 35epoch:train:1701-1800batch: iter_time=8.480e-05, forward_time=0.319, loss_ctc=48.875, loss_att=49.748, acc=0.778, loss=49.486, backward_time=0.316, grad_norm=38.019, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.712e-04, train_time=1.320
+[gpub011:0/16] 2024-02-08 17:18:10,188 (trainer:762) INFO: 35epoch:train:1801-1900batch: iter_time=8.834e-05, forward_time=0.382, loss_ctc=43.372, loss_att=42.695, acc=0.774, loss=42.898, backward_time=0.319, grad_norm=39.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.712e-04, train_time=1.324
+[gpub011:0/16] 2024-02-08 17:20:21,579 (trainer:762) INFO: 35epoch:train:1901-2000batch: iter_time=8.044e-05, forward_time=0.290, loss_ctc=46.071, loss_att=42.234, acc=0.765, loss=43.385, backward_time=0.295, grad_norm=39.313, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.712e-04, train_time=1.313
+[gpub011:0/16] 2024-02-08 17:22:18,478 (trainer:762) INFO: 35epoch:train:2001-2100batch: iter_time=8.390e-05, forward_time=0.297, loss_ctc=43.134, loss_att=51.213, acc=0.731, loss=48.789, backward_time=0.301, grad_norm=38.940, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.712e-04, train_time=1.170
+[gpub011:0/16] 2024-02-08 17:24:36,344 (trainer:762) INFO: 35epoch:train:2101-2200batch: iter_time=8.788e-05, forward_time=0.322, loss_ctc=40.663, loss_att=42.242, acc=0.751, loss=41.768, backward_time=0.321, grad_norm=38.043, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.712e-04, train_time=1.378
+[gpub011:0/16] 2024-02-08 17:25:30,839 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 17:27:02,300 (trainer:762) INFO: 35epoch:train:2201-2300batch: iter_time=1.320e-04, forward_time=0.380, loss_ctc=43.434, loss_att=39.049, acc=0.765, loss=40.365, backward_time=0.305, grad_norm=40.042, clip=100.000, loss_scale=7.080e+33, optim_step_time=0.095, optim0_lr0=1.711e-04, train_time=1.459
+[gpub011:0/16] 2024-02-08 17:29:03,321 (trainer:762) INFO: 35epoch:train:2301-2400batch: iter_time=8.387e-05, forward_time=0.291, loss_ctc=44.102, loss_att=41.233, acc=0.775, loss=42.094, backward_time=0.296, grad_norm=40.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.711e-04, train_time=1.209
+[gpub011:0/16] 2024-02-08 17:31:20,859 (trainer:762) INFO: 35epoch:train:2401-2500batch: iter_time=7.975e-05, forward_time=0.302, loss_ctc=51.943, loss_att=54.736, acc=0.745, loss=53.898, backward_time=0.316, grad_norm=48.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.711e-04, train_time=1.377
+[gpub011:0/16] 2024-02-08 17:31:40,916 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-08 17:31:59,763 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 17:32:03,292 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fb43a3af0>)
+[gpub011:0/16] 2024-02-08 17:32:03,292 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-08 17:32:03,295 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 17:39:32,486 (trainer:762) INFO: 35epoch:train:2501-2600batch: iter_time=3.575, forward_time=0.383, loss_ctc=50.139, loss_att=50.854, acc=0.759, loss=50.639, backward_time=0.313, grad_norm=41.095, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.711e-04, train_time=4.916
+[gpub011:0/16] 2024-02-08 17:41:32,116 (trainer:762) INFO: 35epoch:train:2601-2700batch: iter_time=8.029e-05, forward_time=0.290, loss_ctc=46.325, loss_att=41.124, acc=0.779, loss=42.685, backward_time=0.296, grad_norm=42.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.711e-04, train_time=1.195
+[gpub011:0/16] 2024-02-08 17:43:52,793 (trainer:762) INFO: 35epoch:train:2701-2800batch: iter_time=8.049e-05, forward_time=0.308, loss_ctc=45.476, loss_att=48.284, acc=0.752, loss=47.441, backward_time=0.319, grad_norm=40.880, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.711e-04, train_time=1.408
+[gpub011:0/16] 2024-02-08 17:46:02,904 (trainer:762) INFO: 35epoch:train:2801-2900batch: iter_time=9.127e-05, forward_time=0.363, loss_ctc=43.619, loss_att=42.446, acc=0.770, loss=42.798, backward_time=0.307, grad_norm=38.418, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.710e-04, train_time=1.301
+[gpub011:0/16] 2024-02-08 17:48:14,859 (trainer:762) INFO: 35epoch:train:2901-3000batch: iter_time=8.487e-05, forward_time=0.294, loss_ctc=53.133, loss_att=55.330, acc=0.758, loss=54.671, backward_time=0.299, grad_norm=44.340, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.710e-04, train_time=1.319
+[gpub011:0/16] 2024-02-08 17:50:22,774 (trainer:762) INFO: 35epoch:train:3001-3100batch: iter_time=8.263e-05, forward_time=0.337, loss_ctc=44.930, loss_att=46.663, acc=0.767, loss=46.143, backward_time=0.329, grad_norm=41.407, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.710e-04, train_time=1.278
+[gpub011:0/16] 2024-02-08 17:52:34,000 (trainer:762) INFO: 35epoch:train:3101-3200batch: iter_time=9.230e-05, forward_time=0.354, loss_ctc=43.908, loss_att=40.863, acc=0.784, loss=41.776, backward_time=0.313, grad_norm=35.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.710e-04, train_time=1.314
+[gpub011:0/16] 2024-02-08 17:54:50,430 (trainer:762) INFO: 35epoch:train:3201-3300batch: iter_time=8.369e-05, forward_time=0.292, loss_ctc=46.005, loss_att=50.265, acc=0.746, loss=48.987, backward_time=0.299, grad_norm=43.439, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.710e-04, train_time=1.364
+[gpub011:0/16] 2024-02-08 17:57:05,040 (trainer:762) INFO: 35epoch:train:3301-3400batch: iter_time=7.984e-05, forward_time=0.414, loss_ctc=38.651, loss_att=40.649, acc=0.741, loss=40.050, backward_time=0.312, grad_norm=36.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.119, optim0_lr0=1.710e-04, train_time=1.346
+[gpub011:0/16] 2024-02-08 17:59:10,068 (trainer:762) INFO: 35epoch:train:3401-3500batch: iter_time=7.879e-05, forward_time=0.292, loss_ctc=46.035, loss_att=45.117, acc=0.768, loss=45.393, backward_time=0.297, grad_norm=37.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.709e-04, train_time=1.251
+[gpub011:0/16] 2024-02-08 18:01:17,351 (trainer:762) INFO: 35epoch:train:3501-3600batch: iter_time=7.976e-05, forward_time=0.287, loss_ctc=41.551, loss_att=40.274, acc=0.757, loss=40.657, backward_time=0.294, grad_norm=44.736, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.709e-04, train_time=1.272
+[gpub011:0/16] 2024-02-08 18:03:41,592 (trainer:762) INFO: 35epoch:train:3601-3700batch: iter_time=8.113e-05, forward_time=0.433, loss_ctc=47.286, loss_att=45.590, acc=0.763, loss=46.098, backward_time=0.318, grad_norm=41.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.709e-04, train_time=1.442
+[gpub011:0/16] 2024-02-08 18:05:03,845 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-08 18:05:23,141 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 18:05:26,674 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fbd4f7dc0>)
+[gpub011:0/16] 2024-02-08 18:05:26,674 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-08 18:05:26,677 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 18:11:51,122 (trainer:762) INFO: 35epoch:train:3701-3800batch: iter_time=3.584, forward_time=0.295, loss_ctc=48.234, loss_att=53.429, acc=0.753, loss=51.870, backward_time=0.298, grad_norm=41.268, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.709e-04, train_time=4.895
+[gpub011:0/16] 2024-02-08 18:13:42,827 (trainer:762) INFO: 35epoch:train:3801-3900batch: iter_time=7.839e-05, forward_time=0.290, loss_ctc=49.058, loss_att=50.735, acc=0.753, loss=50.232, backward_time=0.298, grad_norm=41.582, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.709e-04, train_time=1.116
+[gpub011:0/16] 2024-02-08 18:15:57,529 (trainer:762) INFO: 35epoch:train:3901-4000batch: iter_time=7.921e-05, forward_time=0.443, loss_ctc=45.346, loss_att=37.555, acc=0.770, loss=39.892, backward_time=0.313, grad_norm=42.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.709e-04, train_time=1.347
+[gpub011:0/16] 2024-02-08 18:18:33,323 (trainer:762) INFO: 35epoch:train:4001-4100batch: iter_time=7.846e-05, forward_time=0.292, loss_ctc=44.924, loss_att=48.963, acc=0.747, loss=47.751, backward_time=0.296, grad_norm=39.556, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.708e-04, train_time=1.558
+[gpub011:0/16] 2024-02-08 18:20:24,789 (trainer:762) INFO: 35epoch:train:4101-4200batch: iter_time=7.726e-05, forward_time=0.292, loss_ctc=48.219, loss_att=50.584, acc=0.741, loss=49.874, backward_time=0.299, grad_norm=46.282, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.708e-04, train_time=1.114
+[gpub011:0/16] 2024-02-08 18:22:42,244 (trainer:762) INFO: 35epoch:train:4201-4300batch: iter_time=7.857e-05, forward_time=0.425, loss_ctc=48.725, loss_att=50.447, acc=0.770, loss=49.931, backward_time=0.341, grad_norm=39.195, clip=100.000, loss_scale=8.463e+33, optim_step_time=0.097, optim0_lr0=1.708e-04, train_time=1.374
+[gpub011:0/16] 2024-02-08 18:25:07,818 (trainer:762) INFO: 35epoch:train:4301-4400batch: iter_time=7.837e-05, forward_time=0.289, loss_ctc=42.966, loss_att=42.549, acc=0.766, loss=42.674, backward_time=0.295, grad_norm=39.043, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.708e-04, train_time=1.455
+[gpub011:0/16] 2024-02-08 18:27:06,502 (trainer:762) INFO: 35epoch:train:4401-4500batch: iter_time=7.984e-05, forward_time=0.289, loss_ctc=45.580, loss_att=40.855, acc=0.762, loss=42.273, backward_time=0.294, grad_norm=44.341, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.708e-04, train_time=1.187
+[gpub011:0/16] 2024-02-08 18:29:04,625 (trainer:762) INFO: 35epoch:train:4501-4600batch: iter_time=7.953e-05, forward_time=0.321, loss_ctc=42.602, loss_att=49.514, acc=0.730, loss=47.440, backward_time=0.297, grad_norm=37.960, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.708e-04, train_time=1.181
+[gpub011:0/16] 2024-02-08 18:31:53,222 (trainer:762) INFO: 35epoch:train:4601-4700batch: iter_time=8.233e-05, forward_time=0.371, loss_ctc=40.408, loss_att=41.999, acc=0.738, loss=41.522, backward_time=0.348, grad_norm=37.590, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.707e-04, train_time=1.686
+[gpub011:0/16] 2024-02-08 18:33:49,324 (trainer:762) INFO: 35epoch:train:4701-4800batch: iter_time=7.895e-05, forward_time=0.287, loss_ctc=42.501, loss_att=39.387, acc=0.757, loss=40.321, backward_time=0.294, grad_norm=38.000, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.707e-04, train_time=1.160
+[gpub011:0/16] 2024-02-08 18:36:12,163 (trainer:762) INFO: 35epoch:train:4801-4900batch: iter_time=7.938e-05, forward_time=0.290, loss_ctc=43.502, loss_att=39.887, acc=0.773, loss=40.971, backward_time=0.294, grad_norm=37.693, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.707e-04, train_time=1.429
+[gpub011:0/16] 2024-02-08 18:38:14,232 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 18:38:15,315 (trainer:762) INFO: 35epoch:train:4901-5000batch: iter_time=7.865e-05, forward_time=0.379, loss_ctc=51.467, loss_att=54.552, acc=0.743, loss=53.627, backward_time=0.330, grad_norm=47.873, clip=100.000, loss_scale=1.033e+34, optim_step_time=0.097, optim0_lr0=1.707e-04, train_time=1.231
+[gpub011:0/16] 2024-02-08 18:38:35,380 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-08 18:38:54,823 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 18:38:58,418 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9835c4f9a0>)
+[gpub011:0/16] 2024-02-08 18:38:58,418 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-08 18:38:58,423 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 18:46:28,070 (trainer:762) INFO: 35epoch:train:5001-5100batch: iter_time=3.451, forward_time=0.330, loss_ctc=50.569, loss_att=51.223, acc=0.749, loss=51.027, backward_time=0.303, grad_norm=42.476, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.707e-04, train_time=4.927
+[gpub011:0/16] 2024-02-08 18:48:20,726 (trainer:762) INFO: 35epoch:train:5101-5200batch: iter_time=8.608e-05, forward_time=0.289, loss_ctc=45.866, loss_att=40.864, acc=0.774, loss=42.364, backward_time=0.296, grad_norm=41.105, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.707e-04, train_time=1.126
+[gpub011:0/16] 2024-02-08 18:50:18,959 (trainer:762) INFO: 35epoch:train:5201-5300batch: iter_time=8.365e-05, forward_time=0.304, loss_ctc=44.938, loss_att=48.647, acc=0.742, loss=47.535, backward_time=0.300, grad_norm=40.375, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.706e-04, train_time=1.182
+[gpub011:0/16] 2024-02-08 18:52:55,990 (trainer:762) INFO: 35epoch:train:5301-5400batch: iter_time=8.328e-05, forward_time=0.402, loss_ctc=43.532, loss_att=41.342, acc=0.762, loss=41.999, backward_time=0.317, grad_norm=40.071, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.706e-04, train_time=1.570
+[gpub011:0/16] 2024-02-08 18:54:56,362 (trainer:762) INFO: 35epoch:train:5401-5500batch: iter_time=8.131e-05, forward_time=0.292, loss_ctc=52.406, loss_att=54.709, acc=0.751, loss=54.018, backward_time=0.298, grad_norm=42.069, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.706e-04, train_time=1.203
+[gpub011:0/16] 2024-02-08 18:57:03,482 (trainer:762) INFO: 35epoch:train:5501-5600batch: iter_time=2.317e-04, forward_time=0.297, loss_ctc=44.690, loss_att=46.318, acc=0.758, loss=45.829, backward_time=0.298, grad_norm=42.482, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.706e-04, train_time=1.269
+[gpub011:0/16] 2024-02-08 18:59:36,343 (trainer:762) INFO: 35epoch:train:5601-5700batch: iter_time=8.516e-05, forward_time=0.414, loss_ctc=43.490, loss_att=39.381, acc=0.779, loss=40.613, backward_time=0.329, grad_norm=35.217, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.706e-04, train_time=1.530
+[gpub011:0/16] 2024-02-08 19:01:34,739 (trainer:762) INFO: 35epoch:train:5701-5800batch: iter_time=8.438e-05, forward_time=0.290, loss_ctc=45.381, loss_att=49.720, acc=0.739, loss=48.418, backward_time=0.296, grad_norm=43.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.706e-04, train_time=1.184
+[gpub011:0/16] 2024-02-08 19:03:54,885 (trainer:762) INFO: 35epoch:train:5801-5900batch: iter_time=9.150e-05, forward_time=0.389, loss_ctc=37.943, loss_att=40.112, acc=0.732, loss=39.461, backward_time=0.322, grad_norm=37.861, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.705e-04, train_time=1.401
+[gpub011:0/16] 2024-02-08 19:06:03,793 (trainer:762) INFO: 35epoch:train:5901-6000batch: iter_time=8.036e-05, forward_time=0.291, loss_ctc=45.848, loss_att=43.967, acc=0.758, loss=44.531, backward_time=0.296, grad_norm=39.358, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.705e-04, train_time=1.289
+[gpub011:0/16] 2024-02-08 19:08:01,907 (trainer:762) INFO: 35epoch:train:6001-6100batch: iter_time=8.391e-05, forward_time=0.288, loss_ctc=40.421, loss_att=39.351, acc=0.749, loss=39.672, backward_time=0.294, grad_norm=42.696, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.705e-04, train_time=1.181
+[gpub011:0/16] 2024-02-08 19:10:35,013 (trainer:762) INFO: 35epoch:train:6101-6200batch: iter_time=8.144e-05, forward_time=0.373, loss_ctc=46.913, loss_att=45.078, acc=0.759, loss=45.628, backward_time=0.350, grad_norm=42.332, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.705e-04, train_time=1.531
+[gpub011:0/16] 2024-02-08 19:11:50,464 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-08 19:12:09,773 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 19:12:13,297 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f982dfbff70>)
+[gpub011:0/16] 2024-02-08 19:12:13,297 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-08 19:12:13,326 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 19:18:14,845 (trainer:762) INFO: 35epoch:train:6201-6300batch: iter_time=3.329, forward_time=0.294, loss_ctc=48.464, loss_att=51.029, acc=0.759, loss=50.259, backward_time=0.299, grad_norm=40.231, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.705e-04, train_time=4.598
+[gpub011:0/16] 2024-02-08 19:20:28,023 (trainer:762) INFO: 35epoch:train:6301-6400batch: iter_time=7.953e-05, forward_time=0.325, loss_ctc=48.459, loss_att=49.437, acc=0.758, loss=49.144, backward_time=0.297, grad_norm=42.080, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.705e-04, train_time=1.332
+[gpub011:0/16] 2024-02-08 19:22:36,518 (trainer:762) INFO: 35epoch:train:6401-6500batch: iter_time=8.840e-05, forward_time=0.387, loss_ctc=45.396, loss_att=37.598, acc=0.770, loss=39.937, backward_time=0.311, grad_norm=39.310, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.704e-04, train_time=1.284
+[gpub011:0/16] 2024-02-08 19:24:48,639 (trainer:762) INFO: 35epoch:train:6501-6600batch: iter_time=9.132e-05, forward_time=0.315, loss_ctc=44.760, loss_att=48.182, acc=0.748, loss=47.156, backward_time=0.307, grad_norm=40.693, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.704e-04, train_time=1.321
+[gpub011:0/16] 2024-02-08 19:27:20,669 (trainer:762) INFO: 35epoch:train:6601-6700batch: iter_time=3.974e-04, forward_time=0.382, loss_ctc=47.662, loss_att=50.106, acc=0.742, loss=49.373, backward_time=0.336, grad_norm=43.269, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.704e-04, train_time=1.520
+[gpub011:0/16] 2024-02-08 19:29:12,393 (trainer:762) INFO: 35epoch:train:6701-6800batch: iter_time=9.214e-05, forward_time=0.292, loss_ctc=48.333, loss_att=49.782, acc=0.771, loss=49.347, backward_time=0.299, grad_norm=39.791, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.704e-04, train_time=1.117
+[gpub011:0/16] 2024-02-08 19:31:04,653 (trainer:762) INFO: 35epoch:train:6801-6900batch: iter_time=9.251e-05, forward_time=0.290, loss_ctc=42.841, loss_att=42.201, acc=0.769, loss=42.393, backward_time=0.297, grad_norm=39.374, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.704e-04, train_time=1.122
+[gpub011:0/16] 2024-02-08 19:34:01,732 (trainer:762) INFO: 35epoch:train:6901-7000batch: iter_time=9.549e-05, forward_time=0.363, loss_ctc=45.481, loss_att=40.899, acc=0.762, loss=42.274, backward_time=0.384, grad_norm=43.082, clip=100.000, loss_scale=5.244e+33, optim_step_time=0.107, optim0_lr0=1.704e-04, train_time=1.770
+[gpub011:0/16] 2024-02-08 19:35:53,883 (trainer:762) INFO: 35epoch:train:7001-7100batch: iter_time=8.743e-05, forward_time=0.291, loss_ctc=42.317, loss_att=49.556, acc=0.727, loss=47.384, backward_time=0.297, grad_norm=38.802, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.703e-04, train_time=1.122
+[gpub011:0/16] 2024-02-08 19:37:45,968 (trainer:762) INFO: 35epoch:train:7101-7200batch: iter_time=9.046e-05, forward_time=0.288, loss_ctc=40.176, loss_att=40.876, acc=0.742, loss=40.666, backward_time=0.295, grad_norm=37.388, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.703e-04, train_time=1.121
+[gpub011:0/16] 2024-02-08 19:40:53,828 (trainer:762) INFO: 35epoch:train:7201-7300batch: iter_time=8.949e-05, forward_time=0.403, loss_ctc=41.628, loss_att=39.644, acc=0.756, loss=40.239, backward_time=0.324, grad_norm=38.920, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.703e-04, train_time=1.878
+[gpub011:0/16] 2024-02-08 19:42:46,853 (trainer:762) INFO: 35epoch:train:7301-7400batch: iter_time=8.687e-05, forward_time=0.289, loss_ctc=42.846, loss_att=39.231, acc=0.774, loss=40.315, backward_time=0.296, grad_norm=38.829, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.703e-04, train_time=1.130
+[gpub011:0/16] 2024-02-08 19:44:41,931 (trainer:762) INFO: 35epoch:train:7401-7500batch: iter_time=8.511e-05, forward_time=0.293, loss_ctc=51.566, loss_att=54.427, acc=0.744, loss=53.569, backward_time=0.299, grad_norm=47.217, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.703e-04, train_time=1.151
+[gpub011:0/16] 2024-02-08 19:45:03,709 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-08 19:45:23,393 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 19:45:27,208 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f982d3c4640>)
+[gpub011:0/16] 2024-02-08 19:45:27,208 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-08 19:45:27,748 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 19:51:52,611 (trainer:762) INFO: 35epoch:train:7501-7600batch: iter_time=3.107, forward_time=0.341, loss_ctc=49.761, loss_att=50.355, acc=0.752, loss=50.177, backward_time=0.307, grad_norm=42.413, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.703e-04, train_time=4.307
+[gpub011:0/16] 2024-02-08 19:53:49,125 (trainer:762) INFO: 35epoch:train:7601-7700batch: iter_time=8.426e-05, forward_time=0.323, loss_ctc=45.900, loss_att=40.864, acc=0.776, loss=42.375, backward_time=0.336, grad_norm=39.454, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.702e-04, train_time=1.165
+[gpub011:0/16] 2024-02-08 19:56:15,008 (trainer:762) INFO: 35epoch:train:7701-7800batch: iter_time=8.606e-05, forward_time=0.337, loss_ctc=45.050, loss_att=48.289, acc=0.746, loss=47.317, backward_time=0.330, grad_norm=40.558, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.702e-04, train_time=1.458
+[gpub011:0/16] 2024-02-08 19:58:28,807 (trainer:762) INFO: 35epoch:train:7801-7900batch: iter_time=1.849e-04, forward_time=0.312, loss_ctc=43.123, loss_att=39.631, acc=0.769, loss=40.678, backward_time=0.333, grad_norm=37.318, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.702e-04, train_time=1.338
+[gpub011:0/16] 2024-02-08 20:00:39,251 (trainer:762) INFO: 35epoch:train:7901-8000batch: iter_time=8.809e-05, forward_time=0.367, loss_ctc=52.111, loss_att=54.588, acc=0.753, loss=53.845, backward_time=0.317, grad_norm=43.291, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.702e-04, train_time=1.304
+[gpub011:0/16] 2024-02-08 20:03:10,595 (trainer:762) INFO: 35epoch:train:8001-8100batch: iter_time=8.537e-05, forward_time=0.328, loss_ctc=44.870, loss_att=46.034, acc=0.760, loss=45.685, backward_time=0.374, grad_norm=44.282, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.702e-04, train_time=1.514
+[gpub011:0/16] 2024-02-08 20:03:44,239 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 20:05:29,812 (trainer:762) INFO: 35epoch:train:8101-8200batch: iter_time=2.051e-04, forward_time=0.375, loss_ctc=43.310, loss_att=39.193, acc=0.779, loss=40.428, backward_time=0.306, grad_norm=37.288, clip=100.000, loss_scale=6.713e+33, optim_step_time=0.095, optim0_lr0=1.702e-04, train_time=1.392
+[gpub011:0/16] 2024-02-08 20:08:02,413 (trainer:762) INFO: 35epoch:train:8201-8300batch: iter_time=0.001, forward_time=0.348, loss_ctc=45.059, loss_att=48.986, acc=0.744, loss=47.808, backward_time=0.306, grad_norm=42.348, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.124, optim0_lr0=1.701e-04, train_time=1.525
+[gpub011:0/16] 2024-02-08 20:10:03,073 (trainer:762) INFO: 35epoch:train:8301-8400batch: iter_time=7.997e-05, forward_time=0.319, loss_ctc=37.702, loss_att=39.469, acc=0.734, loss=38.939, backward_time=0.338, grad_norm=36.804, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.701e-04, train_time=1.207
+[gpub011:0/16] 2024-02-08 20:12:16,342 (trainer:762) INFO: 35epoch:train:8401-8500batch: iter_time=2.990e-04, forward_time=0.336, loss_ctc=45.461, loss_att=43.895, acc=0.760, loss=44.365, backward_time=0.318, grad_norm=38.243, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.701e-04, train_time=1.333
+[gpub011:0/16] 2024-02-08 20:14:41,798 (trainer:762) INFO: 35epoch:train:8501-8600batch: iter_time=8.409e-05, forward_time=0.287, loss_ctc=39.820, loss_att=38.713, acc=0.754, loss=39.045, backward_time=0.292, grad_norm=43.365, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.701e-04, train_time=1.453
+[gpub011:0/16] 2024-02-08 20:16:41,978 (trainer:762) INFO: 35epoch:train:8601-8700batch: iter_time=2.369e-04, forward_time=0.318, loss_ctc=46.927, loss_att=44.429, acc=0.762, loss=45.178, backward_time=0.330, grad_norm=42.378, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.701e-04, train_time=1.202
+[gpub011:0/16] 2024-02-08 20:18:17,327 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-08 20:18:36,562 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 20:18:40,842 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f982d3c75b0>)
+[gpub011:0/16] 2024-02-08 20:18:40,842 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-08 20:18:40,845 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 20:24:37,310 (trainer:762) INFO: 35epoch:train:8701-8800batch: iter_time=3.280, forward_time=0.338, loss_ctc=47.680, loss_att=53.307, acc=0.763, loss=51.619, backward_time=0.313, grad_norm=42.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.701e-04, train_time=4.754
+[gpub011:0/16] 2024-02-08 20:26:39,425 (trainer:762) INFO: 35epoch:train:8801-8900batch: iter_time=7.482e-05, forward_time=0.344, loss_ctc=48.656, loss_att=49.518, acc=0.768, loss=49.259, backward_time=0.312, grad_norm=41.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.700e-04, train_time=1.221
+[gpub011:0/16] 2024-02-08 20:29:16,345 (trainer:762) INFO: 35epoch:train:8901-9000batch: iter_time=8.769e-05, forward_time=0.357, loss_ctc=44.505, loss_att=36.520, acc=0.777, loss=38.915, backward_time=0.305, grad_norm=39.443, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.700e-04, train_time=1.569
+[gpub011:0/16] 2024-02-08 20:31:34,613 (trainer:762) INFO: 35epoch:train:9001-9100batch: iter_time=8.683e-05, forward_time=0.291, loss_ctc=44.470, loss_att=50.375, acc=0.759, loss=48.603, backward_time=0.299, grad_norm=38.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.700e-04, train_time=1.383
+[gpub011:0/16] 2024-02-08 20:33:50,613 (trainer:762) INFO: 35epoch:train:9101-9200batch: iter_time=8.822e-05, forward_time=0.342, loss_ctc=47.384, loss_att=51.545, acc=0.751, loss=50.297, backward_time=0.313, grad_norm=45.408, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.700e-04, train_time=1.360
+[gpub011:0/16] 2024-02-08 20:35:57,317 (trainer:762) INFO: 35epoch:train:9201-9300batch: iter_time=9.331e-05, forward_time=0.342, loss_ctc=48.142, loss_att=50.090, acc=0.780, loss=49.506, backward_time=0.322, grad_norm=39.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.700e-04, train_time=1.267
+[gpub011:0/16] 2024-02-08 20:38:22,997 (trainer:762) INFO: 35epoch:train:9301-9400batch: iter_time=8.361e-05, forward_time=0.290, loss_ctc=42.351, loss_att=42.275, acc=0.779, loss=42.298, backward_time=0.295, grad_norm=37.306, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.700e-04, train_time=1.457
+[gpub011:0/16] 2024-02-08 20:40:51,183 (trainer:762) INFO: 35epoch:train:9401-9500batch: iter_time=1.713e-04, forward_time=0.379, loss_ctc=45.294, loss_att=42.202, acc=0.768, loss=43.130, backward_time=0.335, grad_norm=42.347, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.700e-04, train_time=1.482
+[gpub011:0/16] 2024-02-08 20:43:01,096 (trainer:762) INFO: 35epoch:train:9501-9600batch: iter_time=8.735e-05, forward_time=0.292, loss_ctc=42.472, loss_att=51.098, acc=0.735, loss=48.510, backward_time=0.297, grad_norm=38.998, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.699e-04, train_time=1.299
+[gpub011:0/16] 2024-02-08 20:44:57,302 (trainer:762) INFO: 35epoch:train:9601-9700batch: iter_time=9.006e-05, forward_time=0.288, loss_ctc=39.746, loss_att=41.994, acc=0.755, loss=41.320, backward_time=0.295, grad_norm=36.572, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.699e-04, train_time=1.162
+[gpub011:0/16] 2024-02-08 20:47:42,163 (trainer:762) INFO: 35epoch:train:9701-9800batch: iter_time=0.010, forward_time=0.393, loss_ctc=41.527, loss_att=38.538, acc=0.768, loss=39.434, backward_time=0.314, grad_norm=37.824, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.699e-04, train_time=1.648
+[gpub011:0/16] 2024-02-08 20:49:52,743 (trainer:762) INFO: 35epoch:train:9801-9900batch: iter_time=8.658e-05, forward_time=0.291, loss_ctc=42.643, loss_att=40.686, acc=0.778, loss=41.273, backward_time=0.296, grad_norm=37.929, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.699e-04, train_time=1.306
+[gpub011:0/16] 2024-02-08 20:51:59,050 (trainer:762) INFO: 35epoch:train:9901-10000batch: iter_time=8.262e-05, forward_time=0.296, loss_ctc=50.652, loss_att=54.438, acc=0.750, loss=53.302, backward_time=0.299, grad_norm=47.787, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.699e-04, train_time=1.263
+[gpub011:0/16] 2024-02-08 20:52:19,078 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-08 20:52:38,476 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 20:52:42,006 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a8f8fb610>)
+[gpub011:0/16] 2024-02-08 20:52:42,006 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-08 20:52:42,120 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 20:58:59,059 (trainer:762) INFO: 35epoch:train:10001-10100batch: iter_time=2.983, forward_time=0.379, loss_ctc=49.569, loss_att=51.501, acc=0.750, loss=50.921, backward_time=0.313, grad_norm=42.977, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.699e-04, train_time=4.199
+[gpub011:0/16] 2024-02-08 21:01:11,683 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 21:01:33,144 (trainer:762) INFO: 35epoch:train:10101-10200batch: iter_time=7.809e-05, forward_time=0.290, loss_ctc=45.616, loss_att=41.025, acc=0.775, loss=42.402, backward_time=0.295, grad_norm=39.773, clip=100.000, loss_scale=8.549e+33, optim_step_time=0.092, optim0_lr0=1.698e-04, train_time=1.541
+[gpub011:0/16] 2024-02-08 21:03:44,595 (trainer:762) INFO: 35epoch:train:10201-10300batch: iter_time=9.795e-04, forward_time=0.407, loss_ctc=44.973, loss_att=48.413, acc=0.744, loss=47.381, backward_time=0.342, grad_norm=38.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.698e-04, train_time=1.314
+[gpub011:0/16] 2024-02-08 21:05:36,345 (trainer:762) INFO: 35epoch:train:10301-10400batch: iter_time=8.080e-05, forward_time=0.291, loss_ctc=43.072, loss_att=40.852, acc=0.764, loss=41.518, backward_time=0.297, grad_norm=37.487, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.698e-04, train_time=1.116
+[gpub011:0/16] 2024-02-08 21:08:09,914 (trainer:762) INFO: 35epoch:train:10401-10500batch: iter_time=8.249e-05, forward_time=0.292, loss_ctc=51.402, loss_att=54.047, acc=0.754, loss=53.253, backward_time=0.297, grad_norm=43.113, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.698e-04, train_time=1.537
+[gpub011:0/16] 2024-02-08 21:10:14,198 (trainer:762) INFO: 35epoch:train:10501-10600batch: iter_time=8.035e-05, forward_time=0.371, loss_ctc=44.532, loss_att=46.128, acc=0.760, loss=45.650, backward_time=0.343, grad_norm=41.565, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.698e-04, train_time=1.243
+[gpub011:0/16] 2024-02-08 21:12:06,424 (trainer:762) INFO: 35epoch:train:10601-10700batch: iter_time=8.089e-05, forward_time=0.291, loss_ctc=43.144, loss_att=39.375, acc=0.779, loss=40.506, backward_time=0.297, grad_norm=36.696, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.698e-04, train_time=1.121
+[gpub011:0/16] 2024-02-08 21:14:55,310 (trainer:762) INFO: 35epoch:train:10701-10800batch: iter_time=3.108e-04, forward_time=0.350, loss_ctc=45.068, loss_att=49.838, acc=0.741, loss=48.407, backward_time=0.362, grad_norm=45.312, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.697e-04, train_time=1.689
+[gpub011:0/16] 2024-02-08 21:16:44,179 (trainer:762) INFO: 35epoch:train:10801-10900batch: iter_time=8.138e-05, forward_time=0.286, loss_ctc=37.492, loss_att=39.801, acc=0.735, loss=39.108, backward_time=0.292, grad_norm=37.464, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.697e-04, train_time=1.089
+[gpub011:0/16] 2024-02-08 21:18:36,060 (trainer:762) INFO: 35epoch:train:10901-11000batch: iter_time=8.235e-05, forward_time=0.292, loss_ctc=45.287, loss_att=43.543, acc=0.761, loss=44.066, backward_time=0.298, grad_norm=38.633, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.697e-04, train_time=1.118
+[gpub011:0/16] 2024-02-08 21:21:30,418 (trainer:762) INFO: 35epoch:train:11001-11100batch: iter_time=3.702e-04, forward_time=0.475, loss_ctc=39.672, loss_att=38.998, acc=0.752, loss=39.200, backward_time=0.314, grad_norm=43.315, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.697e-04, train_time=1.744
+[gpub011:0/16] 2024-02-08 21:23:24,395 (trainer:762) INFO: 35epoch:train:11101-11200batch: iter_time=8.145e-05, forward_time=0.289, loss_ctc=46.528, loss_att=44.699, acc=0.761, loss=45.248, backward_time=0.296, grad_norm=42.214, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.697e-04, train_time=1.140
+[gpub011:0/16] 2024-02-08 21:24:40,482 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-08 21:24:59,557 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 21:25:03,073 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97d6cdba90>)
+[gpub011:0/16] 2024-02-08 21:25:03,073 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-08 21:25:03,076 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 21:31:05,288 (trainer:762) INFO: 35epoch:train:11201-11300batch: iter_time=3.367, forward_time=0.391, loss_ctc=47.328, loss_att=52.009, acc=0.766, loss=50.605, backward_time=0.320, grad_norm=43.259, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.697e-04, train_time=4.608
+[gpub011:0/16] 2024-02-08 21:33:23,575 (trainer:762) INFO: 35epoch:train:11301-11400batch: iter_time=8.186e-05, forward_time=0.291, loss_ctc=47.756, loss_att=49.340, acc=0.766, loss=48.865, backward_time=0.298, grad_norm=40.045, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.696e-04, train_time=1.383
+[gpub011:0/16] 2024-02-08 21:35:31,876 (trainer:762) INFO: 35epoch:train:11401-11500batch: iter_time=7.973e-05, forward_time=0.419, loss_ctc=44.977, loss_att=36.929, acc=0.775, loss=39.343, backward_time=0.314, grad_norm=39.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.696e-04, train_time=1.282
+[gpub011:0/16] 2024-02-08 21:37:09,795 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-08 21:37:42,660 (trainer:762) INFO: 35epoch:train:11501-11600batch: iter_time=8.252e-05, forward_time=0.293, loss_ctc=44.107, loss_att=50.076, acc=0.760, loss=48.286, backward_time=0.299, grad_norm=39.493, clip=100.000, loss_scale=4.484e+33, optim_step_time=0.093, optim0_lr0=1.696e-04, train_time=1.308
+[gpub011:0/16] 2024-02-08 21:40:07,349 (trainer:762) INFO: 35epoch:train:11601-11700batch: iter_time=8.242e-05, forward_time=0.387, loss_ctc=46.879, loss_att=49.905, acc=0.755, loss=48.997, backward_time=0.331, grad_norm=42.329, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.696e-04, train_time=1.447
+[gpub011:0/16] 2024-02-08 21:42:42,046 (trainer:762) INFO: 35epoch:train:11701-11800batch: iter_time=8.491e-05, forward_time=0.418, loss_ctc=47.768, loss_att=49.376, acc=0.782, loss=48.893, backward_time=0.342, grad_norm=40.306, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.696e-04, train_time=1.547
+[gpub011:0/16] 2024-02-08 21:44:43,863 (trainer:762) INFO: 35epoch:train:11801-11900batch: iter_time=8.267e-05, forward_time=0.290, loss_ctc=41.948, loss_att=42.357, acc=0.778, loss=42.234, backward_time=0.298, grad_norm=39.718, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.696e-04, train_time=1.217
+[gpub011:0/16] 2024-02-08 21:47:13,000 (trainer:762) INFO: 35epoch:train:11901-12000batch: iter_time=7.876e-05, forward_time=0.370, loss_ctc=44.741, loss_att=42.366, acc=0.767, loss=43.079, backward_time=0.364, grad_norm=44.210, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.695e-04, train_time=1.492
+[gpub011:0/16] 2024-02-08 21:49:26,143 (trainer:762) INFO: 35epoch:train:12001-12100batch: iter_time=8.484e-05, forward_time=0.315, loss_ctc=41.803, loss_att=50.899, acc=0.733, loss=48.170, backward_time=0.296, grad_norm=39.359, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.695e-04, train_time=1.331
+[gpub011:0/16] 2024-02-08 21:51:48,444 (trainer:762) INFO: 35epoch:train:12101-12200batch: iter_time=4.157e-04, forward_time=0.390, loss_ctc=39.469, loss_att=41.190, acc=0.758, loss=40.673, backward_time=0.310, grad_norm=37.683, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.105, optim0_lr0=1.695e-04, train_time=1.423
+[gpub011:0/16] 2024-02-08 21:54:01,179 (trainer:762) INFO: 35epoch:train:12201-12300batch: iter_time=8.327e-05, forward_time=0.313, loss_ctc=41.039, loss_att=38.570, acc=0.772, loss=39.310, backward_time=0.299, grad_norm=39.219, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.695e-04, train_time=1.328
+[gpub011:0/16] 2024-02-08 21:56:00,264 (trainer:762) INFO: 35epoch:train:12301-12400batch: iter_time=8.400e-05, forward_time=0.291, loss_ctc=42.720, loss_att=41.181, acc=0.777, loss=41.643, backward_time=0.298, grad_norm=39.490, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.695e-04, train_time=1.190
+[gpub011:0/16] 2024-02-08 21:58:22,275 (trainer:762) INFO: 35epoch:train:12401-12500batch: iter_time=8.745e-05, forward_time=0.293, loss_ctc=50.480, loss_att=54.874, acc=0.747, loss=53.556, backward_time=0.299, grad_norm=45.189, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.695e-04, train_time=1.420
+[gpub011:0/16] 2024-02-08 21:58:42,377 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-08 21:59:01,622 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 21:59:05,228 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98186ba560>)
+[gpub011:0/16] 2024-02-08 21:59:05,229 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-08 21:59:05,234 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 22:05:22,645 (trainer:762) INFO: 35epoch:train:12501-12600batch: iter_time=2.997, forward_time=0.376, loss_ctc=49.779, loss_att=50.558, acc=0.764, loss=50.324, backward_time=0.315, grad_norm=40.296, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.694e-04, train_time=4.203
+[gpub011:0/16] 2024-02-08 22:07:34,873 (trainer:762) INFO: 35epoch:train:12601-12700batch: iter_time=8.326e-05, forward_time=0.289, loss_ctc=45.690, loss_att=40.432, acc=0.786, loss=42.009, backward_time=0.296, grad_norm=38.250, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.694e-04, train_time=1.322
+[gpub011:0/16] 2024-02-08 22:09:49,105 (trainer:762) INFO: 35epoch:train:12701-12800batch: iter_time=8.435e-05, forward_time=0.290, loss_ctc=44.429, loss_att=47.755, acc=0.756, loss=46.757, backward_time=0.296, grad_norm=39.915, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.694e-04, train_time=1.343
+[gpub011:0/16] 2024-02-08 22:11:56,596 (trainer:762) INFO: 35epoch:train:12801-12900batch: iter_time=3.370e-04, forward_time=0.398, loss_ctc=42.789, loss_att=42.137, acc=0.773, loss=42.333, backward_time=0.320, grad_norm=38.203, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=1.694e-04, train_time=1.274
+[gpub011:0/16] 2024-02-08 22:14:03,801 (trainer:762) INFO: 35epoch:train:12901-13000batch: iter_time=8.662e-05, forward_time=0.294, loss_ctc=51.502, loss_att=54.399, acc=0.758, loss=53.530, backward_time=0.301, grad_norm=44.360, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.694e-04, train_time=1.272
+[gpub011:0/16] 2024-02-08 22:16:27,743 (trainer:762) INFO: 35epoch:train:13001-13100batch: iter_time=9.292e-05, forward_time=0.290, loss_ctc=44.022, loss_att=45.925, acc=0.771, loss=45.354, backward_time=0.296, grad_norm=42.108, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.694e-04, train_time=1.439
+[gpub011:0/16] 2024-02-08 22:18:35,170 (trainer:762) INFO: 35epoch:train:13101-13200batch: iter_time=7.688e-04, forward_time=0.400, loss_ctc=43.482, loss_att=40.484, acc=0.788, loss=41.384, backward_time=0.331, grad_norm=36.618, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.693e-04, train_time=1.274
+[gpub011:0/16] 2024-02-08 22:20:48,146 (trainer:762) INFO: 35epoch:train:13201-13300batch: iter_time=8.557e-05, forward_time=0.316, loss_ctc=44.772, loss_att=48.718, acc=0.754, loss=47.534, backward_time=0.299, grad_norm=45.214, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.693e-04, train_time=1.329
+[gpub011:0/16] 2024-02-08 22:23:12,375 (trainer:762) INFO: 35epoch:train:13301-13400batch: iter_time=8.709e-05, forward_time=0.287, loss_ctc=37.393, loss_att=40.554, acc=0.743, loss=39.606, backward_time=0.291, grad_norm=36.669, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.693e-04, train_time=1.443
+[gpub011:0/16] 2024-02-08 22:25:01,457 (trainer:762) INFO: 35epoch:train:13401-13500batch: iter_time=8.873e-05, forward_time=0.292, loss_ctc=45.378, loss_att=45.411, acc=0.767, loss=45.401, backward_time=0.299, grad_norm=37.310, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.693e-04, train_time=1.090
+[gpub011:0/16] 2024-02-08 22:27:36,472 (trainer:762) INFO: 35epoch:train:13501-13600batch: iter_time=9.312e-05, forward_time=0.373, loss_ctc=39.617, loss_att=39.914, acc=0.760, loss=39.825, backward_time=0.340, grad_norm=41.268, clip=100.000, loss_scale=3.297e+33, optim_step_time=0.100, optim0_lr0=1.693e-04, train_time=1.550
+[gpub011:0/16] 2024-02-08 22:29:49,637 (trainer:762) INFO: 35epoch:train:13601-13700batch: iter_time=8.684e-05, forward_time=0.290, loss_ctc=46.498, loss_att=44.716, acc=0.768, loss=45.250, backward_time=0.296, grad_norm=41.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.693e-04, train_time=1.331
+[gpub011:0/16] 2024-02-08 22:31:04,564 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-08 22:31:24,126 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 22:31:27,651 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97e72efdc0>)
+[gpub011:0/16] 2024-02-08 22:31:27,651 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-08 22:31:27,670 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 22:38:02,019 (trainer:762) INFO: 35epoch:train:13701-13800batch: iter_time=3.447, forward_time=0.415, loss_ctc=47.376, loss_att=52.585, acc=0.757, loss=51.022, backward_time=0.321, grad_norm=40.850, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.693e-04, train_time=4.924
+[gpub011:0/16] 2024-02-08 22:40:26,706 (trainer:762) INFO: 35epoch:train:13801-13900batch: iter_time=8.454e-05, forward_time=0.290, loss_ctc=48.164, loss_att=50.192, acc=0.755, loss=49.584, backward_time=0.296, grad_norm=43.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.692e-04, train_time=1.446
+[gpub011:0/16] 2024-02-08 22:42:35,164 (trainer:762) INFO: 35epoch:train:13901-14000batch: iter_time=2.080e-04, forward_time=0.384, loss_ctc=44.964, loss_att=36.956, acc=0.774, loss=39.358, backward_time=0.330, grad_norm=39.673, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.692e-04, train_time=1.285
+[gpub011:0/16] 2024-02-08 22:45:16,456 (trainer:762) INFO: 35epoch:train:14001-14100batch: iter_time=8.961e-05, forward_time=0.337, loss_ctc=44.514, loss_att=48.795, acc=0.750, loss=47.511, backward_time=0.369, grad_norm=39.825, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.692e-04, train_time=1.613
+[gpub011:0/16] 2024-02-08 22:47:29,304 (trainer:762) INFO: 35epoch:train:14101-14200batch: iter_time=8.062e-05, forward_time=0.289, loss_ctc=47.058, loss_att=50.348, acc=0.744, loss=49.361, backward_time=0.296, grad_norm=42.994, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.692e-04, train_time=1.328
+[gpub011:0/16] 2024-02-08 22:49:56,249 (trainer:762) INFO: 35epoch:train:14201-14300batch: iter_time=2.434e-04, forward_time=0.359, loss_ctc=47.510, loss_att=49.310, acc=0.773, loss=48.770, backward_time=0.385, grad_norm=39.118, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.692e-04, train_time=1.469
+[gpub011:0/16] 2024-02-08 22:52:19,182 (trainer:762) INFO: 35epoch:train:14301-14400batch: iter_time=8.250e-05, forward_time=0.291, loss_ctc=41.962, loss_att=41.535, acc=0.771, loss=41.663, backward_time=0.295, grad_norm=38.412, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.692e-04, train_time=1.430
+[gpub011:0/16] 2024-02-08 22:54:32,164 (trainer:762) INFO: 35epoch:train:14401-14500batch: iter_time=8.527e-05, forward_time=0.365, loss_ctc=44.822, loss_att=40.087, acc=0.765, loss=41.507, backward_time=0.415, grad_norm=41.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.691e-04, train_time=1.330
+[gpub011:0/16] 2024-02-08 22:56:54,728 (trainer:762) INFO: 35epoch:train:14501-14600batch: iter_time=8.515e-05, forward_time=0.289, loss_ctc=41.578, loss_att=48.627, acc=0.732, loss=46.512, backward_time=0.295, grad_norm=38.349, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.691e-04, train_time=1.425
+[gpub011:0/16] 2024-02-08 22:59:26,260 (trainer:762) INFO: 35epoch:train:14601-14700batch: iter_time=8.721e-05, forward_time=0.403, loss_ctc=39.454, loss_att=40.896, acc=0.742, loss=40.464, backward_time=0.343, grad_norm=36.429, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.691e-04, train_time=1.515
+[gpub011:0/16] 2024-02-08 23:01:16,628 (trainer:762) INFO: 35epoch:train:14701-14800batch: iter_time=7.930e-05, forward_time=0.289, loss_ctc=40.869, loss_att=38.929, acc=0.761, loss=39.511, backward_time=0.295, grad_norm=38.398, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.691e-04, train_time=1.104
+[gpub011:0/16] 2024-02-08 23:03:52,005 (trainer:762) INFO: 35epoch:train:14801-14900batch: iter_time=3.847e-04, forward_time=0.354, loss_ctc=42.701, loss_att=39.480, acc=0.775, loss=40.446, backward_time=0.339, grad_norm=39.202, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.691e-04, train_time=1.553
+[gpub011:0/16] 2024-02-08 23:06:08,993 (trainer:762) INFO: 35epoch:train:14901-15000batch: iter_time=8.532e-05, forward_time=0.292, loss_ctc=50.772, loss_att=54.006, acc=0.746, loss=53.036, backward_time=0.298, grad_norm=48.033, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.691e-04, train_time=1.370
+[gpub011:0/16] 2024-02-08 23:44:34,678 (trainer:361) INFO: 35epoch results: [train] iter_time=0.264, forward_time=0.331, loss_ctc=45.292, loss_att=45.681, acc=0.759, loss=45.564, backward_time=0.313, grad_norm=41.074, clip=100.000, loss_scale=5.891e+33, optim_step_time=0.096, optim0_lr0=1.703e-04, train_time=1.599, time=6 hours, 40 minutes and 18.97 seconds, total_count=555000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=35.486, cer_ctc=0.180, loss_att=39.286, acc=0.674, cer=0.360, wer=1.000, loss=38.146, time=38 minutes and 1.37 seconds, total_count=172827, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-08 23:44:44,073 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-08 23:44:44,229 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/30epoch.pth
+[gpub011:0/16] 2024-02-08 23:44:44,230 (trainer:290) INFO: 36/45epoch started. Estimated time to finish: 3 days, 2 hours and 37 minutes
+[gpub011:0/16] 2024-02-08 23:44:44,240 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-08 23:45:02,702 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-08 23:45:06,037 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fb7386fb0>)
+[gpub011:0/16] 2024-02-08 23:45:06,037 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-08 23:45:06,041 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-08 23:52:11,802 (trainer:762) INFO: 36epoch:train:1-100batch: iter_time=3.271, forward_time=0.380, loss_ctc=55.520, loss_att=53.362, acc=0.730, loss=54.010, backward_time=0.312, grad_norm=56.776, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.690e-04, train_time=4.475
+[gpub011:0/16] 2024-02-08 23:54:07,592 (trainer:762) INFO: 36epoch:train:101-200batch: iter_time=8.964e-05, forward_time=0.289, loss_ctc=50.599, loss_att=47.254, acc=0.740, loss=48.257, backward_time=0.296, grad_norm=46.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.690e-04, train_time=1.158
+[gpub011:0/16] 2024-02-08 23:56:49,513 (trainer:762) INFO: 36epoch:train:201-300batch: iter_time=4.791e-04, forward_time=0.377, loss_ctc=56.336, loss_att=42.323, acc=0.748, loss=46.527, backward_time=0.326, grad_norm=46.904, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.690e-04, train_time=1.619
+[gpub011:0/16] 2024-02-08 23:58:56,273 (trainer:762) INFO: 36epoch:train:301-400batch: iter_time=9.297e-05, forward_time=0.292, loss_ctc=48.413, loss_att=51.873, acc=0.740, loss=50.835, backward_time=0.298, grad_norm=40.777, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.690e-04, train_time=1.267
+[gpub011:0/16] 2024-02-09 00:01:14,702 (trainer:762) INFO: 36epoch:train:401-500batch: iter_time=9.008e-05, forward_time=0.290, loss_ctc=48.716, loss_att=50.913, acc=0.743, loss=50.254, backward_time=0.294, grad_norm=42.127, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.690e-04, train_time=1.384
+[gpub011:0/16] 2024-02-09 00:03:37,206 (trainer:762) INFO: 36epoch:train:501-600batch: iter_time=9.277e-05, forward_time=0.361, loss_ctc=48.919, loss_att=51.870, acc=0.754, loss=50.984, backward_time=0.381, grad_norm=43.525, clip=100.000, loss_scale=6.594e+33, optim_step_time=0.098, optim0_lr0=1.690e-04, train_time=1.423
+[gpub011:0/16] 2024-02-09 00:05:54,635 (trainer:762) INFO: 36epoch:train:601-700batch: iter_time=8.838e-05, forward_time=0.293, loss_ctc=56.065, loss_att=50.269, acc=0.753, loss=52.008, backward_time=0.296, grad_norm=47.324, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.689e-04, train_time=1.375
+[gpub011:0/16] 2024-02-09 00:08:05,254 (trainer:762) INFO: 36epoch:train:701-800batch: iter_time=8.938e-05, forward_time=0.288, loss_ctc=46.160, loss_att=42.762, acc=0.752, loss=43.781, backward_time=0.295, grad_norm=43.400, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.689e-04, train_time=1.306
+[gpub011:0/16] 2024-02-09 00:10:09,330 (trainer:762) INFO: 36epoch:train:801-900batch: iter_time=9.187e-05, forward_time=0.292, loss_ctc=54.466, loss_att=53.232, acc=0.752, loss=53.602, backward_time=0.298, grad_norm=45.339, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.689e-04, train_time=1.241
+[gpub011:0/16] 2024-02-09 00:12:41,213 (trainer:762) INFO: 36epoch:train:901-1000batch: iter_time=9.122e-05, forward_time=0.413, loss_ctc=52.733, loss_att=47.595, acc=0.742, loss=49.136, backward_time=0.336, grad_norm=50.043, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.689e-04, train_time=1.518
+[gpub011:0/16] 2024-02-09 00:14:50,094 (trainer:762) INFO: 36epoch:train:1001-1100batch: iter_time=8.550e-05, forward_time=0.289, loss_ctc=49.904, loss_att=46.845, acc=0.741, loss=47.762, backward_time=0.295, grad_norm=45.971, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.689e-04, train_time=1.289
+[gpub011:0/16] 2024-02-09 00:17:08,511 (trainer:762) INFO: 36epoch:train:1101-1200batch: iter_time=8.717e-05, forward_time=0.289, loss_ctc=46.462, loss_att=44.912, acc=0.741, loss=45.377, backward_time=0.295, grad_norm=43.725, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.689e-04, train_time=1.384
+[gpub011:0/16] 2024-02-09 00:18:31,406 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-09 00:18:50,673 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 00:18:54,148 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a8607d5a0>)
+[gpub011:0/16] 2024-02-09 00:18:54,148 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-09 00:18:54,238 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 00:24:58,558 (trainer:762) INFO: 36epoch:train:1201-1300batch: iter_time=3.478, forward_time=0.290, loss_ctc=48.101, loss_att=52.399, acc=0.719, loss=51.109, backward_time=0.296, grad_norm=47.966, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.688e-04, train_time=4.700
+[gpub011:0/16] 2024-02-09 00:27:24,774 (trainer:762) INFO: 36epoch:train:1301-1400batch: iter_time=8.900e-05, forward_time=0.427, loss_ctc=56.982, loss_att=51.472, acc=0.743, loss=53.125, backward_time=0.334, grad_norm=60.528, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.688e-04, train_time=1.461
+[gpub011:0/16] 2024-02-09 00:28:33,511 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 00:29:37,505 (trainer:762) INFO: 36epoch:train:1401-1500batch: iter_time=8.068e-05, forward_time=0.292, loss_ctc=56.077, loss_att=47.273, acc=0.743, loss=49.914, backward_time=0.295, grad_norm=43.703, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.093, optim0_lr0=1.688e-04, train_time=1.328
+[gpub011:0/16] 2024-02-09 00:31:40,012 (trainer:762) INFO: 36epoch:train:1501-1600batch: iter_time=8.082e-05, forward_time=0.289, loss_ctc=44.167, loss_att=40.885, acc=0.747, loss=41.870, backward_time=0.293, grad_norm=39.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.688e-04, train_time=1.225
+[gpub011:0/16] 2024-02-09 00:34:01,153 (trainer:762) INFO: 36epoch:train:1601-1700batch: iter_time=8.832e-05, forward_time=0.347, loss_ctc=51.756, loss_att=58.441, acc=0.734, loss=56.436, backward_time=0.298, grad_norm=44.187, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.688e-04, train_time=1.411
+[gpub011:0/16] 2024-02-09 00:36:15,401 (trainer:762) INFO: 36epoch:train:1701-1800batch: iter_time=9.066e-05, forward_time=0.387, loss_ctc=46.158, loss_att=47.449, acc=0.766, loss=47.061, backward_time=0.326, grad_norm=42.165, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.688e-04, train_time=1.342
+[gpub011:0/16] 2024-02-09 00:38:38,223 (trainer:762) INFO: 36epoch:train:1801-1900batch: iter_time=8.663e-05, forward_time=0.291, loss_ctc=53.608, loss_att=53.314, acc=0.743, loss=53.402, backward_time=0.295, grad_norm=49.441, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.688e-04, train_time=1.429
+[gpub011:0/16] 2024-02-09 00:40:49,260 (trainer:762) INFO: 36epoch:train:1901-2000batch: iter_time=8.303e-05, forward_time=0.290, loss_ctc=46.562, loss_att=42.687, acc=0.760, loss=43.850, backward_time=0.296, grad_norm=40.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.687e-04, train_time=1.310
+[gpub011:0/16] 2024-02-09 00:43:05,340 (trainer:762) INFO: 36epoch:train:2001-2100batch: iter_time=8.693e-05, forward_time=0.441, loss_ctc=48.471, loss_att=48.608, acc=0.750, loss=48.567, backward_time=0.337, grad_norm=42.602, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.687e-04, train_time=1.361
+[gpub011:0/16] 2024-02-09 00:45:17,960 (trainer:762) INFO: 36epoch:train:2101-2200batch: iter_time=8.677e-05, forward_time=0.298, loss_ctc=54.855, loss_att=50.429, acc=0.755, loss=51.757, backward_time=0.297, grad_norm=46.671, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.687e-04, train_time=1.325
+[gpub011:0/16] 2024-02-09 00:47:34,465 (trainer:762) INFO: 36epoch:train:2201-2300batch: iter_time=8.432e-05, forward_time=0.288, loss_ctc=48.152, loss_att=46.316, acc=0.748, loss=46.867, backward_time=0.293, grad_norm=49.159, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.687e-04, train_time=1.366
+[gpub011:0/16] 2024-02-09 00:49:46,284 (trainer:762) INFO: 36epoch:train:2301-2400batch: iter_time=8.544e-05, forward_time=0.304, loss_ctc=46.856, loss_att=43.854, acc=0.736, loss=44.755, backward_time=0.294, grad_norm=39.808, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.687e-04, train_time=1.318
+[gpub011:0/16] 2024-02-09 00:51:53,169 (trainer:762) INFO: 36epoch:train:2401-2500batch: iter_time=8.054e-05, forward_time=0.367, loss_ctc=49.134, loss_att=51.075, acc=0.734, loss=50.493, backward_time=0.371, grad_norm=45.702, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.687e-04, train_time=1.269
+[gpub011:0/16] 2024-02-09 00:52:13,215 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-09 00:52:32,084 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 00:52:35,972 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a56e9e080>)
+[gpub011:0/16] 2024-02-09 00:52:35,972 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-09 00:52:35,976 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 00:59:00,011 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 00:59:27,008 (trainer:762) INFO: 36epoch:train:2501-2600batch: iter_time=3.202, forward_time=0.291, loss_ctc=54.506, loss_att=55.920, acc=0.737, loss=55.496, backward_time=0.297, grad_norm=57.865, clip=100.000, loss_scale=4.694e+33, optim_step_time=0.093, optim0_lr0=1.686e-04, train_time=4.538
+[gpub011:0/16] 2024-02-09 01:01:20,099 (trainer:762) INFO: 36epoch:train:2601-2700batch: iter_time=8.055e-05, forward_time=0.290, loss_ctc=49.114, loss_att=47.359, acc=0.758, loss=47.885, backward_time=0.298, grad_norm=43.253, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.686e-04, train_time=1.131
+[gpub011:0/16] 2024-02-09 01:03:27,042 (trainer:762) INFO: 36epoch:train:2701-2800batch: iter_time=8.253e-05, forward_time=0.293, loss_ctc=52.924, loss_att=43.449, acc=0.757, loss=46.291, backward_time=0.306, grad_norm=41.616, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.686e-04, train_time=1.269
+[gpub011:0/16] 2024-02-09 01:05:57,483 (trainer:762) INFO: 36epoch:train:2801-2900batch: iter_time=8.604e-05, forward_time=0.393, loss_ctc=47.349, loss_att=51.400, acc=0.754, loss=50.184, backward_time=0.329, grad_norm=42.042, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.686e-04, train_time=1.504
+[gpub011:0/16] 2024-02-09 01:07:53,434 (trainer:762) INFO: 36epoch:train:2901-3000batch: iter_time=8.326e-05, forward_time=0.293, loss_ctc=47.600, loss_att=50.837, acc=0.755, loss=49.866, backward_time=0.300, grad_norm=40.037, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.686e-04, train_time=1.159
+[gpub011:0/16] 2024-02-09 01:10:06,429 (trainer:762) INFO: 36epoch:train:3001-3100batch: iter_time=8.609e-05, forward_time=0.291, loss_ctc=48.052, loss_att=51.569, acc=0.759, loss=50.514, backward_time=0.297, grad_norm=41.270, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.686e-04, train_time=1.330
+[gpub011:0/16] 2024-02-09 01:12:20,306 (trainer:762) INFO: 36epoch:train:3101-3200batch: iter_time=8.816e-05, forward_time=0.323, loss_ctc=52.594, loss_att=49.027, acc=0.763, loss=50.097, backward_time=0.299, grad_norm=43.330, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.685e-04, train_time=1.338
+[gpub011:0/16] 2024-02-09 01:14:46,256 (trainer:762) INFO: 36epoch:train:3201-3300batch: iter_time=8.569e-05, forward_time=0.390, loss_ctc=44.558, loss_att=42.779, acc=0.761, loss=43.313, backward_time=0.321, grad_norm=40.512, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.685e-04, train_time=1.460
+[gpub011:0/16] 2024-02-09 01:16:44,830 (trainer:762) INFO: 36epoch:train:3301-3400batch: iter_time=8.315e-05, forward_time=0.292, loss_ctc=52.906, loss_att=52.481, acc=0.759, loss=52.609, backward_time=0.299, grad_norm=45.585, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.685e-04, train_time=1.185
+[gpub011:0/16] 2024-02-09 01:19:05,916 (trainer:762) INFO: 36epoch:train:3401-3500batch: iter_time=8.533e-05, forward_time=0.291, loss_ctc=51.175, loss_att=47.719, acc=0.755, loss=48.756, backward_time=0.298, grad_norm=48.129, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.685e-04, train_time=1.411
+[gpub011:0/16] 2024-02-09 01:21:34,588 (trainer:762) INFO: 36epoch:train:3501-3600batch: iter_time=2.603e-04, forward_time=0.405, loss_ctc=49.019, loss_att=46.140, acc=0.755, loss=47.003, backward_time=0.321, grad_norm=42.607, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.685e-04, train_time=1.486
+[gpub011:0/16] 2024-02-09 01:23:32,582 (trainer:762) INFO: 36epoch:train:3601-3700batch: iter_time=8.298e-05, forward_time=0.311, loss_ctc=44.895, loss_att=43.068, acc=0.765, loss=43.616, backward_time=0.298, grad_norm=43.369, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.685e-04, train_time=1.180
+[gpub011:0/16] 2024-02-09 01:24:47,147 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-09 01:25:06,361 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 01:25:09,923 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a8fb4b640>)
+[gpub011:0/16] 2024-02-09 01:25:09,923 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-09 01:25:09,929 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 01:31:06,653 (trainer:762) INFO: 36epoch:train:3701-3800batch: iter_time=3.315, forward_time=0.291, loss_ctc=47.427, loss_att=52.648, acc=0.730, loss=51.082, backward_time=0.298, grad_norm=46.952, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.685e-04, train_time=4.540
+[gpub011:0/16] 2024-02-09 01:33:03,153 (trainer:762) INFO: 36epoch:train:3801-3900batch: iter_time=7.524e-05, forward_time=0.292, loss_ctc=57.404, loss_att=51.324, acc=0.745, loss=53.148, backward_time=0.299, grad_norm=59.083, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.684e-04, train_time=1.165
+[gpub011:0/16] 2024-02-09 01:35:33,556 (trainer:762) INFO: 36epoch:train:3901-4000batch: iter_time=8.638e-05, forward_time=0.380, loss_ctc=53.847, loss_att=46.714, acc=0.747, loss=48.854, backward_time=0.329, grad_norm=40.049, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=1.684e-04, train_time=1.504
+[gpub011:0/16] 2024-02-09 01:37:47,942 (trainer:762) INFO: 36epoch:train:4001-4100batch: iter_time=8.612e-05, forward_time=0.289, loss_ctc=44.265, loss_att=40.842, acc=0.749, loss=41.869, backward_time=0.295, grad_norm=38.564, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.684e-04, train_time=1.344
+[gpub011:0/16] 2024-02-09 01:39:42,290 (trainer:762) INFO: 36epoch:train:4101-4200batch: iter_time=8.378e-05, forward_time=0.294, loss_ctc=51.437, loss_att=58.772, acc=0.735, loss=56.571, backward_time=0.301, grad_norm=42.729, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.684e-04, train_time=1.143
+[gpub011:0/16] 2024-02-09 01:41:51,923 (trainer:762) INFO: 36epoch:train:4201-4300batch: iter_time=8.737e-05, forward_time=0.292, loss_ctc=45.574, loss_att=48.067, acc=0.767, loss=47.319, backward_time=0.298, grad_norm=39.616, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.684e-04, train_time=1.297
+[gpub011:0/16] 2024-02-09 01:44:20,870 (trainer:762) INFO: 36epoch:train:4301-4400batch: iter_time=8.161e-05, forward_time=0.382, loss_ctc=53.060, loss_att=52.078, acc=0.748, loss=52.372, backward_time=0.350, grad_norm=46.966, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.684e-04, train_time=1.489
+[gpub011:0/16] 2024-02-09 01:46:36,180 (trainer:762) INFO: 36epoch:train:4401-4500batch: iter_time=8.336e-05, forward_time=0.290, loss_ctc=45.752, loss_att=42.426, acc=0.761, loss=43.424, backward_time=0.298, grad_norm=40.938, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.683e-04, train_time=1.353
+[gpub011:0/16] 2024-02-09 01:48:33,280 (trainer:762) INFO: 36epoch:train:4501-4600batch: iter_time=8.351e-05, forward_time=0.292, loss_ctc=47.685, loss_att=47.724, acc=0.754, loss=47.712, backward_time=0.297, grad_norm=43.272, clip=100.000, loss_scale=3.089e+33, optim_step_time=0.093, optim0_lr0=1.683e-04, train_time=1.170
+[gpub011:0/16] 2024-02-09 01:51:09,693 (trainer:762) INFO: 36epoch:train:4601-4700batch: iter_time=8.292e-05, forward_time=0.293, loss_ctc=53.800, loss_att=50.058, acc=0.757, loss=51.180, backward_time=0.296, grad_norm=44.899, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.683e-04, train_time=1.565
+[gpub011:0/16] 2024-02-09 01:53:32,822 (trainer:762) INFO: 36epoch:train:4701-4800batch: iter_time=4.465e-04, forward_time=0.392, loss_ctc=48.096, loss_att=45.918, acc=0.750, loss=46.571, backward_time=0.344, grad_norm=49.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.683e-04, train_time=1.430
+[gpub011:0/16] 2024-02-09 01:55:26,157 (trainer:762) INFO: 36epoch:train:4801-4900batch: iter_time=7.846e-05, forward_time=0.290, loss_ctc=47.308, loss_att=44.309, acc=0.735, loss=45.209, backward_time=0.297, grad_norm=41.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.683e-04, train_time=1.134
+[gpub011:0/16] 2024-02-09 01:57:41,164 (trainer:762) INFO: 36epoch:train:4901-5000batch: iter_time=7.952e-05, forward_time=0.291, loss_ctc=48.801, loss_att=50.712, acc=0.736, loss=50.138, backward_time=0.298, grad_norm=44.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.683e-04, train_time=1.350
+[gpub011:0/16] 2024-02-09 01:58:01,216 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-09 01:58:20,271 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 01:58:23,884 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9ae7a8e2c0>)
+[gpub011:0/16] 2024-02-09 01:58:23,885 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-09 01:58:23,890 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 02:05:26,136 (trainer:762) INFO: 36epoch:train:5001-5100batch: iter_time=3.439, forward_time=0.297, loss_ctc=53.107, loss_att=54.339, acc=0.742, loss=53.970, backward_time=0.298, grad_norm=52.821, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.682e-04, train_time=4.649
+[gpub011:0/16] 2024-02-09 02:08:02,507 (trainer:762) INFO: 36epoch:train:5101-5200batch: iter_time=8.593e-05, forward_time=0.421, loss_ctc=48.974, loss_att=46.831, acc=0.759, loss=47.474, backward_time=0.329, grad_norm=44.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.682e-04, train_time=1.563
+[gpub011:0/16] 2024-02-09 02:10:19,060 (trainer:762) INFO: 36epoch:train:5201-5300batch: iter_time=8.461e-05, forward_time=0.309, loss_ctc=52.532, loss_att=42.091, acc=0.767, loss=45.223, backward_time=0.295, grad_norm=44.576, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.682e-04, train_time=1.366
+[gpub011:0/16] 2024-02-09 02:12:26,367 (trainer:762) INFO: 36epoch:train:5301-5400batch: iter_time=8.307e-05, forward_time=0.313, loss_ctc=46.797, loss_att=51.406, acc=0.754, loss=50.023, backward_time=0.302, grad_norm=40.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.682e-04, train_time=1.273
+[gpub011:0/16] 2024-02-09 02:14:54,577 (trainer:762) INFO: 36epoch:train:5401-5500batch: iter_time=8.193e-05, forward_time=0.396, loss_ctc=47.376, loss_att=50.371, acc=0.757, loss=49.473, backward_time=0.338, grad_norm=41.183, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.682e-04, train_time=1.482
+[gpub011:0/16] 2024-02-09 02:17:05,356 (trainer:762) INFO: 36epoch:train:5501-5600batch: iter_time=8.462e-05, forward_time=0.291, loss_ctc=47.125, loss_att=51.086, acc=0.761, loss=49.898, backward_time=0.297, grad_norm=40.527, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.682e-04, train_time=1.307
+[gpub011:0/16] 2024-02-09 02:19:09,691 (trainer:762) INFO: 36epoch:train:5601-5700batch: iter_time=8.419e-05, forward_time=0.293, loss_ctc=52.339, loss_att=48.507, acc=0.764, loss=49.657, backward_time=0.299, grad_norm=46.183, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.681e-04, train_time=1.244
+[gpub011:0/16] 2024-02-09 02:21:20,416 (trainer:762) INFO: 36epoch:train:5701-5800batch: iter_time=4.550e-04, forward_time=0.306, loss_ctc=44.406, loss_att=42.061, acc=0.765, loss=42.764, backward_time=0.315, grad_norm=41.123, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.681e-04, train_time=1.307
+[gpub011:0/16] 2024-02-09 02:23:49,665 (trainer:762) INFO: 36epoch:train:5801-5900batch: iter_time=9.714e-05, forward_time=0.366, loss_ctc=52.464, loss_att=52.667, acc=0.758, loss=52.606, backward_time=0.356, grad_norm=47.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.681e-04, train_time=1.492
+[gpub011:0/16] 2024-02-09 02:25:59,996 (trainer:762) INFO: 36epoch:train:5901-6000batch: iter_time=9.393e-05, forward_time=0.291, loss_ctc=50.617, loss_att=47.233, acc=0.756, loss=48.248, backward_time=0.296, grad_norm=47.394, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.681e-04, train_time=1.302
+[gpub011:0/16] 2024-02-09 02:28:00,361 (trainer:762) INFO: 36epoch:train:6001-6100batch: iter_time=9.778e-05, forward_time=0.300, loss_ctc=48.771, loss_att=46.422, acc=0.757, loss=47.126, backward_time=0.305, grad_norm=42.977, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.681e-04, train_time=1.204
+[gpub011:0/16] 2024-02-09 02:30:26,124 (trainer:762) INFO: 36epoch:train:6101-6200batch: iter_time=1.033e-04, forward_time=0.295, loss_ctc=45.089, loss_att=43.606, acc=0.764, loss=44.051, backward_time=0.304, grad_norm=42.082, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.681e-04, train_time=1.457
+[gpub011:0/16] 2024-02-09 02:32:08,881 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-09 02:32:27,803 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 02:32:31,400 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97dc69f7c0>)
+[gpub011:0/16] 2024-02-09 02:32:31,400 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-09 02:32:31,445 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 02:38:03,191 (trainer:762) INFO: 36epoch:train:6201-6300batch: iter_time=3.123, forward_time=0.398, loss_ctc=46.611, loss_att=51.149, acc=0.738, loss=49.787, backward_time=0.328, grad_norm=47.752, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.681e-04, train_time=4.570
+[gpub011:0/16] 2024-02-09 02:40:24,300 (trainer:762) INFO: 36epoch:train:6301-6400batch: iter_time=7.356e-05, forward_time=0.292, loss_ctc=56.083, loss_att=52.301, acc=0.758, loss=53.436, backward_time=0.297, grad_norm=58.010, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.680e-04, train_time=1.412
+[gpub011:0/16] 2024-02-09 02:42:37,801 (trainer:762) INFO: 36epoch:train:6401-6500batch: iter_time=7.824e-05, forward_time=0.299, loss_ctc=54.068, loss_att=48.038, acc=0.761, loss=49.847, backward_time=0.316, grad_norm=39.138, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.680e-04, train_time=1.335
+[gpub011:0/16] 2024-02-09 02:45:02,071 (trainer:762) INFO: 36epoch:train:6501-6600batch: iter_time=7.890e-05, forward_time=0.372, loss_ctc=43.773, loss_att=39.718, acc=0.765, loss=40.935, backward_time=0.332, grad_norm=36.901, clip=100.000, loss_scale=6.179e+33, optim_step_time=0.096, optim0_lr0=1.680e-04, train_time=1.442
+[gpub011:0/16] 2024-02-09 02:47:16,630 (trainer:762) INFO: 36epoch:train:6601-6700batch: iter_time=8.212e-05, forward_time=0.296, loss_ctc=51.119, loss_att=58.545, acc=0.743, loss=56.317, backward_time=0.300, grad_norm=44.941, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.680e-04, train_time=1.346
+[gpub011:0/16] 2024-02-09 02:49:40,162 (trainer:762) INFO: 36epoch:train:6701-6800batch: iter_time=8.255e-05, forward_time=0.295, loss_ctc=45.344, loss_att=46.958, acc=0.779, loss=46.474, backward_time=0.309, grad_norm=40.406, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.680e-04, train_time=1.436
+[gpub011:0/16] 2024-02-09 02:50:03,568 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 02:51:37,313 (trainer:762) INFO: 36epoch:train:6801-6900batch: iter_time=8.243e-05, forward_time=0.292, loss_ctc=52.181, loss_att=52.838, acc=0.745, loss=52.641, backward_time=0.305, grad_norm=46.028, clip=100.000, loss_scale=6.084e+33, optim_step_time=0.093, optim0_lr0=1.680e-04, train_time=1.171
+[gpub011:0/16] 2024-02-09 02:54:10,879 (trainer:762) INFO: 36epoch:train:6901-7000batch: iter_time=8.233e-05, forward_time=0.372, loss_ctc=45.753, loss_att=42.516, acc=0.767, loss=43.487, backward_time=0.354, grad_norm=39.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.679e-04, train_time=1.534
+[gpub011:0/16] 2024-02-09 02:56:16,445 (trainer:762) INFO: 36epoch:train:7001-7100batch: iter_time=8.998e-05, forward_time=0.291, loss_ctc=47.515, loss_att=47.655, acc=0.760, loss=47.613, backward_time=0.298, grad_norm=43.300, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.679e-04, train_time=1.257
+[gpub011:0/16] 2024-02-09 02:58:26,437 (trainer:762) INFO: 36epoch:train:7101-7200batch: iter_time=3.471e-04, forward_time=0.307, loss_ctc=53.570, loss_att=50.378, acc=0.765, loss=51.335, backward_time=0.307, grad_norm=46.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.679e-04, train_time=1.300
+[gpub011:0/16] 2024-02-09 02:59:24,103 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 03:00:57,942 (trainer:762) INFO: 36epoch:train:7201-7300batch: iter_time=8.977e-05, forward_time=0.346, loss_ctc=47.366, loss_att=45.623, acc=0.759, loss=46.146, backward_time=0.364, grad_norm=48.513, clip=100.000, loss_scale=3.488e+33, optim_step_time=0.099, optim0_lr0=1.679e-04, train_time=1.515
+[gpub011:0/16] 2024-02-09 03:03:12,196 (trainer:762) INFO: 36epoch:train:7301-7400batch: iter_time=8.211e-05, forward_time=0.292, loss_ctc=46.213, loss_att=43.332, acc=0.756, loss=44.197, backward_time=0.298, grad_norm=37.922, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.679e-04, train_time=1.342
+[gpub011:0/16] 2024-02-09 03:05:06,826 (trainer:762) INFO: 36epoch:train:7401-7500batch: iter_time=8.766e-05, forward_time=0.293, loss_ctc=49.288, loss_att=51.321, acc=0.751, loss=50.711, backward_time=0.300, grad_norm=45.149, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.679e-04, train_time=1.147
+[gpub011:0/16] 2024-02-09 03:05:26,867 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-09 03:05:45,944 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 03:05:49,802 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f984154b670>)
+[gpub011:0/16] 2024-02-09 03:05:49,802 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-09 03:05:49,805 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 03:12:52,955 (trainer:762) INFO: 36epoch:train:7501-7600batch: iter_time=3.297, forward_time=0.397, loss_ctc=51.885, loss_att=52.806, acc=0.745, loss=52.530, backward_time=0.323, grad_norm=52.772, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.678e-04, train_time=4.661
+[gpub011:0/16] 2024-02-09 03:15:03,795 (trainer:762) INFO: 36epoch:train:7601-7700batch: iter_time=8.766e-05, forward_time=0.291, loss_ctc=47.691, loss_att=44.900, acc=0.765, loss=45.737, backward_time=0.297, grad_norm=40.373, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.678e-04, train_time=1.308
+[gpub011:0/16] 2024-02-09 03:17:14,734 (trainer:762) INFO: 36epoch:train:7701-7800batch: iter_time=9.259e-05, forward_time=0.291, loss_ctc=52.036, loss_att=42.755, acc=0.762, loss=45.539, backward_time=0.297, grad_norm=43.246, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.678e-04, train_time=1.308
+[gpub011:0/16] 2024-02-09 03:19:32,346 (trainer:762) INFO: 36epoch:train:7801-7900batch: iter_time=8.845e-05, forward_time=0.398, loss_ctc=46.887, loss_att=50.513, acc=0.757, loss=49.425, backward_time=0.322, grad_norm=41.730, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.678e-04, train_time=1.377
+[gpub011:0/16] 2024-02-09 03:21:52,659 (trainer:762) INFO: 36epoch:train:7901-8000batch: iter_time=8.540e-05, forward_time=0.318, loss_ctc=46.984, loss_att=49.547, acc=0.761, loss=48.779, backward_time=0.311, grad_norm=40.453, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.678e-04, train_time=1.403
+[gpub011:0/16] 2024-02-09 03:24:02,503 (trainer:762) INFO: 36epoch:train:8001-8100batch: iter_time=8.291e-05, forward_time=0.293, loss_ctc=47.303, loss_att=51.014, acc=0.761, loss=49.901, backward_time=0.297, grad_norm=40.787, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.678e-04, train_time=1.298
+[gpub011:0/16] 2024-02-09 03:26:04,026 (trainer:762) INFO: 36epoch:train:8101-8200batch: iter_time=8.477e-05, forward_time=0.293, loss_ctc=52.040, loss_att=47.945, acc=0.766, loss=49.173, backward_time=0.312, grad_norm=42.624, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.678e-04, train_time=1.214
+[gpub011:0/16] 2024-02-09 03:28:35,935 (trainer:762) INFO: 36epoch:train:8201-8300batch: iter_time=8.356e-05, forward_time=0.396, loss_ctc=44.276, loss_att=42.250, acc=0.764, loss=42.858, backward_time=0.314, grad_norm=42.029, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.677e-04, train_time=1.520
+[gpub011:0/16] 2024-02-09 03:30:41,338 (trainer:762) INFO: 36epoch:train:8301-8400batch: iter_time=8.035e-05, forward_time=0.314, loss_ctc=51.569, loss_att=52.570, acc=0.759, loss=52.270, backward_time=0.320, grad_norm=47.231, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.677e-04, train_time=1.254
+[gpub011:0/16] 2024-02-09 03:33:01,294 (trainer:762) INFO: 36epoch:train:8401-8500batch: iter_time=8.082e-05, forward_time=0.292, loss_ctc=49.700, loss_att=46.366, acc=0.759, loss=47.366, backward_time=0.297, grad_norm=45.794, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.677e-04, train_time=1.399
+[gpub011:0/16] 2024-02-09 03:35:33,653 (trainer:762) INFO: 36epoch:train:8501-8600batch: iter_time=9.179e-05, forward_time=0.349, loss_ctc=48.466, loss_att=45.751, acc=0.758, loss=46.566, backward_time=0.348, grad_norm=42.307, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.677e-04, train_time=1.523
+[gpub011:0/16] 2024-02-09 03:37:33,924 (trainer:762) INFO: 36epoch:train:8601-8700batch: iter_time=8.777e-05, forward_time=0.317, loss_ctc=44.471, loss_att=42.005, acc=0.768, loss=42.745, backward_time=0.306, grad_norm=40.442, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.677e-04, train_time=1.202
+[gpub011:0/16] 2024-02-09 03:39:05,003 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-09 03:39:24,741 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 03:39:28,700 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9826c4f880>)
+[gpub011:0/16] 2024-02-09 03:39:28,700 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-09 03:39:28,703 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 03:45:08,061 (trainer:762) INFO: 36epoch:train:8701-8800batch: iter_time=3.213, forward_time=0.326, loss_ctc=45.926, loss_att=52.251, acc=0.732, loss=50.354, backward_time=0.299, grad_norm=47.047, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.677e-04, train_time=4.541
+[gpub011:0/16] 2024-02-09 03:47:10,638 (trainer:762) INFO: 36epoch:train:8801-8900batch: iter_time=8.694e-05, forward_time=0.313, loss_ctc=55.615, loss_att=52.200, acc=0.745, loss=53.224, backward_time=0.329, grad_norm=54.329, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.676e-04, train_time=1.225
+[gpub011:0/16] 2024-02-09 03:49:24,999 (trainer:762) INFO: 36epoch:train:8901-9000batch: iter_time=8.772e-05, forward_time=0.309, loss_ctc=53.188, loss_att=47.498, acc=0.744, loss=49.205, backward_time=0.302, grad_norm=42.566, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.676e-04, train_time=1.344
+[gpub011:0/16] 2024-02-09 03:51:41,427 (trainer:762) INFO: 36epoch:train:9001-9100batch: iter_time=9.216e-05, forward_time=0.335, loss_ctc=44.067, loss_att=41.382, acc=0.748, loss=42.187, backward_time=0.296, grad_norm=39.271, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.676e-04, train_time=1.364
+[gpub011:0/16] 2024-02-09 03:53:39,954 (trainer:762) INFO: 36epoch:train:9101-9200batch: iter_time=8.061e-05, forward_time=0.325, loss_ctc=50.350, loss_att=58.482, acc=0.736, loss=56.042, backward_time=0.319, grad_norm=43.634, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.676e-04, train_time=1.185
+[gpub011:0/16] 2024-02-09 03:55:55,669 (trainer:762) INFO: 36epoch:train:9201-9300batch: iter_time=8.730e-05, forward_time=0.291, loss_ctc=45.359, loss_att=48.118, acc=0.767, loss=47.290, backward_time=0.298, grad_norm=39.181, clip=100.000, loss_scale=4.284e+33, optim_step_time=0.093, optim0_lr0=1.676e-04, train_time=1.357
+[gpub011:0/16] 2024-02-09 03:58:24,080 (trainer:762) INFO: 36epoch:train:9301-9400batch: iter_time=8.834e-05, forward_time=0.335, loss_ctc=51.928, loss_att=51.821, acc=0.749, loss=51.853, backward_time=0.338, grad_norm=43.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.676e-04, train_time=1.484
+[gpub011:0/16] 2024-02-09 04:00:26,746 (trainer:762) INFO: 36epoch:train:9401-9500batch: iter_time=8.174e-05, forward_time=0.304, loss_ctc=45.134, loss_att=42.637, acc=0.764, loss=43.386, backward_time=0.298, grad_norm=38.708, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=1.227
+[gpub011:0/16] 2024-02-09 04:02:38,744 (trainer:762) INFO: 36epoch:train:9501-9600batch: iter_time=8.406e-05, forward_time=0.316, loss_ctc=47.120, loss_att=47.974, acc=0.754, loss=47.718, backward_time=0.307, grad_norm=43.212, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=1.319
+[gpub011:0/16] 2024-02-09 04:05:03,879 (trainer:762) INFO: 36epoch:train:9601-9700batch: iter_time=8.626e-05, forward_time=0.311, loss_ctc=52.899, loss_att=50.155, acc=0.758, loss=50.978, backward_time=0.310, grad_norm=46.515, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=1.451
+[gpub011:0/16] 2024-02-09 04:07:24,887 (trainer:762) INFO: 36epoch:train:9701-9800batch: iter_time=8.251e-05, forward_time=0.317, loss_ctc=46.839, loss_att=45.652, acc=0.753, loss=46.008, backward_time=0.305, grad_norm=49.403, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.675e-04, train_time=1.411
+[gpub011:0/16] 2024-02-09 04:09:27,447 (trainer:762) INFO: 36epoch:train:9801-9900batch: iter_time=8.501e-05, forward_time=0.338, loss_ctc=46.158, loss_att=44.065, acc=0.739, loss=44.693, backward_time=0.310, grad_norm=41.158, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=1.225
+[gpub011:0/16] 2024-02-09 04:11:41,343 (trainer:762) INFO: 36epoch:train:9901-10000batch: iter_time=8.025e-05, forward_time=0.309, loss_ctc=48.195, loss_att=50.695, acc=0.738, loss=49.945, backward_time=0.316, grad_norm=44.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=1.339
+[gpub011:0/16] 2024-02-09 04:12:01,472 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-09 04:12:20,566 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 04:12:24,337 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97cfb7b610>)
+[gpub011:0/16] 2024-02-09 04:12:24,337 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-09 04:12:24,342 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 04:19:13,158 (trainer:762) INFO: 36epoch:train:10001-10100batch: iter_time=3.263, forward_time=0.341, loss_ctc=52.419, loss_att=53.766, acc=0.745, loss=53.362, backward_time=0.310, grad_norm=58.034, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.675e-04, train_time=4.518
+[gpub011:0/16] 2024-02-09 04:21:29,348 (trainer:762) INFO: 36epoch:train:10101-10200batch: iter_time=8.807e-05, forward_time=0.317, loss_ctc=48.184, loss_att=46.483, acc=0.762, loss=46.994, backward_time=0.299, grad_norm=43.901, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.674e-04, train_time=1.362
+[gpub011:0/16] 2024-02-09 04:23:51,313 (trainer:762) INFO: 36epoch:train:10201-10300batch: iter_time=8.955e-05, forward_time=0.336, loss_ctc=52.263, loss_att=41.740, acc=0.769, loss=44.897, backward_time=0.323, grad_norm=43.353, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.674e-04, train_time=1.419
+[gpub011:0/16] 2024-02-09 04:25:59,413 (trainer:762) INFO: 36epoch:train:10301-10400batch: iter_time=8.629e-05, forward_time=0.297, loss_ctc=46.727, loss_att=50.958, acc=0.756, loss=49.689, backward_time=0.305, grad_norm=41.266, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.674e-04, train_time=1.281
+[gpub011:0/16] 2024-02-09 04:28:15,050 (trainer:762) INFO: 36epoch:train:10401-10500batch: iter_time=2.441e-04, forward_time=0.331, loss_ctc=47.026, loss_att=49.926, acc=0.759, loss=49.056, backward_time=0.307, grad_norm=41.092, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.674e-04, train_time=1.356
+[gpub011:0/16] 2024-02-09 04:30:39,218 (trainer:762) INFO: 36epoch:train:10501-10600batch: iter_time=8.732e-05, forward_time=0.359, loss_ctc=46.766, loss_att=50.186, acc=0.764, loss=49.160, backward_time=0.321, grad_norm=40.802, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.674e-04, train_time=1.441
+[gpub011:0/16] 2024-02-09 04:32:46,084 (trainer:762) INFO: 36epoch:train:10601-10700batch: iter_time=9.261e-05, forward_time=0.321, loss_ctc=52.749, loss_att=48.272, acc=0.767, loss=49.615, backward_time=0.331, grad_norm=43.978, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.674e-04, train_time=1.269
+[gpub011:0/16] 2024-02-09 04:35:01,180 (trainer:762) INFO: 36epoch:train:10701-10800batch: iter_time=9.316e-05, forward_time=0.294, loss_ctc=43.773, loss_att=41.692, acc=0.767, loss=42.317, backward_time=0.295, grad_norm=41.522, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.673e-04, train_time=1.350
+[gpub011:0/16] 2024-02-09 04:37:22,142 (trainer:762) INFO: 36epoch:train:10801-10900batch: iter_time=2.150e-04, forward_time=0.351, loss_ctc=51.907, loss_att=52.620, acc=0.759, loss=52.406, backward_time=0.333, grad_norm=44.626, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.673e-04, train_time=1.409
+[gpub011:0/16] 2024-02-09 04:39:25,705 (trainer:762) INFO: 36epoch:train:10901-11000batch: iter_time=8.729e-05, forward_time=0.331, loss_ctc=49.757, loss_att=46.093, acc=0.759, loss=47.192, backward_time=0.316, grad_norm=45.785, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.673e-04, train_time=1.236
+[gpub011:0/16] 2024-02-09 04:41:48,903 (trainer:762) INFO: 36epoch:train:11001-11100batch: iter_time=8.552e-05, forward_time=0.327, loss_ctc=48.127, loss_att=46.086, acc=0.759, loss=46.698, backward_time=0.344, grad_norm=42.917, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.673e-04, train_time=1.431
+[gpub011:0/16] 2024-02-09 04:44:08,003 (trainer:762) INFO: 36epoch:train:11101-11200batch: iter_time=9.159e-05, forward_time=0.321, loss_ctc=44.298, loss_att=43.084, acc=0.765, loss=43.448, backward_time=0.329, grad_norm=39.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.673e-04, train_time=1.391
+[gpub011:0/16] 2024-02-09 04:45:27,957 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-09 04:45:47,327 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 04:45:51,258 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97cfb1b970>)
+[gpub011:0/16] 2024-02-09 04:45:51,258 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-09 04:45:51,261 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 04:51:42,164 (trainer:762) INFO: 36epoch:train:11201-11300batch: iter_time=3.273, forward_time=0.293, loss_ctc=45.631, loss_att=51.658, acc=0.734, loss=49.850, backward_time=0.301, grad_norm=47.757, clip=100.000, loss_scale=8.567e+33, optim_step_time=0.093, optim0_lr0=1.673e-04, train_time=4.541
+[gpub011:0/16] 2024-02-09 04:53:54,494 (trainer:762) INFO: 36epoch:train:11301-11400batch: iter_time=8.269e-05, forward_time=0.370, loss_ctc=55.000, loss_att=51.266, acc=0.748, loss=52.386, backward_time=0.318, grad_norm=56.400, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.673e-04, train_time=1.323
+[gpub011:0/16] 2024-02-09 04:56:08,736 (trainer:762) INFO: 36epoch:train:11401-11500batch: iter_time=8.389e-05, forward_time=0.329, loss_ctc=53.357, loss_att=46.836, acc=0.749, loss=48.792, backward_time=0.304, grad_norm=41.725, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.672e-04, train_time=1.342
+[gpub011:0/16] 2024-02-09 04:58:27,192 (trainer:762) INFO: 36epoch:train:11501-11600batch: iter_time=9.527e-05, forward_time=0.302, loss_ctc=43.676, loss_att=40.509, acc=0.752, loss=41.459, backward_time=0.303, grad_norm=38.104, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.672e-04, train_time=1.384
+[gpub011:0/16] 2024-02-09 05:00:36,735 (trainer:762) INFO: 36epoch:train:11601-11700batch: iter_time=8.950e-05, forward_time=0.317, loss_ctc=50.372, loss_att=58.780, acc=0.735, loss=56.258, backward_time=0.311, grad_norm=44.363, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.672e-04, train_time=1.295
+[gpub011:0/16] 2024-02-09 05:02:48,963 (trainer:762) INFO: 36epoch:train:11701-11800batch: iter_time=9.125e-05, forward_time=0.316, loss_ctc=44.820, loss_att=47.064, acc=0.770, loss=46.391, backward_time=0.313, grad_norm=39.584, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.672e-04, train_time=1.322
+[gpub011:0/16] 2024-02-09 05:05:13,045 (trainer:762) INFO: 36epoch:train:11801-11900batch: iter_time=8.679e-05, forward_time=0.314, loss_ctc=51.224, loss_att=51.263, acc=0.752, loss=51.252, backward_time=0.324, grad_norm=44.226, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.672e-04, train_time=1.441
+[gpub011:0/16] 2024-02-09 05:07:17,397 (trainer:762) INFO: 36epoch:train:11901-12000batch: iter_time=8.630e-05, forward_time=0.290, loss_ctc=45.416, loss_att=42.753, acc=0.762, loss=43.552, backward_time=0.297, grad_norm=40.050, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.672e-04, train_time=1.243
+[gpub011:0/16] 2024-02-09 05:09:26,033 (trainer:762) INFO: 36epoch:train:12001-12100batch: iter_time=8.810e-05, forward_time=0.327, loss_ctc=47.175, loss_att=47.583, acc=0.755, loss=47.461, backward_time=0.343, grad_norm=41.163, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.671e-04, train_time=1.286
+[gpub011:0/16] 2024-02-09 05:11:46,181 (trainer:762) INFO: 36epoch:train:12101-12200batch: iter_time=8.916e-05, forward_time=0.334, loss_ctc=53.143, loss_att=50.182, acc=0.759, loss=51.070, backward_time=0.320, grad_norm=46.633, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.671e-04, train_time=1.401
+[gpub011:0/16] 2024-02-09 05:14:03,835 (trainer:762) INFO: 36epoch:train:12201-12300batch: iter_time=8.596e-05, forward_time=0.294, loss_ctc=47.262, loss_att=45.526, acc=0.753, loss=46.047, backward_time=0.295, grad_norm=47.404, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.671e-04, train_time=1.376
+[gpub011:0/16] 2024-02-09 05:16:05,583 (trainer:762) INFO: 36epoch:train:12301-12400batch: iter_time=8.472e-05, forward_time=0.313, loss_ctc=46.387, loss_att=44.246, acc=0.738, loss=44.888, backward_time=0.302, grad_norm=40.133, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.671e-04, train_time=1.217
+[gpub011:0/16] 2024-02-09 05:18:22,363 (trainer:762) INFO: 36epoch:train:12401-12500batch: iter_time=8.122e-05, forward_time=0.323, loss_ctc=47.720, loss_att=50.703, acc=0.738, loss=49.808, backward_time=0.300, grad_norm=43.746, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.671e-04, train_time=1.368
+[gpub011:0/16] 2024-02-09 05:18:42,409 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-09 05:19:01,509 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 05:19:05,354 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f980dea7580>)
+[gpub011:0/16] 2024-02-09 05:19:05,354 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-09 05:19:05,358 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 05:25:23,666 (trainer:762) INFO: 36epoch:train:12501-12600batch: iter_time=2.966, forward_time=0.326, loss_ctc=52.291, loss_att=54.344, acc=0.743, loss=53.728, backward_time=0.308, grad_norm=56.640, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.671e-04, train_time=4.213
+[gpub011:0/16] 2024-02-09 05:27:43,572 (trainer:762) INFO: 36epoch:train:12601-12700batch: iter_time=8.554e-05, forward_time=0.315, loss_ctc=46.870, loss_att=45.301, acc=0.765, loss=45.772, backward_time=0.306, grad_norm=40.093, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.671e-04, train_time=1.399
+[gpub011:0/16] 2024-02-09 05:30:01,609 (trainer:762) INFO: 36epoch:train:12701-12800batch: iter_time=8.884e-05, forward_time=0.330, loss_ctc=51.709, loss_att=42.659, acc=0.764, loss=45.374, backward_time=0.300, grad_norm=41.783, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.670e-04, train_time=1.380
+[gpub011:0/16] 2024-02-09 05:32:05,385 (trainer:762) INFO: 36epoch:train:12801-12900batch: iter_time=8.385e-05, forward_time=0.320, loss_ctc=46.613, loss_att=50.559, acc=0.759, loss=49.375, backward_time=0.325, grad_norm=38.894, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.670e-04, train_time=1.237
+[gpub011:0/16] 2024-02-09 05:34:25,414 (trainer:762) INFO: 36epoch:train:12901-13000batch: iter_time=9.054e-05, forward_time=0.301, loss_ctc=46.199, loss_att=49.717, acc=0.762, loss=48.661, backward_time=0.309, grad_norm=39.229, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.670e-04, train_time=1.400
+[gpub011:0/16] 2024-02-09 05:36:44,301 (trainer:762) INFO: 36epoch:train:13001-13100batch: iter_time=8.888e-05, forward_time=0.351, loss_ctc=47.001, loss_att=50.354, acc=0.764, loss=49.348, backward_time=0.318, grad_norm=41.938, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.670e-04, train_time=1.389
+[gpub011:0/16] 2024-02-09 05:38:48,063 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 05:38:54,166 (trainer:762) INFO: 36epoch:train:13101-13200batch: iter_time=9.449e-05, forward_time=0.320, loss_ctc=52.217, loss_att=48.017, acc=0.767, loss=49.277, backward_time=0.328, grad_norm=45.318, clip=100.000, loss_scale=1.017e+34, optim_step_time=0.097, optim0_lr0=1.670e-04, train_time=1.299
+[gpub011:0/16] 2024-02-09 05:41:10,373 (trainer:762) INFO: 36epoch:train:13201-13300batch: iter_time=9.177e-05, forward_time=0.290, loss_ctc=43.597, loss_att=42.389, acc=0.765, loss=42.752, backward_time=0.295, grad_norm=42.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.670e-04, train_time=1.362
+[gpub011:0/16] 2024-02-09 05:43:29,887 (trainer:762) INFO: 36epoch:train:13301-13400batch: iter_time=9.321e-05, forward_time=0.363, loss_ctc=51.244, loss_att=52.087, acc=0.760, loss=51.834, backward_time=0.315, grad_norm=44.435, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.669e-04, train_time=1.395
+[gpub011:0/16] 2024-02-09 05:45:33,940 (trainer:762) INFO: 36epoch:train:13401-13500batch: iter_time=9.242e-05, forward_time=0.330, loss_ctc=49.988, loss_att=46.188, acc=0.759, loss=47.328, backward_time=0.313, grad_norm=46.063, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.669e-04, train_time=1.240
+[gpub011:0/16] 2024-02-09 05:48:11,466 (trainer:762) INFO: 36epoch:train:13501-13600batch: iter_time=9.796e-05, forward_time=0.303, loss_ctc=48.321, loss_att=46.018, acc=0.759, loss=46.709, backward_time=0.300, grad_norm=43.919, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.669e-04, train_time=1.575
+[gpub011:0/16] 2024-02-09 05:50:13,514 (trainer:762) INFO: 36epoch:train:13601-13700batch: iter_time=1.003e-04, forward_time=0.319, loss_ctc=44.229, loss_att=42.363, acc=0.771, loss=42.923, backward_time=0.304, grad_norm=39.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.669e-04, train_time=1.220
+[gpub011:0/16] 2024-02-09 05:51:40,507 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-09 05:51:59,910 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 05:52:03,493 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97f93eaa70>)
+[gpub011:0/16] 2024-02-09 05:52:03,493 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-09 05:52:03,524 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 05:57:30,418 (trainer:762) INFO: 36epoch:train:13701-13800batch: iter_time=3.052, forward_time=0.363, loss_ctc=45.686, loss_att=51.870, acc=0.732, loss=50.015, backward_time=0.327, grad_norm=46.847, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.669e-04, train_time=4.369
+[gpub011:0/16] 2024-02-09 05:59:31,713 (trainer:762) INFO: 36epoch:train:13801-13900batch: iter_time=8.158e-05, forward_time=0.311, loss_ctc=54.688, loss_att=51.012, acc=0.747, loss=52.115, backward_time=0.308, grad_norm=58.176, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.669e-04, train_time=1.213
+[gpub011:0/16] 2024-02-09 06:01:54,741 (trainer:762) INFO: 36epoch:train:13901-14000batch: iter_time=3.219e-04, forward_time=0.354, loss_ctc=53.026, loss_att=47.116, acc=0.746, loss=48.889, backward_time=0.335, grad_norm=43.871, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.668e-04, train_time=1.430
+[gpub011:0/16] 2024-02-09 06:04:19,479 (trainer:762) INFO: 36epoch:train:14001-14100batch: iter_time=8.515e-05, forward_time=0.299, loss_ctc=43.135, loss_att=40.486, acc=0.753, loss=41.280, backward_time=0.306, grad_norm=39.223, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.668e-04, train_time=1.447
+[gpub011:0/16] 2024-02-09 06:06:38,699 (trainer:762) INFO: 36epoch:train:14101-14200batch: iter_time=8.543e-05, forward_time=0.379, loss_ctc=49.990, loss_att=58.122, acc=0.739, loss=55.682, backward_time=0.329, grad_norm=43.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.668e-04, train_time=1.392
+[gpub011:0/16] 2024-02-09 06:08:49,549 (trainer:762) INFO: 36epoch:train:14201-14300batch: iter_time=8.697e-05, forward_time=0.304, loss_ctc=44.803, loss_att=47.061, acc=0.768, loss=46.383, backward_time=0.303, grad_norm=40.609, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.668e-04, train_time=1.309
+[gpub011:0/16] 2024-02-09 06:11:30,318 (trainer:762) INFO: 36epoch:train:14301-14400batch: iter_time=8.470e-05, forward_time=0.388, loss_ctc=51.727, loss_att=51.575, acc=0.750, loss=51.621, backward_time=0.322, grad_norm=44.126, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.668e-04, train_time=1.607
+[gpub011:0/16] 2024-02-09 06:13:29,682 (trainer:762) INFO: 36epoch:train:14401-14500batch: iter_time=8.768e-05, forward_time=0.294, loss_ctc=44.924, loss_att=42.385, acc=0.763, loss=43.147, backward_time=0.300, grad_norm=41.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.668e-04, train_time=1.193
+[gpub011:0/16] 2024-02-09 06:16:08,360 (trainer:762) INFO: 36epoch:train:14501-14600batch: iter_time=8.404e-05, forward_time=0.341, loss_ctc=46.910, loss_att=47.860, acc=0.755, loss=47.575, backward_time=0.308, grad_norm=42.065, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.668e-04, train_time=1.587
+[gpub011:0/16] 2024-02-09 06:18:16,135 (trainer:762) INFO: 36epoch:train:14601-14700batch: iter_time=8.577e-05, forward_time=0.338, loss_ctc=52.500, loss_att=49.895, acc=0.760, loss=50.677, backward_time=0.329, grad_norm=44.041, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.667e-04, train_time=1.277
+[gpub011:0/16] 2024-02-09 06:20:42,104 (trainer:762) INFO: 36epoch:train:14701-14800batch: iter_time=8.849e-05, forward_time=0.313, loss_ctc=46.079, loss_att=44.751, acc=0.756, loss=45.149, backward_time=0.300, grad_norm=45.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.667e-04, train_time=1.460
+[gpub011:0/16] 2024-02-09 06:23:12,636 (trainer:762) INFO: 36epoch:train:14801-14900batch: iter_time=0.001, forward_time=0.366, loss_ctc=46.137, loss_att=43.992, acc=0.739, loss=44.636, backward_time=0.361, grad_norm=40.164, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.667e-04, train_time=1.504
+[gpub011:0/16] 2024-02-09 06:25:31,116 (trainer:762) INFO: 36epoch:train:14901-15000batch: iter_time=8.137e-05, forward_time=0.306, loss_ctc=47.736, loss_att=50.310, acc=0.740, loss=49.538, backward_time=0.308, grad_norm=44.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.667e-04, train_time=1.385
+[gpub011:0/16] 2024-02-09 07:03:27,940 (trainer:361) INFO: 36epoch results: [train] iter_time=0.259, forward_time=0.323, loss_ctc=49.145, loss_att=48.232, acc=0.753, loss=48.506, backward_time=0.311, grad_norm=44.299, clip=100.000, loss_scale=5.565e+33, optim_step_time=0.095, optim0_lr0=1.679e-04, train_time=1.603, time=6 hours, 41 minutes and 11.18 seconds, total_count=570000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=35.002, cer_ctc=0.183, loss_att=38.491, acc=0.681, cer=0.336, wer=0.995, loss=37.444, time=37 minutes and 32.31 seconds, total_count=177498, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-09 07:03:36,946 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-09 07:03:37,023 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/25epoch.pth
+[gpub011:0/16] 2024-02-09 07:03:37,023 (trainer:290) INFO: 37/45epoch started. Estimated time to finish: 2 days, 18 hours and 42 minutes
+[gpub011:0/16] 2024-02-09 07:03:37,033 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-09 07:03:55,400 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 07:03:58,782 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fb731a0b0>)
+[gpub011:0/16] 2024-02-09 07:03:58,783 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-09 07:03:58,786 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 07:10:52,677 (trainer:762) INFO: 37epoch:train:1-100batch: iter_time=3.182, forward_time=0.373, loss_ctc=49.240, loss_att=44.451, acc=0.751, loss=45.888, backward_time=0.304, grad_norm=45.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.667e-04, train_time=4.356
+[gpub011:0/16] 2024-02-09 07:12:43,997 (trainer:762) INFO: 37epoch:train:101-200batch: iter_time=9.024e-05, forward_time=0.291, loss_ctc=48.026, loss_att=41.576, acc=0.762, loss=43.511, backward_time=0.303, grad_norm=42.030, clip=100.000, loss_scale=5.400e+33, optim_step_time=0.093, optim0_lr0=1.667e-04, train_time=1.113
+[gpub011:0/16] 2024-02-09 07:15:16,563 (trainer:762) INFO: 37epoch:train:201-300batch: iter_time=5.546e-04, forward_time=0.368, loss_ctc=49.152, loss_att=50.510, acc=0.754, loss=50.103, backward_time=0.349, grad_norm=44.317, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.666e-04, train_time=1.525
+[gpub011:0/16] 2024-02-09 07:17:13,376 (trainer:762) INFO: 37epoch:train:301-400batch: iter_time=9.469e-05, forward_time=0.301, loss_ctc=53.910, loss_att=45.743, acc=0.754, loss=48.193, backward_time=0.298, grad_norm=50.137, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.666e-04, train_time=1.168
+[gpub011:0/16] 2024-02-09 07:19:22,635 (trainer:762) INFO: 37epoch:train:401-500batch: iter_time=1.005e-04, forward_time=0.294, loss_ctc=53.586, loss_att=53.870, acc=0.737, loss=53.785, backward_time=0.301, grad_norm=49.730, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.666e-04, train_time=1.292
+[gpub011:0/16] 2024-02-09 07:21:46,504 (trainer:762) INFO: 37epoch:train:501-600batch: iter_time=9.318e-05, forward_time=0.307, loss_ctc=48.007, loss_att=50.247, acc=0.740, loss=49.575, backward_time=0.323, grad_norm=43.915, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.666e-04, train_time=1.438
+[gpub011:0/16] 2024-02-09 07:23:50,747 (trainer:762) INFO: 37epoch:train:601-700batch: iter_time=1.024e-04, forward_time=0.331, loss_ctc=49.536, loss_att=48.507, acc=0.746, loss=48.816, backward_time=0.329, grad_norm=43.764, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.666e-04, train_time=1.242
+[gpub011:0/16] 2024-02-09 07:25:50,585 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 07:26:03,819 (trainer:762) INFO: 37epoch:train:701-800batch: iter_time=9.640e-05, forward_time=0.291, loss_ctc=42.396, loss_att=35.373, acc=0.777, loss=37.480, backward_time=0.300, grad_norm=37.520, clip=100.000, loss_scale=9.755e+33, optim_step_time=0.093, optim0_lr0=1.666e-04, train_time=1.331
+[gpub011:0/16] 2024-02-09 07:28:12,615 (trainer:762) INFO: 37epoch:train:801-900batch: iter_time=9.698e-05, forward_time=0.312, loss_ctc=45.694, loss_att=41.858, acc=0.747, loss=43.009, backward_time=0.305, grad_norm=42.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.666e-04, train_time=1.288
+[gpub011:0/16] 2024-02-09 07:30:27,141 (trainer:762) INFO: 37epoch:train:901-1000batch: iter_time=9.338e-05, forward_time=0.337, loss_ctc=46.603, loss_att=50.184, acc=0.745, loss=49.110, backward_time=0.332, grad_norm=45.108, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.665e-04, train_time=1.345
+[gpub011:0/16] 2024-02-09 07:32:39,675 (trainer:762) INFO: 37epoch:train:1001-1100batch: iter_time=9.314e-05, forward_time=0.295, loss_ctc=49.048, loss_att=44.901, acc=0.760, loss=46.145, backward_time=0.304, grad_norm=40.738, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.665e-04, train_time=1.325
+[gpub011:0/16] 2024-02-09 07:34:54,548 (trainer:762) INFO: 37epoch:train:1101-1200batch: iter_time=3.494e-04, forward_time=0.370, loss_ctc=43.149, loss_att=39.572, acc=0.755, loss=40.645, backward_time=0.316, grad_norm=38.232, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.665e-04, train_time=1.349
+[gpub011:0/16] 2024-02-09 07:36:11,998 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-09 07:36:31,151 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 07:36:34,659 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065a1a110>)
+[gpub011:0/16] 2024-02-09 07:36:34,659 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-09 07:36:34,682 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 07:42:18,391 (trainer:762) INFO: 37epoch:train:1201-1300batch: iter_time=3.149, forward_time=0.290, loss_ctc=45.962, loss_att=43.875, acc=0.762, loss=44.501, backward_time=0.297, grad_norm=43.657, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.665e-04, train_time=4.438
+[gpub011:0/16] 2024-02-09 07:44:13,572 (trainer:762) INFO: 37epoch:train:1301-1400batch: iter_time=7.391e-05, forward_time=0.290, loss_ctc=48.063, loss_att=42.476, acc=0.757, loss=44.152, backward_time=0.296, grad_norm=42.490, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.665e-04, train_time=1.151
+[gpub011:0/16] 2024-02-09 07:46:16,980 (trainer:762) INFO: 37epoch:train:1401-1500batch: iter_time=8.344e-05, forward_time=0.331, loss_ctc=48.128, loss_att=51.115, acc=0.762, loss=50.219, backward_time=0.324, grad_norm=42.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.665e-04, train_time=1.234
+[gpub011:0/16] 2024-02-09 07:48:35,199 (trainer:762) INFO: 37epoch:train:1501-1600batch: iter_time=8.844e-05, forward_time=0.359, loss_ctc=49.102, loss_att=44.942, acc=0.775, loss=46.190, backward_time=0.311, grad_norm=49.347, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.664e-04, train_time=1.382
+[gpub011:0/16] 2024-02-09 07:50:50,886 (trainer:762) INFO: 37epoch:train:1601-1700batch: iter_time=9.246e-05, forward_time=0.292, loss_ctc=54.404, loss_att=51.382, acc=0.756, loss=52.289, backward_time=0.296, grad_norm=49.812, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.664e-04, train_time=1.357
+[gpub011:0/16] 2024-02-09 07:52:43,018 (trainer:762) INFO: 37epoch:train:1701-1800batch: iter_time=8.910e-05, forward_time=0.296, loss_ctc=51.089, loss_att=51.336, acc=0.754, loss=51.262, backward_time=0.308, grad_norm=43.658, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.664e-04, train_time=1.121
+[gpub011:0/16] 2024-02-09 07:54:43,744 (trainer:762) INFO: 37epoch:train:1801-1900batch: iter_time=9.080e-05, forward_time=0.318, loss_ctc=46.729, loss_att=46.924, acc=0.748, loss=46.865, backward_time=0.320, grad_norm=46.501, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.664e-04, train_time=1.207
+[gpub011:0/16] 2024-02-09 07:56:06,995 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 07:57:19,551 (trainer:762) INFO: 37epoch:train:1901-2000batch: iter_time=9.699e-05, forward_time=0.349, loss_ctc=46.581, loss_att=44.763, acc=0.774, loss=45.308, backward_time=0.319, grad_norm=44.853, clip=100.000, loss_scale=3.540e+33, optim_step_time=0.095, optim0_lr0=1.664e-04, train_time=1.558
+[gpub011:0/16] 2024-02-09 07:59:10,133 (trainer:762) INFO: 37epoch:train:2001-2100batch: iter_time=8.307e-05, forward_time=0.293, loss_ctc=47.877, loss_att=41.666, acc=0.767, loss=43.529, backward_time=0.298, grad_norm=40.406, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.664e-04, train_time=1.106
+[gpub011:0/16] 2024-02-09 08:01:12,388 (trainer:762) INFO: 37epoch:train:2101-2200batch: iter_time=8.757e-05, forward_time=0.293, loss_ctc=40.650, loss_att=38.848, acc=0.768, loss=39.389, backward_time=0.300, grad_norm=36.379, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.664e-04, train_time=1.222
+[gpub011:0/16] 2024-02-09 08:03:59,031 (trainer:762) INFO: 37epoch:train:2201-2300batch: iter_time=9.228e-05, forward_time=0.366, loss_ctc=51.328, loss_att=52.678, acc=0.761, loss=52.273, backward_time=0.334, grad_norm=46.632, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.663e-04, train_time=1.666
+[gpub011:0/16] 2024-02-09 08:05:49,384 (trainer:762) INFO: 37epoch:train:2301-2400batch: iter_time=8.798e-05, forward_time=0.304, loss_ctc=39.110, loss_att=36.333, acc=0.779, loss=37.166, backward_time=0.297, grad_norm=33.691, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.663e-04, train_time=1.103
+[gpub011:0/16] 2024-02-09 08:07:56,908 (trainer:762) INFO: 37epoch:train:2401-2500batch: iter_time=8.799e-05, forward_time=0.291, loss_ctc=47.603, loss_att=43.863, acc=0.751, loss=44.985, backward_time=0.295, grad_norm=43.857, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.663e-04, train_time=1.275
+[gpub011:0/16] 2024-02-09 08:08:16,936 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-09 08:08:35,897 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 08:08:39,408 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97d745afe0>)
+[gpub011:0/16] 2024-02-09 08:08:39,408 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-09 08:08:39,411 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 08:14:48,136 (trainer:762) INFO: 37epoch:train:2501-2600batch: iter_time=2.848, forward_time=0.305, loss_ctc=47.899, loss_att=42.322, acc=0.772, loss=43.995, backward_time=0.324, grad_norm=39.981, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.663e-04, train_time=4.112
+[gpub011:0/16] 2024-02-09 08:16:59,975 (trainer:762) INFO: 37epoch:train:2601-2700batch: iter_time=4.871e-04, forward_time=0.344, loss_ctc=47.882, loss_att=41.495, acc=0.768, loss=43.411, backward_time=0.300, grad_norm=43.408, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.663e-04, train_time=1.318
+[gpub011:0/16] 2024-02-09 08:19:18,916 (trainer:762) INFO: 37epoch:train:2701-2800batch: iter_time=8.385e-05, forward_time=0.299, loss_ctc=48.515, loss_att=50.930, acc=0.767, loss=50.205, backward_time=0.304, grad_norm=40.990, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.663e-04, train_time=1.389
+[gpub011:0/16] 2024-02-09 08:21:29,893 (trainer:762) INFO: 37epoch:train:2801-2900batch: iter_time=8.873e-05, forward_time=0.289, loss_ctc=50.937, loss_att=44.596, acc=0.772, loss=46.498, backward_time=0.296, grad_norm=47.731, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.662e-04, train_time=1.310
+[gpub011:0/16] 2024-02-09 08:23:33,756 (trainer:762) INFO: 37epoch:train:2901-3000batch: iter_time=8.553e-05, forward_time=0.323, loss_ctc=51.448, loss_att=53.639, acc=0.746, loss=52.982, backward_time=0.310, grad_norm=46.245, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.662e-04, train_time=1.238
+[gpub011:0/16] 2024-02-09 08:25:54,735 (trainer:762) INFO: 37epoch:train:3001-3100batch: iter_time=8.626e-05, forward_time=0.377, loss_ctc=47.834, loss_att=49.110, acc=0.765, loss=48.727, backward_time=0.303, grad_norm=42.579, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.662e-04, train_time=1.410
+[gpub011:0/16] 2024-02-09 08:28:12,904 (trainer:762) INFO: 37epoch:train:3101-3200batch: iter_time=8.923e-05, forward_time=0.297, loss_ctc=48.583, loss_att=49.053, acc=0.752, loss=48.912, backward_time=0.300, grad_norm=44.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.662e-04, train_time=1.381
+[gpub011:0/16] 2024-02-09 08:30:05,553 (trainer:762) INFO: 37epoch:train:3201-3300batch: iter_time=8.191e-05, forward_time=0.308, loss_ctc=41.860, loss_att=35.054, acc=0.787, loss=37.096, backward_time=0.295, grad_norm=36.314, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.662e-04, train_time=1.126
+[gpub011:0/16] 2024-02-09 08:32:13,081 (trainer:762) INFO: 37epoch:train:3301-3400batch: iter_time=8.393e-05, forward_time=0.305, loss_ctc=45.233, loss_att=40.887, acc=0.761, loss=42.191, backward_time=0.308, grad_norm=43.258, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.662e-04, train_time=1.275
+[gpub011:0/16] 2024-02-09 08:34:36,304 (trainer:762) INFO: 37epoch:train:3401-3500batch: iter_time=2.265e-04, forward_time=0.376, loss_ctc=45.339, loss_att=48.842, acc=0.756, loss=47.791, backward_time=0.319, grad_norm=44.118, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.662e-04, train_time=1.432
+[gpub011:0/16] 2024-02-09 08:36:40,057 (trainer:762) INFO: 37epoch:train:3501-3600batch: iter_time=8.816e-05, forward_time=0.293, loss_ctc=48.798, loss_att=45.773, acc=0.769, loss=46.680, backward_time=0.300, grad_norm=40.502, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.661e-04, train_time=1.237
+[gpub011:0/16] 2024-02-09 08:38:47,997 (trainer:762) INFO: 37epoch:train:3601-3700batch: iter_time=8.998e-05, forward_time=0.291, loss_ctc=41.912, loss_att=38.253, acc=0.767, loss=39.350, backward_time=0.298, grad_norm=37.584, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.661e-04, train_time=1.279
+[gpub011:0/16] 2024-02-09 08:40:13,188 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-09 08:40:32,857 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 08:40:36,448 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f985695b850>)
+[gpub011:0/16] 2024-02-09 08:40:36,448 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-09 08:40:36,451 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 08:46:37,623 (trainer:762) INFO: 37epoch:train:3701-3800batch: iter_time=3.408, forward_time=0.362, loss_ctc=46.238, loss_att=44.690, acc=0.758, loss=45.154, backward_time=0.304, grad_norm=44.636, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.661e-04, train_time=4.696
+[gpub011:0/16] 2024-02-09 08:48:39,367 (trainer:762) INFO: 37epoch:train:3801-3900batch: iter_time=7.433e-05, forward_time=0.290, loss_ctc=47.497, loss_att=42.335, acc=0.752, loss=43.884, backward_time=0.295, grad_norm=41.922, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.661e-04, train_time=1.217
+[gpub011:0/16] 2024-02-09 08:50:43,155 (trainer:762) INFO: 37epoch:train:3901-4000batch: iter_time=2.623e-04, forward_time=0.298, loss_ctc=47.848, loss_att=51.147, acc=0.753, loss=50.157, backward_time=0.301, grad_norm=42.739, clip=100.000, loss_scale=4.232e+33, optim_step_time=0.095, optim0_lr0=1.661e-04, train_time=1.238
+[gpub011:0/16] 2024-02-09 08:53:00,716 (trainer:762) INFO: 37epoch:train:4001-4100batch: iter_time=8.222e-05, forward_time=0.351, loss_ctc=48.617, loss_att=45.903, acc=0.767, loss=46.717, backward_time=0.343, grad_norm=47.030, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.661e-04, train_time=1.375
+[gpub011:0/16] 2024-02-09 08:54:59,169 (trainer:762) INFO: 37epoch:train:4101-4200batch: iter_time=7.906e-05, forward_time=0.316, loss_ctc=52.302, loss_att=50.583, acc=0.753, loss=51.099, backward_time=0.298, grad_norm=47.179, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.660e-04, train_time=1.184
+[gpub011:0/16] 2024-02-09 08:57:12,739 (trainer:762) INFO: 37epoch:train:4201-4300batch: iter_time=8.015e-05, forward_time=0.295, loss_ctc=50.606, loss_att=50.637, acc=0.748, loss=50.628, backward_time=0.301, grad_norm=43.093, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.660e-04, train_time=1.336
+[gpub011:0/16] 2024-02-09 08:59:18,242 (trainer:762) INFO: 37epoch:train:4301-4400batch: iter_time=8.162e-05, forward_time=0.289, loss_ctc=45.787, loss_att=47.262, acc=0.735, loss=46.820, backward_time=0.297, grad_norm=46.972, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.660e-04, train_time=1.255
+[gpub011:0/16] 2024-02-09 09:01:24,698 (trainer:762) INFO: 37epoch:train:4401-4500batch: iter_time=8.983e-05, forward_time=0.387, loss_ctc=46.260, loss_att=44.020, acc=0.770, loss=44.692, backward_time=0.324, grad_norm=40.403, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.660e-04, train_time=1.264
+[gpub011:0/16] 2024-02-09 09:03:39,990 (trainer:762) INFO: 37epoch:train:4501-4600batch: iter_time=8.415e-05, forward_time=0.290, loss_ctc=47.757, loss_att=41.761, acc=0.760, loss=43.560, backward_time=0.296, grad_norm=43.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.660e-04, train_time=1.353
+[gpub011:0/16] 2024-02-09 09:05:44,416 (trainer:762) INFO: 37epoch:train:4601-4700batch: iter_time=8.799e-05, forward_time=0.292, loss_ctc=39.955, loss_att=38.910, acc=0.768, loss=39.223, backward_time=0.300, grad_norm=34.519, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.660e-04, train_time=1.244
+[gpub011:0/16] 2024-02-09 09:07:41,707 (trainer:762) INFO: 37epoch:train:4701-4800batch: iter_time=8.152e-05, forward_time=0.334, loss_ctc=50.857, loss_att=52.321, acc=0.757, loss=51.882, backward_time=0.328, grad_norm=47.552, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.660e-04, train_time=1.172
+[gpub011:0/16] 2024-02-09 09:09:58,196 (trainer:762) INFO: 37epoch:train:4801-4900batch: iter_time=8.107e-05, forward_time=0.331, loss_ctc=38.906, loss_att=35.950, acc=0.774, loss=36.837, backward_time=0.305, grad_norm=33.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.659e-04, train_time=1.365
+[gpub011:0/16] 2024-02-09 09:12:02,526 (trainer:762) INFO: 37epoch:train:4901-5000batch: iter_time=7.681e-05, forward_time=0.295, loss_ctc=46.627, loss_att=44.501, acc=0.740, loss=45.139, backward_time=0.296, grad_norm=42.112, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.659e-04, train_time=1.243
+[gpub011:0/16] 2024-02-09 09:12:22,670 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-09 09:12:42,028 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 09:12:45,504 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9818b6f7c0>)
+[gpub011:0/16] 2024-02-09 09:12:45,504 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-09 09:12:45,575 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 09:19:30,726 (trainer:762) INFO: 37epoch:train:5001-5100batch: iter_time=3.342, forward_time=0.301, loss_ctc=47.548, loss_att=42.812, acc=0.759, loss=44.233, backward_time=0.299, grad_norm=39.234, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.659e-04, train_time=4.482
+[gpub011:0/16] 2024-02-09 09:22:05,063 (trainer:762) INFO: 37epoch:train:5101-5200batch: iter_time=8.260e-05, forward_time=0.383, loss_ctc=47.276, loss_att=41.139, acc=0.765, loss=42.980, backward_time=0.314, grad_norm=40.498, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.659e-04, train_time=1.542
+[gpub011:0/16] 2024-02-09 09:23:58,742 (trainer:762) INFO: 37epoch:train:5201-5300batch: iter_time=9.032e-05, forward_time=0.292, loss_ctc=47.799, loss_att=50.015, acc=0.758, loss=49.350, backward_time=0.299, grad_norm=41.021, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.659e-04, train_time=1.137
+[gpub011:0/16] 2024-02-09 09:25:59,674 (trainer:762) INFO: 37epoch:train:5301-5400batch: iter_time=8.897e-05, forward_time=0.293, loss_ctc=49.709, loss_att=45.538, acc=0.759, loss=46.789, backward_time=0.300, grad_norm=47.195, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.659e-04, train_time=1.209
+[gpub011:0/16] 2024-02-09 09:28:29,568 (trainer:762) INFO: 37epoch:train:5401-5500batch: iter_time=8.849e-05, forward_time=0.301, loss_ctc=51.331, loss_att=52.028, acc=0.747, loss=51.819, backward_time=0.307, grad_norm=46.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.659e-04, train_time=1.499
+[gpub011:0/16] 2024-02-09 09:30:43,929 (trainer:762) INFO: 37epoch:train:5501-5600batch: iter_time=8.700e-05, forward_time=0.357, loss_ctc=47.467, loss_att=50.005, acc=0.746, loss=49.243, backward_time=0.318, grad_norm=43.395, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.658e-04, train_time=1.343
+[gpub011:0/16] 2024-02-09 09:32:35,948 (trainer:762) INFO: 37epoch:train:5601-5700batch: iter_time=8.844e-05, forward_time=0.291, loss_ctc=47.223, loss_att=47.346, acc=0.752, loss=47.309, backward_time=0.298, grad_norm=39.611, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.658e-04, train_time=1.120
+[gpub011:0/16] 2024-02-09 09:34:38,269 (trainer:762) INFO: 37epoch:train:5701-5800batch: iter_time=9.635e-05, forward_time=0.296, loss_ctc=41.771, loss_att=35.045, acc=0.782, loss=37.062, backward_time=0.299, grad_norm=35.599, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.658e-04, train_time=1.223
+[gpub011:0/16] 2024-02-09 09:37:01,703 (trainer:762) INFO: 37epoch:train:5801-5900batch: iter_time=9.482e-05, forward_time=0.373, loss_ctc=45.009, loss_att=41.587, acc=0.753, loss=42.614, backward_time=0.322, grad_norm=41.041, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.658e-04, train_time=1.435
+[gpub011:0/16] 2024-02-09 09:38:58,795 (trainer:762) INFO: 37epoch:train:5901-6000batch: iter_time=9.213e-05, forward_time=0.291, loss_ctc=45.005, loss_att=48.948, acc=0.750, loss=47.765, backward_time=0.298, grad_norm=43.709, clip=100.000, loss_scale=8.463e+33, optim_step_time=0.093, optim0_lr0=1.658e-04, train_time=1.170
+[gpub011:0/16] 2024-02-09 09:41:12,097 (trainer:762) INFO: 37epoch:train:6001-6100batch: iter_time=9.747e-05, forward_time=0.296, loss_ctc=48.105, loss_att=44.628, acc=0.765, loss=45.671, backward_time=0.300, grad_norm=42.238, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.658e-04, train_time=1.333
+[gpub011:0/16] 2024-02-09 09:43:17,490 (trainer:762) INFO: 37epoch:train:6101-6200batch: iter_time=1.052e-04, forward_time=0.290, loss_ctc=41.789, loss_att=38.982, acc=0.758, loss=39.824, backward_time=0.295, grad_norm=36.072, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.657e-04, train_time=1.254
+[gpub011:0/16] 2024-02-09 09:44:40,115 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-09 09:44:59,498 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 09:45:03,000 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f983fda7f10>)
+[gpub011:0/16] 2024-02-09 09:45:03,000 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-09 09:45:03,003 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 09:50:26,530 (trainer:762) INFO: 37epoch:train:6201-6300batch: iter_time=3.039, forward_time=0.365, loss_ctc=45.550, loss_att=41.465, acc=0.763, loss=42.691, backward_time=0.329, grad_norm=41.350, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.657e-04, train_time=4.290
+[gpub011:0/16] 2024-02-09 09:52:28,710 (trainer:762) INFO: 37epoch:train:6301-6400batch: iter_time=8.986e-05, forward_time=0.294, loss_ctc=47.084, loss_att=40.133, acc=0.757, loss=42.218, backward_time=0.298, grad_norm=42.658, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.657e-04, train_time=1.221
+[gpub011:0/16] 2024-02-09 09:54:42,262 (trainer:762) INFO: 37epoch:train:6401-6500batch: iter_time=9.318e-05, forward_time=0.299, loss_ctc=47.706, loss_att=49.761, acc=0.757, loss=49.144, backward_time=0.299, grad_norm=41.760, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.657e-04, train_time=1.336
+[gpub011:0/16] 2024-02-09 09:56:45,604 (trainer:762) INFO: 37epoch:train:6501-6600batch: iter_time=8.557e-05, forward_time=0.350, loss_ctc=48.253, loss_att=44.839, acc=0.769, loss=45.863, backward_time=0.338, grad_norm=46.096, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.657e-04, train_time=1.233
+[gpub011:0/16] 2024-02-09 09:58:54,167 (trainer:762) INFO: 37epoch:train:6601-6700batch: iter_time=9.278e-05, forward_time=0.295, loss_ctc=52.598, loss_att=50.330, acc=0.755, loss=51.010, backward_time=0.301, grad_norm=45.663, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.657e-04, train_time=1.286
+[gpub011:0/16] 2024-02-09 10:00:08,661 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 10:01:11,788 (trainer:762) INFO: 37epoch:train:6701-6800batch: iter_time=9.487e-05, forward_time=0.293, loss_ctc=50.438, loss_att=50.609, acc=0.749, loss=50.558, backward_time=0.301, grad_norm=42.429, clip=100.000, loss_scale=7.448e+33, optim_step_time=0.093, optim0_lr0=1.657e-04, train_time=1.376
+[gpub011:0/16] 2024-02-09 10:03:23,507 (trainer:762) INFO: 37epoch:train:6801-6900batch: iter_time=9.627e-05, forward_time=0.311, loss_ctc=45.156, loss_att=46.958, acc=0.735, loss=46.417, backward_time=0.326, grad_norm=43.041, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.656e-04, train_time=1.317
+[gpub011:0/16] 2024-02-09 10:05:36,237 (trainer:762) INFO: 37epoch:train:6901-7000batch: iter_time=8.674e-05, forward_time=0.323, loss_ctc=45.892, loss_att=43.419, acc=0.774, loss=44.161, backward_time=0.318, grad_norm=38.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.656e-04, train_time=1.327
+[gpub011:0/16] 2024-02-09 10:07:43,393 (trainer:762) INFO: 37epoch:train:7001-7100batch: iter_time=8.950e-05, forward_time=0.292, loss_ctc=47.104, loss_att=41.417, acc=0.760, loss=43.123, backward_time=0.300, grad_norm=41.808, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.656e-04, train_time=1.272
+[gpub011:0/16] 2024-02-09 10:09:47,045 (trainer:762) INFO: 37epoch:train:7101-7200batch: iter_time=9.048e-05, forward_time=0.291, loss_ctc=39.432, loss_att=38.531, acc=0.771, loss=38.802, backward_time=0.297, grad_norm=35.997, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.656e-04, train_time=1.236
+[gpub011:0/16] 2024-02-09 10:12:00,870 (trainer:762) INFO: 37epoch:train:7201-7300batch: iter_time=8.864e-05, forward_time=0.351, loss_ctc=50.130, loss_att=51.868, acc=0.761, loss=51.346, backward_time=0.334, grad_norm=46.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.656e-04, train_time=1.339
+[gpub011:0/16] 2024-02-09 10:13:53,045 (trainer:762) INFO: 37epoch:train:7301-7400batch: iter_time=8.768e-05, forward_time=0.292, loss_ctc=38.975, loss_att=35.350, acc=0.777, loss=36.437, backward_time=0.298, grad_norm=34.482, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.656e-04, train_time=1.121
+[gpub011:0/16] 2024-02-09 10:16:26,499 (trainer:762) INFO: 37epoch:train:7401-7500batch: iter_time=8.341e-05, forward_time=0.289, loss_ctc=46.269, loss_att=44.168, acc=0.740, loss=44.798, backward_time=0.295, grad_norm=41.968, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.655e-04, train_time=1.535
+[gpub011:0/16] 2024-02-09 10:16:46,527 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-09 10:17:05,750 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 10:17:09,746 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98237a88b0>)
+[gpub011:0/16] 2024-02-09 10:17:09,747 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-09 10:17:09,750 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 10:23:59,408 (trainer:762) INFO: 37epoch:train:7501-7600batch: iter_time=3.320, forward_time=0.359, loss_ctc=47.338, loss_att=45.757, acc=0.770, loss=46.231, backward_time=0.305, grad_norm=44.277, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.655e-04, train_time=4.529
+[gpub011:0/16] 2024-02-09 10:26:07,547 (trainer:762) INFO: 37epoch:train:7601-7700batch: iter_time=8.202e-05, forward_time=0.316, loss_ctc=46.940, loss_att=41.891, acc=0.768, loss=43.406, backward_time=0.322, grad_norm=42.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.655e-04, train_time=1.281
+[gpub011:0/16] 2024-02-09 10:28:25,867 (trainer:762) INFO: 37epoch:train:7701-7800batch: iter_time=8.467e-05, forward_time=0.293, loss_ctc=47.652, loss_att=50.675, acc=0.767, loss=49.768, backward_time=0.303, grad_norm=43.426, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.655e-04, train_time=1.382
+[gpub011:0/16] 2024-02-09 10:30:15,738 (trainer:762) INFO: 37epoch:train:7801-7900batch: iter_time=8.406e-05, forward_time=0.292, loss_ctc=49.272, loss_att=44.555, acc=0.774, loss=45.971, backward_time=0.300, grad_norm=44.957, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.655e-04, train_time=1.099
+[gpub011:0/16] 2024-02-09 10:32:49,873 (trainer:762) INFO: 37epoch:train:7901-8000batch: iter_time=9.106e-05, forward_time=0.370, loss_ctc=50.730, loss_att=53.228, acc=0.751, loss=52.479, backward_time=0.342, grad_norm=46.019, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.655e-04, train_time=1.541
+[gpub011:0/16] 2024-02-09 10:35:02,919 (trainer:762) INFO: 37epoch:train:8001-8100batch: iter_time=8.576e-05, forward_time=0.298, loss_ctc=46.896, loss_att=49.384, acc=0.766, loss=48.638, backward_time=0.303, grad_norm=42.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.655e-04, train_time=1.330
+[gpub011:0/16] 2024-02-09 10:37:05,042 (trainer:762) INFO: 37epoch:train:8101-8200batch: iter_time=8.301e-05, forward_time=0.292, loss_ctc=47.168, loss_att=47.688, acc=0.760, loss=47.532, backward_time=0.300, grad_norm=41.486, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.654e-04, train_time=1.220
+[gpub011:0/16] 2024-02-09 10:38:57,286 (trainer:762) INFO: 37epoch:train:8201-8300batch: iter_time=8.713e-05, forward_time=0.290, loss_ctc=41.628, loss_att=35.120, acc=0.787, loss=37.072, backward_time=0.299, grad_norm=37.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.654e-04, train_time=1.123
+[gpub011:0/16] 2024-02-09 10:41:45,595 (trainer:762) INFO: 37epoch:train:8301-8400batch: iter_time=8.729e-05, forward_time=0.355, loss_ctc=45.129, loss_att=40.835, acc=0.763, loss=42.123, backward_time=0.320, grad_norm=42.392, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.654e-04, train_time=1.683
+[gpub011:0/16] 2024-02-09 10:43:37,792 (trainer:762) INFO: 37epoch:train:8401-8500batch: iter_time=9.214e-05, forward_time=0.294, loss_ctc=43.949, loss_att=48.810, acc=0.759, loss=47.351, backward_time=0.307, grad_norm=43.434, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.654e-04, train_time=1.122
+[gpub011:0/16] 2024-02-09 10:45:36,179 (trainer:762) INFO: 37epoch:train:8501-8600batch: iter_time=8.637e-05, forward_time=0.294, loss_ctc=47.491, loss_att=45.311, acc=0.773, loss=45.965, backward_time=0.300, grad_norm=39.266, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.654e-04, train_time=1.184
+[gpub011:0/16] 2024-02-09 10:48:07,119 (trainer:762) INFO: 37epoch:train:8601-8700batch: iter_time=8.251e-05, forward_time=0.320, loss_ctc=41.151, loss_att=38.066, acc=0.769, loss=38.991, backward_time=0.323, grad_norm=36.265, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.654e-04, train_time=1.509
+[gpub011:0/16] 2024-02-09 10:49:28,898 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-09 10:49:48,214 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 10:49:51,876 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98242c2f20>)
+[gpub011:0/16] 2024-02-09 10:49:51,876 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-09 10:49:51,882 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 10:55:52,434 (trainer:762) INFO: 37epoch:train:8701-8800batch: iter_time=3.460, forward_time=0.324, loss_ctc=45.219, loss_att=41.480, acc=0.774, loss=42.602, backward_time=0.318, grad_norm=41.207, clip=100.000, loss_scale=8.100e+33, optim_step_time=0.094, optim0_lr0=1.654e-04, train_time=4.653
+[gpub011:0/16] 2024-02-09 10:58:24,972 (trainer:762) INFO: 37epoch:train:8801-8900batch: iter_time=7.951e-05, forward_time=0.289, loss_ctc=46.174, loss_att=40.187, acc=0.766, loss=41.983, backward_time=0.297, grad_norm=41.574, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.653e-04, train_time=1.525
+[gpub011:0/16] 2024-02-09 10:59:41,165 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 11:00:30,054 (trainer:762) INFO: 37epoch:train:8901-9000batch: iter_time=8.223e-05, forward_time=0.357, loss_ctc=47.877, loss_att=50.058, acc=0.768, loss=49.404, backward_time=0.339, grad_norm=43.363, clip=100.000, loss_scale=8.339e+33, optim_step_time=0.099, optim0_lr0=1.653e-04, train_time=1.251
+[gpub011:0/16] 2024-02-09 11:02:47,045 (trainer:762) INFO: 37epoch:train:9001-9100batch: iter_time=8.477e-05, forward_time=0.300, loss_ctc=48.263, loss_att=45.199, acc=0.778, loss=46.118, backward_time=0.300, grad_norm=45.989, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.653e-04, train_time=1.370
+[gpub011:0/16] 2024-02-09 11:05:00,840 (trainer:762) INFO: 37epoch:train:9101-9200batch: iter_time=8.616e-05, forward_time=0.294, loss_ctc=51.612, loss_att=50.351, acc=0.763, loss=50.729, backward_time=0.302, grad_norm=46.557, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.653e-04, train_time=1.338
+[gpub011:0/16] 2024-02-09 11:06:14,028 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 11:07:22,479 (trainer:762) INFO: 37epoch:train:9201-9300batch: iter_time=8.587e-05, forward_time=0.321, loss_ctc=49.871, loss_att=49.936, acc=0.760, loss=49.916, backward_time=0.327, grad_norm=40.031, clip=100.000, loss_scale=4.038e+33, optim_step_time=0.096, optim0_lr0=1.653e-04, train_time=1.416
+[gpub011:0/16] 2024-02-09 11:09:25,609 (trainer:762) INFO: 37epoch:train:9301-9400batch: iter_time=8.391e-05, forward_time=0.317, loss_ctc=44.956, loss_att=46.708, acc=0.750, loss=46.182, backward_time=0.330, grad_norm=43.777, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.653e-04, train_time=1.231
+[gpub011:0/16] 2024-02-09 11:11:36,088 (trainer:762) INFO: 37epoch:train:9401-9500batch: iter_time=8.621e-05, forward_time=0.292, loss_ctc=46.711, loss_att=43.918, acc=0.779, loss=44.756, backward_time=0.297, grad_norm=39.518, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.652e-04, train_time=1.305
+[gpub011:0/16] 2024-02-09 11:13:52,963 (trainer:762) INFO: 37epoch:train:9501-9600batch: iter_time=8.692e-05, forward_time=0.291, loss_ctc=47.337, loss_att=41.202, acc=0.772, loss=43.042, backward_time=0.299, grad_norm=40.703, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.652e-04, train_time=1.368
+[gpub011:0/16] 2024-02-09 11:16:07,485 (trainer:762) INFO: 37epoch:train:9601-9700batch: iter_time=8.833e-05, forward_time=0.396, loss_ctc=39.311, loss_att=38.439, acc=0.772, loss=38.700, backward_time=0.320, grad_norm=34.968, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.652e-04, train_time=1.345
+[gpub011:0/16] 2024-02-09 11:18:23,727 (trainer:762) INFO: 37epoch:train:9701-9800batch: iter_time=8.435e-05, forward_time=0.297, loss_ctc=49.538, loss_att=51.869, acc=0.766, loss=51.170, backward_time=0.305, grad_norm=45.534, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.652e-04, train_time=1.362
+[gpub011:0/16] 2024-02-09 11:20:41,809 (trainer:762) INFO: 37epoch:train:9801-9900batch: iter_time=9.338e-05, forward_time=0.296, loss_ctc=38.752, loss_att=35.861, acc=0.784, loss=36.729, backward_time=0.298, grad_norm=33.569, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.652e-04, train_time=1.381
+[gpub011:0/16] 2024-02-09 11:22:44,219 (trainer:762) INFO: 37epoch:train:9901-10000batch: iter_time=8.559e-05, forward_time=0.330, loss_ctc=46.414, loss_att=43.762, acc=0.753, loss=44.558, backward_time=0.313, grad_norm=42.087, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.652e-04, train_time=1.224
+[gpub011:0/16] 2024-02-09 11:23:04,326 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-09 11:23:23,516 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 11:23:27,081 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9809e79de0>)
+[gpub011:0/16] 2024-02-09 11:23:27,081 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-09 11:23:27,116 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 11:30:13,546 (trainer:762) INFO: 37epoch:train:10001-10100batch: iter_time=3.348, forward_time=0.328, loss_ctc=46.718, loss_att=41.608, acc=0.778, loss=43.141, backward_time=0.303, grad_norm=39.801, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.652e-04, train_time=4.493
+[gpub011:0/16] 2024-02-09 11:32:23,546 (trainer:762) INFO: 37epoch:train:10101-10200batch: iter_time=8.100e-05, forward_time=0.289, loss_ctc=46.518, loss_att=40.798, acc=0.771, loss=42.514, backward_time=0.296, grad_norm=41.527, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.651e-04, train_time=1.300
+[gpub011:0/16] 2024-02-09 11:34:50,286 (trainer:762) INFO: 37epoch:train:10201-10300batch: iter_time=8.133e-05, forward_time=0.315, loss_ctc=47.697, loss_att=50.068, acc=0.770, loss=49.357, backward_time=0.340, grad_norm=40.896, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.651e-04, train_time=1.467
+[gpub011:0/16] 2024-02-09 11:36:47,578 (trainer:762) INFO: 37epoch:train:10301-10400batch: iter_time=8.073e-05, forward_time=0.326, loss_ctc=49.654, loss_att=43.959, acc=0.775, loss=45.667, backward_time=0.311, grad_norm=46.348, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.651e-04, train_time=1.173
+[gpub011:0/16] 2024-02-09 11:39:02,822 (trainer:762) INFO: 37epoch:train:10401-10500batch: iter_time=9.687e-05, forward_time=0.294, loss_ctc=50.927, loss_att=53.210, acc=0.749, loss=52.525, backward_time=0.301, grad_norm=45.607, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.651e-04, train_time=1.352
+[gpub011:0/16] 2024-02-09 11:41:18,359 (trainer:762) INFO: 37epoch:train:10501-10600batch: iter_time=8.373e-05, forward_time=0.352, loss_ctc=46.995, loss_att=49.019, acc=0.768, loss=48.412, backward_time=0.314, grad_norm=40.163, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.651e-04, train_time=1.355
+[gpub011:0/16] 2024-02-09 11:43:31,892 (trainer:762) INFO: 37epoch:train:10601-10700batch: iter_time=8.311e-05, forward_time=0.337, loss_ctc=46.921, loss_att=48.090, acc=0.756, loss=47.739, backward_time=0.312, grad_norm=45.217, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.651e-04, train_time=1.335
+[gpub011:0/16] 2024-02-09 11:45:43,898 (trainer:762) INFO: 37epoch:train:10701-10800batch: iter_time=8.522e-05, forward_time=0.294, loss_ctc=41.618, loss_att=34.833, acc=0.789, loss=36.868, backward_time=0.295, grad_norm=38.585, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.651e-04, train_time=1.320
+[gpub011:0/16] 2024-02-09 11:48:03,070 (trainer:762) INFO: 37epoch:train:10801-10900batch: iter_time=8.504e-05, forward_time=0.322, loss_ctc=44.604, loss_att=40.556, acc=0.765, loss=41.770, backward_time=0.307, grad_norm=42.442, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.650e-04, train_time=1.391
+[gpub011:0/16] 2024-02-09 11:50:22,532 (trainer:762) INFO: 37epoch:train:10901-11000batch: iter_time=8.459e-05, forward_time=0.343, loss_ctc=43.960, loss_att=48.820, acc=0.758, loss=47.362, backward_time=0.326, grad_norm=42.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.650e-04, train_time=1.394
+[gpub011:0/16] 2024-02-09 11:52:28,848 (trainer:762) INFO: 37epoch:train:11001-11100batch: iter_time=8.356e-05, forward_time=0.292, loss_ctc=47.378, loss_att=44.822, acc=0.774, loss=45.589, backward_time=0.301, grad_norm=39.997, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.650e-04, train_time=1.263
+[gpub011:0/16] 2024-02-09 11:54:39,882 (trainer:762) INFO: 37epoch:train:11101-11200batch: iter_time=8.328e-05, forward_time=0.322, loss_ctc=41.114, loss_att=37.902, acc=0.770, loss=38.866, backward_time=0.309, grad_norm=37.085, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.650e-04, train_time=1.310
+[gpub011:0/16] 2024-02-09 11:55:55,538 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-09 11:56:15,562 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 11:56:19,206 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98327477c0>)
+[gpub011:0/16] 2024-02-09 11:56:19,206 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-09 11:56:19,210 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 12:02:11,600 (trainer:762) INFO: 37epoch:train:11201-11300batch: iter_time=3.302, forward_time=0.341, loss_ctc=44.598, loss_att=43.782, acc=0.763, loss=44.027, backward_time=0.308, grad_norm=43.751, clip=100.000, loss_scale=3.738e+33, optim_step_time=0.096, optim0_lr0=1.650e-04, train_time=4.517
+[gpub011:0/16] 2024-02-09 12:04:17,074 (trainer:762) INFO: 37epoch:train:11301-11400batch: iter_time=8.608e-05, forward_time=0.290, loss_ctc=46.013, loss_att=41.559, acc=0.756, loss=42.895, backward_time=0.297, grad_norm=41.657, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.650e-04, train_time=1.254
+[gpub011:0/16] 2024-02-09 12:06:29,284 (trainer:762) INFO: 37epoch:train:11401-11500batch: iter_time=8.704e-05, forward_time=0.327, loss_ctc=47.041, loss_att=50.318, acc=0.758, loss=49.335, backward_time=0.313, grad_norm=41.410, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.649e-04, train_time=1.322
+[gpub011:0/16] 2024-02-09 12:08:45,155 (trainer:762) INFO: 37epoch:train:11501-11600batch: iter_time=9.602e-05, forward_time=0.340, loss_ctc=47.365, loss_att=44.565, acc=0.772, loss=45.405, backward_time=0.316, grad_norm=47.592, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.649e-04, train_time=1.358
+[gpub011:0/16] 2024-02-09 12:11:09,464 (trainer:762) INFO: 37epoch:train:11601-11700batch: iter_time=4.163e-04, forward_time=0.292, loss_ctc=51.449, loss_att=50.043, acc=0.756, loss=50.465, backward_time=0.304, grad_norm=45.892, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.649e-04, train_time=1.443
+[gpub011:0/16] 2024-02-09 12:13:07,605 (trainer:762) INFO: 37epoch:train:11701-11800batch: iter_time=8.567e-05, forward_time=0.359, loss_ctc=50.407, loss_att=51.300, acc=0.747, loss=51.032, backward_time=0.315, grad_norm=43.006, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.649e-04, train_time=1.181
+[gpub011:0/16] 2024-02-09 12:15:21,403 (trainer:762) INFO: 37epoch:train:11801-11900batch: iter_time=8.161e-05, forward_time=0.318, loss_ctc=44.680, loss_att=47.433, acc=0.734, loss=46.607, backward_time=0.319, grad_norm=45.782, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.649e-04, train_time=1.338
+[gpub011:0/16] 2024-02-09 12:17:35,814 (trainer:762) INFO: 37epoch:train:11901-12000batch: iter_time=8.517e-05, forward_time=0.291, loss_ctc=46.155, loss_att=43.636, acc=0.775, loss=44.392, backward_time=0.297, grad_norm=38.956, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.649e-04, train_time=1.344
+[gpub011:0/16] 2024-02-09 12:19:43,285 (trainer:762) INFO: 37epoch:train:12001-12100batch: iter_time=8.413e-05, forward_time=0.299, loss_ctc=46.801, loss_att=41.082, acc=0.762, loss=42.798, backward_time=0.299, grad_norm=38.759, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.649e-04, train_time=1.274
+[gpub011:0/16] 2024-02-09 12:22:01,286 (trainer:762) INFO: 37epoch:train:12101-12200batch: iter_time=8.502e-05, forward_time=0.345, loss_ctc=39.326, loss_att=38.256, acc=0.771, loss=38.577, backward_time=0.343, grad_norm=35.491, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.648e-04, train_time=1.380
+[gpub011:0/16] 2024-02-09 12:24:13,993 (trainer:762) INFO: 37epoch:train:12201-12300batch: iter_time=8.631e-05, forward_time=0.302, loss_ctc=49.272, loss_att=51.514, acc=0.761, loss=50.841, backward_time=0.304, grad_norm=46.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.648e-04, train_time=1.327
+[gpub011:0/16] 2024-02-09 12:26:28,464 (trainer:762) INFO: 37epoch:train:12301-12400batch: iter_time=8.456e-05, forward_time=0.289, loss_ctc=38.668, loss_att=35.353, acc=0.777, loss=36.347, backward_time=0.296, grad_norm=32.972, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.648e-04, train_time=1.344
+[gpub011:0/16] 2024-02-09 12:28:33,766 (trainer:762) INFO: 37epoch:train:12401-12500batch: iter_time=8.695e-05, forward_time=0.320, loss_ctc=45.772, loss_att=44.193, acc=0.741, loss=44.667, backward_time=0.323, grad_norm=42.198, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.648e-04, train_time=1.253
+[gpub011:0/16] 2024-02-09 12:28:53,827 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-09 12:29:13,164 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 12:29:16,707 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f986b12ada0>)
+[gpub011:0/16] 2024-02-09 12:29:16,707 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-09 12:29:16,712 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 12:36:04,079 (trainer:762) INFO: 37epoch:train:12501-12600batch: iter_time=3.227, forward_time=0.339, loss_ctc=47.072, loss_att=43.999, acc=0.773, loss=44.921, backward_time=0.307, grad_norm=45.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.648e-04, train_time=4.503
+[gpub011:0/16] 2024-02-09 12:38:04,696 (trainer:762) INFO: 37epoch:train:12601-12700batch: iter_time=7.642e-05, forward_time=0.290, loss_ctc=46.539, loss_att=41.178, acc=0.770, loss=42.787, backward_time=0.298, grad_norm=41.554, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.648e-04, train_time=1.206
+[gpub011:0/16] 2024-02-09 12:40:36,187 (trainer:762) INFO: 37epoch:train:12701-12800batch: iter_time=8.273e-05, forward_time=0.351, loss_ctc=47.366, loss_att=50.469, acc=0.769, loss=49.538, backward_time=0.310, grad_norm=40.734, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.648e-04, train_time=1.515
+[gpub011:0/16] 2024-02-09 12:41:58,447 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 12:42:46,010 (trainer:762) INFO: 37epoch:train:12801-12900batch: iter_time=7.394e-05, forward_time=0.347, loss_ctc=48.741, loss_att=44.404, acc=0.777, loss=45.705, backward_time=0.326, grad_norm=50.129, clip=100.000, loss_scale=4.222e+33, optim_step_time=0.095, optim0_lr0=1.647e-04, train_time=1.298
+[gpub011:0/16] 2024-02-09 12:44:52,402 (trainer:762) INFO: 37epoch:train:12901-13000batch: iter_time=7.680e-05, forward_time=0.299, loss_ctc=50.503, loss_att=52.548, acc=0.752, loss=51.934, backward_time=0.301, grad_norm=46.080, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.647e-04, train_time=1.264
+[gpub011:0/16] 2024-02-09 12:47:16,710 (trainer:762) INFO: 37epoch:train:13001-13100batch: iter_time=7.687e-05, forward_time=0.338, loss_ctc=46.607, loss_att=48.704, acc=0.770, loss=48.075, backward_time=0.320, grad_norm=42.842, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.647e-04, train_time=1.443
+[gpub011:0/16] 2024-02-09 12:49:14,008 (trainer:762) INFO: 37epoch:train:13101-13200batch: iter_time=1.670e-04, forward_time=0.326, loss_ctc=46.800, loss_att=47.314, acc=0.761, loss=47.160, backward_time=0.334, grad_norm=41.149, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.647e-04, train_time=1.173
+[gpub011:0/16] 2024-02-09 12:51:29,002 (trainer:762) INFO: 37epoch:train:13201-13300batch: iter_time=1.713e-04, forward_time=0.292, loss_ctc=41.094, loss_att=34.719, acc=0.789, loss=36.632, backward_time=0.303, grad_norm=36.692, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.647e-04, train_time=1.350
+[gpub011:0/16] 2024-02-09 12:53:59,617 (trainer:762) INFO: 37epoch:train:13301-13400batch: iter_time=7.933e-05, forward_time=0.290, loss_ctc=44.703, loss_att=40.409, acc=0.765, loss=41.697, backward_time=0.295, grad_norm=40.940, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.647e-04, train_time=1.506
+[gpub011:0/16] 2024-02-09 12:56:06,510 (trainer:762) INFO: 37epoch:train:13401-13500batch: iter_time=7.765e-05, forward_time=0.370, loss_ctc=43.652, loss_att=48.493, acc=0.760, loss=47.040, backward_time=0.327, grad_norm=43.856, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.646e-04, train_time=1.269
+[gpub011:0/16] 2024-02-09 12:58:08,781 (trainer:762) INFO: 37epoch:train:13501-13600batch: iter_time=8.325e-05, forward_time=0.293, loss_ctc=47.479, loss_att=45.177, acc=0.773, loss=45.868, backward_time=0.299, grad_norm=40.179, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.646e-04, train_time=1.223
+[gpub011:0/16] 2024-02-09 13:00:31,237 (trainer:762) INFO: 37epoch:train:13601-13700batch: iter_time=3.599e-04, forward_time=0.291, loss_ctc=41.245, loss_att=38.123, acc=0.770, loss=39.059, backward_time=0.299, grad_norm=38.754, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.646e-04, train_time=1.424
+[gpub011:0/16] 2024-02-09 13:01:47,237 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-09 13:02:06,943 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 13:02:10,476 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97df9477c0>)
+[gpub011:0/16] 2024-02-09 13:02:10,476 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-09 13:02:10,545 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 13:07:56,403 (trainer:762) INFO: 37epoch:train:13701-13800batch: iter_time=3.259, forward_time=0.352, loss_ctc=44.650, loss_att=42.731, acc=0.767, loss=43.307, backward_time=0.308, grad_norm=42.008, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.646e-04, train_time=4.452
+[gpub011:0/16] 2024-02-09 13:10:02,236 (trainer:762) INFO: 37epoch:train:13801-13900batch: iter_time=8.755e-05, forward_time=0.291, loss_ctc=46.374, loss_att=41.314, acc=0.756, loss=42.832, backward_time=0.299, grad_norm=41.479, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.646e-04, train_time=1.258
+[gpub011:0/16] 2024-02-09 13:12:07,740 (trainer:762) INFO: 37epoch:train:13901-14000batch: iter_time=8.050e-05, forward_time=0.292, loss_ctc=47.315, loss_att=50.112, acc=0.758, loss=49.273, backward_time=0.300, grad_norm=42.388, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.646e-04, train_time=1.254
+[gpub011:0/16] 2024-02-09 13:14:21,500 (trainer:762) INFO: 37epoch:train:14001-14100batch: iter_time=8.163e-05, forward_time=0.383, loss_ctc=48.013, loss_att=44.493, acc=0.771, loss=45.549, backward_time=0.322, grad_norm=47.361, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.646e-04, train_time=1.337
+[gpub011:0/16] 2024-02-09 13:16:42,945 (trainer:762) INFO: 37epoch:train:14101-14200batch: iter_time=9.263e-05, forward_time=0.294, loss_ctc=51.464, loss_att=50.032, acc=0.756, loss=50.462, backward_time=0.300, grad_norm=46.660, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.645e-04, train_time=1.415
+[gpub011:0/16] 2024-02-09 13:19:06,793 (trainer:762) INFO: 37epoch:train:14201-14300batch: iter_time=9.265e-05, forward_time=0.353, loss_ctc=49.890, loss_att=50.550, acc=0.749, loss=50.352, backward_time=0.362, grad_norm=43.032, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.645e-04, train_time=1.438
+[gpub011:0/16] 2024-02-09 13:21:20,406 (trainer:762) INFO: 37epoch:train:14301-14400batch: iter_time=8.632e-05, forward_time=0.291, loss_ctc=44.120, loss_att=46.828, acc=0.736, loss=46.016, backward_time=0.298, grad_norm=46.418, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.645e-04, train_time=1.336
+[gpub011:0/16] 2024-02-09 13:23:37,568 (trainer:762) INFO: 37epoch:train:14401-14500batch: iter_time=8.927e-05, forward_time=0.381, loss_ctc=45.511, loss_att=42.906, acc=0.777, loss=43.688, backward_time=0.344, grad_norm=38.505, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.645e-04, train_time=1.371
+[gpub011:0/16] 2024-02-09 13:25:45,563 (trainer:762) INFO: 37epoch:train:14501-14600batch: iter_time=8.423e-04, forward_time=0.294, loss_ctc=46.844, loss_att=41.249, acc=0.761, loss=42.927, backward_time=0.300, grad_norm=41.945, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.645e-04, train_time=1.280
+[gpub011:0/16] 2024-02-09 13:28:08,605 (trainer:762) INFO: 37epoch:train:14601-14700batch: iter_time=9.315e-05, forward_time=0.326, loss_ctc=39.204, loss_att=38.154, acc=0.772, loss=38.469, backward_time=0.331, grad_norm=34.412, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.645e-04, train_time=1.431
+[gpub011:0/16] 2024-02-09 13:30:13,283 (trainer:762) INFO: 37epoch:train:14701-14800batch: iter_time=8.447e-05, forward_time=0.297, loss_ctc=48.716, loss_att=51.240, acc=0.762, loss=50.483, backward_time=0.305, grad_norm=46.512, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.645e-04, train_time=1.247
+[gpub011:0/16] 2024-02-09 13:32:18,316 (trainer:762) INFO: 37epoch:train:14801-14900batch: iter_time=9.178e-05, forward_time=0.288, loss_ctc=38.397, loss_att=35.201, acc=0.778, loss=36.160, backward_time=0.294, grad_norm=33.286, clip=100.000, loss_scale=3.557e+33, optim_step_time=0.093, optim0_lr0=1.644e-04, train_time=1.250
+[gpub011:0/16] 2024-02-09 13:34:43,636 (trainer:762) INFO: 37epoch:train:14901-15000batch: iter_time=9.125e-05, forward_time=0.388, loss_ctc=45.949, loss_att=44.229, acc=0.741, loss=44.745, backward_time=0.323, grad_norm=42.534, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.644e-04, train_time=1.453
+[gpub011:0/16] 2024-02-09 14:13:09,591 (trainer:361) INFO: 37epoch results: [train] iter_time=0.259, forward_time=0.317, loss_ctc=46.616, loss_att=44.903, acc=0.762, loss=45.417, backward_time=0.309, grad_norm=42.185, clip=100.000, loss_scale=4.712e+33, optim_step_time=0.095, optim0_lr0=1.655e-04, train_time=1.564, time=6 hours, 31 minutes and 30.46 seconds, total_count=585000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=34.864, cer_ctc=0.179, loss_att=37.953, acc=0.690, cer=0.316, wer=0.989, loss=37.026, time=38 minutes and 1.9 seconds, total_count=182169, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-09 14:13:18,768 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-09 14:13:18,888 (trainer:290) INFO: 38/45epoch started. Estimated time to finish: 2 days, 10 hours and 48 minutes
+[gpub011:0/16] 2024-02-09 14:13:18,898 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-09 14:13:37,064 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 14:13:40,487 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8f91c677f0>)
+[gpub011:0/16] 2024-02-09 14:13:40,487 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-09 14:13:40,491 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 14:20:27,005 (trainer:762) INFO: 38epoch:train:1-100batch: iter_time=3.085, forward_time=0.335, loss_ctc=50.711, loss_att=52.060, acc=0.739, loss=51.656, backward_time=0.316, grad_norm=46.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.644e-04, train_time=4.280
+[gpub011:0/16] 2024-02-09 14:22:23,885 (trainer:762) INFO: 38epoch:train:101-200batch: iter_time=1.067e-04, forward_time=0.310, loss_ctc=39.529, loss_att=41.018, acc=0.767, loss=40.571, backward_time=0.299, grad_norm=37.999, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.644e-04, train_time=1.169
+[gpub011:0/16] 2024-02-09 14:24:52,377 (trainer:762) INFO: 38epoch:train:201-300batch: iter_time=9.012e-05, forward_time=0.335, loss_ctc=50.934, loss_att=55.565, acc=0.742, loss=54.175, backward_time=0.325, grad_norm=47.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.644e-04, train_time=1.484
+[gpub011:0/16] 2024-02-09 14:26:55,294 (trainer:762) INFO: 38epoch:train:301-400batch: iter_time=9.441e-05, forward_time=0.320, loss_ctc=54.399, loss_att=52.075, acc=0.752, loss=52.772, backward_time=0.317, grad_norm=105.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.644e-04, train_time=1.229
+[gpub011:0/16] 2024-02-09 14:29:06,668 (trainer:762) INFO: 38epoch:train:401-500batch: iter_time=8.240e-05, forward_time=0.314, loss_ctc=49.023, loss_att=49.679, acc=0.761, loss=49.482, backward_time=0.329, grad_norm=40.255, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.644e-04, train_time=1.314
+[gpub011:0/16] 2024-02-09 14:31:18,659 (trainer:762) INFO: 38epoch:train:501-600batch: iter_time=9.154e-05, forward_time=0.340, loss_ctc=42.695, loss_att=42.274, acc=0.769, loss=42.400, backward_time=0.300, grad_norm=41.897, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.643e-04, train_time=1.320
+[gpub011:0/16] 2024-02-09 14:33:29,516 (trainer:762) INFO: 38epoch:train:601-700batch: iter_time=8.640e-05, forward_time=0.297, loss_ctc=61.925, loss_att=53.367, acc=0.743, loss=55.934, backward_time=0.304, grad_norm=59.956, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.643e-04, train_time=1.308
+[gpub011:0/16] 2024-02-09 14:35:43,972 (trainer:762) INFO: 38epoch:train:701-800batch: iter_time=8.098e-05, forward_time=0.323, loss_ctc=42.501, loss_att=39.940, acc=0.763, loss=40.708, backward_time=0.323, grad_norm=38.844, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.643e-04, train_time=1.345
+[gpub011:0/16] 2024-02-09 14:37:43,382 (trainer:762) INFO: 38epoch:train:801-900batch: iter_time=5.000e-04, forward_time=0.352, loss_ctc=58.417, loss_att=48.338, acc=0.771, loss=51.362, backward_time=0.308, grad_norm=57.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.643e-04, train_time=1.194
+[gpub011:0/16] 2024-02-09 14:40:03,155 (trainer:762) INFO: 38epoch:train:901-1000batch: iter_time=1.222e-04, forward_time=0.333, loss_ctc=55.732, loss_att=49.492, acc=0.758, loss=51.364, backward_time=0.320, grad_norm=55.106, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.643e-04, train_time=1.397
+[gpub011:0/16] 2024-02-09 14:42:21,918 (trainer:762) INFO: 38epoch:train:1001-1100batch: iter_time=8.742e-05, forward_time=0.324, loss_ctc=45.720, loss_att=46.849, acc=0.750, loss=46.510, backward_time=0.312, grad_norm=43.120, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.643e-04, train_time=1.388
+[gpub011:0/16] 2024-02-09 14:44:31,352 (trainer:762) INFO: 38epoch:train:1101-1200batch: iter_time=1.726e-04, forward_time=0.314, loss_ctc=51.363, loss_att=48.380, acc=0.753, loss=49.275, backward_time=0.324, grad_norm=46.239, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.642e-04, train_time=1.294
+[gpub011:0/16] 2024-02-09 14:45:50,253 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-09 14:46:09,827 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 14:46:13,432 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a64710220>)
+[gpub011:0/16] 2024-02-09 14:46:13,432 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-09 14:46:13,440 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 14:52:01,891 (trainer:762) INFO: 38epoch:train:1201-1300batch: iter_time=3.117, forward_time=0.314, loss_ctc=49.256, loss_att=47.942, acc=0.749, loss=48.336, backward_time=0.304, grad_norm=42.078, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.642e-04, train_time=4.505
+[gpub011:0/16] 2024-02-09 14:54:00,268 (trainer:762) INFO: 38epoch:train:1301-1400batch: iter_time=8.479e-05, forward_time=0.327, loss_ctc=44.285, loss_att=48.232, acc=0.739, loss=47.048, backward_time=0.296, grad_norm=41.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.642e-04, train_time=1.183
+[gpub011:0/16] 2024-02-09 14:56:07,197 (trainer:762) INFO: 38epoch:train:1401-1500batch: iter_time=1.620e-04, forward_time=0.310, loss_ctc=42.172, loss_att=46.034, acc=0.760, loss=44.875, backward_time=0.311, grad_norm=40.599, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.642e-04, train_time=1.269
+[gpub011:0/16] 2024-02-09 14:58:17,707 (trainer:762) INFO: 38epoch:train:1501-1600batch: iter_time=9.265e-05, forward_time=0.305, loss_ctc=51.058, loss_att=46.044, acc=0.762, loss=47.549, backward_time=0.312, grad_norm=48.428, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.642e-04, train_time=1.305
+[gpub011:0/16] 2024-02-09 15:00:29,489 (trainer:762) INFO: 38epoch:train:1601-1700batch: iter_time=8.928e-05, forward_time=0.306, loss_ctc=50.236, loss_att=51.095, acc=0.740, loss=50.837, backward_time=0.297, grad_norm=45.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.642e-04, train_time=1.318
+[gpub011:0/16] 2024-02-09 15:02:26,242 (trainer:762) INFO: 38epoch:train:1701-1800batch: iter_time=9.377e-05, forward_time=0.326, loss_ctc=44.771, loss_att=44.994, acc=0.764, loss=44.927, backward_time=0.307, grad_norm=38.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.642e-04, train_time=1.167
+[gpub011:0/16] 2024-02-09 15:04:29,546 (trainer:762) INFO: 38epoch:train:1801-1900batch: iter_time=9.636e-05, forward_time=0.341, loss_ctc=46.195, loss_att=41.645, acc=0.760, loss=43.010, backward_time=0.317, grad_norm=44.370, clip=100.000, loss_scale=7.113e+33, optim_step_time=0.098, optim0_lr0=1.641e-04, train_time=1.233
+[gpub011:0/16] 2024-02-09 15:04:43,421 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 15:06:42,001 (trainer:762) INFO: 38epoch:train:1901-2000batch: iter_time=8.999e-05, forward_time=0.290, loss_ctc=54.430, loss_att=48.894, acc=0.748, loss=50.555, backward_time=0.294, grad_norm=56.261, clip=100.000, loss_scale=5.507e+33, optim_step_time=0.093, optim0_lr0=1.641e-04, train_time=1.325
+[gpub011:0/16] 2024-02-09 15:08:48,382 (trainer:762) INFO: 38epoch:train:2001-2100batch: iter_time=9.099e-05, forward_time=0.324, loss_ctc=52.708, loss_att=41.329, acc=0.769, loss=44.743, backward_time=0.301, grad_norm=46.918, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.641e-04, train_time=1.264
+[gpub011:0/16] 2024-02-09 15:10:52,687 (trainer:762) INFO: 38epoch:train:2101-2200batch: iter_time=9.298e-05, forward_time=0.328, loss_ctc=55.508, loss_att=52.514, acc=0.747, loss=53.412, backward_time=0.328, grad_norm=54.997, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.641e-04, train_time=1.242
+[gpub011:0/16] 2024-02-09 15:12:58,874 (trainer:762) INFO: 38epoch:train:2201-2300batch: iter_time=1.028e-04, forward_time=0.289, loss_ctc=43.394, loss_att=37.375, acc=0.766, loss=39.181, backward_time=0.293, grad_norm=39.296, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.641e-04, train_time=1.262
+[gpub011:0/16] 2024-02-09 15:15:04,592 (trainer:762) INFO: 38epoch:train:2301-2400batch: iter_time=8.276e-05, forward_time=0.325, loss_ctc=47.701, loss_att=47.883, acc=0.739, loss=47.828, backward_time=0.299, grad_norm=44.932, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.641e-04, train_time=1.257
+[gpub011:0/16] 2024-02-09 15:17:03,692 (trainer:762) INFO: 38epoch:train:2401-2500batch: iter_time=9.112e-05, forward_time=0.305, loss_ctc=48.949, loss_att=46.862, acc=0.761, loss=47.488, backward_time=0.301, grad_norm=43.782, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.641e-04, train_time=1.191
+[gpub011:0/16] 2024-02-09 15:17:23,721 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-09 15:17:43,040 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 15:17:46,847 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fbe586470>)
+[gpub011:0/16] 2024-02-09 15:17:46,848 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-09 15:17:46,851 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 15:24:12,968 (trainer:762) INFO: 38epoch:train:2501-2600batch: iter_time=3.032, forward_time=0.312, loss_ctc=49.458, loss_att=48.455, acc=0.740, loss=48.756, backward_time=0.305, grad_norm=43.699, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.640e-04, train_time=4.292
+[gpub011:0/16] 2024-02-09 15:26:09,883 (trainer:762) INFO: 38epoch:train:2601-2700batch: iter_time=8.713e-05, forward_time=0.325, loss_ctc=38.746, loss_att=38.090, acc=0.766, loss=38.286, backward_time=0.297, grad_norm=36.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.640e-04, train_time=1.169
+[gpub011:0/16] 2024-02-09 15:28:33,570 (trainer:762) INFO: 38epoch:train:2701-2800batch: iter_time=9.269e-05, forward_time=0.296, loss_ctc=49.484, loss_att=53.498, acc=0.742, loss=52.294, backward_time=0.299, grad_norm=43.950, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.640e-04, train_time=1.437
+[gpub011:0/16] 2024-02-09 15:30:47,426 (trainer:762) INFO: 38epoch:train:2801-2900batch: iter_time=9.068e-05, forward_time=0.350, loss_ctc=52.890, loss_att=48.943, acc=0.753, loss=50.127, backward_time=0.316, grad_norm=48.281, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.640e-04, train_time=1.338
+[gpub011:0/16] 2024-02-09 15:32:50,896 (trainer:762) INFO: 38epoch:train:2901-3000batch: iter_time=8.799e-05, forward_time=0.304, loss_ctc=47.596, loss_att=47.651, acc=0.759, loss=47.635, backward_time=0.300, grad_norm=38.406, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.640e-04, train_time=1.235
+[gpub011:0/16] 2024-02-09 15:34:58,555 (trainer:762) INFO: 38epoch:train:3001-3100batch: iter_time=9.639e-05, forward_time=0.323, loss_ctc=42.706, loss_att=41.886, acc=0.767, loss=42.132, backward_time=0.315, grad_norm=40.963, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.640e-04, train_time=1.276
+[gpub011:0/16] 2024-02-09 15:37:05,492 (trainer:762) INFO: 38epoch:train:3101-3200batch: iter_time=9.005e-05, forward_time=0.331, loss_ctc=55.428, loss_att=49.166, acc=0.742, loss=51.045, backward_time=0.315, grad_norm=56.830, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.640e-04, train_time=1.268
+[gpub011:0/16] 2024-02-09 15:37:35,893 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 15:39:26,084 (trainer:762) INFO: 38epoch:train:3201-3300batch: iter_time=5.322e-04, forward_time=0.344, loss_ctc=41.899, loss_att=39.056, acc=0.763, loss=39.909, backward_time=0.298, grad_norm=38.370, clip=100.000, loss_scale=2.885e+33, optim_step_time=0.093, optim0_lr0=1.639e-04, train_time=1.407
+[gpub011:0/16] 2024-02-09 15:41:25,969 (trainer:762) INFO: 38epoch:train:3301-3400batch: iter_time=8.098e-05, forward_time=0.302, loss_ctc=53.344, loss_att=45.212, acc=0.770, loss=47.652, backward_time=0.302, grad_norm=54.793, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.639e-04, train_time=1.199
+[gpub011:0/16] 2024-02-09 15:43:41,244 (trainer:762) INFO: 38epoch:train:3401-3500batch: iter_time=7.968e-05, forward_time=0.351, loss_ctc=52.904, loss_att=46.650, acc=0.760, loss=48.526, backward_time=0.342, grad_norm=51.861, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.639e-04, train_time=1.353
+[gpub011:0/16] 2024-02-09 15:45:57,995 (trainer:762) INFO: 38epoch:train:3501-3600batch: iter_time=9.039e-05, forward_time=0.333, loss_ctc=44.854, loss_att=45.794, acc=0.740, loss=45.512, backward_time=0.295, grad_norm=43.716, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.639e-04, train_time=1.367
+[gpub011:0/16] 2024-02-09 15:47:54,375 (trainer:762) INFO: 38epoch:train:3601-3700batch: iter_time=9.177e-05, forward_time=0.298, loss_ctc=49.612, loss_att=47.022, acc=0.748, loss=47.799, backward_time=0.302, grad_norm=44.762, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.639e-04, train_time=1.164
+[gpub011:0/16] 2024-02-09 15:49:09,838 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-09 15:49:29,202 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 15:49:32,780 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97e21b3ee0>)
+[gpub011:0/16] 2024-02-09 15:49:32,780 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-09 15:49:32,784 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 15:55:24,389 (trainer:762) INFO: 38epoch:train:3701-3800batch: iter_time=3.271, forward_time=0.335, loss_ctc=49.038, loss_att=46.913, acc=0.756, loss=47.551, backward_time=0.307, grad_norm=40.087, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.639e-04, train_time=4.500
+[gpub011:0/16] 2024-02-09 15:57:30,845 (trainer:762) INFO: 38epoch:train:3801-3900batch: iter_time=8.239e-05, forward_time=0.337, loss_ctc=43.727, loss_att=48.240, acc=0.752, loss=46.886, backward_time=0.300, grad_norm=41.496, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.639e-04, train_time=1.264
+[gpub011:0/16] 2024-02-09 15:59:44,123 (trainer:762) INFO: 38epoch:train:3901-4000batch: iter_time=8.366e-05, forward_time=0.292, loss_ctc=41.570, loss_att=47.429, acc=0.762, loss=45.671, backward_time=0.297, grad_norm=39.264, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.638e-04, train_time=1.333
+[gpub011:0/16] 2024-02-09 16:02:00,352 (trainer:762) INFO: 38epoch:train:4001-4100batch: iter_time=9.395e-05, forward_time=0.342, loss_ctc=49.914, loss_att=48.827, acc=0.766, loss=49.153, backward_time=0.322, grad_norm=45.738, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.105, optim0_lr0=1.638e-04, train_time=1.362
+[gpub011:0/16] 2024-02-09 16:05:05,732 (trainer:762) INFO: 38epoch:train:4101-4200batch: iter_time=0.168, forward_time=0.364, loss_ctc=50.203, loss_att=51.515, acc=0.751, loss=51.121, backward_time=0.301, grad_norm=45.164, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.638e-04, train_time=1.853
+[gpub011:0/16] 2024-02-09 16:08:44,826 (trainer:762) INFO: 38epoch:train:4201-4300batch: iter_time=0.391, forward_time=0.290, loss_ctc=44.241, loss_att=44.823, acc=0.773, loss=44.648, backward_time=0.298, grad_norm=38.896, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.638e-04, train_time=2.191
+[gpub011:0/16] 2024-02-09 16:10:55,141 (trainer:762) INFO: 38epoch:train:4301-4400batch: iter_time=8.784e-05, forward_time=0.291, loss_ctc=46.079, loss_att=41.869, acc=0.767, loss=43.132, backward_time=0.301, grad_norm=44.128, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.638e-04, train_time=1.304
+[gpub011:0/16] 2024-02-09 16:14:32,788 (trainer:762) INFO: 38epoch:train:4401-4500batch: iter_time=0.066, forward_time=0.345, loss_ctc=51.686, loss_att=47.962, acc=0.758, loss=49.079, backward_time=0.335, grad_norm=51.167, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.638e-04, train_time=2.176
+[gpub011:0/16] 2024-02-09 16:16:43,874 (trainer:762) INFO: 38epoch:train:4501-4600batch: iter_time=9.928e-05, forward_time=0.322, loss_ctc=51.569, loss_att=42.327, acc=0.771, loss=45.100, backward_time=0.299, grad_norm=49.609, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.637e-04, train_time=1.310
+[gpub011:0/16] 2024-02-09 16:18:57,330 (trainer:762) INFO: 38epoch:train:4601-4700batch: iter_time=9.673e-05, forward_time=0.295, loss_ctc=54.865, loss_att=54.909, acc=0.753, loss=54.895, backward_time=0.301, grad_norm=55.574, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.637e-04, train_time=1.334
+[gpub011:0/16] 2024-02-09 16:22:44,813 (trainer:762) INFO: 38epoch:train:4701-4800batch: iter_time=9.873e-05, forward_time=0.408, loss_ctc=43.067, loss_att=39.059, acc=0.770, loss=40.262, backward_time=0.372, grad_norm=38.875, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=1.637e-04, train_time=2.275
+[gpub011:0/16] 2024-02-09 16:24:55,701 (trainer:762) INFO: 38epoch:train:4801-4900batch: iter_time=8.834e-05, forward_time=0.329, loss_ctc=47.670, loss_att=47.764, acc=0.753, loss=47.736, backward_time=0.305, grad_norm=45.765, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.637e-04, train_time=1.309
+[gpub011:0/16] 2024-02-09 16:26:58,983 (trainer:762) INFO: 38epoch:train:4901-5000batch: iter_time=8.183e-05, forward_time=0.294, loss_ctc=48.772, loss_att=47.362, acc=0.768, loss=47.785, backward_time=0.300, grad_norm=41.241, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.637e-04, train_time=1.232
+[gpub011:0/16] 2024-02-09 16:27:19,063 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-09 16:27:38,564 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 16:27:42,153 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9818b177c0>)
+[gpub011:0/16] 2024-02-09 16:27:42,153 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-09 16:27:42,157 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 16:34:20,670 (trainer:762) INFO: 38epoch:train:5001-5100batch: iter_time=3.186, forward_time=0.352, loss_ctc=49.152, loss_att=50.267, acc=0.734, loss=49.933, backward_time=0.312, grad_norm=45.593, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.637e-04, train_time=4.417
+[gpub011:0/16] 2024-02-09 16:36:29,860 (trainer:762) INFO: 38epoch:train:5101-5200batch: iter_time=7.687e-05, forward_time=0.341, loss_ctc=38.292, loss_att=37.752, acc=0.770, loss=37.914, backward_time=0.297, grad_norm=35.883, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.637e-04, train_time=1.292
+[gpub011:0/16] 2024-02-09 16:38:42,134 (trainer:762) INFO: 38epoch:train:5201-5300batch: iter_time=8.207e-05, forward_time=0.292, loss_ctc=48.734, loss_att=52.724, acc=0.745, loss=51.527, backward_time=0.298, grad_norm=46.544, clip=100.000, loss_scale=4.881e+33, optim_step_time=0.093, optim0_lr0=1.636e-04, train_time=1.323
+[gpub011:0/16] 2024-02-09 16:40:46,920 (trainer:762) INFO: 38epoch:train:5301-5400batch: iter_time=8.249e-05, forward_time=0.297, loss_ctc=52.421, loss_att=49.502, acc=0.751, loss=50.378, backward_time=0.303, grad_norm=46.811, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.636e-04, train_time=1.247
+[gpub011:0/16] 2024-02-09 16:43:01,153 (trainer:762) INFO: 38epoch:train:5401-5500batch: iter_time=8.476e-05, forward_time=0.332, loss_ctc=47.013, loss_att=48.307, acc=0.757, loss=47.919, backward_time=0.343, grad_norm=39.753, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.636e-04, train_time=1.342
+[gpub011:0/16] 2024-02-09 16:45:19,017 (trainer:762) INFO: 38epoch:train:5501-5600batch: iter_time=9.205e-05, forward_time=0.334, loss_ctc=42.241, loss_att=41.661, acc=0.771, loss=41.835, backward_time=0.299, grad_norm=40.473, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.636e-04, train_time=1.379
+[gpub011:0/16] 2024-02-09 16:47:15,234 (trainer:762) INFO: 38epoch:train:5601-5700batch: iter_time=8.709e-05, forward_time=0.290, loss_ctc=54.084, loss_att=49.098, acc=0.744, loss=50.594, backward_time=0.297, grad_norm=57.659, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.636e-04, train_time=1.162
+[gpub011:0/16] 2024-02-09 16:49:28,248 (trainer:762) INFO: 38epoch:train:5701-5800batch: iter_time=2.506e-04, forward_time=0.354, loss_ctc=41.779, loss_att=38.401, acc=0.766, loss=39.415, backward_time=0.325, grad_norm=36.934, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.636e-04, train_time=1.329
+[gpub011:0/16] 2024-02-09 16:51:58,188 (trainer:762) INFO: 38epoch:train:5801-5900batch: iter_time=8.959e-05, forward_time=0.314, loss_ctc=52.517, loss_att=44.544, acc=0.773, loss=46.936, backward_time=0.317, grad_norm=51.083, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.636e-04, train_time=1.500
+[gpub011:0/16] 2024-02-09 16:54:07,352 (trainer:762) INFO: 38epoch:train:5901-6000batch: iter_time=8.947e-05, forward_time=0.291, loss_ctc=52.221, loss_att=46.199, acc=0.762, loss=48.006, backward_time=0.298, grad_norm=55.090, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.635e-04, train_time=1.292
+[gpub011:0/16] 2024-02-09 16:56:18,009 (trainer:762) INFO: 38epoch:train:6001-6100batch: iter_time=0.001, forward_time=0.298, loss_ctc=44.604, loss_att=45.467, acc=0.744, loss=45.208, backward_time=0.296, grad_norm=43.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.635e-04, train_time=1.306
+[gpub011:0/16] 2024-02-09 16:59:00,690 (trainer:762) INFO: 38epoch:train:6101-6200batch: iter_time=8.862e-05, forward_time=0.418, loss_ctc=49.698, loss_att=47.372, acc=0.748, loss=48.070, backward_time=0.316, grad_norm=46.566, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.635e-04, train_time=1.626
+[gpub011:0/16] 2024-02-09 17:00:15,329 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-09 17:00:35,265 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 17:00:39,220 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9820e2f0a0>)
+[gpub011:0/16] 2024-02-09 17:00:39,220 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-09 17:00:39,223 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 17:07:04,015 (trainer:762) INFO: 38epoch:train:6201-6300batch: iter_time=3.685, forward_time=0.290, loss_ctc=48.670, loss_att=45.921, acc=0.753, loss=46.746, backward_time=0.298, grad_norm=42.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.635e-04, train_time=4.834
+[gpub011:0/16] 2024-02-09 17:09:15,198 (trainer:762) INFO: 38epoch:train:6301-6400batch: iter_time=8.381e-05, forward_time=0.290, loss_ctc=43.541, loss_att=45.804, acc=0.748, loss=45.125, backward_time=0.296, grad_norm=40.530, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.635e-04, train_time=1.312
+[gpub011:0/16] 2024-02-09 17:11:40,964 (trainer:762) INFO: 38epoch:train:6401-6500batch: iter_time=8.560e-05, forward_time=0.415, loss_ctc=41.491, loss_att=45.439, acc=0.764, loss=44.255, backward_time=0.323, grad_norm=39.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.635e-04, train_time=1.457
+[gpub011:0/16] 2024-02-09 17:13:45,312 (trainer:762) INFO: 38epoch:train:6501-6600batch: iter_time=8.239e-05, forward_time=0.292, loss_ctc=49.304, loss_att=44.639, acc=0.769, loss=46.039, backward_time=0.299, grad_norm=45.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.635e-04, train_time=1.243
+[gpub011:0/16] 2024-02-09 17:16:23,549 (trainer:762) INFO: 38epoch:train:6601-6700batch: iter_time=2.886e-04, forward_time=0.350, loss_ctc=49.795, loss_att=49.608, acc=0.744, loss=49.664, backward_time=0.384, grad_norm=43.854, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.634e-04, train_time=1.583
+[gpub011:0/16] 2024-02-09 17:18:16,245 (trainer:762) INFO: 38epoch:train:6701-6800batch: iter_time=8.510e-05, forward_time=0.290, loss_ctc=43.432, loss_att=44.354, acc=0.769, loss=44.077, backward_time=0.298, grad_norm=41.570, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.634e-04, train_time=1.126
+[gpub011:0/16] 2024-02-09 17:20:28,757 (trainer:762) INFO: 38epoch:train:6801-6900batch: iter_time=8.384e-05, forward_time=0.289, loss_ctc=45.558, loss_att=40.399, acc=0.765, loss=41.946, backward_time=0.296, grad_norm=40.886, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.634e-04, train_time=1.325
+[gpub011:0/16] 2024-02-09 17:22:43,880 (trainer:762) INFO: 38epoch:train:6901-7000batch: iter_time=9.079e-04, forward_time=0.302, loss_ctc=51.293, loss_att=49.032, acc=0.753, loss=49.711, backward_time=0.296, grad_norm=55.155, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.634e-04, train_time=1.351
+[gpub011:0/16] 2024-02-09 17:24:58,989 (trainer:762) INFO: 38epoch:train:7001-7100batch: iter_time=8.427e-05, forward_time=0.361, loss_ctc=51.655, loss_att=41.537, acc=0.772, loss=44.573, backward_time=0.352, grad_norm=49.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.634e-04, train_time=1.351
+[gpub011:0/16] 2024-02-09 17:27:20,203 (trainer:762) INFO: 38epoch:train:7101-7200batch: iter_time=8.422e-05, forward_time=0.294, loss_ctc=53.391, loss_att=52.400, acc=0.749, loss=52.697, backward_time=0.300, grad_norm=55.303, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.634e-04, train_time=1.412
+[gpub011:0/16] 2024-02-09 17:29:30,969 (trainer:762) INFO: 38epoch:train:7201-7300batch: iter_time=2.710e-04, forward_time=0.291, loss_ctc=42.495, loss_att=36.671, acc=0.772, loss=38.418, backward_time=0.299, grad_norm=39.632, clip=100.000, loss_scale=9.762e+33, optim_step_time=0.093, optim0_lr0=1.634e-04, train_time=1.307
+[gpub011:0/16] 2024-02-09 17:31:52,762 (trainer:762) INFO: 38epoch:train:7301-7400batch: iter_time=1.447e-04, forward_time=0.425, loss_ctc=47.240, loss_att=47.370, acc=0.743, loss=47.331, backward_time=0.329, grad_norm=45.636, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.633e-04, train_time=1.417
+[gpub011:0/16] 2024-02-09 17:34:14,600 (trainer:762) INFO: 38epoch:train:7401-7500batch: iter_time=8.681e-05, forward_time=0.289, loss_ctc=48.514, loss_att=46.311, acc=0.764, loss=46.972, backward_time=0.296, grad_norm=41.543, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.633e-04, train_time=1.419
+[gpub011:0/16] 2024-02-09 17:34:34,670 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-09 17:34:54,083 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 17:34:57,709 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97ccec3970>)
+[gpub011:0/16] 2024-02-09 17:34:57,709 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-09 17:34:57,712 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 17:42:43,116 (trainer:762) INFO: 38epoch:train:7501-7600batch: iter_time=3.697, forward_time=0.291, loss_ctc=47.991, loss_att=50.460, acc=0.745, loss=49.719, backward_time=0.299, grad_norm=44.699, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.633e-04, train_time=5.085
+[gpub011:0/16] 2024-02-09 17:45:14,571 (trainer:762) INFO: 38epoch:train:7601-7700batch: iter_time=8.212e-05, forward_time=0.353, loss_ctc=37.544, loss_att=39.122, acc=0.777, loss=38.649, backward_time=0.363, grad_norm=34.154, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.633e-04, train_time=1.514
+[gpub011:0/16] 2024-02-09 17:47:28,570 (trainer:762) INFO: 38epoch:train:7701-7800batch: iter_time=8.421e-05, forward_time=0.293, loss_ctc=48.761, loss_att=54.865, acc=0.748, loss=53.034, backward_time=0.299, grad_norm=45.068, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.633e-04, train_time=1.339
+[gpub011:0/16] 2024-02-09 17:49:33,547 (trainer:762) INFO: 38epoch:train:7801-7900batch: iter_time=3.547e-04, forward_time=0.297, loss_ctc=51.724, loss_att=49.701, acc=0.760, loss=50.308, backward_time=0.301, grad_norm=48.780, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.633e-04, train_time=1.250
+[gpub011:0/16] 2024-02-09 17:51:35,652 (trainer:762) INFO: 38epoch:train:7901-8000batch: iter_time=8.234e-05, forward_time=0.305, loss_ctc=47.169, loss_att=49.069, acc=0.766, loss=48.499, backward_time=0.307, grad_norm=39.609, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.633e-04, train_time=1.221
+[gpub011:0/16] 2024-02-09 17:54:15,903 (trainer:762) INFO: 38epoch:train:8001-8100batch: iter_time=8.382e-05, forward_time=0.396, loss_ctc=42.255, loss_att=41.560, acc=0.777, loss=41.769, backward_time=0.312, grad_norm=41.249, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.632e-04, train_time=1.601
+[gpub011:0/16] 2024-02-09 17:56:17,803 (trainer:762) INFO: 38epoch:train:8101-8200batch: iter_time=3.785e-04, forward_time=0.318, loss_ctc=53.769, loss_att=50.261, acc=0.749, loss=51.314, backward_time=0.304, grad_norm=57.521, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.632e-04, train_time=1.220
+[gpub011:0/16] 2024-02-09 17:58:38,050 (trainer:762) INFO: 38epoch:train:8201-8300batch: iter_time=8.441e-05, forward_time=0.291, loss_ctc=41.179, loss_att=39.551, acc=0.769, loss=40.039, backward_time=0.297, grad_norm=36.505, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.632e-04, train_time=1.402
+[gpub011:0/16] 2024-02-09 18:01:19,017 (trainer:762) INFO: 38epoch:train:8301-8400batch: iter_time=8.758e-05, forward_time=0.422, loss_ctc=52.701, loss_att=46.783, acc=0.775, loss=48.559, backward_time=0.329, grad_norm=49.383, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.632e-04, train_time=1.609
+[gpub011:0/16] 2024-02-09 18:03:19,242 (trainer:762) INFO: 38epoch:train:8401-8500batch: iter_time=8.269e-05, forward_time=0.295, loss_ctc=52.414, loss_att=48.402, acc=0.764, loss=49.605, backward_time=0.304, grad_norm=49.983, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.632e-04, train_time=1.202
+[gpub011:0/16] 2024-02-09 18:05:36,478 (trainer:762) INFO: 38epoch:train:8501-8600batch: iter_time=8.323e-05, forward_time=0.293, loss_ctc=43.715, loss_att=45.841, acc=0.757, loss=45.203, backward_time=0.298, grad_norm=41.802, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.632e-04, train_time=1.372
+[gpub011:0/16] 2024-02-09 18:08:23,627 (trainer:762) INFO: 38epoch:train:8601-8700batch: iter_time=8.486e-05, forward_time=0.405, loss_ctc=49.582, loss_att=47.493, acc=0.759, loss=48.120, backward_time=0.324, grad_norm=45.766, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.119, optim0_lr0=1.632e-04, train_time=1.671
+[gpub011:0/16] 2024-02-09 18:09:45,183 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-09 18:10:04,948 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 18:10:08,552 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9820e677c0>)
+[gpub011:0/16] 2024-02-09 18:10:08,553 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-09 18:10:08,565 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 18:16:44,049 (trainer:762) INFO: 38epoch:train:8701-8800batch: iter_time=3.356, forward_time=0.291, loss_ctc=48.831, loss_att=47.377, acc=0.754, loss=47.813, backward_time=0.297, grad_norm=41.730, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.631e-04, train_time=5.003
+[gpub011:0/16] 2024-02-09 18:18:43,282 (trainer:762) INFO: 38epoch:train:8801-8900batch: iter_time=1.003e-04, forward_time=0.291, loss_ctc=43.382, loss_att=47.310, acc=0.746, loss=46.132, backward_time=0.301, grad_norm=39.924, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.631e-04, train_time=1.193
+[gpub011:0/16] 2024-02-09 18:21:00,806 (trainer:762) INFO: 38epoch:train:8901-9000batch: iter_time=1.003e-04, forward_time=0.370, loss_ctc=41.275, loss_att=45.814, acc=0.764, loss=44.452, backward_time=0.377, grad_norm=40.314, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.631e-04, train_time=1.375
+[gpub011:0/16] 2024-02-09 18:23:33,619 (trainer:762) INFO: 38epoch:train:9001-9100batch: iter_time=8.283e-05, forward_time=0.319, loss_ctc=49.146, loss_att=44.519, acc=0.769, loss=45.907, backward_time=0.298, grad_norm=45.392, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.631e-04, train_time=1.527
+[gpub011:0/16] 2024-02-09 18:25:37,377 (trainer:762) INFO: 38epoch:train:9101-9200batch: iter_time=8.247e-05, forward_time=0.295, loss_ctc=49.106, loss_att=49.483, acc=0.747, loss=49.370, backward_time=0.301, grad_norm=44.126, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.631e-04, train_time=1.237
+[gpub011:0/16] 2024-02-09 18:26:49,758 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 18:27:49,047 (trainer:762) INFO: 38epoch:train:9201-9300batch: iter_time=9.870e-05, forward_time=0.402, loss_ctc=43.690, loss_att=43.925, acc=0.772, loss=43.855, backward_time=0.322, grad_norm=38.287, clip=100.000, loss_scale=1.374e+34, optim_step_time=0.102, optim0_lr0=1.631e-04, train_time=1.317
+[gpub011:0/16] 2024-02-09 18:30:38,075 (trainer:762) INFO: 38epoch:train:9301-9400batch: iter_time=9.435e-05, forward_time=0.295, loss_ctc=46.015, loss_att=41.595, acc=0.761, loss=42.921, backward_time=0.294, grad_norm=43.855, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.631e-04, train_time=1.690
+[gpub011:0/16] 2024-02-09 18:32:42,573 (trainer:762) INFO: 38epoch:train:9401-9500batch: iter_time=9.541e-05, forward_time=0.293, loss_ctc=50.671, loss_att=47.182, acc=0.752, loss=48.229, backward_time=0.296, grad_norm=52.897, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.630e-04, train_time=1.245
+[gpub011:0/16] 2024-02-09 18:34:56,139 (trainer:762) INFO: 38epoch:train:9501-9600batch: iter_time=8.780e-05, forward_time=0.395, loss_ctc=50.039, loss_att=42.772, acc=0.771, loss=44.952, backward_time=0.320, grad_norm=52.564, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=1.630e-04, train_time=1.335
+[gpub011:0/16] 2024-02-09 18:37:37,033 (trainer:762) INFO: 38epoch:train:9601-9700batch: iter_time=9.266e-05, forward_time=0.295, loss_ctc=53.535, loss_att=52.049, acc=0.748, loss=52.495, backward_time=0.298, grad_norm=53.093, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.630e-04, train_time=1.608
+[gpub011:0/16] 2024-02-09 18:39:55,613 (trainer:762) INFO: 38epoch:train:9701-9800batch: iter_time=9.223e-05, forward_time=0.444, loss_ctc=42.156, loss_att=36.780, acc=0.772, loss=38.393, backward_time=0.320, grad_norm=38.014, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.630e-04, train_time=1.387
+[gpub011:0/16] 2024-02-09 18:41:49,819 (trainer:762) INFO: 38epoch:train:9801-9900batch: iter_time=8.538e-05, forward_time=0.292, loss_ctc=46.561, loss_att=46.498, acc=0.745, loss=46.517, backward_time=0.301, grad_norm=45.370, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.630e-04, train_time=1.141
+[gpub011:0/16] 2024-02-09 18:44:21,774 (trainer:762) INFO: 38epoch:train:9901-10000batch: iter_time=8.048e-05, forward_time=0.301, loss_ctc=47.998, loss_att=46.466, acc=0.765, loss=46.925, backward_time=0.300, grad_norm=40.199, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.630e-04, train_time=1.517
+[gpub011:0/16] 2024-02-09 18:44:41,834 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-09 18:45:01,037 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 18:45:04,632 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97c73fd2a0>)
+[gpub011:0/16] 2024-02-09 18:45:04,632 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-09 18:45:04,635 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 18:52:40,548 (trainer:762) INFO: 38epoch:train:10001-10100batch: iter_time=3.616, forward_time=0.389, loss_ctc=48.348, loss_att=50.029, acc=0.746, loss=49.525, backward_time=0.348, grad_norm=42.721, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.629e-04, train_time=4.990
+[gpub011:0/16] 2024-02-09 18:55:00,661 (trainer:762) INFO: 38epoch:train:10101-10200batch: iter_time=8.480e-05, forward_time=0.288, loss_ctc=37.963, loss_att=39.284, acc=0.778, loss=38.888, backward_time=0.294, grad_norm=34.293, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.629e-04, train_time=1.401
+[gpub011:0/16] 2024-02-09 18:57:02,586 (trainer:762) INFO: 38epoch:train:10201-10300batch: iter_time=8.871e-05, forward_time=0.300, loss_ctc=48.516, loss_att=54.730, acc=0.749, loss=52.866, backward_time=0.308, grad_norm=43.805, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.629e-04, train_time=1.218
+[gpub011:0/16] 2024-02-09 18:59:18,778 (trainer:762) INFO: 38epoch:train:10301-10400batch: iter_time=9.071e-05, forward_time=0.293, loss_ctc=51.807, loss_att=49.079, acc=0.764, loss=49.897, backward_time=0.299, grad_norm=46.071, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.629e-04, train_time=1.363
+[gpub011:0/16] 2024-02-09 19:01:56,187 (trainer:762) INFO: 38epoch:train:10401-10500batch: iter_time=8.723e-05, forward_time=0.356, loss_ctc=47.037, loss_att=47.940, acc=0.771, loss=47.669, backward_time=0.383, grad_norm=39.627, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.629e-04, train_time=1.574
+[gpub011:0/16] 2024-02-09 19:02:58,671 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 19:04:03,443 (trainer:762) INFO: 38epoch:train:10501-10600batch: iter_time=8.952e-05, forward_time=0.293, loss_ctc=41.817, loss_att=41.285, acc=0.777, loss=41.445, backward_time=0.297, grad_norm=40.890, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.094, optim0_lr0=1.629e-04, train_time=1.272
+[gpub011:0/16] 2024-02-09 19:06:04,019 (trainer:762) INFO: 38epoch:train:10601-10700batch: iter_time=8.315e-05, forward_time=0.293, loss_ctc=53.434, loss_att=48.842, acc=0.753, loss=50.220, backward_time=0.299, grad_norm=55.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.629e-04, train_time=1.205
+[gpub011:0/16] 2024-02-09 19:09:03,918 (trainer:762) INFO: 38epoch:train:10701-10800batch: iter_time=9.565e-05, forward_time=0.423, loss_ctc=41.182, loss_att=39.723, acc=0.768, loss=40.161, backward_time=0.320, grad_norm=37.166, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.628e-04, train_time=1.800
+[gpub011:0/16] 2024-02-09 19:11:09,039 (trainer:762) INFO: 38epoch:train:10801-10900batch: iter_time=9.252e-05, forward_time=0.298, loss_ctc=51.683, loss_att=45.453, acc=0.781, loss=47.322, backward_time=0.305, grad_norm=49.354, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.628e-04, train_time=1.251
+[gpub011:0/16] 2024-02-09 19:13:58,869 (trainer:762) INFO: 38epoch:train:10901-11000batch: iter_time=8.669e-05, forward_time=0.292, loss_ctc=51.234, loss_att=47.251, acc=0.765, loss=48.446, backward_time=0.298, grad_norm=46.512, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.628e-04, train_time=1.697
+[gpub011:0/16] 2024-02-09 19:16:09,062 (trainer:762) INFO: 38epoch:train:11001-11100batch: iter_time=8.757e-05, forward_time=0.374, loss_ctc=44.072, loss_att=45.453, acc=0.757, loss=45.039, backward_time=0.381, grad_norm=42.648, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.628e-04, train_time=1.303
+[gpub011:0/16] 2024-02-09 19:18:39,109 (trainer:762) INFO: 38epoch:train:11101-11200batch: iter_time=8.224e-05, forward_time=0.294, loss_ctc=49.022, loss_att=47.191, acc=0.759, loss=47.741, backward_time=0.299, grad_norm=43.473, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.628e-04, train_time=1.500
+[gpub011:0/16] 2024-02-09 19:20:04,774 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-09 19:20:24,130 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 19:20:27,699 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98383b3790>)
+[gpub011:0/16] 2024-02-09 19:20:27,699 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-09 19:20:27,702 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 19:27:18,471 (trainer:762) INFO: 38epoch:train:11201-11300batch: iter_time=3.828, forward_time=0.399, loss_ctc=48.312, loss_att=45.491, acc=0.765, loss=46.337, backward_time=0.318, grad_norm=40.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.628e-04, train_time=5.193
+[gpub011:0/16] 2024-02-09 19:29:36,498 (trainer:762) INFO: 38epoch:train:11301-11400batch: iter_time=8.099e-05, forward_time=0.306, loss_ctc=43.400, loss_att=46.671, acc=0.758, loss=45.690, backward_time=0.298, grad_norm=39.681, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.628e-04, train_time=1.380
+[gpub011:0/16] 2024-02-09 19:31:55,648 (trainer:762) INFO: 38epoch:train:11401-11500batch: iter_time=8.210e-05, forward_time=0.292, loss_ctc=41.333, loss_att=46.696, acc=0.768, loss=45.087, backward_time=0.295, grad_norm=38.731, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.627e-04, train_time=1.391
+[gpub011:0/16] 2024-02-09 19:34:54,114 (trainer:762) INFO: 38epoch:train:11501-11600batch: iter_time=8.917e-05, forward_time=0.420, loss_ctc=49.302, loss_att=47.513, acc=0.769, loss=48.050, backward_time=0.324, grad_norm=42.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.627e-04, train_time=1.784
+[gpub011:0/16] 2024-02-09 19:36:50,133 (trainer:762) INFO: 38epoch:train:11601-11700batch: iter_time=4.398e-04, forward_time=0.310, loss_ctc=49.319, loss_att=49.534, acc=0.760, loss=49.470, backward_time=0.301, grad_norm=41.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.627e-04, train_time=1.160
+[gpub011:0/16] 2024-02-09 19:39:42,988 (trainer:762) INFO: 38epoch:train:11701-11800batch: iter_time=8.287e-05, forward_time=0.292, loss_ctc=43.350, loss_att=44.309, acc=0.776, loss=44.021, backward_time=0.296, grad_norm=37.521, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.627e-04, train_time=1.728
+[gpub011:0/16] 2024-02-09 19:42:00,869 (trainer:762) INFO: 38epoch:train:11801-11900batch: iter_time=8.406e-05, forward_time=0.354, loss_ctc=45.410, loss_att=41.069, acc=0.772, loss=42.371, backward_time=0.362, grad_norm=43.690, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.627e-04, train_time=1.379
+[gpub011:0/16] 2024-02-09 19:44:25,029 (trainer:762) INFO: 38epoch:train:11901-12000batch: iter_time=3.980e-04, forward_time=0.291, loss_ctc=49.426, loss_att=47.852, acc=0.759, loss=48.324, backward_time=0.306, grad_norm=53.348, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.627e-04, train_time=1.441
+[gpub011:0/16] 2024-02-09 19:46:44,478 (trainer:762) INFO: 38epoch:train:12001-12100batch: iter_time=8.192e-05, forward_time=0.298, loss_ctc=50.051, loss_att=41.355, acc=0.776, loss=43.964, backward_time=0.297, grad_norm=48.253, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.627e-04, train_time=1.394
+[gpub011:0/16] 2024-02-09 19:49:01,644 (trainer:762) INFO: 38epoch:train:12101-12200batch: iter_time=8.284e-05, forward_time=0.394, loss_ctc=52.752, loss_att=53.458, acc=0.759, loss=53.246, backward_time=0.362, grad_norm=54.596, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.626e-04, train_time=1.371
+[gpub011:0/16] 2024-02-09 19:51:24,034 (trainer:762) INFO: 38epoch:train:12201-12300batch: iter_time=8.699e-05, forward_time=0.296, loss_ctc=42.092, loss_att=38.462, acc=0.774, loss=39.551, backward_time=0.294, grad_norm=205.017, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.626e-04, train_time=1.424
+[gpub011:0/16] 2024-02-09 19:53:21,648 (trainer:762) INFO: 38epoch:train:12301-12400batch: iter_time=8.272e-05, forward_time=0.295, loss_ctc=46.374, loss_att=46.627, acc=0.756, loss=46.551, backward_time=0.301, grad_norm=43.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.626e-04, train_time=1.176
+[gpub011:0/16] 2024-02-09 19:55:35,814 (trainer:762) INFO: 38epoch:train:12401-12500batch: iter_time=8.124e-05, forward_time=0.396, loss_ctc=48.458, loss_att=47.029, acc=0.771, loss=47.458, backward_time=0.328, grad_norm=41.899, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.626e-04, train_time=1.342
+[gpub011:0/16] 2024-02-09 19:55:55,841 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-09 19:56:15,691 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 19:56:19,529 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9872302200>)
+[gpub011:0/16] 2024-02-09 19:56:19,529 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-09 19:56:19,533 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 20:02:49,736 (trainer:762) INFO: 38epoch:train:12501-12600batch: iter_time=3.063, forward_time=0.293, loss_ctc=48.159, loss_att=50.439, acc=0.737, loss=49.755, backward_time=0.298, grad_norm=45.635, clip=100.000, loss_scale=7.529e+33, optim_step_time=0.093, optim0_lr0=1.626e-04, train_time=4.339
+[gpub011:0/16] 2024-02-09 20:04:04,929 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 20:05:04,806 (trainer:762) INFO: 38epoch:train:12601-12700batch: iter_time=2.654e-04, forward_time=0.374, loss_ctc=37.933, loss_att=38.278, acc=0.769, loss=38.174, backward_time=0.350, grad_norm=35.435, clip=100.000, loss_scale=7.500e+33, optim_step_time=0.097, optim0_lr0=1.626e-04, train_time=1.351
+[gpub011:0/16] 2024-02-09 20:07:20,774 (trainer:762) INFO: 38epoch:train:12701-12800batch: iter_time=8.393e-05, forward_time=0.293, loss_ctc=47.793, loss_att=53.161, acc=0.744, loss=51.551, backward_time=0.297, grad_norm=44.928, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.626e-04, train_time=1.359
+[gpub011:0/16] 2024-02-09 20:09:36,130 (trainer:762) INFO: 38epoch:train:12801-12900batch: iter_time=2.308e-04, forward_time=0.392, loss_ctc=51.231, loss_att=49.296, acc=0.755, loss=49.876, backward_time=0.348, grad_norm=47.658, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.625e-04, train_time=1.353
+[gpub011:0/16] 2024-02-09 20:11:31,531 (trainer:762) INFO: 38epoch:train:12901-13000batch: iter_time=8.257e-05, forward_time=0.300, loss_ctc=46.945, loss_att=47.785, acc=0.760, loss=47.533, backward_time=0.300, grad_norm=39.259, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.625e-04, train_time=1.154
+[gpub011:0/16] 2024-02-09 20:14:16,593 (trainer:762) INFO: 38epoch:train:13001-13100batch: iter_time=2.716e-04, forward_time=0.355, loss_ctc=41.819, loss_att=41.512, acc=0.772, loss=41.604, backward_time=0.353, grad_norm=41.151, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.625e-04, train_time=1.650
+[gpub011:0/16] 2024-02-09 20:16:16,152 (trainer:762) INFO: 38epoch:train:13101-13200batch: iter_time=8.573e-05, forward_time=0.297, loss_ctc=53.535, loss_att=48.613, acc=0.746, loss=50.089, backward_time=0.300, grad_norm=57.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.625e-04, train_time=1.196
+[gpub011:0/16] 2024-02-09 20:18:26,154 (trainer:762) INFO: 38epoch:train:13201-13300batch: iter_time=8.567e-05, forward_time=0.385, loss_ctc=40.897, loss_att=38.422, acc=0.767, loss=39.164, backward_time=0.340, grad_norm=36.303, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.625e-04, train_time=1.300
+[gpub011:0/16] 2024-02-09 20:20:57,099 (trainer:762) INFO: 38epoch:train:13301-13400batch: iter_time=4.039e-04, forward_time=0.292, loss_ctc=51.647, loss_att=45.007, acc=0.774, loss=46.999, backward_time=0.301, grad_norm=51.543, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.625e-04, train_time=1.508
+[gpub011:0/16] 2024-02-09 20:23:12,231 (trainer:762) INFO: 38epoch:train:13401-13500batch: iter_time=8.220e-05, forward_time=0.402, loss_ctc=52.129, loss_att=46.053, acc=0.764, loss=47.876, backward_time=0.337, grad_norm=52.001, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.625e-04, train_time=1.351
+[gpub011:0/16] 2024-02-09 20:25:27,890 (trainer:762) INFO: 38epoch:train:13501-13600batch: iter_time=2.495e-04, forward_time=0.299, loss_ctc=43.480, loss_att=45.537, acc=0.745, loss=44.920, backward_time=0.298, grad_norm=41.915, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.624e-04, train_time=1.357
+[gpub011:0/16] 2024-02-09 20:27:46,822 (trainer:762) INFO: 38epoch:train:13601-13700batch: iter_time=8.175e-05, forward_time=0.386, loss_ctc=49.022, loss_att=46.921, acc=0.750, loss=47.551, backward_time=0.335, grad_norm=45.396, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.624e-04, train_time=1.388
+[gpub011:0/16] 2024-02-09 20:29:07,703 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-09 20:29:27,683 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 20:29:31,240 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97df033610>)
+[gpub011:0/16] 2024-02-09 20:29:31,240 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-09 20:29:31,244 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 20:35:25,344 (trainer:762) INFO: 38epoch:train:13701-13800batch: iter_time=3.332, forward_time=0.295, loss_ctc=48.113, loss_att=46.007, acc=0.762, loss=46.639, backward_time=0.304, grad_norm=41.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.624e-04, train_time=4.586
+[gpub011:0/16] 2024-02-09 20:37:30,450 (trainer:762) INFO: 38epoch:train:13801-13900batch: iter_time=7.986e-05, forward_time=0.338, loss_ctc=43.065, loss_att=47.601, acc=0.756, loss=46.240, backward_time=0.334, grad_norm=39.587, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.624e-04, train_time=1.251
+[gpub011:0/16] 2024-02-09 20:39:52,485 (trainer:762) INFO: 38epoch:train:13901-14000batch: iter_time=8.009e-05, forward_time=0.327, loss_ctc=40.629, loss_att=46.625, acc=0.770, loss=44.826, backward_time=0.297, grad_norm=40.167, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.624e-04, train_time=1.420
+[gpub011:0/16] 2024-02-09 20:42:34,450 (trainer:762) INFO: 38epoch:train:14001-14100batch: iter_time=8.254e-05, forward_time=0.303, loss_ctc=49.952, loss_att=48.537, acc=0.770, loss=48.962, backward_time=0.297, grad_norm=44.634, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.624e-04, train_time=1.619
+[gpub011:0/16] 2024-02-09 20:44:36,360 (trainer:762) INFO: 38epoch:train:14101-14200batch: iter_time=8.034e-05, forward_time=0.295, loss_ctc=49.437, loss_att=49.509, acc=0.760, loss=49.488, backward_time=0.307, grad_norm=44.012, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.624e-04, train_time=1.219
+[gpub011:0/16] 2024-02-09 20:47:08,505 (trainer:762) INFO: 38epoch:train:14201-14300batch: iter_time=8.197e-05, forward_time=0.355, loss_ctc=43.702, loss_att=44.262, acc=0.778, loss=44.094, backward_time=0.345, grad_norm=39.270, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.623e-04, train_time=1.521
+[gpub011:0/16] 2024-02-09 20:49:28,643 (trainer:762) INFO: 38epoch:train:14301-14400batch: iter_time=1.511e-04, forward_time=0.305, loss_ctc=45.068, loss_att=40.924, acc=0.772, loss=42.167, backward_time=0.300, grad_norm=41.093, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.623e-04, train_time=1.401
+[gpub011:0/16] 2024-02-09 20:51:27,275 (trainer:762) INFO: 38epoch:train:14401-14500batch: iter_time=8.626e-05, forward_time=0.291, loss_ctc=50.158, loss_att=49.154, acc=0.759, loss=49.456, backward_time=0.296, grad_norm=62.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.623e-04, train_time=1.186
+[gpub011:0/16] 2024-02-09 20:54:04,825 (trainer:762) INFO: 38epoch:train:14501-14600batch: iter_time=8.210e-05, forward_time=0.368, loss_ctc=50.177, loss_att=41.293, acc=0.775, loss=43.958, backward_time=0.328, grad_norm=50.405, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.623e-04, train_time=1.575
+[gpub011:0/16] 2024-02-09 20:56:04,683 (trainer:762) INFO: 38epoch:train:14601-14700batch: iter_time=8.791e-05, forward_time=0.298, loss_ctc=52.614, loss_att=53.764, acc=0.760, loss=53.419, backward_time=0.311, grad_norm=53.153, clip=100.000, loss_scale=8.048e+33, optim_step_time=0.094, optim0_lr0=1.623e-04, train_time=1.198
+[gpub011:0/16] 2024-02-09 20:58:23,264 (trainer:762) INFO: 38epoch:train:14701-14800batch: iter_time=8.338e-05, forward_time=0.288, loss_ctc=42.115, loss_att=38.256, acc=0.775, loss=39.414, backward_time=0.294, grad_norm=39.707, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.623e-04, train_time=1.386
+[gpub011:0/16] 2024-02-09 21:00:50,740 (trainer:762) INFO: 38epoch:train:14801-14900batch: iter_time=8.478e-05, forward_time=0.375, loss_ctc=46.803, loss_att=47.165, acc=0.756, loss=47.056, backward_time=0.351, grad_norm=46.172, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.623e-04, train_time=1.474
+[gpub011:0/16] 2024-02-09 21:03:07,040 (trainer:762) INFO: 38epoch:train:14901-15000batch: iter_time=8.049e-05, forward_time=0.301, loss_ctc=48.325, loss_att=47.156, acc=0.771, loss=47.507, backward_time=0.298, grad_norm=41.274, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.622e-04, train_time=1.363
+[gpub011:0/16] 2024-02-09 21:41:40,488 (trainer:361) INFO: 38epoch results: [train] iter_time=0.273, forward_time=0.326, loss_ctc=47.705, loss_att=46.264, acc=0.760, loss=46.696, backward_time=0.312, grad_norm=46.262, clip=100.000, loss_scale=6.194e+33, optim_step_time=0.096, optim0_lr0=1.633e-04, train_time=1.639, time=6 hours, 50 minutes and 12.04 seconds, total_count=600000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=35.145, cer_ctc=0.179, loss_att=37.130, acc=0.683, cer=0.345, wer=0.998, loss=36.534, time=38 minutes and 9.34 seconds, total_count=186840, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-09 21:41:50,207 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-09 21:41:50,362 (trainer:290) INFO: 39/45epoch started. Estimated time to finish: 2 days, 3 hours and 37 minutes
+[gpub011:0/16] 2024-02-09 21:41:50,371 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-09 21:42:09,062 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 21:42:12,545 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065b1f7c0>)
+[gpub011:0/16] 2024-02-09 21:42:12,545 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-09 21:42:12,548 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 21:48:31,038 (trainer:762) INFO: 39epoch:train:1-100batch: iter_time=2.861, forward_time=0.339, loss_ctc=51.230, loss_att=46.265, acc=0.760, loss=47.754, backward_time=0.303, grad_norm=47.260, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.622e-04, train_time=4.006
+[gpub011:0/16] 2024-02-09 21:50:24,504 (trainer:762) INFO: 39epoch:train:101-200batch: iter_time=8.253e-05, forward_time=0.293, loss_ctc=45.606, loss_att=38.903, acc=0.784, loss=40.914, backward_time=0.306, grad_norm=42.005, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.622e-04, train_time=1.135
+[gpub011:0/16] 2024-02-09 21:52:47,288 (trainer:762) INFO: 39epoch:train:201-300batch: iter_time=8.473e-05, forward_time=0.294, loss_ctc=54.257, loss_att=54.412, acc=0.755, loss=54.365, backward_time=0.301, grad_norm=56.390, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.622e-04, train_time=1.428
+[gpub011:0/16] 2024-02-09 21:55:01,593 (trainer:762) INFO: 39epoch:train:301-400batch: iter_time=9.345e-05, forward_time=0.377, loss_ctc=55.951, loss_att=48.755, acc=0.761, loss=50.913, backward_time=0.324, grad_norm=50.768, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.622e-04, train_time=1.343
+[gpub011:0/16] 2024-02-09 21:57:03,237 (trainer:762) INFO: 39epoch:train:401-500batch: iter_time=9.004e-05, forward_time=0.290, loss_ctc=43.364, loss_att=39.515, acc=0.770, loss=40.670, backward_time=0.300, grad_norm=40.189, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.622e-04, train_time=1.216
+[gpub011:0/16] 2024-02-09 21:59:07,906 (trainer:762) INFO: 39epoch:train:501-600batch: iter_time=8.967e-05, forward_time=0.293, loss_ctc=50.515, loss_att=52.554, acc=0.734, loss=51.942, backward_time=0.299, grad_norm=46.221, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.622e-04, train_time=1.247
+[gpub011:0/16] 2024-02-09 22:01:15,482 (trainer:762) INFO: 39epoch:train:601-700batch: iter_time=8.648e-05, forward_time=0.303, loss_ctc=40.461, loss_att=43.185, acc=0.776, loss=42.367, backward_time=0.306, grad_norm=37.563, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.621e-04, train_time=1.276
+[gpub011:0/16] 2024-02-09 22:03:39,887 (trainer:762) INFO: 39epoch:train:701-800batch: iter_time=1.039e-04, forward_time=0.337, loss_ctc=47.492, loss_att=46.150, acc=0.765, loss=46.552, backward_time=0.337, grad_norm=41.427, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.621e-04, train_time=1.443
+[gpub011:0/16] 2024-02-09 22:05:35,444 (trainer:762) INFO: 39epoch:train:801-900batch: iter_time=9.205e-05, forward_time=0.290, loss_ctc=48.235, loss_att=37.427, acc=0.763, loss=40.670, backward_time=0.294, grad_norm=45.324, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.621e-04, train_time=1.156
+[gpub011:0/16] 2024-02-09 22:07:46,201 (trainer:762) INFO: 39epoch:train:901-1000batch: iter_time=8.830e-05, forward_time=0.294, loss_ctc=49.381, loss_att=47.409, acc=0.764, loss=48.001, backward_time=0.299, grad_norm=56.574, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.621e-04, train_time=1.307
+[gpub011:0/16] 2024-02-09 22:09:52,452 (trainer:762) INFO: 39epoch:train:1001-1100batch: iter_time=1.983e-04, forward_time=0.306, loss_ctc=38.679, loss_att=35.676, acc=0.771, loss=36.577, backward_time=0.314, grad_norm=37.649, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.621e-04, train_time=1.262
+[gpub011:0/16] 2024-02-09 22:12:10,422 (trainer:762) INFO: 39epoch:train:1101-1200batch: iter_time=9.115e-05, forward_time=0.334, loss_ctc=41.549, loss_att=44.292, acc=0.759, loss=43.469, backward_time=0.332, grad_norm=38.930, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.621e-04, train_time=1.379
+[gpub011:0/16] 2024-02-09 22:13:31,904 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-09 22:13:51,222 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 22:13:55,139 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065bdf9d0>)
+[gpub011:0/16] 2024-02-09 22:13:55,139 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-09 22:13:55,143 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 22:19:47,374 (trainer:762) INFO: 39epoch:train:1201-1300batch: iter_time=3.297, forward_time=0.296, loss_ctc=44.125, loss_att=38.059, acc=0.789, loss=39.879, backward_time=0.296, grad_norm=39.143, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.621e-04, train_time=4.570
+[gpub011:0/16] 2024-02-09 22:21:45,163 (trainer:762) INFO: 39epoch:train:1301-1400batch: iter_time=7.811e-05, forward_time=0.312, loss_ctc=44.882, loss_att=44.634, acc=0.770, loss=44.709, backward_time=0.320, grad_norm=45.891, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.620e-04, train_time=1.178
+[gpub011:0/16] 2024-02-09 22:22:04,735 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 22:24:16,465 (trainer:762) INFO: 39epoch:train:1401-1500batch: iter_time=8.742e-05, forward_time=0.364, loss_ctc=47.104, loss_att=42.119, acc=0.783, loss=43.614, backward_time=0.314, grad_norm=39.689, clip=100.000, loss_scale=5.559e+33, optim_step_time=0.094, optim0_lr0=1.620e-04, train_time=1.512
+[gpub011:0/16] 2024-02-09 22:26:20,296 (trainer:762) INFO: 39epoch:train:1501-1600batch: iter_time=8.833e-05, forward_time=0.301, loss_ctc=57.515, loss_att=53.569, acc=0.758, loss=54.753, backward_time=0.309, grad_norm=54.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.620e-04, train_time=1.238
+[gpub011:0/16] 2024-02-09 22:28:28,657 (trainer:762) INFO: 39epoch:train:1601-1700batch: iter_time=8.749e-05, forward_time=0.312, loss_ctc=50.521, loss_att=47.855, acc=0.763, loss=48.655, backward_time=0.305, grad_norm=48.810, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.620e-04, train_time=1.284
+[gpub011:0/16] 2024-02-09 22:30:40,278 (trainer:762) INFO: 39epoch:train:1701-1800batch: iter_time=9.134e-05, forward_time=0.361, loss_ctc=42.752, loss_att=42.523, acc=0.754, loss=42.592, backward_time=0.312, grad_norm=39.309, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.620e-04, train_time=1.316
+[gpub011:0/16] 2024-02-09 22:32:53,560 (trainer:762) INFO: 39epoch:train:1801-1900batch: iter_time=9.193e-05, forward_time=0.298, loss_ctc=46.024, loss_att=49.665, acc=0.755, loss=48.573, backward_time=0.303, grad_norm=42.499, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.620e-04, train_time=1.333
+[gpub011:0/16] 2024-02-09 22:35:01,354 (trainer:762) INFO: 39epoch:train:1901-2000batch: iter_time=9.245e-05, forward_time=0.308, loss_ctc=46.601, loss_att=47.509, acc=0.775, loss=47.237, backward_time=0.311, grad_norm=41.383, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.620e-04, train_time=1.278
+[gpub011:0/16] 2024-02-09 22:37:02,285 (trainer:762) INFO: 39epoch:train:2001-2100batch: iter_time=8.661e-05, forward_time=0.346, loss_ctc=41.219, loss_att=36.170, acc=0.765, loss=37.684, backward_time=0.309, grad_norm=40.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.620e-04, train_time=1.208
+[gpub011:0/16] 2024-02-09 22:39:22,005 (trainer:762) INFO: 39epoch:train:2101-2200batch: iter_time=8.518e-05, forward_time=0.294, loss_ctc=51.760, loss_att=46.120, acc=0.762, loss=47.812, backward_time=0.304, grad_norm=59.843, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.619e-04, train_time=1.398
+[gpub011:0/16] 2024-02-09 22:41:35,834 (trainer:762) INFO: 39epoch:train:2201-2300batch: iter_time=9.326e-05, forward_time=0.338, loss_ctc=39.030, loss_att=34.986, acc=0.782, loss=36.199, backward_time=0.333, grad_norm=36.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.619e-04, train_time=1.338
+[gpub011:0/16] 2024-02-09 22:43:29,424 (trainer:762) INFO: 39epoch:train:2301-2400batch: iter_time=9.229e-05, forward_time=0.294, loss_ctc=40.681, loss_att=39.551, acc=0.762, loss=39.890, backward_time=0.300, grad_norm=39.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.619e-04, train_time=1.135
+[gpub011:0/16] 2024-02-09 22:45:48,760 (trainer:762) INFO: 39epoch:train:2401-2500batch: iter_time=9.248e-05, forward_time=0.299, loss_ctc=41.337, loss_att=42.639, acc=0.785, loss=42.249, backward_time=0.304, grad_norm=37.969, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.619e-04, train_time=1.394
+[gpub011:0/16] 2024-02-09 22:46:08,789 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-09 22:46:28,403 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 22:46:31,921 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9af6c5bd60>)
+[gpub011:0/16] 2024-02-09 22:46:31,921 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-09 22:46:31,925 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 22:52:34,011 (trainer:762) INFO: 39epoch:train:2501-2600batch: iter_time=2.917, forward_time=0.329, loss_ctc=49.726, loss_att=48.331, acc=0.751, loss=48.750, backward_time=0.303, grad_norm=45.894, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.619e-04, train_time=4.051
+[gpub011:0/16] 2024-02-09 22:54:40,598 (trainer:762) INFO: 39epoch:train:2601-2700batch: iter_time=9.423e-05, forward_time=0.386, loss_ctc=44.648, loss_att=39.264, acc=0.778, loss=40.879, backward_time=0.320, grad_norm=40.951, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.619e-04, train_time=1.267
+[gpub011:0/16] 2024-02-09 22:56:58,420 (trainer:762) INFO: 39epoch:train:2701-2800batch: iter_time=9.791e-05, forward_time=0.294, loss_ctc=51.768, loss_att=55.109, acc=0.751, loss=54.107, backward_time=0.316, grad_norm=53.660, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.619e-04, train_time=1.378
+[gpub011:0/16] 2024-02-09 22:59:00,680 (trainer:762) INFO: 39epoch:train:2801-2900batch: iter_time=8.521e-05, forward_time=0.350, loss_ctc=52.141, loss_att=47.014, acc=0.752, loss=48.552, backward_time=0.334, grad_norm=49.741, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.618e-04, train_time=1.222
+[gpub011:0/16] 2024-02-09 23:00:59,797 (trainer:762) INFO: 39epoch:train:2901-3000batch: iter_time=8.234e-05, forward_time=0.289, loss_ctc=42.124, loss_att=39.565, acc=0.767, loss=40.333, backward_time=0.295, grad_norm=39.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.618e-04, train_time=1.192
+[gpub011:0/16] 2024-02-09 23:03:41,591 (trainer:762) INFO: 39epoch:train:3001-3100batch: iter_time=8.419e-05, forward_time=0.415, loss_ctc=48.894, loss_att=51.347, acc=0.733, loss=50.611, backward_time=0.317, grad_norm=45.520, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.618e-04, train_time=1.618
+[gpub011:0/16] 2024-02-09 23:05:33,660 (trainer:762) INFO: 39epoch:train:3101-3200batch: iter_time=8.268e-05, forward_time=0.289, loss_ctc=40.070, loss_att=42.753, acc=0.774, loss=41.948, backward_time=0.296, grad_norm=38.758, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.618e-04, train_time=1.120
+[gpub011:0/16] 2024-02-09 23:07:50,214 (trainer:762) INFO: 39epoch:train:3201-3300batch: iter_time=4.685e-04, forward_time=0.389, loss_ctc=46.966, loss_att=46.869, acc=0.759, loss=46.899, backward_time=0.346, grad_norm=42.042, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.618e-04, train_time=1.364
+[gpub011:0/16] 2024-02-09 23:09:55,854 (trainer:762) INFO: 39epoch:train:3301-3400batch: iter_time=9.047e-05, forward_time=0.295, loss_ctc=46.309, loss_att=37.251, acc=0.761, loss=39.968, backward_time=0.295, grad_norm=41.944, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.618e-04, train_time=1.257
+[gpub011:0/16] 2024-02-09 23:12:21,256 (trainer:762) INFO: 39epoch:train:3401-3500batch: iter_time=4.084e-04, forward_time=0.370, loss_ctc=47.382, loss_att=47.109, acc=0.760, loss=47.191, backward_time=0.306, grad_norm=51.913, clip=100.000, loss_scale=9.969e+33, optim_step_time=0.099, optim0_lr0=1.618e-04, train_time=1.454
+[gpub011:0/16] 2024-02-09 23:14:22,742 (trainer:762) INFO: 39epoch:train:3501-3600batch: iter_time=8.515e-05, forward_time=0.318, loss_ctc=38.266, loss_att=34.994, acc=0.773, loss=35.976, backward_time=0.304, grad_norm=35.513, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.617e-04, train_time=1.214
+[gpub011:0/16] 2024-02-09 23:16:36,766 (trainer:762) INFO: 39epoch:train:3601-3700batch: iter_time=3.908e-04, forward_time=0.336, loss_ctc=40.553, loss_att=45.029, acc=0.745, loss=43.686, backward_time=0.301, grad_norm=38.335, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.617e-04, train_time=1.341
+[gpub011:0/16] 2024-02-09 23:18:02,378 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-09 23:18:21,723 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 23:18:25,310 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9a9fdf4430>)
+[gpub011:0/16] 2024-02-09 23:18:25,310 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-09 23:18:25,314 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 23:24:28,541 (trainer:762) INFO: 39epoch:train:3701-3800batch: iter_time=3.349, forward_time=0.353, loss_ctc=43.912, loss_att=38.571, acc=0.786, loss=40.174, backward_time=0.312, grad_norm=39.527, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.617e-04, train_time=4.718
+[gpub011:0/16] 2024-02-09 23:26:26,464 (trainer:762) INFO: 39epoch:train:3801-3900batch: iter_time=7.888e-05, forward_time=0.291, loss_ctc=43.472, loss_att=44.953, acc=0.770, loss=44.508, backward_time=0.297, grad_norm=42.719, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.617e-04, train_time=1.178
+[gpub011:0/16] 2024-02-09 23:28:03,676 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-09 23:28:51,033 (trainer:762) INFO: 39epoch:train:3901-4000batch: iter_time=2.464e-04, forward_time=0.402, loss_ctc=46.487, loss_att=41.862, acc=0.786, loss=43.249, backward_time=0.321, grad_norm=39.849, clip=100.000, loss_scale=8.129e+33, optim_step_time=0.104, optim0_lr0=1.617e-04, train_time=1.446
+[gpub011:0/16] 2024-02-09 23:31:03,205 (trainer:762) INFO: 39epoch:train:4001-4100batch: iter_time=8.557e-05, forward_time=0.293, loss_ctc=56.488, loss_att=53.710, acc=0.758, loss=54.543, backward_time=0.300, grad_norm=51.973, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.617e-04, train_time=1.322
+[gpub011:0/16] 2024-02-09 23:33:09,542 (trainer:762) INFO: 39epoch:train:4101-4200batch: iter_time=8.163e-05, forward_time=0.291, loss_ctc=49.148, loss_att=47.614, acc=0.764, loss=48.074, backward_time=0.297, grad_norm=45.680, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.617e-04, train_time=1.262
+[gpub011:0/16] 2024-02-09 23:35:31,288 (trainer:762) INFO: 39epoch:train:4201-4300batch: iter_time=0.002, forward_time=0.432, loss_ctc=42.828, loss_att=42.926, acc=0.754, loss=42.897, backward_time=0.311, grad_norm=39.790, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.616e-04, train_time=1.418
+[gpub011:0/16] 2024-02-09 23:37:45,397 (trainer:762) INFO: 39epoch:train:4301-4400batch: iter_time=8.206e-05, forward_time=0.293, loss_ctc=45.862, loss_att=48.942, acc=0.759, loss=48.018, backward_time=0.297, grad_norm=42.802, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.616e-04, train_time=1.341
+[gpub011:0/16] 2024-02-09 23:39:42,286 (trainer:762) INFO: 39epoch:train:4401-4500batch: iter_time=3.285e-04, forward_time=0.321, loss_ctc=46.324, loss_att=47.257, acc=0.777, loss=46.977, backward_time=0.316, grad_norm=40.317, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.616e-04, train_time=1.169
+[gpub011:0/16] 2024-02-09 23:41:58,525 (trainer:762) INFO: 39epoch:train:4501-4600batch: iter_time=8.102e-05, forward_time=0.397, loss_ctc=40.517, loss_att=35.706, acc=0.769, loss=37.149, backward_time=0.311, grad_norm=37.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.616e-04, train_time=1.362
+[gpub011:0/16] 2024-02-09 23:44:28,284 (trainer:762) INFO: 39epoch:train:4601-4700batch: iter_time=8.132e-05, forward_time=0.291, loss_ctc=50.809, loss_att=46.738, acc=0.760, loss=47.959, backward_time=0.296, grad_norm=55.302, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.616e-04, train_time=1.498
+[gpub011:0/16] 2024-02-09 23:46:36,707 (trainer:762) INFO: 39epoch:train:4701-4800batch: iter_time=3.609e-04, forward_time=0.390, loss_ctc=38.764, loss_att=35.339, acc=0.779, loss=36.367, backward_time=0.321, grad_norm=35.955, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.128, optim0_lr0=1.616e-04, train_time=1.283
+[gpub011:0/16] 2024-02-09 23:48:36,922 (trainer:762) INFO: 39epoch:train:4801-4900batch: iter_time=8.214e-05, forward_time=0.288, loss_ctc=40.392, loss_att=39.220, acc=0.765, loss=39.571, backward_time=0.294, grad_norm=39.616, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.616e-04, train_time=1.203
+[gpub011:0/16] 2024-02-09 23:50:47,623 (trainer:762) INFO: 39epoch:train:4901-5000batch: iter_time=8.224e-05, forward_time=0.291, loss_ctc=41.085, loss_att=43.153, acc=0.784, loss=42.533, backward_time=0.297, grad_norm=37.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.615e-04, train_time=1.307
+[gpub011:0/16] 2024-02-09 23:51:07,671 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-09 23:51:26,920 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-09 23:51:30,513 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9add22d810>)
+[gpub011:0/16] 2024-02-09 23:51:30,514 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-09 23:51:30,517 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-09 23:58:55,253 (trainer:762) INFO: 39epoch:train:5001-5100batch: iter_time=3.398, forward_time=0.454, loss_ctc=49.720, loss_att=45.565, acc=0.768, loss=46.811, backward_time=0.314, grad_norm=43.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.615e-04, train_time=4.875
+[gpub011:0/16] 2024-02-10 00:01:08,608 (trainer:762) INFO: 39epoch:train:5101-5200batch: iter_time=8.096e-05, forward_time=0.294, loss_ctc=43.664, loss_att=38.064, acc=0.789, loss=39.744, backward_time=0.297, grad_norm=39.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.615e-04, train_time=1.334
+[gpub011:0/16] 2024-02-10 00:03:27,524 (trainer:762) INFO: 39epoch:train:5201-5300batch: iter_time=8.429e-05, forward_time=0.294, loss_ctc=52.296, loss_att=55.019, acc=0.760, loss=54.202, backward_time=0.301, grad_norm=56.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.615e-04, train_time=1.388
+[gpub011:0/16] 2024-02-10 00:05:36,471 (trainer:762) INFO: 39epoch:train:5301-5400batch: iter_time=2.809e-04, forward_time=0.417, loss_ctc=51.391, loss_att=47.675, acc=0.765, loss=48.790, backward_time=0.324, grad_norm=52.090, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.615e-04, train_time=1.290
+[gpub011:0/16] 2024-02-10 00:07:56,038 (trainer:762) INFO: 39epoch:train:5401-5500batch: iter_time=8.928e-05, forward_time=0.290, loss_ctc=42.493, loss_att=39.968, acc=0.772, loss=40.725, backward_time=0.294, grad_norm=40.591, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.615e-04, train_time=1.394
+[gpub011:0/16] 2024-02-10 00:10:06,573 (trainer:762) INFO: 39epoch:train:5501-5600batch: iter_time=8.236e-05, forward_time=0.293, loss_ctc=48.590, loss_att=51.113, acc=0.742, loss=50.356, backward_time=0.298, grad_norm=44.894, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.615e-04, train_time=1.306
+[gpub011:0/16] 2024-02-10 00:12:17,014 (trainer:762) INFO: 39epoch:train:5601-5700batch: iter_time=8.987e-05, forward_time=0.409, loss_ctc=39.515, loss_att=41.704, acc=0.784, loss=41.047, backward_time=0.338, grad_norm=38.091, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.614e-04, train_time=1.303
+[gpub011:0/16] 2024-02-10 00:14:38,078 (trainer:762) INFO: 39epoch:train:5701-5800batch: iter_time=8.660e-05, forward_time=0.292, loss_ctc=46.401, loss_att=45.890, acc=0.769, loss=46.043, backward_time=0.298, grad_norm=39.626, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.614e-04, train_time=1.410
+[gpub011:0/16] 2024-02-10 00:16:48,501 (trainer:762) INFO: 39epoch:train:5801-5900batch: iter_time=8.497e-05, forward_time=0.291, loss_ctc=46.065, loss_att=37.028, acc=0.767, loss=39.739, backward_time=0.295, grad_norm=42.757, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.614e-04, train_time=1.305
+[gpub011:0/16] 2024-02-10 00:18:57,309 (trainer:762) INFO: 39epoch:train:5901-6000batch: iter_time=8.564e-05, forward_time=0.291, loss_ctc=47.108, loss_att=46.423, acc=0.765, loss=46.628, backward_time=0.297, grad_norm=50.020, clip=100.000, loss_scale=7.425e+33, optim_step_time=0.093, optim0_lr0=1.614e-04, train_time=1.288
+[gpub011:0/16] 2024-02-10 00:21:20,195 (trainer:762) INFO: 39epoch:train:6001-6100batch: iter_time=8.294e-05, forward_time=0.372, loss_ctc=38.293, loss_att=35.289, acc=0.776, loss=36.190, backward_time=0.400, grad_norm=38.596, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.614e-04, train_time=1.428
+[gpub011:0/16] 2024-02-10 00:22:00,647 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 00:23:33,256 (trainer:762) INFO: 39epoch:train:6101-6200batch: iter_time=8.357e-05, forward_time=0.291, loss_ctc=40.440, loss_att=44.462, acc=0.762, loss=43.256, backward_time=0.296, grad_norm=37.909, clip=100.000, loss_scale=7.080e+33, optim_step_time=0.093, optim0_lr0=1.614e-04, train_time=1.331
+[gpub011:0/16] 2024-02-10 00:24:50,319 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-10 00:25:09,939 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 00:25:13,520 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9860c8b3d0>)
+[gpub011:0/16] 2024-02-10 00:25:13,520 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-10 00:25:13,524 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 00:31:11,232 (trainer:762) INFO: 39epoch:train:6201-6300batch: iter_time=3.334, forward_time=0.288, loss_ctc=43.378, loss_att=38.085, acc=0.791, loss=39.673, backward_time=0.295, grad_norm=39.915, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.614e-04, train_time=4.580
+[gpub011:0/16] 2024-02-10 00:33:12,837 (trainer:762) INFO: 39epoch:train:6301-6400batch: iter_time=8.088e-05, forward_time=0.355, loss_ctc=43.425, loss_att=45.955, acc=0.757, loss=45.196, backward_time=0.332, grad_norm=44.629, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.613e-04, train_time=1.215
+[gpub011:0/16] 2024-02-10 00:35:29,581 (trainer:762) INFO: 39epoch:train:6401-6500batch: iter_time=8.127e-05, forward_time=0.291, loss_ctc=46.433, loss_att=42.171, acc=0.780, loss=43.450, backward_time=0.298, grad_norm=39.177, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.613e-04, train_time=1.367
+[gpub011:0/16] 2024-02-10 00:37:46,088 (trainer:762) INFO: 39epoch:train:6501-6600batch: iter_time=8.426e-05, forward_time=0.291, loss_ctc=56.168, loss_att=55.579, acc=0.749, loss=55.756, backward_time=0.298, grad_norm=52.556, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.613e-04, train_time=1.365
+[gpub011:0/16] 2024-02-10 00:40:13,868 (trainer:762) INFO: 39epoch:train:6601-6700batch: iter_time=3.600e-04, forward_time=0.407, loss_ctc=48.826, loss_att=46.997, acc=0.754, loss=47.546, backward_time=0.319, grad_norm=46.793, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.127, optim0_lr0=1.613e-04, train_time=1.478
+[gpub011:0/16] 2024-02-10 00:42:27,044 (trainer:762) INFO: 39epoch:train:6701-6800batch: iter_time=8.934e-05, forward_time=0.289, loss_ctc=42.549, loss_att=41.777, acc=0.758, loss=42.008, backward_time=0.294, grad_norm=37.333, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.613e-04, train_time=1.331
+[gpub011:0/16] 2024-02-10 00:44:29,006 (trainer:762) INFO: 39epoch:train:6801-6900batch: iter_time=8.788e-05, forward_time=0.290, loss_ctc=45.177, loss_att=49.419, acc=0.753, loss=48.147, backward_time=0.297, grad_norm=41.830, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.613e-04, train_time=1.219
+[gpub011:0/16] 2024-02-10 00:47:05,905 (trainer:762) INFO: 39epoch:train:6901-7000batch: iter_time=8.965e-05, forward_time=0.444, loss_ctc=45.801, loss_att=47.374, acc=0.770, loss=46.902, backward_time=0.331, grad_norm=39.927, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.613e-04, train_time=1.569
+[gpub011:0/16] 2024-02-10 00:49:11,623 (trainer:762) INFO: 39epoch:train:7001-7100batch: iter_time=8.766e-05, forward_time=0.289, loss_ctc=40.005, loss_att=36.195, acc=0.758, loss=37.338, backward_time=0.294, grad_norm=37.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.612e-04, train_time=1.257
+[gpub011:0/16] 2024-02-10 00:51:09,870 (trainer:762) INFO: 39epoch:train:7101-7200batch: iter_time=8.524e-05, forward_time=0.291, loss_ctc=50.013, loss_att=45.495, acc=0.766, loss=46.850, backward_time=0.297, grad_norm=55.571, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.612e-04, train_time=1.182
+[gpub011:0/16] 2024-02-10 00:53:42,890 (trainer:762) INFO: 39epoch:train:7201-7300batch: iter_time=8.686e-05, forward_time=0.288, loss_ctc=38.310, loss_att=34.873, acc=0.775, loss=35.904, backward_time=0.294, grad_norm=36.610, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.612e-04, train_time=1.530
+[gpub011:0/16] 2024-02-10 00:55:53,324 (trainer:762) INFO: 39epoch:train:7301-7400batch: iter_time=8.558e-05, forward_time=0.398, loss_ctc=39.837, loss_att=38.634, acc=0.762, loss=38.995, backward_time=0.325, grad_norm=40.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.612e-04, train_time=1.305
+[gpub011:0/16] 2024-02-10 00:58:11,508 (trainer:762) INFO: 39epoch:train:7401-7500batch: iter_time=8.506e-05, forward_time=0.289, loss_ctc=40.583, loss_att=43.676, acc=0.774, loss=42.748, backward_time=0.296, grad_norm=38.792, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.612e-04, train_time=1.382
+[gpub011:0/16] 2024-02-10 00:58:31,536 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-10 00:58:51,268 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 00:58:55,082 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9860c4b5b0>)
+[gpub011:0/16] 2024-02-10 00:58:55,082 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-10 00:58:55,085 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 01:05:38,160 (trainer:762) INFO: 39epoch:train:7501-7600batch: iter_time=3.287, forward_time=0.358, loss_ctc=49.898, loss_att=45.335, acc=0.761, loss=46.704, backward_time=0.312, grad_norm=43.212, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.612e-04, train_time=4.466
+[gpub011:0/16] 2024-02-10 01:07:41,913 (trainer:762) INFO: 39epoch:train:7601-7700batch: iter_time=8.489e-05, forward_time=0.291, loss_ctc=43.472, loss_att=38.062, acc=0.782, loss=39.685, backward_time=0.297, grad_norm=39.839, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.612e-04, train_time=1.237
+[gpub011:0/16] 2024-02-10 01:10:18,513 (trainer:762) INFO: 39epoch:train:7701-7800batch: iter_time=8.651e-05, forward_time=0.400, loss_ctc=50.287, loss_att=53.690, acc=0.755, loss=52.669, backward_time=0.316, grad_norm=53.694, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.611e-04, train_time=1.566
+[gpub011:0/16] 2024-02-10 01:12:09,310 (trainer:762) INFO: 39epoch:train:7801-7900batch: iter_time=8.750e-05, forward_time=0.289, loss_ctc=51.472, loss_att=46.125, acc=0.756, loss=47.729, backward_time=0.295, grad_norm=48.009, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.611e-04, train_time=1.108
+[gpub011:0/16] 2024-02-10 01:14:25,265 (trainer:762) INFO: 39epoch:train:7901-8000batch: iter_time=8.317e-05, forward_time=0.289, loss_ctc=42.377, loss_att=39.334, acc=0.769, loss=40.247, backward_time=0.295, grad_norm=39.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.611e-04, train_time=1.359
+[gpub011:0/16] 2024-02-10 01:16:56,588 (trainer:762) INFO: 39epoch:train:8001-8100batch: iter_time=8.065e-05, forward_time=0.341, loss_ctc=48.311, loss_att=51.337, acc=0.732, loss=50.429, backward_time=0.385, grad_norm=45.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.611e-04, train_time=1.513
+[gpub011:0/16] 2024-02-10 01:19:05,043 (trainer:762) INFO: 39epoch:train:8101-8200batch: iter_time=8.332e-05, forward_time=0.288, loss_ctc=39.574, loss_att=41.909, acc=0.778, loss=41.209, backward_time=0.295, grad_norm=39.146, clip=100.000, loss_scale=8.463e+33, optim_step_time=0.093, optim0_lr0=1.611e-04, train_time=1.284
+[gpub011:0/16] 2024-02-10 01:21:06,008 (trainer:762) INFO: 39epoch:train:8201-8300batch: iter_time=8.081e-05, forward_time=0.291, loss_ctc=45.816, loss_att=46.059, acc=0.763, loss=45.986, backward_time=0.298, grad_norm=49.489, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.611e-04, train_time=1.210
+[gpub011:0/16] 2024-02-10 01:23:50,492 (trainer:762) INFO: 39epoch:train:8301-8400batch: iter_time=4.294e-04, forward_time=0.399, loss_ctc=45.325, loss_att=36.954, acc=0.764, loss=39.466, backward_time=0.317, grad_norm=42.901, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.111, optim0_lr0=1.611e-04, train_time=1.645
+[gpub011:0/16] 2024-02-10 01:25:39,100 (trainer:762) INFO: 39epoch:train:8401-8500batch: iter_time=8.310e-05, forward_time=0.290, loss_ctc=45.823, loss_att=45.348, acc=0.764, loss=45.491, backward_time=0.297, grad_norm=50.671, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.611e-04, train_time=1.086
+[gpub011:0/16] 2024-02-10 01:27:42,999 (trainer:762) INFO: 39epoch:train:8501-8600batch: iter_time=8.115e-05, forward_time=0.290, loss_ctc=38.212, loss_att=34.960, acc=0.774, loss=35.935, backward_time=0.294, grad_norm=36.365, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.610e-04, train_time=1.238
+[gpub011:0/16] 2024-02-10 01:30:22,866 (trainer:762) INFO: 39epoch:train:8601-8700batch: iter_time=8.865e-05, forward_time=0.331, loss_ctc=40.436, loss_att=44.567, acc=0.749, loss=43.328, backward_time=0.351, grad_norm=38.036, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.610e-04, train_time=1.599
+[gpub011:0/16] 2024-02-10 01:31:41,117 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-10 01:32:00,530 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 01:32:04,106 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f985f53bcd0>)
+[gpub011:0/16] 2024-02-10 01:32:04,106 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-10 01:32:04,165 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 01:38:09,898 (trainer:762) INFO: 39epoch:train:8701-8800batch: iter_time=3.293, forward_time=0.288, loss_ctc=43.622, loss_att=38.667, acc=0.787, loss=40.154, backward_time=0.295, grad_norm=38.703, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.610e-04, train_time=4.670
+[gpub011:0/16] 2024-02-10 01:40:31,208 (trainer:762) INFO: 39epoch:train:8801-8900batch: iter_time=8.538e-05, forward_time=0.290, loss_ctc=43.122, loss_att=45.513, acc=0.769, loss=44.796, backward_time=0.297, grad_norm=45.582, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.610e-04, train_time=1.413
+[gpub011:0/16] 2024-02-10 01:43:08,336 (trainer:762) INFO: 39epoch:train:8901-9000batch: iter_time=1.016e-04, forward_time=0.384, loss_ctc=46.232, loss_att=42.040, acc=0.787, loss=43.298, backward_time=0.385, grad_norm=37.836, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.610e-04, train_time=1.571
+[gpub011:0/16] 2024-02-10 01:44:56,975 (trainer:762) INFO: 39epoch:train:9001-9100batch: iter_time=9.926e-05, forward_time=0.292, loss_ctc=54.939, loss_att=53.858, acc=0.758, loss=54.182, backward_time=0.300, grad_norm=52.518, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.610e-04, train_time=1.086
+[gpub011:0/16] 2024-02-10 01:47:04,743 (trainer:762) INFO: 39epoch:train:9101-9200batch: iter_time=9.150e-05, forward_time=0.292, loss_ctc=48.799, loss_att=47.748, acc=0.766, loss=48.063, backward_time=0.298, grad_norm=46.971, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.610e-04, train_time=1.277
+[gpub011:0/16] 2024-02-10 01:49:47,861 (trainer:762) INFO: 39epoch:train:9201-9300batch: iter_time=9.298e-05, forward_time=0.330, loss_ctc=41.723, loss_att=42.071, acc=0.758, loss=41.967, backward_time=0.360, grad_norm=39.474, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.609e-04, train_time=1.631
+[gpub011:0/16] 2024-02-10 01:51:57,816 (trainer:762) INFO: 39epoch:train:9301-9400batch: iter_time=9.857e-05, forward_time=0.294, loss_ctc=45.408, loss_att=49.185, acc=0.761, loss=48.051, backward_time=0.298, grad_norm=42.634, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.609e-04, train_time=1.299
+[gpub011:0/16] 2024-02-10 01:54:08,928 (trainer:762) INFO: 39epoch:train:9401-9500batch: iter_time=8.879e-05, forward_time=0.294, loss_ctc=45.851, loss_att=47.388, acc=0.780, loss=46.927, backward_time=0.301, grad_norm=40.033, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.609e-04, train_time=1.311
+[gpub011:0/16] 2024-02-10 01:56:36,297 (trainer:762) INFO: 39epoch:train:9501-9600batch: iter_time=8.394e-05, forward_time=0.289, loss_ctc=40.140, loss_att=35.824, acc=0.769, loss=37.118, backward_time=0.294, grad_norm=41.767, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.609e-04, train_time=1.473
+[gpub011:0/16] 2024-02-10 01:58:46,000 (trainer:762) INFO: 39epoch:train:9601-9700batch: iter_time=9.301e-05, forward_time=0.382, loss_ctc=49.776, loss_att=46.609, acc=0.762, loss=47.559, backward_time=0.355, grad_norm=56.975, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.609e-04, train_time=1.297
+[gpub011:0/16] 2024-02-10 02:00:52,756 (trainer:762) INFO: 39epoch:train:9701-9800batch: iter_time=8.300e-05, forward_time=0.290, loss_ctc=38.582, loss_att=35.477, acc=0.782, loss=36.409, backward_time=0.297, grad_norm=36.842, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.609e-04, train_time=1.267
+[gpub011:0/16] 2024-02-10 02:03:22,859 (trainer:762) INFO: 39epoch:train:9801-9900batch: iter_time=9.055e-05, forward_time=0.289, loss_ctc=40.105, loss_att=39.095, acc=0.768, loss=39.398, backward_time=0.295, grad_norm=37.928, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.609e-04, train_time=1.501
+[gpub011:0/16] 2024-02-10 02:05:39,255 (trainer:762) INFO: 39epoch:train:9901-10000batch: iter_time=8.521e-05, forward_time=0.358, loss_ctc=40.820, loss_att=43.723, acc=0.786, loss=42.852, backward_time=0.355, grad_norm=36.299, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.608e-04, train_time=1.364
+[gpub011:0/16] 2024-02-10 02:05:59,284 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-10 02:06:19,072 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 02:06:22,642 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97e3919fc0>)
+[gpub011:0/16] 2024-02-10 02:06:22,642 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-10 02:06:22,673 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 02:13:18,313 (trainer:762) INFO: 39epoch:train:10001-10100batch: iter_time=3.287, forward_time=0.289, loss_ctc=48.725, loss_att=46.264, acc=0.758, loss=47.002, backward_time=0.295, grad_norm=45.010, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.608e-04, train_time=4.590
+[gpub011:0/16] 2024-02-10 02:14:50,780 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 02:15:55,528 (trainer:762) INFO: 39epoch:train:10101-10200batch: iter_time=8.386e-05, forward_time=0.290, loss_ctc=43.086, loss_att=38.495, acc=0.781, loss=39.872, backward_time=0.296, grad_norm=40.259, clip=100.000, loss_scale=1.374e+34, optim_step_time=0.095, optim0_lr0=1.608e-04, train_time=1.572
+[gpub011:0/16] 2024-02-10 02:18:03,366 (trainer:762) INFO: 39epoch:train:10201-10300batch: iter_time=3.089e-04, forward_time=0.359, loss_ctc=50.182, loss_att=53.452, acc=0.756, loss=52.471, backward_time=0.343, grad_norm=52.409, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.608e-04, train_time=1.278
+[gpub011:0/16] 2024-02-10 02:20:12,419 (trainer:762) INFO: 39epoch:train:10301-10400batch: iter_time=9.213e-05, forward_time=0.291, loss_ctc=51.420, loss_att=46.114, acc=0.757, loss=47.706, backward_time=0.298, grad_norm=48.864, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.608e-04, train_time=1.290
+[gpub011:0/16] 2024-02-10 02:22:37,210 (trainer:762) INFO: 39epoch:train:10401-10500batch: iter_time=9.090e-05, forward_time=0.289, loss_ctc=41.839, loss_att=39.505, acc=0.769, loss=40.205, backward_time=0.296, grad_norm=39.680, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.608e-04, train_time=1.448
+[gpub011:0/16] 2024-02-10 02:25:03,558 (trainer:762) INFO: 39epoch:train:10501-10600batch: iter_time=9.009e-05, forward_time=0.406, loss_ctc=47.637, loss_att=51.231, acc=0.733, loss=50.153, backward_time=0.319, grad_norm=79.526, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.608e-04, train_time=1.463
+[gpub011:0/16] 2024-02-10 02:25:37,726 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 02:26:56,398 (trainer:762) INFO: 39epoch:train:10601-10700batch: iter_time=9.002e-05, forward_time=0.291, loss_ctc=39.763, loss_att=42.403, acc=0.778, loss=41.611, backward_time=0.297, grad_norm=37.898, clip=100.000, loss_scale=6.713e+33, optim_step_time=0.093, optim0_lr0=1.607e-04, train_time=1.128
+[gpub011:0/16] 2024-02-10 02:29:39,552 (trainer:762) INFO: 39epoch:train:10701-10800batch: iter_time=8.437e-05, forward_time=0.294, loss_ctc=46.061, loss_att=46.213, acc=0.762, loss=46.167, backward_time=0.296, grad_norm=40.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.607e-04, train_time=1.631
+[gpub011:0/16] 2024-02-10 02:31:28,105 (trainer:762) INFO: 39epoch:train:10801-10900batch: iter_time=8.936e-05, forward_time=0.289, loss_ctc=45.310, loss_att=36.823, acc=0.765, loss=39.369, backward_time=0.295, grad_norm=43.039, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.607e-04, train_time=1.085
+[gpub011:0/16] 2024-02-10 02:33:57,266 (trainer:762) INFO: 39epoch:train:10901-11000batch: iter_time=9.268e-05, forward_time=0.421, loss_ctc=45.950, loss_att=45.110, acc=0.764, loss=45.362, backward_time=0.339, grad_norm=51.738, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.607e-04, train_time=1.491
+[gpub011:0/16] 2024-02-10 02:36:20,823 (trainer:762) INFO: 39epoch:train:11001-11100batch: iter_time=8.907e-05, forward_time=0.287, loss_ctc=37.805, loss_att=34.720, acc=0.775, loss=35.646, backward_time=0.291, grad_norm=37.844, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.607e-04, train_time=1.435
+[gpub011:0/16] 2024-02-10 02:38:28,657 (trainer:762) INFO: 39epoch:train:11101-11200batch: iter_time=8.846e-05, forward_time=0.289, loss_ctc=39.463, loss_att=43.999, acc=0.752, loss=42.638, backward_time=0.295, grad_norm=37.187, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.607e-04, train_time=1.278
+[gpub011:0/16] 2024-02-10 02:40:08,240 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-10 02:40:27,728 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 02:40:31,302 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f985f6ec310>)
+[gpub011:0/16] 2024-02-10 02:40:31,302 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-10 02:40:31,305 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 02:46:20,494 (trainer:762) INFO: 39epoch:train:11201-11300batch: iter_time=3.224, forward_time=0.403, loss_ctc=43.297, loss_att=38.125, acc=0.789, loss=39.677, backward_time=0.317, grad_norm=39.042, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.607e-04, train_time=4.718
+[gpub011:0/16] 2024-02-10 02:47:51,281 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 02:48:08,780 (trainer:762) INFO: 39epoch:train:11301-11400batch: iter_time=9.003e-05, forward_time=0.291, loss_ctc=43.075, loss_att=45.003, acc=0.773, loss=44.425, backward_time=0.298, grad_norm=40.920, clip=100.000, loss_scale=4.773e+33, optim_step_time=0.093, optim0_lr0=1.606e-04, train_time=1.083
+[gpub011:0/16] 2024-02-10 02:50:34,157 (trainer:762) INFO: 39epoch:train:11401-11500batch: iter_time=9.390e-05, forward_time=0.293, loss_ctc=45.888, loss_att=41.732, acc=0.788, loss=42.979, backward_time=0.298, grad_norm=39.903, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.606e-04, train_time=1.453
+[gpub011:0/16] 2024-02-10 02:53:08,851 (trainer:762) INFO: 39epoch:train:11501-11600batch: iter_time=9.612e-05, forward_time=0.389, loss_ctc=55.454, loss_att=52.495, acc=0.764, loss=53.383, backward_time=0.338, grad_norm=52.833, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.606e-04, train_time=1.547
+[gpub011:0/16] 2024-02-10 02:55:22,089 (trainer:762) INFO: 39epoch:train:11601-11700batch: iter_time=8.145e-05, forward_time=0.291, loss_ctc=48.482, loss_att=48.088, acc=0.766, loss=48.207, backward_time=0.298, grad_norm=46.449, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.606e-04, train_time=1.332
+[gpub011:0/16] 2024-02-10 02:57:26,040 (trainer:762) INFO: 39epoch:train:11701-11800batch: iter_time=9.534e-05, forward_time=0.293, loss_ctc=42.008, loss_att=42.172, acc=0.760, loss=42.123, backward_time=0.296, grad_norm=39.371, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.606e-04, train_time=1.239
+[gpub011:0/16] 2024-02-10 03:00:04,830 (trainer:762) INFO: 39epoch:train:11801-11900batch: iter_time=8.849e-05, forward_time=0.368, loss_ctc=45.263, loss_att=49.403, acc=0.758, loss=48.161, backward_time=0.328, grad_norm=44.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.606e-04, train_time=1.588
+[gpub011:0/16] 2024-02-10 03:02:11,900 (trainer:762) INFO: 39epoch:train:11901-12000batch: iter_time=9.022e-05, forward_time=0.294, loss_ctc=45.444, loss_att=47.322, acc=0.780, loss=46.758, backward_time=0.300, grad_norm=44.455, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.606e-04, train_time=1.271
+[gpub011:0/16] 2024-02-10 03:04:42,785 (trainer:762) INFO: 39epoch:train:12001-12100batch: iter_time=9.363e-05, forward_time=0.290, loss_ctc=39.766, loss_att=35.540, acc=0.769, loss=36.808, backward_time=0.293, grad_norm=38.884, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.606e-04, train_time=1.508
+[gpub011:0/16] 2024-02-10 03:06:40,270 (trainer:762) INFO: 39epoch:train:12101-12200batch: iter_time=9.078e-05, forward_time=0.292, loss_ctc=49.272, loss_att=45.439, acc=0.768, loss=46.589, backward_time=0.297, grad_norm=51.836, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.605e-04, train_time=1.175
+[gpub011:0/16] 2024-02-10 03:09:17,512 (trainer:762) INFO: 39epoch:train:12201-12300batch: iter_time=8.569e-05, forward_time=0.394, loss_ctc=38.515, loss_att=34.872, acc=0.785, loss=35.965, backward_time=0.318, grad_norm=34.650, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.605e-04, train_time=1.572
+[gpub011:0/16] 2024-02-10 03:11:18,619 (trainer:762) INFO: 39epoch:train:12301-12400batch: iter_time=8.624e-05, forward_time=0.288, loss_ctc=39.892, loss_att=38.967, acc=0.768, loss=39.245, backward_time=0.294, grad_norm=37.626, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.605e-04, train_time=1.211
+[gpub011:0/16] 2024-02-10 03:13:30,272 (trainer:762) INFO: 39epoch:train:12401-12500batch: iter_time=8.769e-05, forward_time=0.293, loss_ctc=40.496, loss_att=42.266, acc=0.790, loss=41.735, backward_time=0.298, grad_norm=35.976, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.605e-04, train_time=1.316
+[gpub011:0/16] 2024-02-10 03:13:50,301 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-10 03:14:10,338 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 03:14:13,929 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9889837790>)
+[gpub011:0/16] 2024-02-10 03:14:13,929 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-10 03:14:13,933 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 03:20:50,611 (trainer:762) INFO: 39epoch:train:12501-12600batch: iter_time=3.145, forward_time=0.385, loss_ctc=49.039, loss_att=46.908, acc=0.757, loss=47.547, backward_time=0.315, grad_norm=45.026, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.605e-04, train_time=4.403
+[gpub011:0/16] 2024-02-10 03:23:05,260 (trainer:762) INFO: 39epoch:train:12601-12700batch: iter_time=8.251e-05, forward_time=0.291, loss_ctc=42.780, loss_att=38.192, acc=0.782, loss=39.569, backward_time=0.295, grad_norm=38.984, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.605e-04, train_time=1.346
+[gpub011:0/16] 2024-02-10 03:25:33,355 (trainer:762) INFO: 39epoch:train:12701-12800batch: iter_time=8.782e-05, forward_time=0.324, loss_ctc=49.680, loss_att=53.779, acc=0.756, loss=52.549, backward_time=0.347, grad_norm=50.333, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.605e-04, train_time=1.481
+[gpub011:0/16] 2024-02-10 03:27:35,522 (trainer:762) INFO: 39epoch:train:12801-12900batch: iter_time=9.367e-05, forward_time=0.307, loss_ctc=50.877, loss_att=46.212, acc=0.756, loss=47.611, backward_time=0.307, grad_norm=50.083, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.604e-04, train_time=1.221
+[gpub011:0/16] 2024-02-10 03:30:08,251 (trainer:762) INFO: 39epoch:train:12901-13000batch: iter_time=9.151e-05, forward_time=0.313, loss_ctc=42.000, loss_att=39.024, acc=0.771, loss=39.917, backward_time=0.293, grad_norm=38.674, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.604e-04, train_time=1.527
+[gpub011:0/16] 2024-02-10 03:31:57,688 (trainer:762) INFO: 39epoch:train:13001-13100batch: iter_time=9.072e-05, forward_time=0.292, loss_ctc=48.088, loss_att=50.554, acc=0.736, loss=49.814, backward_time=0.299, grad_norm=46.664, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.604e-04, train_time=1.094
+[gpub011:0/16] 2024-02-10 03:34:34,701 (trainer:762) INFO: 39epoch:train:13101-13200batch: iter_time=9.594e-05, forward_time=0.355, loss_ctc=39.473, loss_att=42.287, acc=0.777, loss=41.443, backward_time=0.334, grad_norm=37.124, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.604e-04, train_time=1.569
+[gpub011:0/16] 2024-02-10 03:36:44,400 (trainer:762) INFO: 39epoch:train:13201-13300batch: iter_time=8.518e-05, forward_time=0.292, loss_ctc=45.794, loss_att=46.033, acc=0.761, loss=45.961, backward_time=0.299, grad_norm=40.116, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.604e-04, train_time=1.298
+[gpub011:0/16] 2024-02-10 03:38:50,840 (trainer:762) INFO: 39epoch:train:13301-13400batch: iter_time=8.645e-05, forward_time=0.288, loss_ctc=45.277, loss_att=36.886, acc=0.767, loss=39.403, backward_time=0.294, grad_norm=41.921, clip=100.000, loss_scale=3.012e+33, optim_step_time=0.093, optim0_lr0=1.604e-04, train_time=1.264
+[gpub011:0/16] 2024-02-10 03:41:23,067 (trainer:762) INFO: 39epoch:train:13401-13500batch: iter_time=2.546e-04, forward_time=0.371, loss_ctc=45.652, loss_att=45.442, acc=0.764, loss=45.505, backward_time=0.321, grad_norm=52.128, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.604e-04, train_time=1.522
+[gpub011:0/16] 2024-02-10 03:43:31,018 (trainer:762) INFO: 39epoch:train:13501-13600batch: iter_time=8.595e-05, forward_time=0.304, loss_ctc=37.617, loss_att=34.310, acc=0.777, loss=35.302, backward_time=0.298, grad_norm=36.949, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.603e-04, train_time=1.278
+[gpub011:0/16] 2024-02-10 03:45:36,008 (trainer:762) INFO: 39epoch:train:13601-13700batch: iter_time=9.307e-05, forward_time=0.292, loss_ctc=39.690, loss_att=44.270, acc=0.750, loss=42.896, backward_time=0.296, grad_norm=39.588, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.603e-04, train_time=1.251
+[gpub011:0/16] 2024-02-10 03:47:04,174 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-10 03:47:24,270 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 03:47:27,858 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98131cff70>)
+[gpub011:0/16] 2024-02-10 03:47:27,858 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-10 03:47:27,863 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 03:53:25,625 (trainer:762) INFO: 39epoch:train:13701-13800batch: iter_time=3.351, forward_time=0.369, loss_ctc=43.168, loss_att=37.323, acc=0.791, loss=39.077, backward_time=0.309, grad_norm=39.851, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.603e-04, train_time=4.696
+[gpub011:0/16] 2024-02-10 03:55:29,821 (trainer:762) INFO: 39epoch:train:13801-13900batch: iter_time=7.898e-05, forward_time=0.289, loss_ctc=42.685, loss_att=43.863, acc=0.762, loss=43.510, backward_time=0.294, grad_norm=42.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.603e-04, train_time=1.241
+[gpub011:0/16] 2024-02-10 03:57:55,214 (trainer:762) INFO: 39epoch:train:13901-14000batch: iter_time=8.020e-05, forward_time=0.292, loss_ctc=46.580, loss_att=41.624, acc=0.782, loss=43.111, backward_time=0.297, grad_norm=39.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.603e-04, train_time=1.455
+[gpub011:0/16] 2024-02-10 04:00:08,434 (trainer:762) INFO: 39epoch:train:14001-14100batch: iter_time=8.190e-05, forward_time=0.374, loss_ctc=55.530, loss_att=54.878, acc=0.753, loss=55.074, backward_time=0.319, grad_norm=54.883, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.603e-04, train_time=1.332
+[gpub011:0/16] 2024-02-10 04:02:41,470 (trainer:762) INFO: 39epoch:train:14101-14200batch: iter_time=8.147e-05, forward_time=0.290, loss_ctc=47.886, loss_att=45.913, acc=0.758, loss=46.505, backward_time=0.294, grad_norm=47.388, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.603e-04, train_time=1.529
+[gpub011:0/16] 2024-02-10 04:04:36,726 (trainer:762) INFO: 39epoch:train:14201-14300batch: iter_time=7.910e-05, forward_time=0.291, loss_ctc=42.109, loss_att=42.358, acc=0.757, loss=42.283, backward_time=0.295, grad_norm=39.799, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.603e-04, train_time=1.154
+[gpub011:0/16] 2024-02-10 04:07:04,421 (trainer:762) INFO: 39epoch:train:14301-14400batch: iter_time=8.043e-05, forward_time=0.358, loss_ctc=44.918, loss_att=49.361, acc=0.751, loss=48.028, backward_time=0.354, grad_norm=42.519, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.602e-04, train_time=1.477
+[gpub011:0/16] 2024-02-10 04:09:26,568 (trainer:762) INFO: 39epoch:train:14401-14500batch: iter_time=8.183e-05, forward_time=0.293, loss_ctc=45.725, loss_att=47.083, acc=0.773, loss=46.675, backward_time=0.298, grad_norm=41.187, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.602e-04, train_time=1.420
+[gpub011:0/16] 2024-02-10 04:11:37,801 (trainer:762) INFO: 39epoch:train:14501-14600batch: iter_time=0.001, forward_time=0.373, loss_ctc=39.647, loss_att=36.005, acc=0.760, loss=37.098, backward_time=0.365, grad_norm=38.810, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.602e-04, train_time=1.314
+[gpub011:0/16] 2024-02-10 04:13:54,282 (trainer:762) INFO: 39epoch:train:14601-14700batch: iter_time=8.217e-05, forward_time=0.290, loss_ctc=49.896, loss_att=46.064, acc=0.763, loss=47.214, backward_time=0.295, grad_norm=53.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.602e-04, train_time=1.364
+[gpub011:0/16] 2024-02-10 04:16:07,508 (trainer:762) INFO: 39epoch:train:14701-14800batch: iter_time=8.027e-05, forward_time=0.289, loss_ctc=38.358, loss_att=34.267, acc=0.777, loss=35.494, backward_time=0.292, grad_norm=36.361, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.602e-04, train_time=1.333
+[gpub011:0/16] 2024-02-10 04:18:28,426 (trainer:762) INFO: 39epoch:train:14801-14900batch: iter_time=8.013e-05, forward_time=0.402, loss_ctc=39.452, loss_att=38.392, acc=0.766, loss=38.710, backward_time=0.318, grad_norm=36.665, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.602e-04, train_time=1.409
+[gpub011:0/16] 2024-02-10 04:20:31,345 (trainer:762) INFO: 39epoch:train:14901-15000batch: iter_time=7.903e-05, forward_time=0.291, loss_ctc=40.396, loss_att=43.137, acc=0.777, loss=42.315, backward_time=0.297, grad_norm=38.726, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.602e-04, train_time=1.229
+[gpub011:0/16] 2024-02-10 04:58:50,649 (trainer:361) INFO: 39epoch results: [train] iter_time=0.258, forward_time=0.323, loss_ctc=45.121, loss_att=43.705, acc=0.766, loss=44.130, backward_time=0.309, grad_norm=43.358, clip=100.000, loss_scale=6.470e+33, optim_step_time=0.096, optim0_lr0=1.612e-04, train_time=1.595, time=6 hours, 39 minutes and 5.81 seconds, total_count=615000, gpu_max_cached_mem_GB=43.805, [valid] loss_ctc=36.305, cer_ctc=0.183, loss_att=39.322, acc=0.689, cer=0.319, wer=0.988, loss=38.417, time=37 minutes and 54.25 seconds, total_count=191511, gpu_max_cached_mem_GB=43.805
+[gpub011:0/16] 2024-02-10 04:59:00,298 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-10 04:59:00,411 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/31epoch.pth
+[gpub011:0/16] 2024-02-10 04:59:00,411 (trainer:290) INFO: 40/45epoch started. Estimated time to finish: 1 day, 20 hours and 9 minutes
+[gpub011:0/16] 2024-02-10 04:59:00,421 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-10 04:59:18,877 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 04:59:22,323 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8fb7820e50>)
+[gpub011:0/16] 2024-02-10 04:59:22,323 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-10 04:59:22,326 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 05:05:48,189 (trainer:762) INFO: 40epoch:train:1-100batch: iter_time=2.960, forward_time=0.321, loss_ctc=47.859, loss_att=39.866, acc=0.778, loss=42.264, backward_time=0.302, grad_norm=43.408, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.601e-04, train_time=4.077
+[gpub011:0/16] 2024-02-10 05:07:44,743 (trainer:762) INFO: 40epoch:train:101-200batch: iter_time=8.501e-05, forward_time=0.290, loss_ctc=44.612, loss_att=49.821, acc=0.743, loss=48.258, backward_time=0.298, grad_norm=53.263, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.601e-04, train_time=1.165
+[gpub011:0/16] 2024-02-10 05:09:56,898 (trainer:762) INFO: 40epoch:train:201-300batch: iter_time=9.110e-05, forward_time=0.297, loss_ctc=40.028, loss_att=35.508, acc=0.781, loss=36.864, backward_time=0.304, grad_norm=39.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.601e-04, train_time=1.321
+[gpub011:0/16] 2024-02-10 05:12:18,491 (trainer:762) INFO: 40epoch:train:301-400batch: iter_time=9.413e-05, forward_time=0.347, loss_ctc=57.058, loss_att=55.220, acc=0.736, loss=55.771, backward_time=0.356, grad_norm=55.957, clip=100.000, loss_scale=6.023e+33, optim_step_time=0.100, optim0_lr0=1.601e-04, train_time=1.416
+[gpub011:0/16] 2024-02-10 05:14:26,094 (trainer:762) INFO: 40epoch:train:401-500batch: iter_time=9.017e-05, forward_time=0.290, loss_ctc=51.434, loss_att=50.801, acc=0.741, loss=50.991, backward_time=0.296, grad_norm=48.900, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.601e-04, train_time=1.276
+[gpub011:0/16] 2024-02-10 05:16:31,408 (trainer:762) INFO: 40epoch:train:501-600batch: iter_time=9.055e-05, forward_time=0.314, loss_ctc=46.312, loss_att=44.663, acc=0.786, loss=45.158, backward_time=0.313, grad_norm=39.673, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.601e-04, train_time=1.253
+[gpub011:0/16] 2024-02-10 05:19:01,160 (trainer:762) INFO: 40epoch:train:601-700batch: iter_time=8.549e-05, forward_time=0.330, loss_ctc=50.798, loss_att=49.519, acc=0.761, loss=49.902, backward_time=0.333, grad_norm=50.091, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.601e-04, train_time=1.497
+[gpub011:0/16] 2024-02-10 05:21:00,372 (trainer:762) INFO: 40epoch:train:701-800batch: iter_time=8.548e-05, forward_time=0.292, loss_ctc=45.749, loss_att=45.250, acc=0.763, loss=45.400, backward_time=0.299, grad_norm=43.157, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=1.192
+[gpub011:0/16] 2024-02-10 05:21:37,902 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 05:23:28,288 (trainer:762) INFO: 40epoch:train:801-900batch: iter_time=8.541e-05, forward_time=0.379, loss_ctc=45.896, loss_att=41.039, acc=0.769, loss=42.496, backward_time=0.369, grad_norm=41.588, clip=100.000, loss_scale=6.661e+33, optim_step_time=0.100, optim0_lr0=1.600e-04, train_time=1.479
+[gpub011:0/16] 2024-02-10 05:25:28,280 (trainer:762) INFO: 40epoch:train:901-1000batch: iter_time=8.864e-05, forward_time=0.293, loss_ctc=54.202, loss_att=54.408, acc=0.743, loss=54.346, backward_time=0.300, grad_norm=55.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=1.200
+[gpub011:0/16] 2024-02-10 05:27:53,511 (trainer:762) INFO: 40epoch:train:1001-1100batch: iter_time=8.846e-05, forward_time=0.297, loss_ctc=51.177, loss_att=48.684, acc=0.758, loss=49.432, backward_time=0.304, grad_norm=48.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=1.452
+[gpub011:0/16] 2024-02-10 05:30:04,769 (trainer:762) INFO: 40epoch:train:1101-1200batch: iter_time=2.671e-04, forward_time=0.373, loss_ctc=49.649, loss_att=47.832, acc=0.760, loss=48.377, backward_time=0.350, grad_norm=43.589, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.600e-04, train_time=1.312
+[gpub011:0/16] 2024-02-10 05:31:25,943 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-10 05:31:45,508 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 05:31:49,033 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa065aa9f90>)
+[gpub011:0/16] 2024-02-10 05:31:49,033 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-10 05:31:49,036 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 05:37:31,677 (trainer:762) INFO: 40epoch:train:1201-1300batch: iter_time=3.173, forward_time=0.287, loss_ctc=43.020, loss_att=40.580, acc=0.772, loss=41.312, backward_time=0.293, grad_norm=40.874, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=4.469
+[gpub011:0/16] 2024-02-10 05:39:22,219 (trainer:762) INFO: 40epoch:train:1301-1400batch: iter_time=7.902e-05, forward_time=0.291, loss_ctc=45.839, loss_att=47.758, acc=0.755, loss=47.182, backward_time=0.299, grad_norm=47.553, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=1.105
+[gpub011:0/16] 2024-02-10 05:41:50,013 (trainer:762) INFO: 40epoch:train:1401-1500batch: iter_time=8.607e-05, forward_time=0.376, loss_ctc=41.367, loss_att=40.148, acc=0.755, loss=40.514, backward_time=0.315, grad_norm=48.052, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.599e-04, train_time=1.478
+[gpub011:0/16] 2024-02-10 05:43:57,490 (trainer:762) INFO: 40epoch:train:1501-1600batch: iter_time=8.410e-05, forward_time=0.299, loss_ctc=42.765, loss_att=40.158, acc=0.761, loss=40.940, backward_time=0.297, grad_norm=41.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.599e-04, train_time=1.274
+[gpub011:0/16] 2024-02-10 05:45:53,781 (trainer:762) INFO: 40epoch:train:1601-1700batch: iter_time=8.217e-05, forward_time=0.311, loss_ctc=49.072, loss_att=51.427, acc=0.746, loss=50.720, backward_time=0.296, grad_norm=51.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.599e-04, train_time=1.163
+[gpub011:0/16] 2024-02-10 05:48:01,828 (trainer:762) INFO: 40epoch:train:1701-1800batch: iter_time=8.115e-05, forward_time=0.299, loss_ctc=53.802, loss_att=47.505, acc=0.748, loss=49.394, backward_time=0.301, grad_norm=50.240, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.599e-04, train_time=1.280
+[gpub011:0/16] 2024-02-10 05:50:30,627 (trainer:762) INFO: 40epoch:train:1801-1900batch: iter_time=8.987e-05, forward_time=0.355, loss_ctc=49.631, loss_att=48.324, acc=0.766, loss=48.716, backward_time=0.350, grad_norm=44.273, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.599e-04, train_time=1.487
+[gpub011:0/16] 2024-02-10 05:52:39,323 (trainer:762) INFO: 40epoch:train:1901-2000batch: iter_time=8.699e-05, forward_time=0.290, loss_ctc=43.836, loss_att=42.679, acc=0.768, loss=43.026, backward_time=0.296, grad_norm=37.103, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.599e-04, train_time=1.286
+[gpub011:0/16] 2024-02-10 05:54:30,468 (trainer:762) INFO: 40epoch:train:2001-2100batch: iter_time=8.920e-05, forward_time=0.292, loss_ctc=44.581, loss_att=39.988, acc=0.770, loss=41.366, backward_time=0.299, grad_norm=40.405, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.599e-04, train_time=1.112
+[gpub011:0/16] 2024-02-10 05:56:51,817 (trainer:762) INFO: 40epoch:train:2101-2200batch: iter_time=9.374e-05, forward_time=0.307, loss_ctc=47.843, loss_att=45.968, acc=0.758, loss=46.530, backward_time=0.314, grad_norm=45.330, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.599e-04, train_time=1.413
+[gpub011:0/16] 2024-02-10 05:59:06,558 (trainer:762) INFO: 40epoch:train:2201-2300batch: iter_time=2.226e-04, forward_time=0.346, loss_ctc=51.285, loss_att=49.553, acc=0.739, loss=50.073, backward_time=0.310, grad_norm=55.799, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.598e-04, train_time=1.347
+[gpub011:0/16] 2024-02-10 06:01:12,799 (trainer:762) INFO: 40epoch:train:2301-2400batch: iter_time=8.177e-05, forward_time=0.294, loss_ctc=51.448, loss_att=51.157, acc=0.753, loss=51.244, backward_time=0.300, grad_norm=45.269, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.262
+[gpub011:0/16] 2024-02-10 06:03:19,530 (trainer:762) INFO: 40epoch:train:2401-2500batch: iter_time=8.089e-05, forward_time=0.290, loss_ctc=43.890, loss_att=44.249, acc=0.766, loss=44.141, backward_time=0.296, grad_norm=41.920, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.268
+[gpub011:0/16] 2024-02-10 06:03:39,571 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-10 06:03:59,013 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 06:04:02,546 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9aa1988d60>)
+[gpub011:0/16] 2024-02-10 06:04:02,547 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-10 06:04:02,550 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 06:10:22,097 (trainer:762) INFO: 40epoch:train:2501-2600batch: iter_time=2.902, forward_time=0.416, loss_ctc=45.764, loss_att=37.583, acc=0.777, loss=40.037, backward_time=0.313, grad_norm=40.886, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.598e-04, train_time=4.225
+[gpub011:0/16] 2024-02-10 06:12:25,437 (trainer:762) INFO: 40epoch:train:2601-2700batch: iter_time=7.925e-05, forward_time=0.290, loss_ctc=42.692, loss_att=45.899, acc=0.742, loss=44.937, backward_time=0.295, grad_norm=48.220, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.233
+[gpub011:0/16] 2024-02-10 06:14:33,242 (trainer:762) INFO: 40epoch:train:2701-2800batch: iter_time=8.166e-05, forward_time=0.288, loss_ctc=39.367, loss_att=34.455, acc=0.782, loss=35.929, backward_time=0.294, grad_norm=37.537, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.278
+[gpub011:0/16] 2024-02-10 06:16:51,544 (trainer:762) INFO: 40epoch:train:2801-2900batch: iter_time=8.312e-05, forward_time=0.335, loss_ctc=53.517, loss_att=54.230, acc=0.733, loss=54.016, backward_time=0.308, grad_norm=52.410, clip=100.000, loss_scale=8.879e+33, optim_step_time=0.094, optim0_lr0=1.598e-04, train_time=1.383
+[gpub011:0/16] 2024-02-10 06:17:55,290 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 06:19:07,140 (trainer:762) INFO: 40epoch:train:2901-3000batch: iter_time=8.458e-05, forward_time=0.378, loss_ctc=49.251, loss_att=47.635, acc=0.744, loss=48.120, backward_time=0.301, grad_norm=50.665, clip=100.000, loss_scale=7.395e+33, optim_step_time=0.095, optim0_lr0=1.597e-04, train_time=1.356
+[gpub011:0/16] 2024-02-10 06:21:20,686 (trainer:762) INFO: 40epoch:train:3001-3100batch: iter_time=8.486e-05, forward_time=0.292, loss_ctc=45.515, loss_att=44.082, acc=0.780, loss=44.512, backward_time=0.298, grad_norm=39.068, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.335
+[gpub011:0/16] 2024-02-10 06:23:41,170 (trainer:762) INFO: 40epoch:train:3101-3200batch: iter_time=8.225e-05, forward_time=0.290, loss_ctc=49.093, loss_att=47.347, acc=0.760, loss=47.871, backward_time=0.295, grad_norm=43.774, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.404
+[gpub011:0/16] 2024-02-10 06:25:47,756 (trainer:762) INFO: 40epoch:train:3201-3300batch: iter_time=8.370e-05, forward_time=0.376, loss_ctc=44.717, loss_att=44.017, acc=0.765, loss=44.227, backward_time=0.370, grad_norm=41.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.597e-04, train_time=1.267
+[gpub011:0/16] 2024-02-10 06:28:07,374 (trainer:762) INFO: 40epoch:train:3301-3400batch: iter_time=9.391e-05, forward_time=0.290, loss_ctc=43.938, loss_att=39.412, acc=0.766, loss=40.769, backward_time=0.294, grad_norm=39.830, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.597e-04, train_time=1.396
+[gpub011:0/16] 2024-02-10 06:30:15,702 (trainer:762) INFO: 40epoch:train:3401-3500batch: iter_time=9.092e-05, forward_time=0.291, loss_ctc=52.731, loss_att=52.187, acc=0.747, loss=52.350, backward_time=0.298, grad_norm=57.639, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.597e-04, train_time=1.283
+[gpub011:0/16] 2024-02-10 06:32:10,245 (trainer:762) INFO: 40epoch:train:3501-3600batch: iter_time=8.536e-05, forward_time=0.292, loss_ctc=50.160, loss_att=46.981, acc=0.750, loss=47.935, backward_time=0.299, grad_norm=46.091, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.597e-04, train_time=1.144
+[gpub011:0/16] 2024-02-10 06:34:33,519 (trainer:762) INFO: 40epoch:train:3601-3700batch: iter_time=9.227e-05, forward_time=0.419, loss_ctc=48.352, loss_att=46.646, acc=0.764, loss=47.158, backward_time=0.322, grad_norm=42.479, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.597e-04, train_time=1.434
+[gpub011:0/16] 2024-02-10 06:35:51,784 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-10 06:36:11,322 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 06:36:15,091 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9ad87f3d00>)
+[gpub011:0/16] 2024-02-10 06:36:15,092 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-10 06:36:15,095 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 06:41:46,746 (trainer:762) INFO: 40epoch:train:3701-3800batch: iter_time=3.081, forward_time=0.288, loss_ctc=42.204, loss_att=39.331, acc=0.772, loss=40.193, backward_time=0.295, grad_norm=41.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=4.332
+[gpub011:0/16] 2024-02-10 06:43:53,060 (trainer:762) INFO: 40epoch:train:3801-3900batch: iter_time=7.947e-05, forward_time=0.291, loss_ctc=46.097, loss_att=46.494, acc=0.759, loss=46.375, backward_time=0.297, grad_norm=44.665, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.263
+[gpub011:0/16] 2024-02-10 06:45:59,166 (trainer:762) INFO: 40epoch:train:3901-4000batch: iter_time=7.865e-05, forward_time=0.410, loss_ctc=41.259, loss_att=39.380, acc=0.758, loss=39.944, backward_time=0.314, grad_norm=45.143, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.596e-04, train_time=1.261
+[gpub011:0/16] 2024-02-10 06:48:07,849 (trainer:762) INFO: 40epoch:train:4001-4100batch: iter_time=8.177e-05, forward_time=0.289, loss_ctc=42.500, loss_att=39.855, acc=0.762, loss=40.648, backward_time=0.294, grad_norm=40.939, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.596e-04, train_time=1.287
+[gpub011:0/16] 2024-02-10 06:50:14,328 (trainer:762) INFO: 40epoch:train:4101-4200batch: iter_time=8.220e-05, forward_time=0.290, loss_ctc=47.666, loss_att=50.917, acc=0.750, loss=49.942, backward_time=0.295, grad_norm=49.555, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.596e-04, train_time=1.265
+[gpub011:0/16] 2024-02-10 06:52:25,327 (trainer:762) INFO: 40epoch:train:4201-4300batch: iter_time=7.866e-05, forward_time=0.363, loss_ctc=52.258, loss_att=46.566, acc=0.753, loss=48.274, backward_time=0.377, grad_norm=46.457, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.596e-04, train_time=1.309
+[gpub011:0/16] 2024-02-10 06:54:42,261 (trainer:762) INFO: 40epoch:train:4301-4400batch: iter_time=7.839e-05, forward_time=0.291, loss_ctc=48.917, loss_att=47.375, acc=0.771, loss=47.838, backward_time=0.296, grad_norm=46.680, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.596e-04, train_time=1.370
+[gpub011:0/16] 2024-02-10 06:56:44,613 (trainer:762) INFO: 40epoch:train:4401-4500batch: iter_time=7.917e-05, forward_time=0.291, loss_ctc=43.347, loss_att=42.116, acc=0.771, loss=42.486, backward_time=0.297, grad_norm=36.707, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.595e-04, train_time=1.223
+[gpub011:0/16] 2024-02-10 06:58:39,581 (trainer:762) INFO: 40epoch:train:4501-4600batch: iter_time=7.760e-05, forward_time=0.312, loss_ctc=43.896, loss_att=39.483, acc=0.773, loss=40.807, backward_time=0.317, grad_norm=39.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.595e-04, train_time=1.149
+[gpub011:0/16] 2024-02-10 07:01:02,431 (trainer:762) INFO: 40epoch:train:4601-4700batch: iter_time=7.874e-05, forward_time=0.370, loss_ctc=47.282, loss_att=45.675, acc=0.760, loss=46.157, backward_time=0.336, grad_norm=44.750, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.595e-04, train_time=1.428
+[gpub011:0/16] 2024-02-10 07:03:22,335 (trainer:762) INFO: 40epoch:train:4701-4800batch: iter_time=7.807e-05, forward_time=0.290, loss_ctc=51.687, loss_att=48.303, acc=0.741, loss=49.318, backward_time=0.295, grad_norm=57.202, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.595e-04, train_time=1.399
+[gpub011:0/16] 2024-02-10 07:05:19,756 (trainer:762) INFO: 40epoch:train:4801-4900batch: iter_time=8.353e-05, forward_time=0.294, loss_ctc=51.203, loss_att=50.034, acc=0.756, loss=50.385, backward_time=0.300, grad_norm=45.743, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=1.174
+[gpub011:0/16] 2024-02-10 07:07:42,473 (trainer:762) INFO: 40epoch:train:4901-5000batch: iter_time=7.878e-05, forward_time=0.373, loss_ctc=43.226, loss_att=44.113, acc=0.768, loss=43.847, backward_time=0.358, grad_norm=41.774, clip=100.000, loss_scale=8.152e+33, optim_step_time=0.098, optim0_lr0=1.595e-04, train_time=1.427
+[gpub011:0/16] 2024-02-10 07:08:02,516 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-10 07:08:21,892 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 07:08:25,483 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9f3a4cf610>)
+[gpub011:0/16] 2024-02-10 07:08:25,483 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-10 07:08:25,486 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 07:15:12,903 (trainer:762) INFO: 40epoch:train:5001-5100batch: iter_time=3.274, forward_time=0.289, loss_ctc=44.438, loss_att=38.663, acc=0.785, loss=40.396, backward_time=0.297, grad_norm=40.158, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=4.504
+[gpub011:0/16] 2024-02-10 07:15:42,788 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 07:17:25,845 (trainer:762) INFO: 40epoch:train:5101-5200batch: iter_time=8.333e-05, forward_time=0.291, loss_ctc=42.167, loss_att=46.770, acc=0.755, loss=45.389, backward_time=0.297, grad_norm=49.635, clip=100.000, loss_scale=6.556e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.330
+[gpub011:0/16] 2024-02-10 07:19:58,057 (trainer:762) INFO: 40epoch:train:5201-5300batch: iter_time=8.889e-05, forward_time=0.345, loss_ctc=39.111, loss_att=34.800, acc=0.789, loss=36.093, backward_time=0.401, grad_norm=36.245, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.594e-04, train_time=1.522
+[gpub011:0/16] 2024-02-10 07:21:50,657 (trainer:762) INFO: 40epoch:train:5301-5400batch: iter_time=9.097e-05, forward_time=0.292, loss_ctc=52.331, loss_att=53.403, acc=0.745, loss=53.081, backward_time=0.299, grad_norm=52.803, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.126
+[gpub011:0/16] 2024-02-10 07:24:03,477 (trainer:762) INFO: 40epoch:train:5401-5500batch: iter_time=1.025e-04, forward_time=0.334, loss_ctc=47.843, loss_att=48.884, acc=0.749, loss=48.572, backward_time=0.308, grad_norm=46.012, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.594e-04, train_time=1.328
+[gpub011:0/16] 2024-02-10 07:26:21,907 (trainer:762) INFO: 40epoch:train:5501-5600batch: iter_time=8.611e-05, forward_time=0.352, loss_ctc=44.946, loss_att=43.766, acc=0.792, loss=44.120, backward_time=0.303, grad_norm=36.836, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.384
+[gpub011:0/16] 2024-02-10 07:28:37,935 (trainer:762) INFO: 40epoch:train:5601-5700batch: iter_time=8.578e-05, forward_time=0.292, loss_ctc=48.825, loss_att=48.630, acc=0.767, loss=48.688, backward_time=0.297, grad_norm=48.954, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.360
+[gpub011:0/16] 2024-02-10 07:30:54,714 (trainer:762) INFO: 40epoch:train:5701-5800batch: iter_time=9.969e-05, forward_time=0.380, loss_ctc=44.247, loss_att=43.998, acc=0.769, loss=44.073, backward_time=0.317, grad_norm=41.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.594e-04, train_time=1.368
+[gpub011:0/16] 2024-02-10 07:33:01,624 (trainer:762) INFO: 40epoch:train:5801-5900batch: iter_time=9.047e-05, forward_time=0.294, loss_ctc=43.565, loss_att=40.313, acc=0.775, loss=41.288, backward_time=0.296, grad_norm=39.694, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.269
+[gpub011:0/16] 2024-02-10 07:35:16,660 (trainer:762) INFO: 40epoch:train:5901-6000batch: iter_time=8.382e-05, forward_time=0.292, loss_ctc=52.916, loss_att=52.517, acc=0.748, loss=52.637, backward_time=0.297, grad_norm=60.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.350
+[gpub011:0/16] 2024-02-10 07:37:25,815 (trainer:762) INFO: 40epoch:train:6001-6100batch: iter_time=8.725e-05, forward_time=0.366, loss_ctc=50.091, loss_att=48.218, acc=0.762, loss=48.780, backward_time=0.353, grad_norm=47.402, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.593e-04, train_time=1.291
+[gpub011:0/16] 2024-02-10 07:39:56,541 (trainer:762) INFO: 40epoch:train:6101-6200batch: iter_time=8.685e-05, forward_time=0.295, loss_ctc=47.942, loss_att=46.514, acc=0.766, loss=46.943, backward_time=0.297, grad_norm=41.501, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.507
+[gpub011:0/16] 2024-02-10 07:41:31,267 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-10 07:41:50,619 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 07:41:54,234 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9ad9d65180>)
+[gpub011:0/16] 2024-02-10 07:41:54,234 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-10 07:41:54,239 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 07:48:01,006 (trainer:762) INFO: 40epoch:train:6201-6300batch: iter_time=3.296, forward_time=0.363, loss_ctc=41.541, loss_att=39.401, acc=0.781, loss=40.043, backward_time=0.322, grad_norm=39.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.593e-04, train_time=4.844
+[gpub011:0/16] 2024-02-10 07:50:08,391 (trainer:762) INFO: 40epoch:train:6301-6400batch: iter_time=8.143e-05, forward_time=0.338, loss_ctc=45.121, loss_att=45.900, acc=0.777, loss=45.666, backward_time=0.306, grad_norm=42.630, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.593e-04, train_time=1.274
+[gpub011:0/16] 2024-02-10 07:52:39,426 (trainer:762) INFO: 40epoch:train:6401-6500batch: iter_time=8.121e-05, forward_time=0.312, loss_ctc=40.935, loss_att=41.142, acc=0.762, loss=41.080, backward_time=0.293, grad_norm=47.505, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.510
+[gpub011:0/16] 2024-02-10 07:54:44,863 (trainer:762) INFO: 40epoch:train:6501-6600batch: iter_time=8.285e-05, forward_time=0.354, loss_ctc=42.211, loss_att=39.446, acc=0.770, loss=40.275, backward_time=0.327, grad_norm=41.307, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.593e-04, train_time=1.254
+[gpub011:0/16] 2024-02-10 07:56:51,773 (trainer:762) INFO: 40epoch:train:6601-6700batch: iter_time=8.430e-05, forward_time=0.325, loss_ctc=46.973, loss_att=52.849, acc=0.755, loss=51.086, backward_time=0.323, grad_norm=48.162, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.592e-04, train_time=1.269
+[gpub011:0/16] 2024-02-10 07:59:16,780 (trainer:762) INFO: 40epoch:train:6701-6800batch: iter_time=8.288e-05, forward_time=0.291, loss_ctc=51.479, loss_att=46.107, acc=0.761, loss=47.719, backward_time=0.297, grad_norm=47.217, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.450
+[gpub011:0/16] 2024-02-10 08:01:30,008 (trainer:762) INFO: 40epoch:train:6801-6900batch: iter_time=8.145e-05, forward_time=0.307, loss_ctc=48.202, loss_att=46.925, acc=0.778, loss=47.308, backward_time=0.312, grad_norm=47.509, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.592e-04, train_time=1.332
+[gpub011:0/16] 2024-02-10 08:04:33,479 (trainer:762) INFO: 40epoch:train:6901-7000batch: iter_time=8.448e-05, forward_time=0.401, loss_ctc=43.639, loss_att=43.929, acc=0.775, loss=43.842, backward_time=0.331, grad_norm=37.366, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.592e-04, train_time=1.834
+[gpub011:0/16] 2024-02-10 08:06:25,477 (trainer:762) INFO: 40epoch:train:7001-7100batch: iter_time=7.978e-05, forward_time=0.292, loss_ctc=43.892, loss_att=40.377, acc=0.778, loss=41.431, backward_time=0.298, grad_norm=40.130, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.120
+[gpub011:0/16] 2024-02-10 08:08:29,361 (trainer:762) INFO: 40epoch:train:7101-7200batch: iter_time=8.186e-05, forward_time=0.311, loss_ctc=46.853, loss_att=46.104, acc=0.766, loss=46.329, backward_time=0.307, grad_norm=43.592, clip=100.000, loss_scale=8.983e+33, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.238
+[gpub011:0/16] 2024-02-10 08:11:07,081 (trainer:762) INFO: 40epoch:train:7201-7300batch: iter_time=1.261e-04, forward_time=0.320, loss_ctc=50.591, loss_att=49.461, acc=0.751, loss=49.800, backward_time=0.338, grad_norm=55.727, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.592e-04, train_time=1.576
+[gpub011:0/16] 2024-02-10 08:13:06,195 (trainer:762) INFO: 40epoch:train:7301-7400batch: iter_time=8.465e-05, forward_time=0.319, loss_ctc=50.870, loss_att=50.635, acc=0.763, loss=50.706, backward_time=0.326, grad_norm=45.318, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.191
+[gpub011:0/16] 2024-02-10 08:15:31,918 (trainer:762) INFO: 40epoch:train:7401-7500batch: iter_time=8.202e-05, forward_time=0.291, loss_ctc=42.861, loss_att=44.499, acc=0.771, loss=44.007, backward_time=0.295, grad_norm=40.829, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.591e-04, train_time=1.457
+[gpub011:0/16] 2024-02-10 08:15:51,951 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-10 08:16:11,384 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-10 08:16:14,956 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97c22dba60>)
+[gpub011:0/16] 2024-02-10 08:16:14,956 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-10 08:16:15,009 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-10 08:23:06,031 (trainer:762) INFO: 40epoch:train:7501-7600batch: iter_time=3.183, forward_time=0.352, loss_ctc=44.399, loss_att=36.902, acc=0.792, loss=39.151, backward_time=0.314, grad_norm=41.651, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.591e-04, train_time=4.541
+[gpub011:0/16] 2024-02-10 08:25:31,223 (trainer:762) INFO: 40epoch:train:7601-7700batch: iter_time=8.613e-05, forward_time=0.342, loss_ctc=42.221, loss_att=44.539, acc=0.764, loss=43.843, backward_time=0.303, grad_norm=48.636, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.591e-04, train_time=1.452
+[gpub011:0/16] 2024-02-10 08:27:49,427 (trainer:762) INFO: 40epoch:train:7701-7800batch: iter_time=8.577e-05, forward_time=0.289, loss_ctc=38.844, loss_att=34.166, acc=0.790, loss=35.569, backward_time=0.295, grad_norm=36.977, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.591e-04, train_time=1.382
+[gpub011:0/16] 2024-02-10 08:28:16,890 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-10 08:30:28,504 (trainer:762) INFO: 40epoch:train:7801-7900batch: iter_time=9.123e-05, forward_time=0.338, loss_ctc=52.514, loss_att=53.222, acc=0.747, loss=53.009, backward_time=0.345, grad_norm=51.442, clip=100.000, loss_scale=6.241e+33, optim_step_time=0.103, optim0_lr0=1.591e-04, train_time=1.591
+[gpub011:0/16] 2024-02-10 08:32:56,829 (trainer:762) INFO: 40epoch:train:7901-8000batch: iter_time=8.773e-05, forward_time=0.333, loss_ctc=47.856, loss_att=49.608, acc=0.747, loss=49.082, backward_time=0.302, grad_norm=47.291, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.591e-04, train_time=1.483
+[gpub011:0/16] 2024-02-10 08:35:20,763 (trainer:762) INFO: 40epoch:train:8001-8100batch: iter_time=8.871e-05, forward_time=0.292, loss_ctc=45.018, loss_att=43.464, acc=0.792, loss=43.930, backward_time=0.298, grad_norm=37.813, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.591e-04, train_time=1.439
+[gpub011:0/16] 2024-02-10 08:37:55,624 (trainer:762) INFO: 40epoch:train:8101-8200batch: iter_time=8.806e-05, forward_time=0.350, loss_ctc=48.936, loss_att=49.011, acc=0.765, loss=48.988, backward_time=0.323, grad_norm=42.997, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.590e-04, train_time=1.548
+[gpub011:0/16] 2024-02-10 08:40:27,311 (trainer:762) INFO: 40epoch:train:8201-8300batch: iter_time=9.276e-05, forward_time=0.290, loss_ctc=43.803, loss_att=43.285, acc=0.772, loss=43.440, backward_time=0.296, grad_norm=39.630, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.590e-04, train_time=1.517
+[gpub011:0/16] 2024-02-10 08:43:14,386 (trainer:762) INFO: 40epoch:train:8301-8400batch: iter_time=9.201e-05, forward_time=0.324, loss_ctc=43.245, loss_att=39.430, acc=0.777, loss=40.575, backward_time=0.314, grad_norm=39.310, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.590e-04, train_time=1.668
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2955233.0 ON gpub011 CANCELLED AT 2024-02-10T08:43:24 DUE TO TIME LIMIT ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.10.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.10.log
new file mode 100644
index 0000000000000000000000000000000000000000..c00b9cf89c64d5d7dd6ad13678ea1fa55d628468
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.10.log
@@ -0,0 +1,2592 @@
+# Running on gpub058.delta.ncsa.illinois.edu
+# Started at Fri Feb 2 20:29:30 CST 2024
+# SLURMD_NODENAME=gpub058
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2932506
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1707100152
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2932506
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[058,061-062,071]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1706927352
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[058,061-062,071]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2024661
+# SLURM_TOPOLOGY_ADDR=ss00.ss11.gpub058
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_8c2850e3-53c6-4718-bc5e-c97ca6838673 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_8c2850e3-53c6-4718-bc5e-c97ca6838673
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_8c2850e3-53c6-4718-bc5e-c97ca6838673
+[gpub058:0/16] 2024-02-02 20:31:09,702 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub058:0/16] 2024-02-02 20:31:19,708 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:31:29,714 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:31:39,726 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:31:49,729 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:31:59,736 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:32:09,737 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:32:19,759 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=8, timeout=0:30:00)
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_8c2850e3-53c6-4718-bc5e-c97ca6838673
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_8c2850e3-53c6-4718-bc5e-c97ca6838673
+[gpub058:0/16] 2024-02-02 20:32:30,011 (distributed_c10d:337) INFO: Waiting in store based barrier to initialize process group for rank: 0, key: store_based_barrier_key:1 (world_size=16, worker_count=10, timeout=0:30:00)
+[gpub058:0/16] 2024-02-02 20:32:31,239 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub058:0/16] 2024-02-02 20:32:31,326 (s2t:464) INFO: Vocabulary size: 50002
+[gpub058:0/16] 2024-02-02 20:32:37,473 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub058:0/16] 2024-02-02 20:32:37,479 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub058:0/16] 2024-02-02 20:32:37,480 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub058:0/16] 2024-02-02 20:32:37,480 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub058:0/16] 2024-02-02 20:32:37,492 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub058:0/16] 2024-02-02 20:32:43,001 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 20:32:44,177 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7ff326e12380>)
+[gpub058:0/16] 2024-02-02 20:32:44,177 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub058:0/16] 2024-02-02 20:32:44,178 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-02 20:32:59,169 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub058:2024730:2024730 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2024730:2024730 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2024730:2024730 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub058:0/16] 2024-02-02 20:33:04,760 (trainer:284) INFO: 22/45epoch started
+[gpub058:0/16] 2024-02-02 20:33:04,804 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-02 20:33:22,661 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 20:33:26,008 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7ff13e42d450>)
+[gpub058:0/16] 2024-02-02 20:33:26,008 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-02 20:33:26,011 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub062:1470456:1470456 [0] NCCL INFO cudaDriverVersion 12020
+gpub062:1470456:1470456 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1470456:1470456 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1470456:1470508 [0] NCCL INFO NET/IB : No device found.
+gpub062:1470456:1470508 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1470456:1470508 [0] NCCL INFO Using network Socket
+gpub062:1470456:1470508 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub062:1470456:1470508 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub062:1470456:1470508 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub062:1470456:1470508 [0] NCCL INFO Connected all rings
+gpub062:1470456:1470508 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/Socket/1
+gpub062:1470456:1470508 [0] NCCL INFO Connected all trees
+gpub062:1470456:1470508 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1470456:1470508 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1470456:1470508 [0] NCCL INFO comm 0x14bf05c0 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub062:1470458:1470458 [2] NCCL INFO cudaDriverVersion 12020
+gpub062:1470458:1470458 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1470458:1470458 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1470458:1470509 [2] NCCL INFO NET/IB : No device found.
+gpub062:1470458:1470509 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1470458:1470509 [2] NCCL INFO Using network Socket
+gpub062:1470458:1470509 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub062:1470458:1470509 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub062:1470458:1470509 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub062:1470458:1470509 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub062:1470458:1470509 [2] NCCL INFO Connected all rings
+gpub062:1470458:1470509 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub062:1470458:1470509 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub062:1470458:1470509 [2] NCCL INFO Connected all trees
+gpub062:1470458:1470509 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1470458:1470509 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1470458:1470509 [2] NCCL INFO comm 0xb6031f0 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub062:1470457:1470457 [1] NCCL INFO cudaDriverVersion 12020
+gpub062:1470457:1470457 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1470457:1470457 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1470457:1470510 [1] NCCL INFO NET/IB : No device found.
+gpub062:1470457:1470510 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1470457:1470510 [1] NCCL INFO Using network Socket
+gpub062:1470457:1470510 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub062:1470457:1470510 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub062:1470457:1470510 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub062:1470457:1470510 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub062:1470457:1470510 [1] NCCL INFO Connected all rings
+gpub062:1470457:1470510 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/Socket/1
+gpub062:1470457:1470510 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/Socket/1
+gpub062:1470457:1470510 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub062:1470457:1470510 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub062:1470457:1470510 [1] NCCL INFO Connected all trees
+gpub062:1470457:1470510 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1470457:1470510 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1470457:1470510 [1] NCCL INFO comm 0xd3ed1d0 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub071:440844:440844 [3] NCCL INFO cudaDriverVersion 12020
+gpub071:440844:440844 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.171<0>
+gpub071:440844:440844 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub071:440844:440892 [3] NCCL INFO NET/IB : No device found.
+gpub071:440844:440892 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.171<0> [1]hsn0:141.142.145.171<0> [2]eth0:fe80::d48a:52dd:fe61:8ea0%eth0<0>
+gpub071:440844:440892 [3] NCCL INFO Using network Socket
+gpub071:440844:440892 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub071:440844:440892 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub071:440844:440892 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub071:440844:440892 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub071:440844:440892 [3] NCCL INFO Connected all rings
+gpub071:440844:440892 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub071:440844:440892 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub071:440844:440892 [3] NCCL INFO Connected all trees
+gpub071:440844:440892 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub071:440844:440892 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub071:440844:440892 [3] NCCL INFO comm 0xe33bf10 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub062:1470459:1470459 [3] NCCL INFO cudaDriverVersion 12020
+gpub062:1470459:1470459 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1470459:1470459 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1470459:1470511 [3] NCCL INFO NET/IB : No device found.
+gpub062:1470459:1470511 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1470459:1470511 [3] NCCL INFO Using network Socket
+gpub062:1470459:1470511 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub062:1470459:1470511 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub062:1470459:1470511 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1470459:1470511 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1470459:1470511 [3] NCCL INFO Connected all rings
+gpub062:1470459:1470511 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub062:1470459:1470511 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub062:1470459:1470511 [3] NCCL INFO Connected all trees
+gpub062:1470459:1470511 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1470459:1470511 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1470459:1470511 [3] NCCL INFO comm 0x16654700 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub061:2429081:2429081 [0] NCCL INFO cudaDriverVersion 12020
+gpub061:2429081:2429081 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2429081:2429081 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2429081:2429140 [0] NCCL INFO NET/IB : No device found.
+gpub061:2429081:2429140 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2429081:2429140 [0] NCCL INFO Using network Socket
+gpub061:2429081:2429140 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub061:2429081:2429140 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub061:2429081:2429140 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub061:2429081:2429140 [0] NCCL INFO Connected all rings
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/Socket/1
+gpub061:2429081:2429140 [0] NCCL INFO Connected all trees
+gpub061:2429081:2429140 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2429081:2429140 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2429081:2429140 [0] NCCL INFO comm 0x1af5f510 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub061:2429082:2429082 [1] NCCL INFO cudaDriverVersion 12020
+gpub061:2429082:2429082 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2429082:2429082 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2429082:2429142 [1] NCCL INFO NET/IB : No device found.
+gpub061:2429082:2429142 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2429082:2429142 [1] NCCL INFO Using network Socket
+gpub061:2429082:2429142 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub061:2429082:2429142 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub061:2429082:2429142 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub061:2429082:2429142 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub061:2429082:2429142 [1] NCCL INFO Connected all rings
+gpub061:2429082:2429142 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/Socket/1
+gpub061:2429082:2429142 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/Socket/1
+gpub061:2429082:2429142 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub061:2429082:2429142 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub061:2429082:2429142 [1] NCCL INFO Connected all trees
+gpub061:2429082:2429142 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2429082:2429142 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2429082:2429142 [1] NCCL INFO comm 0xe2568f0 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub061:2429083:2429083 [2] NCCL INFO cudaDriverVersion 12020
+gpub061:2429083:2429083 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2429083:2429083 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2429083:2429141 [2] NCCL INFO NET/IB : No device found.
+gpub061:2429083:2429141 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2429083:2429141 [2] NCCL INFO Using network Socket
+gpub061:2429083:2429141 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub061:2429083:2429141 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub061:2429083:2429141 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub061:2429083:2429141 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub061:2429083:2429141 [2] NCCL INFO Connected all rings
+gpub061:2429083:2429141 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub061:2429083:2429141 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub061:2429083:2429141 [2] NCCL INFO Connected all trees
+gpub061:2429083:2429141 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2429083:2429141 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2429083:2429141 [2] NCCL INFO comm 0x111d07c0 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub071:440843:440843 [2] NCCL INFO cudaDriverVersion 12020
+gpub071:440843:440843 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.171<0>
+gpub071:440843:440843 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub071:440843:440889 [2] NCCL INFO NET/IB : No device found.
+gpub071:440843:440889 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.171<0> [1]hsn0:141.142.145.171<0> [2]eth0:fe80::d48a:52dd:fe61:8ea0%eth0<0>
+gpub071:440843:440889 [2] NCCL INFO Using network Socket
+gpub071:440843:440889 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub071:440843:440889 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub071:440843:440889 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub071:440843:440889 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub071:440843:440889 [2] NCCL INFO Connected all rings
+gpub071:440843:440889 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub058:2024731:2024731 [1] NCCL INFO cudaDriverVersion 12020
+gpub058:2024731:2024731 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2024731:2024731 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2024731:2024796 [1] NCCL INFO NET/IB : No device found.
+gpub058:2024731:2024796 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2024731:2024796 [1] NCCL INFO Using network Socket
+gpub058:2024731:2024796 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub058:2024731:2024796 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub058:2024731:2024796 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub058:2024731:2024796 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub058:2024731:2024796 [1] NCCL INFO Connected all rings
+gpub058:2024731:2024796 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub071:440843:440889 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub071:440843:440889 [2] NCCL INFO Connected all trees
+gpub071:440843:440889 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub071:440843:440889 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub071:440843:440889 [2] NCCL INFO comm 0x15c0b660 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub058:2024731:2024796 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub058:2024731:2024796 [1] NCCL INFO Connected all trees
+gpub058:2024731:2024796 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2024731:2024796 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2024731:2024796 [1] NCCL INFO comm 0xcc917a0 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub071:440841:440841 [0] NCCL INFO cudaDriverVersion 12020
+gpub071:440841:440841 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.171<0>
+gpub071:440841:440841 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub071:440841:440890 [0] NCCL INFO NET/IB : No device found.
+gpub071:440841:440890 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.171<0> [1]hsn0:141.142.145.171<0> [2]eth0:fe80::d48a:52dd:fe61:8ea0%eth0<0>
+gpub071:440841:440890 [0] NCCL INFO Using network Socket
+gpub071:440841:440890 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub071:440841:440890 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub071:440841:440890 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub071:440841:440890 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub071:440841:440890 [0] NCCL INFO Connected all rings
+gpub071:440841:440890 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/Socket/1
+gpub071:440841:440890 [0] NCCL INFO Connected all trees
+gpub071:440841:440890 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub071:440841:440890 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub071:440841:440890 [0] NCCL INFO comm 0x14fb51f0 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub071:440842:440842 [1] NCCL INFO cudaDriverVersion 12020
+gpub071:440842:440842 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.171<0>
+gpub071:440842:440842 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub071:440842:440891 [1] NCCL INFO NET/IB : No device found.
+gpub071:440842:440891 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.171<0> [1]hsn0:141.142.145.171<0> [2]eth0:fe80::d48a:52dd:fe61:8ea0%eth0<0>
+gpub071:440842:440891 [1] NCCL INFO Using network Socket
+gpub071:440842:440891 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub071:440842:440891 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub071:440842:440891 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub071:440842:440891 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub071:440842:440891 [1] NCCL INFO Connected all rings
+gpub071:440842:440891 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub071:440842:440891 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub071:440842:440891 [1] NCCL INFO Connected all trees
+gpub071:440842:440891 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub071:440842:440891 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub071:440842:440891 [1] NCCL INFO comm 0xf72c140 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub061:2429084:2429084 [3] NCCL INFO cudaDriverVersion 12020
+gpub061:2429084:2429084 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2429084:2429084 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2429084:2429143 [3] NCCL INFO NET/IB : No device found.
+gpub061:2429084:2429143 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2429084:2429143 [3] NCCL INFO Using network Socket
+gpub061:2429084:2429143 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub061:2429084:2429143 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub061:2429084:2429143 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub061:2429084:2429143 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub061:2429084:2429143 [3] NCCL INFO Connected all rings
+gpub061:2429084:2429143 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub061:2429084:2429143 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub061:2429084:2429143 [3] NCCL INFO Connected all trees
+gpub061:2429084:2429143 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2429084:2429143 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2429084:2429143 [3] NCCL INFO comm 0x10f63180 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub058:2024730:2024793 [0] NCCL INFO NET/IB : No device found.
+gpub058:2024730:2024793 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2024730:2024793 [0] NCCL INFO Using network Socket
+gpub058:2024730:2024793 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub058:2024730:2024793 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub058:2024730:2024793 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub058:2024730:2024793 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub058:2024730:2024793 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub058:2024730:2024793 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub058:2024730:2024793 [0] NCCL INFO Connected all rings
+gpub058:2024730:2024793 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2024730:2024793 [0] NCCL INFO Connected all trees
+gpub058:2024730:2024793 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2024730:2024793 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2024730:2024793 [0] NCCL INFO comm 0x7474f0c0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub058:2024732:2024732 [2] NCCL INFO cudaDriverVersion 12020
+gpub058:2024732:2024732 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2024732:2024732 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2024732:2024794 [2] NCCL INFO NET/IB : No device found.
+gpub058:2024732:2024794 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2024732:2024794 [2] NCCL INFO Using network Socket
+gpub058:2024732:2024794 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub058:2024732:2024794 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub058:2024732:2024794 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub058:2024732:2024794 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub058:2024732:2024794 [2] NCCL INFO Connected all rings
+gpub058:2024732:2024794 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub058:2024732:2024794 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub058:2024732:2024794 [2] NCCL INFO Connected all trees
+gpub058:2024732:2024794 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2024732:2024794 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2024732:2024794 [2] NCCL INFO comm 0xedc40d0 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub058:2024733:2024733 [3] NCCL INFO cudaDriverVersion 12020
+gpub058:2024733:2024733 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2024733:2024733 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2024733:2024795 [3] NCCL INFO NET/IB : No device found.
+gpub058:2024733:2024795 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2024733:2024795 [3] NCCL INFO Using network Socket
+gpub058:2024733:2024795 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub058:2024733:2024795 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub058:2024733:2024795 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2024733:2024795 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2024733:2024795 [3] NCCL INFO Connected all rings
+gpub058:2024733:2024795 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub058:2024733:2024795 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub058:2024733:2024795 [3] NCCL INFO Connected all trees
+gpub058:2024733:2024795 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2024733:2024795 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2024733:2024795 [3] NCCL INFO comm 0x14398bf0 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+[gpub058:0/16] 2024-02-02 20:42:59,792 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub058:0/16] 2024-02-02 20:46:10,549 (trainer:737) INFO: 22epoch:train:1-100batch: iter_time=3.072, forward_time=0.376, loss_ctc=54.605, loss_att=59.349, acc=0.723, loss=57.926, backward_time=0.411, grad_norm=39.680, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.182e-04, train_time=7.857
+[gpub058:0/16] 2024-02-02 20:50:45,607 (trainer:737) INFO: 22epoch:train:101-200batch: iter_time=9.428e-05, forward_time=0.350, loss_ctc=50.542, loss_att=51.612, acc=0.719, loss=51.291, backward_time=0.425, grad_norm=36.846, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.182e-04, train_time=2.750
+[gpub058:0/16] 2024-02-02 20:54:06,235 (trainer:737) INFO: 22epoch:train:201-300batch: iter_time=3.186e-04, forward_time=0.313, loss_ctc=54.663, loss_att=46.797, acc=0.760, loss=49.157, backward_time=0.407, grad_norm=32.838, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.181e-04, train_time=2.007
+[gpub058:0/16] 2024-02-02 20:57:41,945 (trainer:737) INFO: 22epoch:train:301-400batch: iter_time=8.946e-05, forward_time=0.336, loss_ctc=52.116, loss_att=54.303, acc=0.732, loss=53.647, backward_time=0.420, grad_norm=38.768, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.181e-04, train_time=2.157
+[gpub058:0/16] 2024-02-02 21:00:43,152 (trainer:737) INFO: 22epoch:train:401-500batch: iter_time=1.055e-04, forward_time=0.327, loss_ctc=58.133, loss_att=51.993, acc=0.749, loss=53.835, backward_time=0.421, grad_norm=39.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.181e-04, train_time=1.811
+[gpub058:0/16] 2024-02-02 21:04:20,921 (trainer:737) INFO: 22epoch:train:501-600batch: iter_time=9.733e-05, forward_time=0.344, loss_ctc=52.104, loss_att=50.746, acc=0.735, loss=51.153, backward_time=0.414, grad_norm=35.890, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.180e-04, train_time=2.179
+[gpub058:0/16] 2024-02-02 21:07:35,569 (trainer:737) INFO: 22epoch:train:601-700batch: iter_time=9.290e-05, forward_time=0.287, loss_ctc=55.156, loss_att=46.061, acc=0.741, loss=48.790, backward_time=0.397, grad_norm=42.747, clip=100.000, loss_scale=7.062e+33, optim_step_time=0.091, optim0_lr0=2.180e-04, train_time=1.946
+[gpub058:0/16] 2024-02-02 21:10:53,191 (trainer:737) INFO: 22epoch:train:701-800batch: iter_time=9.677e-05, forward_time=0.395, loss_ctc=56.217, loss_att=50.572, acc=0.749, loss=52.265, backward_time=0.420, grad_norm=39.491, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.180e-04, train_time=1.975
+[gpub058:0/16] 2024-02-02 21:14:01,106 (trainer:737) INFO: 22epoch:train:801-900batch: iter_time=9.549e-05, forward_time=0.289, loss_ctc=52.629, loss_att=55.688, acc=0.735, loss=54.770, backward_time=0.400, grad_norm=38.894, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.179e-04, train_time=1.880
+[gpub058:0/16] 2024-02-02 21:16:59,310 (trainer:737) INFO: 22epoch:train:901-1000batch: iter_time=9.657e-05, forward_time=0.369, loss_ctc=53.588, loss_att=51.105, acc=0.723, loss=51.850, backward_time=0.413, grad_norm=40.169, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.179e-04, train_time=1.782
+[gpub058:0/16] 2024-02-02 21:19:59,397 (trainer:737) INFO: 22epoch:train:1001-1100batch: iter_time=8.136e-05, forward_time=0.301, loss_ctc=60.142, loss_att=58.070, acc=0.733, loss=58.692, backward_time=0.417, grad_norm=40.980, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.179e-04, train_time=1.798
+[gpub058:0/16] 2024-02-02 21:23:15,247 (trainer:737) INFO: 22epoch:train:1101-1200batch: iter_time=8.064e-05, forward_time=0.287, loss_ctc=51.604, loss_att=51.098, acc=0.734, loss=51.250, backward_time=0.399, grad_norm=36.998, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.178e-04, train_time=1.961
+[gpub058:0/16] 2024-02-02 21:25:05,164 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-02 21:25:23,799 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 21:25:27,255 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7ff13e42c490>)
+[gpub058:0/16] 2024-02-02 21:25:27,255 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-02 21:25:27,258 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-02 21:34:16,523 (trainer:737) INFO: 22epoch:train:1201-1300batch: iter_time=3.169, forward_time=0.337, loss_ctc=52.096, loss_att=57.080, acc=0.728, loss=55.585, backward_time=0.418, grad_norm=38.095, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.178e-04, train_time=6.613
+[gpub058:0/16] 2024-02-02 21:37:01,583 (trainer:737) INFO: 22epoch:train:1301-1400batch: iter_time=8.324e-05, forward_time=0.333, loss_ctc=53.304, loss_att=57.492, acc=0.704, loss=56.235, backward_time=0.423, grad_norm=39.679, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.178e-04, train_time=1.650
+[gpub058:0/16] 2024-02-02 21:39:30,357 (trainer:737) INFO: 22epoch:train:1401-1500batch: iter_time=8.313e-05, forward_time=0.287, loss_ctc=42.968, loss_att=37.847, acc=0.752, loss=39.383, backward_time=0.398, grad_norm=30.403, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.177e-04, train_time=1.487
+[gpub058:0/16] 2024-02-02 21:40:45,319 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-02 21:42:07,759 (trainer:737) INFO: 22epoch:train:1501-1600batch: iter_time=8.718e-05, forward_time=0.295, loss_ctc=55.182, loss_att=52.618, acc=0.747, loss=53.387, backward_time=0.408, grad_norm=35.755, clip=100.000, loss_scale=7.395e+33, optim_step_time=0.091, optim0_lr0=2.177e-04, train_time=1.574
+[gpub058:0/16] 2024-02-02 21:45:02,452 (trainer:737) INFO: 22epoch:train:1601-1700batch: iter_time=9.128e-05, forward_time=0.375, loss_ctc=56.111, loss_att=53.970, acc=0.734, loss=54.612, backward_time=0.444, grad_norm=35.374, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.177e-04, train_time=1.747
+[gpub058:0/16] 2024-02-02 21:47:45,115 (trainer:737) INFO: 22epoch:train:1701-1800batch: iter_time=8.693e-05, forward_time=0.291, loss_ctc=54.139, loss_att=48.451, acc=0.743, loss=50.158, backward_time=0.403, grad_norm=35.890, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.176e-04, train_time=1.626
+[gpub058:0/16] 2024-02-02 21:50:08,523 (trainer:737) INFO: 22epoch:train:1801-1900batch: iter_time=8.335e-05, forward_time=0.289, loss_ctc=47.506, loss_att=45.046, acc=0.734, loss=45.784, backward_time=0.402, grad_norm=35.443, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.176e-04, train_time=1.433
+[gpub058:0/16] 2024-02-02 21:52:42,711 (trainer:737) INFO: 22epoch:train:1901-2000batch: iter_time=8.193e-05, forward_time=0.305, loss_ctc=54.524, loss_att=45.149, acc=0.732, loss=47.961, backward_time=0.413, grad_norm=43.650, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.176e-04, train_time=1.542
+[gpub058:0/16] 2024-02-02 21:55:35,849 (trainer:737) INFO: 22epoch:train:2001-2100batch: iter_time=8.690e-05, forward_time=0.342, loss_ctc=59.794, loss_att=59.746, acc=0.733, loss=59.760, backward_time=0.419, grad_norm=39.380, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.175e-04, train_time=1.731
+[gpub058:0/16] 2024-02-02 21:58:03,989 (trainer:737) INFO: 22epoch:train:2101-2200batch: iter_time=8.242e-05, forward_time=0.289, loss_ctc=50.537, loss_att=54.357, acc=0.726, loss=53.211, backward_time=0.402, grad_norm=36.613, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.175e-04, train_time=1.481
+[gpub058:0/16] 2024-02-02 22:00:29,399 (trainer:737) INFO: 22epoch:train:2201-2300batch: iter_time=8.149e-05, forward_time=0.292, loss_ctc=48.640, loss_att=47.881, acc=0.730, loss=48.109, backward_time=0.403, grad_norm=34.322, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.175e-04, train_time=1.453
+[gpub058:0/16] 2024-02-02 22:03:18,954 (trainer:737) INFO: 22epoch:train:2301-2400batch: iter_time=8.335e-05, forward_time=0.374, loss_ctc=55.953, loss_att=56.322, acc=0.731, loss=56.211, backward_time=0.439, grad_norm=38.833, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.174e-04, train_time=1.696
+[gpub058:0/16] 2024-02-02 22:06:00,158 (trainer:737) INFO: 22epoch:train:2401-2500batch: iter_time=7.995e-05, forward_time=0.288, loss_ctc=49.520, loss_att=46.109, acc=0.731, loss=47.132, backward_time=0.399, grad_norm=33.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.174e-04, train_time=1.612
+[gpub058:0/16] 2024-02-02 22:06:20,187 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-02 22:06:38,708 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 22:06:42,670 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7ff12cd23fa0>)
+[gpub058:0/16] 2024-02-02 22:06:42,670 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-02 22:06:42,674 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-02 22:13:32,522 (trainer:737) INFO: 22epoch:train:2501-2600batch: iter_time=2.997, forward_time=0.380, loss_ctc=53.323, loss_att=59.525, acc=0.728, loss=57.664, backward_time=0.426, grad_norm=37.643, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.174e-04, train_time=4.523
+[gpub058:0/16] 2024-02-02 22:15:52,316 (trainer:737) INFO: 22epoch:train:2601-2700batch: iter_time=8.583e-05, forward_time=0.306, loss_ctc=48.222, loss_att=50.977, acc=0.727, loss=50.151, backward_time=0.406, grad_norm=34.965, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.173e-04, train_time=1.398
+[gpub058:0/16] 2024-02-02 22:18:40,112 (trainer:737) INFO: 22epoch:train:2701-2800batch: iter_time=7.923e-05, forward_time=0.310, loss_ctc=52.831, loss_att=45.762, acc=0.766, loss=47.883, backward_time=0.419, grad_norm=33.753, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.173e-04, train_time=1.678
+[gpub058:0/16] 2024-02-02 22:21:05,510 (trainer:737) INFO: 22epoch:train:2801-2900batch: iter_time=8.115e-05, forward_time=0.342, loss_ctc=50.906, loss_att=53.637, acc=0.736, loss=52.818, backward_time=0.425, grad_norm=36.687, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.173e-04, train_time=1.453
+[gpub058:0/16] 2024-02-02 22:23:30,182 (trainer:737) INFO: 22epoch:train:2901-3000batch: iter_time=8.372e-05, forward_time=0.313, loss_ctc=56.109, loss_att=51.681, acc=0.751, loss=53.009, backward_time=0.420, grad_norm=37.704, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.172e-04, train_time=1.447
+[gpub058:0/16] 2024-02-02 22:26:04,977 (trainer:737) INFO: 22epoch:train:3001-3100batch: iter_time=7.420e-05, forward_time=0.355, loss_ctc=49.939, loss_att=49.832, acc=0.739, loss=49.864, backward_time=0.415, grad_norm=34.651, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.172e-04, train_time=1.548
+[gpub058:0/16] 2024-02-02 22:28:38,778 (trainer:737) INFO: 22epoch:train:3101-3200batch: iter_time=5.763e-04, forward_time=0.308, loss_ctc=52.385, loss_att=45.069, acc=0.744, loss=47.264, backward_time=0.408, grad_norm=40.315, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.172e-04, train_time=1.538
+[gpub058:0/16] 2024-02-02 22:31:07,364 (trainer:737) INFO: 22epoch:train:3201-3300batch: iter_time=7.773e-05, forward_time=0.347, loss_ctc=54.468, loss_att=49.829, acc=0.753, loss=51.220, backward_time=0.440, grad_norm=37.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.171e-04, train_time=1.486
+[gpub058:0/16] 2024-02-02 22:33:43,279 (trainer:737) INFO: 22epoch:train:3301-3400batch: iter_time=1.614e-04, forward_time=0.321, loss_ctc=50.544, loss_att=54.836, acc=0.740, loss=53.548, backward_time=0.410, grad_norm=35.949, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.171e-04, train_time=1.558
+[gpub058:0/16] 2024-02-02 22:36:08,226 (trainer:737) INFO: 22epoch:train:3401-3500batch: iter_time=8.242e-05, forward_time=0.333, loss_ctc=51.578, loss_att=50.390, acc=0.729, loss=50.747, backward_time=0.434, grad_norm=38.147, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.170e-04, train_time=1.450
+[gpub058:0/16] 2024-02-02 22:38:47,308 (trainer:737) INFO: 22epoch:train:3501-3600batch: iter_time=0.002, forward_time=0.329, loss_ctc=57.958, loss_att=57.261, acc=0.738, loss=57.470, backward_time=0.420, grad_norm=36.663, clip=100.000, loss_scale=8.152e+33, optim_step_time=0.097, optim0_lr0=2.170e-04, train_time=1.590
+[gpub058:0/16] 2024-02-02 22:41:08,150 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-02 22:41:25,142 (trainer:737) INFO: 22epoch:train:3601-3700batch: iter_time=4.220e-04, forward_time=0.348, loss_ctc=49.342, loss_att=49.126, acc=0.741, loss=49.191, backward_time=0.417, grad_norm=34.289, clip=100.000, loss_scale=9.755e+33, optim_step_time=0.097, optim0_lr0=2.170e-04, train_time=1.579
+[gpub058:0/16] 2024-02-02 22:42:56,635 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-02 22:43:15,347 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 22:43:18,975 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec4b0ef880>)
+[gpub058:0/16] 2024-02-02 22:43:18,975 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-02 22:43:18,978 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-02 22:48:55,576 (trainer:737) INFO: 22epoch:train:3701-3800batch: iter_time=3.057, forward_time=0.350, loss_ctc=51.109, loss_att=55.167, acc=0.736, loss=53.949, backward_time=0.447, grad_norm=36.715, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.169e-04, train_time=4.504
+[gpub058:0/16] 2024-02-02 22:51:29,376 (trainer:737) INFO: 22epoch:train:3801-3900batch: iter_time=7.545e-05, forward_time=0.315, loss_ctc=52.851, loss_att=56.611, acc=0.708, loss=55.483, backward_time=0.402, grad_norm=38.396, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.169e-04, train_time=1.538
+[gpub058:0/16] 2024-02-02 22:54:05,378 (trainer:737) INFO: 22epoch:train:3901-4000batch: iter_time=7.910e-05, forward_time=0.320, loss_ctc=42.545, loss_att=37.765, acc=0.754, loss=39.199, backward_time=0.405, grad_norm=29.364, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.169e-04, train_time=1.560
+[gpub058:0/16] 2024-02-02 22:56:36,674 (trainer:737) INFO: 22epoch:train:4001-4100batch: iter_time=8.114e-05, forward_time=0.304, loss_ctc=54.096, loss_att=51.773, acc=0.751, loss=52.470, backward_time=0.416, grad_norm=35.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.168e-04, train_time=1.512
+[gpub058:0/16] 2024-02-02 22:58:54,859 (trainer:737) INFO: 22epoch:train:4101-4200batch: iter_time=8.093e-05, forward_time=0.333, loss_ctc=55.593, loss_att=53.843, acc=0.737, loss=54.368, backward_time=0.421, grad_norm=36.693, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.168e-04, train_time=1.382
+[gpub058:0/16] 2024-02-02 23:01:38,322 (trainer:737) INFO: 22epoch:train:4201-4300batch: iter_time=2.448e-04, forward_time=0.325, loss_ctc=53.155, loss_att=47.959, acc=0.747, loss=49.518, backward_time=0.405, grad_norm=35.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.168e-04, train_time=1.634
+[gpub058:0/16] 2024-02-02 23:04:03,206 (trainer:737) INFO: 22epoch:train:4301-4400batch: iter_time=8.520e-05, forward_time=0.306, loss_ctc=46.551, loss_att=44.317, acc=0.736, loss=44.987, backward_time=0.421, grad_norm=34.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.167e-04, train_time=1.448
+[gpub058:0/16] 2024-02-02 23:06:37,723 (trainer:737) INFO: 22epoch:train:4401-4500batch: iter_time=8.277e-05, forward_time=0.351, loss_ctc=53.771, loss_att=44.644, acc=0.737, loss=47.382, backward_time=0.413, grad_norm=42.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.167e-04, train_time=1.546
+[gpub058:0/16] 2024-02-02 23:08:54,737 (trainer:737) INFO: 22epoch:train:4501-4600batch: iter_time=1.858e-04, forward_time=0.304, loss_ctc=58.689, loss_att=58.310, acc=0.736, loss=58.424, backward_time=0.419, grad_norm=38.764, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.167e-04, train_time=1.370
+[gpub058:0/16] 2024-02-02 23:11:42,279 (trainer:737) INFO: 22epoch:train:4601-4700batch: iter_time=8.721e-05, forward_time=0.337, loss_ctc=49.167, loss_att=53.840, acc=0.730, loss=52.438, backward_time=0.428, grad_norm=36.468, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.166e-04, train_time=1.674
+[gpub058:0/16] 2024-02-02 23:14:20,570 (trainer:737) INFO: 22epoch:train:4701-4800batch: iter_time=8.879e-04, forward_time=0.347, loss_ctc=48.156, loss_att=48.217, acc=0.731, loss=48.199, backward_time=0.430, grad_norm=35.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.166e-04, train_time=1.584
+[gpub058:0/16] 2024-02-02 23:16:54,500 (trainer:737) INFO: 22epoch:train:4801-4900batch: iter_time=8.658e-05, forward_time=0.350, loss_ctc=54.703, loss_att=54.955, acc=0.734, loss=54.879, backward_time=0.447, grad_norm=42.581, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.166e-04, train_time=1.538
+[gpub058:0/16] 2024-02-02 23:19:28,373 (trainer:737) INFO: 22epoch:train:4901-5000batch: iter_time=8.766e-05, forward_time=0.312, loss_ctc=48.506, loss_att=45.667, acc=0.733, loss=46.519, backward_time=0.410, grad_norm=38.431, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.165e-04, train_time=1.539
+[gpub058:0/16] 2024-02-02 23:19:48,562 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-02 23:20:07,257 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 23:20:10,859 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec4de7c880>)
+[gpub058:0/16] 2024-02-02 23:20:10,860 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-02 23:20:10,892 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-02 23:27:26,455 (trainer:737) INFO: 22epoch:train:5001-5100batch: iter_time=3.273, forward_time=0.311, loss_ctc=53.167, loss_att=58.353, acc=0.726, loss=56.797, backward_time=0.420, grad_norm=38.131, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.165e-04, train_time=4.781
+[gpub058:0/16] 2024-02-02 23:30:06,871 (trainer:737) INFO: 22epoch:train:5101-5200batch: iter_time=8.560e-05, forward_time=0.332, loss_ctc=47.431, loss_att=49.019, acc=0.723, loss=48.542, backward_time=0.411, grad_norm=41.624, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.165e-04, train_time=1.603
+[gpub058:0/16] 2024-02-02 23:32:33,416 (trainer:737) INFO: 22epoch:train:5201-5300batch: iter_time=8.219e-05, forward_time=0.322, loss_ctc=52.616, loss_att=46.062, acc=0.757, loss=48.028, backward_time=0.418, grad_norm=33.911, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.164e-04, train_time=1.466
+[gpub058:0/16] 2024-02-02 23:35:10,154 (trainer:737) INFO: 22epoch:train:5301-5400batch: iter_time=8.672e-05, forward_time=0.307, loss_ctc=50.234, loss_att=52.391, acc=0.736, loss=51.744, backward_time=0.402, grad_norm=37.221, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.164e-04, train_time=1.567
+[gpub058:0/16] 2024-02-02 23:37:33,903 (trainer:737) INFO: 22epoch:train:5401-5500batch: iter_time=8.658e-05, forward_time=0.351, loss_ctc=55.457, loss_att=49.079, acc=0.749, loss=50.992, backward_time=0.456, grad_norm=37.923, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.164e-04, train_time=1.437
+[gpub058:0/16] 2024-02-02 23:39:57,298 (trainer:737) INFO: 22epoch:train:5501-5600batch: iter_time=8.016e-05, forward_time=0.309, loss_ctc=49.281, loss_att=49.787, acc=0.726, loss=49.635, backward_time=0.422, grad_norm=34.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.163e-04, train_time=1.434
+[gpub058:0/16] 2024-02-02 23:42:32,906 (trainer:737) INFO: 22epoch:train:5601-5700batch: iter_time=8.459e-05, forward_time=0.291, loss_ctc=51.749, loss_att=43.251, acc=0.743, loss=45.800, backward_time=0.399, grad_norm=36.677, clip=100.000, loss_scale=5.815e+33, optim_step_time=0.091, optim0_lr0=2.163e-04, train_time=1.556
+[gpub058:0/16] 2024-02-02 23:44:56,499 (trainer:737) INFO: 22epoch:train:5701-5800batch: iter_time=8.799e-05, forward_time=0.327, loss_ctc=53.557, loss_att=48.942, acc=0.750, loss=50.327, backward_time=0.422, grad_norm=35.259, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.163e-04, train_time=1.436
+[gpub058:0/16] 2024-02-02 23:47:29,688 (trainer:737) INFO: 22epoch:train:5801-5900batch: iter_time=8.139e-05, forward_time=0.317, loss_ctc=49.652, loss_att=54.448, acc=0.737, loss=53.009, backward_time=0.416, grad_norm=37.216, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.162e-04, train_time=1.532
+[gpub058:0/16] 2024-02-02 23:50:06,970 (trainer:737) INFO: 22epoch:train:5901-6000batch: iter_time=9.230e-05, forward_time=0.311, loss_ctc=50.648, loss_att=48.999, acc=0.727, loss=49.494, backward_time=0.414, grad_norm=36.041, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.162e-04, train_time=1.573
+[gpub058:0/16] 2024-02-02 23:52:35,113 (trainer:737) INFO: 22epoch:train:6001-6100batch: iter_time=7.936e-05, forward_time=0.308, loss_ctc=56.060, loss_att=56.425, acc=0.737, loss=56.315, backward_time=0.424, grad_norm=36.248, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.162e-04, train_time=1.481
+[gpub058:0/16] 2024-02-02 23:55:05,274 (trainer:737) INFO: 22epoch:train:6101-6200batch: iter_time=8.342e-05, forward_time=0.337, loss_ctc=48.563, loss_att=48.931, acc=0.735, loss=48.821, backward_time=0.415, grad_norm=35.660, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.161e-04, train_time=1.502
+[gpub058:0/16] 2024-02-02 23:56:43,138 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-02 23:57:01,932 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-02 23:57:05,571 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c3532590>)
+[gpub058:0/16] 2024-02-02 23:57:05,571 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-02 23:57:05,574 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 00:02:51,976 (trainer:737) INFO: 22epoch:train:6201-6300batch: iter_time=3.106, forward_time=0.303, loss_ctc=50.699, loss_att=54.610, acc=0.727, loss=53.436, backward_time=0.405, grad_norm=37.948, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.161e-04, train_time=4.667
+[gpub058:0/16] 2024-02-03 00:05:11,597 (trainer:737) INFO: 22epoch:train:6301-6400batch: iter_time=7.676e-05, forward_time=0.318, loss_ctc=51.903, loss_att=55.731, acc=0.711, loss=54.582, backward_time=0.424, grad_norm=36.923, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.161e-04, train_time=1.396
+[gpub058:0/16] 2024-02-03 00:07:36,422 (trainer:737) INFO: 22epoch:train:6401-6500batch: iter_time=7.697e-05, forward_time=0.321, loss_ctc=42.050, loss_att=36.935, acc=0.757, loss=38.469, backward_time=0.406, grad_norm=31.031, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.160e-04, train_time=1.448
+[gpub058:0/16] 2024-02-03 00:10:13,880 (trainer:737) INFO: 22epoch:train:6501-6600batch: iter_time=8.625e-05, forward_time=0.311, loss_ctc=54.107, loss_att=51.416, acc=0.752, loss=52.224, backward_time=0.415, grad_norm=36.022, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.160e-04, train_time=1.574
+[gpub058:0/16] 2024-02-03 00:12:40,044 (trainer:737) INFO: 22epoch:train:6601-6700batch: iter_time=8.824e-05, forward_time=0.365, loss_ctc=54.580, loss_att=52.500, acc=0.741, loss=53.124, backward_time=0.438, grad_norm=36.155, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.160e-04, train_time=1.462
+[gpub058:0/16] 2024-02-03 00:15:09,737 (trainer:737) INFO: 22epoch:train:6701-6800batch: iter_time=2.090e-04, forward_time=0.315, loss_ctc=51.993, loss_att=47.123, acc=0.747, loss=48.584, backward_time=0.419, grad_norm=34.131, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.159e-04, train_time=1.497
+[gpub058:0/16] 2024-02-03 00:17:56,548 (trainer:737) INFO: 22epoch:train:6801-6900batch: iter_time=8.775e-05, forward_time=0.390, loss_ctc=46.408, loss_att=44.353, acc=0.736, loss=44.969, backward_time=0.421, grad_norm=33.629, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.159e-04, train_time=1.668
+[gpub058:0/16] 2024-02-03 00:20:16,108 (trainer:737) INFO: 22epoch:train:6901-7000batch: iter_time=8.490e-05, forward_time=0.297, loss_ctc=52.993, loss_att=43.624, acc=0.742, loss=46.435, backward_time=0.401, grad_norm=41.897, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.159e-04, train_time=1.395
+[gpub058:0/16] 2024-02-03 00:21:04,061 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 00:22:35,644 (trainer:737) INFO: 22epoch:train:7001-7100batch: iter_time=4.279e-04, forward_time=0.319, loss_ctc=58.265, loss_att=58.594, acc=0.735, loss=58.495, backward_time=0.422, grad_norm=38.788, clip=100.000, loss_scale=6.923e+33, optim_step_time=0.095, optim0_lr0=2.158e-04, train_time=1.395
+[gpub058:0/16] 2024-02-03 00:25:20,087 (trainer:737) INFO: 22epoch:train:7101-7200batch: iter_time=8.871e-05, forward_time=0.334, loss_ctc=48.586, loss_att=53.609, acc=0.731, loss=52.102, backward_time=0.424, grad_norm=35.666, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.158e-04, train_time=1.645
+[gpub058:0/16] 2024-02-03 00:27:56,095 (trainer:737) INFO: 22epoch:train:7201-7300batch: iter_time=8.506e-05, forward_time=0.318, loss_ctc=47.090, loss_att=48.316, acc=0.730, loss=47.948, backward_time=0.410, grad_norm=35.579, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.158e-04, train_time=1.560
+[gpub058:0/16] 2024-02-03 00:30:20,457 (trainer:737) INFO: 22epoch:train:7301-7400batch: iter_time=9.981e-05, forward_time=0.338, loss_ctc=53.643, loss_att=54.671, acc=0.738, loss=54.362, backward_time=0.440, grad_norm=35.025, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.157e-04, train_time=1.443
+[gpub058:0/16] 2024-02-03 00:32:56,359 (trainer:737) INFO: 22epoch:train:7401-7500batch: iter_time=8.207e-05, forward_time=0.321, loss_ctc=48.220, loss_att=44.945, acc=0.737, loss=45.927, backward_time=0.413, grad_norm=35.056, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.157e-04, train_time=1.559
+[gpub058:0/16] 2024-02-03 00:33:16,545 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-03 00:33:35,549 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 00:33:39,201 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec7d44af20>)
+[gpub058:0/16] 2024-02-03 00:33:39,202 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-03 00:33:39,205 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 00:40:26,154 (trainer:737) INFO: 22epoch:train:7501-7600batch: iter_time=3.116, forward_time=0.347, loss_ctc=52.300, loss_att=57.848, acc=0.728, loss=56.184, backward_time=0.413, grad_norm=38.196, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.157e-04, train_time=4.498
+[gpub058:0/16] 2024-02-03 00:43:03,242 (trainer:737) INFO: 22epoch:train:7601-7700batch: iter_time=8.548e-05, forward_time=0.370, loss_ctc=47.287, loss_att=47.819, acc=0.728, loss=47.659, backward_time=0.446, grad_norm=34.484, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.156e-04, train_time=1.570
+[gpub058:0/16] 2024-02-03 00:45:26,332 (trainer:737) INFO: 22epoch:train:7701-7800batch: iter_time=8.890e-05, forward_time=0.299, loss_ctc=52.565, loss_att=45.986, acc=0.757, loss=47.960, backward_time=0.407, grad_norm=31.601, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.156e-04, train_time=1.432
+[gpub058:0/16] 2024-02-03 00:47:58,322 (trainer:737) INFO: 22epoch:train:7801-7900batch: iter_time=9.277e-05, forward_time=0.290, loss_ctc=49.670, loss_att=51.401, acc=0.739, loss=50.882, backward_time=0.405, grad_norm=35.967, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.156e-04, train_time=1.520
+[gpub058:0/16] 2024-02-03 00:50:29,782 (trainer:737) INFO: 22epoch:train:7901-8000batch: iter_time=2.038e-04, forward_time=0.366, loss_ctc=55.304, loss_att=48.870, acc=0.749, loss=50.800, backward_time=0.456, grad_norm=36.083, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.155e-04, train_time=1.513
+[gpub058:0/16] 2024-02-03 00:52:46,346 (trainer:737) INFO: 22epoch:train:8001-8100batch: iter_time=8.568e-05, forward_time=0.290, loss_ctc=48.580, loss_att=48.277, acc=0.731, loss=48.368, backward_time=0.405, grad_norm=35.934, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.155e-04, train_time=1.366
+[gpub058:0/16] 2024-02-03 00:55:13,564 (trainer:737) INFO: 22epoch:train:8101-8200batch: iter_time=8.795e-05, forward_time=0.287, loss_ctc=52.379, loss_att=43.114, acc=0.743, loss=45.894, backward_time=0.405, grad_norm=40.289, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.155e-04, train_time=1.472
+[gpub058:0/16] 2024-02-03 00:57:59,347 (trainer:737) INFO: 22epoch:train:8201-8300batch: iter_time=8.282e-05, forward_time=0.386, loss_ctc=53.573, loss_att=48.979, acc=0.750, loss=50.357, backward_time=0.438, grad_norm=38.457, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.154e-04, train_time=1.657
+[gpub058:0/16] 2024-02-03 01:00:15,967 (trainer:737) INFO: 22epoch:train:8301-8400batch: iter_time=8.517e-05, forward_time=0.290, loss_ctc=49.159, loss_att=53.407, acc=0.741, loss=52.133, backward_time=0.404, grad_norm=37.424, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.154e-04, train_time=1.367
+[gpub058:0/16] 2024-02-03 01:02:57,826 (trainer:737) INFO: 22epoch:train:8401-8500batch: iter_time=3.523e-04, forward_time=0.402, loss_ctc=49.704, loss_att=47.822, acc=0.731, loss=48.386, backward_time=0.435, grad_norm=36.285, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.154e-04, train_time=1.618
+[gpub058:0/16] 2024-02-03 01:05:30,104 (trainer:737) INFO: 22epoch:train:8501-8600batch: iter_time=8.976e-05, forward_time=0.290, loss_ctc=56.390, loss_att=55.994, acc=0.739, loss=56.113, backward_time=0.404, grad_norm=40.067, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.153e-04, train_time=1.522
+[gpub058:0/16] 2024-02-03 01:07:47,423 (trainer:737) INFO: 22epoch:train:8601-8700batch: iter_time=8.565e-05, forward_time=0.291, loss_ctc=48.017, loss_att=48.385, acc=0.737, loss=48.274, backward_time=0.405, grad_norm=34.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.153e-04, train_time=1.374
+[gpub058:0/16] 2024-02-03 01:09:27,114 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-03 01:09:45,948 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 01:09:49,636 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec858e8b20>)
+[gpub058:0/16] 2024-02-03 01:09:49,636 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-03 01:09:49,639 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 01:15:42,539 (trainer:737) INFO: 22epoch:train:8701-8800batch: iter_time=3.206, forward_time=0.419, loss_ctc=50.350, loss_att=55.746, acc=0.732, loss=54.127, backward_time=0.430, grad_norm=35.287, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.153e-04, train_time=4.751
+[gpub058:0/16] 2024-02-03 01:18:20,726 (trainer:737) INFO: 22epoch:train:8801-8900batch: iter_time=2.864e-04, forward_time=0.409, loss_ctc=52.057, loss_att=58.505, acc=0.713, loss=56.571, backward_time=0.430, grad_norm=38.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.152e-04, train_time=1.582
+[gpub058:0/16] 2024-02-03 01:21:14,784 (trainer:737) INFO: 22epoch:train:8901-9000batch: iter_time=8.226e-05, forward_time=0.352, loss_ctc=41.931, loss_att=37.602, acc=0.765, loss=38.900, backward_time=0.486, grad_norm=29.814, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=2.152e-04, train_time=1.739
+[gpub058:0/16] 2024-02-03 01:24:00,061 (trainer:737) INFO: 22epoch:train:9001-9100batch: iter_time=6.086e-04, forward_time=0.405, loss_ctc=53.750, loss_att=52.271, acc=0.757, loss=52.715, backward_time=0.429, grad_norm=34.440, clip=100.000, loss_scale=8.619e+33, optim_step_time=0.109, optim0_lr0=2.152e-04, train_time=1.653
+[gpub058:0/16] 2024-02-03 01:26:36,705 (trainer:737) INFO: 22epoch:train:9101-9200batch: iter_time=7.822e-05, forward_time=0.389, loss_ctc=54.552, loss_att=53.830, acc=0.745, loss=54.047, backward_time=0.458, grad_norm=35.089, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.151e-04, train_time=1.567
+[gpub058:0/16] 2024-02-03 01:29:38,596 (trainer:737) INFO: 22epoch:train:9201-9300batch: iter_time=1.676e-04, forward_time=0.369, loss_ctc=52.281, loss_att=49.550, acc=0.757, loss=50.369, backward_time=0.467, grad_norm=34.899, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.151e-04, train_time=1.818
+[gpub058:0/16] 2024-02-03 01:31:49,938 (trainer:737) INFO: 22epoch:train:9301-9400batch: iter_time=7.846e-05, forward_time=0.293, loss_ctc=45.622, loss_att=46.130, acc=0.741, loss=45.977, backward_time=0.404, grad_norm=34.316, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.151e-04, train_time=1.314
+[gpub058:0/16] 2024-02-03 01:34:30,949 (trainer:737) INFO: 22epoch:train:9401-9500batch: iter_time=8.166e-05, forward_time=0.405, loss_ctc=53.151, loss_att=44.191, acc=0.751, loss=46.879, backward_time=0.427, grad_norm=38.780, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.150e-04, train_time=1.609
+[gpub058:0/16] 2024-02-03 01:37:04,742 (trainer:737) INFO: 22epoch:train:9501-9600batch: iter_time=8.798e-05, forward_time=0.296, loss_ctc=57.247, loss_att=58.903, acc=0.746, loss=58.407, backward_time=0.406, grad_norm=37.001, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.150e-04, train_time=1.539
+[gpub058:0/16] 2024-02-03 01:39:41,763 (trainer:737) INFO: 22epoch:train:9601-9700batch: iter_time=3.024e-04, forward_time=0.378, loss_ctc=47.933, loss_att=53.065, acc=0.739, loss=51.526, backward_time=0.449, grad_norm=35.860, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.150e-04, train_time=1.569
+[gpub058:0/16] 2024-02-03 01:41:56,670 (trainer:737) INFO: 22epoch:train:9701-9800batch: iter_time=7.750e-05, forward_time=0.290, loss_ctc=47.317, loss_att=48.845, acc=0.733, loss=48.387, backward_time=0.403, grad_norm=35.385, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.149e-04, train_time=1.347
+[gpub058:0/16] 2024-02-03 01:44:38,127 (trainer:737) INFO: 22epoch:train:9801-9900batch: iter_time=4.114e-04, forward_time=0.354, loss_ctc=54.118, loss_att=55.381, acc=0.743, loss=55.002, backward_time=0.484, grad_norm=38.307, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.149e-04, train_time=1.616
+[gpub058:0/16] 2024-02-03 01:47:15,366 (trainer:737) INFO: 22epoch:train:9901-10000batch: iter_time=8.621e-05, forward_time=0.293, loss_ctc=47.763, loss_att=44.134, acc=0.754, loss=45.223, backward_time=0.402, grad_norm=33.586, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.149e-04, train_time=1.572
+[gpub058:0/16] 2024-02-03 01:47:35,599 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-03 01:47:54,240 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 01:47:57,864 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c8ce1630>)
+[gpub058:0/16] 2024-02-03 01:47:57,865 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-03 01:47:57,868 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 01:54:57,694 (trainer:737) INFO: 22epoch:train:10001-10100batch: iter_time=3.111, forward_time=0.384, loss_ctc=51.723, loss_att=57.270, acc=0.735, loss=55.606, backward_time=0.439, grad_norm=38.253, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.148e-04, train_time=4.624
+[gpub058:0/16] 2024-02-03 01:57:23,563 (trainer:737) INFO: 22epoch:train:10101-10200batch: iter_time=8.089e-05, forward_time=0.288, loss_ctc=46.786, loss_att=50.283, acc=0.731, loss=49.234, backward_time=0.404, grad_norm=33.774, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.148e-04, train_time=1.457
+[gpub058:0/16] 2024-02-03 02:00:06,778 (trainer:737) INFO: 22epoch:train:10201-10300batch: iter_time=8.189e-05, forward_time=0.298, loss_ctc=51.752, loss_att=45.335, acc=0.768, loss=47.260, backward_time=0.406, grad_norm=32.286, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.148e-04, train_time=1.634
+[gpub058:0/16] 2024-02-03 02:02:47,616 (trainer:737) INFO: 22epoch:train:10301-10400batch: iter_time=8.612e-05, forward_time=0.424, loss_ctc=49.480, loss_att=52.527, acc=0.742, loss=51.613, backward_time=0.428, grad_norm=34.832, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.147e-04, train_time=1.608
+[gpub058:0/16] 2024-02-03 02:05:23,968 (trainer:737) INFO: 22epoch:train:10401-10500batch: iter_time=8.649e-05, forward_time=0.291, loss_ctc=54.893, loss_att=50.454, acc=0.757, loss=51.786, backward_time=0.405, grad_norm=37.263, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.147e-04, train_time=1.562
+[gpub058:0/16] 2024-02-03 02:08:00,523 (trainer:737) INFO: 22epoch:train:10501-10600batch: iter_time=8.532e-05, forward_time=0.410, loss_ctc=48.337, loss_att=48.260, acc=0.745, loss=48.283, backward_time=0.436, grad_norm=35.360, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.147e-04, train_time=1.567
+[gpub058:0/16] 2024-02-03 02:10:42,174 (trainer:737) INFO: 22epoch:train:10601-10700batch: iter_time=8.673e-05, forward_time=0.288, loss_ctc=51.698, loss_att=44.120, acc=0.750, loss=46.394, backward_time=0.402, grad_norm=38.163, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.146e-04, train_time=1.616
+[gpub058:0/16] 2024-02-03 02:13:04,355 (trainer:737) INFO: 22epoch:train:10701-10800batch: iter_time=8.389e-05, forward_time=0.333, loss_ctc=53.604, loss_att=49.082, acc=0.759, loss=50.439, backward_time=0.421, grad_norm=35.938, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.146e-04, train_time=1.421
+[gpub058:0/16] 2024-02-03 02:15:54,423 (trainer:737) INFO: 22epoch:train:10801-10900batch: iter_time=9.021e-05, forward_time=0.384, loss_ctc=49.008, loss_att=54.406, acc=0.744, loss=52.786, backward_time=0.423, grad_norm=33.964, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.146e-04, train_time=1.701
+[gpub058:0/16] 2024-02-03 02:18:20,796 (trainer:737) INFO: 22epoch:train:10901-11000batch: iter_time=8.535e-05, forward_time=0.287, loss_ctc=49.829, loss_att=49.195, acc=0.734, loss=49.385, backward_time=0.400, grad_norm=37.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.145e-04, train_time=1.464
+[gpub058:0/16] 2024-02-03 02:20:17,061 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 02:20:49,844 (trainer:737) INFO: 22epoch:train:11001-11100batch: iter_time=8.282e-05, forward_time=0.354, loss_ctc=55.549, loss_att=56.173, acc=0.741, loss=55.986, backward_time=0.476, grad_norm=39.487, clip=100.000, loss_scale=1.458e+34, optim_step_time=0.102, optim0_lr0=2.145e-04, train_time=1.490
+[gpub058:0/16] 2024-02-03 02:21:14,432 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 02:23:25,304 (trainer:737) INFO: 22epoch:train:11101-11200batch: iter_time=9.115e-05, forward_time=0.289, loss_ctc=48.161, loss_att=48.919, acc=0.743, loss=48.692, backward_time=0.403, grad_norm=33.252, clip=100.000, loss_scale=6.136e+33, optim_step_time=0.091, optim0_lr0=2.145e-04, train_time=1.555
+[gpub058:0/16] 2024-02-03 02:24:53,219 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-03 02:25:12,724 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 02:25:16,295 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c8c5f610>)
+[gpub058:0/16] 2024-02-03 02:25:16,296 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-03 02:25:16,299 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 02:30:54,566 (trainer:737) INFO: 22epoch:train:11201-11300batch: iter_time=3.060, forward_time=0.330, loss_ctc=49.675, loss_att=55.146, acc=0.737, loss=53.505, backward_time=0.411, grad_norm=37.522, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.144e-04, train_time=4.492
+[gpub058:0/16] 2024-02-03 02:33:41,864 (trainer:737) INFO: 22epoch:train:11301-11400batch: iter_time=7.636e-05, forward_time=0.330, loss_ctc=51.704, loss_att=56.672, acc=0.711, loss=55.181, backward_time=0.422, grad_norm=38.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.144e-04, train_time=1.672
+[gpub058:0/16] 2024-02-03 02:36:08,831 (trainer:737) INFO: 22epoch:train:11401-11500batch: iter_time=8.306e-05, forward_time=0.360, loss_ctc=41.820, loss_att=37.683, acc=0.757, loss=38.924, backward_time=0.428, grad_norm=29.525, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.144e-04, train_time=1.470
+[gpub058:0/16] 2024-02-03 02:39:01,612 (trainer:737) INFO: 22epoch:train:11501-11600batch: iter_time=9.158e-05, forward_time=0.323, loss_ctc=53.320, loss_att=51.718, acc=0.753, loss=52.198, backward_time=0.410, grad_norm=36.060, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.143e-04, train_time=1.728
+[gpub058:0/16] 2024-02-03 02:41:17,841 (trainer:737) INFO: 22epoch:train:11601-11700batch: iter_time=8.745e-05, forward_time=0.293, loss_ctc=54.218, loss_att=52.272, acc=0.743, loss=52.856, backward_time=0.406, grad_norm=35.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.143e-04, train_time=1.362
+[gpub058:0/16] 2024-02-03 02:44:04,831 (trainer:737) INFO: 22epoch:train:11701-11800batch: iter_time=4.196e-04, forward_time=0.316, loss_ctc=52.450, loss_att=47.782, acc=0.747, loss=49.183, backward_time=0.416, grad_norm=36.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.143e-04, train_time=1.668
+[gpub058:0/16] 2024-02-03 02:46:27,803 (trainer:737) INFO: 22epoch:train:11801-11900batch: iter_time=8.328e-05, forward_time=0.356, loss_ctc=45.977, loss_att=44.310, acc=0.737, loss=44.810, backward_time=0.414, grad_norm=34.076, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.142e-04, train_time=1.431
+[gpub058:0/16] 2024-02-03 02:49:16,193 (trainer:737) INFO: 22epoch:train:11901-12000batch: iter_time=8.476e-05, forward_time=0.288, loss_ctc=52.100, loss_att=43.602, acc=0.744, loss=46.152, backward_time=0.398, grad_norm=40.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.142e-04, train_time=1.683
+[gpub058:0/16] 2024-02-03 02:51:39,496 (trainer:737) INFO: 22epoch:train:12001-12100batch: iter_time=4.064e-04, forward_time=0.329, loss_ctc=56.963, loss_att=58.404, acc=0.737, loss=57.971, backward_time=0.462, grad_norm=39.344, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.142e-04, train_time=1.433
+[gpub058:0/16] 2024-02-03 02:54:05,061 (trainer:737) INFO: 22epoch:train:12101-12200batch: iter_time=8.791e-05, forward_time=0.338, loss_ctc=47.564, loss_att=53.367, acc=0.732, loss=51.626, backward_time=0.418, grad_norm=36.520, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.141e-04, train_time=1.456
+[gpub058:0/16] 2024-02-03 02:56:55,343 (trainer:737) INFO: 22epoch:train:12201-12300batch: iter_time=1.027e-04, forward_time=0.313, loss_ctc=46.717, loss_att=48.209, acc=0.732, loss=47.762, backward_time=0.425, grad_norm=36.644, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.141e-04, train_time=1.702
+[gpub058:0/16] 2024-02-03 02:59:11,081 (trainer:737) INFO: 22epoch:train:12301-12400batch: iter_time=9.116e-05, forward_time=0.312, loss_ctc=53.995, loss_att=55.080, acc=0.738, loss=54.754, backward_time=0.414, grad_norm=37.800, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.141e-04, train_time=1.358
+[gpub058:0/16] 2024-02-03 03:02:04,814 (trainer:737) INFO: 22epoch:train:12401-12500batch: iter_time=8.637e-05, forward_time=0.327, loss_ctc=47.496, loss_att=44.960, acc=0.738, loss=45.721, backward_time=0.431, grad_norm=34.649, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.140e-04, train_time=1.736
+[gpub058:0/16] 2024-02-03 03:02:24,872 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-03 03:02:43,740 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 03:02:47,238 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1e7597610>)
+[gpub058:0/16] 2024-02-03 03:02:47,238 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-03 03:02:47,241 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 03:09:13,002 (trainer:737) INFO: 22epoch:train:12501-12600batch: iter_time=2.802, forward_time=0.324, loss_ctc=51.748, loss_att=58.705, acc=0.733, loss=56.618, backward_time=0.409, grad_norm=37.083, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.140e-04, train_time=4.282
+[gpub058:0/16] 2024-02-03 03:12:05,177 (trainer:737) INFO: 22epoch:train:12601-12700batch: iter_time=8.414e-05, forward_time=0.329, loss_ctc=46.801, loss_att=50.786, acc=0.730, loss=49.591, backward_time=0.416, grad_norm=130.987, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.140e-04, train_time=1.722
+[gpub058:0/16] 2024-02-03 03:14:29,086 (trainer:737) INFO: 22epoch:train:12701-12800batch: iter_time=8.360e-05, forward_time=0.375, loss_ctc=51.528, loss_att=45.076, acc=0.767, loss=47.011, backward_time=0.421, grad_norm=31.587, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.139e-04, train_time=1.439
+[gpub058:0/16] 2024-02-03 03:17:15,074 (trainer:737) INFO: 22epoch:train:12801-12900batch: iter_time=8.793e-05, forward_time=0.290, loss_ctc=49.206, loss_att=52.643, acc=0.741, loss=51.612, backward_time=0.403, grad_norm=36.493, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.139e-04, train_time=1.660
+[gpub058:0/16] 2024-02-03 03:19:51,241 (trainer:737) INFO: 22epoch:train:12901-13000batch: iter_time=1.046e-04, forward_time=0.351, loss_ctc=54.860, loss_att=50.896, acc=0.756, loss=52.085, backward_time=0.418, grad_norm=37.526, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.139e-04, train_time=1.562
+[gpub058:0/16] 2024-02-03 03:22:38,325 (trainer:737) INFO: 22epoch:train:13001-13100batch: iter_time=2.332e-04, forward_time=0.312, loss_ctc=48.788, loss_att=48.619, acc=0.746, loss=48.670, backward_time=0.435, grad_norm=35.273, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.138e-04, train_time=1.670
+[gpub058:0/16] 2024-02-03 03:25:08,371 (trainer:737) INFO: 22epoch:train:13101-13200batch: iter_time=8.872e-05, forward_time=0.289, loss_ctc=50.875, loss_att=44.430, acc=0.750, loss=46.364, backward_time=0.400, grad_norm=38.619, clip=100.000, loss_scale=9.398e+33, optim_step_time=0.091, optim0_lr0=2.138e-04, train_time=1.500
+[gpub058:0/16] 2024-02-03 03:27:47,154 (trainer:737) INFO: 22epoch:train:13201-13300batch: iter_time=1.180e-04, forward_time=0.351, loss_ctc=52.858, loss_att=48.543, acc=0.760, loss=49.837, backward_time=0.414, grad_norm=34.418, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.138e-04, train_time=1.588
+[gpub058:0/16] 2024-02-03 03:30:16,848 (trainer:737) INFO: 22epoch:train:13301-13400batch: iter_time=8.396e-05, forward_time=0.303, loss_ctc=49.060, loss_att=53.704, acc=0.746, loss=52.311, backward_time=0.442, grad_norm=37.799, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.138e-04, train_time=1.496
+[gpub058:0/16] 2024-02-03 03:32:41,460 (trainer:737) INFO: 22epoch:train:13401-13500batch: iter_time=8.930e-05, forward_time=0.290, loss_ctc=49.287, loss_att=48.869, acc=0.737, loss=48.995, backward_time=0.403, grad_norm=36.542, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.137e-04, train_time=1.446
+[gpub058:0/16] 2024-02-03 03:35:36,710 (trainer:737) INFO: 22epoch:train:13501-13600batch: iter_time=9.990e-05, forward_time=0.349, loss_ctc=54.809, loss_att=56.411, acc=0.743, loss=55.931, backward_time=0.418, grad_norm=34.903, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.137e-04, train_time=1.752
+[gpub058:0/16] 2024-02-03 03:38:09,676 (trainer:737) INFO: 22epoch:train:13601-13700batch: iter_time=2.242e-04, forward_time=0.310, loss_ctc=47.673, loss_att=48.559, acc=0.744, loss=48.293, backward_time=0.443, grad_norm=34.067, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.137e-04, train_time=1.528
+[gpub058:0/16] 2024-02-03 03:39:48,161 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-03 03:40:07,004 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 03:40:10,522 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec5601bca0>)
+[gpub058:0/16] 2024-02-03 03:40:10,522 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-03 03:40:10,525 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 03:45:48,627 (trainer:737) INFO: 22epoch:train:13701-13800batch: iter_time=3.061, forward_time=0.337, loss_ctc=50.152, loss_att=53.601, acc=0.746, loss=52.566, backward_time=0.409, grad_norm=36.162, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.136e-04, train_time=4.590
+[gpub058:0/16] 2024-02-03 03:47:02,743 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 03:48:23,932 (trainer:737) INFO: 22epoch:train:13801-13900batch: iter_time=7.960e-05, forward_time=0.330, loss_ctc=51.492, loss_att=55.862, acc=0.719, loss=54.551, backward_time=0.431, grad_norm=38.086, clip=100.000, loss_scale=7.710e+33, optim_step_time=0.096, optim0_lr0=2.136e-04, train_time=1.553
+[gpub058:0/16] 2024-02-03 03:51:03,367 (trainer:737) INFO: 22epoch:train:13901-14000batch: iter_time=8.045e-05, forward_time=0.288, loss_ctc=41.146, loss_att=36.387, acc=0.769, loss=37.815, backward_time=0.401, grad_norm=29.354, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.136e-04, train_time=1.592
+[gpub058:0/16] 2024-02-03 03:53:35,257 (trainer:737) INFO: 22epoch:train:14001-14100batch: iter_time=8.033e-05, forward_time=0.421, loss_ctc=52.815, loss_att=51.156, acc=0.759, loss=51.653, backward_time=0.433, grad_norm=35.543, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.135e-04, train_time=1.521
+[gpub058:0/16] 2024-02-03 03:56:03,598 (trainer:737) INFO: 22epoch:train:14101-14200batch: iter_time=7.994e-05, forward_time=0.291, loss_ctc=54.308, loss_att=52.523, acc=0.747, loss=53.059, backward_time=0.405, grad_norm=35.214, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.135e-04, train_time=1.484
+[gpub058:0/16] 2024-02-03 03:58:42,489 (trainer:737) INFO: 22epoch:train:14201-14300batch: iter_time=8.164e-05, forward_time=0.326, loss_ctc=52.227, loss_att=48.222, acc=0.760, loss=49.424, backward_time=0.413, grad_norm=35.670, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.135e-04, train_time=1.588
+[gpub058:0/16] 2024-02-03 04:01:19,509 (trainer:737) INFO: 22epoch:train:14301-14400batch: iter_time=2.542e-04, forward_time=0.336, loss_ctc=45.418, loss_att=45.666, acc=0.741, loss=45.591, backward_time=0.424, grad_norm=33.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.134e-04, train_time=1.570
+[gpub058:0/16] 2024-02-03 04:03:29,767 (trainer:737) INFO: 22epoch:train:14401-14500batch: iter_time=7.956e-05, forward_time=0.288, loss_ctc=51.726, loss_att=42.976, acc=0.751, loss=45.601, backward_time=0.400, grad_norm=41.077, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.134e-04, train_time=1.303
+[gpub058:0/16] 2024-02-03 04:06:10,302 (trainer:737) INFO: 22epoch:train:14501-14600batch: iter_time=8.087e-05, forward_time=0.323, loss_ctc=56.438, loss_att=57.950, acc=0.750, loss=57.496, backward_time=0.421, grad_norm=37.255, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.134e-04, train_time=1.605
+[gpub058:0/16] 2024-02-03 04:08:44,586 (trainer:737) INFO: 22epoch:train:14601-14700batch: iter_time=4.724e-04, forward_time=0.311, loss_ctc=47.218, loss_att=52.884, acc=0.741, loss=51.184, backward_time=0.421, grad_norm=34.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.133e-04, train_time=1.543
+[gpub058:0/16] 2024-02-03 04:10:55,110 (trainer:737) INFO: 22epoch:train:14701-14800batch: iter_time=8.106e-05, forward_time=0.290, loss_ctc=46.451, loss_att=48.028, acc=0.736, loss=47.555, backward_time=0.402, grad_norm=33.503, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.133e-04, train_time=1.305
+[gpub058:0/16] 2024-02-03 04:13:44,423 (trainer:737) INFO: 22epoch:train:14801-14900batch: iter_time=3.924e-04, forward_time=0.371, loss_ctc=53.077, loss_att=54.845, acc=0.745, loss=54.315, backward_time=0.447, grad_norm=35.861, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.133e-04, train_time=1.692
+[gpub058:0/16] 2024-02-03 04:16:09,738 (trainer:737) INFO: 22epoch:train:14901-15000batch: iter_time=7.975e-05, forward_time=0.291, loss_ctc=47.528, loss_att=43.855, acc=0.755, loss=44.957, backward_time=0.403, grad_norm=33.806, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.090, optim0_lr0=2.132e-04, train_time=1.454
+[gpub058:0/16] 2024-02-03 04:53:13,170 (trainer:343) INFO: 22epoch results: [train] iter_time=0.247, forward_time=0.329, loss_ctc=51.350, loss_att=50.500, acc=0.740, loss=50.755, backward_time=0.419, grad_norm=37.038, clip=100.000, loss_scale=7.013e+33, optim_step_time=0.094, optim0_lr0=2.157e-04, train_time=1.852, time=7 hours, 43 minutes and 28.6 seconds, total_count=360000, gpu_max_cached_mem_GB=43.281, [valid] loss_ctc=39.810, cer_ctc=0.204, loss_att=39.841, acc=0.671, cer=0.327, wer=0.992, loss=39.831, time=36 minutes and 39.42 seconds, total_count=112104, gpu_max_cached_mem_GB=43.281
+[gpub058:0/16] 2024-02-03 04:53:28,993 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub058:0/16] 2024-02-03 04:53:29,098 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/14epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/17epoch.pth
+[gpub058:0/16] 2024-02-03 04:53:29,098 (trainer:272) INFO: 23/45epoch started. Estimated time to finish: 1 week, 23 hours and 49 minutes
+[gpub058:0/16] 2024-02-03 04:53:29,108 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-03 04:53:46,653 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 04:53:49,993 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1f90523b0>)
+[gpub058:0/16] 2024-02-03 04:53:49,993 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-03 04:53:49,996 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 05:00:18,603 (trainer:737) INFO: 23epoch:train:1-100batch: iter_time=2.657, forward_time=0.345, loss_ctc=48.675, loss_att=38.911, acc=0.768, loss=41.840, backward_time=0.412, grad_norm=38.161, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.132e-04, train_time=4.095
+[gpub058:0/16] 2024-02-03 05:02:45,266 (trainer:737) INFO: 23epoch:train:101-200batch: iter_time=8.325e-05, forward_time=0.351, loss_ctc=45.886, loss_att=49.549, acc=0.734, loss=48.450, backward_time=0.466, grad_norm=34.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.132e-04, train_time=1.466
+[gpub058:0/16] 2024-02-03 05:05:29,783 (trainer:737) INFO: 23epoch:train:201-300batch: iter_time=2.205e-04, forward_time=0.376, loss_ctc=57.730, loss_att=46.255, acc=0.757, loss=49.697, backward_time=0.453, grad_norm=43.856, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.131e-04, train_time=1.645
+[gpub058:0/16] 2024-02-03 05:07:54,621 (trainer:737) INFO: 23epoch:train:301-400batch: iter_time=7.955e-05, forward_time=0.308, loss_ctc=44.920, loss_att=49.606, acc=0.728, loss=48.200, backward_time=0.420, grad_norm=37.150, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.131e-04, train_time=1.448
+[gpub058:0/16] 2024-02-03 05:10:25,142 (trainer:737) INFO: 23epoch:train:401-500batch: iter_time=8.253e-05, forward_time=0.319, loss_ctc=61.257, loss_att=59.983, acc=0.727, loss=60.366, backward_time=0.416, grad_norm=49.477, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.131e-04, train_time=1.505
+[gpub058:0/16] 2024-02-03 05:13:02,591 (trainer:737) INFO: 23epoch:train:501-600batch: iter_time=5.192e-04, forward_time=0.361, loss_ctc=53.167, loss_att=47.817, acc=0.761, loss=49.422, backward_time=0.453, grad_norm=35.013, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.130e-04, train_time=1.574
+[gpub058:0/16] 2024-02-03 05:15:30,555 (trainer:737) INFO: 23epoch:train:601-700batch: iter_time=5.661e-04, forward_time=0.367, loss_ctc=51.050, loss_att=52.503, acc=0.739, loss=52.067, backward_time=0.442, grad_norm=36.419, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.130e-04, train_time=1.480
+[gpub058:0/16] 2024-02-03 05:18:11,254 (trainer:737) INFO: 23epoch:train:701-800batch: iter_time=8.446e-05, forward_time=0.338, loss_ctc=50.481, loss_att=42.897, acc=0.757, loss=45.172, backward_time=0.454, grad_norm=35.924, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.130e-04, train_time=1.607
+[gpub058:0/16] 2024-02-03 05:20:47,776 (trainer:737) INFO: 23epoch:train:801-900batch: iter_time=4.800e-04, forward_time=0.316, loss_ctc=55.886, loss_att=47.281, acc=0.746, loss=49.863, backward_time=0.423, grad_norm=44.063, clip=100.000, loss_scale=7.840e+33, optim_step_time=0.096, optim0_lr0=2.129e-04, train_time=1.565
+[gpub058:0/16] 2024-02-03 05:23:16,817 (trainer:737) INFO: 23epoch:train:901-1000batch: iter_time=7.480e-04, forward_time=0.365, loss_ctc=50.822, loss_att=53.939, acc=0.747, loss=53.004, backward_time=0.446, grad_norm=35.068, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.129e-04, train_time=1.490
+[gpub058:0/16] 2024-02-03 05:25:48,680 (trainer:737) INFO: 23epoch:train:1001-1100batch: iter_time=4.711e-04, forward_time=0.359, loss_ctc=48.883, loss_att=47.305, acc=0.737, loss=47.778, backward_time=0.451, grad_norm=36.612, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.129e-04, train_time=1.519
+[gpub058:0/16] 2024-02-03 05:28:38,486 (trainer:737) INFO: 23epoch:train:1101-1200batch: iter_time=0.001, forward_time=0.356, loss_ctc=61.897, loss_att=65.780, acc=0.701, loss=64.615, backward_time=0.441, grad_norm=48.801, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.128e-04, train_time=1.698
+[gpub058:0/16] 2024-02-03 05:30:14,713 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-03 05:30:33,659 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 05:30:37,275 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec0ff47f40>)
+[gpub058:0/16] 2024-02-03 05:30:37,275 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-03 05:30:37,279 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 05:36:10,430 (trainer:737) INFO: 23epoch:train:1201-1300batch: iter_time=3.038, forward_time=0.348, loss_ctc=44.341, loss_att=37.322, acc=0.761, loss=39.428, backward_time=0.418, grad_norm=35.045, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.128e-04, train_time=4.520
+[gpub058:0/16] 2024-02-03 05:37:30,573 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 05:38:35,542 (trainer:737) INFO: 23epoch:train:1301-1400batch: iter_time=7.993e-05, forward_time=0.324, loss_ctc=46.230, loss_att=42.961, acc=0.748, loss=43.942, backward_time=0.412, grad_norm=36.547, clip=100.000, loss_scale=8.234e+33, optim_step_time=0.093, optim0_lr0=2.128e-04, train_time=1.450
+[gpub058:0/16] 2024-02-03 05:41:11,618 (trainer:737) INFO: 23epoch:train:1401-1500batch: iter_time=8.753e-05, forward_time=0.346, loss_ctc=54.703, loss_att=52.927, acc=0.746, loss=53.459, backward_time=0.430, grad_norm=39.896, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.128e-04, train_time=1.560
+[gpub058:0/16] 2024-02-03 05:43:25,030 (trainer:737) INFO: 23epoch:train:1501-1600batch: iter_time=9.607e-05, forward_time=0.290, loss_ctc=45.957, loss_att=46.148, acc=0.749, loss=46.091, backward_time=0.403, grad_norm=35.084, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.127e-04, train_time=1.335
+[gpub058:0/16] 2024-02-03 05:45:47,956 (trainer:737) INFO: 23epoch:train:1601-1700batch: iter_time=9.147e-05, forward_time=0.318, loss_ctc=47.807, loss_att=51.435, acc=0.725, loss=50.346, backward_time=0.418, grad_norm=36.134, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.127e-04, train_time=1.429
+[gpub058:0/16] 2024-02-03 05:48:26,955 (trainer:737) INFO: 23epoch:train:1701-1800batch: iter_time=8.991e-05, forward_time=0.329, loss_ctc=58.813, loss_att=54.698, acc=0.751, loss=55.933, backward_time=0.417, grad_norm=47.770, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.127e-04, train_time=1.589
+[gpub058:0/16] 2024-02-03 05:50:52,027 (trainer:737) INFO: 23epoch:train:1801-1900batch: iter_time=9.214e-05, forward_time=0.331, loss_ctc=49.506, loss_att=47.621, acc=0.753, loss=48.186, backward_time=0.419, grad_norm=32.265, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.126e-04, train_time=1.450
+[gpub058:0/16] 2024-02-03 05:53:32,329 (trainer:737) INFO: 23epoch:train:1901-2000batch: iter_time=8.735e-05, forward_time=0.327, loss_ctc=50.877, loss_att=47.230, acc=0.749, loss=48.324, backward_time=0.410, grad_norm=35.214, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.126e-04, train_time=1.604
+[gpub058:0/16] 2024-02-03 05:55:57,176 (trainer:737) INFO: 23epoch:train:2001-2100batch: iter_time=8.986e-05, forward_time=0.308, loss_ctc=53.709, loss_att=44.980, acc=0.759, loss=47.598, backward_time=0.411, grad_norm=38.431, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.126e-04, train_time=1.448
+[gpub058:0/16] 2024-02-03 05:58:23,072 (trainer:737) INFO: 23epoch:train:2101-2200batch: iter_time=1.027e-04, forward_time=0.336, loss_ctc=53.783, loss_att=54.688, acc=0.738, loss=54.416, backward_time=0.429, grad_norm=37.919, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.125e-04, train_time=1.458
+[gpub058:0/16] 2024-02-03 06:00:56,544 (trainer:737) INFO: 23epoch:train:2201-2300batch: iter_time=8.169e-05, forward_time=0.316, loss_ctc=45.687, loss_att=45.971, acc=0.754, loss=45.886, backward_time=0.406, grad_norm=31.252, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.125e-04, train_time=1.534
+[gpub058:0/16] 2024-02-03 06:03:21,161 (trainer:737) INFO: 23epoch:train:2301-2400batch: iter_time=8.976e-05, forward_time=0.309, loss_ctc=59.312, loss_att=56.915, acc=0.716, loss=57.634, backward_time=0.411, grad_norm=43.625, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.125e-04, train_time=1.447
+[gpub058:0/16] 2024-02-03 06:05:54,295 (trainer:737) INFO: 23epoch:train:2401-2500batch: iter_time=8.717e-05, forward_time=0.348, loss_ctc=45.848, loss_att=50.824, acc=0.728, loss=49.331, backward_time=0.421, grad_norm=38.792, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.124e-04, train_time=1.530
+[gpub058:0/16] 2024-02-03 06:06:14,493 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-03 06:06:33,044 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 06:06:36,533 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec4d20b850>)
+[gpub058:0/16] 2024-02-03 06:06:36,533 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-03 06:06:36,573 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 06:13:22,739 (trainer:737) INFO: 23epoch:train:2501-2600batch: iter_time=2.925, forward_time=0.384, loss_ctc=47.007, loss_att=40.988, acc=0.754, loss=42.794, backward_time=0.421, grad_norm=38.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.124e-04, train_time=4.485
+[gpub058:0/16] 2024-02-03 06:15:52,397 (trainer:737) INFO: 23epoch:train:2601-2700batch: iter_time=8.190e-05, forward_time=0.330, loss_ctc=44.493, loss_att=48.528, acc=0.732, loss=47.317, backward_time=0.409, grad_norm=33.229, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.124e-04, train_time=1.496
+[gpub058:0/16] 2024-02-03 06:18:16,438 (trainer:737) INFO: 23epoch:train:2701-2800batch: iter_time=8.206e-05, forward_time=0.319, loss_ctc=52.554, loss_att=47.571, acc=0.746, loss=49.066, backward_time=0.459, grad_norm=42.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.123e-04, train_time=1.441
+[gpub058:0/16] 2024-02-03 06:21:04,294 (trainer:737) INFO: 23epoch:train:2801-2900batch: iter_time=9.180e-05, forward_time=0.325, loss_ctc=43.902, loss_att=49.065, acc=0.729, loss=47.516, backward_time=0.400, grad_norm=35.526, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.123e-04, train_time=1.678
+[gpub058:0/16] 2024-02-03 06:23:29,796 (trainer:737) INFO: 23epoch:train:2901-3000batch: iter_time=8.959e-05, forward_time=0.337, loss_ctc=59.261, loss_att=57.702, acc=0.723, loss=58.170, backward_time=0.462, grad_norm=53.841, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.123e-04, train_time=1.454
+[gpub058:0/16] 2024-02-03 06:25:59,420 (trainer:737) INFO: 23epoch:train:3001-3100batch: iter_time=9.289e-05, forward_time=0.304, loss_ctc=51.844, loss_att=48.126, acc=0.749, loss=49.242, backward_time=0.406, grad_norm=35.373, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.122e-04, train_time=1.497
+[gpub058:0/16] 2024-02-03 06:28:39,274 (trainer:737) INFO: 23epoch:train:3101-3200batch: iter_time=9.075e-05, forward_time=0.347, loss_ctc=50.173, loss_att=50.581, acc=0.735, loss=50.459, backward_time=0.423, grad_norm=34.861, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.122e-04, train_time=1.598
+[gpub058:0/16] 2024-02-03 06:31:00,704 (trainer:737) INFO: 23epoch:train:3201-3300batch: iter_time=8.518e-05, forward_time=0.351, loss_ctc=48.529, loss_att=41.865, acc=0.760, loss=43.864, backward_time=0.416, grad_norm=36.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.122e-04, train_time=1.414
+[gpub058:0/16] 2024-02-03 06:33:42,771 (trainer:737) INFO: 23epoch:train:3301-3400batch: iter_time=9.129e-05, forward_time=0.316, loss_ctc=53.505, loss_att=44.814, acc=0.754, loss=47.421, backward_time=0.433, grad_norm=47.057, clip=100.000, loss_scale=7.321e+33, optim_step_time=0.094, optim0_lr0=2.121e-04, train_time=1.621
+[gpub058:0/16] 2024-02-03 06:36:19,245 (trainer:737) INFO: 23epoch:train:3401-3500batch: iter_time=8.467e-05, forward_time=0.372, loss_ctc=49.996, loss_att=53.699, acc=0.745, loss=52.588, backward_time=0.415, grad_norm=34.859, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.121e-04, train_time=1.564
+[gpub058:0/16] 2024-02-03 06:39:02,498 (trainer:737) INFO: 23epoch:train:3501-3600batch: iter_time=9.442e-05, forward_time=0.342, loss_ctc=47.484, loss_att=47.010, acc=0.737, loss=47.152, backward_time=0.410, grad_norm=35.697, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.121e-04, train_time=1.633
+[gpub058:0/16] 2024-02-03 06:41:25,546 (trainer:737) INFO: 23epoch:train:3601-3700batch: iter_time=4.998e-04, forward_time=0.344, loss_ctc=60.367, loss_att=63.216, acc=0.703, loss=62.361, backward_time=0.424, grad_norm=46.216, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.120e-04, train_time=1.430
+[gpub058:0/16] 2024-02-03 06:42:50,350 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-03 06:43:08,989 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 06:43:12,547 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c441d600>)
+[gpub058:0/16] 2024-02-03 06:43:12,547 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-03 06:43:12,550 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 06:48:58,420 (trainer:737) INFO: 23epoch:train:3701-3800batch: iter_time=3.116, forward_time=0.287, loss_ctc=43.119, loss_att=37.469, acc=0.762, loss=39.164, backward_time=0.402, grad_norm=36.154, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.120e-04, train_time=4.528
+[gpub058:0/16] 2024-02-03 06:50:49,843 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 06:51:31,389 (trainer:737) INFO: 23epoch:train:3801-3900batch: iter_time=2.734e-04, forward_time=0.406, loss_ctc=45.700, loss_att=43.597, acc=0.748, loss=44.228, backward_time=0.441, grad_norm=36.633, clip=100.000, loss_scale=8.706e+33, optim_step_time=0.099, optim0_lr0=2.120e-04, train_time=1.529
+[gpub058:0/16] 2024-02-03 06:54:10,145 (trainer:737) INFO: 23epoch:train:3901-4000batch: iter_time=8.373e-05, forward_time=0.290, loss_ctc=53.533, loss_att=53.411, acc=0.747, loss=53.448, backward_time=0.404, grad_norm=41.603, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.120e-04, train_time=1.588
+[gpub058:0/16] 2024-02-03 06:56:54,487 (trainer:737) INFO: 23epoch:train:4001-4100batch: iter_time=9.821e-05, forward_time=0.417, loss_ctc=45.640, loss_att=46.120, acc=0.752, loss=45.976, backward_time=0.427, grad_norm=33.993, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.119e-04, train_time=1.643
+[gpub058:0/16] 2024-02-03 06:59:10,439 (trainer:737) INFO: 23epoch:train:4101-4200batch: iter_time=9.241e-05, forward_time=0.289, loss_ctc=46.960, loss_att=51.335, acc=0.728, loss=50.022, backward_time=0.404, grad_norm=37.058, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.119e-04, train_time=1.359
+[gpub058:0/16] 2024-02-03 07:01:45,158 (trainer:737) INFO: 23epoch:train:4201-4300batch: iter_time=8.915e-05, forward_time=0.292, loss_ctc=58.358, loss_att=55.705, acc=0.752, loss=56.501, backward_time=0.406, grad_norm=47.202, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.119e-04, train_time=1.548
+[gpub058:0/16] 2024-02-03 07:04:29,008 (trainer:737) INFO: 23epoch:train:4301-4400batch: iter_time=8.590e-05, forward_time=0.379, loss_ctc=48.343, loss_att=47.139, acc=0.758, loss=47.500, backward_time=0.462, grad_norm=31.400, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.118e-04, train_time=1.638
+[gpub058:0/16] 2024-02-03 07:06:59,686 (trainer:737) INFO: 23epoch:train:4401-4500batch: iter_time=8.382e-05, forward_time=0.289, loss_ctc=50.454, loss_att=47.068, acc=0.751, loss=48.084, backward_time=0.402, grad_norm=37.204, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.118e-04, train_time=1.507
+[gpub058:0/16] 2024-02-03 07:09:35,198 (trainer:737) INFO: 23epoch:train:4501-4600batch: iter_time=9.267e-05, forward_time=0.290, loss_ctc=53.104, loss_att=44.300, acc=0.761, loss=46.941, backward_time=0.404, grad_norm=38.836, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.118e-04, train_time=1.554
+[gpub058:0/16] 2024-02-03 07:12:05,321 (trainer:737) INFO: 23epoch:train:4601-4700batch: iter_time=4.562e-04, forward_time=0.409, loss_ctc=52.534, loss_att=53.918, acc=0.741, loss=53.503, backward_time=0.420, grad_norm=37.643, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.117e-04, train_time=1.501
+[gpub058:0/16] 2024-02-03 07:14:50,431 (trainer:737) INFO: 23epoch:train:4701-4800batch: iter_time=9.082e-05, forward_time=0.323, loss_ctc=45.391, loss_att=45.851, acc=0.755, loss=45.713, backward_time=0.416, grad_norm=31.343, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.117e-04, train_time=1.651
+[gpub058:0/16] 2024-02-03 07:17:06,947 (trainer:737) INFO: 23epoch:train:4801-4900batch: iter_time=8.539e-05, forward_time=0.291, loss_ctc=57.959, loss_att=56.451, acc=0.718, loss=56.903, backward_time=0.406, grad_norm=42.835, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.117e-04, train_time=1.364
+[gpub058:0/16] 2024-02-03 07:19:50,651 (trainer:737) INFO: 23epoch:train:4901-5000batch: iter_time=8.657e-05, forward_time=0.356, loss_ctc=45.005, loss_att=50.952, acc=0.729, loss=49.168, backward_time=0.415, grad_norm=36.871, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.116e-04, train_time=1.637
+[gpub058:0/16] 2024-02-03 07:20:10,679 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-03 07:20:29,717 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 07:20:33,610 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9fe3bfcd0>)
+[gpub058:0/16] 2024-02-03 07:20:33,610 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-03 07:20:33,613 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 07:27:25,966 (trainer:737) INFO: 23epoch:train:5001-5100batch: iter_time=3.126, forward_time=0.332, loss_ctc=46.601, loss_att=37.950, acc=0.773, loss=40.545, backward_time=0.413, grad_norm=34.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.116e-04, train_time=4.553
+[gpub058:0/16] 2024-02-03 07:30:05,687 (trainer:737) INFO: 23epoch:train:5101-5200batch: iter_time=8.513e-05, forward_time=0.355, loss_ctc=43.960, loss_att=49.003, acc=0.740, loss=47.490, backward_time=0.405, grad_norm=33.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.116e-04, train_time=1.597
+[gpub058:0/16] 2024-02-03 07:32:39,410 (trainer:737) INFO: 23epoch:train:5201-5300batch: iter_time=8.114e-05, forward_time=0.308, loss_ctc=52.702, loss_att=45.907, acc=0.763, loss=47.946, backward_time=0.470, grad_norm=41.395, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.115e-04, train_time=1.537
+[gpub058:0/16] 2024-02-03 07:35:22,549 (trainer:737) INFO: 23epoch:train:5301-5400batch: iter_time=8.295e-05, forward_time=0.288, loss_ctc=43.336, loss_att=47.760, acc=0.737, loss=46.433, backward_time=0.402, grad_norm=34.297, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.115e-04, train_time=1.631
+[gpub058:0/16] 2024-02-03 07:37:55,614 (trainer:737) INFO: 23epoch:train:5401-5500batch: iter_time=9.229e-05, forward_time=0.360, loss_ctc=56.256, loss_att=58.962, acc=0.732, loss=58.150, backward_time=0.475, grad_norm=43.907, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.115e-04, train_time=1.530
+[gpub058:0/16] 2024-02-03 07:40:43,807 (trainer:737) INFO: 23epoch:train:5501-5600batch: iter_time=1.051e-04, forward_time=0.291, loss_ctc=51.651, loss_att=46.295, acc=0.770, loss=47.902, backward_time=0.405, grad_norm=33.469, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.114e-04, train_time=1.682
+[gpub058:0/16] 2024-02-03 07:43:07,727 (trainer:737) INFO: 23epoch:train:5601-5700batch: iter_time=8.195e-05, forward_time=0.364, loss_ctc=49.778, loss_att=51.610, acc=0.742, loss=51.061, backward_time=0.412, grad_norm=32.890, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.114e-04, train_time=1.439
+[gpub058:0/16] 2024-02-03 07:45:45,537 (trainer:737) INFO: 23epoch:train:5701-5800batch: iter_time=8.749e-05, forward_time=0.311, loss_ctc=47.851, loss_att=41.233, acc=0.766, loss=43.218, backward_time=0.446, grad_norm=34.604, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.114e-04, train_time=1.577
+[gpub058:0/16] 2024-02-03 07:48:20,539 (trainer:737) INFO: 23epoch:train:5801-5900batch: iter_time=9.100e-05, forward_time=0.291, loss_ctc=53.371, loss_att=46.389, acc=0.753, loss=48.484, backward_time=0.405, grad_norm=39.072, clip=100.000, loss_scale=6.854e+33, optim_step_time=0.092, optim0_lr0=2.114e-04, train_time=1.550
+[gpub058:0/16] 2024-02-03 07:50:58,815 (trainer:737) INFO: 23epoch:train:5901-6000batch: iter_time=8.510e-05, forward_time=0.339, loss_ctc=49.083, loss_att=53.002, acc=0.753, loss=51.826, backward_time=0.423, grad_norm=35.392, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.113e-04, train_time=1.583
+[gpub058:0/16] 2024-02-03 07:53:14,443 (trainer:737) INFO: 23epoch:train:6001-6100batch: iter_time=9.282e-05, forward_time=0.289, loss_ctc=47.612, loss_att=46.412, acc=0.743, loss=46.772, backward_time=0.403, grad_norm=35.477, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.113e-04, train_time=1.356
+[gpub058:0/16] 2024-02-03 07:56:15,835 (trainer:737) INFO: 23epoch:train:6101-6200batch: iter_time=8.948e-05, forward_time=0.293, loss_ctc=59.393, loss_att=64.002, acc=0.707, loss=62.619, backward_time=0.407, grad_norm=45.644, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.113e-04, train_time=1.814
+[gpub058:0/16] 2024-02-03 07:57:54,876 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-03 07:58:14,008 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 07:58:17,927 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d862feb0>)
+[gpub058:0/16] 2024-02-03 07:58:17,927 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-03 07:58:17,930 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 08:03:58,208 (trainer:737) INFO: 23epoch:train:6201-6300batch: iter_time=3.110, forward_time=0.381, loss_ctc=42.941, loss_att=37.469, acc=0.764, loss=39.110, backward_time=0.443, grad_norm=35.328, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.112e-04, train_time=4.623
+[gpub058:0/16] 2024-02-03 08:06:38,213 (trainer:737) INFO: 23epoch:train:6301-6400batch: iter_time=2.536e-04, forward_time=0.330, loss_ctc=44.702, loss_att=44.591, acc=0.734, loss=44.625, backward_time=0.405, grad_norm=38.617, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.112e-04, train_time=1.599
+[gpub058:0/16] 2024-02-03 08:08:23,368 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 08:09:19,501 (trainer:737) INFO: 23epoch:train:6401-6500batch: iter_time=0.001, forward_time=0.388, loss_ctc=52.479, loss_att=52.195, acc=0.747, loss=52.280, backward_time=0.424, grad_norm=41.081, clip=100.000, loss_scale=8.759e+33, optim_step_time=0.094, optim0_lr0=2.112e-04, train_time=1.613
+[gpub058:0/16] 2024-02-03 08:11:36,089 (trainer:737) INFO: 23epoch:train:6501-6600batch: iter_time=8.831e-05, forward_time=0.289, loss_ctc=44.734, loss_att=46.825, acc=0.743, loss=46.198, backward_time=0.403, grad_norm=34.953, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.111e-04, train_time=1.366
+[gpub058:0/16] 2024-02-03 08:14:24,549 (trainer:737) INFO: 23epoch:train:6601-6700batch: iter_time=3.664e-04, forward_time=0.375, loss_ctc=46.478, loss_att=48.809, acc=0.730, loss=48.110, backward_time=0.420, grad_norm=40.809, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.111e-04, train_time=1.684
+[gpub058:0/16] 2024-02-03 08:16:44,015 (trainer:737) INFO: 23epoch:train:6701-6800batch: iter_time=9.023e-05, forward_time=0.292, loss_ctc=56.880, loss_att=56.872, acc=0.738, loss=56.875, backward_time=0.406, grad_norm=48.068, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.111e-04, train_time=1.395
+[gpub058:0/16] 2024-02-03 08:19:09,689 (trainer:737) INFO: 23epoch:train:6801-6900batch: iter_time=9.810e-05, forward_time=0.290, loss_ctc=48.515, loss_att=46.516, acc=0.749, loss=47.115, backward_time=0.404, grad_norm=33.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.110e-04, train_time=1.457
+[gpub058:0/16] 2024-02-03 08:22:01,086 (trainer:737) INFO: 23epoch:train:6901-7000batch: iter_time=9.444e-05, forward_time=0.354, loss_ctc=49.944, loss_att=47.110, acc=0.747, loss=47.960, backward_time=0.480, grad_norm=37.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.110e-04, train_time=1.713
+[gpub058:0/16] 2024-02-03 08:24:15,174 (trainer:737) INFO: 23epoch:train:7001-7100batch: iter_time=9.158e-05, forward_time=0.291, loss_ctc=52.235, loss_att=43.790, acc=0.764, loss=46.324, backward_time=0.405, grad_norm=40.832, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.110e-04, train_time=1.341
+[gpub058:0/16] 2024-02-03 08:27:00,077 (trainer:737) INFO: 23epoch:train:7101-7200batch: iter_time=9.847e-05, forward_time=0.419, loss_ctc=52.127, loss_att=52.346, acc=0.736, loss=52.280, backward_time=0.423, grad_norm=37.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.109e-04, train_time=1.649
+[gpub058:0/16] 2024-02-03 08:29:33,325 (trainer:737) INFO: 23epoch:train:7201-7300batch: iter_time=9.375e-05, forward_time=0.289, loss_ctc=45.139, loss_att=45.854, acc=0.753, loss=45.639, backward_time=0.403, grad_norm=31.659, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.109e-04, train_time=1.532
+[gpub058:0/16] 2024-02-03 08:32:16,751 (trainer:737) INFO: 23epoch:train:7301-7400batch: iter_time=3.396e-04, forward_time=0.394, loss_ctc=57.735, loss_att=56.133, acc=0.716, loss=56.613, backward_time=0.445, grad_norm=45.415, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.109e-04, train_time=1.634
+[gpub058:0/16] 2024-02-03 08:34:35,099 (trainer:737) INFO: 23epoch:train:7401-7500batch: iter_time=8.331e-05, forward_time=0.289, loss_ctc=44.955, loss_att=50.288, acc=0.721, loss=48.688, backward_time=0.401, grad_norm=39.371, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.109e-04, train_time=1.383
+[gpub058:0/16] 2024-02-03 08:34:55,128 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-03 08:35:14,078 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 08:35:17,655 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d39242b0>)
+[gpub058:0/16] 2024-02-03 08:35:17,655 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-03 08:35:17,658 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 08:42:12,843 (trainer:737) INFO: 23epoch:train:7501-7600batch: iter_time=3.012, forward_time=0.424, loss_ctc=45.417, loss_att=38.268, acc=0.762, loss=40.413, backward_time=0.421, grad_norm=39.848, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.108e-04, train_time=4.577
+[gpub058:0/16] 2024-02-03 08:44:33,355 (trainer:737) INFO: 23epoch:train:7601-7700batch: iter_time=8.293e-05, forward_time=0.288, loss_ctc=44.116, loss_att=47.915, acc=0.734, loss=46.775, backward_time=0.401, grad_norm=33.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.108e-04, train_time=1.405
+[gpub058:0/16] 2024-02-03 08:46:33,922 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 08:47:15,556 (trainer:737) INFO: 23epoch:train:7701-7800batch: iter_time=3.212e-04, forward_time=0.319, loss_ctc=52.228, loss_att=48.645, acc=0.750, loss=49.720, backward_time=0.424, grad_norm=42.865, clip=100.000, loss_scale=4.851e+33, optim_step_time=0.100, optim0_lr0=2.108e-04, train_time=1.622
+[gpub058:0/16] 2024-02-03 08:49:37,690 (trainer:737) INFO: 23epoch:train:7801-7900batch: iter_time=8.386e-05, forward_time=0.288, loss_ctc=42.803, loss_att=47.083, acc=0.739, loss=45.799, backward_time=0.402, grad_norm=34.061, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.107e-04, train_time=1.421
+[gpub058:0/16] 2024-02-03 08:52:01,216 (trainer:737) INFO: 23epoch:train:7901-8000batch: iter_time=9.791e-05, forward_time=0.290, loss_ctc=55.992, loss_att=57.118, acc=0.727, loss=56.780, backward_time=0.404, grad_norm=43.116, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.107e-04, train_time=1.435
+[gpub058:0/16] 2024-02-03 08:54:35,556 (trainer:737) INFO: 23epoch:train:8001-8100batch: iter_time=8.756e-05, forward_time=0.291, loss_ctc=51.431, loss_att=47.426, acc=0.753, loss=48.628, backward_time=0.405, grad_norm=35.343, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.107e-04, train_time=1.543
+[gpub058:0/16] 2024-02-03 08:57:29,256 (trainer:737) INFO: 23epoch:train:8101-8200batch: iter_time=8.970e-05, forward_time=0.397, loss_ctc=49.364, loss_att=50.439, acc=0.738, loss=50.116, backward_time=0.425, grad_norm=34.197, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.106e-04, train_time=1.736
+[gpub058:0/16] 2024-02-03 09:00:04,172 (trainer:737) INFO: 23epoch:train:8201-8300batch: iter_time=8.590e-05, forward_time=0.287, loss_ctc=47.800, loss_att=41.145, acc=0.764, loss=43.142, backward_time=0.399, grad_norm=35.069, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.106e-04, train_time=1.549
+[gpub058:0/16] 2024-02-03 09:02:39,167 (trainer:737) INFO: 23epoch:train:8301-8400batch: iter_time=9.282e-05, forward_time=0.298, loss_ctc=52.418, loss_att=44.165, acc=0.757, loss=46.641, backward_time=0.405, grad_norm=38.836, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.106e-04, train_time=1.550
+[gpub058:0/16] 2024-02-03 09:05:22,678 (trainer:737) INFO: 23epoch:train:8401-8500batch: iter_time=8.872e-05, forward_time=0.377, loss_ctc=48.722, loss_att=52.823, acc=0.749, loss=51.593, backward_time=0.435, grad_norm=34.460, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.105e-04, train_time=1.635
+[gpub058:0/16] 2024-02-03 09:07:39,636 (trainer:737) INFO: 23epoch:train:8501-8600batch: iter_time=8.577e-05, forward_time=0.288, loss_ctc=47.146, loss_att=46.618, acc=0.739, loss=46.777, backward_time=0.401, grad_norm=35.309, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.105e-04, train_time=1.369
+[gpub058:0/16] 2024-02-03 09:10:25,362 (trainer:737) INFO: 23epoch:train:8601-8700batch: iter_time=8.359e-05, forward_time=0.292, loss_ctc=58.526, loss_att=62.375, acc=0.704, loss=61.220, backward_time=0.406, grad_norm=44.867, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.105e-04, train_time=1.657
+[gpub058:0/16] 2024-02-03 09:12:07,716 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-03 09:12:26,863 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 09:12:30,501 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d27da2c0>)
+[gpub058:0/16] 2024-02-03 09:12:30,502 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-03 09:12:30,505 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 09:18:18,224 (trainer:737) INFO: 23epoch:train:8701-8800batch: iter_time=3.180, forward_time=0.368, loss_ctc=42.549, loss_att=37.751, acc=0.765, loss=39.191, backward_time=0.466, grad_norm=35.566, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=2.104e-04, train_time=4.728
+[gpub058:0/16] 2024-02-03 09:20:43,182 (trainer:737) INFO: 23epoch:train:8801-8900batch: iter_time=7.316e-05, forward_time=0.288, loss_ctc=44.568, loss_att=43.168, acc=0.754, loss=43.588, backward_time=0.401, grad_norm=36.856, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.104e-04, train_time=1.449
+[gpub058:0/16] 2024-02-03 09:23:36,758 (trainer:737) INFO: 23epoch:train:8901-9000batch: iter_time=8.170e-05, forward_time=0.390, loss_ctc=52.024, loss_att=53.554, acc=0.747, loss=53.095, backward_time=0.453, grad_norm=43.161, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.104e-04, train_time=1.736
+[gpub058:0/16] 2024-02-03 09:25:56,400 (trainer:737) INFO: 23epoch:train:9001-9100batch: iter_time=8.444e-05, forward_time=0.290, loss_ctc=44.873, loss_att=45.483, acc=0.756, loss=45.300, backward_time=0.403, grad_norm=35.606, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.104e-04, train_time=1.396
+[gpub058:0/16] 2024-02-03 09:28:28,452 (trainer:737) INFO: 23epoch:train:9101-9200batch: iter_time=8.976e-05, forward_time=0.346, loss_ctc=46.172, loss_att=51.522, acc=0.728, loss=49.917, backward_time=0.420, grad_norm=37.349, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.103e-04, train_time=1.520
+[gpub058:0/16] 2024-02-03 09:31:02,885 (trainer:737) INFO: 23epoch:train:9201-9300batch: iter_time=8.624e-05, forward_time=0.353, loss_ctc=57.149, loss_att=54.577, acc=0.757, loss=55.348, backward_time=0.419, grad_norm=48.652, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.103e-04, train_time=1.544
+[gpub058:0/16] 2024-02-03 09:33:35,067 (trainer:737) INFO: 23epoch:train:9301-9400batch: iter_time=8.637e-05, forward_time=0.290, loss_ctc=47.913, loss_att=47.300, acc=0.759, loss=47.484, backward_time=0.401, grad_norm=33.729, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.103e-04, train_time=1.521
+[gpub058:0/16] 2024-02-03 09:36:21,224 (trainer:737) INFO: 23epoch:train:9401-9500batch: iter_time=8.027e-05, forward_time=0.355, loss_ctc=49.887, loss_att=46.598, acc=0.755, loss=47.585, backward_time=0.420, grad_norm=35.244, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.102e-04, train_time=1.661
+[gpub058:0/16] 2024-02-03 09:38:43,584 (trainer:737) INFO: 23epoch:train:9501-9600batch: iter_time=3.822e-04, forward_time=0.334, loss_ctc=51.946, loss_att=44.156, acc=0.762, loss=46.493, backward_time=0.422, grad_norm=38.506, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.102e-04, train_time=1.423
+[gpub058:0/16] 2024-02-03 09:41:38,010 (trainer:737) INFO: 23epoch:train:9601-9700batch: iter_time=7.941e-05, forward_time=0.292, loss_ctc=51.311, loss_att=54.192, acc=0.743, loss=53.328, backward_time=0.405, grad_norm=38.033, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.102e-04, train_time=1.744
+[gpub058:0/16] 2024-02-03 09:44:11,050 (trainer:737) INFO: 23epoch:train:9701-9800batch: iter_time=4.824e-04, forward_time=0.410, loss_ctc=44.541, loss_att=45.440, acc=0.759, loss=45.170, backward_time=0.451, grad_norm=30.853, clip=100.000, loss_scale=2.934e+33, optim_step_time=0.096, optim0_lr0=2.101e-04, train_time=1.531
+[gpub058:0/16] 2024-02-03 09:46:36,083 (trainer:737) INFO: 23epoch:train:9801-9900batch: iter_time=7.666e-05, forward_time=0.290, loss_ctc=56.622, loss_att=56.446, acc=0.719, loss=56.499, backward_time=0.404, grad_norm=44.563, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.101e-04, train_time=1.450
+[gpub058:0/16] 2024-02-03 09:49:22,348 (trainer:737) INFO: 23epoch:train:9901-10000batch: iter_time=7.812e-05, forward_time=0.289, loss_ctc=44.462, loss_att=50.984, acc=0.730, loss=49.028, backward_time=0.401, grad_norm=39.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.101e-04, train_time=1.663
+[gpub058:0/16] 2024-02-03 09:49:42,377 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-03 09:50:01,362 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 09:50:04,969 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d27db160>)
+[gpub058:0/16] 2024-02-03 09:50:04,969 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-03 09:50:04,972 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 09:57:17,786 (trainer:737) INFO: 23epoch:train:10001-10100batch: iter_time=3.153, forward_time=0.408, loss_ctc=45.568, loss_att=38.862, acc=0.763, loss=40.874, backward_time=0.426, grad_norm=36.967, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.100e-04, train_time=4.754
+[gpub058:0/16] 2024-02-03 09:59:54,603 (trainer:737) INFO: 23epoch:train:10101-10200batch: iter_time=8.011e-05, forward_time=0.287, loss_ctc=44.147, loss_att=48.039, acc=0.736, loss=46.871, backward_time=0.399, grad_norm=34.265, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.100e-04, train_time=1.567
+[gpub058:0/16] 2024-02-03 10:02:25,490 (trainer:737) INFO: 23epoch:train:10201-10300batch: iter_time=8.078e-05, forward_time=0.295, loss_ctc=51.456, loss_att=46.709, acc=0.753, loss=48.133, backward_time=0.403, grad_norm=42.803, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.100e-04, train_time=1.510
+[gpub058:0/16] 2024-02-03 10:05:02,051 (trainer:737) INFO: 23epoch:train:10301-10400batch: iter_time=8.244e-05, forward_time=0.358, loss_ctc=42.271, loss_att=48.269, acc=0.735, loss=46.469, backward_time=0.447, grad_norm=36.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.100e-04, train_time=1.565
+[gpub058:0/16] 2024-02-03 10:07:36,249 (trainer:737) INFO: 23epoch:train:10401-10500batch: iter_time=8.111e-05, forward_time=0.290, loss_ctc=55.145, loss_att=57.538, acc=0.727, loss=56.820, backward_time=0.403, grad_norm=62.494, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.099e-04, train_time=1.541
+[gpub058:0/16] 2024-02-03 10:10:18,653 (trainer:737) INFO: 23epoch:train:10501-10600batch: iter_time=8.538e-05, forward_time=0.301, loss_ctc=51.409, loss_att=47.329, acc=0.752, loss=48.553, backward_time=0.415, grad_norm=34.064, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.099e-04, train_time=1.623
+[gpub058:0/16] 2024-02-03 10:12:47,095 (trainer:737) INFO: 23epoch:train:10601-10700batch: iter_time=7.756e-04, forward_time=0.392, loss_ctc=49.673, loss_att=50.097, acc=0.741, loss=49.970, backward_time=0.418, grad_norm=36.125, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.099e-04, train_time=1.486
+[gpub058:0/16] 2024-02-03 10:15:32,174 (trainer:737) INFO: 23epoch:train:10701-10800batch: iter_time=8.213e-05, forward_time=0.288, loss_ctc=47.416, loss_att=41.274, acc=0.764, loss=43.117, backward_time=0.399, grad_norm=34.363, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.098e-04, train_time=1.650
+[gpub058:0/16] 2024-02-03 10:17:51,955 (trainer:737) INFO: 23epoch:train:10801-10900batch: iter_time=8.310e-05, forward_time=0.291, loss_ctc=52.572, loss_att=44.135, acc=0.758, loss=46.666, backward_time=0.409, grad_norm=40.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.098e-04, train_time=1.398
+[gpub058:0/16] 2024-02-03 10:20:38,653 (trainer:737) INFO: 23epoch:train:10901-11000batch: iter_time=4.566e-04, forward_time=0.366, loss_ctc=48.690, loss_att=53.107, acc=0.749, loss=51.782, backward_time=0.458, grad_norm=34.499, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.098e-04, train_time=1.667
+[gpub058:0/16] 2024-02-03 10:23:12,935 (trainer:737) INFO: 23epoch:train:11001-11100batch: iter_time=8.714e-05, forward_time=0.288, loss_ctc=46.975, loss_att=46.264, acc=0.741, loss=46.478, backward_time=0.403, grad_norm=36.505, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.097e-04, train_time=1.543
+[gpub058:0/16] 2024-02-03 10:25:54,383 (trainer:737) INFO: 23epoch:train:11101-11200batch: iter_time=8.212e-05, forward_time=0.310, loss_ctc=58.572, loss_att=62.046, acc=0.708, loss=61.004, backward_time=0.405, grad_norm=47.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.097e-04, train_time=1.614
+[gpub058:0/16] 2024-02-03 10:27:23,107 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-03 10:27:42,187 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 10:27:45,735 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9dd383e50>)
+[gpub058:0/16] 2024-02-03 10:27:45,735 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-03 10:27:45,738 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 10:33:23,084 (trainer:737) INFO: 23epoch:train:11201-11300batch: iter_time=3.056, forward_time=0.374, loss_ctc=42.346, loss_att=36.715, acc=0.763, loss=38.404, backward_time=0.425, grad_norm=36.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.097e-04, train_time=4.487
+[gpub058:0/16] 2024-02-03 10:35:46,878 (trainer:737) INFO: 23epoch:train:11301-11400batch: iter_time=8.346e-05, forward_time=0.290, loss_ctc=44.158, loss_att=41.855, acc=0.742, loss=42.546, backward_time=0.399, grad_norm=36.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.096e-04, train_time=1.438
+[gpub058:0/16] 2024-02-03 10:38:33,758 (trainer:737) INFO: 23epoch:train:11401-11500batch: iter_time=8.306e-05, forward_time=0.309, loss_ctc=52.388, loss_att=50.375, acc=0.748, loss=50.979, backward_time=0.428, grad_norm=41.941, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.096e-04, train_time=1.668
+[gpub058:0/16] 2024-02-03 10:40:51,068 (trainer:737) INFO: 23epoch:train:11501-11600batch: iter_time=6.465e-04, forward_time=0.334, loss_ctc=44.619, loss_att=45.560, acc=0.748, loss=45.277, backward_time=0.432, grad_norm=33.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.096e-04, train_time=1.373
+[gpub058:0/16] 2024-02-03 10:43:33,348 (trainer:737) INFO: 23epoch:train:11601-11700batch: iter_time=8.422e-05, forward_time=0.288, loss_ctc=45.173, loss_att=48.822, acc=0.729, loss=47.727, backward_time=0.401, grad_norm=37.036, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.096e-04, train_time=1.622
+[gpub058:0/16] 2024-02-03 10:46:04,201 (trainer:737) INFO: 23epoch:train:11701-11800batch: iter_time=8.409e-05, forward_time=0.328, loss_ctc=56.321, loss_att=54.655, acc=0.742, loss=55.155, backward_time=0.422, grad_norm=42.758, clip=100.000, loss_scale=5.867e+33, optim_step_time=0.095, optim0_lr0=2.095e-04, train_time=1.508
+[gpub058:0/16] 2024-02-03 10:48:38,884 (trainer:737) INFO: 23epoch:train:11801-11900batch: iter_time=2.896e-04, forward_time=0.350, loss_ctc=47.887, loss_att=45.556, acc=0.752, loss=46.256, backward_time=0.417, grad_norm=32.197, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.095e-04, train_time=1.547
+[gpub058:0/16] 2024-02-03 10:51:07,103 (trainer:737) INFO: 23epoch:train:11901-12000batch: iter_time=8.832e-05, forward_time=0.289, loss_ctc=49.604, loss_att=46.414, acc=0.752, loss=47.371, backward_time=0.402, grad_norm=35.506, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.095e-04, train_time=1.482
+[gpub058:0/16] 2024-02-03 10:53:55,876 (trainer:737) INFO: 23epoch:train:12001-12100batch: iter_time=8.670e-05, forward_time=0.337, loss_ctc=51.387, loss_att=43.974, acc=0.766, loss=46.198, backward_time=0.414, grad_norm=36.073, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.094e-04, train_time=1.688
+[gpub058:0/16] 2024-02-03 10:56:48,302 (trainer:737) INFO: 23epoch:train:12101-12200batch: iter_time=6.227e-04, forward_time=0.362, loss_ctc=51.948, loss_att=52.246, acc=0.739, loss=52.156, backward_time=0.423, grad_norm=39.474, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.094e-04, train_time=1.723
+[gpub058:0/16] 2024-02-03 10:59:05,607 (trainer:737) INFO: 23epoch:train:12201-12300batch: iter_time=8.387e-05, forward_time=0.290, loss_ctc=44.691, loss_att=45.443, acc=0.756, loss=45.218, backward_time=0.403, grad_norm=33.582, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.094e-04, train_time=1.374
+[gpub058:0/16] 2024-02-03 11:01:52,314 (trainer:737) INFO: 23epoch:train:12301-12400batch: iter_time=9.694e-05, forward_time=0.341, loss_ctc=56.691, loss_att=55.138, acc=0.718, loss=55.604, backward_time=0.422, grad_norm=44.931, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.093e-04, train_time=1.667
+[gpub058:0/16] 2024-02-03 11:04:27,903 (trainer:737) INFO: 23epoch:train:12401-12500batch: iter_time=2.313e-04, forward_time=0.355, loss_ctc=44.426, loss_att=49.678, acc=0.725, loss=48.102, backward_time=0.433, grad_norm=37.754, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.093e-04, train_time=1.555
+[gpub058:0/16] 2024-02-03 11:04:47,931 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-03 11:05:06,758 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 11:05:10,333 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe99c4c5c90>)
+[gpub058:0/16] 2024-02-03 11:05:10,333 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-03 11:05:10,336 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 11:12:34,049 (trainer:737) INFO: 23epoch:train:12501-12600batch: iter_time=3.259, forward_time=0.318, loss_ctc=45.326, loss_att=38.783, acc=0.773, loss=40.746, backward_time=0.402, grad_norm=35.838, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.093e-04, train_time=4.861
+[gpub058:0/16] 2024-02-03 11:15:15,888 (trainer:737) INFO: 23epoch:train:12601-12700batch: iter_time=7.736e-05, forward_time=0.357, loss_ctc=43.673, loss_att=50.354, acc=0.740, loss=48.349, backward_time=0.416, grad_norm=33.882, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.092e-04, train_time=1.618
+[gpub058:0/16] 2024-02-03 11:17:36,737 (trainer:737) INFO: 23epoch:train:12701-12800batch: iter_time=7.794e-05, forward_time=0.290, loss_ctc=51.037, loss_att=46.430, acc=0.764, loss=47.812, backward_time=0.405, grad_norm=41.213, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.092e-04, train_time=1.406
+[gpub058:0/16] 2024-02-03 11:20:22,577 (trainer:737) INFO: 23epoch:train:12801-12900batch: iter_time=8.277e-05, forward_time=0.290, loss_ctc=42.590, loss_att=47.885, acc=0.738, loss=46.297, backward_time=0.418, grad_norm=36.544, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.092e-04, train_time=1.660
+[gpub058:0/16] 2024-02-03 11:23:11,746 (trainer:737) INFO: 23epoch:train:12901-13000batch: iter_time=8.914e-05, forward_time=0.363, loss_ctc=55.164, loss_att=58.613, acc=0.737, loss=57.578, backward_time=0.426, grad_norm=46.901, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.092e-04, train_time=1.691
+[gpub058:0/16] 2024-02-03 11:25:33,276 (trainer:737) INFO: 23epoch:train:13001-13100batch: iter_time=7.977e-05, forward_time=0.292, loss_ctc=51.294, loss_att=47.229, acc=0.768, loss=48.448, backward_time=0.406, grad_norm=34.267, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.091e-04, train_time=1.415
+[gpub058:0/16] 2024-02-03 11:28:18,396 (trainer:737) INFO: 23epoch:train:13101-13200batch: iter_time=8.154e-05, forward_time=0.307, loss_ctc=49.462, loss_att=51.969, acc=0.745, loss=51.217, backward_time=0.414, grad_norm=34.307, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.091e-04, train_time=1.650
+[gpub058:0/16] 2024-02-03 11:30:39,509 (trainer:737) INFO: 23epoch:train:13201-13300batch: iter_time=7.805e-05, forward_time=0.373, loss_ctc=47.383, loss_att=41.217, acc=0.767, loss=43.066, backward_time=0.423, grad_norm=35.651, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.091e-04, train_time=1.412
+[gpub058:0/16] 2024-02-03 11:33:19,472 (trainer:737) INFO: 23epoch:train:13301-13400batch: iter_time=8.542e-05, forward_time=0.290, loss_ctc=51.638, loss_att=46.363, acc=0.754, loss=47.946, backward_time=0.403, grad_norm=40.452, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.090e-04, train_time=1.599
+[gpub058:0/16] 2024-02-03 11:35:59,003 (trainer:737) INFO: 23epoch:train:13401-13500batch: iter_time=8.183e-05, forward_time=0.299, loss_ctc=48.436, loss_att=52.703, acc=0.755, loss=51.423, backward_time=0.430, grad_norm=33.872, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.090e-04, train_time=1.594
+[gpub058:0/16] 2024-02-03 11:38:50,058 (trainer:737) INFO: 23epoch:train:13501-13600batch: iter_time=9.657e-05, forward_time=0.363, loss_ctc=46.525, loss_att=46.259, acc=0.744, loss=46.339, backward_time=0.433, grad_norm=34.992, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.090e-04, train_time=1.711
+[gpub058:0/16] 2024-02-03 11:41:22,234 (trainer:737) INFO: 23epoch:train:13601-13700batch: iter_time=8.477e-05, forward_time=0.293, loss_ctc=58.241, loss_att=64.049, acc=0.708, loss=62.307, backward_time=0.407, grad_norm=47.642, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.089e-04, train_time=1.521
+[gpub058:0/16] 2024-02-03 11:43:21,883 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-03 11:43:41,327 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 11:43:44,922 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe984e38880>)
+[gpub058:0/16] 2024-02-03 11:43:44,922 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-03 11:43:44,925 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 11:50:09,900 (trainer:737) INFO: 23epoch:train:13701-13800batch: iter_time=3.315, forward_time=0.420, loss_ctc=42.088, loss_att=36.631, acc=0.765, loss=38.268, backward_time=0.425, grad_norm=36.108, clip=100.000, loss_scale=1.173e+34, optim_step_time=0.097, optim0_lr0=2.089e-04, train_time=5.277
+[gpub058:0/16] 2024-02-03 11:51:01,477 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 11:52:27,473 (trainer:737) INFO: 23epoch:train:13801-13900batch: iter_time=8.364e-05, forward_time=0.287, loss_ctc=43.830, loss_att=43.220, acc=0.738, loss=43.403, backward_time=0.399, grad_norm=37.771, clip=100.000, loss_scale=1.416e+34, optim_step_time=0.092, optim0_lr0=2.089e-04, train_time=1.376
+[gpub058:0/16] 2024-02-03 11:55:14,769 (trainer:737) INFO: 23epoch:train:13901-14000batch: iter_time=8.066e-05, forward_time=0.395, loss_ctc=53.037, loss_att=51.184, acc=0.749, loss=51.740, backward_time=0.438, grad_norm=41.141, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.088e-04, train_time=1.673
+[gpub058:0/16] 2024-02-03 11:57:16,265 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 11:57:52,156 (trainer:737) INFO: 23epoch:train:14001-14100batch: iter_time=8.057e-05, forward_time=0.310, loss_ctc=44.564, loss_att=45.975, acc=0.747, loss=45.552, backward_time=0.401, grad_norm=35.049, clip=100.000, loss_scale=9.021e+33, optim_step_time=0.092, optim0_lr0=2.088e-04, train_time=1.574
+[gpub058:0/16] 2024-02-03 12:00:24,886 (trainer:737) INFO: 23epoch:train:14101-14200batch: iter_time=8.338e-05, forward_time=0.289, loss_ctc=45.513, loss_att=48.550, acc=0.733, loss=47.639, backward_time=0.401, grad_norm=38.050, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.088e-04, train_time=1.527
+[gpub058:0/16] 2024-02-03 12:02:59,241 (trainer:737) INFO: 23epoch:train:14201-14300batch: iter_time=9.038e-05, forward_time=0.398, loss_ctc=55.457, loss_att=54.539, acc=0.741, loss=54.814, backward_time=0.434, grad_norm=45.526, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.088e-04, train_time=1.543
+[gpub058:0/16] 2024-02-03 12:05:28,359 (trainer:737) INFO: 23epoch:train:14301-14400batch: iter_time=8.927e-05, forward_time=0.289, loss_ctc=47.540, loss_att=45.726, acc=0.752, loss=46.270, backward_time=0.402, grad_norm=33.313, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.087e-04, train_time=1.491
+[gpub058:0/16] 2024-02-03 12:07:56,782 (trainer:737) INFO: 23epoch:train:14401-14500batch: iter_time=8.405e-05, forward_time=0.290, loss_ctc=50.001, loss_att=46.576, acc=0.750, loss=47.604, backward_time=0.404, grad_norm=36.670, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.087e-04, train_time=1.484
+[gpub058:0/16] 2024-02-03 12:10:35,246 (trainer:737) INFO: 23epoch:train:14501-14600batch: iter_time=8.240e-05, forward_time=0.381, loss_ctc=51.641, loss_att=43.497, acc=0.767, loss=45.941, backward_time=0.450, grad_norm=38.766, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.087e-04, train_time=1.584
+[gpub058:0/16] 2024-02-03 12:13:03,873 (trainer:737) INFO: 23epoch:train:14601-14700batch: iter_time=8.236e-05, forward_time=0.291, loss_ctc=51.524, loss_att=51.898, acc=0.739, loss=51.786, backward_time=0.405, grad_norm=38.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.086e-04, train_time=1.486
+[gpub058:0/16] 2024-02-03 12:15:42,677 (trainer:737) INFO: 23epoch:train:14701-14800batch: iter_time=2.236e-04, forward_time=0.290, loss_ctc=44.382, loss_att=45.147, acc=0.755, loss=44.917, backward_time=0.404, grad_norm=32.411, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.086e-04, train_time=1.588
+[gpub058:0/16] 2024-02-03 12:18:13,143 (trainer:737) INFO: 23epoch:train:14801-14900batch: iter_time=8.564e-05, forward_time=0.376, loss_ctc=56.018, loss_att=54.883, acc=0.721, loss=55.223, backward_time=0.452, grad_norm=44.005, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.086e-04, train_time=1.504
+[gpub058:0/16] 2024-02-03 12:20:38,165 (trainer:737) INFO: 23epoch:train:14901-15000batch: iter_time=8.003e-05, forward_time=0.289, loss_ctc=43.824, loss_att=49.462, acc=0.726, loss=47.771, backward_time=0.402, grad_norm=37.833, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.085e-04, train_time=1.450
+[gpub058:0/16] 2024-02-03 12:58:54,819 (trainer:343) INFO: 23epoch results: [train] iter_time=0.246, forward_time=0.329, loss_ctc=49.724, loss_att=48.692, acc=0.745, loss=49.002, backward_time=0.419, grad_norm=38.258, clip=100.000, loss_scale=6.231e+33, optim_step_time=0.095, optim0_lr0=2.108e-04, train_time=1.788, time=7 hours, 27 minutes and 32.77 seconds, total_count=375000, gpu_max_cached_mem_GB=43.281, [valid] loss_ctc=42.465, cer_ctc=0.207, loss_att=43.033, acc=0.664, cer=0.341, wer=0.993, loss=42.862, time=37 minutes and 52.7 seconds, total_count=116775, gpu_max_cached_mem_GB=43.281
+[gpub058:0/16] 2024-02-03 12:59:04,853 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub058:0/16] 2024-02-03 12:59:04,876 (trainer:272) INFO: 24/45epoch started. Estimated time to finish: 1 week, 12 hours and 46 minutes
+[gpub058:0/16] 2024-02-03 12:59:04,889 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-03 12:59:23,667 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 12:59:27,402 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1901af460>)
+[gpub058:0/16] 2024-02-03 12:59:27,402 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-03 12:59:27,405 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 13:06:15,352 (trainer:737) INFO: 24epoch:train:1-100batch: iter_time=2.845, forward_time=0.352, loss_ctc=50.175, loss_att=50.524, acc=0.738, loss=50.419, backward_time=0.413, grad_norm=37.072, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.085e-04, train_time=4.304
+[gpub058:0/16] 2024-02-03 13:08:50,403 (trainer:737) INFO: 24epoch:train:101-200batch: iter_time=2.293e-04, forward_time=0.355, loss_ctc=47.259, loss_att=43.000, acc=0.753, loss=44.278, backward_time=0.444, grad_norm=35.270, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.085e-04, train_time=1.550
+[gpub058:0/16] 2024-02-03 13:11:33,501 (trainer:737) INFO: 24epoch:train:201-300batch: iter_time=8.119e-05, forward_time=0.352, loss_ctc=51.331, loss_att=50.891, acc=0.735, loss=51.023, backward_time=0.420, grad_norm=39.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.085e-04, train_time=1.631
+[gpub058:0/16] 2024-02-03 13:14:21,069 (trainer:737) INFO: 24epoch:train:301-400batch: iter_time=8.229e-05, forward_time=0.343, loss_ctc=46.483, loss_att=46.575, acc=0.729, loss=46.547, backward_time=0.438, grad_norm=35.024, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.084e-04, train_time=1.675
+[gpub058:0/16] 2024-02-03 13:17:16,240 (trainer:737) INFO: 24epoch:train:401-500batch: iter_time=4.872e-04, forward_time=0.346, loss_ctc=49.660, loss_att=55.072, acc=0.733, loss=53.449, backward_time=0.437, grad_norm=37.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.084e-04, train_time=1.751
+[gpub058:0/16] 2024-02-03 13:19:53,818 (trainer:737) INFO: 24epoch:train:501-600batch: iter_time=8.355e-05, forward_time=0.372, loss_ctc=46.149, loss_att=44.927, acc=0.740, loss=45.293, backward_time=0.471, grad_norm=36.599, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.084e-04, train_time=1.576
+[gpub058:0/16] 2024-02-03 13:22:44,071 (trainer:737) INFO: 24epoch:train:601-700batch: iter_time=8.395e-05, forward_time=0.340, loss_ctc=45.721, loss_att=49.090, acc=0.734, loss=48.079, backward_time=0.426, grad_norm=35.211, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.083e-04, train_time=1.701
+[gpub058:0/16] 2024-02-03 13:25:09,614 (trainer:737) INFO: 24epoch:train:701-800batch: iter_time=2.257e-04, forward_time=0.320, loss_ctc=48.389, loss_att=47.318, acc=0.738, loss=47.639, backward_time=0.413, grad_norm=98.350, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.083e-04, train_time=1.456
+[gpub058:0/16] 2024-02-03 13:28:05,634 (trainer:737) INFO: 24epoch:train:801-900batch: iter_time=4.741e-04, forward_time=0.401, loss_ctc=51.690, loss_att=54.504, acc=0.713, loss=53.660, backward_time=0.437, grad_norm=39.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.083e-04, train_time=1.759
+[gpub058:0/16] 2024-02-03 13:31:04,193 (trainer:737) INFO: 24epoch:train:901-1000batch: iter_time=4.756e-04, forward_time=0.387, loss_ctc=57.463, loss_att=52.060, acc=0.732, loss=53.681, backward_time=0.439, grad_norm=51.901, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.082e-04, train_time=1.787
+[gpub058:0/16] 2024-02-03 13:33:42,963 (trainer:737) INFO: 24epoch:train:1001-1100batch: iter_time=8.326e-05, forward_time=0.366, loss_ctc=50.373, loss_att=52.115, acc=0.744, loss=51.593, backward_time=0.456, grad_norm=36.286, clip=100.000, loss_scale=6.542e+33, optim_step_time=0.100, optim0_lr0=2.082e-04, train_time=1.587
+[gpub058:0/16] 2024-02-03 13:36:32,272 (trainer:737) INFO: 24epoch:train:1101-1200batch: iter_time=4.115e-04, forward_time=0.327, loss_ctc=52.085, loss_att=42.168, acc=0.764, loss=45.143, backward_time=0.413, grad_norm=40.559, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.082e-04, train_time=1.694
+[gpub058:0/16] 2024-02-03 13:38:16,364 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-03 13:38:35,295 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 13:38:38,945 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe187e76b90>)
+[gpub058:0/16] 2024-02-03 13:38:38,945 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-03 13:38:38,948 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 13:44:44,299 (trainer:737) INFO: 24epoch:train:1201-1300batch: iter_time=3.260, forward_time=0.376, loss_ctc=57.261, loss_att=62.858, acc=0.714, loss=61.179, backward_time=0.437, grad_norm=42.682, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.082e-04, train_time=4.920
+[gpub058:0/16] 2024-02-03 13:47:24,705 (trainer:737) INFO: 24epoch:train:1301-1400batch: iter_time=8.574e-05, forward_time=0.333, loss_ctc=46.474, loss_att=42.675, acc=0.753, loss=43.815, backward_time=0.409, grad_norm=38.063, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.081e-04, train_time=1.604
+[gpub058:0/16] 2024-02-03 13:50:07,566 (trainer:737) INFO: 24epoch:train:1401-1500batch: iter_time=8.603e-05, forward_time=0.348, loss_ctc=50.702, loss_att=55.884, acc=0.742, loss=54.329, backward_time=0.438, grad_norm=39.573, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.081e-04, train_time=1.628
+[gpub058:0/16] 2024-02-03 13:52:36,198 (trainer:737) INFO: 24epoch:train:1501-1600batch: iter_time=9.179e-05, forward_time=0.290, loss_ctc=45.503, loss_att=42.781, acc=0.755, loss=43.597, backward_time=0.403, grad_norm=36.199, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.081e-04, train_time=1.486
+[gpub058:0/16] 2024-02-03 13:54:47,270 (trainer:737) INFO: 24epoch:train:1601-1700batch: iter_time=9.484e-05, forward_time=0.291, loss_ctc=48.482, loss_att=53.460, acc=0.729, loss=51.967, backward_time=0.412, grad_norm=34.934, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.080e-04, train_time=1.311
+[gpub058:0/16] 2024-02-03 13:57:30,504 (trainer:737) INFO: 24epoch:train:1701-1800batch: iter_time=9.309e-05, forward_time=0.365, loss_ctc=45.546, loss_att=48.838, acc=0.759, loss=47.850, backward_time=0.458, grad_norm=35.985, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.080e-04, train_time=1.632
+[gpub058:0/16] 2024-02-03 14:00:09,245 (trainer:737) INFO: 24epoch:train:1801-1900batch: iter_time=9.402e-05, forward_time=0.288, loss_ctc=43.166, loss_att=38.133, acc=0.754, loss=39.643, backward_time=0.400, grad_norm=33.451, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.080e-04, train_time=1.587
+[gpub058:0/16] 2024-02-03 14:02:23,503 (trainer:737) INFO: 24epoch:train:1901-2000batch: iter_time=9.437e-05, forward_time=0.302, loss_ctc=50.123, loss_att=59.077, acc=0.723, loss=56.391, backward_time=0.408, grad_norm=36.323, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.079e-04, train_time=1.343
+[gpub058:0/16] 2024-02-03 14:05:17,867 (trainer:737) INFO: 24epoch:train:2001-2100batch: iter_time=9.292e-05, forward_time=0.329, loss_ctc=45.664, loss_att=48.663, acc=0.737, loss=47.763, backward_time=0.437, grad_norm=38.871, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.079e-04, train_time=1.744
+[gpub058:0/16] 2024-02-03 14:05:58,174 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 14:07:34,427 (trainer:737) INFO: 24epoch:train:2101-2200batch: iter_time=9.150e-05, forward_time=0.330, loss_ctc=51.021, loss_att=51.510, acc=0.738, loss=51.363, backward_time=0.408, grad_norm=37.640, clip=100.000, loss_scale=6.661e+33, optim_step_time=0.092, optim0_lr0=2.079e-04, train_time=1.365
+[gpub058:0/16] 2024-02-03 14:09:54,996 (trainer:737) INFO: 24epoch:train:2201-2300batch: iter_time=8.379e-05, forward_time=0.295, loss_ctc=59.173, loss_att=57.892, acc=0.753, loss=58.277, backward_time=0.408, grad_norm=52.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.079e-04, train_time=1.405
+[gpub058:0/16] 2024-02-03 14:12:47,009 (trainer:737) INFO: 24epoch:train:2301-2400batch: iter_time=8.676e-05, forward_time=0.353, loss_ctc=52.187, loss_att=49.489, acc=0.763, loss=50.298, backward_time=0.422, grad_norm=38.302, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.078e-04, train_time=1.720
+[gpub058:0/16] 2024-02-03 14:15:07,047 (trainer:737) INFO: 24epoch:train:2401-2500batch: iter_time=8.537e-05, forward_time=0.302, loss_ctc=53.081, loss_att=53.724, acc=0.721, loss=53.531, backward_time=0.411, grad_norm=39.582, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.078e-04, train_time=1.400
+[gpub058:0/16] 2024-02-03 14:15:27,074 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-03 14:15:46,137 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 14:15:49,656 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9ca61f4f0>)
+[gpub058:0/16] 2024-02-03 14:15:49,657 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-03 14:15:49,660 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 14:23:20,637 (trainer:737) INFO: 24epoch:train:2501-2600batch: iter_time=3.216, forward_time=0.303, loss_ctc=47.587, loss_att=49.463, acc=0.740, loss=48.900, backward_time=0.406, grad_norm=38.770, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.078e-04, train_time=4.935
+[gpub058:0/16] 2024-02-03 14:26:00,165 (trainer:737) INFO: 24epoch:train:2601-2700batch: iter_time=0.045, forward_time=0.372, loss_ctc=46.159, loss_att=42.381, acc=0.759, loss=43.514, backward_time=0.420, grad_norm=35.776, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.077e-04, train_time=1.596
+[gpub058:0/16] 2024-02-03 14:29:44,262 (trainer:737) INFO: 24epoch:train:2701-2800batch: iter_time=0.036, forward_time=0.312, loss_ctc=50.341, loss_att=52.306, acc=0.734, loss=51.717, backward_time=0.457, grad_norm=39.222, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.077e-04, train_time=2.241
+[gpub058:0/16] 2024-02-03 14:32:14,257 (trainer:737) INFO: 24epoch:train:2801-2900batch: iter_time=8.050e-05, forward_time=0.300, loss_ctc=45.951, loss_att=46.445, acc=0.732, loss=46.296, backward_time=0.405, grad_norm=36.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.077e-04, train_time=1.500
+[gpub058:0/16] 2024-02-03 14:34:32,025 (trainer:737) INFO: 24epoch:train:2901-3000batch: iter_time=8.631e-05, forward_time=0.296, loss_ctc=48.234, loss_att=53.824, acc=0.737, loss=52.147, backward_time=0.409, grad_norm=36.370, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.076e-04, train_time=1.363
+[gpub058:0/16] 2024-02-03 14:38:44,483 (trainer:737) INFO: 24epoch:train:3001-3100batch: iter_time=0.006, forward_time=0.383, loss_ctc=44.962, loss_att=43.934, acc=0.746, loss=44.243, backward_time=0.432, grad_norm=34.495, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.076e-04, train_time=2.538
+[gpub058:0/16] 2024-02-03 14:41:20,494 (trainer:737) INFO: 24epoch:train:3101-3200batch: iter_time=8.701e-05, forward_time=0.306, loss_ctc=44.784, loss_att=48.762, acc=0.736, loss=47.569, backward_time=0.406, grad_norm=34.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.076e-04, train_time=1.560
+[gpub058:0/16] 2024-02-03 14:43:34,182 (trainer:737) INFO: 24epoch:train:3201-3300batch: iter_time=8.098e-05, forward_time=0.291, loss_ctc=47.222, loss_att=46.554, acc=0.742, loss=46.755, backward_time=0.418, grad_norm=36.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.076e-04, train_time=1.337
+[gpub058:0/16] 2024-02-03 14:47:28,620 (trainer:737) INFO: 24epoch:train:3301-3400batch: iter_time=0.091, forward_time=0.388, loss_ctc=50.979, loss_att=54.705, acc=0.713, loss=53.587, backward_time=0.431, grad_norm=40.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.075e-04, train_time=2.343
+[gpub058:0/16] 2024-02-03 14:49:56,127 (trainer:737) INFO: 24epoch:train:3401-3500batch: iter_time=8.377e-05, forward_time=0.307, loss_ctc=54.501, loss_att=51.812, acc=0.737, loss=52.618, backward_time=0.408, grad_norm=49.651, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.075e-04, train_time=1.476
+[gpub058:0/16] 2024-02-03 14:52:35,300 (trainer:737) INFO: 24epoch:train:3501-3600batch: iter_time=8.381e-05, forward_time=0.291, loss_ctc=49.440, loss_att=51.478, acc=0.747, loss=50.867, backward_time=0.405, grad_norm=36.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.075e-04, train_time=1.592
+[gpub058:0/16] 2024-02-03 14:54:54,607 (trainer:737) INFO: 24epoch:train:3601-3700batch: iter_time=8.204e-05, forward_time=0.356, loss_ctc=50.651, loss_att=42.628, acc=0.762, loss=45.035, backward_time=0.425, grad_norm=40.204, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.074e-04, train_time=1.393
+[gpub058:0/16] 2024-02-03 14:56:25,836 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-03 14:56:44,585 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 14:56:48,223 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec14e4b790>)
+[gpub058:0/16] 2024-02-03 14:56:48,223 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-03 14:56:48,226 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 15:02:47,187 (trainer:737) INFO: 24epoch:train:3701-3800batch: iter_time=3.255, forward_time=0.325, loss_ctc=55.596, loss_att=61.837, acc=0.717, loss=59.964, backward_time=0.423, grad_norm=39.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.074e-04, train_time=4.725
+[gpub058:0/16] 2024-02-03 15:05:29,084 (trainer:737) INFO: 24epoch:train:3801-3900batch: iter_time=7.637e-05, forward_time=0.294, loss_ctc=45.553, loss_att=42.102, acc=0.757, loss=43.137, backward_time=0.406, grad_norm=36.936, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.074e-04, train_time=1.619
+[gpub058:0/16] 2024-02-03 15:07:54,202 (trainer:737) INFO: 24epoch:train:3901-4000batch: iter_time=7.853e-05, forward_time=0.365, loss_ctc=50.291, loss_att=54.840, acc=0.743, loss=53.475, backward_time=0.423, grad_norm=39.109, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.073e-04, train_time=1.451
+[gpub058:0/16] 2024-02-03 15:10:38,588 (trainer:737) INFO: 24epoch:train:4001-4100batch: iter_time=8.046e-05, forward_time=0.298, loss_ctc=44.587, loss_att=41.997, acc=0.758, loss=42.774, backward_time=0.410, grad_norm=33.212, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.073e-04, train_time=1.644
+[gpub058:0/16] 2024-02-03 15:12:51,237 (trainer:737) INFO: 24epoch:train:4101-4200batch: iter_time=8.190e-05, forward_time=0.308, loss_ctc=47.749, loss_att=52.378, acc=0.734, loss=50.990, backward_time=0.414, grad_norm=34.213, clip=100.000, loss_scale=8.879e+33, optim_step_time=0.092, optim0_lr0=2.073e-04, train_time=1.326
+[gpub058:0/16] 2024-02-03 15:15:38,389 (trainer:737) INFO: 24epoch:train:4201-4300batch: iter_time=8.299e-05, forward_time=0.346, loss_ctc=44.791, loss_att=48.640, acc=0.763, loss=47.485, backward_time=0.419, grad_norm=34.478, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.073e-04, train_time=1.671
+[gpub058:0/16] 2024-02-03 15:17:56,100 (trainer:737) INFO: 24epoch:train:4301-4400batch: iter_time=8.707e-05, forward_time=0.325, loss_ctc=42.888, loss_att=38.008, acc=0.757, loss=39.472, backward_time=0.411, grad_norm=32.218, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.072e-04, train_time=1.377
+[gpub058:0/16] 2024-02-03 15:20:18,278 (trainer:737) INFO: 24epoch:train:4401-4500batch: iter_time=8.514e-05, forward_time=0.292, loss_ctc=49.065, loss_att=58.280, acc=0.724, loss=55.516, backward_time=0.406, grad_norm=36.560, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.072e-04, train_time=1.421
+[gpub058:0/16] 2024-02-03 15:23:08,912 (trainer:737) INFO: 24epoch:train:4501-4600batch: iter_time=8.621e-05, forward_time=0.361, loss_ctc=45.215, loss_att=48.474, acc=0.739, loss=47.496, backward_time=0.428, grad_norm=36.174, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.072e-04, train_time=1.706
+[gpub058:0/16] 2024-02-03 15:25:26,753 (trainer:737) INFO: 24epoch:train:4601-4700batch: iter_time=8.379e-05, forward_time=0.313, loss_ctc=50.250, loss_att=51.259, acc=0.739, loss=50.957, backward_time=0.413, grad_norm=36.980, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.071e-04, train_time=1.378
+[gpub058:0/16] 2024-02-03 15:28:20,692 (trainer:737) INFO: 24epoch:train:4701-4800batch: iter_time=8.484e-05, forward_time=0.407, loss_ctc=55.291, loss_att=53.415, acc=0.755, loss=53.978, backward_time=0.434, grad_norm=44.553, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.071e-04, train_time=1.739
+[gpub058:0/16] 2024-02-03 15:30:45,513 (trainer:737) INFO: 24epoch:train:4801-4900batch: iter_time=8.562e-05, forward_time=0.297, loss_ctc=51.355, loss_att=49.199, acc=0.765, loss=49.845, backward_time=0.407, grad_norm=37.482, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.071e-04, train_time=1.448
+[gpub058:0/16] 2024-02-03 15:33:25,828 (trainer:737) INFO: 24epoch:train:4901-5000batch: iter_time=7.854e-05, forward_time=0.380, loss_ctc=52.553, loss_att=53.140, acc=0.723, loss=52.964, backward_time=0.425, grad_norm=40.339, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.071e-04, train_time=1.603
+[gpub058:0/16] 2024-02-03 15:33:45,856 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-03 15:34:04,968 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 15:34:08,559 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d1c543a0>)
+[gpub058:0/16] 2024-02-03 15:34:08,559 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-03 15:34:08,563 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 15:41:06,856 (trainer:737) INFO: 24epoch:train:5001-5100batch: iter_time=3.142, forward_time=0.319, loss_ctc=47.400, loss_att=47.230, acc=0.757, loss=47.281, backward_time=0.409, grad_norm=35.108, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.070e-04, train_time=4.610
+[gpub058:0/16] 2024-02-03 15:43:56,807 (trainer:737) INFO: 24epoch:train:5101-5200batch: iter_time=8.053e-05, forward_time=0.396, loss_ctc=45.610, loss_att=42.886, acc=0.756, loss=43.703, backward_time=0.421, grad_norm=33.760, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.070e-04, train_time=1.700
+[gpub058:0/16] 2024-02-03 15:44:11,914 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 15:46:22,350 (trainer:737) INFO: 24epoch:train:5201-5300batch: iter_time=8.942e-05, forward_time=0.384, loss_ctc=49.333, loss_att=52.534, acc=0.747, loss=51.574, backward_time=0.421, grad_norm=36.616, clip=100.000, loss_scale=5.717e+33, optim_step_time=0.096, optim0_lr0=2.070e-04, train_time=1.455
+[gpub058:0/16] 2024-02-03 15:49:19,032 (trainer:737) INFO: 24epoch:train:5301-5400batch: iter_time=8.662e-05, forward_time=0.381, loss_ctc=44.905, loss_att=46.824, acc=0.744, loss=46.248, backward_time=0.421, grad_norm=35.013, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.069e-04, train_time=1.767
+[gpub058:0/16] 2024-02-03 15:51:35,970 (trainer:737) INFO: 24epoch:train:5401-5500batch: iter_time=2.314e-04, forward_time=0.306, loss_ctc=47.340, loss_att=52.180, acc=0.751, loss=50.728, backward_time=0.411, grad_norm=33.837, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.069e-04, train_time=1.369
+[gpub058:0/16] 2024-02-03 15:54:06,741 (trainer:737) INFO: 24epoch:train:5501-5600batch: iter_time=7.996e-05, forward_time=0.412, loss_ctc=44.801, loss_att=43.844, acc=0.754, loss=44.131, backward_time=0.444, grad_norm=35.050, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.069e-04, train_time=1.508
+[gpub058:0/16] 2024-02-03 15:56:53,082 (trainer:737) INFO: 24epoch:train:5601-5700batch: iter_time=8.414e-05, forward_time=0.297, loss_ctc=44.682, loss_att=49.225, acc=0.745, loss=47.862, backward_time=0.403, grad_norm=34.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.068e-04, train_time=1.663
+[gpub058:0/16] 2024-02-03 15:59:33,568 (trainer:737) INFO: 24epoch:train:5701-5800batch: iter_time=8.065e-05, forward_time=0.394, loss_ctc=46.768, loss_att=48.832, acc=0.741, loss=48.213, backward_time=0.459, grad_norm=33.568, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.068e-04, train_time=1.605
+[gpub058:0/16] 2024-02-03 16:02:08,191 (trainer:737) INFO: 24epoch:train:5801-5900batch: iter_time=7.800e-05, forward_time=0.346, loss_ctc=49.982, loss_att=55.268, acc=0.724, loss=53.682, backward_time=0.417, grad_norm=40.986, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.068e-04, train_time=1.546
+[gpub058:0/16] 2024-02-03 16:05:05,411 (trainer:737) INFO: 24epoch:train:5901-6000batch: iter_time=8.304e-05, forward_time=0.386, loss_ctc=54.547, loss_att=52.350, acc=0.747, loss=53.009, backward_time=0.422, grad_norm=50.759, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.068e-04, train_time=1.772
+[gpub058:0/16] 2024-02-03 16:07:44,537 (trainer:737) INFO: 24epoch:train:6001-6100batch: iter_time=8.254e-05, forward_time=0.383, loss_ctc=48.893, loss_att=51.889, acc=0.754, loss=50.990, backward_time=0.443, grad_norm=34.096, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.067e-04, train_time=1.591
+[gpub058:0/16] 2024-02-03 16:10:25,893 (trainer:737) INFO: 24epoch:train:6101-6200batch: iter_time=3.960e-04, forward_time=0.308, loss_ctc=49.443, loss_att=42.491, acc=0.770, loss=44.577, backward_time=0.440, grad_norm=39.675, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.067e-04, train_time=1.613
+[gpub058:0/16] 2024-02-03 16:12:03,798 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-03 16:12:22,467 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 16:12:26,090 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec447ab7c0>)
+[gpub058:0/16] 2024-02-03 16:12:26,090 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-03 16:12:26,093 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 16:18:14,228 (trainer:737) INFO: 24epoch:train:6201-6300batch: iter_time=3.136, forward_time=0.374, loss_ctc=55.669, loss_att=59.973, acc=0.728, loss=58.682, backward_time=0.443, grad_norm=39.506, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.067e-04, train_time=4.684
+[gpub058:0/16] 2024-02-03 16:21:23,746 (trainer:737) INFO: 24epoch:train:6301-6400batch: iter_time=8.445e-05, forward_time=0.405, loss_ctc=44.937, loss_att=40.707, acc=0.760, loss=41.976, backward_time=0.423, grad_norm=34.873, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.119, optim0_lr0=2.066e-04, train_time=1.895
+[gpub058:0/16] 2024-02-03 16:23:54,237 (trainer:737) INFO: 24epoch:train:6401-6500batch: iter_time=7.508e-05, forward_time=0.404, loss_ctc=49.909, loss_att=53.572, acc=0.745, loss=52.473, backward_time=0.447, grad_norm=36.013, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.066e-04, train_time=1.505
+[gpub058:0/16] 2024-02-03 16:27:02,992 (trainer:737) INFO: 24epoch:train:6501-6600batch: iter_time=2.485e-04, forward_time=0.372, loss_ctc=44.604, loss_att=42.053, acc=0.757, loss=42.818, backward_time=0.431, grad_norm=34.941, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.066e-04, train_time=1.887
+[gpub058:0/16] 2024-02-03 16:29:52,810 (trainer:737) INFO: 24epoch:train:6601-6700batch: iter_time=8.814e-05, forward_time=0.462, loss_ctc=47.550, loss_att=52.694, acc=0.734, loss=51.151, backward_time=0.441, grad_norm=33.912, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.066e-04, train_time=1.698
+[gpub058:0/16] 2024-02-03 16:33:01,146 (trainer:737) INFO: 24epoch:train:6701-6800batch: iter_time=9.158e-04, forward_time=0.424, loss_ctc=44.871, loss_att=48.458, acc=0.763, loss=47.382, backward_time=0.430, grad_norm=35.066, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=2.065e-04, train_time=1.882
+[gpub058:0/16] 2024-02-03 16:35:50,005 (trainer:737) INFO: 24epoch:train:6801-6900batch: iter_time=8.879e-05, forward_time=0.294, loss_ctc=42.675, loss_att=37.800, acc=0.760, loss=39.262, backward_time=0.404, grad_norm=58.051, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.065e-04, train_time=1.690
+[gpub058:0/16] 2024-02-03 16:38:53,332 (trainer:737) INFO: 24epoch:train:6901-7000batch: iter_time=8.701e-05, forward_time=0.409, loss_ctc=49.692, loss_att=59.382, acc=0.722, loss=56.475, backward_time=0.425, grad_norm=38.211, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.065e-04, train_time=1.832
+[gpub058:0/16] 2024-02-03 16:42:02,245 (trainer:737) INFO: 24epoch:train:7001-7100batch: iter_time=4.026e-04, forward_time=0.444, loss_ctc=44.683, loss_att=48.270, acc=0.737, loss=47.194, backward_time=0.422, grad_norm=37.851, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.064e-04, train_time=1.890
+[gpub058:0/16] 2024-02-03 16:44:15,340 (trainer:737) INFO: 24epoch:train:7101-7200batch: iter_time=8.643e-05, forward_time=0.297, loss_ctc=49.876, loss_att=50.326, acc=0.744, loss=50.191, backward_time=0.405, grad_norm=36.826, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.064e-04, train_time=1.330
+[gpub058:0/16] 2024-02-03 16:47:16,356 (trainer:737) INFO: 24epoch:train:7201-7300batch: iter_time=2.637e-04, forward_time=0.401, loss_ctc=55.453, loss_att=53.706, acc=0.757, loss=54.230, backward_time=0.421, grad_norm=46.917, clip=100.000, loss_scale=9.813e+33, optim_step_time=0.094, optim0_lr0=2.064e-04, train_time=1.811
+[gpub058:0/16] 2024-02-03 16:49:28,570 (trainer:737) INFO: 24epoch:train:7301-7400batch: iter_time=8.402e-05, forward_time=0.293, loss_ctc=50.542, loss_att=48.668, acc=0.768, loss=49.230, backward_time=0.410, grad_norm=38.310, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.063e-04, train_time=1.322
+[gpub058:0/16] 2024-02-03 16:52:05,900 (trainer:737) INFO: 24epoch:train:7401-7500batch: iter_time=2.946e-04, forward_time=0.365, loss_ctc=52.136, loss_att=52.942, acc=0.725, loss=52.700, backward_time=0.449, grad_norm=40.749, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.063e-04, train_time=1.573
+[gpub058:0/16] 2024-02-03 16:52:26,066 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-03 16:52:46,086 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 16:52:49,696 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec37af9ff0>)
+[gpub058:0/16] 2024-02-03 16:52:49,696 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-03 16:52:49,700 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 17:00:02,342 (trainer:737) INFO: 24epoch:train:7501-7600batch: iter_time=3.053, forward_time=0.288, loss_ctc=47.236, loss_att=47.622, acc=0.758, loss=47.506, backward_time=0.403, grad_norm=37.050, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.063e-04, train_time=4.764
+[gpub058:0/16] 2024-02-03 17:02:35,432 (trainer:737) INFO: 24epoch:train:7601-7700batch: iter_time=3.354e-04, forward_time=0.400, loss_ctc=45.265, loss_att=42.701, acc=0.757, loss=43.470, backward_time=0.432, grad_norm=33.917, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.063e-04, train_time=1.531
+[gpub058:0/16] 2024-02-03 17:05:44,076 (trainer:737) INFO: 24epoch:train:7701-7800batch: iter_time=8.423e-05, forward_time=0.293, loss_ctc=49.854, loss_att=51.621, acc=0.750, loss=51.091, backward_time=0.407, grad_norm=37.846, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.062e-04, train_time=1.886
+[gpub058:0/16] 2024-02-03 17:07:54,606 (trainer:737) INFO: 24epoch:train:7801-7900batch: iter_time=8.698e-05, forward_time=0.296, loss_ctc=45.100, loss_att=45.868, acc=0.748, loss=45.638, backward_time=0.403, grad_norm=35.277, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.062e-04, train_time=1.305
+[gpub058:0/16] 2024-02-03 17:10:57,694 (trainer:737) INFO: 24epoch:train:7901-8000batch: iter_time=9.106e-05, forward_time=0.397, loss_ctc=47.129, loss_att=52.191, acc=0.753, loss=50.672, backward_time=0.429, grad_norm=34.125, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.062e-04, train_time=1.830
+[gpub058:0/16] 2024-02-03 17:13:12,065 (trainer:737) INFO: 24epoch:train:8001-8100batch: iter_time=8.813e-05, forward_time=0.294, loss_ctc=44.685, loss_att=43.386, acc=0.758, loss=43.776, backward_time=0.403, grad_norm=34.992, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.061e-04, train_time=1.344
+[gpub058:0/16] 2024-02-03 17:15:56,146 (trainer:737) INFO: 24epoch:train:8101-8200batch: iter_time=2.467e-04, forward_time=0.387, loss_ctc=44.659, loss_att=48.066, acc=0.748, loss=47.044, backward_time=0.446, grad_norm=33.357, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.061e-04, train_time=1.641
+[gpub058:0/16] 2024-02-03 17:18:41,409 (trainer:737) INFO: 24epoch:train:8201-8300batch: iter_time=8.094e-05, forward_time=0.290, loss_ctc=46.583, loss_att=48.421, acc=0.744, loss=47.870, backward_time=0.404, grad_norm=34.707, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.061e-04, train_time=1.652
+[gpub058:0/16] 2024-02-03 17:20:58,495 (trainer:737) INFO: 24epoch:train:8301-8400batch: iter_time=8.239e-05, forward_time=0.329, loss_ctc=49.925, loss_att=55.443, acc=0.726, loss=53.787, backward_time=0.418, grad_norm=39.620, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.061e-04, train_time=1.370
+[gpub058:0/16] 2024-02-03 17:23:57,691 (trainer:737) INFO: 24epoch:train:8401-8500batch: iter_time=3.870e-04, forward_time=0.385, loss_ctc=53.928, loss_att=51.887, acc=0.750, loss=52.500, backward_time=0.429, grad_norm=51.232, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.060e-04, train_time=1.792
+[gpub058:0/16] 2024-02-03 17:24:44,631 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 17:26:32,028 (trainer:737) INFO: 24epoch:train:8501-8600batch: iter_time=8.029e-05, forward_time=0.297, loss_ctc=48.494, loss_att=52.188, acc=0.754, loss=51.080, backward_time=0.407, grad_norm=35.897, clip=100.000, loss_scale=6.084e+33, optim_step_time=0.092, optim0_lr0=2.060e-04, train_time=1.544
+[gpub058:0/16] 2024-02-03 17:29:18,859 (trainer:737) INFO: 24epoch:train:8601-8700batch: iter_time=9.965e-05, forward_time=0.439, loss_ctc=49.236, loss_att=42.463, acc=0.771, loss=44.495, backward_time=0.427, grad_norm=38.497, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.060e-04, train_time=1.668
+[gpub058:0/16] 2024-02-03 17:30:44,834 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-03 17:31:04,022 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 17:31:07,695 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec3afd7e80>)
+[gpub058:0/16] 2024-02-03 17:31:07,696 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-03 17:31:07,699 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 17:37:12,084 (trainer:737) INFO: 24epoch:train:8701-8800batch: iter_time=3.167, forward_time=0.297, loss_ctc=55.351, loss_att=61.790, acc=0.716, loss=59.858, backward_time=0.407, grad_norm=41.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.059e-04, train_time=4.732
+[gpub058:0/16] 2024-02-03 17:39:53,321 (trainer:737) INFO: 24epoch:train:8801-8900batch: iter_time=7.767e-05, forward_time=0.381, loss_ctc=44.589, loss_att=41.926, acc=0.756, loss=42.725, backward_time=0.441, grad_norm=35.557, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.059e-04, train_time=1.611
+[gpub058:0/16] 2024-02-03 17:42:25,091 (trainer:737) INFO: 24epoch:train:8901-9000batch: iter_time=8.109e-05, forward_time=0.298, loss_ctc=49.489, loss_att=53.300, acc=0.742, loss=52.156, backward_time=0.404, grad_norm=38.143, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.059e-04, train_time=1.517
+[gpub058:0/16] 2024-02-03 17:44:55,968 (trainer:737) INFO: 24epoch:train:9001-9100batch: iter_time=8.269e-05, forward_time=0.288, loss_ctc=44.115, loss_att=42.963, acc=0.745, loss=43.309, backward_time=0.401, grad_norm=35.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.058e-04, train_time=1.510
+[gpub058:0/16] 2024-02-03 17:47:41,233 (trainer:737) INFO: 24epoch:train:9101-9200batch: iter_time=9.050e-05, forward_time=0.412, loss_ctc=47.523, loss_att=52.749, acc=0.727, loss=51.181, backward_time=0.423, grad_norm=36.536, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.058e-04, train_time=1.652
+[gpub058:0/16] 2024-02-03 17:50:10,239 (trainer:737) INFO: 24epoch:train:9201-9300batch: iter_time=9.408e-05, forward_time=0.307, loss_ctc=44.281, loss_att=48.520, acc=0.754, loss=47.248, backward_time=0.413, grad_norm=36.133, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.058e-04, train_time=1.489
+[gpub058:0/16] 2024-02-03 17:52:39,552 (trainer:737) INFO: 24epoch:train:9301-9400batch: iter_time=8.587e-05, forward_time=0.296, loss_ctc=42.664, loss_att=38.226, acc=0.751, loss=39.557, backward_time=0.399, grad_norm=34.493, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.058e-04, train_time=1.494
+[gpub058:0/16] 2024-02-03 17:55:28,029 (trainer:737) INFO: 24epoch:train:9401-9500batch: iter_time=8.526e-05, forward_time=0.413, loss_ctc=48.848, loss_att=56.880, acc=0.725, loss=54.470, backward_time=0.428, grad_norm=38.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.057e-04, train_time=1.684
+[gpub058:0/16] 2024-02-03 17:57:39,626 (trainer:737) INFO: 24epoch:train:9501-9600batch: iter_time=8.882e-05, forward_time=0.293, loss_ctc=44.276, loss_att=48.104, acc=0.733, loss=46.955, backward_time=0.403, grad_norm=36.351, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.057e-04, train_time=1.316
+[gpub058:0/16] 2024-02-03 18:00:25,047 (trainer:737) INFO: 24epoch:train:9601-9700batch: iter_time=8.607e-05, forward_time=0.297, loss_ctc=50.062, loss_att=49.045, acc=0.728, loss=49.350, backward_time=0.406, grad_norm=39.295, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.057e-04, train_time=1.654
+[gpub058:0/16] 2024-02-03 18:03:06,267 (trainer:737) INFO: 24epoch:train:9701-9800batch: iter_time=0.002, forward_time=0.401, loss_ctc=55.371, loss_att=52.973, acc=0.752, loss=53.692, backward_time=0.424, grad_norm=47.205, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.056e-04, train_time=1.612
+[gpub058:0/16] 2024-02-03 18:05:16,870 (trainer:737) INFO: 24epoch:train:9801-9900batch: iter_time=8.809e-05, forward_time=0.299, loss_ctc=49.646, loss_att=47.978, acc=0.763, loss=48.478, backward_time=0.405, grad_norm=38.266, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.056e-04, train_time=1.306
+[gpub058:0/16] 2024-02-03 18:08:00,294 (trainer:737) INFO: 24epoch:train:9901-10000batch: iter_time=9.122e-05, forward_time=0.308, loss_ctc=52.019, loss_att=53.119, acc=0.718, loss=52.789, backward_time=0.430, grad_norm=41.042, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.056e-04, train_time=1.634
+[gpub058:0/16] 2024-02-03 18:08:20,549 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-03 18:08:39,661 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 18:08:43,206 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9de843640>)
+[gpub058:0/16] 2024-02-03 18:08:43,206 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-03 18:08:43,209 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 18:15:39,301 (trainer:737) INFO: 24epoch:train:10001-10100batch: iter_time=3.169, forward_time=0.366, loss_ctc=46.983, loss_att=47.896, acc=0.747, loss=47.622, backward_time=0.416, grad_norm=36.291, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.056e-04, train_time=4.590
+[gpub058:0/16] 2024-02-03 18:18:22,290 (trainer:737) INFO: 24epoch:train:10101-10200batch: iter_time=7.882e-05, forward_time=0.297, loss_ctc=45.273, loss_att=41.688, acc=0.762, loss=42.764, backward_time=0.400, grad_norm=34.346, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.055e-04, train_time=1.630
+[gpub058:0/16] 2024-02-03 18:20:51,170 (trainer:737) INFO: 24epoch:train:10201-10300batch: iter_time=9.233e-05, forward_time=0.414, loss_ctc=49.450, loss_att=51.228, acc=0.739, loss=50.695, backward_time=0.442, grad_norm=38.358, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.055e-04, train_time=1.489
+[gpub058:0/16] 2024-02-03 18:23:37,571 (trainer:737) INFO: 24epoch:train:10301-10400batch: iter_time=9.190e-05, forward_time=0.293, loss_ctc=44.499, loss_att=45.467, acc=0.737, loss=45.177, backward_time=0.400, grad_norm=34.694, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.055e-04, train_time=1.664
+[gpub058:0/16] 2024-02-03 18:26:06,139 (trainer:737) INFO: 24epoch:train:10401-10500batch: iter_time=8.510e-05, forward_time=0.295, loss_ctc=47.076, loss_att=53.068, acc=0.741, loss=51.270, backward_time=0.404, grad_norm=36.614, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.054e-04, train_time=1.486
+[gpub058:0/16] 2024-02-03 18:28:51,755 (trainer:737) INFO: 24epoch:train:10501-10600batch: iter_time=8.587e-05, forward_time=0.344, loss_ctc=44.598, loss_att=43.409, acc=0.749, loss=43.766, backward_time=0.468, grad_norm=35.470, clip=100.000, loss_scale=9.450e+33, optim_step_time=0.100, optim0_lr0=2.054e-04, train_time=1.656
+[gpub058:0/16] 2024-02-03 18:31:28,060 (trainer:737) INFO: 24epoch:train:10601-10700batch: iter_time=8.310e-05, forward_time=0.288, loss_ctc=44.064, loss_att=48.367, acc=0.738, loss=47.076, backward_time=0.409, grad_norm=36.131, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.054e-04, train_time=1.563
+[gpub058:0/16] 2024-02-03 18:34:10,844 (trainer:737) INFO: 24epoch:train:10701-10800batch: iter_time=8.938e-05, forward_time=0.290, loss_ctc=46.337, loss_att=46.073, acc=0.746, loss=46.152, backward_time=0.403, grad_norm=32.955, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.054e-04, train_time=1.627
+[gpub058:0/16] 2024-02-03 18:36:46,412 (trainer:737) INFO: 24epoch:train:10801-10900batch: iter_time=3.209e-04, forward_time=0.412, loss_ctc=49.381, loss_att=53.555, acc=0.719, loss=52.303, backward_time=0.447, grad_norm=39.028, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.053e-04, train_time=1.554
+[gpub058:0/16] 2024-02-03 18:39:42,674 (trainer:737) INFO: 24epoch:train:10901-11000batch: iter_time=9.056e-05, forward_time=0.297, loss_ctc=54.214, loss_att=50.100, acc=0.740, loss=51.334, backward_time=0.404, grad_norm=49.457, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.053e-04, train_time=1.764
+[gpub058:0/16] 2024-02-03 18:42:06,257 (trainer:737) INFO: 24epoch:train:11001-11100batch: iter_time=8.998e-05, forward_time=0.298, loss_ctc=48.042, loss_att=50.867, acc=0.750, loss=50.020, backward_time=0.404, grad_norm=34.969, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.053e-04, train_time=1.436
+[gpub058:0/16] 2024-02-03 18:45:07,128 (trainer:737) INFO: 24epoch:train:11101-11200batch: iter_time=8.882e-05, forward_time=0.405, loss_ctc=48.909, loss_att=42.141, acc=0.766, loss=44.171, backward_time=0.427, grad_norm=42.649, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.052e-04, train_time=1.807
+[gpub058:0/16] 2024-02-03 18:46:34,206 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-03 18:46:53,459 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 18:46:57,087 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9bccdf5b0>)
+[gpub058:0/16] 2024-02-03 18:46:57,087 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-03 18:46:57,090 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 18:53:04,004 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 18:53:06,652 (trainer:737) INFO: 24epoch:train:11201-11300batch: iter_time=3.321, forward_time=0.294, loss_ctc=55.348, loss_att=60.903, acc=0.717, loss=59.237, backward_time=0.405, grad_norm=38.556, clip=100.000, loss_scale=1.028e+34, optim_step_time=0.092, optim0_lr0=2.052e-04, train_time=4.796
+[gpub058:0/16] 2024-02-03 18:55:24,527 (trainer:737) INFO: 24epoch:train:11301-11400batch: iter_time=8.317e-05, forward_time=0.324, loss_ctc=44.743, loss_att=41.233, acc=0.756, loss=42.286, backward_time=0.423, grad_norm=36.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.052e-04, train_time=1.378
+[gpub058:0/16] 2024-02-03 18:58:32,215 (trainer:737) INFO: 24epoch:train:11401-11500batch: iter_time=8.455e-05, forward_time=0.332, loss_ctc=49.500, loss_att=50.765, acc=0.749, loss=50.386, backward_time=0.410, grad_norm=37.837, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.052e-04, train_time=1.876
+[gpub058:0/16] 2024-02-03 19:00:42,173 (trainer:737) INFO: 24epoch:train:11501-11600batch: iter_time=9.300e-05, forward_time=0.288, loss_ctc=44.529, loss_att=41.366, acc=0.748, loss=42.315, backward_time=0.401, grad_norm=34.757, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.051e-04, train_time=1.300
+[gpub058:0/16] 2024-02-03 19:03:38,492 (trainer:737) INFO: 24epoch:train:11601-11700batch: iter_time=8.667e-05, forward_time=0.335, loss_ctc=47.115, loss_att=51.446, acc=0.729, loss=50.147, backward_time=0.419, grad_norm=34.679, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.051e-04, train_time=1.763
+[gpub058:0/16] 2024-02-03 19:06:42,930 (trainer:737) INFO: 24epoch:train:11701-11800batch: iter_time=8.763e-05, forward_time=0.357, loss_ctc=44.030, loss_att=48.105, acc=0.754, loss=46.883, backward_time=0.414, grad_norm=35.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.051e-04, train_time=1.843
+[gpub058:0/16] 2024-02-03 19:09:07,084 (trainer:737) INFO: 24epoch:train:11801-11900batch: iter_time=8.394e-05, forward_time=0.286, loss_ctc=42.146, loss_att=37.797, acc=0.752, loss=39.101, backward_time=0.399, grad_norm=33.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.050e-04, train_time=1.442
+[gpub058:0/16] 2024-02-03 19:12:23,668 (trainer:737) INFO: 24epoch:train:11901-12000batch: iter_time=5.321e-04, forward_time=0.387, loss_ctc=48.384, loss_att=55.714, acc=0.728, loss=53.515, backward_time=0.425, grad_norm=36.123, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.050e-04, train_time=1.963
+[gpub058:0/16] 2024-02-03 19:14:56,721 (trainer:737) INFO: 24epoch:train:12001-12100batch: iter_time=8.127e-05, forward_time=0.290, loss_ctc=44.290, loss_att=48.293, acc=0.733, loss=47.092, backward_time=0.406, grad_norm=37.639, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.050e-04, train_time=1.530
+[gpub058:0/16] 2024-02-03 19:17:28,046 (trainer:737) INFO: 24epoch:train:12101-12200batch: iter_time=8.245e-05, forward_time=0.289, loss_ctc=49.849, loss_att=47.216, acc=0.734, loss=48.006, backward_time=0.405, grad_norm=39.244, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.050e-04, train_time=1.514
+[gpub058:0/16] 2024-02-03 19:20:32,061 (trainer:737) INFO: 24epoch:train:12201-12300batch: iter_time=5.529e-04, forward_time=0.348, loss_ctc=54.538, loss_att=53.120, acc=0.753, loss=53.545, backward_time=0.479, grad_norm=48.346, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.049e-04, train_time=1.840
+[gpub058:0/16] 2024-02-03 19:23:04,244 (trainer:737) INFO: 24epoch:train:12301-12400batch: iter_time=8.902e-05, forward_time=0.291, loss_ctc=49.364, loss_att=47.364, acc=0.765, loss=47.964, backward_time=0.407, grad_norm=38.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.049e-04, train_time=1.521
+[gpub058:0/16] 2024-02-03 19:25:37,655 (trainer:737) INFO: 24epoch:train:12401-12500batch: iter_time=8.089e-05, forward_time=0.291, loss_ctc=51.821, loss_att=51.598, acc=0.724, loss=51.665, backward_time=0.407, grad_norm=40.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.049e-04, train_time=1.535
+[gpub058:0/16] 2024-02-03 19:25:57,684 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-03 19:26:17,045 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 19:26:20,652 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe959f87d30>)
+[gpub058:0/16] 2024-02-03 19:26:20,652 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-03 19:26:20,655 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 19:33:53,022 (trainer:737) INFO: 24epoch:train:12501-12600batch: iter_time=3.319, forward_time=0.377, loss_ctc=46.813, loss_att=47.716, acc=0.746, loss=47.445, backward_time=0.417, grad_norm=34.963, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.048e-04, train_time=4.953
+[gpub058:0/16] 2024-02-03 19:36:35,364 (trainer:737) INFO: 24epoch:train:12601-12700batch: iter_time=7.872e-05, forward_time=0.289, loss_ctc=45.281, loss_att=41.755, acc=0.763, loss=42.812, backward_time=0.403, grad_norm=35.145, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.048e-04, train_time=1.623
+[gpub058:0/16] 2024-02-03 19:39:11,528 (trainer:737) INFO: 24epoch:train:12701-12800batch: iter_time=8.095e-05, forward_time=0.329, loss_ctc=49.847, loss_att=50.103, acc=0.744, loss=50.026, backward_time=0.453, grad_norm=38.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.048e-04, train_time=1.562
+[gpub058:0/16] 2024-02-03 19:41:45,164 (trainer:737) INFO: 24epoch:train:12801-12900batch: iter_time=8.290e-05, forward_time=0.291, loss_ctc=44.787, loss_att=45.030, acc=0.739, loss=44.958, backward_time=0.409, grad_norm=35.019, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.048e-04, train_time=1.536
+[gpub058:0/16] 2024-02-03 19:44:27,580 (trainer:737) INFO: 24epoch:train:12901-13000batch: iter_time=8.760e-05, forward_time=0.290, loss_ctc=46.786, loss_att=53.061, acc=0.742, loss=51.179, backward_time=0.402, grad_norm=35.624, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.047e-04, train_time=1.624
+[gpub058:0/16] 2024-02-03 19:47:19,051 (trainer:737) INFO: 24epoch:train:13001-13100batch: iter_time=2.781e-04, forward_time=0.354, loss_ctc=44.114, loss_att=43.651, acc=0.747, loss=43.790, backward_time=0.480, grad_norm=35.624, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.047e-04, train_time=1.715
+[gpub058:0/16] 2024-02-03 19:49:51,840 (trainer:737) INFO: 24epoch:train:13101-13200batch: iter_time=8.872e-05, forward_time=0.290, loss_ctc=44.125, loss_att=48.182, acc=0.741, loss=46.965, backward_time=0.401, grad_norm=34.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.047e-04, train_time=1.527
+[gpub058:0/16] 2024-02-03 19:52:29,628 (trainer:737) INFO: 24epoch:train:13201-13300batch: iter_time=8.455e-05, forward_time=0.289, loss_ctc=46.562, loss_att=46.274, acc=0.746, loss=46.361, backward_time=0.401, grad_norm=33.331, clip=100.000, loss_scale=5.296e+33, optim_step_time=0.092, optim0_lr0=2.046e-04, train_time=1.577
+[gpub058:0/16] 2024-02-03 19:55:17,460 (trainer:737) INFO: 24epoch:train:13301-13400batch: iter_time=6.495e-04, forward_time=0.419, loss_ctc=48.685, loss_att=52.906, acc=0.721, loss=51.640, backward_time=0.424, grad_norm=40.242, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.046e-04, train_time=1.679
+[gpub058:0/16] 2024-02-03 19:57:55,008 (trainer:737) INFO: 24epoch:train:13401-13500batch: iter_time=8.422e-05, forward_time=0.289, loss_ctc=53.082, loss_att=49.837, acc=0.742, loss=50.810, backward_time=0.408, grad_norm=46.790, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.046e-04, train_time=1.575
+[gpub058:0/16] 2024-02-03 20:00:45,911 (trainer:737) INFO: 24epoch:train:13501-13600batch: iter_time=8.489e-05, forward_time=0.292, loss_ctc=47.932, loss_att=50.503, acc=0.751, loss=49.732, backward_time=0.404, grad_norm=36.238, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.046e-04, train_time=1.707
+[gpub058:0/16] 2024-02-03 20:03:28,848 (trainer:737) INFO: 24epoch:train:13601-13700batch: iter_time=8.862e-05, forward_time=0.399, loss_ctc=48.427, loss_att=41.377, acc=0.770, loss=43.492, backward_time=0.441, grad_norm=38.642, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.045e-04, train_time=1.631
+[gpub058:0/16] 2024-02-03 20:05:25,100 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-03 20:05:44,346 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 20:05:47,911 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe97ff06f50>)
+[gpub058:0/16] 2024-02-03 20:05:47,911 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-03 20:05:48,055 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 20:11:56,171 (trainer:737) INFO: 24epoch:train:13701-13800batch: iter_time=3.214, forward_time=0.290, loss_ctc=55.052, loss_att=60.370, acc=0.725, loss=58.774, backward_time=0.403, grad_norm=40.340, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.045e-04, train_time=5.073
+[gpub058:0/16] 2024-02-03 20:15:08,185 (trainer:737) INFO: 24epoch:train:13801-13900batch: iter_time=8.562e-05, forward_time=0.372, loss_ctc=44.474, loss_att=41.762, acc=0.761, loss=42.575, backward_time=0.428, grad_norm=34.972, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.045e-04, train_time=1.920
+[gpub058:0/16] 2024-02-03 20:17:48,999 (trainer:737) INFO: 24epoch:train:13901-14000batch: iter_time=8.108e-05, forward_time=0.296, loss_ctc=49.222, loss_att=55.075, acc=0.746, loss=53.319, backward_time=0.407, grad_norm=39.689, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.044e-04, train_time=1.608
+[gpub058:0/16] 2024-02-03 20:20:34,478 (trainer:737) INFO: 24epoch:train:14001-14100batch: iter_time=8.341e-05, forward_time=0.289, loss_ctc=44.251, loss_att=42.342, acc=0.757, loss=42.915, backward_time=0.401, grad_norm=34.909, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.044e-04, train_time=1.654
+[gpub058:0/16] 2024-02-03 20:23:26,867 (trainer:737) INFO: 24epoch:train:14101-14200batch: iter_time=8.028e-05, forward_time=0.416, loss_ctc=46.736, loss_att=51.995, acc=0.737, loss=50.417, backward_time=0.434, grad_norm=36.554, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.044e-04, train_time=1.724
+[gpub058:0/16] 2024-02-03 20:25:47,653 (trainer:737) INFO: 24epoch:train:14201-14300batch: iter_time=8.773e-05, forward_time=0.291, loss_ctc=44.050, loss_att=48.783, acc=0.763, loss=47.363, backward_time=0.404, grad_norm=35.169, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.044e-04, train_time=1.407
+[gpub058:0/16] 2024-02-03 20:28:53,446 (trainer:737) INFO: 24epoch:train:14301-14400batch: iter_time=9.256e-05, forward_time=0.286, loss_ctc=41.713, loss_att=37.358, acc=0.762, loss=38.665, backward_time=0.398, grad_norm=33.280, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.043e-04, train_time=1.857
+[gpub058:0/16] 2024-02-03 20:31:52,810 (trainer:737) INFO: 24epoch:train:14401-14500batch: iter_time=8.539e-05, forward_time=0.365, loss_ctc=48.004, loss_att=59.148, acc=0.724, loss=55.805, backward_time=0.459, grad_norm=36.480, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.043e-04, train_time=1.794
+[gpub058:0/16] 2024-02-03 20:34:44,159 (trainer:737) INFO: 24epoch:train:14501-14600batch: iter_time=8.340e-05, forward_time=0.288, loss_ctc=43.724, loss_att=47.952, acc=0.741, loss=46.684, backward_time=0.401, grad_norm=35.428, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.043e-04, train_time=1.713
+[gpub058:0/16] 2024-02-03 20:37:56,322 (trainer:737) INFO: 24epoch:train:14601-14700batch: iter_time=8.695e-05, forward_time=0.389, loss_ctc=49.474, loss_att=50.647, acc=0.744, loss=50.295, backward_time=0.440, grad_norm=36.562, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.042e-04, train_time=1.922
+[gpub058:0/16] 2024-02-03 20:40:34,696 (trainer:737) INFO: 24epoch:train:14701-14800batch: iter_time=8.227e-05, forward_time=0.291, loss_ctc=54.558, loss_att=53.131, acc=0.759, loss=53.559, backward_time=0.406, grad_norm=47.286, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.042e-04, train_time=1.584
+[gpub058:0/16] 2024-02-03 20:43:46,423 (trainer:737) INFO: 24epoch:train:14801-14900batch: iter_time=8.172e-05, forward_time=0.390, loss_ctc=48.836, loss_att=48.508, acc=0.770, loss=48.606, backward_time=0.461, grad_norm=39.383, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.042e-04, train_time=1.917
+[gpub058:0/16] 2024-02-03 20:46:38,645 (trainer:737) INFO: 24epoch:train:14901-15000batch: iter_time=8.132e-05, forward_time=0.291, loss_ctc=51.345, loss_att=52.698, acc=0.727, loss=52.292, backward_time=0.407, grad_norm=39.764, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.042e-04, train_time=1.722
+[gpub058:0/16] 2024-02-03 21:24:59,484 (trainer:343) INFO: 24epoch results: [train] iter_time=0.255, forward_time=0.337, loss_ctc=48.334, loss_att=49.224, acc=0.744, loss=48.957, backward_time=0.420, grad_norm=38.298, clip=100.000, loss_scale=7.243e+33, optim_step_time=0.096, optim0_lr0=2.063e-04, train_time=1.870, time=7 hours, 47 minutes and 58.02 seconds, total_count=390000, gpu_max_cached_mem_GB=43.281, [valid] loss_ctc=39.756, cer_ctc=0.202, loss_att=41.411, acc=0.663, cer=0.327, wer=0.995, loss=40.915, time=37 minutes and 56.35 seconds, total_count=121446, gpu_max_cached_mem_GB=43.281
+[gpub058:0/16] 2024-02-03 21:25:10,898 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub058:0/16] 2024-02-03 21:25:10,945 (trainer:272) INFO: 25/45epoch started. Estimated time to finish: 1 week, 6 hours and 4 minutes
+[gpub058:0/16] 2024-02-03 21:25:10,954 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-03 21:25:29,421 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 21:25:33,138 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe185d08100>)
+[gpub058:0/16] 2024-02-03 21:25:33,138 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-03 21:25:33,141 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 21:32:39,656 (trainer:737) INFO: 25epoch:train:1-100batch: iter_time=3.015, forward_time=0.365, loss_ctc=44.779, loss_att=49.013, acc=0.739, loss=47.742, backward_time=0.411, grad_norm=38.287, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.041e-04, train_time=4.487
+[gpub058:0/16] 2024-02-03 21:35:13,938 (trainer:737) INFO: 25epoch:train:101-200batch: iter_time=2.041e-04, forward_time=0.392, loss_ctc=47.097, loss_att=38.440, acc=0.769, loss=41.037, backward_time=0.442, grad_norm=35.373, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.041e-04, train_time=1.542
+[gpub058:0/16] 2024-02-03 21:37:58,967 (trainer:737) INFO: 25epoch:train:201-300batch: iter_time=8.136e-05, forward_time=0.308, loss_ctc=46.894, loss_att=44.822, acc=0.761, loss=45.443, backward_time=0.421, grad_norm=35.860, clip=100.000, loss_scale=1.059e+34, optim_step_time=0.095, optim0_lr0=2.041e-04, train_time=1.650
+[gpub058:0/16] 2024-02-03 21:38:31,862 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 21:40:36,150 (trainer:737) INFO: 25epoch:train:301-400batch: iter_time=7.995e-05, forward_time=0.382, loss_ctc=60.282, loss_att=57.073, acc=0.729, loss=58.036, backward_time=0.428, grad_norm=46.972, clip=100.000, loss_scale=1.248e+34, optim_step_time=0.107, optim0_lr0=2.040e-04, train_time=1.572
+[gpub058:0/16] 2024-02-03 21:43:11,933 (trainer:737) INFO: 25epoch:train:401-500batch: iter_time=8.441e-05, forward_time=0.409, loss_ctc=48.332, loss_att=40.516, acc=0.748, loss=42.861, backward_time=0.458, grad_norm=39.056, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.040e-04, train_time=1.557
+[gpub058:0/16] 2024-02-03 21:45:43,597 (trainer:737) INFO: 25epoch:train:501-600batch: iter_time=1.724e-04, forward_time=0.321, loss_ctc=41.549, loss_att=41.175, acc=0.753, loss=41.287, backward_time=0.408, grad_norm=33.569, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.040e-04, train_time=1.516
+[gpub058:0/16] 2024-02-03 21:48:18,736 (trainer:737) INFO: 25epoch:train:601-700batch: iter_time=8.655e-05, forward_time=0.381, loss_ctc=52.466, loss_att=51.839, acc=0.734, loss=52.027, backward_time=0.437, grad_norm=40.070, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.040e-04, train_time=1.551
+[gpub058:0/16] 2024-02-03 21:51:12,964 (trainer:737) INFO: 25epoch:train:701-800batch: iter_time=8.679e-05, forward_time=0.351, loss_ctc=56.025, loss_att=53.938, acc=0.738, loss=54.564, backward_time=0.429, grad_norm=41.639, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.039e-04, train_time=1.742
+[gpub058:0/16] 2024-02-03 21:53:39,662 (trainer:737) INFO: 25epoch:train:801-900batch: iter_time=2.862e-04, forward_time=0.391, loss_ctc=56.006, loss_att=55.827, acc=0.741, loss=55.881, backward_time=0.443, grad_norm=46.199, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.039e-04, train_time=1.467
+[gpub058:0/16] 2024-02-03 21:56:40,954 (trainer:737) INFO: 25epoch:train:901-1000batch: iter_time=5.603e-04, forward_time=0.378, loss_ctc=51.883, loss_att=47.338, acc=0.751, loss=48.702, backward_time=0.460, grad_norm=41.179, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.039e-04, train_time=1.812
+[gpub058:0/16] 2024-02-03 21:59:25,895 (trainer:737) INFO: 25epoch:train:1001-1100batch: iter_time=8.882e-05, forward_time=0.418, loss_ctc=52.403, loss_att=47.385, acc=0.749, loss=48.890, backward_time=0.460, grad_norm=38.586, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.038e-04, train_time=1.650
+[gpub058:0/16] 2024-02-03 22:02:09,009 (trainer:737) INFO: 25epoch:train:1101-1200batch: iter_time=1.924e-04, forward_time=0.336, loss_ctc=53.728, loss_att=56.763, acc=0.743, loss=55.853, backward_time=0.414, grad_norm=37.852, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.038e-04, train_time=1.631
+[gpub058:0/16] 2024-02-03 22:03:55,040 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-03 22:04:14,039 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 22:04:17,650 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe187b14520>)
+[gpub058:0/16] 2024-02-03 22:04:17,650 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-03 22:04:17,653 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 22:10:41,577 (trainer:737) INFO: 25epoch:train:1201-1300batch: iter_time=3.527, forward_time=0.367, loss_ctc=46.027, loss_att=52.129, acc=0.734, loss=50.299, backward_time=0.432, grad_norm=37.250, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.038e-04, train_time=5.125
+[gpub058:0/16] 2024-02-03 22:13:18,580 (trainer:737) INFO: 25epoch:train:1301-1400batch: iter_time=8.042e-05, forward_time=0.307, loss_ctc=43.533, loss_att=37.971, acc=0.772, loss=39.639, backward_time=0.405, grad_norm=55.251, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.038e-04, train_time=1.570
+[gpub058:0/16] 2024-02-03 22:16:04,889 (trainer:737) INFO: 25epoch:train:1401-1500batch: iter_time=8.676e-05, forward_time=0.413, loss_ctc=47.729, loss_att=45.148, acc=0.759, loss=45.922, backward_time=0.436, grad_norm=36.541, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.037e-04, train_time=1.663
+[gpub058:0/16] 2024-02-03 22:18:31,990 (trainer:737) INFO: 25epoch:train:1501-1600batch: iter_time=8.126e-05, forward_time=0.291, loss_ctc=50.535, loss_att=45.310, acc=0.759, loss=46.878, backward_time=0.406, grad_norm=38.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.037e-04, train_time=1.471
+[gpub058:0/16] 2024-02-03 22:21:11,245 (trainer:737) INFO: 25epoch:train:1601-1700batch: iter_time=8.420e-05, forward_time=0.308, loss_ctc=49.159, loss_att=47.523, acc=0.740, loss=48.014, backward_time=0.410, grad_norm=36.425, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.037e-04, train_time=1.593
+[gpub058:0/16] 2024-02-03 22:22:43,523 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-03 22:23:49,747 (trainer:737) INFO: 25epoch:train:1701-1800batch: iter_time=4.658e-04, forward_time=0.375, loss_ctc=49.400, loss_att=46.980, acc=0.743, loss=47.706, backward_time=0.429, grad_norm=37.616, clip=100.000, loss_scale=8.182e+33, optim_step_time=0.100, optim0_lr0=2.036e-04, train_time=1.585
+[gpub058:0/16] 2024-02-03 22:26:17,933 (trainer:737) INFO: 25epoch:train:1801-1900batch: iter_time=8.465e-05, forward_time=0.311, loss_ctc=51.117, loss_att=47.292, acc=0.742, loss=48.439, backward_time=0.411, grad_norm=40.334, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.036e-04, train_time=1.481
+[gpub058:0/16] 2024-02-03 22:28:59,074 (trainer:737) INFO: 25epoch:train:1901-2000batch: iter_time=8.313e-05, forward_time=0.311, loss_ctc=48.077, loss_att=50.444, acc=0.742, loss=49.734, backward_time=0.406, grad_norm=35.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.036e-04, train_time=1.612
+[gpub058:0/16] 2024-02-03 22:31:38,061 (trainer:737) INFO: 25epoch:train:2001-2100batch: iter_time=8.326e-05, forward_time=0.407, loss_ctc=57.757, loss_att=51.488, acc=0.743, loss=53.369, backward_time=0.445, grad_norm=44.687, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.036e-04, train_time=1.589
+[gpub058:0/16] 2024-02-03 22:34:23,137 (trainer:737) INFO: 25epoch:train:2101-2200batch: iter_time=8.137e-05, forward_time=0.318, loss_ctc=46.553, loss_att=50.620, acc=0.749, loss=49.400, backward_time=0.409, grad_norm=35.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.035e-04, train_time=1.651
+[gpub058:0/16] 2024-02-03 22:36:50,916 (trainer:737) INFO: 25epoch:train:2201-2300batch: iter_time=8.103e-05, forward_time=0.306, loss_ctc=53.501, loss_att=46.196, acc=0.747, loss=48.388, backward_time=0.415, grad_norm=44.445, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.035e-04, train_time=1.477
+[gpub058:0/16] 2024-02-03 22:39:55,175 (trainer:737) INFO: 25epoch:train:2301-2400batch: iter_time=8.085e-05, forward_time=0.393, loss_ctc=55.561, loss_att=55.237, acc=0.754, loss=55.334, backward_time=0.461, grad_norm=37.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.035e-04, train_time=1.843
+[gpub058:0/16] 2024-02-03 22:42:22,001 (trainer:737) INFO: 25epoch:train:2401-2500batch: iter_time=7.909e-05, forward_time=0.304, loss_ctc=47.600, loss_att=49.971, acc=0.743, loss=49.260, backward_time=0.405, grad_norm=35.467, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.035e-04, train_time=1.468
+[gpub058:0/16] 2024-02-03 22:42:42,029 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-03 22:43:01,069 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 22:43:04,712 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c35f3cd0>)
+[gpub058:0/16] 2024-02-03 22:43:04,712 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-03 22:43:04,715 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 22:50:35,883 (trainer:737) INFO: 25epoch:train:2501-2600batch: iter_time=3.302, forward_time=0.402, loss_ctc=44.148, loss_att=48.782, acc=0.742, loss=47.392, backward_time=0.427, grad_norm=39.106, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.034e-04, train_time=4.938
+[gpub058:0/16] 2024-02-03 22:53:21,232 (trainer:737) INFO: 25epoch:train:2601-2700batch: iter_time=8.005e-05, forward_time=0.290, loss_ctc=45.940, loss_att=37.851, acc=0.771, loss=40.278, backward_time=0.402, grad_norm=34.079, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.034e-04, train_time=1.654
+[gpub058:0/16] 2024-02-03 22:56:14,486 (trainer:737) INFO: 25epoch:train:2701-2800batch: iter_time=8.119e-05, forward_time=0.421, loss_ctc=45.641, loss_att=44.120, acc=0.761, loss=44.576, backward_time=0.428, grad_norm=34.640, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.034e-04, train_time=1.732
+[gpub058:0/16] 2024-02-03 22:58:46,050 (trainer:737) INFO: 25epoch:train:2801-2900batch: iter_time=8.117e-05, forward_time=0.398, loss_ctc=56.629, loss_att=56.206, acc=0.733, loss=56.333, backward_time=0.420, grad_norm=40.922, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=2.033e-04, train_time=1.516
+[gpub058:0/16] 2024-02-03 23:01:47,074 (trainer:737) INFO: 25epoch:train:2901-3000batch: iter_time=8.998e-05, forward_time=0.320, loss_ctc=45.218, loss_att=38.816, acc=0.755, loss=40.736, backward_time=0.422, grad_norm=34.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.033e-04, train_time=1.810
+[gpub058:0/16] 2024-02-03 23:04:25,417 (trainer:737) INFO: 25epoch:train:3001-3100batch: iter_time=2.165e-04, forward_time=0.473, loss_ctc=41.266, loss_att=41.366, acc=0.753, loss=41.336, backward_time=0.421, grad_norm=33.889, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.033e-04, train_time=1.584
+[gpub058:0/16] 2024-02-03 23:07:11,678 (trainer:737) INFO: 25epoch:train:3101-3200batch: iter_time=9.226e-05, forward_time=0.426, loss_ctc=50.359, loss_att=50.614, acc=0.739, loss=50.538, backward_time=0.435, grad_norm=40.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.033e-04, train_time=1.662
+[gpub058:0/16] 2024-02-03 23:10:10,172 (trainer:737) INFO: 25epoch:train:3201-3300batch: iter_time=8.469e-05, forward_time=0.291, loss_ctc=53.507, loss_att=52.805, acc=0.746, loss=53.015, backward_time=0.403, grad_norm=38.455, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.032e-04, train_time=1.785
+[gpub058:0/16] 2024-02-03 23:12:40,640 (trainer:737) INFO: 25epoch:train:3301-3400batch: iter_time=8.636e-05, forward_time=0.432, loss_ctc=54.615, loss_att=55.664, acc=0.742, loss=55.350, backward_time=0.428, grad_norm=48.483, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.032e-04, train_time=1.504
+[gpub058:0/16] 2024-02-03 23:15:25,753 (trainer:737) INFO: 25epoch:train:3401-3500batch: iter_time=9.265e-05, forward_time=0.290, loss_ctc=50.346, loss_att=46.536, acc=0.753, loss=47.679, backward_time=0.401, grad_norm=38.825, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.032e-04, train_time=1.652
+[gpub058:0/16] 2024-02-03 23:18:02,665 (trainer:737) INFO: 25epoch:train:3501-3600batch: iter_time=8.805e-05, forward_time=0.294, loss_ctc=50.634, loss_att=45.793, acc=0.756, loss=47.246, backward_time=0.410, grad_norm=37.836, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.031e-04, train_time=1.569
+[gpub058:0/16] 2024-02-03 23:20:57,167 (trainer:737) INFO: 25epoch:train:3601-3700batch: iter_time=0.001, forward_time=0.463, loss_ctc=52.819, loss_att=55.786, acc=0.746, loss=54.896, backward_time=0.427, grad_norm=37.634, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.031e-04, train_time=1.744
+[gpub058:0/16] 2024-02-03 23:22:28,993 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-03 23:22:48,185 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-03 23:22:51,922 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe95f2d2380>)
+[gpub058:0/16] 2024-02-03 23:22:51,922 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-03 23:22:51,926 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-03 23:29:08,842 (trainer:737) INFO: 25epoch:train:3701-3800batch: iter_time=3.249, forward_time=0.296, loss_ctc=45.081, loss_att=51.101, acc=0.738, loss=49.295, backward_time=0.403, grad_norm=37.976, clip=100.000, loss_scale=7.373e+33, optim_step_time=0.092, optim0_lr0=2.031e-04, train_time=4.917
+[gpub058:0/16] 2024-02-03 23:31:36,483 (trainer:737) INFO: 25epoch:train:3801-3900batch: iter_time=8.313e-05, forward_time=0.292, loss_ctc=43.068, loss_att=38.001, acc=0.772, loss=39.521, backward_time=0.411, grad_norm=33.229, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.031e-04, train_time=1.476
+[gpub058:0/16] 2024-02-03 23:34:17,969 (trainer:737) INFO: 25epoch:train:3901-4000batch: iter_time=5.559e-04, forward_time=0.392, loss_ctc=46.819, loss_att=45.037, acc=0.760, loss=45.571, backward_time=0.415, grad_norm=36.330, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.030e-04, train_time=1.614
+[gpub058:0/16] 2024-02-03 23:36:59,738 (trainer:737) INFO: 25epoch:train:4001-4100batch: iter_time=8.345e-05, forward_time=0.290, loss_ctc=48.956, loss_att=45.135, acc=0.760, loss=46.281, backward_time=0.405, grad_norm=34.919, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.030e-04, train_time=1.617
+[gpub058:0/16] 2024-02-03 23:39:34,894 (trainer:737) INFO: 25epoch:train:4101-4200batch: iter_time=8.007e-05, forward_time=0.289, loss_ctc=47.812, loss_att=46.503, acc=0.747, loss=46.895, backward_time=0.402, grad_norm=34.626, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.030e-04, train_time=1.552
+[gpub058:0/16] 2024-02-03 23:42:09,646 (trainer:737) INFO: 25epoch:train:4201-4300batch: iter_time=7.946e-05, forward_time=0.433, loss_ctc=48.173, loss_att=46.242, acc=0.749, loss=46.821, backward_time=0.440, grad_norm=37.249, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.029e-04, train_time=1.547
+[gpub058:0/16] 2024-02-03 23:44:46,537 (trainer:737) INFO: 25epoch:train:4301-4400batch: iter_time=8.729e-05, forward_time=0.291, loss_ctc=49.977, loss_att=46.326, acc=0.746, loss=47.421, backward_time=0.403, grad_norm=37.772, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.029e-04, train_time=1.569
+[gpub058:0/16] 2024-02-03 23:47:13,234 (trainer:737) INFO: 25epoch:train:4401-4500batch: iter_time=8.448e-05, forward_time=0.291, loss_ctc=47.433, loss_att=49.874, acc=0.744, loss=49.142, backward_time=0.406, grad_norm=36.394, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.029e-04, train_time=1.466
+[gpub058:0/16] 2024-02-03 23:49:46,517 (trainer:737) INFO: 25epoch:train:4501-4600batch: iter_time=8.604e-05, forward_time=0.292, loss_ctc=57.529, loss_att=52.281, acc=0.746, loss=53.855, backward_time=0.411, grad_norm=46.724, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.029e-04, train_time=1.534
+[gpub058:0/16] 2024-02-03 23:52:49,678 (trainer:737) INFO: 25epoch:train:4601-4700batch: iter_time=9.784e-05, forward_time=0.433, loss_ctc=46.057, loss_att=50.306, acc=0.750, loss=49.032, backward_time=0.427, grad_norm=36.046, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.028e-04, train_time=1.831
+[gpub058:0/16] 2024-02-03 23:55:21,666 (trainer:737) INFO: 25epoch:train:4701-4800batch: iter_time=8.503e-05, forward_time=0.289, loss_ctc=52.508, loss_att=45.744, acc=0.750, loss=47.773, backward_time=0.402, grad_norm=41.175, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.028e-04, train_time=1.519
+[gpub058:0/16] 2024-02-03 23:57:45,235 (trainer:737) INFO: 25epoch:train:4801-4900batch: iter_time=8.599e-05, forward_time=0.300, loss_ctc=54.924, loss_att=54.285, acc=0.757, loss=54.476, backward_time=0.410, grad_norm=35.596, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.028e-04, train_time=1.436
+[gpub058:0/16] 2024-02-04 00:00:52,031 (trainer:737) INFO: 25epoch:train:4901-5000batch: iter_time=8.174e-05, forward_time=0.371, loss_ctc=47.378, loss_att=49.485, acc=0.749, loss=48.853, backward_time=0.453, grad_norm=36.264, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.028e-04, train_time=1.868
+[gpub058:0/16] 2024-02-04 00:01:12,059 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-04 00:01:31,442 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 00:01:35,088 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec5a25b160>)
+[gpub058:0/16] 2024-02-04 00:01:35,088 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-04 00:01:35,091 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 00:08:40,920 (trainer:737) INFO: 25epoch:train:5001-5100batch: iter_time=3.294, forward_time=0.304, loss_ctc=43.247, loss_att=50.542, acc=0.736, loss=48.353, backward_time=0.407, grad_norm=37.062, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.027e-04, train_time=4.688
+[gpub058:0/16] 2024-02-04 00:11:23,981 (trainer:737) INFO: 25epoch:train:5101-5200batch: iter_time=8.485e-05, forward_time=0.403, loss_ctc=45.204, loss_att=38.848, acc=0.764, loss=40.755, backward_time=0.420, grad_norm=34.790, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.027e-04, train_time=1.631
+[gpub058:0/16] 2024-02-04 00:14:09,468 (trainer:737) INFO: 25epoch:train:5201-5300batch: iter_time=8.556e-05, forward_time=0.289, loss_ctc=45.281, loss_att=43.018, acc=0.761, loss=43.697, backward_time=0.409, grad_norm=33.669, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.027e-04, train_time=1.652
+[gpub058:0/16] 2024-02-04 00:17:11,367 (trainer:737) INFO: 25epoch:train:5301-5400batch: iter_time=9.014e-05, forward_time=0.474, loss_ctc=55.747, loss_att=53.700, acc=0.730, loss=54.314, backward_time=0.433, grad_norm=42.338, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.026e-04, train_time=1.821
+[gpub058:0/16] 2024-02-04 00:19:53,779 (trainer:737) INFO: 25epoch:train:5401-5500batch: iter_time=8.717e-05, forward_time=0.446, loss_ctc=44.481, loss_att=39.981, acc=0.744, loss=41.331, backward_time=0.441, grad_norm=38.311, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.026e-04, train_time=1.624
+[gpub058:0/16] 2024-02-04 00:22:57,522 (trainer:737) INFO: 25epoch:train:5501-5600batch: iter_time=9.114e-05, forward_time=0.288, loss_ctc=40.555, loss_att=42.008, acc=0.746, loss=41.572, backward_time=0.398, grad_norm=35.019, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.026e-04, train_time=1.837
+[gpub058:0/16] 2024-02-04 00:25:35,258 (trainer:737) INFO: 25epoch:train:5601-5700batch: iter_time=8.599e-05, forward_time=0.450, loss_ctc=50.088, loss_att=51.315, acc=0.734, loss=50.947, backward_time=0.450, grad_norm=38.349, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.026e-04, train_time=1.577
+[gpub058:0/16] 2024-02-04 00:28:33,326 (trainer:737) INFO: 25epoch:train:5701-5800batch: iter_time=9.910e-05, forward_time=0.337, loss_ctc=53.424, loss_att=52.511, acc=0.739, loss=52.785, backward_time=0.463, grad_norm=38.581, clip=100.000, loss_scale=1.475e+34, optim_step_time=0.096, optim0_lr0=2.025e-04, train_time=1.780
+[gpub058:0/16] 2024-02-04 00:30:23,519 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 00:30:58,182 (trainer:737) INFO: 25epoch:train:5801-5900batch: iter_time=9.467e-05, forward_time=0.290, loss_ctc=53.208, loss_att=55.206, acc=0.729, loss=54.607, backward_time=0.404, grad_norm=48.406, clip=100.000, loss_scale=1.804e+34, optim_step_time=0.092, optim0_lr0=2.025e-04, train_time=1.448
+[gpub058:0/16] 2024-02-04 00:34:19,848 (trainer:737) INFO: 25epoch:train:5901-6000batch: iter_time=9.350e-05, forward_time=0.425, loss_ctc=50.061, loss_att=46.996, acc=0.743, loss=47.916, backward_time=0.428, grad_norm=39.272, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.025e-04, train_time=2.016
+[gpub058:0/16] 2024-02-04 00:37:07,521 (trainer:737) INFO: 25epoch:train:6001-6100batch: iter_time=8.892e-05, forward_time=0.343, loss_ctc=50.687, loss_att=46.328, acc=0.746, loss=47.636, backward_time=0.474, grad_norm=38.316, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.024e-04, train_time=1.677
+[gpub058:0/16] 2024-02-04 00:39:56,004 (trainer:737) INFO: 25epoch:train:6101-6200batch: iter_time=9.071e-05, forward_time=0.291, loss_ctc=52.737, loss_att=55.782, acc=0.742, loss=54.869, backward_time=0.402, grad_norm=37.869, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.024e-04, train_time=1.685
+[gpub058:0/16] 2024-02-04 00:42:18,670 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-04 00:42:38,030 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 00:42:41,638 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe95f2db610>)
+[gpub058:0/16] 2024-02-04 00:42:41,638 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-04 00:42:41,641 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 00:52:24,382 (trainer:737) INFO: 25epoch:train:6201-6300batch: iter_time=3.341, forward_time=0.401, loss_ctc=44.648, loss_att=51.658, acc=0.732, loss=49.555, backward_time=0.423, grad_norm=35.982, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.024e-04, train_time=7.484
+[gpub058:0/16] 2024-02-04 00:55:13,953 (trainer:737) INFO: 25epoch:train:6301-6400batch: iter_time=8.041e-05, forward_time=0.364, loss_ctc=42.921, loss_att=38.179, acc=0.769, loss=39.602, backward_time=0.451, grad_norm=33.331, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.024e-04, train_time=1.695
+[gpub058:0/16] 2024-02-04 00:57:44,321 (trainer:737) INFO: 25epoch:train:6401-6500batch: iter_time=8.447e-05, forward_time=0.290, loss_ctc=46.946, loss_att=43.616, acc=0.756, loss=44.615, backward_time=0.404, grad_norm=35.665, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.023e-04, train_time=1.503
+[gpub058:0/16] 2024-02-04 01:00:19,699 (trainer:737) INFO: 25epoch:train:6501-6600batch: iter_time=9.264e-05, forward_time=0.426, loss_ctc=48.533, loss_att=45.690, acc=0.754, loss=46.543, backward_time=0.428, grad_norm=36.136, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.023e-04, train_time=1.554
+[gpub058:0/16] 2024-02-04 01:03:06,349 (trainer:737) INFO: 25epoch:train:6601-6700batch: iter_time=8.512e-05, forward_time=0.291, loss_ctc=47.671, loss_att=43.318, acc=0.743, loss=44.624, backward_time=0.402, grad_norm=35.142, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.023e-04, train_time=1.666
+[gpub058:0/16] 2024-02-04 01:05:29,536 (trainer:737) INFO: 25epoch:train:6701-6800batch: iter_time=1.019e-04, forward_time=0.291, loss_ctc=47.113, loss_att=46.771, acc=0.737, loss=46.874, backward_time=0.401, grad_norm=37.753, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.023e-04, train_time=1.432
+[gpub058:0/16] 2024-02-04 01:08:14,770 (trainer:737) INFO: 25epoch:train:6801-6900batch: iter_time=9.136e-05, forward_time=0.374, loss_ctc=49.495, loss_att=45.599, acc=0.751, loss=46.768, backward_time=0.480, grad_norm=37.412, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.022e-04, train_time=1.652
+[gpub058:0/16] 2024-02-04 01:11:05,362 (trainer:737) INFO: 25epoch:train:6901-7000batch: iter_time=8.780e-05, forward_time=0.290, loss_ctc=47.404, loss_att=50.071, acc=0.736, loss=49.271, backward_time=0.406, grad_norm=35.468, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.022e-04, train_time=1.706
+[gpub058:0/16] 2024-02-04 01:13:51,844 (trainer:737) INFO: 25epoch:train:7001-7100batch: iter_time=8.197e-05, forward_time=0.384, loss_ctc=55.934, loss_att=50.768, acc=0.737, loss=52.318, backward_time=0.478, grad_norm=47.547, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.022e-04, train_time=1.664
+[gpub058:0/16] 2024-02-04 01:16:25,696 (trainer:737) INFO: 25epoch:train:7101-7200batch: iter_time=9.088e-05, forward_time=0.291, loss_ctc=45.747, loss_att=49.255, acc=0.737, loss=48.202, backward_time=0.402, grad_norm=36.535, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.021e-04, train_time=1.539
+[gpub058:0/16] 2024-02-04 01:19:37,860 (trainer:737) INFO: 25epoch:train:7201-7300batch: iter_time=5.639e-04, forward_time=0.402, loss_ctc=51.918, loss_att=45.439, acc=0.743, loss=47.383, backward_time=0.512, grad_norm=42.785, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.021e-04, train_time=1.921
+[gpub058:0/16] 2024-02-04 01:22:07,952 (trainer:737) INFO: 25epoch:train:7301-7400batch: iter_time=8.648e-05, forward_time=0.291, loss_ctc=54.540, loss_att=53.874, acc=0.752, loss=54.074, backward_time=0.407, grad_norm=34.656, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.021e-04, train_time=1.500
+[gpub058:0/16] 2024-02-04 01:24:43,967 (trainer:737) INFO: 25epoch:train:7401-7500batch: iter_time=8.045e-05, forward_time=0.294, loss_ctc=46.710, loss_att=49.515, acc=0.743, loss=48.674, backward_time=0.427, grad_norm=35.516, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.021e-04, train_time=1.559
+[gpub058:0/16] 2024-02-04 01:25:03,995 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-04 01:25:22,855 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 01:25:26,709 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9b9e11570>)
+[gpub058:0/16] 2024-02-04 01:25:26,709 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-04 01:25:26,712 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 01:31:47,054 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 01:33:14,138 (trainer:737) INFO: 25epoch:train:7501-7600batch: iter_time=3.644, forward_time=0.358, loss_ctc=42.862, loss_att=49.095, acc=0.745, loss=47.225, backward_time=0.413, grad_norm=35.315, clip=100.000, loss_scale=7.133e+33, optim_step_time=0.107, optim0_lr0=2.020e-04, train_time=5.102
+[gpub058:0/16] 2024-02-04 01:35:38,916 (trainer:737) INFO: 25epoch:train:7601-7700batch: iter_time=8.324e-05, forward_time=0.290, loss_ctc=45.113, loss_att=37.576, acc=0.777, loss=39.837, backward_time=0.403, grad_norm=33.411, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.020e-04, train_time=1.448
+[gpub058:0/16] 2024-02-04 01:38:19,277 (trainer:737) INFO: 25epoch:train:7701-7800batch: iter_time=8.972e-05, forward_time=0.378, loss_ctc=45.152, loss_att=43.746, acc=0.767, loss=44.168, backward_time=0.440, grad_norm=34.132, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.020e-04, train_time=1.603
+[gpub058:0/16] 2024-02-04 01:40:52,138 (trainer:737) INFO: 25epoch:train:7801-7900batch: iter_time=9.209e-05, forward_time=0.292, loss_ctc=54.847, loss_att=55.980, acc=0.736, loss=55.640, backward_time=0.410, grad_norm=40.325, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.020e-04, train_time=1.528
+[gpub058:0/16] 2024-02-04 01:43:27,240 (trainer:737) INFO: 25epoch:train:7901-8000batch: iter_time=8.981e-05, forward_time=0.290, loss_ctc=44.776, loss_att=39.579, acc=0.753, loss=41.138, backward_time=0.405, grad_norm=34.153, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.019e-04, train_time=1.551
+[gpub058:0/16] 2024-02-04 01:46:01,178 (trainer:737) INFO: 25epoch:train:8001-8100batch: iter_time=9.604e-04, forward_time=0.296, loss_ctc=40.671, loss_att=41.152, acc=0.756, loss=41.008, backward_time=0.405, grad_norm=34.425, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.019e-04, train_time=1.539
+[gpub058:0/16] 2024-02-04 01:48:56,033 (trainer:737) INFO: 25epoch:train:8101-8200batch: iter_time=8.908e-05, forward_time=0.415, loss_ctc=48.977, loss_att=50.277, acc=0.742, loss=49.887, backward_time=0.444, grad_norm=37.934, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.019e-04, train_time=1.749
+[gpub058:0/16] 2024-02-04 01:49:48,732 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 01:51:21,003 (trainer:737) INFO: 25epoch:train:8201-8300batch: iter_time=8.263e-05, forward_time=0.292, loss_ctc=53.186, loss_att=52.511, acc=0.745, loss=52.714, backward_time=0.409, grad_norm=39.744, clip=100.000, loss_scale=3.593e+33, optim_step_time=0.095, optim0_lr0=2.018e-04, train_time=1.449
+[gpub058:0/16] 2024-02-04 01:53:55,040 (trainer:737) INFO: 25epoch:train:8301-8400batch: iter_time=0.002, forward_time=0.292, loss_ctc=52.883, loss_att=55.854, acc=0.748, loss=54.963, backward_time=0.405, grad_norm=45.913, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.018e-04, train_time=1.540
+[gpub058:0/16] 2024-02-04 01:56:53,696 (trainer:737) INFO: 25epoch:train:8401-8500batch: iter_time=2.168e-04, forward_time=0.437, loss_ctc=49.695, loss_att=46.202, acc=0.756, loss=47.250, backward_time=0.466, grad_norm=38.197, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.018e-04, train_time=1.786
+[gpub058:0/16] 2024-02-04 01:59:39,085 (trainer:737) INFO: 25epoch:train:8501-8600batch: iter_time=8.705e-05, forward_time=0.296, loss_ctc=49.595, loss_att=45.701, acc=0.757, loss=46.869, backward_time=0.406, grad_norm=36.404, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.018e-04, train_time=1.654
+[gpub058:0/16] 2024-02-04 02:02:05,680 (trainer:737) INFO: 25epoch:train:8601-8700batch: iter_time=8.068e-05, forward_time=0.293, loss_ctc=52.505, loss_att=55.807, acc=0.748, loss=54.816, backward_time=0.405, grad_norm=37.867, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.017e-04, train_time=1.466
+[gpub058:0/16] 2024-02-04 02:03:42,011 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-04 02:04:01,274 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 02:04:05,163 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9b9e13610>)
+[gpub058:0/16] 2024-02-04 02:04:05,163 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-04 02:04:05,167 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 02:10:26,661 (trainer:737) INFO: 25epoch:train:8701-8800batch: iter_time=3.395, forward_time=0.369, loss_ctc=44.650, loss_att=52.088, acc=0.732, loss=49.856, backward_time=0.413, grad_norm=39.555, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.017e-04, train_time=5.010
+[gpub058:0/16] 2024-02-04 02:12:50,157 (trainer:737) INFO: 25epoch:train:8801-8900batch: iter_time=0.003, forward_time=0.290, loss_ctc=42.626, loss_att=38.136, acc=0.771, loss=39.483, backward_time=0.400, grad_norm=33.264, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.017e-04, train_time=1.435
+[gpub058:0/16] 2024-02-04 02:15:26,554 (trainer:737) INFO: 25epoch:train:8901-9000batch: iter_time=8.488e-05, forward_time=0.289, loss_ctc=46.888, loss_att=43.580, acc=0.758, loss=44.572, backward_time=0.402, grad_norm=36.416, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.017e-04, train_time=1.563
+[gpub058:0/16] 2024-02-04 02:18:09,602 (trainer:737) INFO: 25epoch:train:9001-9100batch: iter_time=4.219e-04, forward_time=0.359, loss_ctc=47.971, loss_att=45.253, acc=0.755, loss=46.068, backward_time=0.487, grad_norm=36.502, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.016e-04, train_time=1.629
+[gpub058:0/16] 2024-02-04 02:20:51,130 (trainer:737) INFO: 25epoch:train:9101-9200batch: iter_time=8.687e-05, forward_time=0.288, loss_ctc=46.986, loss_att=44.255, acc=0.743, loss=45.074, backward_time=0.401, grad_norm=39.269, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.016e-04, train_time=1.616
+[gpub058:0/16] 2024-02-04 02:23:17,931 (trainer:737) INFO: 25epoch:train:9201-9300batch: iter_time=8.649e-05, forward_time=0.291, loss_ctc=46.953, loss_att=46.475, acc=0.739, loss=46.618, backward_time=0.408, grad_norm=36.423, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.016e-04, train_time=1.468
+[gpub058:0/16] 2024-02-04 02:25:48,812 (trainer:737) INFO: 25epoch:train:9301-9400batch: iter_time=5.314e-04, forward_time=0.289, loss_ctc=49.670, loss_att=46.439, acc=0.747, loss=47.408, backward_time=0.402, grad_norm=40.016, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.015e-04, train_time=1.508
+[gpub058:0/16] 2024-02-04 02:28:53,087 (trainer:737) INFO: 25epoch:train:9401-9500batch: iter_time=8.984e-05, forward_time=0.385, loss_ctc=47.181, loss_att=50.822, acc=0.733, loss=49.730, backward_time=0.473, grad_norm=36.901, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.015e-04, train_time=1.841
+[gpub058:0/16] 2024-02-04 02:31:18,018 (trainer:737) INFO: 25epoch:train:9501-9600batch: iter_time=9.044e-05, forward_time=0.291, loss_ctc=55.045, loss_att=51.207, acc=0.736, loss=52.359, backward_time=0.405, grad_norm=47.997, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.015e-04, train_time=1.450
+[gpub058:0/16] 2024-02-04 02:32:45,108 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 02:34:05,746 (trainer:737) INFO: 25epoch:train:9601-9700batch: iter_time=9.115e-05, forward_time=0.289, loss_ctc=45.674, loss_att=49.334, acc=0.738, loss=48.236, backward_time=0.403, grad_norm=36.640, clip=100.000, loss_scale=1.875e+33, optim_step_time=0.093, optim0_lr0=2.015e-04, train_time=1.677
+[gpub058:0/16] 2024-02-04 02:36:56,967 (trainer:737) INFO: 25epoch:train:9701-9800batch: iter_time=2.379e-04, forward_time=0.386, loss_ctc=51.926, loss_att=45.408, acc=0.745, loss=47.363, backward_time=0.453, grad_norm=41.969, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.100, optim0_lr0=2.014e-04, train_time=1.712
+[gpub058:0/16] 2024-02-04 02:39:54,546 (trainer:737) INFO: 25epoch:train:9801-9900batch: iter_time=9.081e-05, forward_time=0.300, loss_ctc=53.811, loss_att=53.978, acc=0.754, loss=53.928, backward_time=0.404, grad_norm=37.660, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.014e-04, train_time=1.774
+[gpub058:0/16] 2024-02-04 02:42:08,649 (trainer:737) INFO: 25epoch:train:9901-10000batch: iter_time=8.299e-05, forward_time=0.290, loss_ctc=46.271, loss_att=49.712, acc=0.744, loss=48.679, backward_time=0.404, grad_norm=36.443, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.014e-04, train_time=1.343
+[gpub058:0/16] 2024-02-04 02:42:28,677 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-04 02:42:48,730 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 02:42:52,661 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9c25d2800>)
+[gpub058:0/16] 2024-02-04 02:42:52,661 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-04 02:42:52,665 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 02:50:42,165 (trainer:737) INFO: 25epoch:train:10001-10100batch: iter_time=3.471, forward_time=0.363, loss_ctc=42.461, loss_att=47.922, acc=0.744, loss=46.283, backward_time=0.512, grad_norm=36.472, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.014e-04, train_time=5.135
+[gpub058:0/16] 2024-02-04 02:53:09,032 (trainer:737) INFO: 25epoch:train:10101-10200batch: iter_time=8.167e-05, forward_time=0.290, loss_ctc=44.969, loss_att=37.719, acc=0.765, loss=39.894, backward_time=0.401, grad_norm=33.526, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.013e-04, train_time=1.469
+[gpub058:0/16] 2024-02-04 02:56:09,177 (trainer:737) INFO: 25epoch:train:10201-10300batch: iter_time=8.844e-05, forward_time=0.291, loss_ctc=44.632, loss_att=41.987, acc=0.763, loss=42.780, backward_time=0.402, grad_norm=35.537, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.104, optim0_lr0=2.013e-04, train_time=1.798
+[gpub058:0/16] 2024-02-04 02:58:39,646 (trainer:737) INFO: 25epoch:train:10301-10400batch: iter_time=8.311e-05, forward_time=0.289, loss_ctc=55.075, loss_att=53.593, acc=0.731, loss=54.037, backward_time=0.403, grad_norm=40.886, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.013e-04, train_time=1.508
+[gpub058:0/16] 2024-02-04 03:01:39,507 (trainer:737) INFO: 25epoch:train:10401-10500batch: iter_time=8.766e-05, forward_time=0.408, loss_ctc=43.474, loss_att=38.642, acc=0.747, loss=40.092, backward_time=0.443, grad_norm=33.309, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.096, optim0_lr0=2.012e-04, train_time=1.798
+[gpub058:0/16] 2024-02-04 03:04:19,286 (trainer:737) INFO: 25epoch:train:10501-10600batch: iter_time=8.774e-05, forward_time=0.296, loss_ctc=40.552, loss_att=41.128, acc=0.750, loss=40.955, backward_time=0.404, grad_norm=36.966, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.095, optim0_lr0=2.012e-04, train_time=1.597
+[gpub058:0/16] 2024-02-04 03:07:03,554 (trainer:737) INFO: 25epoch:train:10601-10700batch: iter_time=2.616e-04, forward_time=0.299, loss_ctc=49.253, loss_att=49.512, acc=0.739, loss=49.434, backward_time=0.406, grad_norm=38.573, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.012e-04, train_time=1.642
+[gpub058:0/16] 2024-02-04 03:09:59,044 (trainer:737) INFO: 25epoch:train:10701-10800batch: iter_time=9.297e-05, forward_time=0.377, loss_ctc=53.208, loss_att=52.126, acc=0.741, loss=52.451, backward_time=0.453, grad_norm=60.945, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.101, optim0_lr0=2.012e-04, train_time=1.756
+[gpub058:0/16] 2024-02-04 03:12:49,662 (trainer:737) INFO: 25epoch:train:10801-10900batch: iter_time=9.329e-05, forward_time=0.297, loss_ctc=53.605, loss_att=53.791, acc=0.735, loss=53.735, backward_time=0.404, grad_norm=46.151, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.094, optim0_lr0=2.011e-04, train_time=1.706
+[gpub058:0/16] 2024-02-04 03:15:24,576 (trainer:737) INFO: 25epoch:train:10901-11000batch: iter_time=8.863e-05, forward_time=0.290, loss_ctc=49.323, loss_att=45.680, acc=0.747, loss=46.773, backward_time=0.403, grad_norm=37.743, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.011e-04, train_time=1.549
+[gpub058:0/16] 2024-02-04 03:18:04,464 (trainer:737) INFO: 25epoch:train:11001-11100batch: iter_time=0.001, forward_time=0.291, loss_ctc=49.630, loss_att=45.911, acc=0.747, loss=47.026, backward_time=0.407, grad_norm=37.319, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.011e-04, train_time=1.597
+[gpub058:0/16] 2024-02-04 03:20:56,716 (trainer:737) INFO: 25epoch:train:11101-11200batch: iter_time=8.577e-05, forward_time=0.386, loss_ctc=52.236, loss_att=55.349, acc=0.744, loss=54.415, backward_time=0.455, grad_norm=37.527, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.096, optim0_lr0=2.011e-04, train_time=1.724
+[gpub058:0/16] 2024-02-04 03:22:22,748 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-04 03:22:42,016 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 03:22:45,952 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9abbebd90>)
+[gpub058:0/16] 2024-02-04 03:22:45,952 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-04 03:22:45,955 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 03:28:40,139 (trainer:737) INFO: 25epoch:train:11201-11300batch: iter_time=3.174, forward_time=0.291, loss_ctc=43.919, loss_att=52.114, acc=0.735, loss=49.656, backward_time=0.406, grad_norm=35.901, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.010e-04, train_time=4.634
+[gpub058:0/16] 2024-02-04 03:31:17,515 (trainer:737) INFO: 25epoch:train:11301-11400batch: iter_time=7.534e-05, forward_time=0.291, loss_ctc=42.542, loss_att=38.426, acc=0.775, loss=39.661, backward_time=0.401, grad_norm=31.673, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.010e-04, train_time=1.574
+[gpub058:0/16] 2024-02-04 03:34:09,926 (trainer:737) INFO: 25epoch:train:11401-11500batch: iter_time=8.641e-05, forward_time=0.407, loss_ctc=46.179, loss_att=46.197, acc=0.761, loss=46.191, backward_time=0.453, grad_norm=37.708, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.095, optim0_lr0=2.010e-04, train_time=1.723
+[gpub058:0/16] 2024-02-04 03:37:02,004 (trainer:737) INFO: 25epoch:train:11501-11600batch: iter_time=9.001e-05, forward_time=0.295, loss_ctc=47.441, loss_att=45.054, acc=0.763, loss=45.770, backward_time=0.405, grad_norm=34.749, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.097, optim0_lr0=2.009e-04, train_time=1.721
+[gpub058:0/16] 2024-02-04 03:39:20,317 (trainer:737) INFO: 25epoch:train:11601-11700batch: iter_time=8.737e-05, forward_time=0.293, loss_ctc=46.918, loss_att=46.882, acc=0.748, loss=46.893, backward_time=0.405, grad_norm=34.923, clip=100.000, loss_scale=2.012e+33, optim_step_time=0.092, optim0_lr0=2.009e-04, train_time=1.383
+[gpub058:0/16] 2024-02-04 03:42:22,387 (trainer:737) INFO: 25epoch:train:11701-11800batch: iter_time=8.906e-05, forward_time=0.394, loss_ctc=46.702, loss_att=45.600, acc=0.751, loss=45.930, backward_time=0.465, grad_norm=36.550, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.009e-04, train_time=1.820
+[gpub058:0/16] 2024-02-04 03:45:12,389 (trainer:737) INFO: 25epoch:train:11801-11900batch: iter_time=8.892e-05, forward_time=0.293, loss_ctc=49.012, loss_att=46.495, acc=0.746, loss=47.250, backward_time=0.404, grad_norm=37.891, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.009e-04, train_time=1.700
+[gpub058:0/16] 2024-02-04 03:47:32,088 (trainer:737) INFO: 25epoch:train:11901-12000batch: iter_time=8.791e-05, forward_time=0.292, loss_ctc=46.519, loss_att=50.107, acc=0.746, loss=49.031, backward_time=0.406, grad_norm=35.931, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.008e-04, train_time=1.397
+[gpub058:0/16] 2024-02-04 03:50:42,176 (trainer:737) INFO: 25epoch:train:12001-12100batch: iter_time=8.250e-05, forward_time=0.430, loss_ctc=56.370, loss_att=51.072, acc=0.747, loss=52.661, backward_time=0.434, grad_norm=49.779, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.008e-04, train_time=1.899
+[gpub058:0/16] 2024-02-04 03:53:33,027 (trainer:737) INFO: 25epoch:train:12101-12200batch: iter_time=9.490e-05, forward_time=0.290, loss_ctc=45.527, loss_att=50.700, acc=0.751, loss=49.148, backward_time=0.407, grad_norm=35.704, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.008e-04, train_time=1.710
+[gpub058:0/16] 2024-02-04 03:55:57,739 (trainer:737) INFO: 25epoch:train:12201-12300batch: iter_time=8.158e-05, forward_time=0.290, loss_ctc=51.840, loss_att=45.665, acc=0.753, loss=47.517, backward_time=0.404, grad_norm=50.803, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.008e-04, train_time=1.447
+[gpub058:0/16] 2024-02-04 03:59:08,325 (trainer:737) INFO: 25epoch:train:12301-12400batch: iter_time=8.189e-05, forward_time=0.400, loss_ctc=53.573, loss_att=54.281, acc=0.760, loss=54.069, backward_time=0.448, grad_norm=34.777, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.007e-04, train_time=1.905
+[gpub058:0/16] 2024-02-04 04:01:50,143 (trainer:737) INFO: 25epoch:train:12401-12500batch: iter_time=2.251e-04, forward_time=0.292, loss_ctc=46.180, loss_att=49.203, acc=0.750, loss=48.296, backward_time=0.404, grad_norm=35.045, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.007e-04, train_time=1.618
+[gpub058:0/16] 2024-02-04 04:02:10,737 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-04 04:02:30,372 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 04:02:34,497 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9f2618790>)
+[gpub058:0/16] 2024-02-04 04:02:34,497 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-04 04:02:34,500 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 04:10:09,902 (trainer:737) INFO: 25epoch:train:12501-12600batch: iter_time=3.312, forward_time=0.289, loss_ctc=42.471, loss_att=48.995, acc=0.741, loss=47.038, backward_time=0.408, grad_norm=39.673, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.007e-04, train_time=4.998
+[gpub058:0/16] 2024-02-04 04:13:12,029 (trainer:737) INFO: 25epoch:train:12601-12700batch: iter_time=8.166e-05, forward_time=0.410, loss_ctc=44.996, loss_att=37.940, acc=0.767, loss=40.057, backward_time=0.438, grad_norm=35.885, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=2.006e-04, train_time=1.820
+[gpub058:0/16] 2024-02-04 04:16:00,017 (trainer:737) INFO: 25epoch:train:12701-12800batch: iter_time=8.576e-05, forward_time=0.289, loss_ctc=44.469, loss_att=42.040, acc=0.765, loss=42.769, backward_time=0.403, grad_norm=33.951, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.006e-04, train_time=1.680
+[gpub058:0/16] 2024-02-04 04:18:42,306 (trainer:737) INFO: 25epoch:train:12801-12900batch: iter_time=3.521e-04, forward_time=0.295, loss_ctc=54.178, loss_att=52.577, acc=0.733, loss=53.057, backward_time=0.404, grad_norm=40.249, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.006e-04, train_time=1.623
+[gpub058:0/16] 2024-02-04 04:21:41,356 (trainer:737) INFO: 25epoch:train:12901-13000batch: iter_time=8.351e-05, forward_time=0.408, loss_ctc=43.584, loss_att=38.972, acc=0.748, loss=40.355, backward_time=0.459, grad_norm=33.946, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.006e-04, train_time=1.790
+[gpub058:0/16] 2024-02-04 04:24:28,200 (trainer:737) INFO: 25epoch:train:13001-13100batch: iter_time=8.476e-05, forward_time=0.292, loss_ctc=40.546, loss_att=41.349, acc=0.749, loss=41.108, backward_time=0.401, grad_norm=34.266, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.005e-04, train_time=1.668
+[gpub058:0/16] 2024-02-04 04:27:12,274 (trainer:737) INFO: 25epoch:train:13101-13200batch: iter_time=1.652e-04, forward_time=0.292, loss_ctc=48.635, loss_att=49.828, acc=0.737, loss=49.470, backward_time=0.405, grad_norm=37.072, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.005e-04, train_time=1.640
+[gpub058:0/16] 2024-02-04 04:30:27,035 (trainer:737) INFO: 25epoch:train:13201-13300batch: iter_time=8.346e-05, forward_time=0.425, loss_ctc=51.977, loss_att=51.507, acc=0.742, loss=51.648, backward_time=0.444, grad_norm=41.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.005e-04, train_time=1.948
+[gpub058:0/16] 2024-02-04 04:33:06,870 (trainer:737) INFO: 25epoch:train:13301-13400batch: iter_time=8.262e-05, forward_time=0.291, loss_ctc=53.262, loss_att=53.815, acc=0.735, loss=53.649, backward_time=0.403, grad_norm=47.055, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.005e-04, train_time=1.598
+[gpub058:0/16] 2024-02-04 04:36:08,247 (trainer:737) INFO: 25epoch:train:13401-13500batch: iter_time=3.444e-04, forward_time=0.289, loss_ctc=49.580, loss_att=46.108, acc=0.745, loss=47.150, backward_time=0.401, grad_norm=37.640, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.004e-04, train_time=1.813
+[gpub058:0/16] 2024-02-04 04:38:58,926 (trainer:737) INFO: 25epoch:train:13501-13600batch: iter_time=8.852e-05, forward_time=0.396, loss_ctc=49.370, loss_att=45.434, acc=0.751, loss=46.615, backward_time=0.496, grad_norm=36.754, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.004e-04, train_time=1.707
+[gpub058:0/16] 2024-02-04 04:41:45,563 (trainer:737) INFO: 25epoch:train:13601-13700batch: iter_time=8.918e-05, forward_time=0.294, loss_ctc=52.203, loss_att=55.486, acc=0.744, loss=54.501, backward_time=0.403, grad_norm=37.336, clip=100.000, loss_scale=4.024e+33, optim_step_time=0.098, optim0_lr0=2.004e-04, train_time=1.665
+[gpub058:0/16] 2024-02-04 04:43:20,616 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-04 04:43:39,993 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 04:43:43,587 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9ac7775e0>)
+[gpub058:0/16] 2024-02-04 04:43:43,587 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-04 04:43:43,591 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 04:49:53,638 (trainer:737) INFO: 25epoch:train:13701-13800batch: iter_time=3.324, forward_time=0.379, loss_ctc=43.559, loss_att=50.528, acc=0.736, loss=48.437, backward_time=0.418, grad_norm=37.773, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.004e-04, train_time=4.881
+[gpub058:0/16] 2024-02-04 04:52:49,440 (trainer:737) INFO: 25epoch:train:13801-13900batch: iter_time=8.402e-05, forward_time=0.288, loss_ctc=42.249, loss_att=37.597, acc=0.773, loss=38.993, backward_time=0.399, grad_norm=33.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.003e-04, train_time=1.758
+[gpub058:0/16] 2024-02-04 04:55:27,128 (trainer:737) INFO: 25epoch:train:13901-14000batch: iter_time=7.968e-05, forward_time=0.295, loss_ctc=46.259, loss_att=42.874, acc=0.759, loss=43.890, backward_time=0.404, grad_norm=36.314, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.003e-04, train_time=1.577
+[gpub058:0/16] 2024-02-04 04:58:34,633 (trainer:737) INFO: 25epoch:train:14001-14100batch: iter_time=8.413e-05, forward_time=0.404, loss_ctc=47.800, loss_att=45.173, acc=0.757, loss=45.961, backward_time=0.464, grad_norm=35.610, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.003e-04, train_time=1.875
+[gpub058:0/16] 2024-02-04 05:01:23,672 (trainer:737) INFO: 25epoch:train:14101-14200batch: iter_time=1.222e-04, forward_time=0.292, loss_ctc=47.112, loss_att=42.719, acc=0.746, loss=44.037, backward_time=0.402, grad_norm=34.884, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.002e-04, train_time=1.690
+[gpub058:0/16] 2024-02-04 05:04:01,340 (trainer:737) INFO: 25epoch:train:14201-14300batch: iter_time=2.850e-04, forward_time=0.296, loss_ctc=46.855, loss_att=46.431, acc=0.740, loss=46.558, backward_time=0.408, grad_norm=39.896, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.002e-04, train_time=1.576
+[gpub058:0/16] 2024-02-04 05:07:38,357 (trainer:737) INFO: 25epoch:train:14301-14400batch: iter_time=9.645e-05, forward_time=0.378, loss_ctc=49.178, loss_att=45.541, acc=0.752, loss=46.633, backward_time=0.483, grad_norm=36.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.002e-04, train_time=2.169
+[gpub058:0/16] 2024-02-04 05:10:11,755 (trainer:737) INFO: 25epoch:train:14401-14500batch: iter_time=9.796e-05, forward_time=0.291, loss_ctc=46.641, loss_att=50.154, acc=0.737, loss=49.100, backward_time=0.403, grad_norm=36.071, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.002e-04, train_time=1.535
+[gpub058:0/16] 2024-02-04 05:12:57,592 (trainer:737) INFO: 25epoch:train:14501-14600batch: iter_time=1.330e-04, forward_time=0.308, loss_ctc=56.767, loss_att=50.561, acc=0.739, loss=52.423, backward_time=0.404, grad_norm=47.550, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.001e-04, train_time=1.658
+[gpub058:0/16] 2024-02-04 05:16:29,655 (trainer:737) INFO: 25epoch:train:14601-14700batch: iter_time=8.829e-05, forward_time=0.385, loss_ctc=45.288, loss_att=49.321, acc=0.736, loss=48.111, backward_time=0.427, grad_norm=36.488, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=2.001e-04, train_time=2.120
+[gpub058:0/16] 2024-02-04 05:19:09,652 (trainer:737) INFO: 25epoch:train:14701-14800batch: iter_time=8.022e-05, forward_time=0.292, loss_ctc=51.063, loss_att=44.862, acc=0.746, loss=46.722, backward_time=0.403, grad_norm=39.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.001e-04, train_time=1.600
+[gpub058:0/16] 2024-02-04 05:22:29,223 (trainer:737) INFO: 25epoch:train:14801-14900batch: iter_time=6.499e-04, forward_time=0.466, loss_ctc=54.240, loss_att=53.555, acc=0.754, loss=53.760, backward_time=0.424, grad_norm=36.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.001e-04, train_time=1.995
+[gpub058:0/16] 2024-02-04 05:25:25,168 (trainer:737) INFO: 25epoch:train:14901-15000batch: iter_time=8.168e-05, forward_time=0.292, loss_ctc=45.863, loss_att=49.066, acc=0.745, loss=48.105, backward_time=0.403, grad_norm=37.147, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.000e-04, train_time=1.759
+[gpub058:0/16] 2024-02-04 06:04:15,182 (trainer:343) INFO: 25epoch results: [train] iter_time=0.267, forward_time=0.339, loss_ctc=48.695, loss_att=47.619, acc=0.748, loss=47.942, backward_time=0.422, grad_norm=38.229, clip=100.000, loss_scale=6.095e+33, optim_step_time=0.096, optim0_lr0=2.021e-04, train_time=1.921, time=8 hours and 37.94 seconds, total_count=405000, gpu_max_cached_mem_GB=43.281, [valid] loss_ctc=38.945, cer_ctc=0.196, loss_att=39.437, acc=0.680, cer=0.315, wer=0.993, loss=39.289, time=38 minutes and 25.98 seconds, total_count=126117, gpu_max_cached_mem_GB=43.281
+[gpub058:0/16] 2024-02-04 06:04:25,602 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub058:0/16] 2024-02-04 06:04:25,658 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/19epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/20epoch.pth
+[gpub058:0/16] 2024-02-04 06:04:25,658 (trainer:272) INFO: 26/45epoch started. Estimated time to finish: 6 days, 23 hours and 36 minutes
+[gpub058:0/16] 2024-02-04 06:04:25,668 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-04 06:04:44,164 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 06:04:47,616 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe187be26e0>)
+[gpub058:0/16] 2024-02-04 06:04:47,616 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-04 06:04:47,619 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 06:10:51,539 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 06:11:39,815 (trainer:737) INFO: 26epoch:train:1-100batch: iter_time=2.877, forward_time=0.399, loss_ctc=42.997, loss_att=39.840, acc=0.753, loss=40.787, backward_time=0.410, grad_norm=34.904, clip=100.000, loss_scale=4.274e+33, optim_step_time=0.093, optim0_lr0=2.000e-04, train_time=4.341
+[gpub058:0/16] 2024-02-04 06:14:07,999 (trainer:737) INFO: 26epoch:train:101-200batch: iter_time=8.280e-05, forward_time=0.289, loss_ctc=48.175, loss_att=44.470, acc=0.732, loss=45.582, backward_time=0.403, grad_norm=38.764, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.000e-04, train_time=1.482
+[gpub058:0/16] 2024-02-04 06:16:29,175 (trainer:737) INFO: 26epoch:train:201-300batch: iter_time=8.096e-05, forward_time=0.291, loss_ctc=45.944, loss_att=41.575, acc=0.750, loss=42.886, backward_time=0.404, grad_norm=37.171, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.000e-04, train_time=1.409
+[gpub058:0/16] 2024-02-04 06:19:25,824 (trainer:737) INFO: 26epoch:train:301-400batch: iter_time=8.780e-05, forward_time=0.403, loss_ctc=49.179, loss_att=54.813, acc=0.737, loss=53.122, backward_time=0.448, grad_norm=37.761, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.117, optim0_lr0=1.999e-04, train_time=1.768
+[gpub058:0/16] 2024-02-04 06:21:46,054 (trainer:737) INFO: 26epoch:train:401-500batch: iter_time=8.175e-05, forward_time=0.289, loss_ctc=48.180, loss_att=43.175, acc=0.752, loss=44.676, backward_time=0.403, grad_norm=36.593, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.999e-04, train_time=1.402
+[gpub058:0/16] 2024-02-04 06:24:18,569 (trainer:737) INFO: 26epoch:train:501-600batch: iter_time=8.643e-05, forward_time=0.289, loss_ctc=53.651, loss_att=44.773, acc=0.757, loss=47.436, backward_time=0.403, grad_norm=40.225, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.999e-04, train_time=1.525
+[gpub058:0/16] 2024-02-04 06:26:44,698 (trainer:737) INFO: 26epoch:train:601-700batch: iter_time=7.967e-05, forward_time=0.320, loss_ctc=50.965, loss_att=54.554, acc=0.725, loss=53.478, backward_time=0.409, grad_norm=38.592, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.998e-04, train_time=1.460
+[gpub058:0/16] 2024-02-04 06:29:39,384 (trainer:737) INFO: 26epoch:train:701-800batch: iter_time=8.690e-05, forward_time=0.352, loss_ctc=45.157, loss_att=44.738, acc=0.731, loss=44.864, backward_time=0.449, grad_norm=36.758, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.998e-04, train_time=1.748
+[gpub058:0/16] 2024-02-04 06:32:00,304 (trainer:737) INFO: 26epoch:train:801-900batch: iter_time=9.442e-05, forward_time=0.288, loss_ctc=49.692, loss_att=42.318, acc=0.748, loss=44.531, backward_time=0.402, grad_norm=45.053, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.998e-04, train_time=1.409
+[gpub058:0/16] 2024-02-04 06:34:45,002 (trainer:737) INFO: 26epoch:train:901-1000batch: iter_time=9.040e-05, forward_time=0.302, loss_ctc=59.491, loss_att=59.083, acc=0.724, loss=59.205, backward_time=0.418, grad_norm=48.333, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.998e-04, train_time=1.645
+[gpub058:0/16] 2024-02-04 06:37:25,868 (trainer:737) INFO: 26epoch:train:1001-1100batch: iter_time=1.032e-04, forward_time=0.290, loss_ctc=53.671, loss_att=52.135, acc=0.722, loss=52.596, backward_time=0.440, grad_norm=48.378, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.997e-04, train_time=1.610
+[gpub058:0/16] 2024-02-04 06:40:23,497 (trainer:737) INFO: 26epoch:train:1101-1200batch: iter_time=9.966e-05, forward_time=0.403, loss_ctc=61.127, loss_att=63.294, acc=0.712, loss=62.644, backward_time=0.431, grad_norm=41.967, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.997e-04, train_time=1.776
+[gpub058:0/16] 2024-02-04 06:42:08,427 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-04 06:42:27,452 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 06:42:31,005 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c50a8d60>)
+[gpub058:0/16] 2024-02-04 06:42:31,005 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-04 06:42:31,008 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 06:48:50,663 (trainer:737) INFO: 26epoch:train:1201-1300batch: iter_time=3.305, forward_time=0.288, loss_ctc=41.678, loss_att=44.027, acc=0.744, loss=43.322, backward_time=0.399, grad_norm=33.238, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.997e-04, train_time=5.071
+[gpub058:0/16] 2024-02-04 06:51:07,784 (trainer:737) INFO: 26epoch:train:1301-1400batch: iter_time=8.380e-05, forward_time=0.296, loss_ctc=46.864, loss_att=43.055, acc=0.751, loss=44.198, backward_time=0.406, grad_norm=35.907, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.997e-04, train_time=1.371
+[gpub058:0/16] 2024-02-04 06:54:13,739 (trainer:737) INFO: 26epoch:train:1401-1500batch: iter_time=8.285e-05, forward_time=0.423, loss_ctc=44.759, loss_att=41.968, acc=0.744, loss=42.805, backward_time=0.439, grad_norm=35.511, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.996e-04, train_time=1.860
+[gpub058:0/16] 2024-02-04 06:57:02,797 (trainer:737) INFO: 26epoch:train:1501-1600batch: iter_time=8.266e-05, forward_time=0.292, loss_ctc=47.121, loss_att=44.181, acc=0.756, loss=45.063, backward_time=0.402, grad_norm=35.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.996e-04, train_time=1.690
+[gpub058:0/16] 2024-02-04 06:59:37,238 (trainer:737) INFO: 26epoch:train:1601-1700batch: iter_time=8.528e-05, forward_time=0.292, loss_ctc=46.141, loss_att=51.108, acc=0.742, loss=49.618, backward_time=0.413, grad_norm=37.077, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.996e-04, train_time=1.544
+[gpub058:0/16] 2024-02-04 07:02:14,661 (trainer:737) INFO: 26epoch:train:1701-1800batch: iter_time=7.974e-05, forward_time=0.304, loss_ctc=53.706, loss_att=45.265, acc=0.751, loss=47.798, backward_time=0.407, grad_norm=40.875, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.996e-04, train_time=1.574
+[gpub058:0/16] 2024-02-04 07:04:57,729 (trainer:737) INFO: 26epoch:train:1801-1900batch: iter_time=8.273e-05, forward_time=0.402, loss_ctc=50.855, loss_att=47.155, acc=0.752, loss=48.265, backward_time=0.413, grad_norm=38.611, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.995e-04, train_time=1.631
+[gpub058:0/16] 2024-02-04 07:07:51,696 (trainer:737) INFO: 26epoch:train:1901-2000batch: iter_time=8.613e-05, forward_time=0.288, loss_ctc=44.829, loss_att=48.337, acc=0.739, loss=47.284, backward_time=0.401, grad_norm=35.006, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.995e-04, train_time=1.739
+[gpub058:0/16] 2024-02-04 07:10:11,945 (trainer:737) INFO: 26epoch:train:2001-2100batch: iter_time=8.319e-05, forward_time=0.302, loss_ctc=46.037, loss_att=43.363, acc=0.736, loss=44.165, backward_time=0.406, grad_norm=39.127, clip=100.000, loss_scale=3.505e+33, optim_step_time=0.097, optim0_lr0=1.995e-04, train_time=1.402
+[gpub058:0/16] 2024-02-04 07:13:14,368 (trainer:737) INFO: 26epoch:train:2101-2200batch: iter_time=8.227e-05, forward_time=0.413, loss_ctc=52.216, loss_att=49.429, acc=0.739, loss=50.265, backward_time=0.438, grad_norm=43.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.994e-04, train_time=1.824
+[gpub058:0/16] 2024-02-04 07:15:52,216 (trainer:737) INFO: 26epoch:train:2201-2300batch: iter_time=8.653e-05, forward_time=0.287, loss_ctc=49.317, loss_att=47.519, acc=0.732, loss=48.058, backward_time=0.404, grad_norm=40.936, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.994e-04, train_time=1.579
+[gpub058:0/16] 2024-02-04 07:18:30,058 (trainer:737) INFO: 26epoch:train:2301-2400batch: iter_time=8.896e-05, forward_time=0.308, loss_ctc=58.260, loss_att=63.861, acc=0.711, loss=62.180, backward_time=0.410, grad_norm=46.176, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.994e-04, train_time=1.578
+[gpub058:0/16] 2024-02-04 07:21:09,530 (trainer:737) INFO: 26epoch:train:2401-2500batch: iter_time=8.091e-05, forward_time=0.373, loss_ctc=53.494, loss_att=55.686, acc=0.729, loss=55.029, backward_time=0.460, grad_norm=38.076, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.994e-04, train_time=1.594
+[gpub058:0/16] 2024-02-04 07:21:29,582 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-04 07:21:49,072 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 07:21:52,671 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7ff12c326bf0>)
+[gpub058:0/16] 2024-02-04 07:21:52,671 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-04 07:21:52,674 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 07:29:23,752 (trainer:737) INFO: 26epoch:train:2501-2600batch: iter_time=3.405, forward_time=0.297, loss_ctc=41.943, loss_att=38.699, acc=0.762, loss=39.672, backward_time=0.399, grad_norm=34.398, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.993e-04, train_time=4.942
+[gpub058:0/16] 2024-02-04 07:32:14,525 (trainer:737) INFO: 26epoch:train:2601-2700batch: iter_time=7.796e-05, forward_time=0.297, loss_ctc=46.636, loss_att=43.623, acc=0.736, loss=44.527, backward_time=0.409, grad_norm=39.112, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.993e-04, train_time=1.708
+[gpub058:0/16] 2024-02-04 07:35:27,338 (trainer:737) INFO: 26epoch:train:2701-2800batch: iter_time=8.196e-05, forward_time=0.362, loss_ctc=44.611, loss_att=41.213, acc=0.756, loss=42.232, backward_time=0.413, grad_norm=34.407, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.993e-04, train_time=1.928
+[gpub058:0/16] 2024-02-04 07:38:00,344 (trainer:737) INFO: 26epoch:train:2801-2900batch: iter_time=8.086e-05, forward_time=0.369, loss_ctc=48.082, loss_att=54.261, acc=0.741, loss=52.408, backward_time=0.418, grad_norm=36.431, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.993e-04, train_time=1.529
+[gpub058:0/16] 2024-02-04 07:40:53,027 (trainer:737) INFO: 26epoch:train:2901-3000batch: iter_time=7.872e-05, forward_time=0.288, loss_ctc=47.037, loss_att=41.818, acc=0.757, loss=43.384, backward_time=0.401, grad_norm=35.115, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.992e-04, train_time=1.727
+[gpub058:0/16] 2024-02-04 07:43:51,703 (trainer:737) INFO: 26epoch:train:3001-3100batch: iter_time=8.057e-05, forward_time=0.294, loss_ctc=52.217, loss_att=43.780, acc=0.761, loss=46.311, backward_time=0.421, grad_norm=43.464, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.992e-04, train_time=1.787
+[gpub058:0/16] 2024-02-04 07:46:42,512 (trainer:737) INFO: 26epoch:train:3101-3200batch: iter_time=8.446e-05, forward_time=0.404, loss_ctc=48.840, loss_att=53.450, acc=0.728, loss=52.067, backward_time=0.469, grad_norm=39.953, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.992e-04, train_time=1.708
+[gpub058:0/16] 2024-02-04 07:49:36,470 (trainer:737) INFO: 26epoch:train:3201-3300batch: iter_time=8.117e-05, forward_time=0.288, loss_ctc=45.166, loss_att=44.739, acc=0.733, loss=44.867, backward_time=0.399, grad_norm=38.194, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.992e-04, train_time=1.739
+[gpub058:0/16] 2024-02-04 07:52:19,286 (trainer:737) INFO: 26epoch:train:3301-3400batch: iter_time=8.603e-05, forward_time=0.303, loss_ctc=47.176, loss_att=40.976, acc=0.754, loss=42.836, backward_time=0.405, grad_norm=40.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.991e-04, train_time=1.627
+[gpub058:0/16] 2024-02-04 07:54:55,093 (trainer:737) INFO: 26epoch:train:3401-3500batch: iter_time=8.124e-05, forward_time=0.369, loss_ctc=56.462, loss_att=58.065, acc=0.726, loss=57.584, backward_time=0.460, grad_norm=41.161, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.991e-04, train_time=1.559
+[gpub058:0/16] 2024-02-04 07:58:02,037 (trainer:737) INFO: 26epoch:train:3501-3600batch: iter_time=8.315e-05, forward_time=0.304, loss_ctc=51.082, loss_att=51.264, acc=0.726, loss=51.209, backward_time=0.402, grad_norm=45.637, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.991e-04, train_time=1.869
+[gpub058:0/16] 2024-02-04 08:00:24,581 (trainer:737) INFO: 26epoch:train:3601-3700batch: iter_time=8.136e-05, forward_time=0.304, loss_ctc=59.511, loss_att=62.028, acc=0.717, loss=61.273, backward_time=0.409, grad_norm=40.177, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.991e-04, train_time=1.425
+[gpub058:0/16] 2024-02-04 08:01:54,279 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-04 08:02:13,289 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 08:02:17,352 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fec1c933970>)
+[gpub058:0/16] 2024-02-04 08:02:17,352 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-04 08:02:17,356 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 08:08:37,044 (trainer:737) INFO: 26epoch:train:3701-3800batch: iter_time=3.390, forward_time=0.396, loss_ctc=40.288, loss_att=44.776, acc=0.751, loss=43.430, backward_time=0.427, grad_norm=33.584, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.990e-04, train_time=4.924
+[gpub058:0/16] 2024-02-04 08:10:51,186 (trainer:737) INFO: 26epoch:train:3801-3900batch: iter_time=8.191e-05, forward_time=0.290, loss_ctc=46.260, loss_att=45.089, acc=0.749, loss=45.440, backward_time=0.404, grad_norm=35.724, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.990e-04, train_time=1.341
+[gpub058:0/16] 2024-02-04 08:13:41,882 (trainer:737) INFO: 26epoch:train:3901-4000batch: iter_time=8.572e-05, forward_time=0.303, loss_ctc=43.398, loss_att=42.149, acc=0.752, loss=42.524, backward_time=0.407, grad_norm=34.421, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.990e-04, train_time=1.707
+[gpub058:0/16] 2024-02-04 08:16:22,845 (trainer:737) INFO: 26epoch:train:4001-4100batch: iter_time=9.220e-05, forward_time=0.389, loss_ctc=46.585, loss_att=44.853, acc=0.760, loss=45.372, backward_time=0.450, grad_norm=35.826, clip=100.000, loss_scale=7.010e+33, optim_step_time=0.096, optim0_lr0=1.989e-04, train_time=1.609
+[gpub058:0/16] 2024-02-04 08:18:54,381 (trainer:737) INFO: 26epoch:train:4101-4200batch: iter_time=9.151e-05, forward_time=0.290, loss_ctc=45.596, loss_att=51.141, acc=0.751, loss=49.478, backward_time=0.402, grad_norm=37.500, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.989e-04, train_time=1.515
+[gpub058:0/16] 2024-02-04 08:21:36,637 (trainer:737) INFO: 26epoch:train:4201-4300batch: iter_time=8.990e-05, forward_time=0.308, loss_ctc=52.332, loss_att=47.114, acc=0.752, loss=48.680, backward_time=0.408, grad_norm=39.984, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.989e-04, train_time=1.622
+[gpub058:0/16] 2024-02-04 08:24:16,578 (trainer:737) INFO: 26epoch:train:4301-4400batch: iter_time=8.099e-04, forward_time=0.412, loss_ctc=49.317, loss_att=47.463, acc=0.755, loss=48.019, backward_time=0.436, grad_norm=38.438, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.989e-04, train_time=1.599
+[gpub058:0/16] 2024-02-04 08:25:58,764 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 08:26:50,806 (trainer:737) INFO: 26epoch:train:4401-4500batch: iter_time=8.880e-05, forward_time=0.298, loss_ctc=44.556, loss_att=48.497, acc=0.753, loss=47.315, backward_time=0.408, grad_norm=34.418, clip=100.000, loss_scale=8.969e+33, optim_step_time=0.095, optim0_lr0=1.988e-04, train_time=1.542
+[gpub058:0/16] 2024-02-04 08:29:30,356 (trainer:737) INFO: 26epoch:train:4501-4600batch: iter_time=8.681e-05, forward_time=0.356, loss_ctc=45.654, loss_att=46.034, acc=0.744, loss=45.920, backward_time=0.434, grad_norm=38.111, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.988e-04, train_time=1.596
+[gpub058:0/16] 2024-02-04 08:31:50,825 (trainer:737) INFO: 26epoch:train:4601-4700batch: iter_time=9.409e-05, forward_time=0.291, loss_ctc=50.666, loss_att=49.498, acc=0.744, loss=49.848, backward_time=0.405, grad_norm=41.012, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.988e-04, train_time=1.404
+[gpub058:0/16] 2024-02-04 08:34:09,086 (trainer:737) INFO: 26epoch:train:4701-4800batch: iter_time=3.820e-04, forward_time=0.298, loss_ctc=48.124, loss_att=49.244, acc=0.744, loss=48.908, backward_time=0.409, grad_norm=40.175, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.988e-04, train_time=1.381
+[gpub058:0/16] 2024-02-04 08:37:24,534 (trainer:737) INFO: 26epoch:train:4801-4900batch: iter_time=9.432e-05, forward_time=0.426, loss_ctc=56.738, loss_att=66.829, acc=0.709, loss=63.802, backward_time=0.453, grad_norm=44.198, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.987e-04, train_time=1.956
+[gpub058:0/16] 2024-02-04 08:39:40,073 (trainer:737) INFO: 26epoch:train:4901-5000batch: iter_time=8.310e-05, forward_time=0.293, loss_ctc=53.583, loss_att=55.226, acc=0.741, loss=54.733, backward_time=0.411, grad_norm=39.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.987e-04, train_time=1.355
+[gpub058:0/16] 2024-02-04 08:40:00,187 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-04 08:40:19,704 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 08:40:23,365 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1e7553610>)
+[gpub058:0/16] 2024-02-04 08:40:23,366 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-04 08:40:23,369 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 08:47:47,099 (trainer:737) INFO: 26epoch:train:5001-5100batch: iter_time=3.360, forward_time=0.395, loss_ctc=41.662, loss_att=39.157, acc=0.762, loss=39.908, backward_time=0.414, grad_norm=33.193, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.987e-04, train_time=4.870
+[gpub058:0/16] 2024-02-04 08:50:24,480 (trainer:737) INFO: 26epoch:train:5101-5200batch: iter_time=8.135e-05, forward_time=0.289, loss_ctc=45.716, loss_att=44.288, acc=0.737, loss=44.716, backward_time=0.401, grad_norm=38.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.987e-04, train_time=1.574
+[gpub058:0/16] 2024-02-04 08:53:22,222 (trainer:737) INFO: 26epoch:train:5201-5300batch: iter_time=0.108, forward_time=0.379, loss_ctc=43.861, loss_att=40.525, acc=0.759, loss=41.526, backward_time=0.466, grad_norm=34.392, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.986e-04, train_time=1.777
+[gpub058:0/16] 2024-02-04 08:58:00,277 (trainer:737) INFO: 26epoch:train:5301-5400batch: iter_time=0.045, forward_time=0.290, loss_ctc=47.232, loss_att=54.736, acc=0.739, loss=52.485, backward_time=0.401, grad_norm=37.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.986e-04, train_time=2.780
+[gpub058:0/16] 2024-02-04 09:00:19,063 (trainer:737) INFO: 26epoch:train:5401-5500batch: iter_time=7.852e-05, forward_time=0.289, loss_ctc=47.165, loss_att=42.506, acc=0.756, loss=43.904, backward_time=0.403, grad_norm=34.238, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.986e-04, train_time=1.388
+[gpub058:0/16] 2024-02-04 09:05:39,622 (trainer:737) INFO: 26epoch:train:5501-5600batch: iter_time=8.935e-04, forward_time=0.423, loss_ctc=50.669, loss_att=43.387, acc=0.766, loss=45.572, backward_time=0.482, grad_norm=38.338, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.986e-04, train_time=3.205
+[gpub058:0/16] 2024-02-04 09:08:09,632 (trainer:737) INFO: 26epoch:train:5601-5700batch: iter_time=8.241e-05, forward_time=0.292, loss_ctc=47.834, loss_att=53.221, acc=0.731, loss=51.605, backward_time=0.404, grad_norm=38.049, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.985e-04, train_time=1.500
+[gpub058:0/16] 2024-02-04 09:11:18,999 (trainer:737) INFO: 26epoch:train:5701-5800batch: iter_time=9.639e-05, forward_time=0.287, loss_ctc=44.208, loss_att=44.811, acc=0.734, loss=44.630, backward_time=0.400, grad_norm=36.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.985e-04, train_time=1.893
+[gpub058:0/16] 2024-02-04 09:15:20,005 (trainer:737) INFO: 26epoch:train:5801-5900batch: iter_time=8.340e-05, forward_time=0.499, loss_ctc=46.103, loss_att=41.024, acc=0.754, loss=42.547, backward_time=0.452, grad_norm=40.395, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.985e-04, train_time=2.409
+[gpub058:0/16] 2024-02-04 09:18:24,680 (trainer:737) INFO: 26epoch:train:5901-6000batch: iter_time=8.895e-05, forward_time=0.295, loss_ctc=55.294, loss_att=57.743, acc=0.729, loss=57.009, backward_time=0.405, grad_norm=43.373, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.984e-04, train_time=1.848
+[gpub058:0/16] 2024-02-04 09:21:14,178 (trainer:737) INFO: 26epoch:train:6001-6100batch: iter_time=7.153e-04, forward_time=0.305, loss_ctc=50.067, loss_att=51.151, acc=0.728, loss=50.826, backward_time=0.458, grad_norm=45.001, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.984e-04, train_time=1.694
+[gpub058:0/16] 2024-02-04 09:23:51,222 (trainer:737) INFO: 26epoch:train:6101-6200batch: iter_time=8.656e-05, forward_time=0.384, loss_ctc=59.535, loss_att=62.192, acc=0.715, loss=61.395, backward_time=0.417, grad_norm=41.198, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.984e-04, train_time=1.571
+[gpub058:0/16] 2024-02-04 09:25:16,473 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-04 09:25:35,751 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 09:25:39,336 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe96915f820>)
+[gpub058:0/16] 2024-02-04 09:25:39,336 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-04 09:25:39,339 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 09:32:02,690 (trainer:737) INFO: 26epoch:train:6201-6300batch: iter_time=3.427, forward_time=0.289, loss_ctc=40.603, loss_att=44.265, acc=0.752, loss=43.167, backward_time=0.402, grad_norm=33.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.984e-04, train_time=4.915
+[gpub058:0/16] 2024-02-04 09:34:44,931 (trainer:737) INFO: 26epoch:train:6301-6400batch: iter_time=8.789e-05, forward_time=0.422, loss_ctc=45.489, loss_att=44.552, acc=0.753, loss=44.833, backward_time=0.426, grad_norm=35.236, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.983e-04, train_time=1.622
+[gpub058:0/16] 2024-02-04 09:37:28,931 (trainer:737) INFO: 26epoch:train:6401-6500batch: iter_time=8.031e-05, forward_time=0.290, loss_ctc=43.028, loss_att=41.779, acc=0.753, loss=42.154, backward_time=0.404, grad_norm=35.064, clip=100.000, loss_scale=6.594e+33, optim_step_time=0.092, optim0_lr0=1.983e-04, train_time=1.640
+[gpub058:0/16] 2024-02-04 09:40:00,185 (trainer:737) INFO: 26epoch:train:6501-6600batch: iter_time=7.913e-05, forward_time=0.291, loss_ctc=46.363, loss_att=44.379, acc=0.765, loss=44.974, backward_time=0.404, grad_norm=36.956, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.983e-04, train_time=1.513
+[gpub058:0/16] 2024-02-04 09:42:37,561 (trainer:737) INFO: 26epoch:train:6601-6700batch: iter_time=8.432e-05, forward_time=0.377, loss_ctc=45.380, loss_att=50.399, acc=0.754, loss=48.893, backward_time=0.454, grad_norm=37.432, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.983e-04, train_time=1.572
+[gpub058:0/16] 2024-02-04 09:45:26,547 (trainer:737) INFO: 26epoch:train:6701-6800batch: iter_time=8.191e-05, forward_time=0.290, loss_ctc=51.512, loss_att=46.483, acc=0.755, loss=47.992, backward_time=0.403, grad_norm=40.172, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.982e-04, train_time=1.691
+[gpub058:0/16] 2024-02-04 09:48:07,509 (trainer:737) INFO: 26epoch:train:6801-6900batch: iter_time=8.404e-05, forward_time=0.323, loss_ctc=48.365, loss_att=47.655, acc=0.755, loss=47.868, backward_time=0.407, grad_norm=37.171, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.982e-04, train_time=1.610
+[gpub058:0/16] 2024-02-04 09:50:35,857 (trainer:737) INFO: 26epoch:train:6901-7000batch: iter_time=8.153e-05, forward_time=0.420, loss_ctc=43.967, loss_att=48.246, acc=0.755, loss=46.963, backward_time=0.438, grad_norm=33.767, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.982e-04, train_time=1.482
+[gpub058:0/16] 2024-02-04 09:52:50,481 (trainer:737) INFO: 26epoch:train:7001-7100batch: iter_time=7.645e-05, forward_time=0.291, loss_ctc=44.538, loss_att=44.972, acc=0.747, loss=44.842, backward_time=0.404, grad_norm=40.764, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.982e-04, train_time=1.347
+[gpub058:0/16] 2024-02-04 09:56:06,665 (trainer:737) INFO: 26epoch:train:7101-7200batch: iter_time=8.113e-05, forward_time=0.425, loss_ctc=50.402, loss_att=48.836, acc=0.747, loss=49.306, backward_time=0.437, grad_norm=43.191, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.981e-04, train_time=1.962
+[gpub058:0/16] 2024-02-04 09:56:47,480 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 09:58:25,895 (trainer:737) INFO: 26epoch:train:7201-7300batch: iter_time=9.247e-05, forward_time=0.290, loss_ctc=48.127, loss_att=48.715, acc=0.745, loss=48.539, backward_time=0.405, grad_norm=42.169, clip=100.000, loss_scale=6.608e+33, optim_step_time=0.092, optim0_lr0=1.981e-04, train_time=1.391
+[gpub058:0/16] 2024-02-04 10:01:10,036 (trainer:737) INFO: 26epoch:train:7301-7400batch: iter_time=8.514e-05, forward_time=0.359, loss_ctc=56.387, loss_att=66.384, acc=0.712, loss=63.385, backward_time=0.454, grad_norm=47.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.981e-04, train_time=1.642
+[gpub058:0/16] 2024-02-04 10:04:06,350 (trainer:737) INFO: 26epoch:train:7401-7500batch: iter_time=8.808e-05, forward_time=0.327, loss_ctc=53.493, loss_att=55.064, acc=0.741, loss=54.593, backward_time=0.414, grad_norm=36.975, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.981e-04, train_time=1.763
+[gpub058:0/16] 2024-02-04 10:04:26,378 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-04 10:04:45,481 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 10:04:49,088 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d3c2bf10>)
+[gpub058:0/16] 2024-02-04 10:04:49,088 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-04 10:04:49,091 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 10:12:29,665 (trainer:737) INFO: 26epoch:train:7501-7600batch: iter_time=3.549, forward_time=0.288, loss_ctc=41.426, loss_att=38.479, acc=0.767, loss=39.363, backward_time=0.401, grad_norm=32.340, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.980e-04, train_time=5.031
+[gpub058:0/16] 2024-02-04 10:15:29,643 (trainer:737) INFO: 26epoch:train:7601-7700batch: iter_time=8.464e-05, forward_time=0.330, loss_ctc=45.081, loss_att=43.168, acc=0.749, loss=43.742, backward_time=0.430, grad_norm=38.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.980e-04, train_time=1.801
+[gpub058:0/16] 2024-02-04 10:18:05,204 (trainer:737) INFO: 26epoch:train:7701-7800batch: iter_time=9.493e-05, forward_time=0.339, loss_ctc=43.292, loss_att=40.140, acc=0.769, loss=41.086, backward_time=0.413, grad_norm=32.840, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.980e-04, train_time=1.554
+[gpub058:0/16] 2024-02-04 10:20:52,014 (trainer:737) INFO: 26epoch:train:7801-7900batch: iter_time=8.890e-05, forward_time=0.291, loss_ctc=46.753, loss_att=53.505, acc=0.745, loss=51.479, backward_time=0.405, grad_norm=36.634, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.980e-04, train_time=1.668
+[gpub058:0/16] 2024-02-04 10:24:07,836 (trainer:737) INFO: 26epoch:train:7901-8000batch: iter_time=9.619e-05, forward_time=0.397, loss_ctc=46.688, loss_att=42.180, acc=0.771, loss=43.532, backward_time=0.442, grad_norm=35.494, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.979e-04, train_time=1.959
+[gpub058:0/16] 2024-02-04 10:26:30,602 (trainer:737) INFO: 26epoch:train:8001-8100batch: iter_time=8.859e-05, forward_time=0.307, loss_ctc=50.507, loss_att=45.156, acc=0.766, loss=46.761, backward_time=0.405, grad_norm=38.860, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.979e-04, train_time=1.428
+[gpub058:0/16] 2024-02-04 10:29:34,091 (trainer:737) INFO: 26epoch:train:8101-8200batch: iter_time=1.021e-04, forward_time=0.410, loss_ctc=47.701, loss_att=53.364, acc=0.737, loss=51.665, backward_time=0.432, grad_norm=41.491, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.979e-04, train_time=1.834
+[gpub058:0/16] 2024-02-04 10:32:23,109 (trainer:737) INFO: 26epoch:train:8201-8300batch: iter_time=1.046e-04, forward_time=0.290, loss_ctc=44.800, loss_att=46.734, acc=0.743, loss=46.154, backward_time=0.401, grad_norm=38.172, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.979e-04, train_time=1.691
+[gpub058:0/16] 2024-02-04 10:34:58,368 (trainer:737) INFO: 26epoch:train:8301-8400batch: iter_time=9.801e-05, forward_time=0.289, loss_ctc=46.347, loss_att=41.325, acc=0.762, loss=42.832, backward_time=0.401, grad_norm=40.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.978e-04, train_time=1.552
+[gpub058:0/16] 2024-02-04 10:38:04,464 (trainer:737) INFO: 26epoch:train:8401-8500batch: iter_time=4.707e-04, forward_time=0.371, loss_ctc=54.995, loss_att=57.724, acc=0.744, loss=56.905, backward_time=0.478, grad_norm=43.011, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.978e-04, train_time=1.859
+[gpub058:0/16] 2024-02-04 10:40:34,215 (trainer:737) INFO: 26epoch:train:8501-8600batch: iter_time=9.592e-05, forward_time=0.294, loss_ctc=49.871, loss_att=52.502, acc=0.736, loss=51.713, backward_time=0.405, grad_norm=47.030, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.978e-04, train_time=1.498
+[gpub058:0/16] 2024-02-04 10:43:36,360 (trainer:737) INFO: 26epoch:train:8601-8700batch: iter_time=9.454e-05, forward_time=0.293, loss_ctc=59.530, loss_att=63.268, acc=0.720, loss=62.147, backward_time=0.407, grad_norm=40.246, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.978e-04, train_time=1.821
+[gpub058:0/16] 2024-02-04 10:45:17,736 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-04 10:45:37,372 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 10:45:40,986 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9d5357610>)
+[gpub058:0/16] 2024-02-04 10:45:40,986 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-04 10:45:40,989 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 10:51:44,028 (trainer:737) INFO: 26epoch:train:8701-8800batch: iter_time=3.319, forward_time=0.355, loss_ctc=40.097, loss_att=43.190, acc=0.757, loss=42.262, backward_time=0.412, grad_norm=34.619, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.977e-04, train_time=4.876
+[gpub058:0/16] 2024-02-04 10:54:21,941 (trainer:737) INFO: 26epoch:train:8801-8900batch: iter_time=8.397e-05, forward_time=0.289, loss_ctc=44.782, loss_att=43.107, acc=0.754, loss=43.610, backward_time=0.402, grad_norm=35.731, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.977e-04, train_time=1.579
+[gpub058:0/16] 2024-02-04 10:57:08,582 (trainer:737) INFO: 26epoch:train:8901-9000batch: iter_time=8.523e-05, forward_time=0.377, loss_ctc=42.733, loss_att=41.640, acc=0.748, loss=41.968, backward_time=0.426, grad_norm=35.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.977e-04, train_time=1.666
+[gpub058:0/16] 2024-02-04 10:59:37,896 (trainer:737) INFO: 26epoch:train:9001-9100batch: iter_time=8.323e-05, forward_time=0.290, loss_ctc=45.826, loss_att=44.141, acc=0.760, loss=44.646, backward_time=0.403, grad_norm=35.175, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.976e-04, train_time=1.493
+[gpub058:0/16] 2024-02-04 11:02:45,577 (trainer:737) INFO: 26epoch:train:9101-9200batch: iter_time=8.890e-05, forward_time=0.438, loss_ctc=44.924, loss_att=50.647, acc=0.746, loss=48.930, backward_time=0.418, grad_norm=36.608, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.976e-04, train_time=1.877
+[gpub058:0/16] 2024-02-04 11:05:00,065 (trainer:737) INFO: 26epoch:train:9201-9300batch: iter_time=7.732e-05, forward_time=0.289, loss_ctc=51.000, loss_att=44.685, acc=0.756, loss=46.579, backward_time=0.404, grad_norm=40.408, clip=100.000, loss_scale=8.931e+33, optim_step_time=0.093, optim0_lr0=1.976e-04, train_time=1.345
+[gpub058:0/16] 2024-02-04 11:08:13,806 (trainer:737) INFO: 26epoch:train:9301-9400batch: iter_time=0.002, forward_time=0.435, loss_ctc=48.071, loss_att=46.718, acc=0.756, loss=47.124, backward_time=0.437, grad_norm=37.340, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.116, optim0_lr0=1.976e-04, train_time=1.937
+[gpub058:0/16] 2024-02-04 11:10:32,577 (trainer:737) INFO: 26epoch:train:9401-9500batch: iter_time=8.478e-05, forward_time=0.290, loss_ctc=43.878, loss_att=47.981, acc=0.742, loss=46.750, backward_time=0.403, grad_norm=34.942, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.975e-04, train_time=1.388
+[gpub058:0/16] 2024-02-04 11:13:45,483 (trainer:737) INFO: 26epoch:train:9501-9600batch: iter_time=7.010e-04, forward_time=0.403, loss_ctc=44.700, loss_att=43.871, acc=0.738, loss=44.119, backward_time=0.436, grad_norm=38.774, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.115, optim0_lr0=1.975e-04, train_time=1.928
+[gpub058:0/16] 2024-02-04 11:16:16,412 (trainer:737) INFO: 26epoch:train:9601-9700batch: iter_time=8.279e-05, forward_time=0.291, loss_ctc=49.934, loss_att=48.650, acc=0.744, loss=49.035, backward_time=0.403, grad_norm=40.204, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.975e-04, train_time=1.509
+[gpub058:0/16] 2024-02-04 11:19:15,141 (trainer:737) INFO: 26epoch:train:9701-9800batch: iter_time=3.836e-04, forward_time=0.401, loss_ctc=47.636, loss_att=47.284, acc=0.735, loss=47.390, backward_time=0.429, grad_norm=42.165, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.975e-04, train_time=1.787
+[gpub058:0/16] 2024-02-04 11:21:54,197 (trainer:737) INFO: 26epoch:train:9801-9900batch: iter_time=8.589e-05, forward_time=0.292, loss_ctc=55.594, loss_att=63.199, acc=0.715, loss=60.918, backward_time=0.408, grad_norm=43.606, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.974e-04, train_time=1.589
+[gpub058:0/16] 2024-02-04 11:25:10,745 (trainer:737) INFO: 26epoch:train:9901-10000batch: iter_time=5.404e-04, forward_time=0.408, loss_ctc=52.535, loss_att=55.170, acc=0.731, loss=54.380, backward_time=0.502, grad_norm=37.883, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.974e-04, train_time=1.967
+[gpub058:0/16] 2024-02-04 11:25:30,909 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+
+gpub071:440843:440893 [2] proxy.cc:1059 NCCL WARN [Proxy Service] Poll failed with error 1
+[gpub058:0/16] 2024-02-04 11:25:50,166 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 11:25:53,808 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9f90aba00>)
+[gpub058:0/16] 2024-02-04 11:25:53,808 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-04 11:25:53,811 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 11:33:56,421 (trainer:737) INFO: 26epoch:train:10001-10100batch: iter_time=3.610, forward_time=0.287, loss_ctc=41.068, loss_att=39.059, acc=0.767, loss=39.662, backward_time=0.400, grad_norm=33.398, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.974e-04, train_time=5.257
+[gpub058:0/16] 2024-02-04 11:37:52,630 (trainer:737) INFO: 26epoch:train:10101-10200batch: iter_time=8.846e-05, forward_time=0.438, loss_ctc=45.272, loss_att=44.298, acc=0.745, loss=44.591, backward_time=0.422, grad_norm=38.510, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.110, optim0_lr0=1.974e-04, train_time=2.362
+[gpub058:0/16] 2024-02-04 11:40:47,786 (trainer:737) INFO: 26epoch:train:10201-10300batch: iter_time=8.720e-05, forward_time=0.288, loss_ctc=43.414, loss_att=40.328, acc=0.769, loss=41.253, backward_time=0.402, grad_norm=32.730, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.973e-04, train_time=1.746
+[gpub058:0/16] 2024-02-04 11:44:45,437 (trainer:737) INFO: 26epoch:train:10301-10400batch: iter_time=8.969e-05, forward_time=0.488, loss_ctc=46.786, loss_att=54.454, acc=0.744, loss=52.154, backward_time=0.433, grad_norm=37.583, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.973e-04, train_time=2.381
+[gpub058:0/16] 2024-02-04 11:46:23,252 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 11:48:04,037 (trainer:737) INFO: 26epoch:train:10401-10500batch: iter_time=9.001e-05, forward_time=0.287, loss_ctc=46.895, loss_att=42.134, acc=0.771, loss=43.562, backward_time=0.402, grad_norm=34.899, clip=100.000, loss_scale=7.972e+33, optim_step_time=0.092, optim0_lr0=1.973e-04, train_time=1.986
+[gpub058:0/16] 2024-02-04 11:50:50,498 (trainer:737) INFO: 26epoch:train:10501-10600batch: iter_time=8.496e-05, forward_time=0.289, loss_ctc=51.626, loss_att=45.361, acc=0.766, loss=47.241, backward_time=0.404, grad_norm=39.867, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.973e-04, train_time=1.664
+[gpub058:0/16] 2024-02-04 11:54:09,194 (trainer:737) INFO: 26epoch:train:10601-10700batch: iter_time=9.287e-05, forward_time=0.443, loss_ctc=47.440, loss_att=53.727, acc=0.739, loss=51.841, backward_time=0.428, grad_norm=39.866, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.972e-04, train_time=1.986
+[gpub058:0/16] 2024-02-04 11:56:56,785 (trainer:737) INFO: 26epoch:train:10701-10800batch: iter_time=9.830e-05, forward_time=0.289, loss_ctc=43.641, loss_att=46.782, acc=0.743, loss=45.840, backward_time=0.402, grad_norm=38.524, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.972e-04, train_time=1.676
+[gpub058:0/16] 2024-02-04 11:59:37,437 (trainer:737) INFO: 26epoch:train:10801-10900batch: iter_time=9.746e-05, forward_time=0.292, loss_ctc=45.859, loss_att=40.728, acc=0.764, loss=42.268, backward_time=0.404, grad_norm=39.566, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.972e-04, train_time=1.605
+[gpub058:0/16] 2024-02-04 12:03:04,246 (trainer:737) INFO: 26epoch:train:10901-11000batch: iter_time=3.958e-04, forward_time=0.394, loss_ctc=54.530, loss_att=57.975, acc=0.743, loss=56.941, backward_time=0.440, grad_norm=45.973, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.972e-04, train_time=2.069
+[gpub058:0/16] 2024-02-04 12:05:36,891 (trainer:737) INFO: 26epoch:train:11001-11100batch: iter_time=9.953e-05, forward_time=0.292, loss_ctc=49.606, loss_att=51.812, acc=0.738, loss=51.150, backward_time=0.406, grad_norm=43.944, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.971e-04, train_time=1.527
+[gpub058:0/16] 2024-02-04 12:08:39,643 (trainer:737) INFO: 26epoch:train:11101-11200batch: iter_time=9.933e-05, forward_time=0.495, loss_ctc=58.973, loss_att=62.861, acc=0.719, loss=61.695, backward_time=0.428, grad_norm=39.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.971e-04, train_time=1.827
+[gpub058:0/16] 2024-02-04 12:10:09,778 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-04 12:10:28,967 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 12:10:32,621 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9bb633640>)
+[gpub058:0/16] 2024-02-04 12:10:32,621 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-04 12:10:32,625 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 12:17:51,865 (trainer:737) INFO: 26epoch:train:11201-11300batch: iter_time=3.692, forward_time=0.470, loss_ctc=39.697, loss_att=41.973, acc=0.764, loss=41.290, backward_time=0.428, grad_norm=31.870, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.971e-04, train_time=5.521
+[gpub058:0/16] 2024-02-04 12:21:08,509 (trainer:737) INFO: 26epoch:train:11301-11400batch: iter_time=8.223e-05, forward_time=0.349, loss_ctc=45.424, loss_att=43.273, acc=0.756, loss=43.919, backward_time=0.501, grad_norm=36.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.971e-04, train_time=1.967
+[gpub058:0/16] 2024-02-04 12:24:50,642 (trainer:737) INFO: 26epoch:train:11401-11500batch: iter_time=9.410e-05, forward_time=0.434, loss_ctc=42.690, loss_att=40.427, acc=0.760, loss=41.106, backward_time=0.525, grad_norm=108.573, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.970e-04, train_time=2.221
+[gpub058:0/16] 2024-02-04 12:27:33,049 (trainer:737) INFO: 26epoch:train:11501-11600batch: iter_time=9.396e-05, forward_time=0.448, loss_ctc=45.825, loss_att=43.639, acc=0.766, loss=44.295, backward_time=0.431, grad_norm=34.745, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.970e-04, train_time=1.623
+[gpub058:0/16] 2024-02-04 12:31:01,751 (trainer:737) INFO: 26epoch:train:11601-11700batch: iter_time=9.045e-05, forward_time=0.375, loss_ctc=44.946, loss_att=49.731, acc=0.757, loss=48.295, backward_time=0.443, grad_norm=34.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.970e-04, train_time=2.086
+[gpub058:0/16] 2024-02-04 12:33:52,734 (trainer:737) INFO: 26epoch:train:11701-11800batch: iter_time=4.614e-04, forward_time=0.453, loss_ctc=50.916, loss_att=46.514, acc=0.756, loss=47.835, backward_time=0.416, grad_norm=39.679, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.970e-04, train_time=1.710
+[gpub058:0/16] 2024-02-04 12:36:50,548 (trainer:737) INFO: 26epoch:train:11801-11900batch: iter_time=8.521e-05, forward_time=0.347, loss_ctc=47.503, loss_att=47.182, acc=0.757, loss=47.278, backward_time=0.406, grad_norm=37.602, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.969e-04, train_time=1.779
+[gpub058:0/16] 2024-02-04 12:40:14,931 (trainer:737) INFO: 26epoch:train:11901-12000batch: iter_time=0.002, forward_time=0.419, loss_ctc=43.798, loss_att=47.755, acc=0.756, loss=46.568, backward_time=0.445, grad_norm=33.749, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.969e-04, train_time=2.041
+[gpub058:0/16] 2024-02-04 12:43:28,644 (trainer:737) INFO: 26epoch:train:12001-12100batch: iter_time=9.048e-04, forward_time=0.475, loss_ctc=44.265, loss_att=44.139, acc=0.752, loss=44.177, backward_time=0.430, grad_norm=40.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.969e-04, train_time=1.939
+[gpub058:0/16] 2024-02-04 12:46:49,595 (trainer:737) INFO: 26epoch:train:12101-12200batch: iter_time=3.534e-04, forward_time=0.462, loss_ctc=49.764, loss_att=47.549, acc=0.752, loss=48.213, backward_time=0.424, grad_norm=40.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.969e-04, train_time=2.009
+[gpub058:0/16] 2024-02-04 12:49:47,693 (trainer:737) INFO: 26epoch:train:12201-12300batch: iter_time=1.016e-04, forward_time=0.289, loss_ctc=48.076, loss_att=47.581, acc=0.751, loss=47.729, backward_time=0.408, grad_norm=39.102, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.968e-04, train_time=1.782
+[gpub058:0/16] 2024-02-04 12:53:00,929 (trainer:737) INFO: 26epoch:train:12301-12400batch: iter_time=0.002, forward_time=0.389, loss_ctc=55.062, loss_att=65.017, acc=0.716, loss=62.030, backward_time=0.488, grad_norm=45.061, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.119, optim0_lr0=1.968e-04, train_time=1.931
+[gpub058:0/16] 2024-02-04 12:56:08,930 (trainer:737) INFO: 26epoch:train:12401-12500batch: iter_time=3.487e-04, forward_time=0.424, loss_ctc=52.524, loss_att=54.308, acc=0.745, loss=53.773, backward_time=0.443, grad_norm=35.269, clip=100.000, loss_scale=7.581e+33, optim_step_time=0.104, optim0_lr0=1.968e-04, train_time=1.881
+[gpub058:0/16] 2024-02-04 12:56:28,981 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-04 12:56:48,777 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 12:56:52,456 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9609cb9a0>)
+[gpub058:0/16] 2024-02-04 12:56:52,456 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-04 12:56:52,459 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 13:05:03,917 (trainer:737) INFO: 26epoch:train:12501-12600batch: iter_time=3.418, forward_time=0.458, loss_ctc=41.206, loss_att=39.964, acc=0.764, loss=40.336, backward_time=0.429, grad_norm=31.844, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.968e-04, train_time=5.350
+[gpub058:0/16] 2024-02-04 13:07:52,742 (trainer:737) INFO: 26epoch:train:12601-12700batch: iter_time=8.748e-05, forward_time=0.289, loss_ctc=44.826, loss_att=44.246, acc=0.737, loss=44.420, backward_time=0.404, grad_norm=38.417, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.967e-04, train_time=1.688
+[gpub058:0/16] 2024-02-04 13:10:20,652 (trainer:737) INFO: 26epoch:train:12701-12800batch: iter_time=8.733e-05, forward_time=0.317, loss_ctc=43.501, loss_att=40.957, acc=0.761, loss=41.720, backward_time=0.402, grad_norm=33.293, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.967e-04, train_time=1.479
+[gpub058:0/16] 2024-02-04 13:13:30,706 (trainer:737) INFO: 26epoch:train:12801-12900batch: iter_time=8.504e-05, forward_time=0.430, loss_ctc=46.835, loss_att=53.675, acc=0.743, loss=51.623, backward_time=0.472, grad_norm=37.730, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.967e-04, train_time=1.900
+[gpub058:0/16] 2024-02-04 13:15:54,150 (trainer:737) INFO: 26epoch:train:12901-13000batch: iter_time=9.035e-05, forward_time=0.289, loss_ctc=45.642, loss_att=41.857, acc=0.761, loss=42.992, backward_time=0.402, grad_norm=34.741, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.967e-04, train_time=1.434
+[gpub058:0/16] 2024-02-04 13:19:00,010 (trainer:737) INFO: 26epoch:train:13001-13100batch: iter_time=2.798e-04, forward_time=0.309, loss_ctc=49.799, loss_att=43.380, acc=0.765, loss=45.305, backward_time=0.409, grad_norm=43.473, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.966e-04, train_time=1.858
+[gpub058:0/16] 2024-02-04 13:22:10,004 (trainer:737) INFO: 26epoch:train:13101-13200batch: iter_time=8.700e-05, forward_time=0.376, loss_ctc=46.515, loss_att=53.829, acc=0.730, loss=51.635, backward_time=0.453, grad_norm=38.416, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.966e-04, train_time=1.900
+[gpub058:0/16] 2024-02-04 13:24:41,152 (trainer:737) INFO: 26epoch:train:13201-13300batch: iter_time=9.233e-05, forward_time=0.287, loss_ctc=43.961, loss_att=45.055, acc=0.735, loss=44.727, backward_time=0.399, grad_norm=35.874, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.966e-04, train_time=1.511
+[gpub058:0/16] 2024-02-04 13:27:55,078 (trainer:737) INFO: 26epoch:train:13301-13400batch: iter_time=9.250e-05, forward_time=0.355, loss_ctc=45.674, loss_att=40.447, acc=0.758, loss=42.015, backward_time=0.409, grad_norm=38.940, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.966e-04, train_time=1.938
+[gpub058:0/16] 2024-02-04 13:31:05,068 (trainer:737) INFO: 26epoch:train:13401-13500batch: iter_time=9.323e-05, forward_time=0.395, loss_ctc=54.087, loss_att=56.797, acc=0.731, loss=55.984, backward_time=0.430, grad_norm=41.565, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.965e-04, train_time=1.900
+[gpub058:0/16] 2024-02-04 13:33:59,500 (trainer:737) INFO: 26epoch:train:13501-13600batch: iter_time=9.900e-05, forward_time=0.290, loss_ctc=48.855, loss_att=50.531, acc=0.730, loss=50.028, backward_time=0.403, grad_norm=45.967, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.965e-04, train_time=1.744
+[gpub058:0/16] 2024-02-04 13:36:57,829 (trainer:737) INFO: 26epoch:train:13601-13700batch: iter_time=9.453e-05, forward_time=0.414, loss_ctc=58.355, loss_att=61.635, acc=0.719, loss=60.651, backward_time=0.448, grad_norm=38.348, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.965e-04, train_time=1.783
+[gpub058:0/16] 2024-02-04 13:38:29,023 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-04 13:38:48,150 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 13:38:51,808 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe99eb1f6d0>)
+[gpub058:0/16] 2024-02-04 13:38:51,808 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-04 13:38:51,811 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 13:45:38,362 (trainer:737) INFO: 26epoch:train:13701-13800batch: iter_time=3.455, forward_time=0.314, loss_ctc=39.669, loss_att=43.416, acc=0.756, loss=42.292, backward_time=0.416, grad_norm=34.396, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.964e-04, train_time=5.205
+[gpub058:0/16] 2024-02-04 13:48:23,639 (trainer:737) INFO: 26epoch:train:13801-13900batch: iter_time=8.475e-05, forward_time=0.430, loss_ctc=44.566, loss_att=43.738, acc=0.755, loss=43.986, backward_time=0.463, grad_norm=35.413, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.964e-04, train_time=1.652
+[gpub058:0/16] 2024-02-04 13:51:03,322 (trainer:737) INFO: 26epoch:train:13901-14000batch: iter_time=8.321e-05, forward_time=0.390, loss_ctc=42.465, loss_att=40.525, acc=0.758, loss=41.107, backward_time=0.416, grad_norm=34.252, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.964e-04, train_time=1.597
+[gpub058:0/16] 2024-02-04 13:54:04,857 (trainer:737) INFO: 26epoch:train:14001-14100batch: iter_time=9.713e-05, forward_time=0.349, loss_ctc=46.165, loss_att=43.796, acc=0.766, loss=44.507, backward_time=0.412, grad_norm=36.419, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.964e-04, train_time=1.815
+[gpub058:0/16] 2024-02-04 13:56:44,186 (trainer:737) INFO: 26epoch:train:14101-14200batch: iter_time=8.735e-05, forward_time=0.464, loss_ctc=45.257, loss_att=50.531, acc=0.755, loss=48.949, backward_time=0.440, grad_norm=37.412, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.963e-04, train_time=1.593
+[gpub058:0/16] 2024-02-04 13:59:39,066 (trainer:737) INFO: 26epoch:train:14201-14300batch: iter_time=9.620e-05, forward_time=0.294, loss_ctc=51.236, loss_att=45.839, acc=0.758, loss=47.458, backward_time=0.406, grad_norm=40.844, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.963e-04, train_time=1.749
+[gpub058:0/16] 2024-02-04 14:02:26,156 (trainer:737) INFO: 26epoch:train:14301-14400batch: iter_time=9.146e-04, forward_time=0.417, loss_ctc=47.442, loss_att=46.923, acc=0.758, loss=47.078, backward_time=0.446, grad_norm=37.061, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.963e-04, train_time=1.669
+[gpub058:0/16] 2024-02-04 14:05:57,279 (trainer:737) INFO: 26epoch:train:14401-14500batch: iter_time=9.090e-04, forward_time=0.499, loss_ctc=43.850, loss_att=47.333, acc=0.757, loss=46.288, backward_time=0.497, grad_norm=35.169, clip=100.000, loss_scale=1.516e+34, optim_step_time=0.113, optim0_lr0=1.963e-04, train_time=2.112
+[gpub058:0/16] 2024-02-04 14:08:09,424 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 14:08:44,161 (trainer:737) INFO: 26epoch:train:14501-14600batch: iter_time=8.479e-05, forward_time=0.432, loss_ctc=44.176, loss_att=44.531, acc=0.751, loss=44.425, backward_time=0.430, grad_norm=38.504, clip=100.000, loss_scale=1.951e+34, optim_step_time=0.095, optim0_lr0=1.962e-04, train_time=1.668
+[gpub058:0/16] 2024-02-04 14:11:59,611 (trainer:737) INFO: 26epoch:train:14601-14700batch: iter_time=3.506e-04, forward_time=0.443, loss_ctc=49.891, loss_att=48.669, acc=0.749, loss=49.036, backward_time=0.433, grad_norm=43.798, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.962e-04, train_time=1.955
+[gpub058:0/16] 2024-02-04 14:12:40,061 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 14:14:38,321 (trainer:737) INFO: 26epoch:train:14701-14800batch: iter_time=9.822e-05, forward_time=0.289, loss_ctc=47.346, loss_att=48.141, acc=0.749, loss=47.902, backward_time=0.402, grad_norm=38.768, clip=100.000, loss_scale=6.399e+33, optim_step_time=0.092, optim0_lr0=1.962e-04, train_time=1.588
+[gpub058:0/16] 2024-02-04 14:17:45,381 (trainer:737) INFO: 26epoch:train:14801-14900batch: iter_time=7.922e-04, forward_time=0.474, loss_ctc=55.140, loss_att=65.623, acc=0.715, loss=62.478, backward_time=0.428, grad_norm=43.764, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.962e-04, train_time=1.869
+[gpub058:0/16] 2024-02-04 14:21:06,550 (trainer:737) INFO: 26epoch:train:14901-15000batch: iter_time=3.854e-04, forward_time=0.478, loss_ctc=51.927, loss_att=54.100, acc=0.744, loss=53.448, backward_time=0.422, grad_norm=40.179, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.961e-04, train_time=2.011
+[gpub058:0/16] 2024-02-04 15:00:16,689 (trainer:343) INFO: 26epoch results: [train] iter_time=0.273, forward_time=0.350, loss_ctc=48.034, loss_att=48.133, acc=0.746, loss=48.103, backward_time=0.422, grad_norm=38.932, clip=100.000, loss_scale=6.550e+33, optim_step_time=0.096, optim0_lr0=1.981e-04, train_time=1.986, time=8 hours, 17 minutes and 4.67 seconds, total_count=420000, gpu_max_cached_mem_GB=43.281, [valid] loss_ctc=37.613, cer_ctc=0.192, loss_att=39.718, acc=0.669, cer=0.324, wer=0.996, loss=39.086, time=38 minutes and 46.09 seconds, total_count=130788, gpu_max_cached_mem_GB=43.281
+[gpub058:0/16] 2024-02-04 15:00:27,267 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub058:0/16] 2024-02-04 15:00:27,338 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/18epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/21epoch.pth
+[gpub058:0/16] 2024-02-04 15:00:27,338 (trainer:272) INFO: 27/45epoch started. Estimated time to finish: 6 days, 17 hours and 20 minutes
+[gpub058:0/16] 2024-02-04 15:00:27,347 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-04 15:00:45,587 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 15:00:49,047 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1901af460>)
+[gpub058:0/16] 2024-02-04 15:00:49,047 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-04 15:00:49,050 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 15:07:51,988 (trainer:737) INFO: 27epoch:train:1-100batch: iter_time=2.917, forward_time=0.385, loss_ctc=54.767, loss_att=57.365, acc=0.711, loss=56.586, backward_time=0.414, grad_norm=43.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.961e-04, train_time=4.446
+[gpub058:0/16] 2024-02-04 15:10:32,064 (trainer:737) INFO: 27epoch:train:101-200batch: iter_time=8.308e-05, forward_time=0.292, loss_ctc=51.994, loss_att=60.561, acc=0.727, loss=57.991, backward_time=0.406, grad_norm=42.761, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.961e-04, train_time=1.600
+[gpub058:0/16] 2024-02-04 15:13:03,816 (trainer:737) INFO: 27epoch:train:201-300batch: iter_time=7.986e-05, forward_time=0.372, loss_ctc=49.813, loss_att=44.933, acc=0.762, loss=46.397, backward_time=0.454, grad_norm=37.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.961e-04, train_time=1.518
+[gpub058:0/16] 2024-02-04 15:16:00,077 (trainer:737) INFO: 27epoch:train:301-400batch: iter_time=9.039e-05, forward_time=0.290, loss_ctc=47.558, loss_att=43.511, acc=0.775, loss=44.725, backward_time=0.402, grad_norm=36.157, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.960e-04, train_time=1.762
+[gpub058:0/16] 2024-02-04 15:18:52,534 (trainer:737) INFO: 27epoch:train:401-500batch: iter_time=8.665e-05, forward_time=0.376, loss_ctc=46.229, loss_att=43.130, acc=0.761, loss=44.060, backward_time=0.434, grad_norm=35.222, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.960e-04, train_time=1.724
+[gpub058:0/16] 2024-02-04 15:21:27,591 (trainer:737) INFO: 27epoch:train:501-600batch: iter_time=9.145e-05, forward_time=0.329, loss_ctc=42.137, loss_att=36.115, acc=0.761, loss=37.921, backward_time=0.421, grad_norm=33.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.960e-04, train_time=1.551
+[gpub058:0/16] 2024-02-04 15:24:05,110 (trainer:737) INFO: 27epoch:train:601-700batch: iter_time=8.252e-05, forward_time=0.343, loss_ctc=48.946, loss_att=40.719, acc=0.751, loss=43.187, backward_time=0.428, grad_norm=38.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.960e-04, train_time=1.575
+[gpub058:0/16] 2024-02-04 15:26:41,817 (trainer:737) INFO: 27epoch:train:701-800batch: iter_time=9.203e-05, forward_time=0.311, loss_ctc=50.154, loss_att=49.336, acc=0.744, loss=49.582, backward_time=0.407, grad_norm=38.778, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.959e-04, train_time=1.566
+[gpub058:0/16] 2024-02-04 15:29:23,788 (trainer:737) INFO: 27epoch:train:801-900batch: iter_time=1.968e-04, forward_time=0.306, loss_ctc=40.491, loss_att=42.932, acc=0.749, loss=42.200, backward_time=0.444, grad_norm=35.619, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.959e-04, train_time=1.620
+[gpub058:0/16] 2024-02-04 15:32:05,491 (trainer:737) INFO: 27epoch:train:901-1000batch: iter_time=8.553e-05, forward_time=0.335, loss_ctc=51.601, loss_att=50.203, acc=0.721, loss=50.622, backward_time=0.425, grad_norm=42.651, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.959e-04, train_time=1.617
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2932506.0 ON gpub058 CANCELLED AT 2024-02-04T15:34:22 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.11.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.11.log
new file mode 100644
index 0000000000000000000000000000000000000000..9a1c8649c357c02f91dede400feeb1dc9f569fe2
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.11.log
@@ -0,0 +1,2281 @@
+# Running on gpub022.delta.ncsa.illinois.edu
+# Started at Thu Feb 1 10:53:31 CST 2024
+# SLURMD_NODENAME=gpub022
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2913929
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1706979193
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2913929
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[022,058,062,085]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1706806393
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[022,058,062,085]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=1741645
+# SLURM_TOPOLOGY_ADDR=ss00.ss10.gpub022
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_faee9aed-7ec4-484d-9dae-05f165fc470e 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_faee9aed-7ec4-484d-9dae-05f165fc470e
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessi_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_faee9aed-7ec4-484d-9dae-05f165fc470e
+ng_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_faee9aed-7ec4-484d-9dae-05f165fc470e
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_faee9aed-7ec4-484d-9dae-05f165fc470e
+[gpub022:0/16] 2024-02-01 10:57:21,085 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub022:0/16] 2024-02-01 10:57:21,202 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub022:0/16] 2024-02-01 10:57:21,264 (s2t:464) INFO: Vocabulary size: 50002
+[gpub022:0/16] 2024-02-01 10:57:30,965 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub022:0/16] 2024-02-01 10:57:30,971 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub022:0/16] 2024-02-01 10:57:30,971 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub022:0/16] 2024-02-01 10:57:30,971 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub022:0/16] 2024-02-01 10:57:30,985 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub022:0/16] 2024-02-01 10:57:36,611 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 10:57:37,536 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe9666e6380>)
+[gpub022:0/16] 2024-02-01 10:57:37,536 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub022:0/16] 2024-02-01 10:57:37,537 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 10:57:49,866 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub022:1741720:1741720 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.122<0>
+gpub022:1741720:1741720 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub022:1741720:1741720 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub022:0/16] 2024-02-01 10:57:55,128 (trainer:284) INFO: 18/45epoch started
+[gpub022:0/16] 2024-02-01 10:57:55,173 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub022:0/16] 2024-02-01 10:58:13,046 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 10:58:16,468 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe77ef914b0>)
+[gpub022:0/16] 2024-02-01 10:58:16,468 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub022:0/16] 2024-02-01 10:58:16,471 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub062:1370885:1370885 [2] NCCL INFO cudaDriverVersion 12020
+gpub062:1370885:1370885 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1370885:1370885 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1370885:1370939 [2] NCCL INFO NET/IB : No device found.
+gpub062:1370885:1370939 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1370885:1370939 [2] NCCL INFO Using network Socket
+gpub062:1370885:1370939 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub062:1370885:1370939 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub062:1370885:1370939 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub062:1370885:1370939 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub062:1370885:1370939 [2] NCCL INFO Connected all rings
+gpub062:1370885:1370939 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub062:1370885:1370939 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub062:1370885:1370939 [2] NCCL INFO Connected all trees
+gpub062:1370885:1370939 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1370885:1370939 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1370885:1370939 [2] NCCL INFO comm 0x1ac32740 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub085:2091268:2091268 [1] NCCL INFO cudaDriverVersion 12020
+gpub085:2091268:2091268 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.185<0>
+gpub085:2091268:2091268 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub085:2091268:2091322 [1] NCCL INFO NET/IB : No device found.
+gpub085:2091268:2091322 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.185<0> [1]hsn0:141.142.145.185<0> [2]eth0:fe80::1703:8949:2817:7411%eth0<0>
+gpub085:2091268:2091322 [1] NCCL INFO Using network Socket
+gpub085:2091268:2091322 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub085:2091268:2091322 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub085:2091268:2091322 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub085:2091268:2091322 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub085:2091268:2091322 [1] NCCL INFO Connected all rings
+gpub085:2091268:2091322 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub085:2091268:2091322 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub085:2091268:2091322 [1] NCCL INFO Connected all trees
+gpub085:2091268:2091322 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub085:2091268:2091322 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub085:2091268:2091322 [1] NCCL INFO comm 0x135025c0 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub085:2091269:2091269 [2] NCCL INFO cudaDriverVersion 12020
+gpub085:2091269:2091269 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.185<0>
+gpub085:2091269:2091269 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub085:2091269:2091323 [2] NCCL INFO NET/IB : No device found.
+gpub085:2091269:2091323 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.185<0> [1]hsn0:141.142.145.185<0> [2]eth0:fe80::1703:8949:2817:7411%eth0<0>
+gpub085:2091269:2091323 [2] NCCL INFO Using network Socket
+gpub085:2091269:2091323 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub085:2091269:2091323 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub085:2091269:2091323 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub085:2091269:2091323 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub085:2091269:2091323 [2] NCCL INFO Connected all rings
+gpub085:2091269:2091323 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub085:2091269:2091323 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub085:2091269:2091323 [2] NCCL INFO Connected all trees
+gpub085:2091269:2091323 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub085:2091269:2091323 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub085:2091269:2091323 [2] NCCL INFO comm 0x1312b960 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub085:2091270:2091270 [3] NCCL INFO cudaDriverVersion 12020
+gpub085:2091270:2091270 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.185<0>
+gpub085:2091270:2091270 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub085:2091270:2091324 [3] NCCL INFO NET/IB : No device found.
+gpub085:2091270:2091324 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.185<0> [1]hsn0:141.142.145.185<0> [2]eth0:fe80::1703:8949:2817:7411%eth0<0>
+gpub085:2091270:2091324 [3] NCCL INFO Using network Socket
+gpub085:2091270:2091324 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub085:2091270:2091324 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub085:2091270:2091324 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub085:2091270:2091324 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub085:2091270:2091324 [3] NCCL INFO Connected all rings
+gpub085:2091270:2091324 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub085:2091270:2091324 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub085:2091270:2091324 [3] NCCL INFO Connected all trees
+gpub085:2091270:2091324 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub085:2091270:2091324 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub085:2091270:2091324 [3] NCCL INFO comm 0x16005160 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub062:1370886:1370886 [3] NCCL INFO cudaDriverVersion 12020
+gpub062:1370886:1370886 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1370886:1370886 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1370886:1370941 [3] NCCL INFO NET/IB : No device found.
+gpub062:1370886:1370941 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1370886:1370941 [3] NCCL INFO Using network Socket
+gpub062:1370886:1370941 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub062:1370886:1370941 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub062:1370886:1370941 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1370886:1370941 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1370886:1370941 [3] NCCL INFO Connected all rings
+gpub062:1370886:1370941 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub062:1370886:1370941 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub062:1370886:1370941 [3] NCCL INFO Connected all trees
+gpub062:1370886:1370941 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1370886:1370941 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1370886:1370941 [3] NCCL INFO comm 0x17dedec0 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub062:1370884:1370884 [1] NCCL INFO cudaDriverVersion 12020
+gpub062:1370884:1370884 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1370884:1370884 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1370884:1370942 [1] NCCL INFO NET/IB : No device found.
+gpub062:1370884:1370942 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1370884:1370942 [1] NCCL INFO Using network Socket
+gpub062:1370884:1370942 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub062:1370884:1370942 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub062:1370884:1370942 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub062:1370884:1370942 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub062:1370884:1370942 [1] NCCL INFO Connected all rings
+gpub062:1370884:1370942 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/Socket/1
+gpub062:1370884:1370942 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/Socket/1
+gpub062:1370884:1370942 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub062:1370884:1370942 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub062:1370884:1370942 [1] NCCL INFO Connected all trees
+gpub062:1370884:1370942 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1370884:1370942 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1370884:1370942 [1] NCCL INFO comm 0x18b7c9b0 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub085:2091267:2091267 [0] NCCL INFO cudaDriverVersion 12020
+gpub085:2091267:2091267 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.185<0>
+gpub085:2091267:2091267 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub085:2091267:2091325 [0] NCCL INFO NET/IB : No device found.
+gpub085:2091267:2091325 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.185<0> [1]hsn0:141.142.145.185<0> [2]eth0:fe80::1703:8949:2817:7411%eth0<0>
+gpub085:2091267:2091325 [0] NCCL INFO Using network Socket
+gpub085:2091267:2091325 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub085:2091267:2091325 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub085:2091267:2091325 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub085:2091267:2091325 [0] NCCL INFO Connected all rings
+gpub085:2091267:2091325 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/Socket/1
+gpub085:2091267:2091325 [0] NCCL INFO Connected all trees
+gpub085:2091267:2091325 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub085:2091267:2091325 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub085:2091267:2091325 [0] NCCL INFO comm 0x155be520 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub062:1370883:1370883 [0] NCCL INFO cudaDriverVersion 12020
+gpub062:1370883:1370883 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.162<0>
+gpub062:1370883:1370883 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub062:1370883:1370940 [0] NCCL INFO NET/IB : No device found.
+gpub062:1370883:1370940 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.162<0> [1]hsn0:141.142.145.162<0> [2]eth0:fe80::221:e9:de39:a135%eth0<0>
+gpub062:1370883:1370940 [0] NCCL INFO Using network Socket
+gpub062:1370883:1370940 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub062:1370883:1370940 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub062:1370883:1370940 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub062:1370883:1370940 [0] NCCL INFO Connected all rings
+gpub062:1370883:1370940 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/Socket/1
+gpub062:1370883:1370940 [0] NCCL INFO Connected all trees
+gpub062:1370883:1370940 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub062:1370883:1370940 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub062:1370883:1370940 [0] NCCL INFO comm 0x16a32520 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub058:1925470:1925470 [2] NCCL INFO cudaDriverVersion 12020
+gpub058:1925470:1925470 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:1925470:1925470 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:1925470:1925523 [2] NCCL INFO NET/IB : No device found.
+gpub058:1925470:1925523 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:1925470:1925523 [2] NCCL INFO Using network Socket
+gpub058:1925470:1925523 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub058:1925470:1925523 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub058:1925470:1925523 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub058:1925470:1925523 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub058:1925470:1925523 [2] NCCL INFO Connected all rings
+gpub058:1925470:1925523 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub058:1925470:1925523 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub058:1925470:1925523 [2] NCCL INFO Connected all trees
+gpub058:1925470:1925523 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:1925470:1925523 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:1925470:1925523 [2] NCCL INFO comm 0x1cf649b0 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub058:1925471:1925471 [3] NCCL INFO cudaDriverVersion 12020
+gpub058:1925471:1925471 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:1925471:1925471 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:1925471:1925524 [3] NCCL INFO NET/IB : No device found.
+gpub058:1925471:1925524 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:1925471:1925524 [3] NCCL INFO Using network Socket
+gpub058:1925471:1925524 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub058:1925471:1925524 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub058:1925471:1925524 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub058:1925471:1925524 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub058:1925471:1925524 [3] NCCL INFO Connected all rings
+gpub058:1925471:1925524 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub058:1925471:1925524 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub058:1925471:1925524 [3] NCCL INFO Connected all trees
+gpub058:1925471:1925524 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:1925471:1925524 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:1925471:1925524 [3] NCCL INFO comm 0x1a09fc10 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub022:1741722:1741722 [2] NCCL INFO cudaDriverVersion 12020
+gpub022:1741722:1741722 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.122<0>
+gpub022:1741722:1741722 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub022:1741722:1741773 [2] NCCL INFO NET/IB : No device found.
+gpub022:1741722:1741773 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.122<0> [1]hsn0:141.142.145.122<0>
+gpub022:1741722:1741773 [2] NCCL INFO Using network Socket
+gpub022:1741722:1741773 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub022:1741722:1741773 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub022:1741722:1741773 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub022:1741722:1741773 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub022:1741722:1741773 [2] NCCL INFO Connected all rings
+gpub022:1741722:1741773 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub058:1925468:1925468 [0] NCCL INFO cudaDriverVersion 12020
+gpub058:1925468:1925468 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:1925468:1925468 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:1925468:1925525 [0] NCCL INFO NET/IB : No device found.
+gpub058:1925468:1925525 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:1925468:1925525 [0] NCCL INFO Using network Socket
+gpub058:1925468:1925525 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub058:1925468:1925525 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub058:1925468:1925525 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub022:1741722:1741773 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub022:1741722:1741773 [2] NCCL INFO Connected all trees
+gpub022:1741722:1741773 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub022:1741722:1741773 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub022:1741722:1741773 [2] NCCL INFO comm 0x15a46340 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub058:1925468:1925525 [0] NCCL INFO Connected all rings
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/Socket/1
+gpub058:1925468:1925525 [0] NCCL INFO Connected all trees
+gpub058:1925468:1925525 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:1925468:1925525 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:1925468:1925525 [0] NCCL INFO comm 0x1822ca70 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub058:1925469:1925469 [1] NCCL INFO cudaDriverVersion 12020
+gpub058:1925469:1925469 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:1925469:1925469 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:1925469:1925526 [1] NCCL INFO NET/IB : No device found.
+gpub058:1925469:1925526 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:1925469:1925526 [1] NCCL INFO Using network Socket
+gpub058:1925469:1925526 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub058:1925469:1925526 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub058:1925469:1925526 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub058:1925469:1925526 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub058:1925469:1925526 [1] NCCL INFO Connected all rings
+gpub058:1925469:1925526 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/Socket/1
+gpub058:1925469:1925526 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/Socket/1
+gpub058:1925469:1925526 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub058:1925469:1925526 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub058:1925469:1925526 [1] NCCL INFO Connected all trees
+gpub058:1925469:1925526 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:1925469:1925526 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:1925469:1925526 [1] NCCL INFO comm 0x18578770 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub022:1741720:1741770 [0] NCCL INFO NET/IB : No device found.
+gpub022:1741720:1741770 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.122<0> [1]hsn0:141.142.145.122<0>
+gpub022:1741720:1741770 [0] NCCL INFO Using network Socket
+gpub022:1741720:1741770 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub022:1741720:1741770 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub022:1741720:1741770 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub022:1741720:1741770 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub022:1741720:1741770 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub022:1741720:1741770 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub022:1741720:1741770 [0] NCCL INFO Connected all rings
+gpub022:1741720:1741770 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub022:1741720:1741770 [0] NCCL INFO Connected all trees
+gpub022:1741720:1741770 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub022:1741720:1741770 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub022:1741720:1741770 [0] NCCL INFO comm 0x14a37600 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub022:1741721:1741721 [1] NCCL INFO cudaDriverVersion 12020
+gpub022:1741721:1741721 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.122<0>
+gpub022:1741721:1741721 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub022:1741721:1741772 [1] NCCL INFO NET/IB : No device found.
+gpub022:1741721:1741772 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.122<0> [1]hsn0:141.142.145.122<0>
+gpub022:1741721:1741772 [1] NCCL INFO Using network Socket
+gpub022:1741721:1741772 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub022:1741721:1741772 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub022:1741721:1741772 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub022:1741721:1741772 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub022:1741721:1741772 [1] NCCL INFO Connected all rings
+gpub022:1741721:1741772 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub022:1741721:1741772 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub022:1741721:1741772 [1] NCCL INFO Connected all trees
+gpub022:1741721:1741772 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub022:1741721:1741772 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub022:1741721:1741772 [1] NCCL INFO comm 0x13440400 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub022:1741723:1741723 [3] NCCL INFO cudaDriverVersion 12020
+gpub022:1741723:1741723 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.122<0>
+gpub022:1741723:1741723 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub022:1741723:1741771 [3] NCCL INFO NET/IB : No device found.
+gpub022:1741723:1741771 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.122<0> [1]hsn0:141.142.145.122<0>
+gpub022:1741723:1741771 [3] NCCL INFO Using network Socket
+gpub022:1741723:1741771 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub022:1741723:1741771 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub022:1741723:1741771 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub022:1741723:1741771 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub022:1741723:1741771 [3] NCCL INFO Connected all rings
+gpub022:1741723:1741771 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub022:1741723:1741771 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub022:1741723:1741771 [3] NCCL INFO Connected all trees
+gpub022:1741723:1741771 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub022:1741723:1741771 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub022:1741723:1741771 [3] NCCL INFO comm 0xf49a900 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+[gpub022:0/16] 2024-02-01 11:07:00,776 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub022:0/16] 2024-02-01 11:09:51,496 (trainer:737) INFO: 18epoch:train:1-100batch: iter_time=4.046, forward_time=0.519, loss_ctc=46.466, loss_att=47.483, acc=0.742, loss=47.178, backward_time=0.455, grad_norm=31.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.425e-04, train_time=7.147
+[gpub022:0/16] 2024-02-01 11:12:45,266 (trainer:737) INFO: 18epoch:train:101-200batch: iter_time=4.226e-04, forward_time=0.415, loss_ctc=54.999, loss_att=52.041, acc=0.726, loss=52.928, backward_time=0.455, grad_norm=33.639, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.425e-04, train_time=1.753
+[gpub022:0/16] 2024-02-01 11:15:46,586 (trainer:737) INFO: 18epoch:train:201-300batch: iter_time=0.001, forward_time=0.496, loss_ctc=45.115, loss_att=40.852, acc=0.730, loss=42.131, backward_time=0.488, grad_norm=31.469, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=2.424e-04, train_time=1.813
+[gpub022:0/16] 2024-02-01 11:18:55,440 (trainer:737) INFO: 18epoch:train:301-400batch: iter_time=1.043e-04, forward_time=0.474, loss_ctc=54.709, loss_att=51.740, acc=0.727, loss=52.630, backward_time=0.458, grad_norm=35.145, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.424e-04, train_time=1.888
+[gpub022:0/16] 2024-02-01 11:21:48,449 (trainer:737) INFO: 18epoch:train:401-500batch: iter_time=6.817e-04, forward_time=0.523, loss_ctc=63.351, loss_att=53.063, acc=0.731, loss=56.149, backward_time=0.484, grad_norm=57.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.423e-04, train_time=1.731
+[gpub022:0/16] 2024-02-01 11:24:38,360 (trainer:737) INFO: 18epoch:train:501-600batch: iter_time=3.971e-04, forward_time=0.445, loss_ctc=51.655, loss_att=56.947, acc=0.716, loss=55.359, backward_time=0.446, grad_norm=35.599, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.423e-04, train_time=1.699
+[gpub022:0/16] 2024-02-01 11:27:29,655 (trainer:737) INFO: 18epoch:train:601-700batch: iter_time=4.565e-04, forward_time=0.546, loss_ctc=62.433, loss_att=51.183, acc=0.734, loss=54.558, backward_time=0.462, grad_norm=39.300, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.422e-04, train_time=1.712
+[gpub022:0/16] 2024-02-01 11:30:17,698 (trainer:737) INFO: 18epoch:train:701-800batch: iter_time=2.025e-04, forward_time=0.446, loss_ctc=50.113, loss_att=49.351, acc=0.735, loss=49.580, backward_time=0.433, grad_norm=33.964, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.422e-04, train_time=1.681
+[gpub022:0/16] 2024-02-01 11:33:02,286 (trainer:737) INFO: 18epoch:train:801-900batch: iter_time=5.816e-04, forward_time=0.391, loss_ctc=59.098, loss_att=52.623, acc=0.734, loss=54.566, backward_time=0.477, grad_norm=38.176, clip=100.000, loss_scale=9.242e+33, optim_step_time=0.103, optim0_lr0=2.421e-04, train_time=1.646
+[gpub022:0/16] 2024-02-01 11:35:43,414 (trainer:737) INFO: 18epoch:train:901-1000batch: iter_time=4.848e-04, forward_time=0.365, loss_ctc=50.237, loss_att=49.477, acc=0.745, loss=49.705, backward_time=0.493, grad_norm=32.129, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.421e-04, train_time=1.611
+[gpub022:0/16] 2024-02-01 11:38:43,681 (trainer:737) INFO: 18epoch:train:1001-1100batch: iter_time=4.796e-04, forward_time=0.403, loss_ctc=56.200, loss_att=55.943, acc=0.729, loss=56.020, backward_time=0.464, grad_norm=36.639, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.420e-04, train_time=1.801
+[gpub022:0/16] 2024-02-01 11:41:37,382 (trainer:737) INFO: 18epoch:train:1101-1200batch: iter_time=2.010e-04, forward_time=0.418, loss_ctc=48.253, loss_att=44.765, acc=0.719, loss=45.811, backward_time=0.448, grad_norm=34.894, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.420e-04, train_time=1.738
+[gpub022:0/16] 2024-02-01 11:43:11,742 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub022:0/16] 2024-02-01 11:43:30,286 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 11:43:33,923 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe204e6f7c0>)
+[gpub022:0/16] 2024-02-01 11:43:33,923 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub022:0/16] 2024-02-01 11:43:33,926 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 11:53:24,853 (trainer:737) INFO: 18epoch:train:1201-1300batch: iter_time=3.383, forward_time=0.433, loss_ctc=45.716, loss_att=47.383, acc=0.727, loss=46.883, backward_time=0.451, grad_norm=31.351, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.420e-04, train_time=7.075
+[gpub022:0/16] 2024-02-01 11:56:37,030 (trainer:737) INFO: 18epoch:train:1301-1400batch: iter_time=8.825e-05, forward_time=0.386, loss_ctc=53.141, loss_att=48.761, acc=0.745, loss=50.075, backward_time=0.457, grad_norm=31.905, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.419e-04, train_time=1.922
+[gpub022:0/16] 2024-02-01 11:59:24,365 (trainer:737) INFO: 18epoch:train:1401-1500batch: iter_time=9.048e-05, forward_time=0.377, loss_ctc=48.752, loss_att=44.621, acc=0.742, loss=45.860, backward_time=0.470, grad_norm=30.073, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.419e-04, train_time=1.673
+[gpub022:0/16] 2024-02-01 12:02:10,692 (trainer:737) INFO: 18epoch:train:1501-1600batch: iter_time=9.158e-05, forward_time=0.368, loss_ctc=43.387, loss_att=38.956, acc=0.727, loss=40.285, backward_time=0.449, grad_norm=32.587, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.418e-04, train_time=1.663
+[gpub022:0/16] 2024-02-01 12:05:33,236 (trainer:737) INFO: 18epoch:train:1601-1700batch: iter_time=2.274e-04, forward_time=0.514, loss_ctc=56.705, loss_att=55.931, acc=0.727, loss=56.163, backward_time=0.475, grad_norm=47.712, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.418e-04, train_time=2.025
+[gpub022:0/16] 2024-02-01 12:08:32,824 (trainer:737) INFO: 18epoch:train:1701-1800batch: iter_time=0.001, forward_time=0.395, loss_ctc=58.414, loss_att=51.942, acc=0.739, loss=53.883, backward_time=0.474, grad_norm=34.973, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.417e-04, train_time=1.795
+[gpub022:0/16] 2024-02-01 12:12:10,647 (trainer:737) INFO: 18epoch:train:1801-1900batch: iter_time=5.196e-04, forward_time=0.573, loss_ctc=52.384, loss_att=50.784, acc=0.735, loss=51.264, backward_time=0.496, grad_norm=35.042, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.111, optim0_lr0=2.417e-04, train_time=2.179
+[gpub022:0/16] 2024-02-01 12:15:08,305 (trainer:737) INFO: 18epoch:train:1901-2000batch: iter_time=3.750e-04, forward_time=0.385, loss_ctc=55.973, loss_att=48.891, acc=0.732, loss=51.016, backward_time=0.481, grad_norm=36.302, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.416e-04, train_time=1.776
+[gpub022:0/16] 2024-02-01 12:17:50,982 (trainer:737) INFO: 18epoch:train:2001-2100batch: iter_time=2.930e-04, forward_time=0.412, loss_ctc=55.744, loss_att=55.374, acc=0.736, loss=55.485, backward_time=0.466, grad_norm=36.727, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.416e-04, train_time=1.627
+[gpub022:0/16] 2024-02-01 12:20:45,876 (trainer:737) INFO: 18epoch:train:2101-2200batch: iter_time=8.576e-05, forward_time=0.401, loss_ctc=49.662, loss_att=47.217, acc=0.734, loss=47.950, backward_time=0.439, grad_norm=33.168, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.415e-04, train_time=1.750
+[gpub022:0/16] 2024-02-01 12:23:45,744 (trainer:737) INFO: 18epoch:train:2201-2300batch: iter_time=4.022e-04, forward_time=0.436, loss_ctc=54.243, loss_att=50.634, acc=0.753, loss=51.717, backward_time=0.441, grad_norm=33.689, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.415e-04, train_time=1.798
+[gpub022:0/16] 2024-02-01 12:25:20,805 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 12:26:52,687 (trainer:737) INFO: 18epoch:train:2301-2400batch: iter_time=4.990e-04, forward_time=0.409, loss_ctc=54.007, loss_att=55.198, acc=0.712, loss=54.841, backward_time=0.442, grad_norm=40.621, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.099, optim0_lr0=2.414e-04, train_time=1.869
+[gpub022:0/16] 2024-02-01 12:29:40,107 (trainer:737) INFO: 18epoch:train:2401-2500batch: iter_time=3.334e-04, forward_time=0.425, loss_ctc=44.624, loss_att=42.682, acc=0.732, loss=43.265, backward_time=0.443, grad_norm=29.779, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.414e-04, train_time=1.674
+[gpub022:0/16] 2024-02-01 12:30:00,430 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub022:0/16] 2024-02-01 12:30:19,175 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 12:30:23,037 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe76d568310>)
+[gpub022:0/16] 2024-02-01 12:30:23,037 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub022:0/16] 2024-02-01 12:30:23,042 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 12:41:34,413 (trainer:737) INFO: 18epoch:train:2501-2600batch: iter_time=3.332, forward_time=0.319, loss_ctc=45.704, loss_att=47.858, acc=0.734, loss=47.212, backward_time=0.446, grad_norm=32.112, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.413e-04, train_time=7.144
+[gpub022:0/16] 2024-02-01 12:44:47,806 (trainer:737) INFO: 18epoch:train:2601-2700batch: iter_time=9.059e-05, forward_time=0.399, loss_ctc=53.511, loss_att=50.016, acc=0.719, loss=51.065, backward_time=0.421, grad_norm=32.100, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.413e-04, train_time=1.934
+[gpub022:0/16] 2024-02-01 12:47:46,692 (trainer:737) INFO: 18epoch:train:2701-2800batch: iter_time=9.547e-05, forward_time=0.403, loss_ctc=42.835, loss_att=39.133, acc=0.736, loss=40.244, backward_time=0.457, grad_norm=31.739, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.412e-04, train_time=1.788
+[gpub022:0/16] 2024-02-01 12:50:36,093 (trainer:737) INFO: 18epoch:train:2801-2900batch: iter_time=5.605e-04, forward_time=0.413, loss_ctc=53.035, loss_att=51.078, acc=0.721, loss=51.665, backward_time=0.441, grad_norm=34.700, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.412e-04, train_time=1.694
+[gpub022:0/16] 2024-02-01 12:53:04,964 (trainer:737) INFO: 18epoch:train:2901-3000batch: iter_time=4.883e-04, forward_time=0.321, loss_ctc=57.212, loss_att=51.049, acc=0.728, loss=52.898, backward_time=0.409, grad_norm=44.774, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.412e-04, train_time=1.488
+[gpub022:0/16] 2024-02-01 12:55:51,580 (trainer:737) INFO: 18epoch:train:3001-3100batch: iter_time=9.796e-05, forward_time=0.400, loss_ctc=50.235, loss_att=56.004, acc=0.711, loss=54.274, backward_time=0.477, grad_norm=35.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.411e-04, train_time=1.667
+[gpub022:0/16] 2024-02-01 12:58:50,628 (trainer:737) INFO: 18epoch:train:3101-3200batch: iter_time=2.460e-04, forward_time=0.308, loss_ctc=59.679, loss_att=48.622, acc=0.737, loss=51.939, backward_time=0.413, grad_norm=37.046, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.411e-04, train_time=1.790
+[gpub022:0/16] 2024-02-01 13:01:48,985 (trainer:737) INFO: 18epoch:train:3201-3300batch: iter_time=6.294e-04, forward_time=0.400, loss_ctc=48.024, loss_att=46.648, acc=0.728, loss=47.061, backward_time=0.453, grad_norm=32.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.410e-04, train_time=1.783
+[gpub022:0/16] 2024-02-01 13:04:34,725 (trainer:737) INFO: 18epoch:train:3301-3400batch: iter_time=9.824e-05, forward_time=0.380, loss_ctc=56.817, loss_att=51.441, acc=0.730, loss=53.054, backward_time=0.458, grad_norm=35.837, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.410e-04, train_time=1.657
+[gpub022:0/16] 2024-02-01 13:07:21,297 (trainer:737) INFO: 18epoch:train:3401-3500batch: iter_time=7.920e-04, forward_time=0.379, loss_ctc=48.926, loss_att=46.838, acc=0.741, loss=47.465, backward_time=0.454, grad_norm=32.731, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.409e-04, train_time=1.666
+[gpub022:0/16] 2024-02-01 13:10:10,638 (trainer:737) INFO: 18epoch:train:3501-3600batch: iter_time=1.050e-04, forward_time=0.309, loss_ctc=55.158, loss_att=55.168, acc=0.725, loss=55.165, backward_time=0.406, grad_norm=37.344, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.409e-04, train_time=1.693
+[gpub022:0/16] 2024-02-01 13:12:52,135 (trainer:737) INFO: 18epoch:train:3601-3700batch: iter_time=4.880e-04, forward_time=0.377, loss_ctc=46.553, loss_att=43.691, acc=0.716, loss=44.549, backward_time=0.484, grad_norm=34.937, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.408e-04, train_time=1.615
+[gpub022:0/16] 2024-02-01 13:14:32,584 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub022:0/16] 2024-02-01 13:14:51,182 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 13:14:54,680 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe204e8cdc0>)
+[gpub022:0/16] 2024-02-01 13:14:54,680 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub022:0/16] 2024-02-01 13:14:54,685 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 13:24:29,092 (trainer:737) INFO: 18epoch:train:3701-3800batch: iter_time=5.151, forward_time=0.471, loss_ctc=44.719, loss_att=48.120, acc=0.714, loss=47.100, backward_time=0.458, grad_norm=32.221, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.408e-04, train_time=6.969
+[gpub022:0/16] 2024-02-01 13:28:02,099 (trainer:737) INFO: 18epoch:train:3801-3900batch: iter_time=0.002, forward_time=0.611, loss_ctc=51.734, loss_att=46.755, acc=0.738, loss=48.249, backward_time=0.515, grad_norm=32.531, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=2.407e-04, train_time=2.130
+[gpub022:0/16] 2024-02-01 13:31:14,981 (trainer:737) INFO: 18epoch:train:3901-4000batch: iter_time=4.878e-04, forward_time=0.476, loss_ctc=48.361, loss_att=44.549, acc=0.735, loss=45.693, backward_time=0.537, grad_norm=30.835, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=2.407e-04, train_time=1.928
+[gpub022:0/16] 2024-02-01 13:34:31,168 (trainer:737) INFO: 18epoch:train:4001-4100batch: iter_time=3.124e-04, forward_time=0.601, loss_ctc=42.754, loss_att=39.246, acc=0.721, loss=40.299, backward_time=0.571, grad_norm=31.523, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.117, optim0_lr0=2.406e-04, train_time=1.962
+[gpub022:0/16] 2024-02-01 13:37:38,124 (trainer:737) INFO: 18epoch:train:4101-4200batch: iter_time=6.845e-04, forward_time=0.629, loss_ctc=55.222, loss_att=54.130, acc=0.724, loss=54.458, backward_time=0.514, grad_norm=41.521, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=2.406e-04, train_time=1.869
+[gpub022:0/16] 2024-02-01 13:40:36,200 (trainer:737) INFO: 18epoch:train:4201-4300batch: iter_time=2.561e-04, forward_time=0.536, loss_ctc=56.510, loss_att=51.163, acc=0.736, loss=52.767, backward_time=0.487, grad_norm=35.046, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.114, optim0_lr0=2.406e-04, train_time=1.780
+[gpub022:0/16] 2024-02-01 13:43:46,179 (trainer:737) INFO: 18epoch:train:4301-4400batch: iter_time=7.856e-04, forward_time=0.543, loss_ctc=51.580, loss_att=49.804, acc=0.729, loss=50.337, backward_time=0.539, grad_norm=33.734, clip=100.000, loss_scale=7.529e+33, optim_step_time=0.109, optim0_lr0=2.405e-04, train_time=1.900
+[gpub022:0/16] 2024-02-01 13:46:47,387 (trainer:737) INFO: 18epoch:train:4401-4500batch: iter_time=4.705e-04, forward_time=0.531, loss_ctc=53.390, loss_att=47.620, acc=0.726, loss=49.351, backward_time=0.523, grad_norm=35.965, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.405e-04, train_time=1.812
+[gpub022:0/16] 2024-02-01 13:49:51,823 (trainer:737) INFO: 18epoch:train:4501-4600batch: iter_time=7.254e-04, forward_time=0.524, loss_ctc=54.679, loss_att=53.386, acc=0.734, loss=53.774, backward_time=0.487, grad_norm=36.717, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.404e-04, train_time=1.844
+[gpub022:0/16] 2024-02-01 13:52:43,675 (trainer:737) INFO: 18epoch:train:4601-4700batch: iter_time=4.744e-04, forward_time=0.496, loss_ctc=48.621, loss_att=45.977, acc=0.724, loss=46.770, backward_time=0.450, grad_norm=31.911, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.404e-04, train_time=1.718
+[gpub022:0/16] 2024-02-01 13:56:05,786 (trainer:737) INFO: 18epoch:train:4701-4800batch: iter_time=5.492e-04, forward_time=0.525, loss_ctc=54.049, loss_att=48.727, acc=0.757, loss=50.323, backward_time=0.505, grad_norm=33.531, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.403e-04, train_time=2.021
+[gpub022:0/16] 2024-02-01 13:59:03,660 (trainer:737) INFO: 18epoch:train:4801-4900batch: iter_time=0.001, forward_time=0.560, loss_ctc=52.783, loss_att=54.046, acc=0.706, loss=53.667, backward_time=0.499, grad_norm=38.868, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.110, optim0_lr0=2.403e-04, train_time=1.778
+[gpub022:0/16] 2024-02-01 14:01:19,229 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 14:01:46,705 (trainer:737) INFO: 18epoch:train:4901-5000batch: iter_time=4.153e-04, forward_time=0.440, loss_ctc=44.030, loss_att=43.703, acc=0.714, loss=43.801, backward_time=0.452, grad_norm=30.207, clip=100.000, loss_scale=9.336e+33, optim_step_time=0.103, optim0_lr0=2.402e-04, train_time=1.630
+[gpub022:0/16] 2024-02-01 14:02:07,104 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub022:0/16] 2024-02-01 14:02:25,381 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 14:02:28,958 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe76ca66740>)
+[gpub022:0/16] 2024-02-01 14:02:28,958 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub022:0/16] 2024-02-01 14:02:29,167 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 14:10:35,135 (trainer:737) INFO: 18epoch:train:5001-5100batch: iter_time=3.400, forward_time=0.525, loss_ctc=44.726, loss_att=45.727, acc=0.748, loss=45.426, backward_time=0.467, grad_norm=31.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.402e-04, train_time=5.285
+[gpub022:0/16] 2024-02-01 14:13:37,812 (trainer:737) INFO: 18epoch:train:5101-5200batch: iter_time=1.998e-04, forward_time=0.542, loss_ctc=52.339, loss_att=50.060, acc=0.735, loss=50.743, backward_time=0.518, grad_norm=30.676, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.123, optim0_lr0=2.401e-04, train_time=1.827
+[gpub022:0/16] 2024-02-01 14:16:16,973 (trainer:737) INFO: 18epoch:train:5201-5300batch: iter_time=5.770e-04, forward_time=0.445, loss_ctc=42.368, loss_att=38.062, acc=0.741, loss=39.354, backward_time=0.454, grad_norm=30.738, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.401e-04, train_time=1.590
+[gpub022:0/16] 2024-02-01 14:18:55,583 (trainer:737) INFO: 18epoch:train:5301-5400batch: iter_time=2.243e-04, forward_time=0.398, loss_ctc=52.167, loss_att=50.368, acc=0.736, loss=50.908, backward_time=0.465, grad_norm=33.531, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.400e-04, train_time=1.586
+[gpub022:0/16] 2024-02-01 14:19:36,396 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 14:22:00,491 (trainer:737) INFO: 18epoch:train:5401-5500batch: iter_time=3.262e-04, forward_time=0.440, loss_ctc=56.966, loss_att=50.850, acc=0.738, loss=52.684, backward_time=0.444, grad_norm=43.780, clip=100.000, loss_scale=3.252e+33, optim_step_time=0.100, optim0_lr0=2.400e-04, train_time=1.849
+[gpub022:0/16] 2024-02-01 14:24:41,046 (trainer:737) INFO: 18epoch:train:5501-5600batch: iter_time=2.583e-04, forward_time=0.424, loss_ctc=50.203, loss_att=55.557, acc=0.723, loss=53.951, backward_time=0.463, grad_norm=34.582, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.400e-04, train_time=1.606
+[gpub022:0/16] 2024-02-01 14:27:21,446 (trainer:737) INFO: 18epoch:train:5601-5700batch: iter_time=0.001, forward_time=0.414, loss_ctc=57.978, loss_att=49.854, acc=0.738, loss=52.291, backward_time=0.482, grad_norm=35.675, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.399e-04, train_time=1.604
+[gpub022:0/16] 2024-02-01 14:30:08,500 (trainer:737) INFO: 18epoch:train:5701-5800batch: iter_time=3.693e-04, forward_time=0.424, loss_ctc=48.199, loss_att=47.643, acc=0.744, loss=47.810, backward_time=0.473, grad_norm=31.027, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.399e-04, train_time=1.670
+[gpub022:0/16] 2024-02-01 14:32:51,000 (trainer:737) INFO: 18epoch:train:5801-5900batch: iter_time=3.874e-04, forward_time=0.372, loss_ctc=56.319, loss_att=51.664, acc=0.740, loss=53.061, backward_time=0.442, grad_norm=37.910, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.398e-04, train_time=1.625
+[gpub022:0/16] 2024-02-01 14:35:34,752 (trainer:737) INFO: 18epoch:train:5901-6000batch: iter_time=2.337e-04, forward_time=0.531, loss_ctc=48.656, loss_att=48.294, acc=0.751, loss=48.402, backward_time=0.459, grad_norm=31.391, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.398e-04, train_time=1.637
+[gpub022:0/16] 2024-02-01 14:38:16,940 (trainer:737) INFO: 18epoch:train:6001-6100batch: iter_time=4.129e-04, forward_time=0.384, loss_ctc=54.557, loss_att=55.376, acc=0.733, loss=55.131, backward_time=0.455, grad_norm=35.836, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.397e-04, train_time=1.622
+[gpub022:0/16] 2024-02-01 14:41:03,811 (trainer:737) INFO: 18epoch:train:6101-6200batch: iter_time=1.770e-04, forward_time=0.387, loss_ctc=45.659, loss_att=44.386, acc=0.724, loss=44.767, backward_time=0.428, grad_norm=33.532, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=2.397e-04, train_time=1.668
+[gpub022:0/16] 2024-02-01 14:42:39,827 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub022:0/16] 2024-02-01 14:42:58,303 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 14:43:02,367 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe2c1e6bee0>)
+[gpub022:0/16] 2024-02-01 14:43:02,368 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub022:0/16] 2024-02-01 14:43:02,371 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 14:51:41,179 (trainer:737) INFO: 18epoch:train:6201-6300batch: iter_time=3.506, forward_time=0.341, loss_ctc=44.400, loss_att=47.922, acc=0.726, loss=46.866, backward_time=0.410, grad_norm=32.011, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.396e-04, train_time=6.374
+[gpub022:0/16] 2024-02-01 14:54:21,033 (trainer:737) INFO: 18epoch:train:6301-6400batch: iter_time=8.929e-05, forward_time=0.295, loss_ctc=51.706, loss_att=47.537, acc=0.751, loss=48.787, backward_time=0.405, grad_norm=31.924, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.396e-04, train_time=1.598
+[gpub022:0/16] 2024-02-01 14:57:15,751 (trainer:737) INFO: 18epoch:train:6401-6500batch: iter_time=8.671e-05, forward_time=0.294, loss_ctc=48.551, loss_att=44.087, acc=0.745, loss=45.427, backward_time=0.405, grad_norm=30.907, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.395e-04, train_time=1.747
+[gpub022:0/16] 2024-02-01 15:00:01,020 (trainer:737) INFO: 18epoch:train:6501-6600batch: iter_time=7.717e-04, forward_time=0.382, loss_ctc=42.283, loss_att=37.509, acc=0.735, loss=38.941, backward_time=0.436, grad_norm=30.969, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=2.395e-04, train_time=1.652
+[gpub022:0/16] 2024-02-01 15:02:28,519 (trainer:737) INFO: 18epoch:train:6601-6700batch: iter_time=8.786e-05, forward_time=0.292, loss_ctc=55.225, loss_att=55.297, acc=0.731, loss=55.275, backward_time=0.406, grad_norm=43.972, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.394e-04, train_time=1.474
+[gpub022:0/16] 2024-02-01 15:05:02,068 (trainer:737) INFO: 18epoch:train:6701-6800batch: iter_time=8.663e-05, forward_time=0.296, loss_ctc=56.050, loss_att=50.976, acc=0.742, loss=52.498, backward_time=0.403, grad_norm=37.400, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.394e-04, train_time=1.536
+[gpub022:0/16] 2024-02-01 15:07:52,357 (trainer:737) INFO: 18epoch:train:6801-6900batch: iter_time=8.609e-05, forward_time=0.299, loss_ctc=50.304, loss_att=49.199, acc=0.740, loss=49.530, backward_time=0.404, grad_norm=32.537, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.394e-04, train_time=1.703
+[gpub022:0/16] 2024-02-01 15:10:45,862 (trainer:737) INFO: 18epoch:train:6901-7000batch: iter_time=3.008e-04, forward_time=0.430, loss_ctc=52.460, loss_att=48.139, acc=0.735, loss=49.435, backward_time=0.451, grad_norm=33.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.393e-04, train_time=1.735
+[gpub022:0/16] 2024-02-01 15:13:04,459 (trainer:737) INFO: 18epoch:train:7001-7100batch: iter_time=8.972e-05, forward_time=0.292, loss_ctc=54.641, loss_att=54.628, acc=0.740, loss=54.632, backward_time=0.404, grad_norm=36.980, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.393e-04, train_time=1.385
+[gpub022:0/16] 2024-02-01 15:15:33,325 (trainer:737) INFO: 18epoch:train:7101-7200batch: iter_time=8.448e-05, forward_time=0.293, loss_ctc=48.606, loss_att=46.359, acc=0.740, loss=47.033, backward_time=0.404, grad_norm=30.737, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.392e-04, train_time=1.489
+[gpub022:0/16] 2024-02-01 15:18:27,163 (trainer:737) INFO: 18epoch:train:7201-7300batch: iter_time=8.927e-05, forward_time=0.316, loss_ctc=53.661, loss_att=50.104, acc=0.758, loss=51.171, backward_time=0.409, grad_norm=32.791, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.392e-04, train_time=1.738
+[gpub022:0/16] 2024-02-01 15:21:23,995 (trainer:737) INFO: 18epoch:train:7301-7400batch: iter_time=3.474e-04, forward_time=0.430, loss_ctc=52.210, loss_att=54.493, acc=0.714, loss=53.808, backward_time=0.442, grad_norm=38.626, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.391e-04, train_time=1.768
+[gpub022:0/16] 2024-02-01 15:23:38,699 (trainer:737) INFO: 18epoch:train:7401-7500batch: iter_time=1.009e-04, forward_time=0.288, loss_ctc=43.496, loss_att=42.993, acc=0.733, loss=43.144, backward_time=0.403, grad_norm=30.973, clip=100.000, loss_scale=4.517e+33, optim_step_time=0.092, optim0_lr0=2.391e-04, train_time=1.346
+[gpub022:0/16] 2024-02-01 15:23:58,728 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub022:0/16] 2024-02-01 15:24:17,662 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 15:24:21,190 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85998b5e0>)
+[gpub022:0/16] 2024-02-01 15:24:21,190 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub022:0/16] 2024-02-01 15:24:21,193 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 15:32:01,290 (trainer:737) INFO: 18epoch:train:7501-7600batch: iter_time=3.267, forward_time=0.293, loss_ctc=44.966, loss_att=47.270, acc=0.737, loss=46.579, backward_time=0.407, grad_norm=34.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.390e-04, train_time=5.027
+[gpub022:0/16] 2024-02-01 15:34:33,752 (trainer:737) INFO: 18epoch:train:7601-7700batch: iter_time=8.055e-05, forward_time=0.420, loss_ctc=52.584, loss_att=49.237, acc=0.723, loss=50.241, backward_time=0.444, grad_norm=33.461, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.390e-04, train_time=1.524
+[gpub022:0/16] 2024-02-01 15:37:05,908 (trainer:737) INFO: 18epoch:train:7701-7800batch: iter_time=8.664e-05, forward_time=0.291, loss_ctc=42.649, loss_att=38.348, acc=0.742, loss=39.638, backward_time=0.403, grad_norm=29.991, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.389e-04, train_time=1.521
+[gpub022:0/16] 2024-02-01 15:39:32,951 (trainer:737) INFO: 18epoch:train:7801-7900batch: iter_time=8.901e-05, forward_time=0.295, loss_ctc=51.709, loss_att=50.120, acc=0.725, loss=50.597, backward_time=0.404, grad_norm=33.836, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.389e-04, train_time=1.470
+[gpub022:0/16] 2024-02-01 15:41:48,242 (trainer:737) INFO: 18epoch:train:7901-8000batch: iter_time=9.018e-05, forward_time=0.301, loss_ctc=56.787, loss_att=50.953, acc=0.731, loss=52.703, backward_time=0.405, grad_norm=45.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.389e-04, train_time=1.353
+[gpub022:0/16] 2024-02-01 15:44:42,132 (trainer:737) INFO: 18epoch:train:8001-8100batch: iter_time=6.428e-04, forward_time=0.408, loss_ctc=50.156, loss_att=55.390, acc=0.713, loss=53.820, backward_time=0.468, grad_norm=36.054, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.388e-04, train_time=1.739
+[gpub022:0/16] 2024-02-01 15:47:14,207 (trainer:737) INFO: 18epoch:train:8101-8200batch: iter_time=9.009e-05, forward_time=0.291, loss_ctc=56.807, loss_att=48.024, acc=0.741, loss=50.659, backward_time=0.404, grad_norm=34.794, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.388e-04, train_time=1.520
+[gpub022:0/16] 2024-02-01 15:49:32,432 (trainer:737) INFO: 18epoch:train:8201-8300batch: iter_time=7.945e-05, forward_time=0.321, loss_ctc=47.885, loss_att=46.122, acc=0.731, loss=46.650, backward_time=0.403, grad_norm=32.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.387e-04, train_time=1.381
+[gpub022:0/16] 2024-02-01 15:52:16,561 (trainer:737) INFO: 18epoch:train:8301-8400batch: iter_time=9.548e-05, forward_time=0.298, loss_ctc=56.331, loss_att=51.528, acc=0.735, loss=52.969, backward_time=0.406, grad_norm=36.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.387e-04, train_time=1.642
+[gpub022:0/16] 2024-02-01 15:55:12,494 (trainer:737) INFO: 18epoch:train:8401-8500batch: iter_time=8.955e-05, forward_time=0.414, loss_ctc=48.518, loss_att=47.070, acc=0.742, loss=47.504, backward_time=0.455, grad_norm=32.091, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.386e-04, train_time=1.759
+[gpub022:0/16] 2024-02-01 15:57:40,881 (trainer:737) INFO: 18epoch:train:8501-8600batch: iter_time=8.531e-05, forward_time=0.290, loss_ctc=54.279, loss_att=54.709, acc=0.731, loss=54.580, backward_time=0.405, grad_norm=37.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.386e-04, train_time=1.483
+[gpub022:0/16] 2024-02-01 16:00:10,064 (trainer:737) INFO: 18epoch:train:8601-8700batch: iter_time=9.312e-05, forward_time=0.294, loss_ctc=45.220, loss_att=44.198, acc=0.714, loss=44.505, backward_time=0.400, grad_norm=33.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.385e-04, train_time=1.491
+[gpub022:0/16] 2024-02-01 16:01:40,251 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub022:0/16] 2024-02-01 16:01:59,046 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 16:02:02,495 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd83bb6cfa0>)
+[gpub022:0/16] 2024-02-01 16:02:02,495 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub022:0/16] 2024-02-01 16:02:02,553 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 16:09:00,028 (trainer:737) INFO: 18epoch:train:8701-8800batch: iter_time=3.672, forward_time=0.362, loss_ctc=44.066, loss_att=47.285, acc=0.717, loss=46.319, backward_time=0.419, grad_norm=31.011, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.385e-04, train_time=5.300
+[gpub022:0/16] 2024-02-01 16:11:15,943 (trainer:737) INFO: 18epoch:train:8801-8900batch: iter_time=7.941e-05, forward_time=0.290, loss_ctc=51.050, loss_att=46.053, acc=0.742, loss=47.552, backward_time=0.405, grad_norm=32.258, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.384e-04, train_time=1.359
+[gpub022:0/16] 2024-02-01 16:13:34,512 (trainer:737) INFO: 18epoch:train:8901-9000batch: iter_time=8.454e-05, forward_time=0.295, loss_ctc=47.489, loss_att=43.743, acc=0.738, loss=44.867, backward_time=0.403, grad_norm=30.626, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.384e-04, train_time=1.385
+[gpub022:0/16] 2024-02-01 16:16:22,864 (trainer:737) INFO: 18epoch:train:9001-9100batch: iter_time=8.157e-05, forward_time=0.379, loss_ctc=42.212, loss_att=38.044, acc=0.727, loss=39.294, backward_time=0.421, grad_norm=31.080, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.384e-04, train_time=1.684
+[gpub022:0/16] 2024-02-01 16:18:43,412 (trainer:737) INFO: 18epoch:train:9101-9200batch: iter_time=8.485e-05, forward_time=0.313, loss_ctc=55.913, loss_att=56.171, acc=0.724, loss=56.093, backward_time=0.406, grad_norm=44.941, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.383e-04, train_time=1.405
+[gpub022:0/16] 2024-02-01 16:20:58,600 (trainer:737) INFO: 18epoch:train:9201-9300batch: iter_time=8.669e-05, forward_time=0.296, loss_ctc=55.014, loss_att=51.309, acc=0.737, loss=52.421, backward_time=0.404, grad_norm=34.195, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.383e-04, train_time=1.351
+[gpub022:0/16] 2024-02-01 16:23:54,196 (trainer:737) INFO: 18epoch:train:9301-9400batch: iter_time=8.351e-05, forward_time=0.364, loss_ctc=50.429, loss_att=49.259, acc=0.733, loss=49.610, backward_time=0.445, grad_norm=33.047, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.382e-04, train_time=1.756
+[gpub022:0/16] 2024-02-01 16:24:51,025 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 16:25:55,999 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 16:26:17,992 (trainer:737) INFO: 18epoch:train:9401-9500batch: iter_time=8.326e-05, forward_time=0.289, loss_ctc=51.812, loss_att=46.914, acc=0.730, loss=48.383, backward_time=0.403, grad_norm=33.490, clip=100.000, loss_scale=5.537e+33, optim_step_time=0.092, optim0_lr0=2.382e-04, train_time=1.438
+[gpub022:0/16] 2024-02-01 16:27:25,017 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 16:28:36,467 (trainer:737) INFO: 18epoch:train:9501-9600batch: iter_time=8.177e-05, forward_time=0.291, loss_ctc=53.787, loss_att=52.586, acc=0.738, loss=52.946, backward_time=0.404, grad_norm=37.410, clip=100.000, loss_scale=1.888e+33, optim_step_time=0.092, optim0_lr0=2.381e-04, train_time=1.385
+[gpub022:0/16] 2024-02-01 16:31:13,316 (trainer:737) INFO: 18epoch:train:9601-9700batch: iter_time=8.024e-05, forward_time=0.294, loss_ctc=48.444, loss_att=46.138, acc=0.728, loss=46.830, backward_time=0.399, grad_norm=32.364, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.381e-04, train_time=1.567
+[gpub022:0/16] 2024-02-01 16:33:55,719 (trainer:737) INFO: 18epoch:train:9701-9800batch: iter_time=8.130e-05, forward_time=0.389, loss_ctc=53.145, loss_att=47.759, acc=0.761, loss=49.375, backward_time=0.439, grad_norm=32.567, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.380e-04, train_time=1.625
+[gpub022:0/16] 2024-02-01 16:36:20,739 (trainer:737) INFO: 18epoch:train:9801-9900batch: iter_time=9.018e-05, forward_time=0.291, loss_ctc=52.104, loss_att=54.008, acc=0.707, loss=53.437, backward_time=0.404, grad_norm=39.162, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.380e-04, train_time=1.450
+[gpub022:0/16] 2024-02-01 16:38:55,609 (trainer:737) INFO: 18epoch:train:9901-10000batch: iter_time=8.735e-05, forward_time=0.294, loss_ctc=42.969, loss_att=42.764, acc=0.722, loss=42.826, backward_time=0.400, grad_norm=29.788, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.380e-04, train_time=1.548
+[gpub022:0/16] 2024-02-01 16:39:15,649 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub022:0/16] 2024-02-01 16:39:34,709 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 16:39:38,308 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd86346bfd0>)
+[gpub022:0/16] 2024-02-01 16:39:38,308 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub022:0/16] 2024-02-01 16:39:38,312 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 16:47:36,323 (trainer:737) INFO: 18epoch:train:10001-10100batch: iter_time=3.509, forward_time=0.392, loss_ctc=44.803, loss_att=45.612, acc=0.752, loss=45.369, backward_time=0.420, grad_norm=32.019, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.095, optim0_lr0=2.379e-04, train_time=5.207
+[gpub022:0/16] 2024-02-01 16:50:18,080 (trainer:737) INFO: 18epoch:train:10101-10200batch: iter_time=8.502e-05, forward_time=0.295, loss_ctc=52.663, loss_att=49.926, acc=0.738, loss=50.747, backward_time=0.404, grad_norm=33.050, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.379e-04, train_time=1.617
+[gpub022:0/16] 2024-02-01 16:53:39,153 (trainer:737) INFO: 18epoch:train:10201-10300batch: iter_time=9.233e-05, forward_time=0.388, loss_ctc=42.094, loss_att=38.472, acc=0.741, loss=39.559, backward_time=0.431, grad_norm=30.429, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.101, optim0_lr0=2.378e-04, train_time=2.010
+[gpub022:0/16] 2024-02-01 16:56:09,787 (trainer:737) INFO: 18epoch:train:10301-10400batch: iter_time=9.708e-05, forward_time=0.292, loss_ctc=51.857, loss_att=49.922, acc=0.737, loss=50.503, backward_time=0.406, grad_norm=34.099, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.378e-04, train_time=1.506
+[gpub022:0/16] 2024-02-01 16:59:04,242 (trainer:737) INFO: 18epoch:train:10401-10500batch: iter_time=9.021e-05, forward_time=0.295, loss_ctc=58.126, loss_att=53.032, acc=0.741, loss=54.560, backward_time=0.405, grad_norm=47.221, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.377e-04, train_time=1.744
+[gpub022:0/16] 2024-02-01 17:01:36,554 (trainer:737) INFO: 18epoch:train:10501-10600batch: iter_time=8.200e-04, forward_time=0.396, loss_ctc=49.246, loss_att=55.295, acc=0.727, loss=53.480, backward_time=0.457, grad_norm=34.067, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.102, optim0_lr0=2.377e-04, train_time=1.523
+[gpub022:0/16] 2024-02-01 17:04:27,457 (trainer:737) INFO: 18epoch:train:10601-10700batch: iter_time=8.994e-05, forward_time=0.290, loss_ctc=56.646, loss_att=49.934, acc=0.741, loss=51.948, backward_time=0.403, grad_norm=34.648, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.376e-04, train_time=1.709
+[gpub022:0/16] 2024-02-01 17:05:08,353 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 17:07:18,675 (trainer:737) INFO: 18epoch:train:10701-10800batch: iter_time=8.724e-05, forward_time=0.302, loss_ctc=47.419, loss_att=47.436, acc=0.746, loss=47.431, backward_time=0.402, grad_norm=31.369, clip=100.000, loss_scale=8.457e+32, optim_step_time=0.092, optim0_lr0=2.376e-04, train_time=1.711
+[gpub022:0/16] 2024-02-01 17:10:05,159 (trainer:737) INFO: 18epoch:train:10801-10900batch: iter_time=8.553e-05, forward_time=0.348, loss_ctc=55.973, loss_att=51.687, acc=0.741, loss=52.973, backward_time=0.466, grad_norm=37.382, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.099, optim0_lr0=2.375e-04, train_time=1.665
+[gpub022:0/16] 2024-02-01 17:12:33,571 (trainer:737) INFO: 18epoch:train:10901-11000batch: iter_time=8.895e-05, forward_time=0.290, loss_ctc=48.347, loss_att=47.830, acc=0.753, loss=47.985, backward_time=0.403, grad_norm=31.948, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.375e-04, train_time=1.484
+[gpub022:0/16] 2024-02-01 17:15:22,368 (trainer:737) INFO: 18epoch:train:11001-11100batch: iter_time=9.771e-05, forward_time=0.297, loss_ctc=53.910, loss_att=54.308, acc=0.739, loss=54.188, backward_time=0.405, grad_norm=36.260, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.375e-04, train_time=1.688
+[gpub022:0/16] 2024-02-01 17:18:17,246 (trainer:737) INFO: 18epoch:train:11101-11200batch: iter_time=9.242e-05, forward_time=0.365, loss_ctc=45.150, loss_att=43.469, acc=0.730, loss=43.973, backward_time=0.456, grad_norm=33.670, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.099, optim0_lr0=2.374e-04, train_time=1.749
+[gpub022:0/16] 2024-02-01 17:19:52,973 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub022:0/16] 2024-02-01 17:20:11,669 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 17:20:15,266 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd82ae12140>)
+[gpub022:0/16] 2024-02-01 17:20:15,267 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub022:0/16] 2024-02-01 17:20:15,270 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 17:26:01,065 (trainer:737) INFO: 18epoch:train:11201-11300batch: iter_time=2.997, forward_time=0.411, loss_ctc=44.062, loss_att=47.627, acc=0.726, loss=46.557, backward_time=0.425, grad_norm=31.853, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.095, optim0_lr0=2.374e-04, train_time=4.638
+[gpub022:0/16] 2024-02-01 17:28:41,250 (trainer:737) INFO: 18epoch:train:11301-11400batch: iter_time=7.748e-05, forward_time=0.296, loss_ctc=50.668, loss_att=46.828, acc=0.742, loss=47.980, backward_time=0.405, grad_norm=31.055, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.373e-04, train_time=1.602
+[gpub022:0/16] 2024-02-01 17:31:26,473 (trainer:737) INFO: 18epoch:train:11401-11500batch: iter_time=2.073e-04, forward_time=0.360, loss_ctc=48.065, loss_att=44.662, acc=0.737, loss=45.683, backward_time=0.447, grad_norm=32.719, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.096, optim0_lr0=2.373e-04, train_time=1.652
+[gpub022:0/16] 2024-02-01 17:33:40,369 (trainer:737) INFO: 18epoch:train:11501-11600batch: iter_time=9.395e-05, forward_time=0.286, loss_ctc=41.986, loss_att=37.949, acc=0.729, loss=39.160, backward_time=0.400, grad_norm=31.758, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.372e-04, train_time=1.339
+[gpub022:0/16] 2024-02-01 17:36:07,402 (trainer:737) INFO: 18epoch:train:11601-11700batch: iter_time=9.050e-05, forward_time=0.303, loss_ctc=54.024, loss_att=54.948, acc=0.723, loss=54.671, backward_time=0.407, grad_norm=43.908, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.372e-04, train_time=1.470
+[gpub022:0/16] 2024-02-01 17:38:43,785 (trainer:737) INFO: 18epoch:train:11701-11800batch: iter_time=9.383e-05, forward_time=0.361, loss_ctc=55.160, loss_att=50.888, acc=0.739, loss=52.170, backward_time=0.451, grad_norm=35.654, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.102, optim0_lr0=2.371e-04, train_time=1.563
+[gpub022:0/16] 2024-02-01 17:41:19,109 (trainer:737) INFO: 18epoch:train:11801-11900batch: iter_time=8.492e-05, forward_time=0.290, loss_ctc=49.900, loss_att=48.914, acc=0.734, loss=49.210, backward_time=0.403, grad_norm=33.129, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.371e-04, train_time=1.553
+[gpub022:0/16] 2024-02-01 17:43:47,895 (trainer:737) INFO: 18epoch:train:11901-12000batch: iter_time=8.958e-05, forward_time=0.291, loss_ctc=51.797, loss_att=46.558, acc=0.732, loss=48.129, backward_time=0.409, grad_norm=34.079, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.371e-04, train_time=1.488
+[gpub022:0/16] 2024-02-01 17:46:09,240 (trainer:737) INFO: 18epoch:train:12001-12100batch: iter_time=8.934e-05, forward_time=0.367, loss_ctc=52.887, loss_att=52.346, acc=0.737, loss=52.508, backward_time=0.430, grad_norm=38.132, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.098, optim0_lr0=2.370e-04, train_time=1.412
+[gpub022:0/16] 2024-02-01 17:48:29,984 (trainer:737) INFO: 18epoch:train:12101-12200batch: iter_time=8.550e-05, forward_time=0.288, loss_ctc=48.318, loss_att=46.400, acc=0.728, loss=46.975, backward_time=0.401, grad_norm=33.381, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.370e-04, train_time=1.408
+[gpub022:0/16] 2024-02-01 17:51:14,467 (trainer:737) INFO: 18epoch:train:12201-12300batch: iter_time=9.433e-05, forward_time=0.291, loss_ctc=53.270, loss_att=48.238, acc=0.762, loss=49.748, backward_time=0.414, grad_norm=33.655, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.369e-04, train_time=1.645
+[gpub022:0/16] 2024-02-01 17:53:37,896 (trainer:737) INFO: 18epoch:train:12301-12400batch: iter_time=0.001, forward_time=0.363, loss_ctc=51.651, loss_att=53.590, acc=0.708, loss=53.008, backward_time=0.435, grad_norm=38.387, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.102, optim0_lr0=2.369e-04, train_time=1.433
+[gpub022:0/16] 2024-02-01 17:55:48,481 (trainer:737) INFO: 18epoch:train:12401-12500batch: iter_time=9.029e-05, forward_time=0.289, loss_ctc=43.055, loss_att=42.817, acc=0.719, loss=42.888, backward_time=0.402, grad_norm=30.092, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.092, optim0_lr0=2.368e-04, train_time=1.307
+[gpub022:0/16] 2024-02-01 17:56:08,508 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub022:0/16] 2024-02-01 17:56:27,788 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 17:56:31,325 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd863bc01c0>)
+[gpub022:0/16] 2024-02-01 17:56:31,325 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub022:0/16] 2024-02-01 17:56:31,328 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 18:04:10,788 (trainer:737) INFO: 18epoch:train:12501-12600batch: iter_time=3.459, forward_time=0.299, loss_ctc=44.847, loss_att=45.249, acc=0.752, loss=45.128, backward_time=0.411, grad_norm=31.535, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.368e-04, train_time=5.023
+[gpub022:0/16] 2024-02-01 18:06:42,416 (trainer:737) INFO: 18epoch:train:12601-12700batch: iter_time=7.830e-05, forward_time=0.396, loss_ctc=52.625, loss_att=49.797, acc=0.738, loss=50.646, backward_time=0.426, grad_norm=32.277, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.110, optim0_lr0=2.367e-04, train_time=1.516
+[gpub022:0/16] 2024-02-01 18:09:33,281 (trainer:737) INFO: 18epoch:train:12701-12800batch: iter_time=8.760e-05, forward_time=0.287, loss_ctc=42.113, loss_att=37.960, acc=0.743, loss=39.206, backward_time=0.397, grad_norm=29.981, clip=100.000, loss_scale=1.097e+33, optim_step_time=0.092, optim0_lr0=2.367e-04, train_time=1.707
+[gpub022:0/16] 2024-02-01 18:11:52,411 (trainer:737) INFO: 18epoch:train:12801-12900batch: iter_time=8.633e-05, forward_time=0.302, loss_ctc=50.662, loss_att=49.346, acc=0.738, loss=49.741, backward_time=0.409, grad_norm=32.975, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.367e-04, train_time=1.392
+[gpub022:0/16] 2024-02-01 18:14:43,105 (trainer:737) INFO: 18epoch:train:12901-13000batch: iter_time=8.366e-05, forward_time=0.348, loss_ctc=56.010, loss_att=50.338, acc=0.743, loss=52.039, backward_time=0.432, grad_norm=42.572, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.102, optim0_lr0=2.366e-04, train_time=1.707
+[gpub022:0/16] 2024-02-01 18:17:13,200 (trainer:737) INFO: 18epoch:train:13001-13100batch: iter_time=9.542e-05, forward_time=0.317, loss_ctc=49.081, loss_att=54.965, acc=0.728, loss=53.200, backward_time=0.407, grad_norm=34.555, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.366e-04, train_time=1.501
+[gpub022:0/16] 2024-02-01 18:19:31,104 (trainer:737) INFO: 18epoch:train:13101-13200batch: iter_time=8.557e-05, forward_time=0.291, loss_ctc=55.471, loss_att=48.935, acc=0.744, loss=50.896, backward_time=0.404, grad_norm=33.485, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.365e-04, train_time=1.377
+[gpub022:0/16] 2024-02-01 18:22:07,062 (trainer:737) INFO: 18epoch:train:13201-13300batch: iter_time=8.892e-05, forward_time=0.300, loss_ctc=47.113, loss_att=46.775, acc=0.746, loss=46.876, backward_time=0.417, grad_norm=30.301, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.365e-04, train_time=1.561
+[gpub022:0/16] 2024-02-01 18:24:42,780 (trainer:737) INFO: 18epoch:train:13301-13400batch: iter_time=8.917e-05, forward_time=0.364, loss_ctc=54.952, loss_att=51.132, acc=0.744, loss=52.278, backward_time=0.428, grad_norm=36.411, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.102, optim0_lr0=2.364e-04, train_time=1.557
+[gpub022:0/16] 2024-02-01 18:27:25,253 (trainer:737) INFO: 18epoch:train:13401-13500batch: iter_time=8.739e-05, forward_time=0.290, loss_ctc=47.891, loss_att=47.504, acc=0.754, loss=47.620, backward_time=0.402, grad_norm=31.333, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.364e-04, train_time=1.625
+[gpub022:0/16] 2024-02-01 18:29:36,017 (trainer:737) INFO: 18epoch:train:13501-13600batch: iter_time=8.808e-05, forward_time=0.293, loss_ctc=53.404, loss_att=54.572, acc=0.737, loss=54.222, backward_time=0.406, grad_norm=35.713, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.364e-04, train_time=1.307
+[gpub022:0/16] 2024-02-01 18:32:07,824 (trainer:737) INFO: 18epoch:train:13601-13700batch: iter_time=8.543e-05, forward_time=0.306, loss_ctc=44.550, loss_att=43.870, acc=0.728, loss=44.074, backward_time=0.417, grad_norm=32.670, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.094, optim0_lr0=2.363e-04, train_time=1.519
+[gpub022:0/16] 2024-02-01 18:34:08,496 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub022:0/16] 2024-02-01 18:34:28,194 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 18:34:32,174 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe208d8f970>)
+[gpub022:0/16] 2024-02-01 18:34:32,174 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub022:0/16] 2024-02-01 18:34:32,178 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 18:41:13,493 (trainer:737) INFO: 18epoch:train:13701-13800batch: iter_time=3.741, forward_time=0.395, loss_ctc=43.634, loss_att=48.315, acc=0.723, loss=46.911, backward_time=0.435, grad_norm=33.939, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.095, optim0_lr0=2.363e-04, train_time=5.456
+[gpub022:0/16] 2024-02-01 18:43:38,832 (trainer:737) INFO: 18epoch:train:13801-13900batch: iter_time=7.520e-05, forward_time=0.291, loss_ctc=50.819, loss_att=47.260, acc=0.740, loss=48.328, backward_time=0.403, grad_norm=32.511, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.362e-04, train_time=1.453
+[gpub022:0/16] 2024-02-01 18:46:27,690 (trainer:737) INFO: 18epoch:train:13901-14000batch: iter_time=7.743e-05, forward_time=0.368, loss_ctc=47.187, loss_att=44.558, acc=0.738, loss=45.347, backward_time=0.501, grad_norm=31.115, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.100, optim0_lr0=2.362e-04, train_time=1.687
+[gpub022:0/16] 2024-02-01 18:49:22,365 (trainer:737) INFO: 18epoch:train:14001-14100batch: iter_time=7.943e-05, forward_time=0.286, loss_ctc=41.527, loss_att=38.570, acc=0.726, loss=39.457, backward_time=0.397, grad_norm=32.121, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.361e-04, train_time=1.747
+[gpub022:0/16] 2024-02-01 18:51:50,412 (trainer:737) INFO: 18epoch:train:14101-14200batch: iter_time=7.848e-05, forward_time=0.355, loss_ctc=55.336, loss_att=54.786, acc=0.726, loss=54.951, backward_time=0.479, grad_norm=46.863, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.361e-04, train_time=1.481
+[gpub022:0/16] 2024-02-01 18:54:30,047 (trainer:737) INFO: 18epoch:train:14201-14300batch: iter_time=8.250e-05, forward_time=0.308, loss_ctc=54.607, loss_att=50.089, acc=0.740, loss=51.444, backward_time=0.402, grad_norm=34.247, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.092, optim0_lr0=2.360e-04, train_time=1.595
+[gpub022:0/16] 2024-02-01 18:57:22,137 (trainer:737) INFO: 18epoch:train:14301-14400batch: iter_time=8.208e-05, forward_time=0.289, loss_ctc=49.652, loss_att=48.952, acc=0.733, loss=49.162, backward_time=0.403, grad_norm=34.929, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.360e-04, train_time=1.721
+[gpub022:0/16] 2024-02-01 19:01:40,179 (trainer:737) INFO: 18epoch:train:14401-14500batch: iter_time=0.132, forward_time=0.867, loss_ctc=50.852, loss_att=46.516, acc=0.731, loss=47.816, backward_time=0.487, grad_norm=33.607, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.100, optim0_lr0=2.360e-04, train_time=2.580
+[gpub022:0/16] 2024-02-01 19:04:24,823 (trainer:737) INFO: 18epoch:train:14501-14600batch: iter_time=7.594e-05, forward_time=0.307, loss_ctc=52.560, loss_att=51.958, acc=0.737, loss=52.138, backward_time=0.404, grad_norm=36.894, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.101, optim0_lr0=2.359e-04, train_time=1.646
+[gpub022:0/16] 2024-02-01 19:06:51,155 (trainer:737) INFO: 18epoch:train:14601-14700batch: iter_time=7.554e-05, forward_time=0.288, loss_ctc=48.345, loss_att=46.124, acc=0.728, loss=46.790, backward_time=0.401, grad_norm=33.522, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.359e-04, train_time=1.464
+[gpub022:0/16] 2024-02-01 19:11:16,255 (trainer:737) INFO: 18epoch:train:14701-14800batch: iter_time=7.576e-05, forward_time=0.875, loss_ctc=52.916, loss_att=47.927, acc=0.761, loss=49.424, backward_time=0.536, grad_norm=33.195, clip=100.000, loss_scale=2.194e+33, optim_step_time=0.150, optim0_lr0=2.358e-04, train_time=2.650
+[gpub022:0/16] 2024-02-01 19:14:41,388 (trainer:737) INFO: 18epoch:train:14801-14900batch: iter_time=7.599e-05, forward_time=0.353, loss_ctc=51.316, loss_att=53.594, acc=0.709, loss=52.911, backward_time=0.409, grad_norm=40.573, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.358e-04, train_time=2.050
+[gpub022:0/16] 2024-02-01 19:16:53,337 (trainer:737) INFO: 18epoch:train:14901-15000batch: iter_time=8.034e-05, forward_time=0.288, loss_ctc=43.031, loss_att=42.695, acc=0.722, loss=42.796, backward_time=0.401, grad_norm=31.140, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.357e-04, train_time=1.321
+[gpub022:0/16] 2024-02-01 19:56:07,228 (trainer:343) INFO: 18epoch results: [train] iter_time=0.291, forward_time=0.381, loss_ctc=50.688, loss_att=48.608, acc=0.733, loss=49.232, backward_time=0.439, grad_norm=34.656, clip=100.000, loss_scale=4.107e+33, optim_step_time=0.098, optim0_lr0=2.391e-04, train_time=1.996, time=8 hours, 19 minutes and 22.1 seconds, total_count=300000, gpu_max_cached_mem_GB=41.891, [valid] loss_ctc=41.877, cer_ctc=0.218, loss_att=42.636, acc=0.657, cer=0.320, wer=0.994, loss=42.408, time=38 minutes and 48.94 seconds, total_count=93420, gpu_max_cached_mem_GB=41.891
+[gpub022:0/16] 2024-02-01 19:56:26,121 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub022:0/16] 2024-02-01 19:56:26,166 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/13epoch.pth
+[gpub022:0/16] 2024-02-01 19:56:26,167 (trainer:272) INFO: 19/45epoch started. Estimated time to finish: 1 week, 3 days and 2 hours
+[gpub022:0/16] 2024-02-01 19:56:26,176 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub022:0/16] 2024-02-01 19:56:43,827 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 19:56:47,152 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85afe3100>)
+[gpub022:0/16] 2024-02-01 19:56:47,152 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub022:0/16] 2024-02-01 19:56:47,155 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 20:03:46,393 (trainer:737) INFO: 19epoch:train:1-100batch: iter_time=2.721, forward_time=0.494, loss_ctc=46.786, loss_att=43.911, acc=0.752, loss=44.774, backward_time=0.456, grad_norm=31.442, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.357e-04, train_time=4.401
+[gpub022:0/16] 2024-02-01 20:06:05,161 (trainer:737) INFO: 19epoch:train:101-200batch: iter_time=9.416e-05, forward_time=0.292, loss_ctc=65.900, loss_att=65.884, acc=0.720, loss=65.889, backward_time=0.407, grad_norm=44.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.356e-04, train_time=1.389
+[gpub022:0/16] 2024-02-01 20:09:00,767 (trainer:737) INFO: 19epoch:train:201-300batch: iter_time=4.911e-04, forward_time=0.434, loss_ctc=52.590, loss_att=50.960, acc=0.736, loss=51.449, backward_time=0.427, grad_norm=32.975, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.356e-04, train_time=1.755
+[gpub022:0/16] 2024-02-01 20:11:29,168 (trainer:737) INFO: 19epoch:train:301-400batch: iter_time=1.045e-04, forward_time=0.289, loss_ctc=47.686, loss_att=41.714, acc=0.754, loss=43.506, backward_time=0.403, grad_norm=31.484, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.356e-04, train_time=1.485
+[gpub022:0/16] 2024-02-01 20:14:08,644 (trainer:737) INFO: 19epoch:train:401-500batch: iter_time=4.683e-04, forward_time=0.354, loss_ctc=58.247, loss_att=58.472, acc=0.712, loss=58.404, backward_time=0.444, grad_norm=42.270, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.110, optim0_lr0=2.355e-04, train_time=1.595
+[gpub022:0/16] 2024-02-01 20:16:35,330 (trainer:737) INFO: 19epoch:train:501-600batch: iter_time=1.086e-04, forward_time=0.334, loss_ctc=57.944, loss_att=50.588, acc=0.729, loss=52.795, backward_time=0.420, grad_norm=39.129, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.355e-04, train_time=1.465
+[gpub022:0/16] 2024-02-01 20:19:11,470 (trainer:737) INFO: 19epoch:train:601-700batch: iter_time=2.057e-04, forward_time=0.371, loss_ctc=51.805, loss_att=49.064, acc=0.744, loss=49.886, backward_time=0.454, grad_norm=33.692, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.354e-04, train_time=1.563
+[gpub022:0/16] 2024-02-01 20:21:55,117 (trainer:737) INFO: 19epoch:train:701-800batch: iter_time=3.612e-04, forward_time=0.425, loss_ctc=45.807, loss_att=41.279, acc=0.753, loss=42.637, backward_time=0.439, grad_norm=31.229, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.354e-04, train_time=1.636
+[gpub022:0/16] 2024-02-01 20:24:42,656 (trainer:737) INFO: 19epoch:train:801-900batch: iter_time=4.417e-04, forward_time=0.405, loss_ctc=63.784, loss_att=56.138, acc=0.720, loss=58.432, backward_time=0.426, grad_norm=55.834, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=2.353e-04, train_time=1.676
+[gpub022:0/16] 2024-02-01 20:27:07,020 (trainer:737) INFO: 19epoch:train:901-1000batch: iter_time=9.694e-05, forward_time=0.301, loss_ctc=53.685, loss_att=49.117, acc=0.754, loss=50.488, backward_time=0.410, grad_norm=34.720, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.353e-04, train_time=1.443
+[gpub022:0/16] 2024-02-01 20:29:47,082 (trainer:737) INFO: 19epoch:train:1001-1100batch: iter_time=1.626e-04, forward_time=0.365, loss_ctc=42.949, loss_att=43.710, acc=0.739, loss=43.482, backward_time=0.437, grad_norm=29.752, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.353e-04, train_time=1.601
+[gpub022:0/16] 2024-02-01 20:32:36,002 (trainer:737) INFO: 19epoch:train:1101-1200batch: iter_time=2.036e-04, forward_time=0.389, loss_ctc=44.059, loss_att=43.595, acc=0.741, loss=43.734, backward_time=0.434, grad_norm=33.321, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=2.352e-04, train_time=1.688
+[gpub022:0/16] 2024-02-01 20:34:05,948 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub022:0/16] 2024-02-01 20:34:24,338 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 20:34:27,827 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85ad68310>)
+[gpub022:0/16] 2024-02-01 20:34:27,827 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub022:0/16] 2024-02-01 20:34:27,859 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 20:40:07,415 (trainer:737) INFO: 19epoch:train:1201-1300batch: iter_time=2.961, forward_time=0.290, loss_ctc=54.447, loss_att=49.153, acc=0.740, loss=50.741, backward_time=0.405, grad_norm=37.743, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.352e-04, train_time=4.515
+[gpub022:0/16] 2024-02-01 20:42:33,333 (trainer:737) INFO: 19epoch:train:1301-1400batch: iter_time=9.638e-05, forward_time=0.294, loss_ctc=60.769, loss_att=56.318, acc=0.742, loss=57.653, backward_time=0.413, grad_norm=38.263, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.351e-04, train_time=1.459
+[gpub022:0/16] 2024-02-01 20:45:09,988 (trainer:737) INFO: 19epoch:train:1401-1500batch: iter_time=9.719e-05, forward_time=0.356, loss_ctc=47.317, loss_att=53.299, acc=0.738, loss=51.504, backward_time=0.465, grad_norm=32.919, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.351e-04, train_time=1.566
+[gpub022:0/16] 2024-02-01 20:47:39,543 (trainer:737) INFO: 19epoch:train:1501-1600batch: iter_time=9.044e-05, forward_time=0.315, loss_ctc=52.142, loss_att=44.481, acc=0.747, loss=46.779, backward_time=0.403, grad_norm=31.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.350e-04, train_time=1.496
+[gpub022:0/16] 2024-02-01 20:50:12,144 (trainer:737) INFO: 19epoch:train:1601-1700batch: iter_time=1.027e-04, forward_time=0.290, loss_ctc=51.684, loss_att=53.714, acc=0.726, loss=53.105, backward_time=0.403, grad_norm=35.665, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.350e-04, train_time=1.526
+[gpub022:0/16] 2024-02-01 20:52:49,340 (trainer:737) INFO: 19epoch:train:1701-1800batch: iter_time=1.002e-04, forward_time=0.390, loss_ctc=55.045, loss_att=48.155, acc=0.727, loss=50.222, backward_time=0.452, grad_norm=44.268, clip=100.000, loss_scale=4.387e+33, optim_step_time=0.101, optim0_lr0=2.350e-04, train_time=1.572
+[gpub022:0/16] 2024-02-01 20:55:10,680 (trainer:737) INFO: 19epoch:train:1801-1900batch: iter_time=9.691e-05, forward_time=0.291, loss_ctc=52.886, loss_att=50.436, acc=0.750, loss=51.171, backward_time=0.409, grad_norm=33.992, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.349e-04, train_time=1.412
+[gpub022:0/16] 2024-02-01 20:57:43,541 (trainer:737) INFO: 19epoch:train:1901-2000batch: iter_time=9.241e-05, forward_time=0.290, loss_ctc=45.936, loss_att=41.773, acc=0.747, loss=43.022, backward_time=0.401, grad_norm=30.972, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.349e-04, train_time=1.530
+[gpub022:0/16] 2024-02-01 21:00:21,323 (trainer:737) INFO: 19epoch:train:2001-2100batch: iter_time=9.287e-05, forward_time=0.335, loss_ctc=53.386, loss_att=50.984, acc=0.733, loss=51.705, backward_time=0.424, grad_norm=37.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.348e-04, train_time=1.578
+[gpub022:0/16] 2024-02-01 21:02:58,370 (trainer:737) INFO: 19epoch:train:2101-2200batch: iter_time=9.513e-05, forward_time=0.362, loss_ctc=61.991, loss_att=56.770, acc=0.737, loss=58.336, backward_time=0.420, grad_norm=56.785, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.348e-04, train_time=1.570
+[gpub022:0/16] 2024-02-01 21:05:27,582 (trainer:737) INFO: 19epoch:train:2201-2300batch: iter_time=1.002e-04, forward_time=0.290, loss_ctc=43.859, loss_att=44.776, acc=0.760, loss=44.501, backward_time=0.404, grad_norm=29.180, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.347e-04, train_time=1.491
+[gpub022:0/16] 2024-02-01 21:07:53,884 (trainer:737) INFO: 19epoch:train:2301-2400batch: iter_time=9.948e-05, forward_time=0.288, loss_ctc=43.180, loss_att=43.506, acc=0.741, loss=43.408, backward_time=0.401, grad_norm=30.794, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.347e-04, train_time=1.464
+[gpub022:0/16] 2024-02-01 21:10:33,156 (trainer:737) INFO: 19epoch:train:2401-2500batch: iter_time=0.001, forward_time=0.335, loss_ctc=51.150, loss_att=46.117, acc=0.737, loss=47.627, backward_time=0.451, grad_norm=39.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.347e-04, train_time=1.592
+[gpub022:0/16] 2024-02-01 21:10:53,180 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub022:0/16] 2024-02-01 21:11:12,071 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 21:11:15,520 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd863a137c0>)
+[gpub022:0/16] 2024-02-01 21:11:15,520 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub022:0/16] 2024-02-01 21:11:15,542 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 21:17:57,431 (trainer:737) INFO: 19epoch:train:2501-2600batch: iter_time=3.010, forward_time=0.316, loss_ctc=45.940, loss_att=44.335, acc=0.746, loss=44.816, backward_time=0.407, grad_norm=33.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.346e-04, train_time=4.443
+[gpub022:0/16] 2024-02-01 21:20:31,401 (trainer:737) INFO: 19epoch:train:2601-2700batch: iter_time=1.026e-04, forward_time=0.293, loss_ctc=60.525, loss_att=63.997, acc=0.722, loss=62.955, backward_time=0.408, grad_norm=43.233, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.346e-04, train_time=1.539
+[gpub022:0/16] 2024-02-01 21:23:04,933 (trainer:737) INFO: 19epoch:train:2701-2800batch: iter_time=9.610e-05, forward_time=0.368, loss_ctc=50.919, loss_att=51.661, acc=0.727, loss=51.438, backward_time=0.428, grad_norm=34.864, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.345e-04, train_time=1.534
+[gpub022:0/16] 2024-02-01 21:25:31,864 (trainer:737) INFO: 19epoch:train:2801-2900batch: iter_time=9.023e-05, forward_time=0.312, loss_ctc=46.596, loss_att=41.668, acc=0.742, loss=43.147, backward_time=0.419, grad_norm=31.501, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.345e-04, train_time=1.470
+[gpub022:0/16] 2024-02-01 21:28:02,571 (trainer:737) INFO: 19epoch:train:2901-3000batch: iter_time=9.741e-05, forward_time=0.290, loss_ctc=56.777, loss_att=58.565, acc=0.709, loss=58.028, backward_time=0.404, grad_norm=43.355, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.344e-04, train_time=1.507
+[gpub022:0/16] 2024-02-01 21:30:22,608 (trainer:737) INFO: 19epoch:train:3001-3100batch: iter_time=9.278e-05, forward_time=0.294, loss_ctc=50.950, loss_att=47.926, acc=0.729, loss=48.833, backward_time=0.404, grad_norm=40.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.344e-04, train_time=1.400
+[gpub022:0/16] 2024-02-01 21:33:03,437 (trainer:737) INFO: 19epoch:train:3101-3200batch: iter_time=9.374e-05, forward_time=0.355, loss_ctc=50.702, loss_att=45.521, acc=0.746, loss=47.076, backward_time=0.467, grad_norm=34.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.344e-04, train_time=1.608
+[gpub022:0/16] 2024-02-01 21:35:24,090 (trainer:737) INFO: 19epoch:train:3201-3300batch: iter_time=9.809e-05, forward_time=0.288, loss_ctc=45.013, loss_att=40.989, acc=0.747, loss=42.196, backward_time=0.402, grad_norm=30.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.343e-04, train_time=1.406
+[gpub022:0/16] 2024-02-01 21:37:51,398 (trainer:737) INFO: 19epoch:train:3301-3400batch: iter_time=9.760e-05, forward_time=0.289, loss_ctc=62.658, loss_att=57.139, acc=0.720, loss=58.795, backward_time=0.404, grad_norm=53.437, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.343e-04, train_time=1.473
+[gpub022:0/16] 2024-02-01 21:40:17,034 (trainer:737) INFO: 19epoch:train:3401-3500batch: iter_time=1.039e-04, forward_time=0.292, loss_ctc=52.640, loss_att=48.100, acc=0.745, loss=49.462, backward_time=0.411, grad_norm=36.515, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.342e-04, train_time=1.456
+[gpub022:0/16] 2024-02-01 21:42:51,295 (trainer:737) INFO: 19epoch:train:3501-3600batch: iter_time=9.250e-05, forward_time=0.372, loss_ctc=42.195, loss_att=43.122, acc=0.741, loss=42.844, backward_time=0.427, grad_norm=30.145, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.342e-04, train_time=1.542
+[gpub022:0/16] 2024-02-01 21:45:22,552 (trainer:737) INFO: 19epoch:train:3601-3700batch: iter_time=9.497e-05, forward_time=0.291, loss_ctc=42.798, loss_att=43.420, acc=0.738, loss=43.233, backward_time=0.401, grad_norm=34.382, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.341e-04, train_time=1.513
+[gpub022:0/16] 2024-02-01 21:46:51,284 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub022:0/16] 2024-02-01 21:47:10,093 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 21:47:13,604 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd857621ea0>)
+[gpub022:0/16] 2024-02-01 21:47:13,604 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub022:0/16] 2024-02-01 21:47:13,626 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 21:53:11,660 (trainer:737) INFO: 19epoch:train:3701-3800batch: iter_time=3.286, forward_time=0.333, loss_ctc=52.791, loss_att=48.799, acc=0.736, loss=49.997, backward_time=0.411, grad_norm=37.475, clip=100.000, loss_scale=8.775e+33, optim_step_time=0.095, optim0_lr0=2.341e-04, train_time=4.691
+[gpub022:0/16] 2024-02-01 21:55:29,589 (trainer:737) INFO: 19epoch:train:3801-3900batch: iter_time=8.091e-05, forward_time=0.296, loss_ctc=58.522, loss_att=57.162, acc=0.727, loss=57.570, backward_time=0.406, grad_norm=39.116, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.341e-04, train_time=1.379
+[gpub022:0/16] 2024-02-01 21:57:53,497 (trainer:737) INFO: 19epoch:train:3901-4000batch: iter_time=7.796e-05, forward_time=0.290, loss_ctc=46.267, loss_att=52.513, acc=0.737, loss=50.639, backward_time=0.402, grad_norm=31.703, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.340e-04, train_time=1.439
+[gpub022:0/16] 2024-02-01 21:58:41,129 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 22:00:34,409 (trainer:737) INFO: 19epoch:train:4001-4100batch: iter_time=9.150e-05, forward_time=0.343, loss_ctc=51.852, loss_att=44.295, acc=0.739, loss=46.562, backward_time=0.461, grad_norm=32.395, clip=100.000, loss_scale=6.923e+33, optim_step_time=0.100, optim0_lr0=2.340e-04, train_time=1.609
+[gpub022:0/16] 2024-02-01 22:02:52,997 (trainer:737) INFO: 19epoch:train:4101-4200batch: iter_time=9.467e-05, forward_time=0.294, loss_ctc=51.120, loss_att=53.508, acc=0.716, loss=52.792, backward_time=0.406, grad_norm=37.570, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.339e-04, train_time=1.386
+[gpub022:0/16] 2024-02-01 22:05:15,316 (trainer:737) INFO: 19epoch:train:4201-4300batch: iter_time=9.623e-05, forward_time=0.291, loss_ctc=52.906, loss_att=46.666, acc=0.734, loss=48.538, backward_time=0.403, grad_norm=43.770, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.339e-04, train_time=1.422
+[gpub022:0/16] 2024-02-01 22:07:50,535 (trainer:737) INFO: 19epoch:train:4301-4400batch: iter_time=8.977e-05, forward_time=0.290, loss_ctc=52.185, loss_att=48.035, acc=0.744, loss=49.280, backward_time=0.404, grad_norm=35.645, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.338e-04, train_time=1.553
+[gpub022:0/16] 2024-02-01 22:10:40,251 (trainer:737) INFO: 19epoch:train:4401-4500batch: iter_time=9.651e-05, forward_time=0.361, loss_ctc=45.661, loss_att=40.253, acc=0.749, loss=41.875, backward_time=0.455, grad_norm=30.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.338e-04, train_time=1.697
+[gpub022:0/16] 2024-02-01 22:12:58,722 (trainer:737) INFO: 19epoch:train:4501-4600batch: iter_time=9.328e-05, forward_time=0.294, loss_ctc=52.162, loss_att=49.372, acc=0.728, loss=50.209, backward_time=0.404, grad_norm=37.028, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.338e-04, train_time=1.384
+[gpub022:0/16] 2024-02-01 22:15:29,467 (trainer:737) INFO: 19epoch:train:4601-4700batch: iter_time=9.178e-05, forward_time=0.295, loss_ctc=60.596, loss_att=56.344, acc=0.729, loss=57.619, backward_time=0.404, grad_norm=54.461, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.337e-04, train_time=1.507
+[gpub022:0/16] 2024-02-01 22:18:06,516 (trainer:737) INFO: 19epoch:train:4701-4800batch: iter_time=8.598e-05, forward_time=0.376, loss_ctc=43.263, loss_att=44.589, acc=0.757, loss=44.191, backward_time=0.444, grad_norm=28.571, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.337e-04, train_time=1.571
+[gpub022:0/16] 2024-02-01 22:20:34,500 (trainer:737) INFO: 19epoch:train:4801-4900batch: iter_time=8.217e-05, forward_time=0.287, loss_ctc=42.374, loss_att=42.710, acc=0.740, loss=42.609, backward_time=0.400, grad_norm=30.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.336e-04, train_time=1.480
+[gpub022:0/16] 2024-02-01 22:22:52,004 (trainer:737) INFO: 19epoch:train:4901-5000batch: iter_time=8.276e-05, forward_time=0.293, loss_ctc=50.103, loss_att=45.866, acc=0.730, loss=47.137, backward_time=0.403, grad_norm=37.620, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.336e-04, train_time=1.375
+[gpub022:0/16] 2024-02-01 22:23:12,308 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub022:0/16] 2024-02-01 22:23:31,028 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 22:23:34,588 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fdffc9a3e80>)
+[gpub022:0/16] 2024-02-01 22:23:34,588 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub022:0/16] 2024-02-01 22:23:34,591 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 22:30:50,139 (trainer:737) INFO: 19epoch:train:5001-5100batch: iter_time=3.295, forward_time=0.368, loss_ctc=45.723, loss_att=41.211, acc=0.756, loss=42.565, backward_time=0.415, grad_norm=30.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.335e-04, train_time=4.781
+[gpub022:0/16] 2024-02-01 22:33:22,026 (trainer:737) INFO: 19epoch:train:5101-5200batch: iter_time=9.378e-05, forward_time=0.292, loss_ctc=60.508, loss_att=63.294, acc=0.723, loss=62.458, backward_time=0.406, grad_norm=41.492, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.335e-04, train_time=1.519
+[gpub022:0/16] 2024-02-01 22:36:05,097 (trainer:737) INFO: 19epoch:train:5201-5300batch: iter_time=9.640e-05, forward_time=0.295, loss_ctc=50.553, loss_att=49.291, acc=0.735, loss=49.669, backward_time=0.402, grad_norm=31.867, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.335e-04, train_time=1.630
+[gpub022:0/16] 2024-02-01 22:38:36,726 (trainer:737) INFO: 19epoch:train:5301-5400batch: iter_time=1.026e-04, forward_time=0.370, loss_ctc=46.196, loss_att=40.075, acc=0.748, loss=41.911, backward_time=0.436, grad_norm=30.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=2.334e-04, train_time=1.517
+[gpub022:0/16] 2024-02-01 22:41:01,289 (trainer:737) INFO: 19epoch:train:5401-5500batch: iter_time=9.306e-05, forward_time=0.321, loss_ctc=55.788, loss_att=56.051, acc=0.716, loss=55.972, backward_time=0.403, grad_norm=39.056, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.334e-04, train_time=1.446
+[gpub022:0/16] 2024-02-01 22:43:30,865 (trainer:737) INFO: 19epoch:train:5501-5600batch: iter_time=1.071e-04, forward_time=0.289, loss_ctc=50.245, loss_att=47.519, acc=0.732, loss=48.337, backward_time=0.404, grad_norm=39.164, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.333e-04, train_time=1.495
+[gpub022:0/16] 2024-02-01 22:46:12,985 (trainer:737) INFO: 19epoch:train:5601-5700batch: iter_time=1.025e-04, forward_time=0.307, loss_ctc=50.620, loss_att=45.268, acc=0.748, loss=46.873, backward_time=0.427, grad_norm=100.497, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.333e-04, train_time=1.621
+[gpub022:0/16] 2024-02-01 22:48:52,585 (trainer:737) INFO: 19epoch:train:5701-5800batch: iter_time=9.555e-05, forward_time=0.393, loss_ctc=44.820, loss_att=41.090, acc=0.748, loss=42.209, backward_time=0.438, grad_norm=29.329, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.332e-04, train_time=1.596
+[gpub022:0/16] 2024-02-01 22:51:15,635 (trainer:737) INFO: 19epoch:train:5801-5900batch: iter_time=1.069e-04, forward_time=0.294, loss_ctc=59.526, loss_att=53.356, acc=0.725, loss=55.207, backward_time=0.403, grad_norm=52.847, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.332e-04, train_time=1.431
+[gpub022:0/16] 2024-02-01 22:53:53,013 (trainer:737) INFO: 19epoch:train:5901-6000batch: iter_time=9.792e-05, forward_time=0.293, loss_ctc=51.746, loss_att=47.421, acc=0.748, loss=48.719, backward_time=0.403, grad_norm=37.128, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.332e-04, train_time=1.573
+[gpub022:0/16] 2024-02-01 22:56:21,485 (trainer:737) INFO: 19epoch:train:6001-6100batch: iter_time=9.489e-05, forward_time=0.365, loss_ctc=41.572, loss_att=42.813, acc=0.744, loss=42.441, backward_time=0.441, grad_norm=28.825, clip=100.000, loss_scale=8.619e+33, optim_step_time=0.100, optim0_lr0=2.331e-04, train_time=1.485
+[gpub022:0/16] 2024-02-01 22:58:50,653 (trainer:737) INFO: 19epoch:train:6101-6200batch: iter_time=9.179e-05, forward_time=0.297, loss_ctc=42.208, loss_att=43.454, acc=0.737, loss=43.080, backward_time=0.402, grad_norm=33.202, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.331e-04, train_time=1.491
+[gpub022:0/16] 2024-02-01 23:00:30,910 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub022:0/16] 2024-02-01 23:00:49,703 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 23:00:53,262 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe052fb76a0>)
+[gpub022:0/16] 2024-02-01 23:00:53,262 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub022:0/16] 2024-02-01 23:00:53,265 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 23:07:16,856 (trainer:737) INFO: 19epoch:train:6201-6300batch: iter_time=3.547, forward_time=0.327, loss_ctc=52.137, loss_att=48.683, acc=0.737, loss=49.719, backward_time=0.409, grad_norm=35.836, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.330e-04, train_time=5.062
+[gpub022:0/16] 2024-02-01 23:09:51,954 (trainer:737) INFO: 19epoch:train:6301-6400batch: iter_time=8.310e-05, forward_time=0.337, loss_ctc=57.930, loss_att=56.350, acc=0.731, loss=56.824, backward_time=0.415, grad_norm=37.234, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.330e-04, train_time=1.551
+[gpub022:0/16] 2024-02-01 23:12:23,270 (trainer:737) INFO: 19epoch:train:6401-6500batch: iter_time=8.149e-05, forward_time=0.293, loss_ctc=46.345, loss_att=53.122, acc=0.737, loss=51.089, backward_time=0.402, grad_norm=32.013, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.329e-04, train_time=1.513
+[gpub022:0/16] 2024-02-01 23:14:58,789 (trainer:737) INFO: 19epoch:train:6501-6600batch: iter_time=1.023e-04, forward_time=0.367, loss_ctc=51.850, loss_att=44.316, acc=0.740, loss=46.576, backward_time=0.427, grad_norm=32.806, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.329e-04, train_time=1.555
+[gpub022:0/16] 2024-02-01 23:17:45,545 (trainer:737) INFO: 19epoch:train:6601-6700batch: iter_time=1.029e-04, forward_time=0.328, loss_ctc=50.764, loss_att=53.739, acc=0.719, loss=52.846, backward_time=0.421, grad_norm=34.629, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.329e-04, train_time=1.667
+[gpub022:0/16] 2024-02-01 23:20:19,944 (trainer:737) INFO: 19epoch:train:6701-6800batch: iter_time=1.030e-04, forward_time=0.291, loss_ctc=51.222, loss_att=45.776, acc=0.736, loss=47.410, backward_time=0.403, grad_norm=43.842, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.328e-04, train_time=1.544
+[gpub022:0/16] 2024-02-01 23:23:05,463 (trainer:737) INFO: 19epoch:train:6801-6900batch: iter_time=1.073e-04, forward_time=0.375, loss_ctc=51.946, loss_att=47.485, acc=0.747, loss=48.823, backward_time=0.424, grad_norm=33.840, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.328e-04, train_time=1.654
+[gpub022:0/16] 2024-02-01 23:26:02,949 (trainer:737) INFO: 19epoch:train:6901-7000batch: iter_time=1.047e-04, forward_time=0.352, loss_ctc=45.561, loss_att=40.273, acc=0.749, loss=41.860, backward_time=0.421, grad_norm=29.911, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.327e-04, train_time=1.775
+[gpub022:0/16] 2024-02-01 23:28:22,010 (trainer:737) INFO: 19epoch:train:7001-7100batch: iter_time=9.962e-05, forward_time=0.289, loss_ctc=51.965, loss_att=48.602, acc=0.732, loss=49.611, backward_time=0.404, grad_norm=37.114, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.327e-04, train_time=1.390
+[gpub022:0/16] 2024-02-01 23:31:37,974 (trainer:737) INFO: 19epoch:train:7101-7200batch: iter_time=3.172e-04, forward_time=0.315, loss_ctc=59.284, loss_att=53.630, acc=0.734, loss=55.327, backward_time=0.420, grad_norm=48.967, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.327e-04, train_time=1.960
+[gpub022:0/16] 2024-02-01 23:33:58,572 (trainer:737) INFO: 19epoch:train:7201-7300batch: iter_time=9.873e-05, forward_time=0.329, loss_ctc=43.206, loss_att=44.113, acc=0.758, loss=43.841, backward_time=0.407, grad_norm=29.093, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.326e-04, train_time=1.406
+[gpub022:0/16] 2024-02-01 23:36:46,883 (trainer:737) INFO: 19epoch:train:7301-7400batch: iter_time=1.065e-04, forward_time=0.327, loss_ctc=42.482, loss_att=43.006, acc=0.738, loss=42.849, backward_time=0.422, grad_norm=29.824, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.326e-04, train_time=1.683
+[gpub022:0/16] 2024-02-01 23:39:18,293 (trainer:737) INFO: 19epoch:train:7401-7500batch: iter_time=9.992e-05, forward_time=0.292, loss_ctc=49.252, loss_att=45.141, acc=0.731, loss=46.375, backward_time=0.402, grad_norm=37.824, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.325e-04, train_time=1.514
+[gpub022:0/16] 2024-02-01 23:39:38,321 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub022:0/16] 2024-02-01 23:39:57,266 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-01 23:40:00,832 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe05a88b790>)
+[gpub022:0/16] 2024-02-01 23:40:00,832 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub022:0/16] 2024-02-01 23:40:00,835 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-01 23:48:34,092 (trainer:737) INFO: 19epoch:train:7501-7600batch: iter_time=3.935, forward_time=0.380, loss_ctc=45.381, loss_att=43.077, acc=0.759, loss=43.768, backward_time=0.418, grad_norm=31.371, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.325e-04, train_time=5.558
+[gpub022:0/16] 2024-02-01 23:50:55,685 (trainer:737) INFO: 19epoch:train:7601-7700batch: iter_time=1.103e-04, forward_time=0.293, loss_ctc=58.330, loss_att=62.993, acc=0.732, loss=61.594, backward_time=0.407, grad_norm=38.096, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.324e-04, train_time=1.416
+[gpub022:0/16] 2024-02-01 23:52:00,009 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-01 23:53:55,123 (trainer:737) INFO: 19epoch:train:7701-7800batch: iter_time=2.481e-04, forward_time=0.425, loss_ctc=50.606, loss_att=50.939, acc=0.740, loss=50.839, backward_time=0.443, grad_norm=33.913, clip=100.000, loss_scale=7.395e+33, optim_step_time=0.101, optim0_lr0=2.324e-04, train_time=1.794
+[gpub022:0/16] 2024-02-01 23:56:21,214 (trainer:737) INFO: 19epoch:train:7801-7900batch: iter_time=9.679e-05, forward_time=0.290, loss_ctc=46.233, loss_att=41.603, acc=0.758, loss=42.992, backward_time=0.404, grad_norm=29.703, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.324e-04, train_time=1.461
+[gpub022:0/16] 2024-02-01 23:58:44,421 (trainer:737) INFO: 19epoch:train:7901-8000batch: iter_time=1.055e-04, forward_time=0.294, loss_ctc=55.025, loss_att=56.759, acc=0.722, loss=56.239, backward_time=0.404, grad_norm=39.883, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.323e-04, train_time=1.432
+[gpub022:0/16] 2024-02-02 00:01:36,741 (trainer:737) INFO: 19epoch:train:8001-8100batch: iter_time=1.082e-04, forward_time=0.405, loss_ctc=49.995, loss_att=49.129, acc=0.734, loss=49.389, backward_time=0.437, grad_norm=38.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.323e-04, train_time=1.723
+[gpub022:0/16] 2024-02-02 00:04:12,911 (trainer:737) INFO: 19epoch:train:8101-8200batch: iter_time=9.360e-05, forward_time=0.291, loss_ctc=50.176, loss_att=48.601, acc=0.749, loss=49.073, backward_time=0.405, grad_norm=32.831, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.322e-04, train_time=1.561
+[gpub022:0/16] 2024-02-02 00:06:43,784 (trainer:737) INFO: 19epoch:train:8201-8300batch: iter_time=8.911e-05, forward_time=0.296, loss_ctc=44.368, loss_att=40.691, acc=0.759, loss=41.794, backward_time=0.414, grad_norm=29.698, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.322e-04, train_time=1.509
+[gpub022:0/16] 2024-02-02 00:09:49,065 (trainer:737) INFO: 19epoch:train:8301-8400batch: iter_time=2.874e-04, forward_time=0.374, loss_ctc=59.843, loss_att=54.425, acc=0.728, loss=56.050, backward_time=0.437, grad_norm=48.741, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.322e-04, train_time=1.852
+[gpub022:0/16] 2024-02-02 00:12:21,959 (trainer:737) INFO: 19epoch:train:8401-8500batch: iter_time=8.975e-05, forward_time=0.292, loss_ctc=51.630, loss_att=48.666, acc=0.757, loss=49.555, backward_time=0.403, grad_norm=32.220, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.321e-04, train_time=1.529
+[gpub022:0/16] 2024-02-02 00:15:07,180 (trainer:737) INFO: 19epoch:train:8501-8600batch: iter_time=9.041e-05, forward_time=0.347, loss_ctc=41.085, loss_att=43.370, acc=0.746, loss=42.684, backward_time=0.453, grad_norm=30.203, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.321e-04, train_time=1.652
+[gpub022:0/16] 2024-02-02 00:17:33,722 (trainer:737) INFO: 19epoch:train:8601-8700batch: iter_time=1.039e-04, forward_time=0.321, loss_ctc=42.174, loss_att=43.152, acc=0.746, loss=42.859, backward_time=0.414, grad_norm=32.490, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.320e-04, train_time=1.465
+[gpub022:0/16] 2024-02-02 00:19:27,910 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub022:0/16] 2024-02-02 00:19:47,120 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 00:19:51,259 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe051d6b5b0>)
+[gpub022:0/16] 2024-02-02 00:19:51,259 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub022:0/16] 2024-02-02 00:19:51,263 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 00:26:32,398 (trainer:737) INFO: 19epoch:train:8701-8800batch: iter_time=3.680, forward_time=0.288, loss_ctc=51.363, loss_att=47.729, acc=0.748, loss=48.819, backward_time=0.403, grad_norm=34.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.320e-04, train_time=5.386
+[gpub022:0/16] 2024-02-02 00:28:43,713 (trainer:737) INFO: 19epoch:train:8801-8900batch: iter_time=9.261e-05, forward_time=0.293, loss_ctc=57.141, loss_att=56.897, acc=0.732, loss=56.971, backward_time=0.408, grad_norm=39.107, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.319e-04, train_time=1.313
+[gpub022:0/16] 2024-02-02 00:31:41,232 (trainer:737) INFO: 19epoch:train:8901-9000batch: iter_time=8.437e-05, forward_time=0.400, loss_ctc=45.903, loss_att=52.659, acc=0.738, loss=50.632, backward_time=0.466, grad_norm=35.977, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.319e-04, train_time=1.775
+[gpub022:0/16] 2024-02-02 00:34:18,894 (trainer:737) INFO: 19epoch:train:9001-9100batch: iter_time=8.357e-05, forward_time=0.289, loss_ctc=51.539, loss_att=44.647, acc=0.741, loss=46.715, backward_time=0.402, grad_norm=34.276, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.319e-04, train_time=1.576
+[gpub022:0/16] 2024-02-02 00:36:34,821 (trainer:737) INFO: 19epoch:train:9101-9200batch: iter_time=7.956e-05, forward_time=0.291, loss_ctc=50.207, loss_att=53.183, acc=0.721, loss=52.291, backward_time=0.404, grad_norm=34.317, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.318e-04, train_time=1.360
+[gpub022:0/16] 2024-02-02 00:39:24,678 (trainer:737) INFO: 19epoch:train:9201-9300batch: iter_time=7.909e-05, forward_time=0.377, loss_ctc=51.097, loss_att=46.234, acc=0.734, loss=47.693, backward_time=0.462, grad_norm=46.858, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.318e-04, train_time=1.698
+[gpub022:0/16] 2024-02-02 00:41:56,859 (trainer:737) INFO: 19epoch:train:9301-9400batch: iter_time=8.178e-05, forward_time=0.290, loss_ctc=51.869, loss_att=47.495, acc=0.746, loss=48.807, backward_time=0.404, grad_norm=35.881, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.317e-04, train_time=1.522
+[gpub022:0/16] 2024-02-02 00:44:29,128 (trainer:737) INFO: 19epoch:train:9401-9500batch: iter_time=7.707e-05, forward_time=0.289, loss_ctc=45.361, loss_att=40.464, acc=0.750, loss=41.933, backward_time=0.402, grad_norm=30.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.317e-04, train_time=1.522
+[gpub022:0/16] 2024-02-02 00:47:00,242 (trainer:737) INFO: 19epoch:train:9501-9600batch: iter_time=9.632e-05, forward_time=0.367, loss_ctc=50.758, loss_att=48.663, acc=0.731, loss=49.291, backward_time=0.439, grad_norm=36.297, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.317e-04, train_time=1.511
+[gpub022:0/16] 2024-02-02 00:49:40,115 (trainer:737) INFO: 19epoch:train:9601-9700batch: iter_time=1.083e-04, forward_time=0.292, loss_ctc=58.793, loss_att=54.433, acc=0.733, loss=55.741, backward_time=0.404, grad_norm=48.878, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.316e-04, train_time=1.598
+[gpub022:0/16] 2024-02-02 00:52:10,690 (trainer:737) INFO: 19epoch:train:9701-9800batch: iter_time=8.419e-05, forward_time=0.288, loss_ctc=43.335, loss_att=43.843, acc=0.759, loss=43.691, backward_time=0.400, grad_norm=28.884, clip=100.000, loss_scale=8.152e+33, optim_step_time=0.093, optim0_lr0=2.316e-04, train_time=1.506
+[gpub022:0/16] 2024-02-02 00:54:25,154 (trainer:737) INFO: 19epoch:train:9801-9900batch: iter_time=8.725e-05, forward_time=0.293, loss_ctc=42.199, loss_att=42.889, acc=0.740, loss=42.682, backward_time=0.403, grad_norm=30.880, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.315e-04, train_time=1.344
+[gpub022:0/16] 2024-02-02 00:57:20,942 (trainer:737) INFO: 19epoch:train:9901-10000batch: iter_time=1.089e-04, forward_time=0.355, loss_ctc=48.469, loss_att=45.492, acc=0.728, loss=46.385, backward_time=0.452, grad_norm=37.311, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.315e-04, train_time=1.758
+[gpub022:0/16] 2024-02-02 00:57:41,252 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub022:0/16] 2024-02-02 00:57:59,986 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 00:58:03,504 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe00f4035e0>)
+[gpub022:0/16] 2024-02-02 00:58:03,504 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub022:0/16] 2024-02-02 00:58:03,507 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 01:05:45,389 (trainer:737) INFO: 19epoch:train:10001-10100batch: iter_time=3.636, forward_time=0.290, loss_ctc=45.686, loss_att=42.376, acc=0.762, loss=43.369, backward_time=0.403, grad_norm=31.223, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.314e-04, train_time=5.044
+[gpub022:0/16] 2024-02-02 01:08:20,108 (trainer:737) INFO: 19epoch:train:10101-10200batch: iter_time=8.690e-05, forward_time=0.315, loss_ctc=58.315, loss_att=62.051, acc=0.734, loss=60.930, backward_time=0.413, grad_norm=39.469, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.314e-04, train_time=1.546
+[gpub022:0/16] 2024-02-02 01:11:05,161 (trainer:737) INFO: 19epoch:train:10201-10300batch: iter_time=8.551e-05, forward_time=0.345, loss_ctc=49.361, loss_att=49.245, acc=0.745, loss=49.280, backward_time=0.458, grad_norm=31.793, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.314e-04, train_time=1.651
+[gpub022:0/16] 2024-02-02 01:13:40,938 (trainer:737) INFO: 19epoch:train:10301-10400batch: iter_time=9.123e-05, forward_time=0.289, loss_ctc=45.798, loss_att=40.588, acc=0.762, loss=42.151, backward_time=0.403, grad_norm=29.830, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.313e-04, train_time=1.558
+[gpub022:0/16] 2024-02-02 01:16:01,839 (trainer:737) INFO: 19epoch:train:10401-10500batch: iter_time=9.357e-05, forward_time=0.298, loss_ctc=54.735, loss_att=56.469, acc=0.722, loss=55.949, backward_time=0.412, grad_norm=39.090, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.313e-04, train_time=1.407
+[gpub022:0/16] 2024-02-02 01:18:47,560 (trainer:737) INFO: 19epoch:train:10501-10600batch: iter_time=9.121e-05, forward_time=0.382, loss_ctc=48.835, loss_att=48.327, acc=0.738, loss=48.479, backward_time=0.440, grad_norm=39.067, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.312e-04, train_time=1.659
+[gpub022:0/16] 2024-02-02 01:21:19,461 (trainer:737) INFO: 19epoch:train:10601-10700batch: iter_time=1.019e-04, forward_time=0.291, loss_ctc=49.721, loss_att=47.693, acc=0.751, loss=48.301, backward_time=0.403, grad_norm=32.654, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.312e-04, train_time=1.519
+[gpub022:0/16] 2024-02-02 01:22:25,271 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 01:24:02,530 (trainer:737) INFO: 19epoch:train:10701-10800batch: iter_time=1.068e-04, forward_time=0.289, loss_ctc=43.977, loss_att=40.406, acc=0.760, loss=41.477, backward_time=0.401, grad_norm=29.811, clip=100.000, loss_scale=7.500e+33, optim_step_time=0.093, optim0_lr0=2.312e-04, train_time=1.631
+[gpub022:0/16] 2024-02-02 01:26:39,391 (trainer:737) INFO: 19epoch:train:10801-10900batch: iter_time=9.522e-05, forward_time=0.402, loss_ctc=60.080, loss_att=55.240, acc=0.732, loss=56.692, backward_time=0.449, grad_norm=46.058, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.311e-04, train_time=1.568
+[gpub022:0/16] 2024-02-02 01:29:02,759 (trainer:737) INFO: 19epoch:train:10901-11000batch: iter_time=8.972e-05, forward_time=0.293, loss_ctc=50.718, loss_att=47.466, acc=0.761, loss=48.442, backward_time=0.406, grad_norm=34.293, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.311e-04, train_time=1.433
+[gpub022:0/16] 2024-02-02 01:31:38,984 (trainer:737) INFO: 19epoch:train:11001-11100batch: iter_time=9.368e-05, forward_time=0.291, loss_ctc=41.159, loss_att=42.735, acc=0.748, loss=42.262, backward_time=0.399, grad_norm=29.577, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.310e-04, train_time=1.562
+[gpub022:0/16] 2024-02-02 01:34:01,792 (trainer:737) INFO: 19epoch:train:11101-11200batch: iter_time=8.723e-05, forward_time=0.357, loss_ctc=41.961, loss_att=42.723, acc=0.747, loss=42.494, backward_time=0.433, grad_norm=31.941, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.310e-04, train_time=1.428
+[gpub022:0/16] 2024-02-02 01:35:48,615 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub022:0/16] 2024-02-02 01:36:07,356 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 01:36:10,886 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe0428354e0>)
+[gpub022:0/16] 2024-02-02 01:36:10,886 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub022:0/16] 2024-02-02 01:36:10,889 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 01:42:24,937 (trainer:737) INFO: 19epoch:train:11201-11300batch: iter_time=3.430, forward_time=0.315, loss_ctc=51.072, loss_att=48.087, acc=0.747, loss=48.983, backward_time=0.416, grad_norm=35.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.310e-04, train_time=5.031
+[gpub022:0/16] 2024-02-02 01:44:41,501 (trainer:737) INFO: 19epoch:train:11301-11400batch: iter_time=8.714e-05, forward_time=0.292, loss_ctc=57.495, loss_att=54.693, acc=0.747, loss=55.534, backward_time=0.407, grad_norm=37.999, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.309e-04, train_time=1.365
+[gpub022:0/16] 2024-02-02 01:47:20,783 (trainer:737) INFO: 19epoch:train:11401-11500batch: iter_time=1.027e-04, forward_time=0.365, loss_ctc=45.680, loss_att=52.373, acc=0.744, loss=50.365, backward_time=0.461, grad_norm=32.650, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.309e-04, train_time=1.593
+[gpub022:0/16] 2024-02-02 01:50:03,670 (trainer:737) INFO: 19epoch:train:11501-11600batch: iter_time=1.215e-04, forward_time=0.291, loss_ctc=51.076, loss_att=44.810, acc=0.751, loss=46.690, backward_time=0.405, grad_norm=32.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.308e-04, train_time=1.629
+[gpub022:0/16] 2024-02-02 01:52:23,515 (trainer:737) INFO: 19epoch:train:11601-11700batch: iter_time=1.137e-04, forward_time=0.292, loss_ctc=49.658, loss_att=52.312, acc=0.734, loss=51.516, backward_time=0.405, grad_norm=33.534, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.308e-04, train_time=1.398
+[gpub022:0/16] 2024-02-02 01:55:10,867 (trainer:737) INFO: 19epoch:train:11701-11800batch: iter_time=1.153e-04, forward_time=0.410, loss_ctc=50.033, loss_att=46.712, acc=0.733, loss=47.708, backward_time=0.434, grad_norm=41.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.307e-04, train_time=1.673
+[gpub022:0/16] 2024-02-02 01:56:52,082 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 01:57:51,090 (trainer:737) INFO: 19epoch:train:11801-11900batch: iter_time=1.004e-04, forward_time=0.290, loss_ctc=52.017, loss_att=49.650, acc=0.755, loss=50.360, backward_time=0.404, grad_norm=34.793, clip=100.000, loss_scale=4.012e+33, optim_step_time=0.093, optim0_lr0=2.307e-04, train_time=1.602
+[gpub022:0/16] 2024-02-02 02:00:10,246 (trainer:737) INFO: 19epoch:train:11901-12000batch: iter_time=9.825e-05, forward_time=0.290, loss_ctc=45.123, loss_att=42.005, acc=0.752, loss=42.940, backward_time=0.404, grad_norm=28.908, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.307e-04, train_time=1.391
+[gpub022:0/16] 2024-02-02 02:03:02,956 (trainer:737) INFO: 19epoch:train:12001-12100batch: iter_time=9.953e-05, forward_time=0.387, loss_ctc=50.842, loss_att=49.167, acc=0.738, loss=49.670, backward_time=0.450, grad_norm=35.119, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=2.306e-04, train_time=1.727
+[gpub022:0/16] 2024-02-02 02:05:35,787 (trainer:737) INFO: 19epoch:train:12101-12200batch: iter_time=9.826e-05, forward_time=0.292, loss_ctc=58.800, loss_att=54.413, acc=0.743, loss=55.729, backward_time=0.405, grad_norm=53.766, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.306e-04, train_time=1.528
+[gpub022:0/16] 2024-02-02 02:08:16,019 (trainer:737) INFO: 19epoch:train:12201-12300batch: iter_time=1.005e-04, forward_time=0.309, loss_ctc=42.671, loss_att=44.475, acc=0.763, loss=43.934, backward_time=0.402, grad_norm=28.787, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.305e-04, train_time=1.602
+[gpub022:0/16] 2024-02-02 02:10:46,903 (trainer:737) INFO: 19epoch:train:12301-12400batch: iter_time=1.068e-04, forward_time=0.358, loss_ctc=42.285, loss_att=42.604, acc=0.747, loss=42.508, backward_time=0.457, grad_norm=29.435, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.305e-04, train_time=1.509
+[gpub022:0/16] 2024-02-02 02:13:27,147 (trainer:737) INFO: 19epoch:train:12401-12500batch: iter_time=9.812e-05, forward_time=0.292, loss_ctc=48.702, loss_att=44.620, acc=0.743, loss=45.845, backward_time=0.403, grad_norm=37.152, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.305e-04, train_time=1.602
+[gpub022:0/16] 2024-02-02 02:13:47,332 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub022:0/16] 2024-02-02 02:14:06,155 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 02:14:09,677 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe090e537c0>)
+[gpub022:0/16] 2024-02-02 02:14:09,677 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub022:0/16] 2024-02-02 02:14:09,680 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 02:21:30,013 (trainer:737) INFO: 19epoch:train:12501-12600batch: iter_time=3.287, forward_time=0.288, loss_ctc=45.028, loss_att=41.191, acc=0.763, loss=42.342, backward_time=0.404, grad_norm=30.494, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.304e-04, train_time=4.828
+[gpub022:0/16] 2024-02-02 02:24:37,682 (trainer:737) INFO: 19epoch:train:12601-12700batch: iter_time=8.689e-05, forward_time=0.390, loss_ctc=58.026, loss_att=61.264, acc=0.735, loss=60.293, backward_time=0.457, grad_norm=37.939, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=2.304e-04, train_time=1.877
+[gpub022:0/16] 2024-02-02 02:24:58,653 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 02:26:53,015 (trainer:737) INFO: 19epoch:train:12701-12800batch: iter_time=8.733e-05, forward_time=0.291, loss_ctc=49.704, loss_att=49.580, acc=0.746, loss=49.617, backward_time=0.404, grad_norm=31.014, clip=100.000, loss_scale=1.469e+33, optim_step_time=0.093, optim0_lr0=2.303e-04, train_time=1.353
+[gpub022:0/16] 2024-02-02 02:29:25,994 (trainer:737) INFO: 19epoch:train:12801-12900batch: iter_time=9.223e-05, forward_time=0.290, loss_ctc=46.068, loss_att=40.563, acc=0.761, loss=42.215, backward_time=0.404, grad_norm=28.964, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.303e-04, train_time=1.530
+[gpub022:0/16] 2024-02-02 02:32:13,656 (trainer:737) INFO: 19epoch:train:12901-13000batch: iter_time=9.104e-05, forward_time=0.351, loss_ctc=54.515, loss_att=56.173, acc=0.724, loss=55.676, backward_time=0.446, grad_norm=38.477, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.100, optim0_lr0=2.303e-04, train_time=1.677
+[gpub022:0/16] 2024-02-02 02:34:43,490 (trainer:737) INFO: 19epoch:train:13001-13100batch: iter_time=8.844e-05, forward_time=0.298, loss_ctc=48.822, loss_att=47.594, acc=0.738, loss=47.963, backward_time=0.406, grad_norm=38.246, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.302e-04, train_time=1.498
+[gpub022:0/16] 2024-02-02 02:37:12,306 (trainer:737) INFO: 19epoch:train:13101-13200batch: iter_time=1.005e-04, forward_time=0.291, loss_ctc=49.627, loss_att=47.959, acc=0.750, loss=48.459, backward_time=0.406, grad_norm=32.737, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.302e-04, train_time=1.488
+[gpub022:0/16] 2024-02-02 02:40:00,962 (trainer:737) INFO: 19epoch:train:13201-13300batch: iter_time=7.429e-04, forward_time=0.339, loss_ctc=44.288, loss_att=40.151, acc=0.760, loss=41.392, backward_time=0.453, grad_norm=30.106, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.100, optim0_lr0=2.301e-04, train_time=1.686
+[gpub022:0/16] 2024-02-02 02:42:16,236 (trainer:737) INFO: 19epoch:train:13301-13400batch: iter_time=7.070e-04, forward_time=0.322, loss_ctc=59.059, loss_att=53.551, acc=0.732, loss=55.203, backward_time=0.411, grad_norm=49.476, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.097, optim0_lr0=2.301e-04, train_time=1.353
+[gpub022:0/16] 2024-02-02 02:45:08,521 (trainer:737) INFO: 19epoch:train:13401-13500batch: iter_time=9.401e-05, forward_time=0.292, loss_ctc=50.908, loss_att=47.357, acc=0.760, loss=48.422, backward_time=0.404, grad_norm=36.621, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.301e-04, train_time=1.723
+[gpub022:0/16] 2024-02-02 02:47:37,563 (trainer:737) INFO: 19epoch:train:13501-13600batch: iter_time=8.361e-05, forward_time=0.367, loss_ctc=41.012, loss_att=42.524, acc=0.750, loss=42.070, backward_time=0.430, grad_norm=29.083, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.101, optim0_lr0=2.300e-04, train_time=1.490
+[gpub022:0/16] 2024-02-02 02:50:01,159 (trainer:737) INFO: 19epoch:train:13601-13700batch: iter_time=5.522e-04, forward_time=0.334, loss_ctc=41.671, loss_att=42.527, acc=0.750, loss=42.270, backward_time=0.419, grad_norm=32.174, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.095, optim0_lr0=2.300e-04, train_time=1.436
+[gpub022:0/16] 2024-02-02 02:51:45,083 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub022:0/16] 2024-02-02 02:52:04,115 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 02:52:07,824 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe03b279180>)
+[gpub022:0/16] 2024-02-02 02:52:07,824 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub022:0/16] 2024-02-02 02:52:07,828 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 02:58:04,910 (trainer:737) INFO: 19epoch:train:13701-13800batch: iter_time=3.222, forward_time=0.291, loss_ctc=50.393, loss_att=47.101, acc=0.751, loss=48.089, backward_time=0.404, grad_norm=34.664, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.299e-04, train_time=4.837
+[gpub022:0/16] 2024-02-02 03:00:43,984 (trainer:737) INFO: 19epoch:train:13801-13900batch: iter_time=2.043e-04, forward_time=0.398, loss_ctc=56.242, loss_att=58.294, acc=0.730, loss=57.678, backward_time=0.436, grad_norm=39.565, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.096, optim0_lr0=2.299e-04, train_time=1.590
+[gpub022:0/16] 2024-02-02 03:03:19,032 (trainer:737) INFO: 19epoch:train:13901-14000batch: iter_time=8.807e-05, forward_time=0.304, loss_ctc=45.497, loss_att=52.649, acc=0.740, loss=50.503, backward_time=0.411, grad_norm=32.818, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.094, optim0_lr0=2.299e-04, train_time=1.551
+[gpub022:0/16] 2024-02-02 03:05:54,300 (trainer:737) INFO: 19epoch:train:14001-14100batch: iter_time=9.751e-05, forward_time=0.289, loss_ctc=50.863, loss_att=44.970, acc=0.740, loss=46.738, backward_time=0.403, grad_norm=34.511, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.094, optim0_lr0=2.298e-04, train_time=1.553
+[gpub022:0/16] 2024-02-02 03:08:56,385 (trainer:737) INFO: 19epoch:train:14101-14200batch: iter_time=1.484e-04, forward_time=0.385, loss_ctc=50.070, loss_att=53.098, acc=0.724, loss=52.190, backward_time=0.427, grad_norm=35.970, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.102, optim0_lr0=2.298e-04, train_time=1.820
+[gpub022:0/16] 2024-02-02 03:11:45,919 (trainer:737) INFO: 19epoch:train:14201-14300batch: iter_time=9.874e-05, forward_time=0.287, loss_ctc=50.190, loss_att=46.021, acc=0.737, loss=47.272, backward_time=0.401, grad_norm=42.176, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.297e-04, train_time=1.696
+[gpub022:0/16] 2024-02-02 03:14:37,252 (trainer:737) INFO: 19epoch:train:14301-14400batch: iter_time=0.001, forward_time=0.392, loss_ctc=51.394, loss_att=47.500, acc=0.748, loss=48.669, backward_time=0.455, grad_norm=34.089, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.297e-04, train_time=1.713
+[gpub022:0/16] 2024-02-02 03:17:32,916 (trainer:737) INFO: 19epoch:train:14401-14500batch: iter_time=8.819e-05, forward_time=0.289, loss_ctc=44.962, loss_att=40.311, acc=0.751, loss=41.706, backward_time=0.401, grad_norm=29.958, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.297e-04, train_time=1.756
+[gpub022:0/16] 2024-02-02 03:19:53,495 (trainer:737) INFO: 19epoch:train:14501-14600batch: iter_time=9.690e-05, forward_time=0.360, loss_ctc=51.443, loss_att=49.503, acc=0.731, loss=50.085, backward_time=0.429, grad_norm=37.968, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.099, optim0_lr0=2.296e-04, train_time=1.406
+[gpub022:0/16] 2024-02-02 03:22:37,177 (trainer:737) INFO: 19epoch:train:14601-14700batch: iter_time=4.917e-04, forward_time=0.314, loss_ctc=58.855, loss_att=54.946, acc=0.734, loss=56.118, backward_time=0.415, grad_norm=49.948, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.097, optim0_lr0=2.296e-04, train_time=1.636
+[gpub022:0/16] 2024-02-02 03:25:23,829 (trainer:737) INFO: 19epoch:train:14701-14800batch: iter_time=1.072e-04, forward_time=0.328, loss_ctc=42.524, loss_att=43.791, acc=0.761, loss=43.411, backward_time=0.419, grad_norm=28.581, clip=100.000, loss_scale=2.414e+33, optim_step_time=0.102, optim0_lr0=2.295e-04, train_time=1.666
+[gpub022:0/16] 2024-02-02 03:27:57,943 (trainer:737) INFO: 19epoch:train:14801-14900batch: iter_time=3.013e-04, forward_time=0.333, loss_ctc=42.092, loss_att=43.417, acc=0.741, loss=43.019, backward_time=0.424, grad_norm=30.493, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.295e-04, train_time=1.541
+[gpub022:0/16] 2024-02-02 03:30:31,927 (trainer:737) INFO: 19epoch:train:14901-15000batch: iter_time=1.003e-04, forward_time=0.321, loss_ctc=48.714, loss_att=45.655, acc=0.729, loss=46.573, backward_time=0.414, grad_norm=38.917, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.294e-04, train_time=1.540
+[gpub022:0/16] 2024-02-02 04:08:16,350 (trainer:343) INFO: 19epoch results: [train] iter_time=0.267, forward_time=0.325, loss_ctc=50.244, loss_att=48.350, acc=0.741, loss=48.918, backward_time=0.419, grad_norm=36.355, clip=100.000, loss_scale=5.218e+33, optim_step_time=0.096, optim0_lr0=2.325e-04, train_time=1.816, time=7 hours, 34 minutes and 30.12 seconds, total_count=315000, gpu_max_cached_mem_GB=41.891, [valid] loss_ctc=40.729, cer_ctc=0.208, loss_att=43.390, acc=0.656, cer=0.299, wer=0.993, loss=42.592, time=37 minutes and 19.77 seconds, total_count=98091, gpu_max_cached_mem_GB=41.891
+[gpub022:0/16] 2024-02-02 04:08:26,811 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub022:0/16] 2024-02-02 04:08:26,975 (trainer:272) INFO: 20/45epoch started. Estimated time to finish: 1 week, 2 days and 7 hours
+[gpub022:0/16] 2024-02-02 04:08:27,170 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub022:0/16] 2024-02-02 04:08:45,084 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 04:08:48,466 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85748c9a0>)
+[gpub022:0/16] 2024-02-02 04:08:48,466 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub022:0/16] 2024-02-02 04:08:48,470 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 04:15:25,586 (trainer:737) INFO: 20epoch:train:1-100batch: iter_time=2.754, forward_time=0.328, loss_ctc=43.687, loss_att=44.593, acc=0.730, loss=44.321, backward_time=0.406, grad_norm=32.312, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.294e-04, train_time=4.185
+[gpub022:0/16] 2024-02-02 04:18:04,680 (trainer:737) INFO: 20epoch:train:101-200batch: iter_time=6.730e-04, forward_time=0.354, loss_ctc=44.083, loss_att=42.887, acc=0.746, loss=43.246, backward_time=0.411, grad_norm=32.758, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.294e-04, train_time=1.590
+[gpub022:0/16] 2024-02-02 04:20:42,961 (trainer:737) INFO: 20epoch:train:201-300batch: iter_time=1.989e-04, forward_time=0.368, loss_ctc=52.078, loss_att=47.727, acc=0.743, loss=49.032, backward_time=0.425, grad_norm=36.511, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.293e-04, train_time=1.583
+[gpub022:0/16] 2024-02-02 04:23:20,699 (trainer:737) INFO: 20epoch:train:301-400batch: iter_time=1.058e-04, forward_time=0.312, loss_ctc=59.159, loss_att=53.034, acc=0.736, loss=54.871, backward_time=0.424, grad_norm=40.122, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.293e-04, train_time=1.577
+[gpub022:0/16] 2024-02-02 04:26:02,099 (trainer:737) INFO: 20epoch:train:401-500batch: iter_time=5.419e-04, forward_time=0.328, loss_ctc=52.431, loss_att=50.398, acc=0.753, loss=51.008, backward_time=0.454, grad_norm=40.787, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.292e-04, train_time=1.611
+[gpub022:0/16] 2024-02-02 04:28:35,659 (trainer:737) INFO: 20epoch:train:501-600batch: iter_time=4.666e-04, forward_time=0.350, loss_ctc=49.950, loss_att=45.587, acc=0.738, loss=46.896, backward_time=0.411, grad_norm=33.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.292e-04, train_time=1.535
+[gpub022:0/16] 2024-02-02 04:31:15,787 (trainer:737) INFO: 20epoch:train:601-700batch: iter_time=9.214e-04, forward_time=0.335, loss_ctc=60.390, loss_att=58.162, acc=0.721, loss=58.830, backward_time=0.442, grad_norm=41.871, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.292e-04, train_time=1.602
+[gpub022:0/16] 2024-02-02 04:33:56,489 (trainer:737) INFO: 20epoch:train:701-800batch: iter_time=4.174e-04, forward_time=0.338, loss_ctc=51.298, loss_att=58.457, acc=0.717, loss=56.310, backward_time=0.415, grad_norm=37.837, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.291e-04, train_time=1.607
+[gpub022:0/16] 2024-02-02 04:36:40,968 (trainer:737) INFO: 20epoch:train:801-900batch: iter_time=4.293e-04, forward_time=0.343, loss_ctc=56.273, loss_att=59.405, acc=0.721, loss=58.465, backward_time=0.445, grad_norm=37.346, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.291e-04, train_time=1.644
+[gpub022:0/16] 2024-02-02 04:39:15,152 (trainer:737) INFO: 20epoch:train:901-1000batch: iter_time=5.310e-04, forward_time=0.333, loss_ctc=66.196, loss_att=60.288, acc=0.711, loss=62.061, backward_time=0.418, grad_norm=42.392, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.290e-04, train_time=1.541
+[gpub022:0/16] 2024-02-02 04:42:01,412 (trainer:737) INFO: 20epoch:train:1001-1100batch: iter_time=6.978e-04, forward_time=0.346, loss_ctc=52.569, loss_att=47.068, acc=0.736, loss=48.718, backward_time=0.423, grad_norm=34.898, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.290e-04, train_time=1.664
+[gpub022:0/16] 2024-02-02 04:44:33,015 (trainer:737) INFO: 20epoch:train:1101-1200batch: iter_time=3.511e-04, forward_time=0.322, loss_ctc=54.076, loss_att=52.745, acc=0.718, loss=53.144, backward_time=0.416, grad_norm=39.824, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.290e-04, train_time=1.514
+[gpub022:0/16] 2024-02-02 04:46:11,919 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub022:0/16] 2024-02-02 04:46:30,803 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 04:46:34,345 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe2b32abf70>)
+[gpub022:0/16] 2024-02-02 04:46:34,345 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub022:0/16] 2024-02-02 04:46:34,435 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 04:52:17,100 (trainer:737) INFO: 20epoch:train:1201-1300batch: iter_time=3.074, forward_time=0.297, loss_ctc=47.802, loss_att=45.092, acc=0.755, loss=45.905, backward_time=0.406, grad_norm=30.654, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.289e-04, train_time=4.642
+[gpub022:0/16] 2024-02-02 04:54:30,176 (trainer:737) INFO: 20epoch:train:1301-1400batch: iter_time=8.401e-05, forward_time=0.296, loss_ctc=46.352, loss_att=48.871, acc=0.748, loss=48.115, backward_time=0.411, grad_norm=33.016, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.289e-04, train_time=1.331
+[gpub022:0/16] 2024-02-02 04:57:07,235 (trainer:737) INFO: 20epoch:train:1401-1500batch: iter_time=3.670e-04, forward_time=0.379, loss_ctc=46.567, loss_att=41.687, acc=0.763, loss=43.151, backward_time=0.432, grad_norm=32.188, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.288e-04, train_time=1.570
+[gpub022:0/16] 2024-02-02 04:59:45,039 (trainer:737) INFO: 20epoch:train:1501-1600batch: iter_time=9.269e-05, forward_time=0.324, loss_ctc=49.825, loss_att=51.840, acc=0.738, loss=51.235, backward_time=0.410, grad_norm=36.161, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.288e-04, train_time=1.577
+[gpub022:0/16] 2024-02-02 05:01:58,765 (trainer:737) INFO: 20epoch:train:1601-1700batch: iter_time=1.052e-04, forward_time=0.295, loss_ctc=55.836, loss_att=53.794, acc=0.753, loss=54.407, backward_time=0.408, grad_norm=38.574, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.288e-04, train_time=1.338
+[gpub022:0/16] 2024-02-02 05:04:31,527 (trainer:737) INFO: 20epoch:train:1701-1800batch: iter_time=6.492e-04, forward_time=0.368, loss_ctc=52.106, loss_att=47.068, acc=0.753, loss=48.580, backward_time=0.439, grad_norm=34.419, clip=100.000, loss_scale=4.829e+33, optim_step_time=0.099, optim0_lr0=2.287e-04, train_time=1.528
+[gpub022:0/16] 2024-02-02 05:07:11,072 (trainer:737) INFO: 20epoch:train:1801-1900batch: iter_time=9.409e-05, forward_time=0.320, loss_ctc=51.681, loss_att=51.575, acc=0.746, loss=51.607, backward_time=0.424, grad_norm=33.245, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.287e-04, train_time=1.595
+[gpub022:0/16] 2024-02-02 05:09:30,155 (trainer:737) INFO: 20epoch:train:1901-2000batch: iter_time=1.030e-04, forward_time=0.292, loss_ctc=56.920, loss_att=58.719, acc=0.722, loss=58.180, backward_time=0.405, grad_norm=40.448, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.286e-04, train_time=1.390
+[gpub022:0/16] 2024-02-02 05:11:51,913 (trainer:737) INFO: 20epoch:train:2001-2100batch: iter_time=4.340e-04, forward_time=0.300, loss_ctc=51.268, loss_att=59.486, acc=0.732, loss=57.021, backward_time=0.419, grad_norm=34.757, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.286e-04, train_time=1.418
+[gpub022:0/16] 2024-02-02 05:14:34,232 (trainer:737) INFO: 20epoch:train:2101-2200batch: iter_time=8.974e-05, forward_time=0.353, loss_ctc=52.717, loss_att=55.029, acc=0.739, loss=54.335, backward_time=0.438, grad_norm=32.794, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.286e-04, train_time=1.623
+[gpub022:0/16] 2024-02-02 05:16:58,035 (trainer:737) INFO: 20epoch:train:2201-2300batch: iter_time=8.969e-05, forward_time=0.318, loss_ctc=64.003, loss_att=57.765, acc=0.733, loss=59.636, backward_time=0.420, grad_norm=40.081, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.285e-04, train_time=1.438
+[gpub022:0/16] 2024-02-02 05:19:21,589 (trainer:737) INFO: 20epoch:train:2301-2400batch: iter_time=3.113e-04, forward_time=0.293, loss_ctc=50.893, loss_att=49.111, acc=0.737, loss=49.645, backward_time=0.412, grad_norm=32.866, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.285e-04, train_time=1.435
+[gpub022:0/16] 2024-02-02 05:21:49,831 (trainer:737) INFO: 20epoch:train:2401-2500batch: iter_time=1.009e-04, forward_time=0.335, loss_ctc=59.208, loss_att=58.950, acc=0.729, loss=59.027, backward_time=0.456, grad_norm=41.123, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.284e-04, train_time=1.482
+[gpub022:0/16] 2024-02-02 05:22:09,859 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub022:0/16] 2024-02-02 05:22:28,533 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 05:22:32,149 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd83bc7ffd0>)
+[gpub022:0/16] 2024-02-02 05:22:32,149 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub022:0/16] 2024-02-02 05:22:32,152 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 05:29:07,951 (trainer:737) INFO: 20epoch:train:2501-2600batch: iter_time=2.918, forward_time=0.305, loss_ctc=42.802, loss_att=44.127, acc=0.750, loss=43.730, backward_time=0.401, grad_norm=30.235, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.284e-04, train_time=4.381
+[gpub022:0/16] 2024-02-02 05:31:30,380 (trainer:737) INFO: 20epoch:train:2601-2700batch: iter_time=3.768e-04, forward_time=0.302, loss_ctc=42.909, loss_att=42.214, acc=0.755, loss=42.422, backward_time=0.411, grad_norm=31.105, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.284e-04, train_time=1.424
+[gpub022:0/16] 2024-02-02 05:33:57,265 (trainer:737) INFO: 20epoch:train:2701-2800batch: iter_time=8.778e-05, forward_time=0.295, loss_ctc=49.604, loss_att=45.661, acc=0.758, loss=46.844, backward_time=0.411, grad_norm=34.217, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.283e-04, train_time=1.468
+[gpub022:0/16] 2024-02-02 05:36:31,000 (trainer:737) INFO: 20epoch:train:2801-2900batch: iter_time=1.015e-04, forward_time=0.372, loss_ctc=55.707, loss_att=51.377, acc=0.752, loss=52.676, backward_time=0.433, grad_norm=37.090, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.283e-04, train_time=1.537
+[gpub022:0/16] 2024-02-02 05:38:51,900 (trainer:737) INFO: 20epoch:train:2901-3000batch: iter_time=1.093e-04, forward_time=0.314, loss_ctc=50.957, loss_att=51.052, acc=0.759, loss=51.023, backward_time=0.408, grad_norm=33.149, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.283e-04, train_time=1.409
+[gpub022:0/16] 2024-02-02 05:41:15,325 (trainer:737) INFO: 20epoch:train:3001-3100batch: iter_time=4.406e-04, forward_time=0.294, loss_ctc=48.908, loss_att=45.431, acc=0.749, loss=46.474, backward_time=0.408, grad_norm=30.674, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.282e-04, train_time=1.434
+[gpub022:0/16] 2024-02-02 05:43:53,774 (trainer:737) INFO: 20epoch:train:3101-3200batch: iter_time=9.799e-05, forward_time=0.333, loss_ctc=57.179, loss_att=57.449, acc=0.733, loss=57.368, backward_time=0.463, grad_norm=37.293, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.282e-04, train_time=1.584
+[gpub022:0/16] 2024-02-02 05:46:26,273 (trainer:737) INFO: 20epoch:train:3201-3300batch: iter_time=1.114e-04, forward_time=0.310, loss_ctc=50.524, loss_att=57.111, acc=0.735, loss=55.135, backward_time=0.410, grad_norm=38.142, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.281e-04, train_time=1.525
+[gpub022:0/16] 2024-02-02 05:48:44,100 (trainer:737) INFO: 20epoch:train:3301-3400batch: iter_time=3.322e-04, forward_time=0.304, loss_ctc=54.303, loss_att=57.955, acc=0.740, loss=56.859, backward_time=0.417, grad_norm=34.439, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.281e-04, train_time=1.378
+[gpub022:0/16] 2024-02-02 05:51:18,750 (trainer:737) INFO: 20epoch:train:3401-3500batch: iter_time=1.119e-04, forward_time=0.299, loss_ctc=63.733, loss_att=61.290, acc=0.720, loss=62.022, backward_time=0.411, grad_norm=39.924, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.281e-04, train_time=1.546
+[gpub022:0/16] 2024-02-02 05:53:53,999 (trainer:737) INFO: 20epoch:train:3501-3600batch: iter_time=1.117e-04, forward_time=0.359, loss_ctc=51.450, loss_att=45.665, acc=0.755, loss=47.401, backward_time=0.434, grad_norm=33.463, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.280e-04, train_time=1.552
+[gpub022:0/16] 2024-02-02 05:56:10,631 (trainer:737) INFO: 20epoch:train:3601-3700batch: iter_time=0.002, forward_time=0.316, loss_ctc=52.303, loss_att=52.574, acc=0.723, loss=52.493, backward_time=0.404, grad_norm=38.646, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.280e-04, train_time=1.366
+[gpub022:0/16] 2024-02-02 05:57:41,452 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub022:0/16] 2024-02-02 05:58:00,211 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 05:58:04,048 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85adb3f10>)
+[gpub022:0/16] 2024-02-02 05:58:04,048 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub022:0/16] 2024-02-02 05:58:04,052 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 06:03:36,308 (trainer:737) INFO: 20epoch:train:3701-3800batch: iter_time=3.045, forward_time=0.297, loss_ctc=46.973, loss_att=47.854, acc=0.751, loss=47.590, backward_time=0.409, grad_norm=30.469, clip=100.000, loss_scale=9.658e+33, optim_step_time=0.094, optim0_lr0=2.279e-04, train_time=4.457
+[gpub022:0/16] 2024-02-02 06:05:57,180 (trainer:737) INFO: 20epoch:train:3801-3900batch: iter_time=8.609e-05, forward_time=0.346, loss_ctc=46.100, loss_att=46.952, acc=0.755, loss=46.696, backward_time=0.423, grad_norm=32.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.279e-04, train_time=1.408
+[gpub022:0/16] 2024-02-02 06:06:03,055 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 06:08:14,283 (trainer:737) INFO: 20epoch:train:3901-4000batch: iter_time=9.268e-05, forward_time=0.307, loss_ctc=46.352, loss_att=40.836, acc=0.767, loss=42.491, backward_time=0.410, grad_norm=33.265, clip=100.000, loss_scale=5.350e+33, optim_step_time=0.093, optim0_lr0=2.279e-04, train_time=1.370
+[gpub022:0/16] 2024-02-02 06:10:56,541 (trainer:737) INFO: 20epoch:train:4001-4100batch: iter_time=7.729e-04, forward_time=0.312, loss_ctc=48.982, loss_att=49.802, acc=0.744, loss=49.556, backward_time=0.419, grad_norm=33.707, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.278e-04, train_time=1.622
+[gpub022:0/16] 2024-02-02 06:13:11,992 (trainer:737) INFO: 20epoch:train:4101-4200batch: iter_time=9.701e-05, forward_time=0.295, loss_ctc=53.890, loss_att=52.115, acc=0.757, loss=52.648, backward_time=0.414, grad_norm=37.700, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.278e-04, train_time=1.355
+[gpub022:0/16] 2024-02-02 06:15:48,812 (trainer:737) INFO: 20epoch:train:4201-4300batch: iter_time=8.522e-05, forward_time=0.332, loss_ctc=51.588, loss_att=46.403, acc=0.757, loss=47.958, backward_time=0.443, grad_norm=33.283, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.277e-04, train_time=1.568
+[gpub022:0/16] 2024-02-02 06:18:18,907 (trainer:737) INFO: 20epoch:train:4301-4400batch: iter_time=9.118e-05, forward_time=0.314, loss_ctc=51.136, loss_att=50.633, acc=0.750, loss=50.784, backward_time=0.410, grad_norm=31.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.277e-04, train_time=1.500
+[gpub022:0/16] 2024-02-02 06:20:39,206 (trainer:737) INFO: 20epoch:train:4401-4500batch: iter_time=2.389e-04, forward_time=0.297, loss_ctc=54.420, loss_att=57.373, acc=0.727, loss=56.487, backward_time=0.411, grad_norm=38.351, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.277e-04, train_time=1.403
+[gpub022:0/16] 2024-02-02 06:23:14,863 (trainer:737) INFO: 20epoch:train:4501-4600batch: iter_time=9.735e-05, forward_time=0.293, loss_ctc=50.334, loss_att=58.751, acc=0.735, loss=56.226, backward_time=0.408, grad_norm=33.699, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.276e-04, train_time=1.557
+[gpub022:0/16] 2024-02-02 06:25:50,501 (trainer:737) INFO: 20epoch:train:4601-4700batch: iter_time=9.006e-05, forward_time=0.339, loss_ctc=52.000, loss_att=53.754, acc=0.742, loss=53.228, backward_time=0.434, grad_norm=33.566, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.276e-04, train_time=1.556
+[gpub022:0/16] 2024-02-02 06:28:08,913 (trainer:737) INFO: 20epoch:train:4701-4800batch: iter_time=9.687e-05, forward_time=0.306, loss_ctc=63.050, loss_att=57.669, acc=0.733, loss=59.284, backward_time=0.413, grad_norm=40.376, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.275e-04, train_time=1.384
+[gpub022:0/16] 2024-02-02 06:30:43,455 (trainer:737) INFO: 20epoch:train:4801-4900batch: iter_time=4.671e-04, forward_time=0.306, loss_ctc=50.069, loss_att=49.114, acc=0.738, loss=49.400, backward_time=0.413, grad_norm=32.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.275e-04, train_time=1.545
+[gpub022:0/16] 2024-02-02 06:33:15,895 (trainer:737) INFO: 20epoch:train:4901-5000batch: iter_time=9.234e-05, forward_time=0.304, loss_ctc=57.411, loss_att=58.213, acc=0.731, loss=57.972, backward_time=0.423, grad_norm=38.633, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.275e-04, train_time=1.525
+[gpub022:0/16] 2024-02-02 06:33:35,950 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub022:0/16] 2024-02-02 06:33:54,979 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 06:33:58,794 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe2c0b9be20>)
+[gpub022:0/16] 2024-02-02 06:33:58,794 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub022:0/16] 2024-02-02 06:33:58,798 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 06:40:36,069 (trainer:737) INFO: 20epoch:train:5001-5100batch: iter_time=3.002, forward_time=0.335, loss_ctc=42.293, loss_att=45.627, acc=0.734, loss=44.627, backward_time=0.408, grad_norm=31.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.274e-04, train_time=4.401
+[gpub022:0/16] 2024-02-02 06:42:59,139 (trainer:737) INFO: 20epoch:train:5101-5200batch: iter_time=1.030e-04, forward_time=0.289, loss_ctc=42.948, loss_att=42.435, acc=0.750, loss=42.589, backward_time=0.405, grad_norm=31.019, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.274e-04, train_time=1.431
+[gpub022:0/16] 2024-02-02 06:45:53,964 (trainer:737) INFO: 20epoch:train:5201-5300batch: iter_time=9.435e-05, forward_time=0.383, loss_ctc=49.204, loss_att=46.358, acc=0.752, loss=47.212, backward_time=0.433, grad_norm=35.444, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.273e-04, train_time=1.747
+[gpub022:0/16] 2024-02-02 06:46:48,878 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 06:48:07,409 (trainer:737) INFO: 20epoch:train:5301-5400batch: iter_time=0.002, forward_time=0.293, loss_ctc=54.818, loss_att=51.827, acc=0.737, loss=52.724, backward_time=0.411, grad_norm=37.801, clip=100.000, loss_scale=3.671e+33, optim_step_time=0.093, optim0_lr0=2.273e-04, train_time=1.335
+[gpub022:0/16] 2024-02-02 06:50:41,885 (trainer:737) INFO: 20epoch:train:5401-5500batch: iter_time=9.991e-05, forward_time=0.343, loss_ctc=50.483, loss_att=49.180, acc=0.759, loss=49.571, backward_time=0.438, grad_norm=33.025, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.273e-04, train_time=1.545
+[gpub022:0/16] 2024-02-02 06:53:22,813 (trainer:737) INFO: 20epoch:train:5501-5600batch: iter_time=9.907e-05, forward_time=0.291, loss_ctc=47.934, loss_att=44.966, acc=0.744, loss=45.857, backward_time=0.406, grad_norm=31.445, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.272e-04, train_time=1.609
+[gpub022:0/16] 2024-02-02 06:55:43,230 (trainer:737) INFO: 20epoch:train:5601-5700batch: iter_time=8.396e-05, forward_time=0.291, loss_ctc=56.370, loss_att=57.211, acc=0.726, loss=56.959, backward_time=0.406, grad_norm=38.372, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.272e-04, train_time=1.404
+[gpub022:0/16] 2024-02-02 06:58:15,066 (trainer:737) INFO: 20epoch:train:5701-5800batch: iter_time=9.339e-05, forward_time=0.403, loss_ctc=48.793, loss_att=57.872, acc=0.724, loss=55.149, backward_time=0.424, grad_norm=35.691, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.272e-04, train_time=1.518
+[gpub022:0/16] 2024-02-02 07:01:09,570 (trainer:737) INFO: 20epoch:train:5801-5900batch: iter_time=8.993e-05, forward_time=0.296, loss_ctc=54.328, loss_att=58.278, acc=0.726, loss=57.093, backward_time=0.408, grad_norm=35.465, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.271e-04, train_time=1.745
+[gpub022:0/16] 2024-02-02 07:03:20,604 (trainer:737) INFO: 20epoch:train:5901-6000batch: iter_time=8.665e-05, forward_time=0.294, loss_ctc=62.813, loss_att=59.595, acc=0.714, loss=60.560, backward_time=0.414, grad_norm=41.847, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.271e-04, train_time=1.310
+[gpub022:0/16] 2024-02-02 07:05:54,198 (trainer:737) INFO: 20epoch:train:6001-6100batch: iter_time=1.005e-04, forward_time=0.359, loss_ctc=50.919, loss_att=45.218, acc=0.746, loss=46.928, backward_time=0.459, grad_norm=34.166, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.270e-04, train_time=1.536
+[gpub022:0/16] 2024-02-02 07:08:31,985 (trainer:737) INFO: 20epoch:train:6101-6200batch: iter_time=8.844e-05, forward_time=0.290, loss_ctc=52.138, loss_att=49.990, acc=0.728, loss=50.634, backward_time=0.404, grad_norm=37.222, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.270e-04, train_time=1.578
+[gpub022:0/16] 2024-02-02 07:09:59,923 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub022:0/16] 2024-02-02 07:10:19,021 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 07:10:22,968 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fdfe6b5b5e0>)
+[gpub022:0/16] 2024-02-02 07:10:22,968 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub022:0/16] 2024-02-02 07:10:22,971 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 07:16:10,263 (trainer:737) INFO: 20epoch:train:6201-6300batch: iter_time=3.131, forward_time=0.370, loss_ctc=46.698, loss_att=44.931, acc=0.749, loss=45.461, backward_time=0.422, grad_norm=29.430, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.270e-04, train_time=4.582
+[gpub022:0/16] 2024-02-02 07:18:51,280 (trainer:737) INFO: 20epoch:train:6301-6400batch: iter_time=9.772e-05, forward_time=0.290, loss_ctc=45.111, loss_att=45.068, acc=0.746, loss=45.081, backward_time=0.403, grad_norm=33.043, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.269e-04, train_time=1.610
+[gpub022:0/16] 2024-02-02 07:21:25,983 (trainer:737) INFO: 20epoch:train:6401-6500batch: iter_time=9.415e-05, forward_time=0.375, loss_ctc=45.620, loss_att=39.439, acc=0.768, loss=41.293, backward_time=0.428, grad_norm=30.806, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=2.269e-04, train_time=1.547
+[gpub022:0/16] 2024-02-02 07:23:56,398 (trainer:737) INFO: 20epoch:train:6501-6600batch: iter_time=9.089e-05, forward_time=0.292, loss_ctc=48.786, loss_att=50.346, acc=0.732, loss=49.878, backward_time=0.406, grad_norm=37.115, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.268e-04, train_time=1.504
+[gpub022:0/16] 2024-02-02 07:26:21,796 (trainer:737) INFO: 20epoch:train:6601-6700batch: iter_time=9.253e-05, forward_time=0.293, loss_ctc=53.387, loss_att=51.988, acc=0.749, loss=52.408, backward_time=0.407, grad_norm=44.189, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.268e-04, train_time=1.454
+[gpub022:0/16] 2024-02-02 07:28:53,938 (trainer:737) INFO: 20epoch:train:6701-6800batch: iter_time=9.299e-05, forward_time=0.341, loss_ctc=50.959, loss_att=44.483, acc=0.757, loss=46.426, backward_time=0.450, grad_norm=33.292, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.268e-04, train_time=1.521
+[gpub022:0/16] 2024-02-02 07:31:38,541 (trainer:737) INFO: 20epoch:train:6801-6900batch: iter_time=1.067e-04, forward_time=0.293, loss_ctc=50.597, loss_att=50.738, acc=0.739, loss=50.696, backward_time=0.406, grad_norm=32.732, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.267e-04, train_time=1.646
+[gpub022:0/16] 2024-02-02 07:33:48,520 (trainer:737) INFO: 20epoch:train:6901-7000batch: iter_time=9.288e-05, forward_time=0.291, loss_ctc=53.457, loss_att=56.937, acc=0.718, loss=55.893, backward_time=0.405, grad_norm=38.770, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.267e-04, train_time=1.300
+[gpub022:0/16] 2024-02-02 07:36:18,388 (trainer:737) INFO: 20epoch:train:7001-7100batch: iter_time=9.905e-05, forward_time=0.328, loss_ctc=49.553, loss_att=57.661, acc=0.721, loss=55.229, backward_time=0.453, grad_norm=34.438, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.266e-04, train_time=1.498
+[gpub022:0/16] 2024-02-02 07:39:02,629 (trainer:737) INFO: 20epoch:train:7101-7200batch: iter_time=9.348e-05, forward_time=0.292, loss_ctc=51.461, loss_att=51.083, acc=0.741, loss=51.197, backward_time=0.406, grad_norm=31.558, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.266e-04, train_time=1.642
+[gpub022:0/16] 2024-02-02 07:41:13,132 (trainer:737) INFO: 20epoch:train:7201-7300batch: iter_time=9.812e-05, forward_time=0.292, loss_ctc=61.001, loss_att=57.132, acc=0.725, loss=58.292, backward_time=0.405, grad_norm=39.089, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.266e-04, train_time=1.305
+[gpub022:0/16] 2024-02-02 07:43:53,188 (trainer:737) INFO: 20epoch:train:7301-7400batch: iter_time=1.038e-04, forward_time=0.377, loss_ctc=50.244, loss_att=47.907, acc=0.742, loss=48.608, backward_time=0.445, grad_norm=33.344, clip=100.000, loss_scale=4.102e+33, optim_step_time=0.101, optim0_lr0=2.265e-04, train_time=1.600
+[gpub022:0/16] 2024-02-02 07:46:35,636 (trainer:737) INFO: 20epoch:train:7401-7500batch: iter_time=8.906e-05, forward_time=0.293, loss_ctc=57.152, loss_att=54.377, acc=0.734, loss=55.210, backward_time=0.407, grad_norm=38.259, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.265e-04, train_time=1.624
+[gpub022:0/16] 2024-02-02 07:46:55,688 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub022:0/16] 2024-02-02 07:47:14,938 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 07:47:18,954 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe038f37970>)
+[gpub022:0/16] 2024-02-02 07:47:18,954 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub022:0/16] 2024-02-02 07:47:18,958 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 07:54:16,581 (trainer:737) INFO: 20epoch:train:7501-7600batch: iter_time=2.959, forward_time=0.402, loss_ctc=42.064, loss_att=42.673, acc=0.742, loss=42.490, backward_time=0.418, grad_norm=30.330, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.265e-04, train_time=4.609
+[gpub022:0/16] 2024-02-02 07:57:03,550 (trainer:737) INFO: 20epoch:train:7601-7700batch: iter_time=8.892e-05, forward_time=0.309, loss_ctc=42.370, loss_att=40.688, acc=0.757, loss=41.193, backward_time=0.435, grad_norm=32.242, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.264e-04, train_time=1.669
+[gpub022:0/16] 2024-02-02 07:59:33,377 (trainer:737) INFO: 20epoch:train:7701-7800batch: iter_time=8.193e-05, forward_time=0.291, loss_ctc=48.824, loss_att=45.383, acc=0.754, loss=46.415, backward_time=0.405, grad_norm=34.832, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.264e-04, train_time=1.499
+[gpub022:0/16] 2024-02-02 08:02:18,337 (trainer:737) INFO: 20epoch:train:7801-7900batch: iter_time=9.057e-05, forward_time=0.416, loss_ctc=54.281, loss_att=50.687, acc=0.742, loss=51.765, backward_time=0.427, grad_norm=37.933, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.263e-04, train_time=1.648
+[gpub022:0/16] 2024-02-02 08:04:53,655 (trainer:737) INFO: 20epoch:train:7901-8000batch: iter_time=8.993e-05, forward_time=0.291, loss_ctc=50.301, loss_att=48.700, acc=0.761, loss=49.180, backward_time=0.403, grad_norm=33.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.263e-04, train_time=1.554
+[gpub022:0/16] 2024-02-02 08:07:22,316 (trainer:737) INFO: 20epoch:train:8001-8100batch: iter_time=0.001, forward_time=0.421, loss_ctc=47.734, loss_att=43.866, acc=0.748, loss=45.026, backward_time=0.423, grad_norm=31.957, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=2.263e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 08:10:07,511 (trainer:737) INFO: 20epoch:train:8101-8200batch: iter_time=1.019e-04, forward_time=0.292, loss_ctc=56.282, loss_att=56.428, acc=0.728, loss=56.384, backward_time=0.406, grad_norm=39.934, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.262e-04, train_time=1.654
+[gpub022:0/16] 2024-02-02 08:12:53,822 (trainer:737) INFO: 20epoch:train:8201-8300batch: iter_time=3.385e-04, forward_time=0.382, loss_ctc=48.546, loss_att=56.184, acc=0.726, loss=53.893, backward_time=0.437, grad_norm=35.762, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.262e-04, train_time=1.663
+[gpub022:0/16] 2024-02-02 08:15:04,415 (trainer:737) INFO: 20epoch:train:8301-8400batch: iter_time=1.003e-04, forward_time=0.292, loss_ctc=53.709, loss_att=56.846, acc=0.728, loss=55.905, backward_time=0.408, grad_norm=34.931, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.261e-04, train_time=1.304
+[gpub022:0/16] 2024-02-02 08:17:55,306 (trainer:737) INFO: 20epoch:train:8401-8500batch: iter_time=1.018e-04, forward_time=0.331, loss_ctc=62.523, loss_att=58.696, acc=0.721, loss=59.845, backward_time=0.416, grad_norm=40.505, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.261e-04, train_time=1.710
+[gpub022:0/16] 2024-02-02 08:20:22,011 (trainer:737) INFO: 20epoch:train:8501-8600batch: iter_time=1.099e-04, forward_time=0.352, loss_ctc=50.990, loss_att=45.723, acc=0.743, loss=47.303, backward_time=0.419, grad_norm=33.393, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.261e-04, train_time=1.466
+[gpub022:0/16] 2024-02-02 08:22:47,509 (trainer:737) INFO: 20epoch:train:8601-8700batch: iter_time=3.835e-04, forward_time=0.292, loss_ctc=50.806, loss_att=50.685, acc=0.725, loss=50.722, backward_time=0.413, grad_norm=37.271, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.260e-04, train_time=1.456
+[gpub022:0/16] 2024-02-02 08:24:26,690 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub022:0/16] 2024-02-02 08:24:45,466 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 08:24:48,921 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe030b3ff40>)
+[gpub022:0/16] 2024-02-02 08:24:48,921 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub022:0/16] 2024-02-02 08:24:48,925 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 08:30:22,956 (trainer:737) INFO: 20epoch:train:8701-8800batch: iter_time=2.959, forward_time=0.388, loss_ctc=46.715, loss_att=44.979, acc=0.758, loss=45.500, backward_time=0.436, grad_norm=30.531, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.260e-04, train_time=4.554
+[gpub022:0/16] 2024-02-02 08:32:37,388 (trainer:737) INFO: 20epoch:train:8801-8900batch: iter_time=2.015e-04, forward_time=0.305, loss_ctc=45.045, loss_att=48.270, acc=0.753, loss=47.302, backward_time=0.412, grad_norm=33.454, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.259e-04, train_time=1.344
+[gpub022:0/16] 2024-02-02 08:35:07,531 (trainer:737) INFO: 20epoch:train:8901-9000batch: iter_time=8.673e-05, forward_time=0.394, loss_ctc=45.635, loss_att=40.960, acc=0.767, loss=42.362, backward_time=0.422, grad_norm=31.277, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.259e-04, train_time=1.501
+[gpub022:0/16] 2024-02-02 08:37:31,006 (trainer:737) INFO: 20epoch:train:9001-9100batch: iter_time=9.016e-05, forward_time=0.292, loss_ctc=48.655, loss_att=50.928, acc=0.740, loss=50.246, backward_time=0.406, grad_norm=34.955, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.259e-04, train_time=1.435
+[gpub022:0/16] 2024-02-02 08:40:13,269 (trainer:737) INFO: 20epoch:train:9101-9200batch: iter_time=1.008e-04, forward_time=0.400, loss_ctc=53.045, loss_att=52.706, acc=0.759, loss=52.808, backward_time=0.443, grad_norm=38.092, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.258e-04, train_time=1.622
+[gpub022:0/16] 2024-02-02 08:42:25,182 (trainer:737) INFO: 20epoch:train:9201-9300batch: iter_time=1.042e-04, forward_time=0.291, loss_ctc=50.783, loss_att=46.089, acc=0.761, loss=47.497, backward_time=0.405, grad_norm=34.326, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.258e-04, train_time=1.319
+[gpub022:0/16] 2024-02-02 08:44:56,494 (trainer:737) INFO: 20epoch:train:9301-9400batch: iter_time=1.032e-04, forward_time=0.294, loss_ctc=50.119, loss_att=50.220, acc=0.752, loss=50.190, backward_time=0.406, grad_norm=32.856, clip=100.000, loss_scale=8.204e+33, optim_step_time=0.093, optim0_lr0=2.258e-04, train_time=1.513
+[gpub022:0/16] 2024-02-02 08:47:41,187 (trainer:737) INFO: 20epoch:train:9401-9500batch: iter_time=9.544e-05, forward_time=0.357, loss_ctc=52.943, loss_att=57.287, acc=0.728, loss=55.984, backward_time=0.465, grad_norm=37.465, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.257e-04, train_time=1.647
+[gpub022:0/16] 2024-02-02 08:50:10,382 (trainer:737) INFO: 20epoch:train:9501-9600batch: iter_time=1.046e-04, forward_time=0.295, loss_ctc=49.731, loss_att=58.604, acc=0.737, loss=55.942, backward_time=0.408, grad_norm=34.356, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.257e-04, train_time=1.491
+[gpub022:0/16] 2024-02-02 08:52:21,505 (trainer:737) INFO: 20epoch:train:9601-9700batch: iter_time=9.536e-05, forward_time=0.298, loss_ctc=51.354, loss_att=54.062, acc=0.746, loss=53.249, backward_time=0.410, grad_norm=32.463, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.256e-04, train_time=1.312
+[gpub022:0/16] 2024-02-02 08:54:49,257 (trainer:737) INFO: 20epoch:train:9701-9800batch: iter_time=5.250e-04, forward_time=0.394, loss_ctc=62.012, loss_att=57.087, acc=0.737, loss=58.565, backward_time=0.445, grad_norm=41.704, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.256e-04, train_time=1.477
+[gpub022:0/16] 2024-02-02 08:57:56,305 (trainer:737) INFO: 20epoch:train:9801-9900batch: iter_time=9.845e-05, forward_time=0.310, loss_ctc=49.805, loss_att=49.403, acc=0.739, loss=49.524, backward_time=0.411, grad_norm=34.074, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.256e-04, train_time=1.871
+[gpub022:0/16] 2024-02-02 09:00:21,416 (trainer:737) INFO: 20epoch:train:9901-10000batch: iter_time=8.600e-05, forward_time=0.368, loss_ctc=55.911, loss_att=57.605, acc=0.733, loss=57.097, backward_time=0.457, grad_norm=37.913, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.255e-04, train_time=1.450
+[gpub022:0/16] 2024-02-02 09:00:41,454 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub022:0/16] 2024-02-02 09:01:00,312 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 09:01:03,900 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fdfb6fc3610>)
+[gpub022:0/16] 2024-02-02 09:01:03,900 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub022:0/16] 2024-02-02 09:01:03,983 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 09:08:44,246 (trainer:737) INFO: 20epoch:train:10001-10100batch: iter_time=3.231, forward_time=0.389, loss_ctc=42.385, loss_att=44.180, acc=0.738, loss=43.642, backward_time=0.415, grad_norm=31.588, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.255e-04, train_time=5.029
+[gpub022:0/16] 2024-02-02 09:11:12,520 (trainer:737) INFO: 20epoch:train:10101-10200batch: iter_time=8.560e-05, forward_time=0.388, loss_ctc=42.633, loss_att=41.289, acc=0.756, loss=41.692, backward_time=0.434, grad_norm=31.823, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.255e-04, train_time=1.482
+[gpub022:0/16] 2024-02-02 09:13:38,663 (trainer:737) INFO: 20epoch:train:10201-10300batch: iter_time=7.971e-05, forward_time=0.363, loss_ctc=48.453, loss_att=45.630, acc=0.753, loss=46.477, backward_time=0.479, grad_norm=35.151, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.254e-04, train_time=1.461
+[gpub022:0/16] 2024-02-02 09:16:34,972 (trainer:737) INFO: 20epoch:train:10301-10400batch: iter_time=9.384e-05, forward_time=0.292, loss_ctc=54.139, loss_att=50.358, acc=0.743, loss=51.492, backward_time=0.405, grad_norm=40.578, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.254e-04, train_time=1.763
+[gpub022:0/16] 2024-02-02 09:19:03,469 (trainer:737) INFO: 20epoch:train:10401-10500batch: iter_time=1.088e-04, forward_time=0.390, loss_ctc=49.939, loss_att=48.489, acc=0.761, loss=48.924, backward_time=0.446, grad_norm=35.421, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.253e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 09:21:30,610 (trainer:737) INFO: 20epoch:train:10501-10600batch: iter_time=9.487e-05, forward_time=0.291, loss_ctc=47.622, loss_att=43.920, acc=0.748, loss=45.031, backward_time=0.404, grad_norm=30.063, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.253e-04, train_time=1.472
+[gpub022:0/16] 2024-02-02 09:24:14,906 (trainer:737) INFO: 20epoch:train:10601-10700batch: iter_time=1.003e-04, forward_time=0.299, loss_ctc=56.036, loss_att=56.550, acc=0.730, loss=56.395, backward_time=0.407, grad_norm=38.860, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.253e-04, train_time=1.643
+[gpub022:0/16] 2024-02-02 09:26:44,324 (trainer:737) INFO: 20epoch:train:10701-10800batch: iter_time=3.871e-04, forward_time=0.396, loss_ctc=48.692, loss_att=56.596, acc=0.727, loss=54.225, backward_time=0.467, grad_norm=127.232, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.252e-04, train_time=1.493
+[gpub022:0/16] 2024-02-02 09:29:13,398 (trainer:737) INFO: 20epoch:train:10801-10900batch: iter_time=1.017e-04, forward_time=0.295, loss_ctc=53.866, loss_att=57.154, acc=0.730, loss=56.168, backward_time=0.408, grad_norm=34.847, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.252e-04, train_time=1.491
+[gpub022:0/16] 2024-02-02 09:32:17,250 (trainer:737) INFO: 20epoch:train:10901-11000batch: iter_time=2.214e-04, forward_time=0.370, loss_ctc=62.614, loss_att=58.522, acc=0.718, loss=59.749, backward_time=0.459, grad_norm=40.601, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.251e-04, train_time=1.837
+[gpub022:0/16] 2024-02-02 09:34:51,304 (trainer:737) INFO: 20epoch:train:11001-11100batch: iter_time=1.026e-04, forward_time=0.291, loss_ctc=50.116, loss_att=45.907, acc=0.744, loss=47.170, backward_time=0.405, grad_norm=34.330, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.251e-04, train_time=1.542
+[gpub022:0/16] 2024-02-02 09:37:47,654 (trainer:737) INFO: 20epoch:train:11101-11200batch: iter_time=9.931e-05, forward_time=0.323, loss_ctc=50.175, loss_att=50.845, acc=0.726, loss=50.644, backward_time=0.433, grad_norm=37.843, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.112, optim0_lr0=2.251e-04, train_time=1.762
+[gpub022:0/16] 2024-02-02 09:39:24,255 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub022:0/16] 2024-02-02 09:39:43,235 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 09:39:46,752 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe0443dff10>)
+[gpub022:0/16] 2024-02-02 09:39:46,752 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub022:0/16] 2024-02-02 09:39:46,833 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 09:45:12,702 (trainer:737) INFO: 20epoch:train:11201-11300batch: iter_time=2.850, forward_time=0.363, loss_ctc=46.183, loss_att=44.282, acc=0.753, loss=44.853, backward_time=0.418, grad_norm=30.190, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.250e-04, train_time=4.450
+[gpub022:0/16] 2024-02-02 09:47:48,706 (trainer:737) INFO: 20epoch:train:11301-11400batch: iter_time=8.781e-05, forward_time=0.292, loss_ctc=45.580, loss_att=45.019, acc=0.748, loss=45.187, backward_time=0.404, grad_norm=31.685, clip=100.000, loss_scale=1.641e+34, optim_step_time=0.093, optim0_lr0=2.250e-04, train_time=1.560
+[gpub022:0/16] 2024-02-02 09:48:15,605 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 09:49:59,317 (trainer:737) INFO: 20epoch:train:11401-11500batch: iter_time=8.470e-05, forward_time=0.299, loss_ctc=45.005, loss_att=39.095, acc=0.770, loss=40.868, backward_time=0.414, grad_norm=30.343, clip=100.000, loss_scale=1.248e+34, optim_step_time=0.094, optim0_lr0=2.250e-04, train_time=1.306
+[gpub022:0/16] 2024-02-02 09:52:49,151 (trainer:737) INFO: 20epoch:train:11501-11600batch: iter_time=9.591e-05, forward_time=0.339, loss_ctc=48.070, loss_att=49.266, acc=0.736, loss=48.907, backward_time=0.423, grad_norm=33.852, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.249e-04, train_time=1.696
+[gpub022:0/16] 2024-02-02 09:55:16,259 (trainer:737) INFO: 20epoch:train:11601-11700batch: iter_time=1.018e-04, forward_time=0.293, loss_ctc=53.138, loss_att=51.402, acc=0.751, loss=51.923, backward_time=0.406, grad_norm=36.416, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.249e-04, train_time=1.473
+[gpub022:0/16] 2024-02-02 09:58:07,592 (trainer:737) INFO: 20epoch:train:11701-11800batch: iter_time=9.239e-05, forward_time=0.310, loss_ctc=50.720, loss_att=44.261, acc=0.758, loss=46.199, backward_time=0.409, grad_norm=32.903, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.248e-04, train_time=1.713
+[gpub022:0/16] 2024-02-02 10:02:02,678 (trainer:737) INFO: 20epoch:train:11801-11900batch: iter_time=0.105, forward_time=0.330, loss_ctc=49.806, loss_att=50.048, acc=0.743, loss=49.975, backward_time=0.457, grad_norm=33.320, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.248e-04, train_time=2.351
+[gpub022:0/16] 2024-02-02 10:04:25,631 (trainer:737) INFO: 20epoch:train:11901-12000batch: iter_time=8.550e-05, forward_time=0.291, loss_ctc=53.639, loss_att=57.454, acc=0.720, loss=56.310, backward_time=0.404, grad_norm=37.173, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.248e-04, train_time=1.427
+[gpub022:0/16] 2024-02-02 10:09:04,714 (trainer:737) INFO: 20epoch:train:12001-12100batch: iter_time=9.028e-05, forward_time=0.359, loss_ctc=49.014, loss_att=57.355, acc=0.722, loss=54.853, backward_time=0.419, grad_norm=34.794, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.247e-04, train_time=2.793
+[gpub022:0/16] 2024-02-02 10:11:35,007 (trainer:737) INFO: 20epoch:train:12101-12200batch: iter_time=8.672e-05, forward_time=0.407, loss_ctc=51.320, loss_att=51.114, acc=0.742, loss=51.176, backward_time=0.429, grad_norm=32.132, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.247e-04, train_time=1.503
+[gpub022:0/16] 2024-02-02 10:14:10,676 (trainer:737) INFO: 20epoch:train:12201-12300batch: iter_time=9.042e-05, forward_time=0.292, loss_ctc=61.055, loss_att=56.677, acc=0.726, loss=57.990, backward_time=0.405, grad_norm=41.272, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.247e-04, train_time=1.556
+[gpub022:0/16] 2024-02-02 10:17:37,398 (trainer:737) INFO: 20epoch:train:12301-12400batch: iter_time=0.069, forward_time=0.534, loss_ctc=49.377, loss_att=47.015, acc=0.745, loss=47.724, backward_time=0.545, grad_norm=32.118, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.246e-04, train_time=2.066
+[gpub022:0/16] 2024-02-02 10:20:50,950 (trainer:737) INFO: 20epoch:train:12401-12500batch: iter_time=9.479e-05, forward_time=0.345, loss_ctc=55.405, loss_att=53.252, acc=0.739, loss=53.898, backward_time=0.504, grad_norm=38.528, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.246e-04, train_time=1.936
+[gpub022:0/16] 2024-02-02 10:21:11,042 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub022:0/16] 2024-02-02 10:21:29,719 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 10:21:33,177 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe0578e3f70>)
+[gpub022:0/16] 2024-02-02 10:21:33,177 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub022:0/16] 2024-02-02 10:21:33,280 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 10:29:34,233 (trainer:737) INFO: 20epoch:train:12501-12600batch: iter_time=3.629, forward_time=0.287, loss_ctc=41.910, loss_att=44.770, acc=0.752, loss=43.912, backward_time=0.399, grad_norm=31.083, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.245e-04, train_time=5.233
+[gpub022:0/16] 2024-02-02 10:32:44,576 (trainer:737) INFO: 20epoch:train:12601-12700batch: iter_time=8.707e-05, forward_time=0.292, loss_ctc=42.161, loss_att=42.545, acc=0.756, loss=42.430, backward_time=0.401, grad_norm=31.215, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.245e-04, train_time=1.903
+[gpub022:0/16] 2024-02-02 10:35:14,258 (trainer:737) INFO: 20epoch:train:12701-12800batch: iter_time=9.321e-05, forward_time=0.398, loss_ctc=48.239, loss_att=45.576, acc=0.763, loss=46.375, backward_time=0.440, grad_norm=34.525, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.245e-04, train_time=1.497
+[gpub022:0/16] 2024-02-02 10:38:06,146 (trainer:737) INFO: 20epoch:train:12801-12900batch: iter_time=1.084e-04, forward_time=0.292, loss_ctc=54.137, loss_att=50.794, acc=0.756, loss=51.797, backward_time=0.406, grad_norm=38.412, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.244e-04, train_time=1.719
+[gpub022:0/16] 2024-02-02 10:38:21,461 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 10:40:21,998 (trainer:737) INFO: 20epoch:train:12901-13000batch: iter_time=1.001e-04, forward_time=0.315, loss_ctc=49.315, loss_att=49.921, acc=0.765, loss=49.740, backward_time=0.408, grad_norm=32.654, clip=100.000, loss_scale=5.664e+33, optim_step_time=0.093, optim0_lr0=2.244e-04, train_time=1.358
+[gpub022:0/16] 2024-02-02 10:43:11,820 (trainer:737) INFO: 20epoch:train:13001-13100batch: iter_time=1.019e-04, forward_time=0.344, loss_ctc=46.788, loss_att=44.607, acc=0.753, loss=45.261, backward_time=0.494, grad_norm=31.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.244e-04, train_time=1.697
+[gpub022:0/16] 2024-02-02 10:45:50,263 (trainer:737) INFO: 20epoch:train:13101-13200batch: iter_time=1.076e-04, forward_time=0.291, loss_ctc=54.980, loss_att=56.180, acc=0.738, loss=55.820, backward_time=0.405, grad_norm=36.763, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.243e-04, train_time=1.585
+[gpub022:0/16] 2024-02-02 10:48:36,173 (trainer:737) INFO: 20epoch:train:13201-13300batch: iter_time=9.931e-05, forward_time=0.294, loss_ctc=47.683, loss_att=56.650, acc=0.740, loss=53.960, backward_time=0.429, grad_norm=34.830, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.243e-04, train_time=1.659
+[gpub022:0/16] 2024-02-02 10:51:17,336 (trainer:737) INFO: 20epoch:train:13301-13400batch: iter_time=9.925e-05, forward_time=0.397, loss_ctc=53.266, loss_att=58.284, acc=0.739, loss=56.779, backward_time=0.454, grad_norm=34.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.242e-04, train_time=1.611
+[gpub022:0/16] 2024-02-02 10:54:02,646 (trainer:737) INFO: 20epoch:train:13401-13500batch: iter_time=1.022e-04, forward_time=0.292, loss_ctc=60.797, loss_att=60.918, acc=0.724, loss=60.882, backward_time=0.405, grad_norm=39.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.242e-04, train_time=1.653
+[gpub022:0/16] 2024-02-02 10:56:37,887 (trainer:737) INFO: 20epoch:train:13501-13600batch: iter_time=9.847e-05, forward_time=0.296, loss_ctc=50.232, loss_att=45.582, acc=0.756, loss=46.977, backward_time=0.417, grad_norm=33.808, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.242e-04, train_time=1.552
+[gpub022:0/16] 2024-02-02 10:59:43,657 (trainer:737) INFO: 20epoch:train:13601-13700batch: iter_time=1.035e-04, forward_time=0.377, loss_ctc=50.204, loss_att=50.841, acc=0.731, loss=50.650, backward_time=0.435, grad_norm=36.335, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.241e-04, train_time=1.857
+[gpub022:0/16] 2024-02-02 11:01:30,580 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub022:0/16] 2024-02-02 11:01:49,609 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 11:01:53,222 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe03017b7f0>)
+[gpub022:0/16] 2024-02-02 11:01:53,222 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub022:0/16] 2024-02-02 11:01:53,312 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 11:08:25,496 (trainer:737) INFO: 20epoch:train:13701-13800batch: iter_time=3.487, forward_time=0.327, loss_ctc=45.696, loss_att=47.476, acc=0.753, loss=46.942, backward_time=0.408, grad_norm=30.825, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.241e-04, train_time=5.218
+[gpub022:0/16] 2024-02-02 11:11:30,373 (trainer:737) INFO: 20epoch:train:13801-13900batch: iter_time=9.173e-05, forward_time=0.296, loss_ctc=44.992, loss_att=46.236, acc=0.759, loss=45.863, backward_time=0.406, grad_norm=31.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.241e-04, train_time=1.849
+[gpub022:0/16] 2024-02-02 11:14:12,113 (trainer:737) INFO: 20epoch:train:13901-14000batch: iter_time=8.634e-05, forward_time=0.360, loss_ctc=45.398, loss_att=40.280, acc=0.769, loss=41.815, backward_time=0.456, grad_norm=31.517, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.240e-04, train_time=1.617
+[gpub022:0/16] 2024-02-02 11:17:20,687 (trainer:737) INFO: 20epoch:train:14001-14100batch: iter_time=8.345e-05, forward_time=0.298, loss_ctc=47.915, loss_att=49.626, acc=0.743, loss=49.113, backward_time=0.415, grad_norm=33.313, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.240e-04, train_time=1.885
+[gpub022:0/16] 2024-02-02 11:20:06,725 (trainer:737) INFO: 20epoch:train:14101-14200batch: iter_time=8.924e-05, forward_time=0.406, loss_ctc=53.356, loss_att=52.013, acc=0.761, loss=52.416, backward_time=0.425, grad_norm=37.698, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.239e-04, train_time=1.661
+[gpub022:0/16] 2024-02-02 11:22:56,380 (trainer:737) INFO: 20epoch:train:14201-14300batch: iter_time=9.373e-05, forward_time=0.292, loss_ctc=50.069, loss_att=45.495, acc=0.762, loss=46.867, backward_time=0.411, grad_norm=31.763, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.239e-04, train_time=1.696
+[gpub022:0/16] 2024-02-02 11:25:41,087 (trainer:737) INFO: 20epoch:train:14301-14400batch: iter_time=8.940e-05, forward_time=0.300, loss_ctc=49.619, loss_att=50.291, acc=0.753, loss=50.089, backward_time=0.410, grad_norm=32.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.239e-04, train_time=1.647
+[gpub022:0/16] 2024-02-02 11:29:10,241 (trainer:737) INFO: 20epoch:train:14401-14500batch: iter_time=9.119e-05, forward_time=0.365, loss_ctc=52.750, loss_att=56.281, acc=0.730, loss=55.222, backward_time=0.423, grad_norm=37.607, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.238e-04, train_time=2.091
+[gpub022:0/16] 2024-02-02 11:31:47,289 (trainer:737) INFO: 20epoch:train:14501-14600batch: iter_time=1.162e-04, forward_time=0.346, loss_ctc=49.310, loss_att=58.031, acc=0.739, loss=55.415, backward_time=0.415, grad_norm=33.336, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.238e-04, train_time=1.570
+[gpub022:0/16] 2024-02-02 11:34:23,549 (trainer:737) INFO: 20epoch:train:14601-14700batch: iter_time=9.032e-05, forward_time=0.293, loss_ctc=51.029, loss_att=53.394, acc=0.749, loss=52.684, backward_time=0.408, grad_norm=33.112, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.238e-04, train_time=1.563
+[gpub022:0/16] 2024-02-02 11:37:46,094 (trainer:737) INFO: 20epoch:train:14701-14800batch: iter_time=9.842e-05, forward_time=0.382, loss_ctc=60.364, loss_att=56.434, acc=0.740, loss=57.613, backward_time=0.430, grad_norm=38.008, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=2.237e-04, train_time=2.024
+[gpub022:0/16] 2024-02-02 11:39:58,552 (trainer:737) INFO: 20epoch:train:14801-14900batch: iter_time=9.689e-05, forward_time=0.290, loss_ctc=48.927, loss_att=48.279, acc=0.741, loss=48.474, backward_time=0.405, grad_norm=32.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.237e-04, train_time=1.325
+[gpub022:0/16] 2024-02-02 11:42:57,968 (trainer:737) INFO: 20epoch:train:14901-15000batch: iter_time=9.887e-04, forward_time=0.412, loss_ctc=55.495, loss_att=57.489, acc=0.733, loss=56.891, backward_time=0.426, grad_norm=38.140, clip=100.000, loss_scale=9.865e+33, optim_step_time=0.113, optim0_lr0=2.236e-04, train_time=1.793
+[gpub022:0/16] 2024-02-02 12:20:24,564 (trainer:343) INFO: 20epoch results: [train] iter_time=0.248, forward_time=0.329, loss_ctc=51.246, loss_att=50.799, acc=0.742, loss=50.933, backward_time=0.422, grad_norm=35.688, clip=100.000, loss_scale=5.935e+33, optim_step_time=0.096, optim0_lr0=2.265e-04, train_time=1.818, time=7 hours, 34 minutes and 54.73 seconds, total_count=330000, gpu_max_cached_mem_GB=41.891, [valid] loss_ctc=42.053, cer_ctc=0.209, loss_att=44.871, acc=0.648, cer=0.343, wer=0.999, loss=44.025, time=37 minutes and 2.53 seconds, total_count=102762, gpu_max_cached_mem_GB=41.891
+[gpub022:0/16] 2024-02-02 12:20:34,267 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub022:0/16] 2024-02-02 12:20:34,364 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/15epoch.pth
+[gpub022:0/16] 2024-02-02 12:20:34,364 (trainer:272) INFO: 21/45epoch started. Estimated time to finish: 1 week, 1 day and 19 hours
+[gpub022:0/16] 2024-02-02 12:20:34,373 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub022:0/16] 2024-02-02 12:20:52,587 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 12:20:55,964 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85af0afe0>)
+[gpub022:0/16] 2024-02-02 12:20:55,965 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub022:0/16] 2024-02-02 12:20:55,968 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 12:28:10,496 (trainer:737) INFO: 21epoch:train:1-100batch: iter_time=3.149, forward_time=0.354, loss_ctc=51.560, loss_att=47.868, acc=0.727, loss=48.976, backward_time=0.427, grad_norm=36.570, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.236e-04, train_time=4.561
+[gpub022:0/16] 2024-02-02 12:30:31,149 (trainer:737) INFO: 21epoch:train:101-200batch: iter_time=9.614e-05, forward_time=0.299, loss_ctc=51.741, loss_att=51.389, acc=0.729, loss=51.495, backward_time=0.405, grad_norm=39.886, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.236e-04, train_time=1.405
+[gpub022:0/16] 2024-02-02 12:33:09,734 (trainer:737) INFO: 21epoch:train:201-300batch: iter_time=1.041e-04, forward_time=0.323, loss_ctc=49.229, loss_att=48.274, acc=0.736, loss=48.560, backward_time=0.451, grad_norm=32.547, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.235e-04, train_time=1.586
+[gpub022:0/16] 2024-02-02 12:35:38,070 (trainer:737) INFO: 21epoch:train:301-400batch: iter_time=1.093e-04, forward_time=0.301, loss_ctc=52.231, loss_att=50.810, acc=0.754, loss=51.237, backward_time=0.409, grad_norm=35.174, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.235e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 12:38:10,921 (trainer:737) INFO: 21epoch:train:401-500batch: iter_time=1.077e-04, forward_time=0.310, loss_ctc=40.017, loss_att=37.275, acc=0.764, loss=38.098, backward_time=0.415, grad_norm=29.238, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.235e-04, train_time=1.528
+[gpub022:0/16] 2024-02-02 12:40:41,843 (trainer:737) INFO: 21epoch:train:501-600batch: iter_time=2.950e-04, forward_time=0.297, loss_ctc=43.078, loss_att=41.332, acc=0.750, loss=41.856, backward_time=0.405, grad_norm=31.591, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.234e-04, train_time=1.509
+[gpub022:0/16] 2024-02-02 12:43:19,281 (trainer:737) INFO: 21epoch:train:601-700batch: iter_time=1.014e-04, forward_time=0.340, loss_ctc=53.639, loss_att=50.162, acc=0.747, loss=51.205, backward_time=0.425, grad_norm=41.818, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.234e-04, train_time=1.574
+[gpub022:0/16] 2024-02-02 12:45:48,756 (trainer:737) INFO: 21epoch:train:701-800batch: iter_time=1.120e-04, forward_time=0.317, loss_ctc=50.383, loss_att=44.596, acc=0.738, loss=46.332, backward_time=0.437, grad_norm=38.133, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.233e-04, train_time=1.495
+[gpub022:0/16] 2024-02-02 12:48:13,689 (trainer:737) INFO: 21epoch:train:801-900batch: iter_time=9.832e-05, forward_time=0.306, loss_ctc=54.606, loss_att=54.311, acc=0.705, loss=54.400, backward_time=0.421, grad_norm=37.141, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.233e-04, train_time=1.449
+[gpub022:0/16] 2024-02-02 12:50:53,543 (trainer:737) INFO: 21epoch:train:901-1000batch: iter_time=1.008e-04, forward_time=0.385, loss_ctc=50.359, loss_att=55.815, acc=0.736, loss=54.178, backward_time=0.415, grad_norm=34.068, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.233e-04, train_time=1.598
+[gpub022:0/16] 2024-02-02 12:53:21,976 (trainer:737) INFO: 21epoch:train:1001-1100batch: iter_time=1.099e-04, forward_time=0.291, loss_ctc=50.225, loss_att=57.285, acc=0.728, loss=55.167, backward_time=0.405, grad_norm=35.525, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.232e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 12:55:58,303 (trainer:737) INFO: 21epoch:train:1101-1200batch: iter_time=1.052e-04, forward_time=0.340, loss_ctc=50.026, loss_att=51.533, acc=0.754, loss=51.081, backward_time=0.422, grad_norm=33.056, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.232e-04, train_time=1.563
+[gpub022:0/16] 2024-02-02 12:57:28,081 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub022:0/16] 2024-02-02 12:57:47,091 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 12:57:50,661 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe2b749bf40>)
+[gpub022:0/16] 2024-02-02 12:57:50,661 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub022:0/16] 2024-02-02 12:57:50,665 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 13:03:32,349 (trainer:737) INFO: 21epoch:train:1201-1300batch: iter_time=3.060, forward_time=0.342, loss_ctc=48.879, loss_att=45.992, acc=0.721, loss=46.858, backward_time=0.409, grad_norm=34.288, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.232e-04, train_time=4.540
+[gpub022:0/16] 2024-02-02 13:06:07,129 (trainer:737) INFO: 21epoch:train:1301-1400batch: iter_time=9.337e-05, forward_time=0.290, loss_ctc=49.816, loss_att=43.476, acc=0.769, loss=45.378, backward_time=0.403, grad_norm=32.447, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.231e-04, train_time=1.547
+[gpub022:0/16] 2024-02-02 13:08:27,979 (trainer:737) INFO: 21epoch:train:1401-1500batch: iter_time=1.077e-04, forward_time=0.290, loss_ctc=47.843, loss_att=48.842, acc=0.739, loss=48.542, backward_time=0.403, grad_norm=34.472, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.231e-04, train_time=1.408
+[gpub022:0/16] 2024-02-02 13:11:00,411 (trainer:737) INFO: 21epoch:train:1501-1600batch: iter_time=1.043e-04, forward_time=0.388, loss_ctc=50.996, loss_att=51.693, acc=0.745, loss=51.484, backward_time=0.455, grad_norm=32.404, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.230e-04, train_time=1.524
+[gpub022:0/16] 2024-02-02 13:11:44,267 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 13:13:28,181 (trainer:737) INFO: 21epoch:train:1601-1700batch: iter_time=1.037e-04, forward_time=0.292, loss_ctc=46.618, loss_att=45.328, acc=0.769, loss=45.715, backward_time=0.406, grad_norm=31.743, clip=100.000, loss_scale=6.661e+33, optim_step_time=0.093, optim0_lr0=2.230e-04, train_time=1.477
+[gpub022:0/16] 2024-02-02 13:15:58,478 (trainer:737) INFO: 21epoch:train:1701-1800batch: iter_time=1.006e-04, forward_time=0.289, loss_ctc=42.014, loss_att=37.727, acc=0.769, loss=39.013, backward_time=0.401, grad_norm=30.700, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.230e-04, train_time=1.503
+[gpub022:0/16] 2024-02-02 13:18:26,487 (trainer:737) INFO: 21epoch:train:1801-1900batch: iter_time=1.029e-04, forward_time=0.314, loss_ctc=47.242, loss_att=50.100, acc=0.751, loss=49.243, backward_time=0.432, grad_norm=35.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.229e-04, train_time=1.480
+[gpub022:0/16] 2024-02-02 13:21:01,181 (trainer:737) INFO: 21epoch:train:1901-2000batch: iter_time=1.183e-04, forward_time=0.366, loss_ctc=52.141, loss_att=46.841, acc=0.755, loss=48.431, backward_time=0.425, grad_norm=35.702, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.229e-04, train_time=1.546
+[gpub022:0/16] 2024-02-02 13:23:25,981 (trainer:737) INFO: 21epoch:train:2001-2100batch: iter_time=1.026e-04, forward_time=0.306, loss_ctc=50.630, loss_att=44.509, acc=0.740, loss=46.345, backward_time=0.402, grad_norm=35.453, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.229e-04, train_time=1.448
+[gpub022:0/16] 2024-02-02 13:25:57,744 (trainer:737) INFO: 21epoch:train:2101-2200batch: iter_time=1.047e-04, forward_time=0.320, loss_ctc=48.337, loss_att=52.646, acc=0.736, loss=51.353, backward_time=0.417, grad_norm=30.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.228e-04, train_time=1.517
+[gpub022:0/16] 2024-02-02 13:28:24,942 (trainer:737) INFO: 21epoch:train:2201-2300batch: iter_time=1.055e-04, forward_time=0.336, loss_ctc=54.530, loss_att=62.780, acc=0.738, loss=60.305, backward_time=0.434, grad_norm=35.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.228e-04, train_time=1.472
+[gpub022:0/16] 2024-02-02 13:30:53,286 (trainer:737) INFO: 21epoch:train:2301-2400batch: iter_time=1.039e-04, forward_time=0.293, loss_ctc=50.852, loss_att=52.375, acc=0.750, loss=51.918, backward_time=0.408, grad_norm=32.084, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.228e-04, train_time=1.483
+[gpub022:0/16] 2024-02-02 13:33:25,751 (trainer:737) INFO: 21epoch:train:2401-2500batch: iter_time=1.062e-04, forward_time=0.289, loss_ctc=42.976, loss_att=44.027, acc=0.749, loss=43.712, backward_time=0.402, grad_norm=31.972, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.227e-04, train_time=1.525
+[gpub022:0/16] 2024-02-02 13:33:45,779 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub022:0/16] 2024-02-02 13:34:04,402 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 13:34:08,003 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd863bcb280>)
+[gpub022:0/16] 2024-02-02 13:34:08,003 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub022:0/16] 2024-02-02 13:34:08,037 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 13:41:25,730 (trainer:737) INFO: 21epoch:train:2501-2600batch: iter_time=3.264, forward_time=0.367, loss_ctc=50.110, loss_att=44.477, acc=0.750, loss=46.167, backward_time=0.468, grad_norm=35.130, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.227e-04, train_time=4.800
+[gpub022:0/16] 2024-02-02 13:44:00,961 (trainer:737) INFO: 21epoch:train:2601-2700batch: iter_time=9.482e-05, forward_time=0.291, loss_ctc=50.837, loss_att=49.332, acc=0.753, loss=49.783, backward_time=0.402, grad_norm=116.480, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.226e-04, train_time=1.552
+[gpub022:0/16] 2024-02-02 13:46:24,107 (trainer:737) INFO: 21epoch:train:2701-2800batch: iter_time=9.674e-05, forward_time=0.292, loss_ctc=47.715, loss_att=46.615, acc=0.748, loss=46.945, backward_time=0.405, grad_norm=31.635, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.226e-04, train_time=1.431
+[gpub022:0/16] 2024-02-02 13:48:49,180 (trainer:737) INFO: 21epoch:train:2801-2900batch: iter_time=2.209e-04, forward_time=0.320, loss_ctc=51.288, loss_att=50.544, acc=0.767, loss=50.767, backward_time=0.438, grad_norm=32.343, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.226e-04, train_time=1.451
+[gpub022:0/16] 2024-02-02 13:51:36,604 (trainer:737) INFO: 21epoch:train:2901-3000batch: iter_time=9.526e-05, forward_time=0.340, loss_ctc=39.646, loss_att=37.346, acc=0.773, loss=38.036, backward_time=0.446, grad_norm=37.116, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.225e-04, train_time=1.674
+[gpub022:0/16] 2024-02-02 13:53:53,083 (trainer:737) INFO: 21epoch:train:3001-3100batch: iter_time=9.862e-05, forward_time=0.290, loss_ctc=42.267, loss_att=41.226, acc=0.760, loss=41.538, backward_time=0.403, grad_norm=30.669, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.225e-04, train_time=1.365
+[gpub022:0/16] 2024-02-02 13:56:16,808 (trainer:737) INFO: 21epoch:train:3101-3200batch: iter_time=8.900e-05, forward_time=0.293, loss_ctc=50.934, loss_att=48.325, acc=0.763, loss=49.108, backward_time=0.405, grad_norm=35.150, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.225e-04, train_time=1.437
+[gpub022:0/16] 2024-02-02 13:58:56,933 (trainer:737) INFO: 21epoch:train:3201-3300batch: iter_time=1.027e-04, forward_time=0.305, loss_ctc=48.666, loss_att=44.255, acc=0.746, loss=45.578, backward_time=0.425, grad_norm=34.257, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.224e-04, train_time=1.601
+[gpub022:0/16] 2024-02-02 14:01:26,361 (trainer:737) INFO: 21epoch:train:3301-3400batch: iter_time=9.514e-05, forward_time=0.329, loss_ctc=53.653, loss_att=51.952, acc=0.726, loss=52.462, backward_time=0.451, grad_norm=34.630, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.224e-04, train_time=1.494
+[gpub022:0/16] 2024-02-02 14:03:46,534 (trainer:737) INFO: 21epoch:train:3401-3500batch: iter_time=9.385e-05, forward_time=0.293, loss_ctc=49.108, loss_att=55.785, acc=0.748, loss=53.782, backward_time=0.407, grad_norm=32.339, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.223e-04, train_time=1.402
+[gpub022:0/16] 2024-02-02 14:06:17,185 (trainer:737) INFO: 21epoch:train:3501-3600batch: iter_time=9.514e-05, forward_time=0.292, loss_ctc=49.679, loss_att=57.405, acc=0.736, loss=55.087, backward_time=0.404, grad_norm=33.537, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.223e-04, train_time=1.506
+[gpub022:0/16] 2024-02-02 14:08:55,190 (trainer:737) INFO: 21epoch:train:3601-3700batch: iter_time=4.177e-04, forward_time=0.393, loss_ctc=49.352, loss_att=51.688, acc=0.761, loss=50.987, backward_time=0.451, grad_norm=33.156, clip=100.000, loss_scale=8.879e+33, optim_step_time=0.105, optim0_lr0=2.223e-04, train_time=1.580
+[gpub022:0/16] 2024-02-02 14:10:22,834 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub022:0/16] 2024-02-02 14:10:42,164 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 14:10:45,800 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe2b682f790>)
+[gpub022:0/16] 2024-02-02 14:10:45,800 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub022:0/16] 2024-02-02 14:10:45,804 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 14:16:57,396 (trainer:737) INFO: 21epoch:train:3701-3800batch: iter_time=3.357, forward_time=0.296, loss_ctc=48.135, loss_att=45.165, acc=0.732, loss=46.056, backward_time=0.406, grad_norm=34.676, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.222e-04, train_time=4.822
+[gpub022:0/16] 2024-02-02 14:19:19,934 (trainer:737) INFO: 21epoch:train:3801-3900batch: iter_time=8.509e-05, forward_time=0.312, loss_ctc=48.782, loss_att=44.826, acc=0.752, loss=46.013, backward_time=0.416, grad_norm=35.074, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.222e-04, train_time=1.425
+[gpub022:0/16] 2024-02-02 14:22:03,748 (trainer:737) INFO: 21epoch:train:3901-4000batch: iter_time=8.683e-05, forward_time=0.328, loss_ctc=47.294, loss_att=49.501, acc=0.729, loss=48.839, backward_time=0.442, grad_norm=34.061, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.222e-04, train_time=1.638
+[gpub022:0/16] 2024-02-02 14:22:48,257 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 14:24:40,064 (trainer:737) INFO: 21epoch:train:4001-4100batch: iter_time=9.147e-05, forward_time=0.301, loss_ctc=50.790, loss_att=51.170, acc=0.737, loss=51.056, backward_time=0.419, grad_norm=32.999, clip=100.000, loss_scale=6.871e+33, optim_step_time=0.094, optim0_lr0=2.221e-04, train_time=1.563
+[gpub022:0/16] 2024-02-02 14:27:17,123 (trainer:737) INFO: 21epoch:train:4101-4200batch: iter_time=9.040e-05, forward_time=0.315, loss_ctc=45.709, loss_att=45.010, acc=0.761, loss=45.220, backward_time=0.417, grad_norm=32.605, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.221e-04, train_time=1.570
+[gpub022:0/16] 2024-02-02 14:29:45,524 (trainer:737) INFO: 21epoch:train:4201-4300batch: iter_time=9.041e-05, forward_time=0.329, loss_ctc=41.093, loss_att=36.693, acc=0.769, loss=38.013, backward_time=0.446, grad_norm=30.488, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.221e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 14:32:33,547 (trainer:737) INFO: 21epoch:train:4301-4400batch: iter_time=1.470e-04, forward_time=0.292, loss_ctc=46.073, loss_att=49.517, acc=0.743, loss=48.484, backward_time=0.410, grad_norm=36.374, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.220e-04, train_time=1.680
+[gpub022:0/16] 2024-02-02 14:34:56,333 (trainer:737) INFO: 21epoch:train:4401-4500batch: iter_time=1.001e-04, forward_time=0.292, loss_ctc=51.258, loss_att=47.081, acc=0.747, loss=48.334, backward_time=0.406, grad_norm=35.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.220e-04, train_time=1.427
+[gpub022:0/16] 2024-02-02 14:37:41,375 (trainer:737) INFO: 21epoch:train:4501-4600batch: iter_time=9.777e-05, forward_time=0.304, loss_ctc=49.076, loss_att=43.200, acc=0.740, loss=44.963, backward_time=0.425, grad_norm=36.012, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.219e-04, train_time=1.650
+[gpub022:0/16] 2024-02-02 14:40:08,619 (trainer:737) INFO: 21epoch:train:4601-4700batch: iter_time=9.609e-05, forward_time=0.319, loss_ctc=48.084, loss_att=54.493, acc=0.718, loss=52.571, backward_time=0.458, grad_norm=32.122, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.219e-04, train_time=1.472
+[gpub022:0/16] 2024-02-02 14:42:58,745 (trainer:737) INFO: 21epoch:train:4701-4800batch: iter_time=9.476e-05, forward_time=0.292, loss_ctc=53.553, loss_att=63.101, acc=0.734, loss=60.237, backward_time=0.407, grad_norm=35.204, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.219e-04, train_time=1.701
+[gpub022:0/16] 2024-02-02 14:45:23,159 (trainer:737) INFO: 21epoch:train:4801-4900batch: iter_time=9.810e-05, forward_time=0.300, loss_ctc=50.286, loss_att=50.398, acc=0.747, loss=50.364, backward_time=0.421, grad_norm=33.315, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.218e-04, train_time=1.443
+[gpub022:0/16] 2024-02-02 14:48:11,585 (trainer:737) INFO: 21epoch:train:4901-5000batch: iter_time=9.409e-05, forward_time=0.339, loss_ctc=42.720, loss_att=45.160, acc=0.738, loss=44.428, backward_time=0.420, grad_norm=32.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.218e-04, train_time=1.685
+[gpub022:0/16] 2024-02-02 14:48:31,613 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub022:0/16] 2024-02-02 14:48:50,989 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 14:48:54,670 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe040e5f760>)
+[gpub022:0/16] 2024-02-02 14:48:54,670 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub022:0/16] 2024-02-02 14:48:54,673 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 14:55:16,509 (trainer:737) INFO: 21epoch:train:5001-5100batch: iter_time=2.788, forward_time=0.317, loss_ctc=49.704, loss_att=44.894, acc=0.736, loss=46.337, backward_time=0.414, grad_norm=34.987, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.218e-04, train_time=4.249
+[gpub022:0/16] 2024-02-02 14:58:06,971 (trainer:737) INFO: 21epoch:train:5101-5200batch: iter_time=9.511e-05, forward_time=0.374, loss_ctc=50.262, loss_att=49.607, acc=0.737, loss=49.803, backward_time=0.416, grad_norm=37.529, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.217e-04, train_time=1.705
+[gpub022:0/16] 2024-02-02 15:00:30,578 (trainer:737) INFO: 21epoch:train:5201-5300batch: iter_time=4.309e-04, forward_time=0.372, loss_ctc=47.960, loss_att=46.830, acc=0.742, loss=47.169, backward_time=0.445, grad_norm=32.627, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.217e-04, train_time=1.436
+[gpub022:0/16] 2024-02-02 15:02:57,420 (trainer:737) INFO: 21epoch:train:5301-5400batch: iter_time=9.048e-05, forward_time=0.304, loss_ctc=51.808, loss_att=49.763, acc=0.761, loss=50.377, backward_time=0.414, grad_norm=36.240, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.217e-04, train_time=1.468
+[gpub022:0/16] 2024-02-02 15:05:36,724 (trainer:737) INFO: 21epoch:train:5401-5500batch: iter_time=9.576e-05, forward_time=0.342, loss_ctc=39.458, loss_att=36.242, acc=0.770, loss=37.207, backward_time=0.435, grad_norm=29.168, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.216e-04, train_time=1.593
+[gpub022:0/16] 2024-02-02 15:07:51,640 (trainer:737) INFO: 21epoch:train:5501-5600batch: iter_time=8.799e-05, forward_time=0.319, loss_ctc=42.001, loss_att=40.708, acc=0.756, loss=41.096, backward_time=0.415, grad_norm=31.103, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.216e-04, train_time=1.349
+[gpub022:0/16] 2024-02-02 15:10:35,269 (trainer:737) INFO: 21epoch:train:5601-5700batch: iter_time=1.118e-04, forward_time=0.330, loss_ctc=50.924, loss_att=48.426, acc=0.757, loss=49.175, backward_time=0.438, grad_norm=36.553, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.215e-04, train_time=1.636
+[gpub022:0/16] 2024-02-02 15:13:15,552 (trainer:737) INFO: 21epoch:train:5701-5800batch: iter_time=9.277e-05, forward_time=0.306, loss_ctc=48.318, loss_att=43.621, acc=0.744, loss=45.030, backward_time=0.411, grad_norm=33.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.215e-04, train_time=1.603
+[gpub022:0/16] 2024-02-02 15:15:41,057 (trainer:737) INFO: 21epoch:train:5801-5900batch: iter_time=8.630e-05, forward_time=0.380, loss_ctc=52.857, loss_att=52.999, acc=0.712, loss=52.956, backward_time=0.416, grad_norm=37.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.215e-04, train_time=1.455
+[gpub022:0/16] 2024-02-02 15:18:29,041 (trainer:737) INFO: 21epoch:train:5901-6000batch: iter_time=9.219e-05, forward_time=0.316, loss_ctc=48.679, loss_att=54.701, acc=0.743, loss=52.894, backward_time=0.419, grad_norm=32.345, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.214e-04, train_time=1.679
+[gpub022:0/16] 2024-02-02 15:21:01,530 (trainer:737) INFO: 21epoch:train:6001-6100batch: iter_time=8.570e-05, forward_time=0.336, loss_ctc=49.557, loss_att=56.770, acc=0.732, loss=54.606, backward_time=0.445, grad_norm=35.168, clip=100.000, loss_scale=8.671e+33, optim_step_time=0.102, optim0_lr0=2.214e-04, train_time=1.525
+[gpub022:0/16] 2024-02-02 15:23:40,695 (trainer:737) INFO: 21epoch:train:6101-6200batch: iter_time=9.441e-05, forward_time=0.313, loss_ctc=48.862, loss_att=51.031, acc=0.757, loss=50.380, backward_time=0.429, grad_norm=32.421, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.214e-04, train_time=1.591
+[gpub022:0/16] 2024-02-02 15:25:15,384 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub022:0/16] 2024-02-02 15:25:34,818 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 15:25:38,291 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe03730ffa0>)
+[gpub022:0/16] 2024-02-02 15:25:38,291 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub022:0/16] 2024-02-02 15:25:38,296 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 15:30:32,770 (trainer:737) INFO: 21epoch:train:6201-6300batch: iter_time=2.595, forward_time=0.346, loss_ctc=48.005, loss_att=45.155, acc=0.723, loss=46.010, backward_time=0.426, grad_norm=33.779, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.213e-04, train_time=4.121
+[gpub022:0/16] 2024-02-02 15:32:45,388 (trainer:737) INFO: 21epoch:train:6301-6400batch: iter_time=8.412e-05, forward_time=0.305, loss_ctc=48.457, loss_att=42.938, acc=0.757, loss=44.594, backward_time=0.409, grad_norm=33.045, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.213e-04, train_time=1.326
+[gpub022:0/16] 2024-02-02 15:35:21,882 (trainer:737) INFO: 21epoch:train:6401-6500batch: iter_time=8.158e-05, forward_time=0.312, loss_ctc=47.333, loss_att=48.781, acc=0.732, loss=48.347, backward_time=0.412, grad_norm=33.945, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.213e-04, train_time=1.564
+[gpub022:0/16] 2024-02-02 15:36:22,775 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 15:37:49,832 (trainer:737) INFO: 21epoch:train:6501-6600batch: iter_time=9.756e-05, forward_time=0.331, loss_ctc=50.914, loss_att=50.773, acc=0.739, loss=50.815, backward_time=0.422, grad_norm=33.246, clip=100.000, loss_scale=7.448e+33, optim_step_time=0.096, optim0_lr0=2.212e-04, train_time=1.480
+[gpub022:0/16] 2024-02-02 15:40:09,574 (trainer:737) INFO: 21epoch:train:6601-6700batch: iter_time=9.218e-05, forward_time=0.311, loss_ctc=45.425, loss_att=44.397, acc=0.766, loss=44.705, backward_time=0.412, grad_norm=30.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.212e-04, train_time=1.397
+[gpub022:0/16] 2024-02-02 15:42:52,684 (trainer:737) INFO: 21epoch:train:6701-6800batch: iter_time=4.538e-04, forward_time=0.316, loss_ctc=41.408, loss_att=36.314, acc=0.770, loss=37.842, backward_time=0.425, grad_norm=29.911, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.211e-04, train_time=1.630
+[gpub022:0/16] 2024-02-02 15:45:10,456 (trainer:737) INFO: 21epoch:train:6801-6900batch: iter_time=1.037e-04, forward_time=0.307, loss_ctc=46.278, loss_att=49.456, acc=0.744, loss=48.502, backward_time=0.407, grad_norm=36.031, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.211e-04, train_time=1.378
+[gpub022:0/16] 2024-02-02 15:47:28,374 (trainer:737) INFO: 21epoch:train:6901-7000batch: iter_time=1.014e-04, forward_time=0.309, loss_ctc=50.971, loss_att=46.064, acc=0.750, loss=47.536, backward_time=0.413, grad_norm=34.703, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.211e-04, train_time=1.379
+[gpub022:0/16] 2024-02-02 15:50:20,958 (trainer:737) INFO: 21epoch:train:7001-7100batch: iter_time=1.784e-04, forward_time=0.377, loss_ctc=48.719, loss_att=42.684, acc=0.742, loss=44.494, backward_time=0.433, grad_norm=33.485, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.210e-04, train_time=1.725
+[gpub022:0/16] 2024-02-02 15:50:36,505 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 15:52:45,862 (trainer:737) INFO: 21epoch:train:7101-7200batch: iter_time=9.972e-05, forward_time=0.290, loss_ctc=47.253, loss_att=53.170, acc=0.721, loss=51.395, backward_time=0.403, grad_norm=31.276, clip=100.000, loss_scale=2.885e+33, optim_step_time=0.093, optim0_lr0=2.210e-04, train_time=1.449
+[gpub022:0/16] 2024-02-02 15:55:23,320 (trainer:737) INFO: 21epoch:train:7201-7300batch: iter_time=9.592e-05, forward_time=0.318, loss_ctc=52.733, loss_att=61.774, acc=0.738, loss=59.062, backward_time=0.415, grad_norm=34.648, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.210e-04, train_time=1.574
+[gpub022:0/16] 2024-02-02 15:57:53,979 (trainer:737) INFO: 21epoch:train:7301-7400batch: iter_time=8.758e-05, forward_time=0.350, loss_ctc=50.015, loss_att=49.987, acc=0.748, loss=49.996, backward_time=0.445, grad_norm=32.072, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.209e-04, train_time=1.506
+[gpub022:0/16] 2024-02-02 16:00:23,127 (trainer:737) INFO: 21epoch:train:7401-7500batch: iter_time=9.089e-05, forward_time=0.289, loss_ctc=42.578, loss_att=45.182, acc=0.739, loss=44.401, backward_time=0.403, grad_norm=31.879, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.209e-04, train_time=1.492
+[gpub022:0/16] 2024-02-02 16:00:43,156 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub022:0/16] 2024-02-02 16:01:02,302 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 16:01:05,832 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe040e1c280>)
+[gpub022:0/16] 2024-02-02 16:01:05,832 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub022:0/16] 2024-02-02 16:01:05,836 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 16:08:09,236 (trainer:737) INFO: 21epoch:train:7501-7600batch: iter_time=3.200, forward_time=0.393, loss_ctc=49.537, loss_att=46.084, acc=0.746, loss=47.120, backward_time=0.429, grad_norm=37.299, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.209e-04, train_time=4.660
+[gpub022:0/16] 2024-02-02 16:10:33,209 (trainer:737) INFO: 21epoch:train:7601-7700batch: iter_time=1.020e-04, forward_time=0.292, loss_ctc=48.738, loss_att=49.240, acc=0.755, loss=49.089, backward_time=0.405, grad_norm=37.021, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.208e-04, train_time=1.440
+[gpub022:0/16] 2024-02-02 16:13:31,364 (trainer:737) INFO: 21epoch:train:7701-7800batch: iter_time=9.727e-05, forward_time=0.340, loss_ctc=47.761, loss_att=47.547, acc=0.748, loss=47.611, backward_time=0.458, grad_norm=33.959, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.208e-04, train_time=1.781
+[gpub022:0/16] 2024-02-02 16:15:54,718 (trainer:737) INFO: 21epoch:train:7801-7900batch: iter_time=1.013e-04, forward_time=0.309, loss_ctc=50.824, loss_att=50.668, acc=0.767, loss=50.715, backward_time=0.411, grad_norm=34.680, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.208e-04, train_time=1.433
+[gpub022:0/16] 2024-02-02 16:18:38,624 (trainer:737) INFO: 21epoch:train:7901-8000batch: iter_time=5.222e-04, forward_time=0.372, loss_ctc=39.267, loss_att=37.285, acc=0.775, loss=37.880, backward_time=0.414, grad_norm=29.655, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.207e-04, train_time=1.639
+[gpub022:0/16] 2024-02-02 16:21:08,344 (trainer:737) INFO: 21epoch:train:8001-8100batch: iter_time=1.047e-04, forward_time=0.289, loss_ctc=41.668, loss_att=41.205, acc=0.762, loss=41.344, backward_time=0.402, grad_norm=29.460, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.207e-04, train_time=1.497
+[gpub022:0/16] 2024-02-02 16:23:39,971 (trainer:737) INFO: 21epoch:train:8101-8200batch: iter_time=1.822e-04, forward_time=0.357, loss_ctc=50.305, loss_att=48.209, acc=0.766, loss=48.837, backward_time=0.478, grad_norm=35.369, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.206e-04, train_time=1.515
+[gpub022:0/16] 2024-02-02 16:26:26,361 (trainer:737) INFO: 21epoch:train:8201-8300batch: iter_time=9.425e-05, forward_time=0.289, loss_ctc=47.898, loss_att=43.214, acc=0.752, loss=44.619, backward_time=0.401, grad_norm=32.714, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.206e-04, train_time=1.665
+[gpub022:0/16] 2024-02-02 16:28:56,023 (trainer:737) INFO: 21epoch:train:8301-8400batch: iter_time=1.004e-04, forward_time=0.291, loss_ctc=52.050, loss_att=52.045, acc=0.725, loss=52.046, backward_time=0.403, grad_norm=36.890, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.206e-04, train_time=1.496
+[gpub022:0/16] 2024-02-02 16:31:37,233 (trainer:737) INFO: 21epoch:train:8401-8500batch: iter_time=9.727e-05, forward_time=0.401, loss_ctc=48.169, loss_att=55.501, acc=0.751, loss=53.302, backward_time=0.421, grad_norm=32.291, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.205e-04, train_time=1.610
+[gpub022:0/16] 2024-02-02 16:34:01,676 (trainer:737) INFO: 21epoch:train:8501-8600batch: iter_time=8.906e-05, forward_time=0.294, loss_ctc=48.836, loss_att=57.361, acc=0.738, loss=54.803, backward_time=0.405, grad_norm=34.444, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.205e-04, train_time=1.446
+[gpub022:0/16] 2024-02-02 16:36:43,531 (trainer:737) INFO: 21epoch:train:8601-8700batch: iter_time=6.487e-04, forward_time=0.362, loss_ctc=48.270, loss_att=50.141, acc=0.765, loss=49.580, backward_time=0.462, grad_norm=32.111, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.205e-04, train_time=1.618
+[gpub022:0/16] 2024-02-02 16:38:33,038 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub022:0/16] 2024-02-02 16:38:52,734 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 16:38:56,509 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe0373bc670>)
+[gpub022:0/16] 2024-02-02 16:38:56,510 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub022:0/16] 2024-02-02 16:38:56,513 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 16:44:18,999 (trainer:737) INFO: 21epoch:train:8701-8800batch: iter_time=3.004, forward_time=0.287, loss_ctc=47.394, loss_att=45.375, acc=0.730, loss=45.981, backward_time=0.401, grad_norm=35.769, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.204e-04, train_time=4.554
+[gpub022:0/16] 2024-02-02 16:47:05,485 (trainer:737) INFO: 21epoch:train:8801-8900batch: iter_time=9.403e-05, forward_time=0.288, loss_ctc=48.027, loss_att=43.515, acc=0.757, loss=44.868, backward_time=0.400, grad_norm=33.538, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.204e-04, train_time=1.665
+[gpub022:0/16] 2024-02-02 16:49:52,146 (trainer:737) INFO: 21epoch:train:8901-9000batch: iter_time=8.334e-05, forward_time=0.369, loss_ctc=46.987, loss_att=48.890, acc=0.733, loss=48.319, backward_time=0.442, grad_norm=33.356, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.204e-04, train_time=1.666
+[gpub022:0/16] 2024-02-02 16:52:34,962 (trainer:737) INFO: 21epoch:train:9001-9100batch: iter_time=9.046e-05, forward_time=0.291, loss_ctc=50.691, loss_att=50.697, acc=0.740, loss=50.695, backward_time=0.404, grad_norm=33.112, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.203e-04, train_time=1.627
+[gpub022:0/16] 2024-02-02 16:53:38,729 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 16:54:49,318 (trainer:737) INFO: 21epoch:train:9101-9200batch: iter_time=9.124e-05, forward_time=0.292, loss_ctc=45.588, loss_att=44.003, acc=0.766, loss=44.478, backward_time=0.406, grad_norm=31.415, clip=100.000, loss_scale=3.514e+33, optim_step_time=0.093, optim0_lr0=2.203e-04, train_time=1.345
+[gpub022:0/16] 2024-02-02 16:57:40,713 (trainer:737) INFO: 21epoch:train:9201-9300batch: iter_time=8.706e-05, forward_time=0.343, loss_ctc=40.761, loss_att=35.875, acc=0.773, loss=37.341, backward_time=0.442, grad_norm=30.573, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.203e-04, train_time=1.714
+[gpub022:0/16] 2024-02-02 17:00:23,829 (trainer:737) INFO: 21epoch:train:9301-9400batch: iter_time=9.011e-05, forward_time=0.290, loss_ctc=45.833, loss_att=49.601, acc=0.744, loss=48.470, backward_time=0.403, grad_norm=39.279, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.202e-04, train_time=1.631
+[gpub022:0/16] 2024-02-02 17:02:38,314 (trainer:737) INFO: 21epoch:train:9401-9500batch: iter_time=8.721e-05, forward_time=0.291, loss_ctc=51.048, loss_att=46.184, acc=0.751, loss=47.643, backward_time=0.406, grad_norm=35.158, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.202e-04, train_time=1.344
+[gpub022:0/16] 2024-02-02 17:05:49,459 (trainer:737) INFO: 21epoch:train:9501-9600batch: iter_time=2.999e-04, forward_time=0.392, loss_ctc=48.423, loss_att=43.166, acc=0.741, loss=44.743, backward_time=0.430, grad_norm=34.091, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.201e-04, train_time=1.912
+[gpub022:0/16] 2024-02-02 17:08:02,343 (trainer:737) INFO: 21epoch:train:9601-9700batch: iter_time=8.953e-05, forward_time=0.291, loss_ctc=47.161, loss_att=53.554, acc=0.723, loss=51.636, backward_time=0.405, grad_norm=30.865, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.201e-04, train_time=1.329
+[gpub022:0/16] 2024-02-02 17:10:34,482 (trainer:737) INFO: 21epoch:train:9701-9800batch: iter_time=1.027e-04, forward_time=0.296, loss_ctc=52.666, loss_att=61.270, acc=0.742, loss=58.689, backward_time=0.409, grad_norm=34.872, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.201e-04, train_time=1.520
+[gpub022:0/16] 2024-02-02 17:13:27,255 (trainer:737) INFO: 21epoch:train:9801-9900batch: iter_time=2.645e-04, forward_time=0.370, loss_ctc=49.378, loss_att=50.030, acc=0.746, loss=49.835, backward_time=0.429, grad_norm=33.465, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.200e-04, train_time=1.728
+[gpub022:0/16] 2024-02-02 17:15:55,642 (trainer:737) INFO: 21epoch:train:9901-10000batch: iter_time=8.740e-05, forward_time=0.290, loss_ctc=42.113, loss_att=44.928, acc=0.740, loss=44.084, backward_time=0.403, grad_norm=52.916, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.200e-04, train_time=1.484
+[gpub022:0/16] 2024-02-02 17:16:15,671 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub022:0/16] 2024-02-02 17:16:35,655 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 17:16:39,153 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe04987bf10>)
+[gpub022:0/16] 2024-02-02 17:16:39,153 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub022:0/16] 2024-02-02 17:16:39,157 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 17:23:55,691 (trainer:737) INFO: 21epoch:train:10001-10100batch: iter_time=3.060, forward_time=0.299, loss_ctc=48.814, loss_att=45.547, acc=0.749, loss=46.527, backward_time=0.420, grad_norm=35.246, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.200e-04, train_time=4.800
+[gpub022:0/16] 2024-02-02 17:26:21,479 (trainer:737) INFO: 21epoch:train:10101-10200batch: iter_time=9.579e-05, forward_time=0.395, loss_ctc=49.626, loss_att=49.492, acc=0.755, loss=49.532, backward_time=0.424, grad_norm=35.701, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.199e-04, train_time=1.458
+[gpub022:0/16] 2024-02-02 17:29:00,080 (trainer:737) INFO: 21epoch:train:10201-10300batch: iter_time=8.838e-05, forward_time=0.292, loss_ctc=46.998, loss_att=46.189, acc=0.751, loss=46.432, backward_time=0.404, grad_norm=31.979, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.199e-04, train_time=1.585
+[gpub022:0/16] 2024-02-02 17:31:43,701 (trainer:737) INFO: 21epoch:train:10301-10400batch: iter_time=8.677e-05, forward_time=0.367, loss_ctc=50.539, loss_att=51.215, acc=0.767, loss=51.012, backward_time=0.450, grad_norm=33.880, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.199e-04, train_time=1.637
+[gpub022:0/16] 2024-02-02 17:34:07,748 (trainer:737) INFO: 21epoch:train:10401-10500batch: iter_time=9.474e-05, forward_time=0.290, loss_ctc=38.635, loss_att=36.654, acc=0.778, loss=37.248, backward_time=0.403, grad_norm=28.235, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.198e-04, train_time=1.440
+[gpub022:0/16] 2024-02-02 17:36:28,533 (trainer:737) INFO: 21epoch:train:10501-10600batch: iter_time=9.733e-05, forward_time=0.291, loss_ctc=41.523, loss_att=41.277, acc=0.762, loss=41.351, backward_time=0.402, grad_norm=31.057, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.198e-04, train_time=1.408
+[gpub022:0/16] 2024-02-02 17:39:31,091 (trainer:737) INFO: 21epoch:train:10601-10700batch: iter_time=1.086e-04, forward_time=0.378, loss_ctc=49.788, loss_att=48.054, acc=0.766, loss=48.574, backward_time=0.435, grad_norm=36.664, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=2.198e-04, train_time=1.825
+[gpub022:0/16] 2024-02-02 17:41:45,223 (trainer:737) INFO: 21epoch:train:10701-10800batch: iter_time=9.939e-05, forward_time=0.291, loss_ctc=47.560, loss_att=43.355, acc=0.752, loss=44.616, backward_time=0.403, grad_norm=34.577, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.197e-04, train_time=1.341
+[gpub022:0/16] 2024-02-02 17:44:33,195 (trainer:737) INFO: 21epoch:train:10801-10900batch: iter_time=9.864e-05, forward_time=0.292, loss_ctc=52.114, loss_att=51.969, acc=0.726, loss=52.012, backward_time=0.406, grad_norm=37.098, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.197e-04, train_time=1.679
+[gpub022:0/16] 2024-02-02 17:47:19,813 (trainer:737) INFO: 21epoch:train:10901-11000batch: iter_time=8.868e-05, forward_time=0.380, loss_ctc=47.664, loss_att=54.464, acc=0.753, loss=52.424, backward_time=0.447, grad_norm=31.747, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.196e-04, train_time=1.666
+[gpub022:0/16] 2024-02-02 17:49:42,267 (trainer:737) INFO: 21epoch:train:11001-11100batch: iter_time=9.480e-05, forward_time=0.293, loss_ctc=48.638, loss_att=56.965, acc=0.741, loss=54.467, backward_time=0.404, grad_norm=35.877, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.196e-04, train_time=1.425
+[gpub022:0/16] 2024-02-02 17:52:38,622 (trainer:737) INFO: 21epoch:train:11101-11200batch: iter_time=8.933e-05, forward_time=0.302, loss_ctc=48.540, loss_att=50.652, acc=0.765, loss=50.018, backward_time=0.408, grad_norm=32.346, clip=100.000, loss_scale=3.946e+33, optim_step_time=0.094, optim0_lr0=2.196e-04, train_time=1.763
+[gpub022:0/16] 2024-02-02 17:54:08,195 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub022:0/16] 2024-02-02 17:54:27,649 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 17:54:31,293 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe0491d2110>)
+[gpub022:0/16] 2024-02-02 17:54:31,293 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub022:0/16] 2024-02-02 17:54:31,296 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 18:00:36,271 (trainer:737) INFO: 21epoch:train:11201-11300batch: iter_time=3.295, forward_time=0.382, loss_ctc=47.342, loss_att=44.408, acc=0.738, loss=45.288, backward_time=0.421, grad_norm=35.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.195e-04, train_time=4.776
+[gpub022:0/16] 2024-02-02 18:03:15,251 (trainer:737) INFO: 21epoch:train:11301-11400batch: iter_time=8.245e-05, forward_time=0.291, loss_ctc=47.495, loss_att=41.663, acc=0.776, loss=43.413, backward_time=0.404, grad_norm=31.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.195e-04, train_time=1.590
+[gpub022:0/16] 2024-02-02 18:06:04,506 (trainer:737) INFO: 21epoch:train:11401-11500batch: iter_time=7.859e-05, forward_time=0.364, loss_ctc=46.800, loss_att=47.962, acc=0.745, loss=47.614, backward_time=0.417, grad_norm=34.232, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.195e-04, train_time=1.692
+[gpub022:0/16] 2024-02-02 18:08:19,993 (trainer:737) INFO: 21epoch:train:11501-11600batch: iter_time=9.063e-05, forward_time=0.294, loss_ctc=50.231, loss_att=50.917, acc=0.750, loss=50.711, backward_time=0.409, grad_norm=31.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.194e-04, train_time=1.355
+[gpub022:0/16] 2024-02-02 18:11:04,237 (trainer:737) INFO: 21epoch:train:11601-11700batch: iter_time=8.813e-05, forward_time=0.295, loss_ctc=45.072, loss_att=44.399, acc=0.776, loss=44.601, backward_time=0.408, grad_norm=31.129, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.194e-04, train_time=1.642
+[gpub022:0/16] 2024-02-02 18:13:45,621 (trainer:737) INFO: 21epoch:train:11701-11800batch: iter_time=9.722e-05, forward_time=0.343, loss_ctc=40.751, loss_att=36.568, acc=0.775, loss=37.823, backward_time=0.451, grad_norm=29.625, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.194e-04, train_time=1.614
+[gpub022:0/16] 2024-02-02 18:16:09,548 (trainer:737) INFO: 21epoch:train:11801-11900batch: iter_time=8.977e-05, forward_time=0.292, loss_ctc=45.510, loss_att=49.196, acc=0.755, loss=48.090, backward_time=0.405, grad_norm=37.965, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.193e-04, train_time=1.439
+[gpub022:0/16] 2024-02-02 18:19:01,564 (trainer:737) INFO: 21epoch:train:11901-12000batch: iter_time=8.695e-05, forward_time=0.321, loss_ctc=50.508, loss_att=45.837, acc=0.759, loss=47.238, backward_time=0.426, grad_norm=35.204, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.193e-04, train_time=1.720
+[gpub022:0/16] 2024-02-02 18:21:45,418 (trainer:737) INFO: 21epoch:train:12001-12100batch: iter_time=8.670e-05, forward_time=0.289, loss_ctc=47.979, loss_att=42.776, acc=0.747, loss=44.337, backward_time=0.401, grad_norm=34.044, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.193e-04, train_time=1.638
+[gpub022:0/16] 2024-02-02 18:24:39,454 (trainer:737) INFO: 21epoch:train:12101-12200batch: iter_time=9.005e-05, forward_time=0.364, loss_ctc=46.868, loss_att=52.053, acc=0.742, loss=50.498, backward_time=0.472, grad_norm=31.262, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.192e-04, train_time=1.740
+[gpub022:0/16] 2024-02-02 18:27:25,497 (trainer:737) INFO: 21epoch:train:12201-12300batch: iter_time=8.847e-05, forward_time=0.295, loss_ctc=52.353, loss_att=62.032, acc=0.742, loss=59.128, backward_time=0.409, grad_norm=33.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.192e-04, train_time=1.660
+[gpub022:0/16] 2024-02-02 18:29:48,772 (trainer:737) INFO: 21epoch:train:12301-12400batch: iter_time=8.954e-05, forward_time=0.293, loss_ctc=49.917, loss_att=51.660, acc=0.754, loss=51.137, backward_time=0.406, grad_norm=34.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.192e-04, train_time=1.432
+[gpub022:0/16] 2024-02-02 18:32:30,934 (trainer:737) INFO: 21epoch:train:12401-12500batch: iter_time=8.350e-05, forward_time=0.325, loss_ctc=42.126, loss_att=43.910, acc=0.753, loss=43.375, backward_time=0.413, grad_norm=32.545, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.191e-04, train_time=1.621
+[gpub022:0/16] 2024-02-02 18:32:51,065 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub022:0/16] 2024-02-02 18:33:10,225 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 18:33:13,913 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fdff4aaee00>)
+[gpub022:0/16] 2024-02-02 18:33:13,913 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub022:0/16] 2024-02-02 18:33:13,916 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 18:40:32,474 (trainer:737) INFO: 21epoch:train:12501-12600batch: iter_time=3.104, forward_time=0.292, loss_ctc=49.226, loss_att=46.338, acc=0.736, loss=47.204, backward_time=0.402, grad_norm=35.179, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.191e-04, train_time=4.815
+[gpub022:0/16] 2024-02-02 18:43:20,534 (trainer:737) INFO: 21epoch:train:12601-12700batch: iter_time=8.506e-05, forward_time=0.360, loss_ctc=48.629, loss_att=49.945, acc=0.736, loss=49.551, backward_time=0.441, grad_norm=36.538, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.191e-04, train_time=1.680
+[gpub022:0/16] 2024-02-02 18:46:06,954 (trainer:737) INFO: 21epoch:train:12701-12800batch: iter_time=8.252e-05, forward_time=0.291, loss_ctc=46.936, loss_att=46.679, acc=0.743, loss=46.756, backward_time=0.406, grad_norm=33.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.190e-04, train_time=1.664
+[gpub022:0/16] 2024-02-02 18:49:03,476 (trainer:737) INFO: 21epoch:train:12801-12900batch: iter_time=8.804e-05, forward_time=0.393, loss_ctc=50.585, loss_att=49.670, acc=0.760, loss=49.945, backward_time=0.430, grad_norm=34.481, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.118, optim0_lr0=2.190e-04, train_time=1.765
+[gpub022:0/16] 2024-02-02 18:51:58,905 (trainer:737) INFO: 21epoch:train:12901-13000batch: iter_time=0.003, forward_time=0.349, loss_ctc=38.794, loss_att=36.719, acc=0.770, loss=37.342, backward_time=0.411, grad_norm=29.410, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.189e-04, train_time=1.754
+[gpub022:0/16] 2024-02-02 18:54:25,929 (trainer:737) INFO: 21epoch:train:13001-13100batch: iter_time=4.718e-04, forward_time=0.320, loss_ctc=41.710, loss_att=40.598, acc=0.755, loss=40.932, backward_time=0.406, grad_norm=31.542, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.189e-04, train_time=1.470
+[gpub022:0/16] 2024-02-02 18:57:16,886 (trainer:737) INFO: 21epoch:train:13101-13200batch: iter_time=8.277e-05, forward_time=0.328, loss_ctc=49.799, loss_att=48.257, acc=0.759, loss=48.719, backward_time=0.459, grad_norm=36.350, clip=100.000, loss_scale=7.892e+33, optim_step_time=0.100, optim0_lr0=2.189e-04, train_time=1.709
+[gpub022:0/16] 2024-02-02 18:59:51,781 (trainer:737) INFO: 21epoch:train:13201-13300batch: iter_time=7.764e-05, forward_time=0.289, loss_ctc=48.161, loss_att=43.784, acc=0.746, loss=45.097, backward_time=0.401, grad_norm=35.131, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.188e-04, train_time=1.549
+[gpub022:0/16] 2024-02-02 19:02:41,189 (trainer:737) INFO: 21epoch:train:13301-13400batch: iter_time=9.021e-05, forward_time=0.352, loss_ctc=52.040, loss_att=52.845, acc=0.711, loss=52.603, backward_time=0.466, grad_norm=37.945, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.188e-04, train_time=1.694
+[gpub022:0/16] 2024-02-02 19:05:16,635 (trainer:737) INFO: 21epoch:train:13401-13500batch: iter_time=8.209e-05, forward_time=0.316, loss_ctc=47.823, loss_att=55.140, acc=0.741, loss=52.945, backward_time=0.412, grad_norm=32.394, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.188e-04, train_time=1.555
+[gpub022:0/16] 2024-02-02 19:07:54,704 (trainer:737) INFO: 21epoch:train:13501-13600batch: iter_time=8.845e-05, forward_time=0.362, loss_ctc=48.577, loss_att=55.837, acc=0.737, loss=53.659, backward_time=0.416, grad_norm=33.691, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.187e-04, train_time=1.580
+[gpub022:0/16] 2024-02-02 19:09:42,109 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub022:0/16] 2024-02-02 19:10:36,179 (trainer:737) INFO: 21epoch:train:13601-13700batch: iter_time=8.247e-05, forward_time=0.292, loss_ctc=48.235, loss_att=51.225, acc=0.760, loss=50.328, backward_time=0.406, grad_norm=33.115, clip=100.000, loss_scale=8.496e+33, optim_step_time=0.093, optim0_lr0=2.187e-04, train_time=1.615
+[gpub022:0/16] 2024-02-02 19:12:17,207 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub022:0/16] 2024-02-02 19:12:36,325 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 19:12:40,184 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fdfef196b90>)
+[gpub022:0/16] 2024-02-02 19:12:40,184 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub022:0/16] 2024-02-02 19:12:40,190 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub022:0/16] 2024-02-02 19:18:24,584 (trainer:737) INFO: 21epoch:train:13701-13800batch: iter_time=3.121, forward_time=0.362, loss_ctc=46.690, loss_att=44.974, acc=0.728, loss=45.489, backward_time=0.411, grad_norm=36.524, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.187e-04, train_time=4.684
+[gpub022:0/16] 2024-02-02 19:20:50,759 (trainer:737) INFO: 21epoch:train:13801-13900batch: iter_time=7.585e-05, forward_time=0.294, loss_ctc=47.945, loss_att=43.304, acc=0.772, loss=44.696, backward_time=0.404, grad_norm=35.267, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.186e-04, train_time=1.462
+[gpub022:0/16] 2024-02-02 19:23:43,688 (trainer:737) INFO: 21epoch:train:13901-14000batch: iter_time=7.962e-05, forward_time=0.337, loss_ctc=46.359, loss_att=47.714, acc=0.747, loss=47.307, backward_time=0.465, grad_norm=32.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.116, optim0_lr0=2.186e-04, train_time=1.729
+[gpub022:0/16] 2024-02-02 19:25:58,472 (trainer:737) INFO: 21epoch:train:14001-14100batch: iter_time=8.911e-05, forward_time=0.295, loss_ctc=49.900, loss_att=51.363, acc=0.748, loss=50.924, backward_time=0.408, grad_norm=35.339, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.186e-04, train_time=1.348
+[gpub022:0/16] 2024-02-02 19:28:48,257 (trainer:737) INFO: 21epoch:train:14101-14200batch: iter_time=1.037e-04, forward_time=0.360, loss_ctc=45.141, loss_att=44.268, acc=0.777, loss=44.530, backward_time=0.419, grad_norm=29.361, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.185e-04, train_time=1.698
+[gpub022:0/16] 2024-02-02 19:31:11,613 (trainer:737) INFO: 21epoch:train:14201-14300batch: iter_time=8.648e-05, forward_time=0.290, loss_ctc=40.558, loss_att=36.473, acc=0.776, loss=37.699, backward_time=0.401, grad_norm=29.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.185e-04, train_time=1.433
+[gpub022:0/16] 2024-02-02 19:33:33,319 (trainer:737) INFO: 21epoch:train:14301-14400batch: iter_time=9.412e-05, forward_time=0.293, loss_ctc=45.386, loss_att=49.367, acc=0.756, loss=48.173, backward_time=0.405, grad_norm=35.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.185e-04, train_time=1.417
+[gpub022:0/16] 2024-02-02 19:36:22,969 (trainer:737) INFO: 21epoch:train:14401-14500batch: iter_time=2.576e-04, forward_time=0.339, loss_ctc=50.649, loss_att=45.694, acc=0.765, loss=47.180, backward_time=0.464, grad_norm=35.381, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.184e-04, train_time=1.696
+[gpub022:0/16] 2024-02-02 19:38:48,534 (trainer:737) INFO: 21epoch:train:14501-14600batch: iter_time=8.970e-05, forward_time=0.289, loss_ctc=47.927, loss_att=43.403, acc=0.745, loss=44.760, backward_time=0.403, grad_norm=34.945, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.184e-04, train_time=1.455
+[gpub022:0/16] 2024-02-02 19:41:05,625 (trainer:737) INFO: 21epoch:train:14601-14700batch: iter_time=9.207e-05, forward_time=0.292, loss_ctc=46.774, loss_att=52.378, acc=0.740, loss=50.697, backward_time=0.405, grad_norm=30.796, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.184e-04, train_time=1.371
+[gpub022:0/16] 2024-02-02 19:43:58,350 (trainer:737) INFO: 21epoch:train:14701-14800batch: iter_time=4.352e-04, forward_time=0.355, loss_ctc=52.385, loss_att=61.827, acc=0.745, loss=58.995, backward_time=0.446, grad_norm=34.476, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.183e-04, train_time=1.727
+[gpub022:0/16] 2024-02-02 19:46:13,421 (trainer:737) INFO: 21epoch:train:14801-14900batch: iter_time=8.995e-05, forward_time=0.293, loss_ctc=49.567, loss_att=51.160, acc=0.756, loss=50.682, backward_time=0.408, grad_norm=32.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.183e-04, train_time=1.350
+[gpub022:0/16] 2024-02-02 19:49:22,455 (trainer:737) INFO: 21epoch:train:14901-15000batch: iter_time=1.002e-04, forward_time=0.356, loss_ctc=41.506, loss_att=43.495, acc=0.752, loss=42.899, backward_time=0.436, grad_norm=30.818, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.183e-04, train_time=1.890
+[gpub022:0/16] 2024-02-02 20:26:47,099 (trainer:343) INFO: 21epoch results: [train] iter_time=0.247, forward_time=0.321, loss_ctc=47.938, loss_att=47.939, acc=0.748, loss=47.938, backward_time=0.420, grad_norm=34.447, clip=100.000, loss_scale=5.558e+33, optim_step_time=0.096, optim0_lr0=2.209e-04, train_time=1.795, time=7 hours, 29 minutes and 11.92 seconds, total_count=345000, gpu_max_cached_mem_GB=41.891, [valid] loss_ctc=41.764, cer_ctc=0.209, loss_att=43.571, acc=0.655, cer=0.321, wer=0.998, loss=43.029, time=37 minutes and 0.61 seconds, total_count=107433, gpu_max_cached_mem_GB=41.891
+[gpub022:0/16] 2024-02-02 20:26:57,139 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub022:0/16] 2024-02-02 20:26:57,255 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/16epoch.pth
+[gpub022:0/16] 2024-02-02 20:26:57,256 (trainer:272) INFO: 22/45epoch started. Estimated time to finish: 1 week, 1 day and 8 hours
+[gpub022:0/16] 2024-02-02 20:26:57,265 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub022:0/16] 2024-02-02 20:27:15,321 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub022:0/16] 2024-02-02 20:27:19,022 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd85af090c0>)
+[gpub022:0/16] 2024-02-02 20:27:19,022 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub022:0/16] 2024-02-02 20:27:19,025 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2913929.0 ON gpub022 CANCELLED AT 2024-02-02T20:28:23 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.12.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.12.log
new file mode 100644
index 0000000000000000000000000000000000000000..0039276a5604214a03fa5ff80e80f27ef711f48c
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.12.log
@@ -0,0 +1,2570 @@
+# Running on gpub010.delta.ncsa.illinois.edu
+# Started at Sun Jan 28 15:01:55 CST 2024
+# SLURMD_NODENAME=gpub010
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2904202
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1706648496
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2904202
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[010,013,072,096]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1706475696
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[010,013,072,096]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=1808814
+# SLURM_TOPOLOGY_ADDR=ss00.ss09.gpub010
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_9ef50019-41d1-40ae-9a78-a780753c8878 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_9ef50019-41d1-40ae-9a78-a780753c8878
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_9ef50019-41d1-40ae-9a78-a780753c8878
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_9ef50019-41d1-40ae-9a78-a780753c8878
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_9ef50019-41d1-40ae-9a78-a780753c8878
+[gpub010:0/16] 2024-01-28 15:05:36,449 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub010:0/16] 2024-01-28 15:05:36,475 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub010:0/16] 2024-01-28 15:05:36,517 (s2t:464) INFO: Vocabulary size: 50002
+[gpub010:0/16] 2024-01-28 15:05:45,374 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub010:0/16] 2024-01-28 15:05:45,380 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub010:0/16] 2024-01-28 15:05:45,380 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub010:0/16] 2024-01-28 15:05:45,380 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub010:0/16] 2024-01-28 15:05:45,399 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub010:0/16] 2024-01-28 15:05:50,729 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 15:05:51,655 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb7de7a6380>)
+[gpub010:0/16] 2024-01-28 15:05:51,655 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub010:0/16] 2024-01-28 15:05:51,657 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 15:06:03,201 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub010:1808893:1808893 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:1808893:1808893 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub010:1808893:1808893 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub010:0/16] 2024-01-28 15:06:08,624 (trainer:284) INFO: 13/45epoch started
+[gpub010:0/16] 2024-01-28 15:06:08,669 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-28 15:06:26,526 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 15:06:29,890 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb5f72292a0>)
+[gpub010:0/16] 2024-01-28 15:06:29,890 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-01-28 15:06:29,893 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub013:1867174:1867174 [0] NCCL INFO cudaDriverVersion 12020
+gpub013:1867174:1867174 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.113<0>
+gpub013:1867174:1867174 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub013:1867174:1867227 [0] NCCL INFO NET/IB : No device found.
+gpub013:1867174:1867227 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.113<0> [1]hsn0:141.142.145.113<0>
+gpub013:1867174:1867227 [0] NCCL INFO Using network Socket
+gpub013:1867174:1867227 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub013:1867174:1867227 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub013:1867174:1867227 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub013:1867174:1867227 [0] NCCL INFO Connected all rings
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/Socket/1
+gpub013:1867174:1867227 [0] NCCL INFO Connected all trees
+gpub013:1867174:1867227 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub013:1867174:1867227 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub013:1867174:1867227 [0] NCCL INFO comm 0x16735ff0 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub072:588459:588459 [2] NCCL INFO cudaDriverVersion 12020
+gpub072:588459:588459 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.172<0>
+gpub072:588459:588459 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub072:588459:588520 [2] NCCL INFO NET/IB : No device found.
+gpub072:588459:588520 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.172<0> [1]hsn0:141.142.145.172<0>
+gpub072:588459:588520 [2] NCCL INFO Using network Socket
+gpub072:588459:588520 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub072:588459:588520 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub072:588459:588520 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub072:588459:588520 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub072:588459:588520 [2] NCCL INFO Connected all rings
+gpub072:588459:588520 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub096:577141:577141 [2] NCCL INFO cudaDriverVersion 12020
+gpub096:577141:577141 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.196<0>
+gpub096:577141:577141 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub096:577141:577201 [2] NCCL INFO NET/IB : No device found.
+gpub096:577141:577201 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.196<0> [1]hsn0:141.142.145.196<0> [2]eth0:fe80::9229:91c2:3352:9f4c%eth0<0>
+gpub096:577141:577201 [2] NCCL INFO Using network Socket
+gpub096:577141:577201 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub096:577141:577201 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub096:577141:577201 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub096:577141:577201 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub096:577141:577201 [2] NCCL INFO Connected all rings
+gpub096:577141:577201 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub013:1867177:1867177 [3] NCCL INFO cudaDriverVersion 12020
+gpub013:1867177:1867177 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.113<0>
+gpub013:1867177:1867177 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub013:1867177:1867226 [3] NCCL INFO NET/IB : No device found.
+gpub013:1867177:1867226 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.113<0> [1]hsn0:141.142.145.113<0>
+gpub013:1867177:1867226 [3] NCCL INFO Using network Socket
+gpub013:1867177:1867226 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub013:1867177:1867226 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub013:1867177:1867226 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub013:1867177:1867226 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub013:1867177:1867226 [3] NCCL INFO Connected all rings
+gpub013:1867177:1867226 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub072:588459:588520 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub072:588459:588520 [2] NCCL INFO Connected all trees
+gpub072:588459:588520 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub072:588459:588520 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub072:588459:588520 [2] NCCL INFO comm 0x1ab85400 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub096:577141:577201 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub096:577141:577201 [2] NCCL INFO Connected all trees
+gpub096:577141:577201 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub096:577141:577201 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub096:577141:577201 [2] NCCL INFO comm 0x14878610 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub013:1867177:1867226 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub013:1867177:1867226 [3] NCCL INFO Connected all trees
+gpub013:1867177:1867226 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub013:1867177:1867226 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub013:1867177:1867226 [3] NCCL INFO comm 0x1400c000 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub072:588460:588460 [3] NCCL INFO cudaDriverVersion 12020
+gpub072:588460:588460 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.172<0>
+gpub072:588460:588460 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub072:588460:588519 [3] NCCL INFO NET/IB : No device found.
+gpub072:588460:588519 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.172<0> [1]hsn0:141.142.145.172<0>
+gpub072:588460:588519 [3] NCCL INFO Using network Socket
+gpub072:588460:588519 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub072:588460:588519 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub072:588460:588519 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub072:588460:588519 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub072:588460:588519 [3] NCCL INFO Connected all rings
+gpub072:588460:588519 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub013:1867175:1867175 [1] NCCL INFO cudaDriverVersion 12020
+gpub013:1867175:1867175 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.113<0>
+gpub013:1867175:1867175 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub013:1867175:1867229 [1] NCCL INFO NET/IB : No device found.
+gpub013:1867175:1867229 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.113<0> [1]hsn0:141.142.145.113<0>
+gpub013:1867175:1867229 [1] NCCL INFO Using network Socket
+gpub013:1867175:1867229 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub013:1867175:1867229 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub013:1867175:1867229 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub013:1867175:1867229 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub013:1867175:1867229 [1] NCCL INFO Connected all rings
+gpub013:1867175:1867229 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/Socket/1
+gpub072:588460:588519 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub072:588460:588519 [3] NCCL INFO Connected all trees
+gpub072:588460:588519 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub072:588460:588519 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub072:588460:588519 [3] NCCL INFO comm 0x17ea23b0 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub013:1867175:1867229 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/Socket/1
+gpub013:1867175:1867229 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub013:1867175:1867229 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub013:1867175:1867229 [1] NCCL INFO Connected all trees
+gpub013:1867175:1867229 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub013:1867175:1867229 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub013:1867175:1867229 [1] NCCL INFO comm 0x155434a0 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub072:588457:588457 [0] NCCL INFO cudaDriverVersion 12020
+gpub072:588457:588457 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.172<0>
+gpub072:588457:588457 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub072:588457:588518 [0] NCCL INFO NET/IB : No device found.
+gpub072:588457:588518 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.172<0> [1]hsn0:141.142.145.172<0>
+gpub072:588457:588518 [0] NCCL INFO Using network Socket
+gpub072:588457:588518 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub072:588457:588518 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub072:588457:588518 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub072:588457:588518 [0] NCCL INFO Connected all rings
+gpub072:588457:588518 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/Socket/1
+gpub072:588457:588518 [0] NCCL INFO Connected all trees
+gpub072:588457:588518 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub072:588457:588518 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub072:588457:588518 [0] NCCL INFO comm 0xf582490 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub010:1808893:1808947 [0] NCCL INFO NET/IB : No device found.
+gpub010:1808893:1808947 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.110<0> [1]hsn0:141.142.145.110<0> [2]eth0:fe80::81a7:cd6e:eca1:d6c4%eth0<0>
+gpub010:1808893:1808947 [0] NCCL INFO Using network Socket
+gpub010:1808893:1808947 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub010:1808893:1808947 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub010:1808893:1808947 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub010:1808893:1808947 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub010:1808893:1808947 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub010:1808893:1808947 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub010:1808893:1808947 [0] NCCL INFO Connected all rings
+gpub010:1808893:1808947 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub010:1808893:1808947 [0] NCCL INFO Connected all trees
+gpub010:1808893:1808947 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:1808893:1808947 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:1808893:1808947 [0] NCCL INFO comm 0x1c26c200 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub010:1808896:1808896 [3] NCCL INFO cudaDriverVersion 12020
+gpub010:1808896:1808896 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:1808896:1808896 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub010:1808896:1808949 [3] NCCL INFO NET/IB : No device found.
+gpub010:1808896:1808949 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.110<0> [1]hsn0:141.142.145.110<0> [2]eth0:fe80::81a7:cd6e:eca1:d6c4%eth0<0>
+gpub010:1808896:1808949 [3] NCCL INFO Using network Socket
+gpub010:1808896:1808949 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub010:1808896:1808949 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub010:1808896:1808949 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub010:1808896:1808949 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub010:1808896:1808949 [3] NCCL INFO Connected all rings
+gpub010:1808896:1808949 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub010:1808896:1808949 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub010:1808896:1808949 [3] NCCL INFO Connected all trees
+gpub010:1808896:1808949 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:1808896:1808949 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:1808896:1808949 [3] NCCL INFO comm 0xda4cb10 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub010:1808895:1808895 [2] NCCL INFO cudaDriverVersion 12020
+gpub010:1808895:1808895 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:1808895:1808895 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub010:1808895:1808950 [2] NCCL INFO NET/IB : No device found.
+gpub010:1808895:1808950 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.110<0> [1]hsn0:141.142.145.110<0> [2]eth0:fe80::81a7:cd6e:eca1:d6c4%eth0<0>
+gpub010:1808895:1808950 [2] NCCL INFO Using network Socket
+gpub010:1808895:1808950 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub010:1808895:1808950 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub010:1808895:1808950 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub010:1808895:1808950 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub010:1808895:1808950 [2] NCCL INFO Connected all rings
+gpub010:1808895:1808950 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub010:1808895:1808950 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub010:1808895:1808950 [2] NCCL INFO Connected all trees
+gpub010:1808895:1808950 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:1808895:1808950 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:1808895:1808950 [2] NCCL INFO comm 0x15436610 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub010:1808894:1808894 [1] NCCL INFO cudaDriverVersion 12020
+gpub010:1808894:1808894 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:1808894:1808894 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub010:1808894:1808948 [1] NCCL INFO NET/IB : No device found.
+gpub010:1808894:1808948 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.110<0> [1]hsn0:141.142.145.110<0> [2]eth0:fe80::81a7:cd6e:eca1:d6c4%eth0<0>
+gpub010:1808894:1808948 [1] NCCL INFO Using network Socket
+gpub010:1808894:1808948 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub010:1808894:1808948 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub010:1808894:1808948 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub010:1808894:1808948 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub010:1808894:1808948 [1] NCCL INFO Connected all rings
+gpub010:1808894:1808948 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub010:1808894:1808948 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub010:1808894:1808948 [1] NCCL INFO Connected all trees
+gpub010:1808894:1808948 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:1808894:1808948 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:1808894:1808948 [1] NCCL INFO comm 0x1ca388d0 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub096:577142:577142 [3] NCCL INFO cudaDriverVersion 12020
+gpub096:577142:577142 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.196<0>
+gpub096:577142:577142 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub096:577142:577200 [3] NCCL INFO NET/IB : No device found.
+gpub096:577142:577200 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.196<0> [1]hsn0:141.142.145.196<0> [2]eth0:fe80::9229:91c2:3352:9f4c%eth0<0>
+gpub096:577142:577200 [3] NCCL INFO Using network Socket
+gpub096:577142:577200 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub096:577142:577200 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub096:577142:577200 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub096:577142:577200 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub096:577142:577200 [3] NCCL INFO Connected all rings
+gpub096:577142:577200 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub096:577142:577200 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub096:577142:577200 [3] NCCL INFO Connected all trees
+gpub096:577142:577200 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub096:577142:577200 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub096:577142:577200 [3] NCCL INFO comm 0x163a2fa0 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub072:588458:588458 [1] NCCL INFO cudaDriverVersion 12020
+gpub072:588458:588458 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.172<0>
+gpub072:588458:588458 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub072:588458:588517 [1] NCCL INFO NET/IB : No device found.
+gpub072:588458:588517 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.172<0> [1]hsn0:141.142.145.172<0>
+gpub072:588458:588517 [1] NCCL INFO Using network Socket
+gpub072:588458:588517 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub072:588458:588517 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub072:588458:588517 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub072:588458:588517 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub072:588458:588517 [1] NCCL INFO Connected all rings
+gpub072:588458:588517 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/Socket/1
+gpub072:588458:588517 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/Socket/1
+gpub072:588458:588517 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub072:588458:588517 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub072:588458:588517 [1] NCCL INFO Connected all trees
+gpub072:588458:588517 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub072:588458:588517 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub072:588458:588517 [1] NCCL INFO comm 0x12e3c790 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub096:577140:577140 [1] NCCL INFO cudaDriverVersion 12020
+gpub096:577140:577140 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.196<0>
+gpub096:577140:577140 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub096:577140:577199 [1] NCCL INFO NET/IB : No device found.
+gpub096:577140:577199 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.196<0> [1]hsn0:141.142.145.196<0> [2]eth0:fe80::9229:91c2:3352:9f4c%eth0<0>
+gpub096:577140:577199 [1] NCCL INFO Using network Socket
+gpub096:577140:577199 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub096:577140:577199 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub096:577140:577199 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub096:577140:577199 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub096:577140:577199 [1] NCCL INFO Connected all rings
+gpub096:577140:577199 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub096:577140:577199 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub096:577140:577199 [1] NCCL INFO Connected all trees
+gpub096:577140:577199 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub096:577140:577199 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub096:577140:577199 [1] NCCL INFO comm 0x1748b460 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub013:1867176:1867176 [2] NCCL INFO cudaDriverVersion 12020
+gpub013:1867176:1867176 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.113<0>
+gpub013:1867176:1867176 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub013:1867176:1867228 [2] NCCL INFO NET/IB : No device found.
+gpub013:1867176:1867228 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.113<0> [1]hsn0:141.142.145.113<0>
+gpub013:1867176:1867228 [2] NCCL INFO Using network Socket
+gpub013:1867176:1867228 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub013:1867176:1867228 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub013:1867176:1867228 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub013:1867176:1867228 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub013:1867176:1867228 [2] NCCL INFO Connected all rings
+gpub013:1867176:1867228 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub013:1867176:1867228 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub013:1867176:1867228 [2] NCCL INFO Connected all trees
+gpub013:1867176:1867228 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub013:1867176:1867228 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub013:1867176:1867228 [2] NCCL INFO comm 0x1579c4f0 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub096:577139:577139 [0] NCCL INFO cudaDriverVersion 12020
+gpub096:577139:577139 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.196<0>
+gpub096:577139:577139 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub096:577139:577202 [0] NCCL INFO NET/IB : No device found.
+gpub096:577139:577202 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.196<0> [1]hsn0:141.142.145.196<0> [2]eth0:fe80::9229:91c2:3352:9f4c%eth0<0>
+gpub096:577139:577202 [0] NCCL INFO Using network Socket
+gpub096:577139:577202 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub096:577139:577202 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub096:577139:577202 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub096:577139:577202 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub096:577139:577202 [0] NCCL INFO Connected all rings
+gpub096:577139:577202 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/Socket/1
+gpub096:577139:577202 [0] NCCL INFO Connected all trees
+gpub096:577139:577202 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub096:577139:577202 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub096:577139:577202 [0] NCCL INFO comm 0x171a4770 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+[gpub010:0/16] 2024-01-28 15:14:27,611 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub010:0/16] 2024-01-28 15:17:37,381 (trainer:737) INFO: 13epoch:train:1-100batch: iter_time=4.851, forward_time=0.604, loss_ctc=66.074, loss_att=62.668, acc=0.690, loss=63.690, backward_time=0.563, grad_norm=38.933, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.119, optim0_lr0=2.886e-04, train_time=6.875
+[gpub010:0/16] 2024-01-28 15:20:43,801 (trainer:737) INFO: 13epoch:train:101-200batch: iter_time=8.525e-04, forward_time=0.569, loss_ctc=51.175, loss_att=54.576, acc=0.704, loss=53.556, backward_time=0.523, grad_norm=31.595, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.111, optim0_lr0=2.886e-04, train_time=1.874
+[gpub010:0/16] 2024-01-28 15:23:49,393 (trainer:737) INFO: 13epoch:train:201-300batch: iter_time=2.730e-04, forward_time=0.485, loss_ctc=69.042, loss_att=61.789, acc=0.704, loss=63.965, backward_time=0.519, grad_norm=43.588, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.102, optim0_lr0=2.885e-04, train_time=1.856
+[gpub010:0/16] 2024-01-28 15:26:52,612 (trainer:737) INFO: 13epoch:train:301-400batch: iter_time=5.821e-04, forward_time=0.535, loss_ctc=68.647, loss_att=62.215, acc=0.726, loss=64.145, backward_time=0.487, grad_norm=35.438, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.104, optim0_lr0=2.884e-04, train_time=1.833
+[gpub010:0/16] 2024-01-28 15:29:41,550 (trainer:737) INFO: 13epoch:train:401-500batch: iter_time=0.001, forward_time=0.489, loss_ctc=69.470, loss_att=58.680, acc=0.690, loss=61.917, backward_time=0.490, grad_norm=39.241, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.103, optim0_lr0=2.883e-04, train_time=1.689
+[gpub010:0/16] 2024-01-28 15:32:42,770 (trainer:737) INFO: 13epoch:train:501-600batch: iter_time=0.001, forward_time=0.512, loss_ctc=45.513, loss_att=40.988, acc=0.726, loss=42.345, backward_time=0.478, grad_norm=29.596, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.109, optim0_lr0=2.882e-04, train_time=1.813
+[gpub010:0/16] 2024-01-28 15:35:44,628 (trainer:737) INFO: 13epoch:train:601-700batch: iter_time=4.440e-04, forward_time=0.469, loss_ctc=60.484, loss_att=56.909, acc=0.721, loss=57.982, backward_time=0.487, grad_norm=32.677, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.100, optim0_lr0=2.882e-04, train_time=1.818
+[gpub010:0/16] 2024-01-28 15:38:42,906 (trainer:737) INFO: 13epoch:train:701-800batch: iter_time=4.578e-04, forward_time=0.534, loss_ctc=50.878, loss_att=49.465, acc=0.747, loss=49.889, backward_time=0.501, grad_norm=27.035, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.102, optim0_lr0=2.881e-04, train_time=1.782
+[gpub010:0/16] 2024-01-28 15:41:28,009 (trainer:737) INFO: 13epoch:train:801-900batch: iter_time=0.001, forward_time=0.459, loss_ctc=60.589, loss_att=56.385, acc=0.736, loss=57.646, backward_time=0.506, grad_norm=31.863, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.101, optim0_lr0=2.880e-04, train_time=1.651
+[gpub010:0/16] 2024-01-28 15:44:18,228 (trainer:737) INFO: 13epoch:train:901-1000batch: iter_time=5.075e-04, forward_time=0.468, loss_ctc=59.426, loss_att=50.820, acc=0.718, loss=53.402, backward_time=0.472, grad_norm=34.556, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.101, optim0_lr0=2.879e-04, train_time=1.702
+[gpub010:0/16] 2024-01-28 15:47:00,984 (trainer:737) INFO: 13epoch:train:1001-1100batch: iter_time=6.327e-04, forward_time=0.421, loss_ctc=60.368, loss_att=50.436, acc=0.727, loss=53.416, backward_time=0.451, grad_norm=36.210, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.101, optim0_lr0=2.878e-04, train_time=1.628
+[gpub010:0/16] 2024-01-28 15:49:57,810 (trainer:737) INFO: 13epoch:train:1101-1200batch: iter_time=5.785e-04, forward_time=0.440, loss_ctc=62.132, loss_att=56.537, acc=0.736, loss=58.216, backward_time=0.449, grad_norm=32.459, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.099, optim0_lr0=2.878e-04, train_time=1.766
+[gpub010:0/16] 2024-01-28 15:51:40,024 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-01-28 15:51:58,482 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 15:52:01,868 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb5f729feb0>)
+[gpub010:0/16] 2024-01-28 15:52:01,869 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-01-28 15:52:01,934 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 15:58:12,623 (trainer:737) INFO: 13epoch:train:1201-1300batch: iter_time=3.247, forward_time=0.348, loss_ctc=67.618, loss_att=62.085, acc=0.693, loss=63.745, backward_time=0.437, grad_norm=40.280, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.094, optim0_lr0=2.877e-04, train_time=4.950
+[gpub010:0/16] 2024-01-28 16:00:59,169 (trainer:737) INFO: 13epoch:train:1301-1400batch: iter_time=8.079e-05, forward_time=0.317, loss_ctc=52.719, loss_att=57.334, acc=0.697, loss=55.950, backward_time=0.403, grad_norm=33.338, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.091, optim0_lr0=2.876e-04, train_time=1.665
+[gpub010:0/16] 2024-01-28 16:03:28,939 (trainer:737) INFO: 13epoch:train:1401-1500batch: iter_time=8.004e-05, forward_time=0.348, loss_ctc=50.977, loss_att=51.454, acc=0.714, loss=51.311, backward_time=0.454, grad_norm=30.279, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.102, optim0_lr0=2.875e-04, train_time=1.497
+[gpub010:0/16] 2024-01-28 16:05:52,557 (trainer:737) INFO: 13epoch:train:1501-1600batch: iter_time=8.779e-05, forward_time=0.295, loss_ctc=77.672, loss_att=67.212, acc=0.721, loss=70.350, backward_time=0.408, grad_norm=44.008, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.091, optim0_lr0=2.874e-04, train_time=1.436
+[gpub010:0/16] 2024-01-28 16:08:36,927 (trainer:737) INFO: 13epoch:train:1601-1700batch: iter_time=8.102e-05, forward_time=0.336, loss_ctc=64.719, loss_att=60.719, acc=0.697, loss=61.919, backward_time=0.416, grad_norm=34.556, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.091, optim0_lr0=2.874e-04, train_time=1.643
+[gpub010:0/16] 2024-01-28 16:11:22,966 (trainer:737) INFO: 13epoch:train:1701-1800batch: iter_time=7.758e-05, forward_time=0.349, loss_ctc=56.681, loss_att=45.850, acc=0.729, loss=49.100, backward_time=0.426, grad_norm=33.011, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.095, optim0_lr0=2.873e-04, train_time=1.660
+[gpub010:0/16] 2024-01-28 16:13:58,576 (trainer:737) INFO: 13epoch:train:1801-1900batch: iter_time=8.334e-05, forward_time=0.296, loss_ctc=48.133, loss_att=40.652, acc=0.737, loss=42.896, backward_time=0.404, grad_norm=27.348, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.091, optim0_lr0=2.872e-04, train_time=1.555
+[gpub010:0/16] 2024-01-28 16:16:29,589 (trainer:737) INFO: 13epoch:train:1901-2000batch: iter_time=8.651e-05, forward_time=0.293, loss_ctc=59.730, loss_att=60.809, acc=0.735, loss=60.485, backward_time=0.407, grad_norm=31.160, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.091, optim0_lr0=2.871e-04, train_time=1.510
+[gpub010:0/16] 2024-01-28 16:18:58,142 (trainer:737) INFO: 13epoch:train:2001-2100batch: iter_time=2.748e-04, forward_time=0.379, loss_ctc=45.455, loss_att=49.816, acc=0.738, loss=48.508, backward_time=0.445, grad_norm=25.639, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.097, optim0_lr0=2.870e-04, train_time=1.484
+[gpub010:0/16] 2024-01-28 16:21:43,971 (trainer:737) INFO: 13epoch:train:2101-2200batch: iter_time=9.702e-05, forward_time=0.310, loss_ctc=65.616, loss_att=51.313, acc=0.738, loss=55.604, backward_time=0.412, grad_norm=35.308, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.092, optim0_lr0=2.870e-04, train_time=1.659
+[gpub010:0/16] 2024-01-28 16:24:04,232 (trainer:737) INFO: 13epoch:train:2201-2300batch: iter_time=8.561e-05, forward_time=0.289, loss_ctc=56.296, loss_att=50.849, acc=0.711, loss=52.483, backward_time=0.400, grad_norm=32.716, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.869e-04, train_time=1.402
+[gpub010:0/16] 2024-01-28 16:26:37,722 (trainer:737) INFO: 13epoch:train:2301-2400batch: iter_time=3.296e-04, forward_time=0.365, loss_ctc=55.955, loss_att=47.137, acc=0.747, loss=49.783, backward_time=0.441, grad_norm=29.908, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.098, optim0_lr0=2.868e-04, train_time=1.535
+[gpub010:0/16] 2024-01-28 16:29:31,460 (trainer:737) INFO: 13epoch:train:2401-2500batch: iter_time=8.652e-05, forward_time=0.334, loss_ctc=66.936, loss_att=64.730, acc=0.713, loss=65.392, backward_time=0.414, grad_norm=38.841, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.093, optim0_lr0=2.867e-04, train_time=1.736
+[gpub010:0/16] 2024-01-28 16:29:51,618 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-01-28 16:30:09,759 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 16:30:13,402 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb5f6c36c80>)
+[gpub010:0/16] 2024-01-28 16:30:13,402 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-28 16:30:13,405 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 16:40:41,895 (trainer:737) INFO: 13epoch:train:2501-2600batch: iter_time=3.211, forward_time=0.285, loss_ctc=61.443, loss_att=59.119, acc=0.698, loss=59.816, backward_time=0.390, grad_norm=36.200, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.867e-04, train_time=6.705
+[gpub010:0/16] 2024-01-28 16:43:44,359 (trainer:737) INFO: 13epoch:train:2601-2700batch: iter_time=8.341e-05, forward_time=0.316, loss_ctc=49.017, loss_att=53.498, acc=0.705, loss=52.154, backward_time=0.415, grad_norm=32.093, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.092, optim0_lr0=2.866e-04, train_time=1.824
+[gpub010:0/16] 2024-01-28 16:46:46,077 (trainer:737) INFO: 13epoch:train:2701-2800batch: iter_time=8.554e-05, forward_time=0.376, loss_ctc=66.238, loss_att=59.018, acc=0.712, loss=61.184, backward_time=0.427, grad_norm=41.129, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.095, optim0_lr0=2.865e-04, train_time=1.817
+[gpub010:0/16] 2024-01-28 16:49:39,727 (trainer:737) INFO: 13epoch:train:2801-2900batch: iter_time=8.474e-05, forward_time=0.294, loss_ctc=64.100, loss_att=58.043, acc=0.736, loss=59.860, backward_time=0.405, grad_norm=34.174, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.864e-04, train_time=1.735
+[gpub010:0/16] 2024-01-28 16:52:31,139 (trainer:737) INFO: 13epoch:train:2901-3000batch: iter_time=9.119e-05, forward_time=0.291, loss_ctc=65.799, loss_att=55.937, acc=0.699, loss=58.895, backward_time=0.400, grad_norm=36.389, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.863e-04, train_time=1.715
+[gpub010:0/16] 2024-01-28 16:55:21,943 (trainer:737) INFO: 13epoch:train:3001-3100batch: iter_time=8.673e-05, forward_time=0.335, loss_ctc=44.791, loss_att=39.977, acc=0.734, loss=41.421, backward_time=0.413, grad_norm=27.763, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.095, optim0_lr0=2.863e-04, train_time=1.708
+[gpub010:0/16] 2024-01-28 16:58:14,693 (trainer:737) INFO: 13epoch:train:3101-3200batch: iter_time=2.071e-04, forward_time=0.372, loss_ctc=59.232, loss_att=55.357, acc=0.728, loss=56.520, backward_time=0.432, grad_norm=31.202, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.095, optim0_lr0=2.862e-04, train_time=1.727
+[gpub010:0/16] 2024-01-28 17:00:49,258 (trainer:737) INFO: 13epoch:train:3201-3300batch: iter_time=8.887e-05, forward_time=0.289, loss_ctc=49.851, loss_att=47.859, acc=0.753, loss=48.456, backward_time=0.401, grad_norm=26.097, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.861e-04, train_time=1.545
+[gpub010:0/16] 2024-01-28 17:03:30,356 (trainer:737) INFO: 13epoch:train:3301-3400batch: iter_time=8.600e-05, forward_time=0.291, loss_ctc=58.625, loss_att=56.177, acc=0.735, loss=56.912, backward_time=0.401, grad_norm=30.290, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.860e-04, train_time=1.610
+[gpub010:0/16] 2024-01-28 17:06:24,568 (trainer:737) INFO: 13epoch:train:3401-3500batch: iter_time=9.140e-05, forward_time=0.322, loss_ctc=57.637, loss_att=49.336, acc=0.725, loss=51.827, backward_time=0.427, grad_norm=34.423, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.096, optim0_lr0=2.859e-04, train_time=1.742
+[gpub010:0/16] 2024-01-28 17:09:14,548 (trainer:737) INFO: 13epoch:train:3501-3600batch: iter_time=8.967e-04, forward_time=0.369, loss_ctc=58.227, loss_att=49.467, acc=0.730, loss=52.095, backward_time=0.429, grad_norm=31.897, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.097, optim0_lr0=2.859e-04, train_time=1.700
+[gpub010:0/16] 2024-01-28 17:11:51,647 (trainer:737) INFO: 13epoch:train:3601-3700batch: iter_time=8.451e-05, forward_time=0.291, loss_ctc=60.345, loss_att=55.535, acc=0.738, loss=56.978, backward_time=0.401, grad_norm=32.072, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.090, optim0_lr0=2.858e-04, train_time=1.571
+[gpub010:0/16] 2024-01-28 17:13:29,320 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-01-28 17:13:47,769 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 17:13:51,250 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb103097610>)
+[gpub010:0/16] 2024-01-28 17:13:51,250 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-01-28 17:13:51,312 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 17:23:36,571 (trainer:737) INFO: 13epoch:train:3701-3800batch: iter_time=3.280, forward_time=0.333, loss_ctc=66.105, loss_att=60.828, acc=0.699, loss=62.411, backward_time=0.404, grad_norm=38.363, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.091, optim0_lr0=2.857e-04, train_time=7.049
+[gpub010:0/16] 2024-01-28 17:27:57,030 (trainer:737) INFO: 13epoch:train:3801-3900batch: iter_time=8.881e-05, forward_time=0.356, loss_ctc=51.281, loss_att=55.376, acc=0.705, loss=54.147, backward_time=0.420, grad_norm=31.643, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.094, optim0_lr0=2.856e-04, train_time=2.604
+[gpub010:0/16] 2024-01-28 17:33:47,588 (trainer:737) INFO: 13epoch:train:3901-4000batch: iter_time=8.507e-05, forward_time=0.289, loss_ctc=50.508, loss_att=50.832, acc=0.718, loss=50.735, backward_time=0.391, grad_norm=29.561, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.090, optim0_lr0=2.856e-04, train_time=3.506
+[gpub010:0/16] 2024-01-28 17:40:34,472 (trainer:737) INFO: 13epoch:train:4001-4100batch: iter_time=8.601e-05, forward_time=0.291, loss_ctc=77.604, loss_att=66.133, acc=0.724, loss=69.574, backward_time=0.396, grad_norm=40.847, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.090, optim0_lr0=2.855e-04, train_time=4.069
+[gpub010:0/16] 2024-01-28 17:46:55,997 (trainer:737) INFO: 13epoch:train:4101-4200batch: iter_time=8.473e-05, forward_time=0.315, loss_ctc=63.386, loss_att=60.644, acc=0.696, loss=61.467, backward_time=0.426, grad_norm=34.317, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.100, optim0_lr0=2.854e-04, train_time=3.815
+[gpub010:0/16] 2024-01-28 17:55:07,801 (trainer:737) INFO: 13epoch:train:4201-4300batch: iter_time=9.727e-05, forward_time=0.331, loss_ctc=54.213, loss_att=45.218, acc=0.733, loss=47.917, backward_time=0.426, grad_norm=31.697, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.094, optim0_lr0=2.853e-04, train_time=4.917
+[gpub010:0/16] 2024-01-28 17:58:03,480 (trainer:737) INFO: 13epoch:train:4301-4400batch: iter_time=9.173e-05, forward_time=0.286, loss_ctc=47.366, loss_att=39.626, acc=0.741, loss=41.948, backward_time=0.397, grad_norm=26.445, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.852e-04, train_time=1.757
+[gpub010:0/16] 2024-01-28 18:00:51,932 (trainer:737) INFO: 13epoch:train:4401-4500batch: iter_time=8.193e-05, forward_time=0.311, loss_ctc=59.164, loss_att=59.782, acc=0.737, loss=59.596, backward_time=0.404, grad_norm=29.540, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.852e-04, train_time=1.684
+[gpub010:0/16] 2024-01-28 18:03:38,150 (trainer:737) INFO: 13epoch:train:4501-4600batch: iter_time=8.421e-05, forward_time=0.327, loss_ctc=44.225, loss_att=48.881, acc=0.740, loss=47.484, backward_time=0.415, grad_norm=25.317, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.094, optim0_lr0=2.851e-04, train_time=1.662
+[gpub010:0/16] 2024-01-28 18:06:25,915 (trainer:737) INFO: 13epoch:train:4601-4700batch: iter_time=8.544e-05, forward_time=0.340, loss_ctc=63.717, loss_att=50.174, acc=0.743, loss=54.237, backward_time=0.422, grad_norm=33.435, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.095, optim0_lr0=2.850e-04, train_time=1.677
+[gpub010:0/16] 2024-01-28 18:09:23,270 (trainer:737) INFO: 13epoch:train:4701-4800batch: iter_time=7.967e-05, forward_time=0.330, loss_ctc=55.196, loss_att=50.683, acc=0.713, loss=52.037, backward_time=0.431, grad_norm=31.583, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.849e-04, train_time=1.773
+[gpub010:0/16] 2024-01-28 18:12:03,701 (trainer:737) INFO: 13epoch:train:4801-4900batch: iter_time=8.647e-05, forward_time=0.291, loss_ctc=54.725, loss_att=45.863, acc=0.753, loss=48.522, backward_time=0.402, grad_norm=29.294, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.849e-04, train_time=1.604
+[gpub010:0/16] 2024-01-28 18:14:52,577 (trainer:737) INFO: 13epoch:train:4901-5000batch: iter_time=9.025e-05, forward_time=0.332, loss_ctc=65.865, loss_att=63.876, acc=0.716, loss=64.473, backward_time=0.418, grad_norm=37.958, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.092, optim0_lr0=2.848e-04, train_time=1.688
+[gpub010:0/16] 2024-01-28 18:15:12,734 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-01-28 18:15:31,399 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 18:15:34,874 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb5e5113e80>)
+[gpub010:0/16] 2024-01-28 18:15:34,874 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-01-28 18:15:34,939 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 18:22:56,838 (trainer:737) INFO: 13epoch:train:5001-5100batch: iter_time=3.312, forward_time=0.355, loss_ctc=60.765, loss_att=59.960, acc=0.691, loss=60.202, backward_time=0.410, grad_norm=37.070, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.094, optim0_lr0=2.847e-04, train_time=4.843
+[gpub010:0/16] 2024-01-28 18:25:18,840 (trainer:737) INFO: 13epoch:train:5101-5200batch: iter_time=8.866e-05, forward_time=0.288, loss_ctc=48.148, loss_att=55.156, acc=0.698, loss=53.054, backward_time=0.400, grad_norm=32.402, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.846e-04, train_time=1.420
+[gpub010:0/16] 2024-01-28 18:27:53,088 (trainer:737) INFO: 13epoch:train:5201-5300batch: iter_time=8.253e-05, forward_time=0.328, loss_ctc=64.990, loss_att=58.262, acc=0.710, loss=60.281, backward_time=0.421, grad_norm=37.657, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.092, optim0_lr0=2.846e-04, train_time=1.542
+[gpub010:0/16] 2024-01-28 18:30:43,332 (trainer:737) INFO: 13epoch:train:5301-5400batch: iter_time=8.722e-05, forward_time=0.384, loss_ctc=62.843, loss_att=58.152, acc=0.721, loss=59.559, backward_time=0.426, grad_norm=34.948, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.094, optim0_lr0=2.845e-04, train_time=1.702
+[gpub010:0/16] 2024-01-28 18:32:59,262 (trainer:737) INFO: 13epoch:train:5401-5500batch: iter_time=8.580e-05, forward_time=0.291, loss_ctc=64.003, loss_att=55.935, acc=0.694, loss=58.356, backward_time=0.402, grad_norm=38.685, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.844e-04, train_time=1.359
+[gpub010:0/16] 2024-01-28 18:35:10,903 (trainer:737) INFO: 13epoch:train:5501-5600batch: iter_time=8.737e-05, forward_time=0.287, loss_ctc=44.220, loss_att=39.934, acc=0.728, loss=41.220, backward_time=0.399, grad_norm=27.795, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.843e-04, train_time=1.316
+[gpub010:0/16] 2024-01-28 18:38:06,073 (trainer:737) INFO: 13epoch:train:5601-5700batch: iter_time=9.770e-05, forward_time=0.387, loss_ctc=58.187, loss_att=56.352, acc=0.709, loss=56.903, backward_time=0.436, grad_norm=32.360, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.099, optim0_lr0=2.842e-04, train_time=1.751
+[gpub010:0/16] 2024-01-28 18:40:31,217 (trainer:737) INFO: 13epoch:train:5701-5800batch: iter_time=9.398e-05, forward_time=0.289, loss_ctc=48.976, loss_att=48.317, acc=0.749, loss=48.515, backward_time=0.401, grad_norm=27.045, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.842e-04, train_time=1.452
+[gpub010:0/16] 2024-01-28 18:42:48,702 (trainer:737) INFO: 13epoch:train:5801-5900batch: iter_time=9.118e-05, forward_time=0.291, loss_ctc=57.098, loss_att=56.026, acc=0.727, loss=56.348, backward_time=0.404, grad_norm=32.225, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.841e-04, train_time=1.375
+[gpub010:0/16] 2024-01-28 18:45:33,097 (trainer:737) INFO: 13epoch:train:5901-6000batch: iter_time=9.025e-05, forward_time=0.288, loss_ctc=56.809, loss_att=49.551, acc=0.715, loss=51.728, backward_time=0.398, grad_norm=33.587, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.091, optim0_lr0=2.840e-04, train_time=1.644
+[gpub010:0/16] 2024-01-28 18:48:12,598 (trainer:737) INFO: 13epoch:train:6001-6100batch: iter_time=1.014e-04, forward_time=0.370, loss_ctc=57.648, loss_att=47.620, acc=0.732, loss=50.628, backward_time=0.508, grad_norm=31.340, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.104, optim0_lr0=2.839e-04, train_time=1.593
+[gpub010:0/16] 2024-01-28 18:50:33,813 (trainer:737) INFO: 13epoch:train:6101-6200batch: iter_time=8.781e-05, forward_time=0.292, loss_ctc=59.354, loss_att=54.881, acc=0.734, loss=56.223, backward_time=0.402, grad_norm=30.647, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.839e-04, train_time=1.413
+[gpub010:0/16] 2024-01-28 18:52:04,542 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-01-28 18:52:23,130 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 18:52:26,593 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb103054700>)
+[gpub010:0/16] 2024-01-28 18:52:26,593 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-01-28 18:52:26,600 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 18:58:38,938 (trainer:737) INFO: 13epoch:train:6201-6300batch: iter_time=3.340, forward_time=0.292, loss_ctc=65.626, loss_att=59.977, acc=0.694, loss=61.672, backward_time=0.403, grad_norm=38.685, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.838e-04, train_time=4.851
+[gpub010:0/16] 2024-01-28 19:01:14,746 (trainer:737) INFO: 13epoch:train:6301-6400batch: iter_time=8.035e-05, forward_time=0.390, loss_ctc=49.747, loss_att=56.351, acc=0.695, loss=54.370, backward_time=0.444, grad_norm=31.860, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.099, optim0_lr0=2.837e-04, train_time=1.558
+[gpub010:0/16] 2024-01-28 19:03:34,199 (trainer:737) INFO: 13epoch:train:6401-6500batch: iter_time=7.810e-05, forward_time=0.298, loss_ctc=49.942, loss_att=49.631, acc=0.721, loss=49.724, backward_time=0.401, grad_norm=28.557, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.836e-04, train_time=1.393
+[gpub010:0/16] 2024-01-28 19:06:17,820 (trainer:737) INFO: 13epoch:train:6501-6600batch: iter_time=9.536e-05, forward_time=0.294, loss_ctc=75.546, loss_att=65.480, acc=0.719, loss=68.500, backward_time=0.406, grad_norm=40.543, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.836e-04, train_time=1.637
+[gpub010:0/16] 2024-01-28 19:08:41,046 (trainer:737) INFO: 13epoch:train:6601-6700batch: iter_time=9.188e-05, forward_time=0.292, loss_ctc=62.715, loss_att=58.455, acc=0.692, loss=59.733, backward_time=0.405, grad_norm=34.122, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.835e-04, train_time=1.432
+[gpub010:0/16] 2024-01-28 19:11:20,122 (trainer:737) INFO: 13epoch:train:6701-6800batch: iter_time=9.489e-05, forward_time=0.427, loss_ctc=54.200, loss_att=44.529, acc=0.731, loss=47.430, backward_time=0.434, grad_norm=32.539, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.096, optim0_lr0=2.834e-04, train_time=1.591
+[gpub010:0/16] 2024-01-28 19:13:58,130 (trainer:737) INFO: 13epoch:train:6801-6900batch: iter_time=9.027e-05, forward_time=0.286, loss_ctc=47.294, loss_att=40.296, acc=0.733, loss=42.396, backward_time=0.396, grad_norm=29.464, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.833e-04, train_time=1.578
+[gpub010:0/16] 2024-01-28 19:16:20,485 (trainer:737) INFO: 13epoch:train:6901-7000batch: iter_time=8.516e-05, forward_time=0.292, loss_ctc=58.198, loss_att=59.947, acc=0.720, loss=59.423, backward_time=0.404, grad_norm=30.001, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.833e-04, train_time=1.425
+[gpub010:0/16] 2024-01-28 19:18:39,276 (trainer:737) INFO: 13epoch:train:7001-7100batch: iter_time=8.240e-05, forward_time=0.288, loss_ctc=44.247, loss_att=49.019, acc=0.733, loss=47.587, backward_time=0.401, grad_norm=25.816, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.832e-04, train_time=1.388
+[gpub010:0/16] 2024-01-28 19:21:21,454 (trainer:737) INFO: 13epoch:train:7101-7200batch: iter_time=8.652e-05, forward_time=0.391, loss_ctc=63.513, loss_att=50.183, acc=0.736, loss=54.182, backward_time=0.425, grad_norm=34.335, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.098, optim0_lr0=2.831e-04, train_time=1.621
+[gpub010:0/16] 2024-01-28 19:23:58,345 (trainer:737) INFO: 13epoch:train:7201-7300batch: iter_time=8.422e-05, forward_time=0.289, loss_ctc=54.810, loss_att=49.236, acc=0.708, loss=50.908, backward_time=0.398, grad_norm=33.155, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.830e-04, train_time=1.568
+[gpub010:0/16] 2024-01-28 19:26:12,968 (trainer:737) INFO: 13epoch:train:7301-7400batch: iter_time=8.096e-05, forward_time=0.291, loss_ctc=54.383, loss_att=46.342, acc=0.749, loss=48.754, backward_time=0.404, grad_norm=29.019, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.830e-04, train_time=1.347
+[gpub010:0/16] 2024-01-28 19:28:35,912 (trainer:737) INFO: 13epoch:train:7401-7500batch: iter_time=8.245e-05, forward_time=0.293, loss_ctc=65.281, loss_att=63.179, acc=0.709, loss=63.809, backward_time=0.406, grad_norm=36.541, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.829e-04, train_time=1.429
+[gpub010:0/16] 2024-01-28 19:28:55,940 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-01-28 19:29:14,522 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 19:29:18,045 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa73e03f7c0>)
+[gpub010:0/16] 2024-01-28 19:29:18,045 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-01-28 19:29:18,085 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 19:36:12,250 (trainer:737) INFO: 13epoch:train:7501-7600batch: iter_time=3.149, forward_time=0.349, loss_ctc=60.377, loss_att=56.856, acc=0.700, loss=57.912, backward_time=0.409, grad_norm=34.993, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.828e-04, train_time=4.562
+[gpub010:0/16] 2024-01-28 19:38:37,210 (trainer:737) INFO: 13epoch:train:7601-7700batch: iter_time=7.672e-05, forward_time=0.289, loss_ctc=47.418, loss_att=53.219, acc=0.702, loss=51.479, backward_time=0.401, grad_norm=29.977, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.827e-04, train_time=1.451
+[gpub010:0/16] 2024-01-28 19:41:20,326 (trainer:737) INFO: 13epoch:train:7701-7800batch: iter_time=8.439e-05, forward_time=0.300, loss_ctc=65.150, loss_att=57.362, acc=0.714, loss=59.698, backward_time=0.416, grad_norm=36.886, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.093, optim0_lr0=2.827e-04, train_time=1.631
+[gpub010:0/16] 2024-01-28 19:43:52,087 (trainer:737) INFO: 13epoch:train:7801-7900batch: iter_time=9.649e-05, forward_time=0.362, loss_ctc=62.873, loss_att=56.750, acc=0.726, loss=58.587, backward_time=0.447, grad_norm=35.601, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.098, optim0_lr0=2.826e-04, train_time=1.517
+[gpub010:0/16] 2024-01-28 19:46:24,625 (trainer:737) INFO: 13epoch:train:7901-8000batch: iter_time=8.346e-05, forward_time=0.288, loss_ctc=63.200, loss_att=54.878, acc=0.698, loss=57.375, backward_time=0.400, grad_norm=37.831, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.091, optim0_lr0=2.825e-04, train_time=1.525
+[gpub010:0/16] 2024-01-28 19:48:56,551 (trainer:737) INFO: 13epoch:train:8001-8100batch: iter_time=8.925e-05, forward_time=0.285, loss_ctc=43.991, loss_att=38.660, acc=0.732, loss=40.259, backward_time=0.396, grad_norm=27.311, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.824e-04, train_time=1.520
+[gpub010:0/16] 2024-01-28 19:51:30,001 (trainer:737) INFO: 13epoch:train:8101-8200batch: iter_time=8.087e-05, forward_time=0.376, loss_ctc=57.917, loss_att=55.459, acc=0.715, loss=56.196, backward_time=0.448, grad_norm=31.065, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.824e-04, train_time=1.534
+[gpub010:0/16] 2024-01-28 19:54:03,511 (trainer:737) INFO: 13epoch:train:8201-8300batch: iter_time=8.135e-05, forward_time=0.289, loss_ctc=49.255, loss_att=47.442, acc=0.750, loss=47.986, backward_time=0.401, grad_norm=26.428, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.823e-04, train_time=1.535
+[gpub010:0/16] 2024-01-28 19:56:28,573 (trainer:737) INFO: 13epoch:train:8301-8400batch: iter_time=8.062e-05, forward_time=0.291, loss_ctc=56.948, loss_att=55.365, acc=0.729, loss=55.840, backward_time=0.402, grad_norm=29.898, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.822e-04, train_time=1.448
+[gpub010:0/16] 2024-01-28 19:59:14,672 (trainer:737) INFO: 13epoch:train:8401-8500batch: iter_time=9.734e-05, forward_time=0.343, loss_ctc=56.655, loss_att=48.997, acc=0.717, loss=51.294, backward_time=0.437, grad_norm=33.028, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.097, optim0_lr0=2.821e-04, train_time=1.663
+[gpub010:0/16] 2024-01-28 20:01:41,223 (trainer:737) INFO: 13epoch:train:8501-8600batch: iter_time=8.348e-05, forward_time=0.291, loss_ctc=58.178, loss_att=47.280, acc=0.734, loss=50.549, backward_time=0.403, grad_norm=33.983, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.821e-04, train_time=1.465
+[gpub010:0/16] 2024-01-28 20:04:13,744 (trainer:737) INFO: 13epoch:train:8601-8700batch: iter_time=8.084e-05, forward_time=0.291, loss_ctc=59.326, loss_att=54.920, acc=0.736, loss=56.242, backward_time=0.404, grad_norm=33.104, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.820e-04, train_time=1.525
+[gpub010:0/16] 2024-01-28 20:05:44,451 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-01-28 20:06:03,300 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 20:06:07,148 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb13fb1fe80>)
+[gpub010:0/16] 2024-01-28 20:06:07,148 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-01-28 20:06:07,151 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 20:12:24,469 (trainer:737) INFO: 13epoch:train:8701-8800batch: iter_time=3.440, forward_time=0.364, loss_ctc=64.510, loss_att=59.058, acc=0.696, loss=60.693, backward_time=0.415, grad_norm=38.870, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.093, optim0_lr0=2.819e-04, train_time=4.907
+[gpub010:0/16] 2024-01-28 20:14:53,350 (trainer:737) INFO: 13epoch:train:8801-8900batch: iter_time=8.183e-05, forward_time=0.287, loss_ctc=49.298, loss_att=55.333, acc=0.697, loss=53.522, backward_time=0.399, grad_norm=32.629, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.818e-04, train_time=1.489
+[gpub010:0/16] 2024-01-28 20:17:06,602 (trainer:737) INFO: 13epoch:train:8901-9000batch: iter_time=7.615e-05, forward_time=0.288, loss_ctc=49.567, loss_att=49.313, acc=0.724, loss=49.389, backward_time=0.402, grad_norm=28.814, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.818e-04, train_time=1.332
+[gpub010:0/16] 2024-01-28 20:19:58,630 (trainer:737) INFO: 13epoch:train:9001-9100batch: iter_time=8.349e-05, forward_time=0.336, loss_ctc=74.111, loss_att=64.901, acc=0.721, loss=67.664, backward_time=0.467, grad_norm=41.567, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.098, optim0_lr0=2.817e-04, train_time=1.720
+[gpub010:0/16] 2024-01-28 20:22:10,286 (trainer:737) INFO: 13epoch:train:9101-9200batch: iter_time=7.856e-05, forward_time=0.292, loss_ctc=61.945, loss_att=57.953, acc=0.694, loss=59.151, backward_time=0.404, grad_norm=35.395, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.816e-04, train_time=1.316
+[gpub010:0/16] 2024-01-28 20:24:45,565 (trainer:737) INFO: 13epoch:train:9201-9300batch: iter_time=7.968e-05, forward_time=0.288, loss_ctc=53.822, loss_att=43.706, acc=0.732, loss=46.741, backward_time=0.400, grad_norm=32.851, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.815e-04, train_time=1.553
+[gpub010:0/16] 2024-01-28 20:27:34,441 (trainer:737) INFO: 13epoch:train:9301-9400batch: iter_time=8.252e-05, forward_time=0.286, loss_ctc=47.383, loss_att=40.069, acc=0.736, loss=42.263, backward_time=0.396, grad_norm=27.698, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.815e-04, train_time=1.689
+[gpub010:0/16] 2024-01-28 20:30:07,537 (trainer:737) INFO: 13epoch:train:9401-9500batch: iter_time=8.056e-05, forward_time=0.392, loss_ctc=58.017, loss_att=59.409, acc=0.723, loss=58.992, backward_time=0.427, grad_norm=29.873, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.104, optim0_lr0=2.814e-04, train_time=1.530
+[gpub010:0/16] 2024-01-28 20:32:39,637 (trainer:737) INFO: 13epoch:train:9501-9600batch: iter_time=8.092e-05, forward_time=0.287, loss_ctc=43.701, loss_att=48.446, acc=0.735, loss=47.022, backward_time=0.398, grad_norm=25.348, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.813e-04, train_time=1.521
+[gpub010:0/16] 2024-01-28 20:35:02,834 (trainer:737) INFO: 13epoch:train:9601-9700batch: iter_time=8.068e-05, forward_time=0.290, loss_ctc=62.641, loss_att=49.762, acc=0.737, loss=53.626, backward_time=0.402, grad_norm=33.173, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.812e-04, train_time=1.432
+[gpub010:0/16] 2024-01-28 20:37:37,387 (trainer:737) INFO: 13epoch:train:9701-9800batch: iter_time=7.859e-05, forward_time=0.288, loss_ctc=54.229, loss_att=48.806, acc=0.711, loss=50.433, backward_time=0.400, grad_norm=32.751, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.812e-04, train_time=1.545
+[gpub010:0/16] 2024-01-28 20:40:31,989 (trainer:737) INFO: 13epoch:train:9801-9900batch: iter_time=8.788e-05, forward_time=0.392, loss_ctc=54.367, loss_att=46.064, acc=0.751, loss=48.555, backward_time=0.442, grad_norm=28.987, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.101, optim0_lr0=2.811e-04, train_time=1.746
+[gpub010:0/16] 2024-01-28 20:43:09,116 (trainer:737) INFO: 13epoch:train:9901-10000batch: iter_time=7.891e-05, forward_time=0.311, loss_ctc=64.899, loss_att=63.057, acc=0.711, loss=63.609, backward_time=0.403, grad_norm=37.033, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.091, optim0_lr0=2.810e-04, train_time=1.571
+[gpub010:0/16] 2024-01-28 20:43:29,196 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-01-28 20:43:47,694 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 20:43:51,252 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa78b131cf0>)
+[gpub010:0/16] 2024-01-28 20:43:51,252 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-01-28 20:43:51,318 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 20:50:57,584 (trainer:737) INFO: 13epoch:train:10001-10100batch: iter_time=3.201, forward_time=0.418, loss_ctc=59.710, loss_att=56.312, acc=0.701, loss=57.331, backward_time=0.421, grad_norm=36.433, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.809e-04, train_time=4.684
+[gpub010:0/16] 2024-01-28 20:53:36,773 (trainer:737) INFO: 13epoch:train:10101-10200batch: iter_time=8.088e-05, forward_time=0.288, loss_ctc=47.401, loss_att=52.140, acc=0.706, loss=50.718, backward_time=0.400, grad_norm=30.501, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.809e-04, train_time=1.591
+[gpub010:0/16] 2024-01-28 20:56:08,764 (trainer:737) INFO: 13epoch:train:10201-10300batch: iter_time=8.133e-05, forward_time=0.402, loss_ctc=63.905, loss_att=56.180, acc=0.717, loss=58.498, backward_time=0.427, grad_norm=35.701, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=2.808e-04, train_time=1.520
+[gpub010:0/16] 2024-01-28 20:58:43,140 (trainer:737) INFO: 13epoch:train:10301-10400batch: iter_time=9.005e-05, forward_time=0.293, loss_ctc=62.421, loss_att=56.155, acc=0.727, loss=58.035, backward_time=0.406, grad_norm=35.034, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.807e-04, train_time=1.544
+[gpub010:0/16] 2024-01-28 21:01:12,040 (trainer:737) INFO: 13epoch:train:10401-10500batch: iter_time=8.408e-05, forward_time=0.289, loss_ctc=62.734, loss_att=53.909, acc=0.701, loss=56.557, backward_time=0.400, grad_norm=38.108, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.806e-04, train_time=1.489
+[gpub010:0/16] 2024-01-28 21:03:37,961 (trainer:737) INFO: 13epoch:train:10501-10600batch: iter_time=8.048e-05, forward_time=0.345, loss_ctc=43.532, loss_att=38.337, acc=0.733, loss=39.895, backward_time=0.441, grad_norm=28.004, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=2.806e-04, train_time=1.459
+[gpub010:0/16] 2024-01-28 21:06:18,705 (trainer:737) INFO: 13epoch:train:10601-10700batch: iter_time=8.538e-05, forward_time=0.305, loss_ctc=57.400, loss_att=54.596, acc=0.717, loss=55.437, backward_time=0.410, grad_norm=31.254, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.805e-04, train_time=1.607
+[gpub010:0/16] 2024-01-28 21:08:30,322 (trainer:737) INFO: 13epoch:train:10701-10800batch: iter_time=9.212e-05, forward_time=0.289, loss_ctc=48.736, loss_att=47.181, acc=0.752, loss=47.648, backward_time=0.403, grad_norm=27.343, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.804e-04, train_time=1.316
+[gpub010:0/16] 2024-01-28 21:11:14,561 (trainer:737) INFO: 13epoch:train:10801-10900batch: iter_time=8.586e-05, forward_time=0.290, loss_ctc=56.227, loss_att=54.946, acc=0.730, loss=55.331, backward_time=0.403, grad_norm=30.564, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.803e-04, train_time=1.642
+[gpub010:0/16] 2024-01-28 21:13:46,652 (trainer:737) INFO: 13epoch:train:10901-11000batch: iter_time=7.681e-04, forward_time=0.397, loss_ctc=55.368, loss_att=48.128, acc=0.720, loss=50.300, backward_time=0.468, grad_norm=31.629, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.803e-04, train_time=1.521
+[gpub010:0/16] 2024-01-28 21:16:28,624 (trainer:737) INFO: 13epoch:train:11001-11100batch: iter_time=8.654e-05, forward_time=0.309, loss_ctc=56.699, loss_att=46.438, acc=0.737, loss=49.516, backward_time=0.403, grad_norm=31.357, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.802e-04, train_time=1.620
+[gpub010:0/16] 2024-01-28 21:18:55,852 (trainer:737) INFO: 13epoch:train:11101-11200batch: iter_time=8.510e-05, forward_time=0.292, loss_ctc=58.723, loss_att=54.597, acc=0.737, loss=55.835, backward_time=0.404, grad_norm=31.190, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.801e-04, train_time=1.472
+[gpub010:0/16] 2024-01-28 21:20:30,703 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-01-28 21:20:49,342 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 21:20:52,908 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa720e2f130>)
+[gpub010:0/16] 2024-01-28 21:20:52,908 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-01-28 21:20:52,912 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 21:27:22,312 (trainer:737) INFO: 13epoch:train:11201-11300batch: iter_time=3.499, forward_time=0.412, loss_ctc=64.301, loss_att=60.724, acc=0.697, loss=61.797, backward_time=0.428, grad_norm=41.112, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.801e-04, train_time=5.064
+[gpub010:0/16] 2024-01-28 21:29:43,493 (trainer:737) INFO: 13epoch:train:11301-11400batch: iter_time=8.157e-05, forward_time=0.288, loss_ctc=48.620, loss_att=56.351, acc=0.704, loss=54.032, backward_time=0.400, grad_norm=33.973, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.090, optim0_lr0=2.800e-04, train_time=1.412
+[gpub010:0/16] 2024-01-28 21:32:14,036 (trainer:737) INFO: 13epoch:train:11401-11500batch: iter_time=8.073e-05, forward_time=0.289, loss_ctc=49.510, loss_att=51.508, acc=0.721, loss=50.908, backward_time=0.400, grad_norm=28.792, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.090, optim0_lr0=2.799e-04, train_time=1.505
+[gpub010:0/16] 2024-01-28 21:35:07,326 (trainer:737) INFO: 13epoch:train:11501-11600batch: iter_time=8.277e-05, forward_time=0.376, loss_ctc=73.333, loss_att=65.240, acc=0.729, loss=67.668, backward_time=0.452, grad_norm=41.372, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.798e-04, train_time=1.733
+[gpub010:0/16] 2024-01-28 21:37:46,412 (trainer:737) INFO: 13epoch:train:11601-11700batch: iter_time=8.148e-05, forward_time=0.293, loss_ctc=61.847, loss_att=60.008, acc=0.703, loss=60.560, backward_time=0.404, grad_norm=33.807, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.798e-04, train_time=1.591
+[gpub010:0/16] 2024-01-28 21:40:20,496 (trainer:737) INFO: 13epoch:train:11701-11800batch: iter_time=8.618e-05, forward_time=0.383, loss_ctc=52.894, loss_att=44.639, acc=0.739, loss=47.115, backward_time=0.459, grad_norm=31.741, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.797e-04, train_time=1.540
+[gpub010:0/16] 2024-01-28 21:42:45,397 (trainer:737) INFO: 13epoch:train:11801-11900batch: iter_time=8.312e-05, forward_time=0.286, loss_ctc=47.160, loss_att=40.264, acc=0.742, loss=42.333, backward_time=0.397, grad_norm=28.750, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.796e-04, train_time=1.449
+[gpub010:0/16] 2024-01-28 21:45:36,101 (trainer:737) INFO: 13epoch:train:11901-12000batch: iter_time=8.175e-05, forward_time=0.294, loss_ctc=57.571, loss_att=59.916, acc=0.741, loss=59.212, backward_time=0.406, grad_norm=29.254, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=2.795e-04, train_time=1.707
+[gpub010:0/16] 2024-01-28 21:48:10,470 (trainer:737) INFO: 13epoch:train:12001-12100batch: iter_time=0.003, forward_time=0.393, loss_ctc=44.022, loss_att=49.269, acc=0.742, loss=47.695, backward_time=0.446, grad_norm=25.166, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.795e-04, train_time=1.543
+[gpub010:0/16] 2024-01-28 21:50:52,795 (trainer:737) INFO: 13epoch:train:12101-12200batch: iter_time=8.311e-05, forward_time=0.290, loss_ctc=61.974, loss_att=49.896, acc=0.744, loss=53.519, backward_time=0.402, grad_norm=32.852, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.794e-04, train_time=1.623
+[gpub010:0/16] 2024-01-28 21:53:33,545 (trainer:737) INFO: 13epoch:train:12201-12300batch: iter_time=8.850e-05, forward_time=0.290, loss_ctc=53.569, loss_att=49.727, acc=0.718, loss=50.879, backward_time=0.400, grad_norm=32.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.793e-04, train_time=1.608
+[gpub010:0/16] 2024-01-28 21:56:16,411 (trainer:737) INFO: 13epoch:train:12301-12400batch: iter_time=3.606e-04, forward_time=0.418, loss_ctc=54.320, loss_att=46.667, acc=0.751, loss=48.963, backward_time=0.465, grad_norm=28.693, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.793e-04, train_time=1.628
+[gpub010:0/16] 2024-01-28 21:58:43,158 (trainer:737) INFO: 13epoch:train:12401-12500batch: iter_time=8.052e-05, forward_time=0.292, loss_ctc=64.562, loss_att=63.567, acc=0.718, loss=63.866, backward_time=0.404, grad_norm=36.921, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.792e-04, train_time=1.467
+[gpub010:0/16] 2024-01-28 21:59:03,432 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-01-28 21:59:22,099 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 21:59:25,585 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa77937b400>)
+[gpub010:0/16] 2024-01-28 21:59:25,585 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-01-28 21:59:25,588 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 22:06:52,977 (trainer:737) INFO: 13epoch:train:12501-12600batch: iter_time=3.491, forward_time=0.349, loss_ctc=59.353, loss_att=58.168, acc=0.697, loss=58.523, backward_time=0.408, grad_norm=36.698, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.791e-04, train_time=4.898
+[gpub010:0/16] 2024-01-28 22:09:13,666 (trainer:737) INFO: 13epoch:train:12601-12700batch: iter_time=7.677e-05, forward_time=0.288, loss_ctc=46.722, loss_att=52.096, acc=0.709, loss=50.484, backward_time=0.401, grad_norm=30.559, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.790e-04, train_time=1.407
+[gpub010:0/16] 2024-01-28 22:12:11,885 (trainer:737) INFO: 13epoch:train:12701-12800batch: iter_time=7.872e-05, forward_time=0.396, loss_ctc=63.273, loss_att=56.541, acc=0.717, loss=58.561, backward_time=0.457, grad_norm=35.627, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.790e-04, train_time=1.782
+[gpub010:0/16] 2024-01-28 22:14:48,121 (trainer:737) INFO: 13epoch:train:12801-12900batch: iter_time=7.845e-05, forward_time=0.294, loss_ctc=61.934, loss_att=55.813, acc=0.729, loss=57.649, backward_time=0.406, grad_norm=34.446, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.789e-04, train_time=1.563
+[gpub010:0/16] 2024-01-28 22:17:27,558 (trainer:737) INFO: 13epoch:train:12901-13000batch: iter_time=7.684e-05, forward_time=0.289, loss_ctc=61.774, loss_att=54.269, acc=0.700, loss=56.520, backward_time=0.404, grad_norm=37.276, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.090, optim0_lr0=2.788e-04, train_time=1.594
+[gpub010:0/16] 2024-01-28 22:20:00,782 (trainer:737) INFO: 13epoch:train:13001-13100batch: iter_time=3.051e-04, forward_time=0.390, loss_ctc=43.209, loss_att=38.676, acc=0.733, loss=40.036, backward_time=0.450, grad_norm=27.685, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.787e-04, train_time=1.532
+[gpub010:0/16] 2024-01-28 22:22:36,454 (trainer:737) INFO: 13epoch:train:13101-13200batch: iter_time=7.705e-05, forward_time=0.291, loss_ctc=57.853, loss_att=54.844, acc=0.718, loss=55.746, backward_time=0.401, grad_norm=31.827, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.787e-04, train_time=1.557
+[gpub010:0/16] 2024-01-28 22:24:59,690 (trainer:737) INFO: 13epoch:train:13201-13300batch: iter_time=7.249e-05, forward_time=0.292, loss_ctc=48.457, loss_att=47.336, acc=0.753, loss=47.672, backward_time=0.400, grad_norm=25.895, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.786e-04, train_time=1.432
+[gpub010:0/16] 2024-01-28 22:27:55,888 (trainer:737) INFO: 13epoch:train:13301-13400batch: iter_time=9.152e-04, forward_time=0.366, loss_ctc=55.520, loss_att=54.889, acc=0.731, loss=55.078, backward_time=0.478, grad_norm=30.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.785e-04, train_time=1.762
+[gpub010:0/16] 2024-01-28 22:30:11,507 (trainer:737) INFO: 13epoch:train:13401-13500batch: iter_time=8.273e-05, forward_time=0.289, loss_ctc=56.365, loss_att=48.875, acc=0.719, loss=51.122, backward_time=0.400, grad_norm=33.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.785e-04, train_time=1.356
+[gpub010:0/16] 2024-01-28 22:33:03,622 (trainer:737) INFO: 13epoch:train:13501-13600batch: iter_time=3.278e-04, forward_time=0.363, loss_ctc=56.265, loss_att=46.559, acc=0.735, loss=49.471, backward_time=0.449, grad_norm=32.138, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.784e-04, train_time=1.720
+[gpub010:0/16] 2024-01-28 22:35:39,518 (trainer:737) INFO: 13epoch:train:13601-13700batch: iter_time=8.175e-05, forward_time=0.291, loss_ctc=58.174, loss_att=54.491, acc=0.738, loss=55.596, backward_time=0.402, grad_norm=32.577, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.783e-04, train_time=1.560
+[gpub010:0/16] 2024-01-28 22:37:09,350 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-01-28 22:37:28,222 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 22:37:31,833 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa71a9237c0>)
+[gpub010:0/16] 2024-01-28 22:37:31,833 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-01-28 22:37:31,836 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-28 22:43:41,307 (trainer:737) INFO: 13epoch:train:13701-13800batch: iter_time=3.369, forward_time=0.351, loss_ctc=63.023, loss_att=59.748, acc=0.698, loss=60.730, backward_time=0.413, grad_norm=36.998, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.782e-04, train_time=4.818
+[gpub010:0/16] 2024-01-28 22:45:57,789 (trainer:737) INFO: 13epoch:train:13801-13900batch: iter_time=8.122e-05, forward_time=0.289, loss_ctc=48.553, loss_att=54.708, acc=0.710, loss=52.861, backward_time=0.401, grad_norm=31.054, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.782e-04, train_time=1.364
+[gpub010:0/16] 2024-01-28 22:48:44,903 (trainer:737) INFO: 13epoch:train:13901-14000batch: iter_time=7.896e-05, forward_time=0.289, loss_ctc=49.170, loss_att=50.544, acc=0.723, loss=50.132, backward_time=0.402, grad_norm=28.003, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.781e-04, train_time=1.671
+[gpub010:0/16] 2024-01-28 22:51:21,783 (trainer:737) INFO: 13epoch:train:14001-14100batch: iter_time=1.832e-04, forward_time=0.365, loss_ctc=73.098, loss_att=64.773, acc=0.731, loss=67.270, backward_time=0.463, grad_norm=41.506, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.780e-04, train_time=1.569
+[gpub010:0/16] 2024-01-28 22:54:15,818 (trainer:737) INFO: 13epoch:train:14101-14200batch: iter_time=8.080e-05, forward_time=0.293, loss_ctc=61.247, loss_att=60.498, acc=0.701, loss=60.722, backward_time=0.407, grad_norm=34.434, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=2.780e-04, train_time=1.739
+[gpub010:0/16] 2024-01-28 22:56:49,375 (trainer:737) INFO: 13epoch:train:14201-14300batch: iter_time=8.535e-05, forward_time=0.384, loss_ctc=52.107, loss_att=44.315, acc=0.740, loss=46.652, backward_time=0.440, grad_norm=32.212, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.779e-04, train_time=1.536
+[gpub010:0/16] 2024-01-28 22:57:47,903 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-28 22:59:19,584 (trainer:737) INFO: 13epoch:train:14301-14400batch: iter_time=8.190e-05, forward_time=0.289, loss_ctc=46.658, loss_att=39.692, acc=0.745, loss=41.782, backward_time=0.397, grad_norm=27.066, clip=100.000, loss_scale=7.343e+33, optim_step_time=0.091, optim0_lr0=2.778e-04, train_time=1.503
+[gpub010:0/16] 2024-01-28 23:01:58,990 (trainer:737) INFO: 13epoch:train:14401-14500batch: iter_time=8.272e-05, forward_time=0.294, loss_ctc=58.129, loss_att=60.026, acc=0.742, loss=59.457, backward_time=0.405, grad_norm=30.574, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.777e-04, train_time=1.593
+[gpub010:0/16] 2024-01-28 23:04:25,455 (trainer:737) INFO: 13epoch:train:14501-14600batch: iter_time=2.568e-04, forward_time=0.391, loss_ctc=43.535, loss_att=48.917, acc=0.745, loss=47.302, backward_time=0.432, grad_norm=26.297, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.777e-04, train_time=1.464
+[gpub010:0/16] 2024-01-28 23:06:57,007 (trainer:737) INFO: 13epoch:train:14601-14700batch: iter_time=8.356e-05, forward_time=0.290, loss_ctc=61.690, loss_att=49.399, acc=0.747, loss=53.087, backward_time=0.401, grad_norm=32.139, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.776e-04, train_time=1.516
+[gpub010:0/16] 2024-01-28 23:09:46,784 (trainer:737) INFO: 13epoch:train:14701-14800batch: iter_time=7.049e-04, forward_time=0.405, loss_ctc=53.258, loss_att=49.566, acc=0.718, loss=50.673, backward_time=0.438, grad_norm=32.056, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.775e-04, train_time=1.698
+[gpub010:0/16] 2024-01-28 23:12:13,087 (trainer:737) INFO: 13epoch:train:14801-14900batch: iter_time=8.989e-05, forward_time=0.303, loss_ctc=53.643, loss_att=45.338, acc=0.757, loss=47.830, backward_time=0.405, grad_norm=28.302, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.775e-04, train_time=1.463
+[gpub010:0/16] 2024-01-28 23:14:48,634 (trainer:737) INFO: 13epoch:train:14901-15000batch: iter_time=8.733e-05, forward_time=0.294, loss_ctc=63.943, loss_att=63.077, acc=0.721, loss=63.336, backward_time=0.405, grad_norm=36.541, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.774e-04, train_time=1.555
+[gpub010:0/16] 2024-01-28 23:54:33,877 (trainer:343) INFO: 13epoch results: [train] iter_time=0.276, forward_time=0.336, loss_ctc=57.285, loss_att=53.229, acc=0.722, loss=54.446, backward_time=0.422, grad_norm=32.776, clip=100.000, loss_scale=1.838e+33, optim_step_time=0.094, optim0_lr0=2.829e-04, train_time=1.954, time=8 hours, 9 minutes and 4 seconds, total_count=225000, gpu_max_cached_mem_GB=42.072, [valid] loss_ctc=45.272, cer_ctc=0.233, loss_att=44.548, acc=0.636, cer=0.355, wer=0.996, loss=44.765, time=39 minutes and 20.93 seconds, total_count=70065, gpu_max_cached_mem_GB=42.072
+[gpub010:0/16] 2024-01-28 23:54:51,648 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub010:0/16] 2024-01-28 23:54:51,682 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/8epoch.pth
+[gpub010:0/16] 2024-01-28 23:54:51,683 (trainer:272) INFO: 14/45epoch started. Estimated time to finish: 1 week, 4 days and 17 hours
+[gpub010:0/16] 2024-01-28 23:54:51,693 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-28 23:55:09,443 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-28 23:55:12,811 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa7306e8880>)
+[gpub010:0/16] 2024-01-28 23:55:12,811 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-01-28 23:55:12,815 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 00:02:32,552 (trainer:737) INFO: 14epoch:train:1-100batch: iter_time=3.164, forward_time=0.354, loss_ctc=60.375, loss_att=66.442, acc=0.706, loss=64.622, backward_time=0.413, grad_norm=35.718, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.773e-04, train_time=4.608
+[gpub010:0/16] 2024-01-29 00:05:10,422 (trainer:737) INFO: 14epoch:train:101-200batch: iter_time=8.944e-05, forward_time=0.292, loss_ctc=63.167, loss_att=57.009, acc=0.700, loss=58.856, backward_time=0.406, grad_norm=35.593, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.772e-04, train_time=1.578
+[gpub010:0/16] 2024-01-29 00:07:59,768 (trainer:737) INFO: 14epoch:train:201-300batch: iter_time=2.635e-04, forward_time=0.377, loss_ctc=52.959, loss_att=56.826, acc=0.704, loss=55.666, backward_time=0.435, grad_norm=32.072, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.772e-04, train_time=1.693
+[gpub010:0/16] 2024-01-29 00:10:34,724 (trainer:737) INFO: 14epoch:train:301-400batch: iter_time=2.620e-04, forward_time=0.294, loss_ctc=51.229, loss_att=46.100, acc=0.729, loss=47.639, backward_time=0.401, grad_norm=30.270, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.771e-04, train_time=1.550
+[gpub010:0/16] 2024-01-29 00:13:04,867 (trainer:737) INFO: 14epoch:train:401-500batch: iter_time=8.509e-05, forward_time=0.349, loss_ctc=57.432, loss_att=44.830, acc=0.722, loss=48.611, backward_time=0.423, grad_norm=34.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.770e-04, train_time=1.501
+[gpub010:0/16] 2024-01-29 00:15:53,365 (trainer:737) INFO: 14epoch:train:501-600batch: iter_time=8.824e-05, forward_time=0.324, loss_ctc=56.293, loss_att=52.099, acc=0.711, loss=53.357, backward_time=0.413, grad_norm=35.006, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.770e-04, train_time=1.685
+[gpub010:0/16] 2024-01-29 00:18:12,725 (trainer:737) INFO: 14epoch:train:601-700batch: iter_time=2.599e-04, forward_time=0.293, loss_ctc=52.805, loss_att=54.229, acc=0.704, loss=53.802, backward_time=0.406, grad_norm=31.371, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.769e-04, train_time=1.394
+[gpub010:0/16] 2024-01-29 00:21:04,683 (trainer:737) INFO: 14epoch:train:701-800batch: iter_time=3.561e-04, forward_time=0.389, loss_ctc=53.525, loss_att=47.530, acc=0.727, loss=49.329, backward_time=0.425, grad_norm=33.649, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.768e-04, train_time=1.719
+[gpub010:0/16] 2024-01-29 00:23:42,638 (trainer:737) INFO: 14epoch:train:801-900batch: iter_time=8.118e-05, forward_time=0.297, loss_ctc=51.713, loss_att=46.619, acc=0.736, loss=48.147, backward_time=0.406, grad_norm=31.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.767e-04, train_time=1.579
+[gpub010:0/16] 2024-01-29 00:26:16,653 (trainer:737) INFO: 14epoch:train:901-1000batch: iter_time=4.201e-04, forward_time=0.389, loss_ctc=62.399, loss_att=48.445, acc=0.747, loss=52.631, backward_time=0.447, grad_norm=41.711, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.767e-04, train_time=1.540
+[gpub010:0/16] 2024-01-29 00:29:00,415 (trainer:737) INFO: 14epoch:train:1001-1100batch: iter_time=9.015e-05, forward_time=0.289, loss_ctc=52.943, loss_att=46.822, acc=0.728, loss=48.658, backward_time=0.404, grad_norm=29.091, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.766e-04, train_time=1.637
+[gpub010:0/16] 2024-01-29 00:31:25,868 (trainer:737) INFO: 14epoch:train:1101-1200batch: iter_time=9.333e-05, forward_time=0.342, loss_ctc=55.514, loss_att=53.064, acc=0.711, loss=53.799, backward_time=0.413, grad_norm=33.431, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.765e-04, train_time=1.454
+[gpub010:0/16] 2024-01-29 00:33:14,917 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-01-29 00:33:33,609 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 00:33:37,105 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa720e9f8e0>)
+[gpub010:0/16] 2024-01-29 00:33:37,105 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-01-29 00:33:37,109 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 00:39:48,645 (trainer:737) INFO: 14epoch:train:1201-1300batch: iter_time=3.271, forward_time=0.349, loss_ctc=51.993, loss_att=57.205, acc=0.713, loss=55.641, backward_time=0.420, grad_norm=29.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.765e-04, train_time=5.028
+[gpub010:0/16] 2024-01-29 00:42:20,978 (trainer:737) INFO: 14epoch:train:1301-1400batch: iter_time=8.350e-05, forward_time=0.299, loss_ctc=62.745, loss_att=61.150, acc=0.707, loss=61.629, backward_time=0.408, grad_norm=36.202, clip=100.000, loss_scale=8.204e+33, optim_step_time=0.093, optim0_lr0=2.764e-04, train_time=1.523
+[gpub010:0/16] 2024-01-29 00:44:58,145 (trainer:737) INFO: 14epoch:train:1401-1500batch: iter_time=7.927e-05, forward_time=0.386, loss_ctc=55.769, loss_att=55.534, acc=0.698, loss=55.605, backward_time=0.443, grad_norm=31.636, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.763e-04, train_time=1.571
+[gpub010:0/16] 2024-01-29 00:47:29,097 (trainer:737) INFO: 14epoch:train:1501-1600batch: iter_time=8.031e-05, forward_time=0.293, loss_ctc=53.039, loss_att=52.522, acc=0.732, loss=52.677, backward_time=0.404, grad_norm=30.209, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.763e-04, train_time=1.509
+[gpub010:0/16] 2024-01-29 00:49:59,697 (trainer:737) INFO: 14epoch:train:1601-1700batch: iter_time=3.671e-04, forward_time=0.307, loss_ctc=50.944, loss_att=41.130, acc=0.742, loss=44.074, backward_time=0.402, grad_norm=30.556, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.762e-04, train_time=1.506
+[gpub010:0/16] 2024-01-29 00:52:28,052 (trainer:737) INFO: 14epoch:train:1701-1800batch: iter_time=8.602e-05, forward_time=0.389, loss_ctc=58.038, loss_att=48.138, acc=0.707, loss=51.108, backward_time=0.444, grad_norm=34.777, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.761e-04, train_time=1.482
+[gpub010:0/16] 2024-01-29 00:54:59,634 (trainer:737) INFO: 14epoch:train:1801-1900batch: iter_time=8.156e-05, forward_time=0.292, loss_ctc=56.117, loss_att=59.793, acc=0.703, loss=58.690, backward_time=0.403, grad_norm=35.480, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.760e-04, train_time=1.516
+[gpub010:0/16] 2024-01-29 00:57:34,292 (trainer:737) INFO: 14epoch:train:1901-2000batch: iter_time=9.819e-04, forward_time=0.293, loss_ctc=51.478, loss_att=48.121, acc=0.713, loss=49.128, backward_time=0.403, grad_norm=30.965, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.760e-04, train_time=1.546
+[gpub010:0/16] 2024-01-29 01:00:07,922 (trainer:737) INFO: 14epoch:train:2001-2100batch: iter_time=8.073e-05, forward_time=0.403, loss_ctc=50.377, loss_att=42.027, acc=0.750, loss=44.532, backward_time=0.442, grad_norm=27.837, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.759e-04, train_time=1.537
+[gpub010:0/16] 2024-01-29 01:02:38,283 (trainer:737) INFO: 14epoch:train:2101-2200batch: iter_time=7.962e-05, forward_time=0.291, loss_ctc=51.859, loss_att=47.262, acc=0.746, loss=48.641, backward_time=0.402, grad_norm=28.202, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.758e-04, train_time=1.502
+[gpub010:0/16] 2024-01-29 01:05:13,895 (trainer:737) INFO: 14epoch:train:2201-2300batch: iter_time=8.148e-05, forward_time=0.293, loss_ctc=53.411, loss_att=45.546, acc=0.730, loss=47.905, backward_time=0.408, grad_norm=39.066, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.758e-04, train_time=1.557
+[gpub010:0/16] 2024-01-29 01:09:44,919 (trainer:737) INFO: 14epoch:train:2301-2400batch: iter_time=8.379e-05, forward_time=0.435, loss_ctc=58.526, loss_att=47.718, acc=0.726, loss=50.961, backward_time=0.448, grad_norm=31.113, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.757e-04, train_time=2.709
+[gpub010:0/16] 2024-01-29 01:11:58,222 (trainer:737) INFO: 14epoch:train:2401-2500batch: iter_time=7.886e-05, forward_time=0.290, loss_ctc=51.047, loss_att=53.487, acc=0.718, loss=52.755, backward_time=0.402, grad_norm=32.065, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.756e-04, train_time=1.333
+[gpub010:0/16] 2024-01-29 01:12:18,349 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-01-29 01:12:36,890 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 01:12:40,423 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa743ea8940>)
+[gpub010:0/16] 2024-01-29 01:12:40,423 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-01-29 01:12:40,426 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 01:24:19,096 (trainer:737) INFO: 14epoch:train:2501-2600batch: iter_time=3.329, forward_time=0.300, loss_ctc=58.132, loss_att=66.475, acc=0.721, loss=63.972, backward_time=0.406, grad_norm=32.428, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.756e-04, train_time=7.409
+[gpub010:0/16] 2024-01-29 01:27:08,652 (trainer:737) INFO: 14epoch:train:2601-2700batch: iter_time=8.856e-05, forward_time=0.390, loss_ctc=61.492, loss_att=58.414, acc=0.706, loss=59.337, backward_time=0.460, grad_norm=35.242, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.755e-04, train_time=1.694
+[gpub010:0/16] 2024-01-29 01:29:46,833 (trainer:737) INFO: 14epoch:train:2701-2800batch: iter_time=9.689e-05, forward_time=0.298, loss_ctc=51.432, loss_att=57.671, acc=0.710, loss=55.799, backward_time=0.404, grad_norm=29.704, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.754e-04, train_time=1.582
+[gpub010:0/16] 2024-01-29 01:32:24,177 (trainer:737) INFO: 14epoch:train:2801-2900batch: iter_time=9.819e-05, forward_time=0.290, loss_ctc=49.712, loss_att=46.630, acc=0.735, loss=47.555, backward_time=0.401, grad_norm=30.275, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.753e-04, train_time=1.574
+[gpub010:0/16] 2024-01-29 01:34:57,101 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 01:35:14,439 (trainer:737) INFO: 14epoch:train:2901-3000batch: iter_time=9.420e-05, forward_time=0.396, loss_ctc=54.209, loss_att=44.734, acc=0.732, loss=47.576, backward_time=0.459, grad_norm=33.024, clip=100.000, loss_scale=9.755e+33, optim_step_time=0.099, optim0_lr0=2.753e-04, train_time=1.702
+[gpub010:0/16] 2024-01-29 01:37:45,219 (trainer:737) INFO: 14epoch:train:3001-3100batch: iter_time=9.258e-05, forward_time=0.293, loss_ctc=54.370, loss_att=50.199, acc=0.728, loss=51.450, backward_time=0.404, grad_norm=30.884, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.752e-04, train_time=1.508
+[gpub010:0/16] 2024-01-29 01:40:26,782 (trainer:737) INFO: 14epoch:train:3101-3200batch: iter_time=9.808e-05, forward_time=0.291, loss_ctc=51.560, loss_att=54.905, acc=0.714, loss=53.901, backward_time=0.406, grad_norm=31.695, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.751e-04, train_time=1.615
+[gpub010:0/16] 2024-01-29 01:43:15,849 (trainer:737) INFO: 14epoch:train:3201-3300batch: iter_time=9.492e-05, forward_time=0.390, loss_ctc=52.299, loss_att=46.953, acc=0.737, loss=48.557, backward_time=0.441, grad_norm=32.154, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.751e-04, train_time=1.691
+[gpub010:0/16] 2024-01-29 01:45:49,335 (trainer:737) INFO: 14epoch:train:3301-3400batch: iter_time=9.692e-05, forward_time=0.289, loss_ctc=50.243, loss_att=45.001, acc=0.751, loss=46.574, backward_time=0.403, grad_norm=37.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.750e-04, train_time=1.530
+[gpub010:0/16] 2024-01-29 01:48:20,941 (trainer:737) INFO: 14epoch:train:3401-3500batch: iter_time=0.001, forward_time=0.299, loss_ctc=59.512, loss_att=49.147, acc=0.761, loss=52.257, backward_time=0.408, grad_norm=36.290, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.749e-04, train_time=1.520
+[gpub010:0/16] 2024-01-29 01:51:10,717 (trainer:737) INFO: 14epoch:train:3501-3600batch: iter_time=8.639e-05, forward_time=0.379, loss_ctc=52.151, loss_att=47.001, acc=0.734, loss=48.546, backward_time=0.460, grad_norm=29.607, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.749e-04, train_time=1.697
+[gpub010:0/16] 2024-01-29 01:53:44,491 (trainer:737) INFO: 14epoch:train:3601-3700batch: iter_time=1.622e-04, forward_time=0.290, loss_ctc=54.217, loss_att=52.350, acc=0.721, loss=52.910, backward_time=0.403, grad_norm=32.360, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.748e-04, train_time=1.536
+[gpub010:0/16] 2024-01-29 01:55:16,354 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-01-29 01:55:35,001 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 01:55:38,461 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb08c6edd80>)
+[gpub010:0/16] 2024-01-29 01:55:38,461 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-01-29 01:55:38,511 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 02:02:07,771 (trainer:737) INFO: 14epoch:train:3701-3800batch: iter_time=3.485, forward_time=0.380, loss_ctc=50.604, loss_att=55.955, acc=0.730, loss=54.350, backward_time=0.420, grad_norm=34.732, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.747e-04, train_time=5.034
+[gpub010:0/16] 2024-01-29 02:05:13,786 (trainer:737) INFO: 14epoch:train:3801-3900batch: iter_time=0.220, forward_time=0.292, loss_ctc=61.436, loss_att=63.348, acc=0.714, loss=62.775, backward_time=0.404, grad_norm=34.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.747e-04, train_time=1.860
+[gpub010:0/16] 2024-01-29 02:08:01,099 (trainer:737) INFO: 14epoch:train:3901-4000batch: iter_time=0.007, forward_time=0.299, loss_ctc=54.837, loss_att=55.021, acc=0.709, loss=54.966, backward_time=0.397, grad_norm=30.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.746e-04, train_time=1.673
+[gpub010:0/16] 2024-01-29 02:10:20,848 (trainer:737) INFO: 14epoch:train:4001-4100batch: iter_time=8.750e-05, forward_time=0.299, loss_ctc=52.102, loss_att=53.086, acc=0.738, loss=52.791, backward_time=0.408, grad_norm=30.320, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.745e-04, train_time=1.397
+[gpub010:0/16] 2024-01-29 02:13:02,111 (trainer:737) INFO: 14epoch:train:4101-4200batch: iter_time=8.850e-05, forward_time=0.391, loss_ctc=50.414, loss_att=40.694, acc=0.748, loss=43.610, backward_time=0.441, grad_norm=31.728, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.744e-04, train_time=1.613
+[gpub010:0/16] 2024-01-29 02:15:37,084 (trainer:737) INFO: 14epoch:train:4201-4300batch: iter_time=8.446e-05, forward_time=0.292, loss_ctc=56.705, loss_att=50.174, acc=0.711, loss=52.133, backward_time=0.407, grad_norm=32.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.744e-04, train_time=1.550
+[gpub010:0/16] 2024-01-29 02:18:13,908 (trainer:737) INFO: 14epoch:train:4301-4400batch: iter_time=8.864e-05, forward_time=0.342, loss_ctc=54.693, loss_att=59.062, acc=0.714, loss=57.751, backward_time=0.455, grad_norm=31.577, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.743e-04, train_time=1.568
+[gpub010:0/16] 2024-01-29 02:20:38,832 (trainer:737) INFO: 14epoch:train:4401-4500batch: iter_time=8.159e-05, forward_time=0.289, loss_ctc=50.358, loss_att=47.532, acc=0.728, loss=48.380, backward_time=0.401, grad_norm=29.415, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.742e-04, train_time=1.449
+[gpub010:0/16] 2024-01-29 02:22:58,764 (trainer:737) INFO: 14epoch:train:4501-4600batch: iter_time=8.804e-05, forward_time=0.302, loss_ctc=49.431, loss_att=41.026, acc=0.759, loss=43.547, backward_time=0.403, grad_norm=28.400, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.742e-04, train_time=1.399
+[gpub010:0/16] 2024-01-29 02:25:50,146 (trainer:737) INFO: 14epoch:train:4601-4700batch: iter_time=8.475e-05, forward_time=0.399, loss_ctc=50.714, loss_att=47.959, acc=0.761, loss=48.786, backward_time=0.426, grad_norm=27.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=2.741e-04, train_time=1.714
+[gpub010:0/16] 2024-01-29 02:28:10,347 (trainer:737) INFO: 14epoch:train:4701-4800batch: iter_time=8.647e-05, forward_time=0.288, loss_ctc=51.455, loss_att=44.774, acc=0.743, loss=46.778, backward_time=0.399, grad_norm=36.047, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.740e-04, train_time=1.402
+[gpub010:0/16] 2024-01-29 02:30:34,291 (trainer:737) INFO: 14epoch:train:4801-4900batch: iter_time=9.054e-05, forward_time=0.293, loss_ctc=57.430, loss_att=47.937, acc=0.732, loss=50.785, backward_time=0.403, grad_norm=30.321, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.740e-04, train_time=1.439
+[gpub010:0/16] 2024-01-29 02:33:13,689 (trainer:737) INFO: 14epoch:train:4901-5000batch: iter_time=2.963e-04, forward_time=0.366, loss_ctc=49.804, loss_att=52.605, acc=0.727, loss=51.764, backward_time=0.458, grad_norm=31.114, clip=100.000, loss_scale=5.815e+33, optim_step_time=0.100, optim0_lr0=2.739e-04, train_time=1.594
+[gpub010:0/16] 2024-01-29 02:33:33,901 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-01-29 02:33:52,658 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 02:33:56,542 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb5e54735b0>)
+[gpub010:0/16] 2024-01-29 02:33:56,543 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-01-29 02:33:56,546 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 02:41:35,278 (trainer:737) INFO: 14epoch:train:5001-5100batch: iter_time=3.226, forward_time=0.291, loss_ctc=56.976, loss_att=65.527, acc=0.713, loss=62.961, backward_time=0.403, grad_norm=32.257, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.738e-04, train_time=5.016
+[gpub010:0/16] 2024-01-29 02:44:10,375 (trainer:737) INFO: 14epoch:train:5101-5200batch: iter_time=8.478e-05, forward_time=0.290, loss_ctc=61.058, loss_att=56.214, acc=0.707, loss=57.668, backward_time=0.400, grad_norm=34.830, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.738e-04, train_time=1.551
+[gpub010:0/16] 2024-01-29 02:44:16,741 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 02:46:41,998 (trainer:737) INFO: 14epoch:train:5201-5300batch: iter_time=8.237e-05, forward_time=0.301, loss_ctc=51.174, loss_att=56.135, acc=0.712, loss=54.646, backward_time=0.403, grad_norm=31.315, clip=100.000, loss_scale=5.402e+33, optim_step_time=0.092, optim0_lr0=2.737e-04, train_time=1.516
+[gpub010:0/16] 2024-01-29 02:49:26,125 (trainer:737) INFO: 14epoch:train:5301-5400batch: iter_time=8.885e-05, forward_time=0.373, loss_ctc=49.163, loss_att=45.337, acc=0.735, loss=46.485, backward_time=0.435, grad_norm=30.797, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.736e-04, train_time=1.641
+[gpub010:0/16] 2024-01-29 02:51:52,234 (trainer:737) INFO: 14epoch:train:5401-5500batch: iter_time=8.460e-05, forward_time=0.290, loss_ctc=54.582, loss_att=43.036, acc=0.732, loss=46.500, backward_time=0.400, grad_norm=32.650, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.736e-04, train_time=1.461
+[gpub010:0/16] 2024-01-29 02:54:19,537 (trainer:737) INFO: 14epoch:train:5501-5600batch: iter_time=8.423e-05, forward_time=0.302, loss_ctc=54.110, loss_att=51.165, acc=0.717, loss=52.048, backward_time=0.404, grad_norm=32.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.735e-04, train_time=1.472
+[gpub010:0/16] 2024-01-29 02:57:05,115 (trainer:737) INFO: 14epoch:train:5601-5700batch: iter_time=8.376e-05, forward_time=0.365, loss_ctc=50.736, loss_att=54.102, acc=0.708, loss=53.092, backward_time=0.455, grad_norm=31.735, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.734e-04, train_time=1.657
+[gpub010:0/16] 2024-01-29 02:59:31,615 (trainer:737) INFO: 14epoch:train:5701-5800batch: iter_time=8.211e-05, forward_time=0.290, loss_ctc=51.793, loss_att=46.375, acc=0.734, loss=48.000, backward_time=0.400, grad_norm=32.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.734e-04, train_time=1.465
+[gpub010:0/16] 2024-01-29 03:01:54,214 (trainer:737) INFO: 14epoch:train:5801-5900batch: iter_time=8.328e-05, forward_time=0.297, loss_ctc=49.486, loss_att=45.509, acc=0.742, loss=46.702, backward_time=0.408, grad_norm=29.796, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.733e-04, train_time=1.426
+[gpub010:0/16] 2024-01-29 03:04:38,329 (trainer:737) INFO: 14epoch:train:5901-6000batch: iter_time=8.356e-05, forward_time=0.298, loss_ctc=58.054, loss_att=47.025, acc=0.753, loss=50.334, backward_time=0.406, grad_norm=36.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.732e-04, train_time=1.640
+[gpub010:0/16] 2024-01-29 03:07:24,134 (trainer:737) INFO: 14epoch:train:6001-6100batch: iter_time=8.330e-05, forward_time=0.429, loss_ctc=51.213, loss_att=45.700, acc=0.736, loss=47.354, backward_time=0.427, grad_norm=30.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.731e-04, train_time=1.656
+[gpub010:0/16] 2024-01-29 03:09:49,105 (trainer:737) INFO: 14epoch:train:6101-6200batch: iter_time=8.276e-05, forward_time=0.290, loss_ctc=53.670, loss_att=51.498, acc=0.719, loss=52.150, backward_time=0.401, grad_norm=32.921, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.731e-04, train_time=1.452
+[gpub010:0/16] 2024-01-29 03:11:17,692 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-01-29 03:11:36,903 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 03:11:40,799 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1310c7ac0>)
+[gpub010:0/16] 2024-01-29 03:11:40,799 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-01-29 03:11:40,802 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 03:17:35,583 (trainer:737) INFO: 14epoch:train:6201-6300batch: iter_time=3.152, forward_time=0.362, loss_ctc=50.609, loss_att=55.752, acc=0.719, loss=54.209, backward_time=0.416, grad_norm=29.938, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.730e-04, train_time=4.665
+[gpub010:0/16] 2024-01-29 03:20:18,792 (trainer:737) INFO: 14epoch:train:6301-6400batch: iter_time=8.459e-05, forward_time=0.291, loss_ctc=61.349, loss_att=60.273, acc=0.713, loss=60.596, backward_time=0.404, grad_norm=34.308, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.729e-04, train_time=1.631
+[gpub010:0/16] 2024-01-29 03:22:55,398 (trainer:737) INFO: 14epoch:train:6401-6500batch: iter_time=8.408e-05, forward_time=0.290, loss_ctc=54.918, loss_att=54.653, acc=0.702, loss=54.732, backward_time=0.401, grad_norm=32.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.729e-04, train_time=1.566
+[gpub010:0/16] 2024-01-29 03:25:43,272 (trainer:737) INFO: 14epoch:train:6501-6600batch: iter_time=8.251e-05, forward_time=0.362, loss_ctc=52.178, loss_att=51.992, acc=0.736, loss=52.048, backward_time=0.441, grad_norm=30.526, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.728e-04, train_time=1.679
+[gpub010:0/16] 2024-01-29 03:28:22,779 (trainer:737) INFO: 14epoch:train:6601-6700batch: iter_time=8.218e-05, forward_time=0.301, loss_ctc=49.645, loss_att=40.035, acc=0.747, loss=42.918, backward_time=0.397, grad_norm=29.395, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.727e-04, train_time=1.595
+[gpub010:0/16] 2024-01-29 03:30:40,710 (trainer:737) INFO: 14epoch:train:6701-6800batch: iter_time=8.294e-05, forward_time=0.289, loss_ctc=57.199, loss_att=47.921, acc=0.712, loss=50.704, backward_time=0.400, grad_norm=34.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.727e-04, train_time=1.379
+[gpub010:0/16] 2024-01-29 03:33:48,936 (trainer:737) INFO: 14epoch:train:6801-6900batch: iter_time=8.852e-05, forward_time=0.359, loss_ctc=54.174, loss_att=58.765, acc=0.707, loss=57.388, backward_time=0.444, grad_norm=33.473, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.726e-04, train_time=1.882
+[gpub010:0/16] 2024-01-29 03:36:26,766 (trainer:737) INFO: 14epoch:train:6901-7000batch: iter_time=9.035e-05, forward_time=0.288, loss_ctc=49.550, loss_att=47.652, acc=0.718, loss=48.222, backward_time=0.399, grad_norm=29.977, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.725e-04, train_time=1.579
+[gpub010:0/16] 2024-01-29 03:38:57,760 (trainer:737) INFO: 14epoch:train:7001-7100batch: iter_time=8.389e-05, forward_time=0.289, loss_ctc=49.224, loss_att=41.205, acc=0.756, loss=43.611, backward_time=0.401, grad_norm=27.497, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.725e-04, train_time=1.510
+[gpub010:0/16] 2024-01-29 03:41:25,294 (trainer:737) INFO: 14epoch:train:7101-7200batch: iter_time=2.210e-04, forward_time=0.315, loss_ctc=50.809, loss_att=46.858, acc=0.751, loss=48.043, backward_time=0.430, grad_norm=29.284, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.724e-04, train_time=1.475
+[gpub010:0/16] 2024-01-29 03:44:18,567 (trainer:737) INFO: 14epoch:train:7201-7300batch: iter_time=9.473e-05, forward_time=0.367, loss_ctc=51.305, loss_att=45.031, acc=0.736, loss=46.913, backward_time=0.407, grad_norm=37.910, clip=100.000, loss_scale=1.012e+34, optim_step_time=0.093, optim0_lr0=2.723e-04, train_time=1.733
+[gpub010:0/16] 2024-01-29 03:46:55,948 (trainer:737) INFO: 14epoch:train:7301-7400batch: iter_time=8.846e-05, forward_time=0.291, loss_ctc=57.499, loss_att=46.971, acc=0.730, loss=50.129, backward_time=0.402, grad_norm=30.140, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.723e-04, train_time=1.574
+[gpub010:0/16] 2024-01-29 03:49:18,220 (trainer:737) INFO: 14epoch:train:7401-7500batch: iter_time=7.783e-05, forward_time=0.310, loss_ctc=49.307, loss_att=52.661, acc=0.721, loss=51.655, backward_time=0.403, grad_norm=30.859, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.722e-04, train_time=1.423
+[gpub010:0/16] 2024-01-29 03:49:38,248 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-01-29 03:49:57,047 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 03:50:00,618 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae3ad4ba90>)
+[gpub010:0/16] 2024-01-29 03:50:00,618 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-01-29 03:50:00,621 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 03:57:43,776 (trainer:737) INFO: 14epoch:train:7501-7600batch: iter_time=3.388, forward_time=0.392, loss_ctc=57.024, loss_att=64.866, acc=0.729, loss=62.514, backward_time=0.422, grad_norm=32.744, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.721e-04, train_time=5.055
+[gpub010:0/16] 2024-01-29 03:59:55,998 (trainer:737) INFO: 14epoch:train:7601-7700batch: iter_time=8.340e-05, forward_time=0.291, loss_ctc=60.021, loss_att=56.757, acc=0.711, loss=57.736, backward_time=0.402, grad_norm=34.048, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.721e-04, train_time=1.322
+[gpub010:0/16] 2024-01-29 04:02:44,776 (trainer:737) INFO: 14epoch:train:7701-7800batch: iter_time=8.651e-05, forward_time=0.382, loss_ctc=50.247, loss_att=56.237, acc=0.716, loss=54.440, backward_time=0.422, grad_norm=30.162, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.119, optim0_lr0=2.720e-04, train_time=1.688
+[gpub010:0/16] 2024-01-29 04:04:13,073 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 04:05:14,030 (trainer:737) INFO: 14epoch:train:7801-7900batch: iter_time=8.810e-05, forward_time=0.289, loss_ctc=48.131, loss_att=45.312, acc=0.743, loss=46.158, backward_time=0.400, grad_norm=42.795, clip=100.000, loss_scale=8.182e+33, optim_step_time=0.092, optim0_lr0=2.719e-04, train_time=1.492
+[gpub010:0/16] 2024-01-29 04:09:11,237 (trainer:737) INFO: 14epoch:train:7901-8000batch: iter_time=8.956e-05, forward_time=0.380, loss_ctc=54.141, loss_att=44.251, acc=0.738, loss=47.218, backward_time=0.437, grad_norm=33.282, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=2.719e-04, train_time=2.372
+[gpub010:0/16] 2024-01-29 04:11:42,059 (trainer:737) INFO: 14epoch:train:8001-8100batch: iter_time=9.537e-05, forward_time=0.290, loss_ctc=53.895, loss_att=50.117, acc=0.730, loss=51.250, backward_time=0.404, grad_norm=30.645, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.718e-04, train_time=1.508
+[gpub010:0/16] 2024-01-29 04:14:29,193 (trainer:737) INFO: 14epoch:train:8101-8200batch: iter_time=9.802e-05, forward_time=0.414, loss_ctc=50.361, loss_att=54.038, acc=0.720, loss=52.935, backward_time=0.423, grad_norm=29.775, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=2.717e-04, train_time=1.671
+[gpub010:0/16] 2024-01-29 04:16:56,962 (trainer:737) INFO: 14epoch:train:8201-8300batch: iter_time=9.563e-05, forward_time=0.290, loss_ctc=51.300, loss_att=45.719, acc=0.744, loss=47.394, backward_time=0.401, grad_norm=28.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.717e-04, train_time=1.478
+[gpub010:0/16] 2024-01-29 04:19:53,453 (trainer:737) INFO: 14epoch:train:8301-8400batch: iter_time=4.007e-04, forward_time=0.458, loss_ctc=49.552, loss_att=44.572, acc=0.754, loss=46.066, backward_time=0.438, grad_norm=28.907, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.716e-04, train_time=1.763
+[gpub010:0/16] 2024-01-29 04:22:17,792 (trainer:737) INFO: 14epoch:train:8401-8500batch: iter_time=8.591e-05, forward_time=0.291, loss_ctc=56.682, loss_att=47.941, acc=0.768, loss=50.563, backward_time=0.403, grad_norm=51.888, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.715e-04, train_time=1.445
+[gpub010:0/16] 2024-01-29 04:24:47,583 (trainer:737) INFO: 14epoch:train:8501-8600batch: iter_time=9.557e-05, forward_time=0.290, loss_ctc=50.974, loss_att=46.689, acc=0.738, loss=47.975, backward_time=0.402, grad_norm=28.945, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.715e-04, train_time=1.498
+[gpub010:0/16] 2024-01-29 04:27:24,926 (trainer:737) INFO: 14epoch:train:8601-8700batch: iter_time=4.733e-04, forward_time=0.466, loss_ctc=53.620, loss_att=51.410, acc=0.728, loss=52.073, backward_time=0.444, grad_norm=31.829, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.714e-04, train_time=1.571
+[gpub010:0/16] 2024-01-29 04:29:14,583 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-01-29 04:29:33,581 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 04:29:37,147 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae6d42bf10>)
+[gpub010:0/16] 2024-01-29 04:29:37,148 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-01-29 04:29:37,151 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 04:35:40,333 (trainer:737) INFO: 14epoch:train:8701-8800batch: iter_time=3.169, forward_time=0.291, loss_ctc=49.880, loss_att=57.078, acc=0.721, loss=54.918, backward_time=0.402, grad_norm=29.645, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.713e-04, train_time=4.956
+[gpub010:0/16] 2024-01-29 04:38:27,515 (trainer:737) INFO: 14epoch:train:8801-8900batch: iter_time=0.002, forward_time=0.431, loss_ctc=60.007, loss_att=60.417, acc=0.714, loss=60.294, backward_time=0.422, grad_norm=34.307, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.713e-04, train_time=1.672
+[gpub010:0/16] 2024-01-29 04:40:48,221 (trainer:737) INFO: 14epoch:train:8901-9000batch: iter_time=8.291e-05, forward_time=0.288, loss_ctc=54.565, loss_att=55.261, acc=0.702, loss=55.052, backward_time=0.399, grad_norm=31.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.712e-04, train_time=1.406
+[gpub010:0/16] 2024-01-29 04:43:45,593 (trainer:737) INFO: 14epoch:train:9001-9100batch: iter_time=9.243e-05, forward_time=0.391, loss_ctc=51.732, loss_att=52.130, acc=0.736, loss=52.011, backward_time=0.461, grad_norm=29.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.711e-04, train_time=1.774
+[gpub010:0/16] 2024-01-29 04:45:59,729 (trainer:737) INFO: 14epoch:train:9101-9200batch: iter_time=8.918e-05, forward_time=0.301, loss_ctc=49.555, loss_att=40.676, acc=0.747, loss=43.340, backward_time=0.400, grad_norm=29.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.711e-04, train_time=1.341
+[gpub010:0/16] 2024-01-29 04:48:34,420 (trainer:737) INFO: 14epoch:train:9201-9300batch: iter_time=9.644e-05, forward_time=0.289, loss_ctc=56.605, loss_att=47.585, acc=0.712, loss=50.291, backward_time=0.400, grad_norm=37.832, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.710e-04, train_time=1.545
+[gpub010:0/16] 2024-01-29 04:51:14,434 (trainer:737) INFO: 14epoch:train:9301-9400batch: iter_time=8.577e-05, forward_time=0.415, loss_ctc=53.754, loss_att=58.817, acc=0.707, loss=57.298, backward_time=0.447, grad_norm=32.614, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.709e-04, train_time=1.601
+[gpub010:0/16] 2024-01-29 04:53:43,472 (trainer:737) INFO: 14epoch:train:9401-9500batch: iter_time=8.606e-05, forward_time=0.288, loss_ctc=49.711, loss_att=47.024, acc=0.724, loss=47.830, backward_time=0.399, grad_norm=29.196, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.709e-04, train_time=1.491
+[gpub010:0/16] 2024-01-29 04:56:19,993 (trainer:737) INFO: 14epoch:train:9501-9600batch: iter_time=8.565e-05, forward_time=0.289, loss_ctc=49.534, loss_att=41.093, acc=0.758, loss=43.626, backward_time=0.402, grad_norm=28.539, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.708e-04, train_time=1.564
+[gpub010:0/16] 2024-01-29 04:58:52,943 (trainer:737) INFO: 14epoch:train:9601-9700batch: iter_time=8.629e-05, forward_time=0.372, loss_ctc=50.350, loss_att=46.919, acc=0.750, loss=47.948, backward_time=0.451, grad_norm=30.052, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=2.707e-04, train_time=1.530
+[gpub010:0/16] 2024-01-29 05:01:15,083 (trainer:737) INFO: 14epoch:train:9701-9800batch: iter_time=8.753e-05, forward_time=0.288, loss_ctc=52.063, loss_att=44.716, acc=0.737, loss=46.920, backward_time=0.399, grad_norm=37.744, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.707e-04, train_time=1.421
+[gpub010:0/16] 2024-01-29 05:04:02,376 (trainer:737) INFO: 14epoch:train:9801-9900batch: iter_time=3.632e-04, forward_time=0.382, loss_ctc=57.097, loss_att=46.956, acc=0.732, loss=49.998, backward_time=0.448, grad_norm=30.198, clip=100.000, loss_scale=7.373e+33, optim_step_time=0.100, optim0_lr0=2.706e-04, train_time=1.671
+[gpub010:0/16] 2024-01-29 05:06:39,288 (trainer:737) INFO: 14epoch:train:9901-10000batch: iter_time=7.882e-05, forward_time=0.301, loss_ctc=49.125, loss_att=52.858, acc=0.721, loss=51.738, backward_time=0.403, grad_norm=29.729, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.705e-04, train_time=1.570
+[gpub010:0/16] 2024-01-29 05:06:59,316 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-01-29 05:07:18,147 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 05:07:21,720 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae6d694f10>)
+[gpub010:0/16] 2024-01-29 05:07:21,720 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-01-29 05:07:21,723 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 05:14:21,918 (trainer:737) INFO: 14epoch:train:10001-10100batch: iter_time=3.109, forward_time=0.293, loss_ctc=56.322, loss_att=64.534, acc=0.729, loss=62.070, backward_time=0.408, grad_norm=31.678, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.705e-04, train_time=4.626
+[gpub010:0/16] 2024-01-29 05:17:23,784 (trainer:737) INFO: 14epoch:train:10101-10200batch: iter_time=8.352e-05, forward_time=0.449, loss_ctc=60.279, loss_att=56.899, acc=0.714, loss=57.913, backward_time=0.435, grad_norm=33.861, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.704e-04, train_time=1.817
+[gpub010:0/16] 2024-01-29 05:19:38,900 (trainer:737) INFO: 14epoch:train:10201-10300batch: iter_time=8.122e-05, forward_time=0.291, loss_ctc=49.510, loss_att=56.098, acc=0.716, loss=54.122, backward_time=0.403, grad_norm=28.531, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.703e-04, train_time=1.352
+[gpub010:0/16] 2024-01-29 05:22:27,279 (trainer:737) INFO: 14epoch:train:10301-10400batch: iter_time=8.345e-05, forward_time=0.400, loss_ctc=47.876, loss_att=45.389, acc=0.742, loss=46.135, backward_time=0.425, grad_norm=28.860, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=2.703e-04, train_time=1.682
+[gpub010:0/16] 2024-01-29 05:25:11,206 (trainer:737) INFO: 14epoch:train:10401-10500batch: iter_time=8.624e-05, forward_time=0.290, loss_ctc=53.590, loss_att=44.387, acc=0.737, loss=47.148, backward_time=0.401, grad_norm=32.635, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.702e-04, train_time=1.640
+[gpub010:0/16] 2024-01-29 05:27:40,469 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 05:27:44,430 (trainer:737) INFO: 14epoch:train:10501-10600batch: iter_time=8.242e-05, forward_time=0.375, loss_ctc=52.778, loss_att=49.019, acc=0.735, loss=50.147, backward_time=0.422, grad_norm=30.223, clip=100.000, loss_scale=1.023e+34, optim_step_time=0.106, optim0_lr0=2.701e-04, train_time=1.532
+[gpub010:0/16] 2024-01-29 05:30:23,024 (trainer:737) INFO: 14epoch:train:10601-10700batch: iter_time=8.590e-05, forward_time=0.290, loss_ctc=49.295, loss_att=54.228, acc=0.719, loss=52.748, backward_time=0.402, grad_norm=30.100, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.701e-04, train_time=1.584
+[gpub010:0/16] 2024-01-29 05:33:03,264 (trainer:737) INFO: 14epoch:train:10701-10800batch: iter_time=8.456e-05, forward_time=0.400, loss_ctc=51.116, loss_att=46.324, acc=0.743, loss=47.762, backward_time=0.435, grad_norm=29.246, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.700e-04, train_time=1.604
+[gpub010:0/16] 2024-01-29 05:33:13,393 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 05:35:46,104 (trainer:737) INFO: 14epoch:train:10801-10900batch: iter_time=8.410e-05, forward_time=0.291, loss_ctc=48.958, loss_att=44.801, acc=0.755, loss=46.048, backward_time=0.402, grad_norm=28.872, clip=100.000, loss_scale=2.675e+33, optim_step_time=0.093, optim0_lr0=2.699e-04, train_time=1.627
+[gpub010:0/16] 2024-01-29 05:38:35,628 (trainer:737) INFO: 14epoch:train:10901-11000batch: iter_time=8.250e-05, forward_time=0.409, loss_ctc=57.000, loss_att=48.096, acc=0.767, loss=50.767, backward_time=0.475, grad_norm=36.801, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.699e-04, train_time=1.696
+[gpub010:0/16] 2024-01-29 05:41:08,076 (trainer:737) INFO: 14epoch:train:11001-11100batch: iter_time=8.553e-05, forward_time=0.293, loss_ctc=50.377, loss_att=45.788, acc=0.741, loss=47.164, backward_time=0.400, grad_norm=29.045, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.698e-04, train_time=1.524
+[gpub010:0/16] 2024-01-29 05:43:44,700 (trainer:737) INFO: 14epoch:train:11101-11200batch: iter_time=8.142e-05, forward_time=0.329, loss_ctc=52.836, loss_att=51.245, acc=0.728, loss=51.722, backward_time=0.405, grad_norm=31.756, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.697e-04, train_time=1.566
+[gpub010:0/16] 2024-01-29 05:45:27,963 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-01-29 05:45:46,973 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 05:45:50,616 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae590a7d30>)
+[gpub010:0/16] 2024-01-29 05:45:50,616 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-29 05:45:50,619 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 05:51:58,905 (trainer:737) INFO: 14epoch:train:11201-11300batch: iter_time=3.232, forward_time=0.367, loss_ctc=49.172, loss_att=54.566, acc=0.735, loss=52.948, backward_time=0.448, grad_norm=28.284, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.697e-04, train_time=4.942
+[gpub010:0/16] 2024-01-29 05:54:42,203 (trainer:737) INFO: 14epoch:train:11301-11400batch: iter_time=8.131e-05, forward_time=0.292, loss_ctc=60.110, loss_att=61.616, acc=0.720, loss=61.164, backward_time=0.405, grad_norm=34.308, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.696e-04, train_time=1.633
+[gpub010:0/16] 2024-01-29 05:57:02,959 (trainer:737) INFO: 14epoch:train:11401-11500batch: iter_time=8.342e-05, forward_time=0.309, loss_ctc=54.185, loss_att=54.776, acc=0.714, loss=54.599, backward_time=0.422, grad_norm=31.954, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.696e-04, train_time=1.407
+[gpub010:0/16] 2024-01-29 05:59:59,813 (trainer:737) INFO: 14epoch:train:11501-11600batch: iter_time=9.015e-05, forward_time=0.385, loss_ctc=51.184, loss_att=51.753, acc=0.745, loss=51.582, backward_time=0.437, grad_norm=29.317, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.695e-04, train_time=1.769
+[gpub010:0/16] 2024-01-29 06:02:28,710 (trainer:737) INFO: 14epoch:train:11601-11700batch: iter_time=8.661e-05, forward_time=0.287, loss_ctc=48.797, loss_att=40.187, acc=0.751, loss=42.770, backward_time=0.399, grad_norm=29.799, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.694e-04, train_time=1.489
+[gpub010:0/16] 2024-01-29 06:04:47,789 (trainer:737) INFO: 14epoch:train:11701-11800batch: iter_time=9.877e-05, forward_time=0.291, loss_ctc=56.723, loss_att=49.135, acc=0.719, loss=51.411, backward_time=0.407, grad_norm=34.604, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.694e-04, train_time=1.391
+[gpub010:0/16] 2024-01-29 06:08:00,614 (trainer:737) INFO: 14epoch:train:11801-11900batch: iter_time=9.885e-05, forward_time=0.400, loss_ctc=52.980, loss_att=58.314, acc=0.718, loss=56.714, backward_time=0.444, grad_norm=31.946, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.693e-04, train_time=1.928
+[gpub010:0/16] 2024-01-29 06:10:12,373 (trainer:737) INFO: 14epoch:train:11901-12000batch: iter_time=9.282e-05, forward_time=0.289, loss_ctc=48.764, loss_att=46.286, acc=0.737, loss=47.029, backward_time=0.400, grad_norm=29.437, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.692e-04, train_time=1.318
+[gpub010:0/16] 2024-01-29 06:12:33,375 (trainer:737) INFO: 14epoch:train:12001-12100batch: iter_time=9.968e-05, forward_time=0.290, loss_ctc=49.405, loss_att=40.598, acc=0.764, loss=43.240, backward_time=0.403, grad_norm=27.912, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.692e-04, train_time=1.410
+[gpub010:0/16] 2024-01-29 06:15:44,980 (trainer:737) INFO: 14epoch:train:12101-12200batch: iter_time=4.902e-04, forward_time=0.406, loss_ctc=49.918, loss_att=47.450, acc=0.764, loss=48.190, backward_time=0.447, grad_norm=28.002, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.691e-04, train_time=1.916
+[gpub010:0/16] 2024-01-29 06:17:58,655 (trainer:737) INFO: 14epoch:train:12201-12300batch: iter_time=9.560e-05, forward_time=0.289, loss_ctc=50.308, loss_att=44.123, acc=0.747, loss=45.979, backward_time=0.400, grad_norm=35.924, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.690e-04, train_time=1.336
+[gpub010:0/16] 2024-01-29 06:20:44,827 (trainer:737) INFO: 14epoch:train:12301-12400batch: iter_time=9.077e-05, forward_time=0.292, loss_ctc=56.404, loss_att=47.053, acc=0.737, loss=49.858, backward_time=0.404, grad_norm=32.678, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.690e-04, train_time=1.662
+[gpub010:0/16] 2024-01-29 06:23:36,020 (trainer:737) INFO: 14epoch:train:12401-12500batch: iter_time=3.089e-04, forward_time=0.437, loss_ctc=48.537, loss_att=52.077, acc=0.731, loss=51.015, backward_time=0.443, grad_norm=30.068, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.689e-04, train_time=1.711
+[gpub010:0/16] 2024-01-29 06:23:56,172 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-01-29 06:24:15,174 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 06:24:18,826 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae591a59f0>)
+[gpub010:0/16] 2024-01-29 06:24:18,826 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-01-29 06:24:18,829 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 06:31:27,390 (trainer:737) INFO: 14epoch:train:12501-12600batch: iter_time=3.099, forward_time=0.293, loss_ctc=56.704, loss_att=61.339, acc=0.735, loss=59.948, backward_time=0.406, grad_norm=30.810, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.688e-04, train_time=4.714
+[gpub010:0/16] 2024-01-29 06:34:16,654 (trainer:737) INFO: 14epoch:train:12601-12700batch: iter_time=9.458e-05, forward_time=0.364, loss_ctc=59.414, loss_att=56.518, acc=0.714, loss=57.387, backward_time=0.436, grad_norm=34.854, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.688e-04, train_time=1.693
+[gpub010:0/16] 2024-01-29 06:36:33,726 (trainer:737) INFO: 14epoch:train:12701-12800batch: iter_time=9.024e-05, forward_time=0.292, loss_ctc=49.450, loss_att=54.924, acc=0.720, loss=53.282, backward_time=0.404, grad_norm=28.857, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.687e-04, train_time=1.371
+[gpub010:0/16] 2024-01-29 06:39:21,598 (trainer:737) INFO: 14epoch:train:12801-12900batch: iter_time=9.276e-05, forward_time=0.294, loss_ctc=47.725, loss_att=44.499, acc=0.744, loss=45.467, backward_time=0.401, grad_norm=28.520, clip=100.000, loss_scale=5.088e+33, optim_step_time=0.092, optim0_lr0=2.686e-04, train_time=1.678
+[gpub010:0/16] 2024-01-29 06:42:15,716 (trainer:737) INFO: 14epoch:train:12901-13000batch: iter_time=9.989e-05, forward_time=0.343, loss_ctc=53.570, loss_att=43.526, acc=0.741, loss=46.539, backward_time=0.444, grad_norm=33.581, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.686e-04, train_time=1.742
+[gpub010:0/16] 2024-01-29 06:44:49,247 (trainer:737) INFO: 14epoch:train:13001-13100batch: iter_time=9.745e-05, forward_time=0.314, loss_ctc=52.621, loss_att=49.058, acc=0.733, loss=50.127, backward_time=0.409, grad_norm=32.058, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.685e-04, train_time=1.534
+[gpub010:0/16] 2024-01-29 06:47:22,063 (trainer:737) INFO: 14epoch:train:13101-13200batch: iter_time=9.468e-05, forward_time=0.290, loss_ctc=49.792, loss_att=53.620, acc=0.725, loss=52.471, backward_time=0.402, grad_norm=31.003, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.684e-04, train_time=1.529
+[gpub010:0/16] 2024-01-29 06:49:59,518 (trainer:737) INFO: 14epoch:train:13201-13300batch: iter_time=5.173e-04, forward_time=0.378, loss_ctc=51.021, loss_att=45.581, acc=0.749, loss=47.213, backward_time=0.432, grad_norm=27.874, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.684e-04, train_time=1.574
+[gpub010:0/16] 2024-01-29 06:52:53,494 (trainer:737) INFO: 14epoch:train:13301-13400batch: iter_time=9.723e-05, forward_time=0.324, loss_ctc=48.966, loss_att=44.041, acc=0.756, loss=45.518, backward_time=0.411, grad_norm=29.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.683e-04, train_time=1.740
+[gpub010:0/16] 2024-01-29 06:55:10,497 (trainer:737) INFO: 14epoch:train:13401-13500batch: iter_time=1.022e-04, forward_time=0.293, loss_ctc=55.049, loss_att=46.999, acc=0.770, loss=49.414, backward_time=0.405, grad_norm=36.876, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.683e-04, train_time=1.369
+[gpub010:0/16] 2024-01-29 06:58:29,593 (trainer:737) INFO: 14epoch:train:13501-13600batch: iter_time=3.288e-04, forward_time=0.400, loss_ctc=50.686, loss_att=46.111, acc=0.740, loss=47.484, backward_time=0.446, grad_norm=29.130, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.682e-04, train_time=1.991
+[gpub010:0/16] 2024-01-29 07:00:47,221 (trainer:737) INFO: 14epoch:train:13601-13700batch: iter_time=9.792e-05, forward_time=0.292, loss_ctc=52.361, loss_att=50.612, acc=0.731, loss=51.137, backward_time=0.404, grad_norm=31.349, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.681e-04, train_time=1.376
+[gpub010:0/16] 2024-01-29 07:02:32,698 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-01-29 07:02:52,241 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 07:02:56,165 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae07753ac0>)
+[gpub010:0/16] 2024-01-29 07:02:56,165 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-01-29 07:02:56,169 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 07:10:03,034 (trainer:737) INFO: 14epoch:train:13701-13800batch: iter_time=3.853, forward_time=0.295, loss_ctc=48.840, loss_att=56.923, acc=0.723, loss=54.498, backward_time=0.404, grad_norm=29.607, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.681e-04, train_time=5.558
+[gpub010:0/16] 2024-01-29 07:13:07,712 (trainer:737) INFO: 14epoch:train:13801-13900batch: iter_time=8.021e-05, forward_time=0.358, loss_ctc=59.683, loss_att=61.435, acc=0.714, loss=60.909, backward_time=0.470, grad_norm=43.139, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.680e-04, train_time=1.847
+[gpub010:0/16] 2024-01-29 07:15:22,340 (trainer:737) INFO: 14epoch:train:13901-14000batch: iter_time=8.105e-05, forward_time=0.307, loss_ctc=53.509, loss_att=55.524, acc=0.704, loss=54.919, backward_time=0.403, grad_norm=31.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.679e-04, train_time=1.346
+[gpub010:0/16] 2024-01-29 07:18:27,568 (trainer:737) INFO: 14epoch:train:14001-14100batch: iter_time=7.981e-05, forward_time=0.293, loss_ctc=50.362, loss_att=52.061, acc=0.740, loss=51.551, backward_time=0.405, grad_norm=30.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.679e-04, train_time=1.852
+[gpub010:0/16] 2024-01-29 07:20:56,319 (trainer:737) INFO: 14epoch:train:14101-14200batch: iter_time=8.369e-05, forward_time=0.352, loss_ctc=48.290, loss_att=39.730, acc=0.750, loss=42.298, backward_time=0.460, grad_norm=29.944, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.678e-04, train_time=1.488
+[gpub010:0/16] 2024-01-29 07:23:38,546 (trainer:737) INFO: 14epoch:train:14201-14300batch: iter_time=8.315e-05, forward_time=0.324, loss_ctc=55.389, loss_att=47.537, acc=0.714, loss=49.892, backward_time=0.409, grad_norm=32.373, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.677e-04, train_time=1.622
+[gpub010:0/16] 2024-01-29 07:26:11,659 (trainer:737) INFO: 14epoch:train:14301-14400batch: iter_time=8.210e-05, forward_time=0.291, loss_ctc=52.429, loss_att=58.556, acc=0.710, loss=56.718, backward_time=0.404, grad_norm=33.285, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.677e-04, train_time=1.531
+[gpub010:0/16] 2024-01-29 07:28:52,194 (trainer:737) INFO: 14epoch:train:14401-14500batch: iter_time=8.350e-05, forward_time=0.290, loss_ctc=48.836, loss_att=47.397, acc=0.722, loss=47.828, backward_time=0.409, grad_norm=30.034, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.676e-04, train_time=1.604
+[gpub010:0/16] 2024-01-29 07:31:37,617 (trainer:737) INFO: 14epoch:train:14501-14600batch: iter_time=8.138e-05, forward_time=0.348, loss_ctc=49.256, loss_att=41.319, acc=0.756, loss=43.700, backward_time=0.468, grad_norm=29.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.676e-04, train_time=1.655
+[gpub010:0/16] 2024-01-29 07:34:17,443 (trainer:737) INFO: 14epoch:train:14601-14700batch: iter_time=8.196e-05, forward_time=0.314, loss_ctc=49.066, loss_att=46.349, acc=0.753, loss=47.164, backward_time=0.408, grad_norm=29.890, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.675e-04, train_time=1.598
+[gpub010:0/16] 2024-01-29 07:36:39,420 (trainer:737) INFO: 14epoch:train:14701-14800batch: iter_time=2.098e-04, forward_time=0.338, loss_ctc=50.595, loss_att=45.158, acc=0.738, loss=46.789, backward_time=0.426, grad_norm=39.118, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.674e-04, train_time=1.419
+[gpub010:0/16] 2024-01-29 07:36:59,283 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 07:39:24,631 (trainer:737) INFO: 14epoch:train:14801-14900batch: iter_time=8.262e-05, forward_time=0.332, loss_ctc=56.323, loss_att=46.495, acc=0.734, loss=49.444, backward_time=0.420, grad_norm=32.549, clip=100.000, loss_scale=5.664e+33, optim_step_time=0.092, optim0_lr0=2.674e-04, train_time=1.652
+[gpub010:0/16] 2024-01-29 07:41:54,084 (trainer:737) INFO: 14epoch:train:14901-15000batch: iter_time=7.957e-05, forward_time=0.290, loss_ctc=48.396, loss_att=51.919, acc=0.727, loss=50.862, backward_time=0.403, grad_norm=31.647, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.673e-04, train_time=1.495
+[gpub010:0/16] 2024-01-29 08:19:34,261 (trainer:343) INFO: 14epoch results: [train] iter_time=0.265, forward_time=0.329, loss_ctc=53.130, loss_att=50.514, acc=0.730, loss=51.299, backward_time=0.417, grad_norm=32.053, clip=100.000, loss_scale=5.975e+33, optim_step_time=0.095, optim0_lr0=2.722e-04, train_time=1.868, time=7 hours, 47 minutes and 26.16 seconds, total_count=240000, gpu_max_cached_mem_GB=42.072, [valid] loss_ctc=45.300, cer_ctc=0.228, loss_att=43.953, acc=0.646, cer=0.326, wer=0.995, loss=44.357, time=37 minutes and 16.23 seconds, total_count=74736, gpu_max_cached_mem_GB=42.072
+[gpub010:0/16] 2024-01-29 08:19:44,183 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub010:0/16] 2024-01-29 08:19:44,210 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/9epoch.pth
+[gpub010:0/16] 2024-01-29 08:19:44,210 (trainer:272) INFO: 15/45epoch started. Estimated time to finish: 1 week, 4 days and 3 hours
+[gpub010:0/16] 2024-01-29 08:19:44,221 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-29 08:20:02,715 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 08:20:06,181 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa7306ebc70>)
+[gpub010:0/16] 2024-01-29 08:20:06,181 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-01-29 08:20:06,185 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 08:27:04,509 (trainer:737) INFO: 15epoch:train:1-100batch: iter_time=2.789, forward_time=0.335, loss_ctc=52.082, loss_att=53.177, acc=0.715, loss=52.848, backward_time=0.410, grad_norm=31.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.672e-04, train_time=4.403
+[gpub010:0/16] 2024-01-29 08:29:22,578 (trainer:737) INFO: 15epoch:train:101-200batch: iter_time=8.335e-05, forward_time=0.302, loss_ctc=53.280, loss_att=43.521, acc=0.730, loss=46.448, backward_time=0.400, grad_norm=36.207, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.672e-04, train_time=1.380
+[gpub010:0/16] 2024-01-29 08:32:20,052 (trainer:737) INFO: 15epoch:train:201-300batch: iter_time=1.353e-04, forward_time=0.321, loss_ctc=54.009, loss_att=54.277, acc=0.718, loss=54.197, backward_time=0.421, grad_norm=32.916, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.671e-04, train_time=1.774
+[gpub010:0/16] 2024-01-29 08:34:56,376 (trainer:737) INFO: 15epoch:train:301-400batch: iter_time=9.242e-05, forward_time=0.324, loss_ctc=57.442, loss_att=54.865, acc=0.701, loss=55.638, backward_time=0.409, grad_norm=40.140, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.670e-04, train_time=1.563
+[gpub010:0/16] 2024-01-29 08:37:34,976 (trainer:737) INFO: 15epoch:train:401-500batch: iter_time=8.472e-05, forward_time=0.294, loss_ctc=61.243, loss_att=58.725, acc=0.717, loss=59.480, backward_time=0.413, grad_norm=33.859, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.670e-04, train_time=1.585
+[gpub010:0/16] 2024-01-29 08:40:13,784 (trainer:737) INFO: 15epoch:train:501-600batch: iter_time=8.783e-05, forward_time=0.376, loss_ctc=53.714, loss_att=49.089, acc=0.735, loss=50.476, backward_time=0.416, grad_norm=31.766, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.669e-04, train_time=1.589
+[gpub010:0/16] 2024-01-29 08:43:03,431 (trainer:737) INFO: 15epoch:train:601-700batch: iter_time=9.257e-05, forward_time=0.315, loss_ctc=62.565, loss_att=64.169, acc=0.704, loss=63.688, backward_time=0.420, grad_norm=36.694, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.669e-04, train_time=1.697
+[gpub010:0/16] 2024-01-29 08:45:51,714 (trainer:737) INFO: 15epoch:train:701-800batch: iter_time=8.972e-05, forward_time=0.318, loss_ctc=56.545, loss_att=50.381, acc=0.724, loss=52.230, backward_time=0.426, grad_norm=33.878, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.668e-04, train_time=1.682
+[gpub010:0/16] 2024-01-29 08:48:31,278 (trainer:737) INFO: 15epoch:train:801-900batch: iter_time=9.359e-05, forward_time=0.334, loss_ctc=55.734, loss_att=47.156, acc=0.765, loss=49.730, backward_time=0.423, grad_norm=30.066, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.667e-04, train_time=1.596
+[gpub010:0/16] 2024-01-29 08:51:05,583 (trainer:737) INFO: 15epoch:train:901-1000batch: iter_time=5.981e-04, forward_time=0.303, loss_ctc=56.656, loss_att=47.296, acc=0.740, loss=50.104, backward_time=0.411, grad_norm=31.814, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.667e-04, train_time=1.543
+[gpub010:0/16] 2024-01-29 08:53:15,406 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 08:53:51,452 (trainer:737) INFO: 15epoch:train:1001-1100batch: iter_time=9.123e-05, forward_time=0.347, loss_ctc=53.336, loss_att=54.683, acc=0.717, loss=54.279, backward_time=0.411, grad_norm=30.268, clip=100.000, loss_scale=4.615e+33, optim_step_time=0.099, optim0_lr0=2.666e-04, train_time=1.657
+[gpub010:0/16] 2024-01-29 08:56:19,852 (trainer:737) INFO: 15epoch:train:1101-1200batch: iter_time=8.600e-05, forward_time=0.328, loss_ctc=56.468, loss_att=58.075, acc=0.718, loss=57.593, backward_time=0.417, grad_norm=31.629, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.665e-04, train_time=1.485
+[gpub010:0/16] 2024-01-29 08:57:51,802 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-01-29 08:58:10,965 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 08:58:14,568 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb07d847070>)
+[gpub010:0/16] 2024-01-29 08:58:14,568 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-01-29 08:58:14,571 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 09:04:19,157 (trainer:737) INFO: 15epoch:train:1201-1300batch: iter_time=3.194, forward_time=0.349, loss_ctc=58.999, loss_att=59.217, acc=0.713, loss=59.152, backward_time=0.415, grad_norm=36.350, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.665e-04, train_time=4.793
+[gpub010:0/16] 2024-01-29 09:06:59,695 (trainer:737) INFO: 15epoch:train:1301-1400batch: iter_time=8.332e-05, forward_time=0.323, loss_ctc=52.082, loss_att=45.948, acc=0.712, loss=47.788, backward_time=0.403, grad_norm=34.895, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.664e-04, train_time=1.604
+[gpub010:0/16] 2024-01-29 09:09:42,637 (trainer:737) INFO: 15epoch:train:1401-1500batch: iter_time=8.423e-05, forward_time=0.304, loss_ctc=53.468, loss_att=53.913, acc=0.708, loss=53.779, backward_time=0.416, grad_norm=32.336, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.663e-04, train_time=1.630
+[gpub010:0/16] 2024-01-29 09:12:38,182 (trainer:737) INFO: 15epoch:train:1501-1600batch: iter_time=6.375e-04, forward_time=0.312, loss_ctc=54.374, loss_att=51.211, acc=0.706, loss=52.160, backward_time=0.446, grad_norm=37.730, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.663e-04, train_time=1.755
+[gpub010:0/16] 2024-01-29 09:15:17,231 (trainer:737) INFO: 15epoch:train:1601-1700batch: iter_time=8.657e-05, forward_time=0.344, loss_ctc=54.852, loss_att=53.415, acc=0.708, loss=53.846, backward_time=0.418, grad_norm=34.517, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.662e-04, train_time=1.590
+[gpub010:0/16] 2024-01-29 09:18:03,233 (trainer:737) INFO: 15epoch:train:1701-1800batch: iter_time=8.852e-05, forward_time=0.299, loss_ctc=58.176, loss_att=53.617, acc=0.726, loss=54.985, backward_time=0.402, grad_norm=32.325, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.662e-04, train_time=1.661
+[gpub010:0/16] 2024-01-29 09:21:00,984 (trainer:737) INFO: 15epoch:train:1801-1900batch: iter_time=8.251e-05, forward_time=0.355, loss_ctc=57.694, loss_att=53.158, acc=0.718, loss=54.519, backward_time=0.417, grad_norm=34.849, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.661e-04, train_time=1.777
+[gpub010:0/16] 2024-01-29 09:23:37,859 (trainer:737) INFO: 15epoch:train:1901-2000batch: iter_time=9.149e-05, forward_time=0.318, loss_ctc=56.249, loss_att=57.624, acc=0.707, loss=57.212, backward_time=0.417, grad_norm=31.423, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.660e-04, train_time=1.568
+[gpub010:0/16] 2024-01-29 09:26:29,604 (trainer:737) INFO: 15epoch:train:2001-2100batch: iter_time=8.324e-05, forward_time=0.292, loss_ctc=56.321, loss_att=49.730, acc=0.738, loss=51.707, backward_time=0.403, grad_norm=34.157, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.660e-04, train_time=1.718
+[gpub010:0/16] 2024-01-29 09:29:13,038 (trainer:737) INFO: 15epoch:train:2101-2200batch: iter_time=9.285e-05, forward_time=0.358, loss_ctc=50.493, loss_att=38.757, acc=0.764, loss=42.278, backward_time=0.434, grad_norm=27.293, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=2.659e-04, train_time=1.633
+[gpub010:0/16] 2024-01-29 09:31:55,935 (trainer:737) INFO: 15epoch:train:2201-2300batch: iter_time=8.818e-05, forward_time=0.304, loss_ctc=55.874, loss_att=52.855, acc=0.717, loss=53.761, backward_time=0.416, grad_norm=33.823, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.658e-04, train_time=1.629
+[gpub010:0/16] 2024-01-29 09:34:59,052 (trainer:737) INFO: 15epoch:train:2301-2400batch: iter_time=8.961e-05, forward_time=0.300, loss_ctc=55.096, loss_att=58.112, acc=0.716, loss=57.207, backward_time=0.402, grad_norm=32.324, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.658e-04, train_time=1.832
+[gpub010:0/16] 2024-01-29 09:37:34,606 (trainer:737) INFO: 15epoch:train:2401-2500batch: iter_time=8.574e-05, forward_time=0.379, loss_ctc=62.381, loss_att=60.784, acc=0.705, loss=61.263, backward_time=0.426, grad_norm=37.478, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.657e-04, train_time=1.555
+[gpub010:0/16] 2024-01-29 09:37:54,817 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-01-29 09:38:13,745 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 09:38:17,326 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7faea86df760>)
+[gpub010:0/16] 2024-01-29 09:38:17,326 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-01-29 09:38:17,329 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 09:47:08,919 (trainer:737) INFO: 15epoch:train:2501-2600batch: iter_time=4.180, forward_time=0.321, loss_ctc=50.335, loss_att=51.504, acc=0.722, loss=51.153, backward_time=0.417, grad_norm=30.874, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.657e-04, train_time=5.743
+[gpub010:0/16] 2024-01-29 09:50:11,848 (trainer:737) INFO: 15epoch:train:2601-2700batch: iter_time=8.994e-04, forward_time=0.292, loss_ctc=49.502, loss_att=41.497, acc=0.736, loss=43.899, backward_time=0.430, grad_norm=32.050, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.656e-04, train_time=1.830
+[gpub010:0/16] 2024-01-29 09:52:44,166 (trainer:737) INFO: 15epoch:train:2701-2800batch: iter_time=8.468e-05, forward_time=0.367, loss_ctc=52.364, loss_att=53.538, acc=0.724, loss=53.186, backward_time=0.432, grad_norm=31.268, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.655e-04, train_time=1.523
+[gpub010:0/16] 2024-01-29 09:55:07,223 (trainer:737) INFO: 15epoch:train:2801-2900batch: iter_time=8.970e-05, forward_time=0.303, loss_ctc=53.689, loss_att=53.710, acc=0.708, loss=53.704, backward_time=0.408, grad_norm=39.311, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.655e-04, train_time=1.431
+[gpub010:0/16] 2024-01-29 09:57:43,584 (trainer:737) INFO: 15epoch:train:2901-3000batch: iter_time=8.747e-05, forward_time=0.300, loss_ctc=61.088, loss_att=58.047, acc=0.721, loss=58.959, backward_time=0.406, grad_norm=32.797, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.654e-04, train_time=1.563
+[gpub010:0/16] 2024-01-29 10:00:36,161 (trainer:737) INFO: 15epoch:train:3001-3100batch: iter_time=8.758e-05, forward_time=0.377, loss_ctc=52.530, loss_att=47.992, acc=0.740, loss=49.354, backward_time=0.442, grad_norm=32.674, clip=100.000, loss_scale=3.167e+33, optim_step_time=0.099, optim0_lr0=2.653e-04, train_time=1.726
+[gpub010:0/16] 2024-01-29 10:03:21,612 (trainer:737) INFO: 15epoch:train:3101-3200batch: iter_time=9.366e-05, forward_time=0.295, loss_ctc=60.394, loss_att=63.211, acc=0.708, loss=62.366, backward_time=0.407, grad_norm=36.151, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.653e-04, train_time=1.655
+[gpub010:0/16] 2024-01-29 10:05:48,899 (trainer:737) INFO: 15epoch:train:3201-3300batch: iter_time=8.762e-05, forward_time=0.338, loss_ctc=55.699, loss_att=50.120, acc=0.726, loss=51.793, backward_time=0.430, grad_norm=32.872, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.652e-04, train_time=1.473
+[gpub010:0/16] 2024-01-29 10:08:25,142 (trainer:737) INFO: 15epoch:train:3301-3400batch: iter_time=8.527e-05, forward_time=0.348, loss_ctc=54.287, loss_att=47.130, acc=0.767, loss=49.277, backward_time=0.422, grad_norm=30.975, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.652e-04, train_time=1.560
+[gpub010:0/16] 2024-01-29 10:10:19,157 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 10:11:08,876 (trainer:737) INFO: 15epoch:train:3401-3500batch: iter_time=8.398e-05, forward_time=0.290, loss_ctc=54.671, loss_att=46.934, acc=0.741, loss=49.255, backward_time=0.402, grad_norm=30.531, clip=100.000, loss_scale=4.327e+33, optim_step_time=0.092, optim0_lr0=2.651e-04, train_time=1.639
+[gpub010:0/16] 2024-01-29 10:13:38,072 (trainer:737) INFO: 15epoch:train:3501-3600batch: iter_time=8.709e-05, forward_time=0.348, loss_ctc=51.749, loss_att=53.077, acc=0.725, loss=52.679, backward_time=0.430, grad_norm=30.219, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.650e-04, train_time=1.492
+[gpub010:0/16] 2024-01-29 10:16:17,490 (trainer:737) INFO: 15epoch:train:3601-3700batch: iter_time=8.438e-05, forward_time=0.335, loss_ctc=55.143, loss_att=57.165, acc=0.722, loss=56.558, backward_time=0.426, grad_norm=30.525, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.650e-04, train_time=1.593
+[gpub010:0/16] 2024-01-29 10:18:14,855 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-01-29 10:18:34,281 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 10:18:38,234 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb0ce912fe0>)
+[gpub010:0/16] 2024-01-29 10:18:38,234 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-01-29 10:18:38,237 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 10:24:23,343 (trainer:737) INFO: 15epoch:train:3701-3800batch: iter_time=3.090, forward_time=0.339, loss_ctc=55.594, loss_att=57.696, acc=0.714, loss=57.065, backward_time=0.407, grad_norm=44.643, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.649e-04, train_time=4.859
+[gpub010:0/16] 2024-01-29 10:26:47,589 (trainer:737) INFO: 15epoch:train:3801-3900batch: iter_time=8.738e-05, forward_time=0.318, loss_ctc=51.268, loss_att=45.221, acc=0.714, loss=47.035, backward_time=0.425, grad_norm=33.140, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.648e-04, train_time=1.442
+[gpub010:0/16] 2024-01-29 10:29:13,169 (trainer:737) INFO: 15epoch:train:3901-4000batch: iter_time=8.723e-05, forward_time=0.324, loss_ctc=52.449, loss_att=53.001, acc=0.710, loss=52.835, backward_time=0.419, grad_norm=31.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.648e-04, train_time=1.455
+[gpub010:0/16] 2024-01-29 10:32:05,306 (trainer:737) INFO: 15epoch:train:4001-4100batch: iter_time=9.727e-05, forward_time=0.307, loss_ctc=52.957, loss_att=50.254, acc=0.711, loss=51.065, backward_time=0.415, grad_norm=36.114, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.647e-04, train_time=1.721
+[gpub010:0/16] 2024-01-29 10:34:32,145 (trainer:737) INFO: 15epoch:train:4101-4200batch: iter_time=8.700e-05, forward_time=0.330, loss_ctc=53.447, loss_att=52.682, acc=0.712, loss=52.911, backward_time=0.417, grad_norm=33.900, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.647e-04, train_time=1.469
+[gpub010:0/16] 2024-01-29 10:37:00,034 (trainer:737) INFO: 15epoch:train:4201-4300batch: iter_time=8.275e-05, forward_time=0.307, loss_ctc=58.176, loss_att=53.832, acc=0.727, loss=55.135, backward_time=0.417, grad_norm=32.334, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.646e-04, train_time=1.478
+[gpub010:0/16] 2024-01-29 10:39:53,608 (trainer:737) INFO: 15epoch:train:4301-4400batch: iter_time=2.219e-04, forward_time=0.327, loss_ctc=56.976, loss_att=52.627, acc=0.724, loss=53.932, backward_time=0.417, grad_norm=33.035, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.645e-04, train_time=1.736
+[gpub010:0/16] 2024-01-29 10:42:20,684 (trainer:737) INFO: 15epoch:train:4401-4500batch: iter_time=8.086e-05, forward_time=0.350, loss_ctc=55.420, loss_att=56.323, acc=0.711, loss=56.052, backward_time=0.431, grad_norm=30.442, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.645e-04, train_time=1.471
+[gpub010:0/16] 2024-01-29 10:44:57,610 (trainer:737) INFO: 15epoch:train:4501-4600batch: iter_time=8.879e-05, forward_time=0.301, loss_ctc=55.517, loss_att=49.211, acc=0.740, loss=51.103, backward_time=0.402, grad_norm=32.654, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.644e-04, train_time=1.569
+[gpub010:0/16] 2024-01-29 10:47:37,592 (trainer:737) INFO: 15epoch:train:4601-4700batch: iter_time=4.752e-04, forward_time=0.339, loss_ctc=49.654, loss_att=38.175, acc=0.767, loss=41.619, backward_time=0.409, grad_norm=27.585, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=2.644e-04, train_time=1.600
+[gpub010:0/16] 2024-01-29 10:50:11,473 (trainer:737) INFO: 15epoch:train:4701-4800batch: iter_time=8.941e-05, forward_time=0.340, loss_ctc=54.223, loss_att=51.729, acc=0.723, loss=52.477, backward_time=0.419, grad_norm=32.753, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.643e-04, train_time=1.539
+[gpub010:0/16] 2024-01-29 10:52:50,334 (trainer:737) INFO: 15epoch:train:4801-4900batch: iter_time=2.018e-04, forward_time=0.357, loss_ctc=53.776, loss_att=57.011, acc=0.720, loss=56.040, backward_time=0.410, grad_norm=30.194, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.642e-04, train_time=1.588
+[gpub010:0/16] 2024-01-29 10:55:14,569 (trainer:737) INFO: 15epoch:train:4901-5000batch: iter_time=3.389e-04, forward_time=0.324, loss_ctc=60.514, loss_att=59.539, acc=0.708, loss=59.832, backward_time=0.408, grad_norm=38.019, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.642e-04, train_time=1.442
+[gpub010:0/16] 2024-01-29 10:55:34,597 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-01-29 10:55:53,724 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 10:55:57,305 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae6fafb610>)
+[gpub010:0/16] 2024-01-29 10:55:57,306 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-01-29 10:55:57,309 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 11:03:28,989 (trainer:737) INFO: 15epoch:train:5001-5100batch: iter_time=3.491, forward_time=0.339, loss_ctc=49.651, loss_att=49.603, acc=0.707, loss=49.617, backward_time=0.412, grad_norm=31.071, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.641e-04, train_time=4.944
+[gpub010:0/16] 2024-01-29 11:06:08,294 (trainer:737) INFO: 15epoch:train:5101-5200batch: iter_time=9.351e-05, forward_time=0.332, loss_ctc=48.836, loss_att=40.560, acc=0.729, loss=43.043, backward_time=0.460, grad_norm=33.108, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=2.640e-04, train_time=1.592
+[gpub010:0/16] 2024-01-29 11:08:40,639 (trainer:737) INFO: 15epoch:train:5201-5300batch: iter_time=8.094e-05, forward_time=0.341, loss_ctc=52.189, loss_att=52.992, acc=0.713, loss=52.751, backward_time=0.417, grad_norm=32.329, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.640e-04, train_time=1.524
+[gpub010:0/16] 2024-01-29 11:11:22,368 (trainer:737) INFO: 15epoch:train:5301-5400batch: iter_time=8.564e-05, forward_time=0.288, loss_ctc=52.739, loss_att=53.370, acc=0.707, loss=53.180, backward_time=0.399, grad_norm=44.088, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.639e-04, train_time=1.617
+[gpub010:0/16] 2024-01-29 11:14:04,874 (trainer:737) INFO: 15epoch:train:5401-5500batch: iter_time=8.607e-05, forward_time=0.339, loss_ctc=59.481, loss_att=57.010, acc=0.711, loss=57.751, backward_time=0.433, grad_norm=33.257, clip=100.000, loss_scale=3.453e+33, optim_step_time=0.102, optim0_lr0=2.639e-04, train_time=1.624
+[gpub010:0/16] 2024-01-29 11:16:34,546 (trainer:737) INFO: 15epoch:train:5501-5600batch: iter_time=8.439e-05, forward_time=0.335, loss_ctc=51.503, loss_att=46.746, acc=0.742, loss=48.173, backward_time=0.427, grad_norm=30.868, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.638e-04, train_time=1.497
+[gpub010:0/16] 2024-01-29 11:19:26,288 (trainer:737) INFO: 15epoch:train:5601-5700batch: iter_time=2.830e-04, forward_time=0.331, loss_ctc=59.335, loss_att=59.962, acc=0.704, loss=59.774, backward_time=0.430, grad_norm=36.486, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.637e-04, train_time=1.717
+[gpub010:0/16] 2024-01-29 11:21:56,588 (trainer:737) INFO: 15epoch:train:5701-5800batch: iter_time=8.758e-05, forward_time=0.328, loss_ctc=54.899, loss_att=48.671, acc=0.723, loss=50.539, backward_time=0.419, grad_norm=32.423, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.637e-04, train_time=1.502
+[gpub010:0/16] 2024-01-29 11:24:23,791 (trainer:737) INFO: 15epoch:train:5801-5900batch: iter_time=8.543e-05, forward_time=0.304, loss_ctc=53.752, loss_att=44.498, acc=0.770, loss=47.274, backward_time=0.410, grad_norm=28.605, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.636e-04, train_time=1.472
+[gpub010:0/16] 2024-01-29 11:27:13,243 (trainer:737) INFO: 15epoch:train:5901-6000batch: iter_time=8.374e-05, forward_time=0.334, loss_ctc=53.643, loss_att=46.449, acc=0.737, loss=48.607, backward_time=0.448, grad_norm=32.437, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.636e-04, train_time=1.694
+[gpub010:0/16] 2024-01-29 11:29:37,884 (trainer:737) INFO: 15epoch:train:6001-6100batch: iter_time=8.435e-05, forward_time=0.346, loss_ctc=51.206, loss_att=53.000, acc=0.718, loss=52.461, backward_time=0.418, grad_norm=30.983, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.635e-04, train_time=1.446
+[gpub010:0/16] 2024-01-29 11:32:24,634 (trainer:737) INFO: 15epoch:train:6101-6200batch: iter_time=2.307e-04, forward_time=0.333, loss_ctc=54.786, loss_att=55.994, acc=0.717, loss=55.632, backward_time=0.414, grad_norm=30.332, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.634e-04, train_time=1.668
+[gpub010:0/16] 2024-01-29 11:34:06,693 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-01-29 11:34:26,077 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 11:34:29,672 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae54ce3c10>)
+[gpub010:0/16] 2024-01-29 11:34:29,672 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-01-29 11:34:29,675 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 11:40:00,519 (trainer:737) INFO: 15epoch:train:6201-6300batch: iter_time=2.942, forward_time=0.339, loss_ctc=54.789, loss_att=55.647, acc=0.718, loss=55.390, backward_time=0.411, grad_norm=34.145, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.634e-04, train_time=4.559
+[gpub010:0/16] 2024-01-29 11:42:16,184 (trainer:737) INFO: 15epoch:train:6301-6400batch: iter_time=9.575e-05, forward_time=0.288, loss_ctc=50.073, loss_att=43.869, acc=0.719, loss=45.730, backward_time=0.399, grad_norm=33.749, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.633e-04, train_time=1.357
+[gpub010:0/16] 2024-01-29 11:45:07,445 (trainer:737) INFO: 15epoch:train:6401-6500batch: iter_time=7.072e-04, forward_time=0.317, loss_ctc=51.213, loss_att=51.370, acc=0.715, loss=51.323, backward_time=0.438, grad_norm=31.424, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.633e-04, train_time=1.712
+[gpub010:0/16] 2024-01-29 11:47:44,069 (trainer:737) INFO: 15epoch:train:6501-6600batch: iter_time=8.959e-05, forward_time=0.330, loss_ctc=51.909, loss_att=50.014, acc=0.711, loss=50.582, backward_time=0.424, grad_norm=40.140, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.632e-04, train_time=1.566
+[gpub010:0/16] 2024-01-29 11:49:57,638 (trainer:737) INFO: 15epoch:train:6601-6700batch: iter_time=8.979e-05, forward_time=0.290, loss_ctc=52.659, loss_att=51.890, acc=0.712, loss=52.121, backward_time=0.402, grad_norm=32.971, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.631e-04, train_time=1.335
+[gpub010:0/16] 2024-01-29 11:52:27,277 (trainer:737) INFO: 15epoch:train:6701-6800batch: iter_time=8.952e-05, forward_time=0.320, loss_ctc=57.650, loss_att=53.232, acc=0.728, loss=54.557, backward_time=0.441, grad_norm=31.821, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.631e-04, train_time=1.496
+[gpub010:0/16] 2024-01-29 11:55:20,359 (trainer:737) INFO: 15epoch:train:6801-6900batch: iter_time=9.056e-05, forward_time=0.348, loss_ctc=56.204, loss_att=51.746, acc=0.726, loss=53.083, backward_time=0.418, grad_norm=32.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.630e-04, train_time=1.730
+[gpub010:0/16] 2024-01-29 11:57:33,401 (trainer:737) INFO: 15epoch:train:6901-7000batch: iter_time=8.673e-05, forward_time=0.292, loss_ctc=54.710, loss_att=55.470, acc=0.714, loss=55.242, backward_time=0.406, grad_norm=30.685, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.630e-04, train_time=1.330
+[gpub010:0/16] 2024-01-29 12:00:02,076 (trainer:737) INFO: 15epoch:train:7001-7100batch: iter_time=8.682e-05, forward_time=0.332, loss_ctc=55.223, loss_att=48.388, acc=0.744, loss=50.439, backward_time=0.435, grad_norm=32.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.629e-04, train_time=1.486
+[gpub010:0/16] 2024-01-29 12:02:57,879 (trainer:737) INFO: 15epoch:train:7101-7200batch: iter_time=8.686e-05, forward_time=0.331, loss_ctc=49.133, loss_att=37.621, acc=0.769, loss=41.074, backward_time=0.412, grad_norm=28.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.628e-04, train_time=1.759
+[gpub010:0/16] 2024-01-29 12:05:17,645 (trainer:737) INFO: 15epoch:train:7201-7300batch: iter_time=8.419e-05, forward_time=0.322, loss_ctc=53.195, loss_att=52.416, acc=0.720, loss=52.650, backward_time=0.406, grad_norm=31.927, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.628e-04, train_time=1.397
+[gpub010:0/16] 2024-01-29 12:08:13,316 (trainer:737) INFO: 15epoch:train:7301-7400batch: iter_time=8.769e-05, forward_time=0.384, loss_ctc=53.950, loss_att=56.048, acc=0.724, loss=55.419, backward_time=0.452, grad_norm=29.764, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.627e-04, train_time=1.756
+[gpub010:0/16] 2024-01-29 12:11:02,196 (trainer:737) INFO: 15epoch:train:7401-7500batch: iter_time=8.526e-05, forward_time=0.292, loss_ctc=59.290, loss_att=58.921, acc=0.712, loss=59.032, backward_time=0.403, grad_norm=35.828, clip=100.000, loss_scale=6.906e+33, optim_step_time=0.092, optim0_lr0=2.626e-04, train_time=1.689
+[gpub010:0/16] 2024-01-29 12:11:22,237 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-01-29 12:11:41,382 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 12:11:44,959 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae7aff3fa0>)
+[gpub010:0/16] 2024-01-29 12:11:44,959 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-29 12:11:44,963 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 12:19:35,360 (trainer:737) INFO: 15epoch:train:7501-7600batch: iter_time=3.606, forward_time=0.407, loss_ctc=49.422, loss_att=52.502, acc=0.721, loss=51.578, backward_time=0.430, grad_norm=32.020, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.626e-04, train_time=5.131
+[gpub010:0/16] 2024-01-29 12:22:02,517 (trainer:737) INFO: 15epoch:train:7601-7700batch: iter_time=8.166e-05, forward_time=0.288, loss_ctc=48.408, loss_att=41.691, acc=0.739, loss=43.706, backward_time=0.399, grad_norm=31.918, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.625e-04, train_time=1.473
+[gpub010:0/16] 2024-01-29 12:24:54,395 (trainer:737) INFO: 15epoch:train:7701-7800batch: iter_time=8.590e-05, forward_time=0.330, loss_ctc=51.282, loss_att=53.626, acc=0.725, loss=52.923, backward_time=0.446, grad_norm=31.146, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.625e-04, train_time=1.719
+[gpub010:0/16] 2024-01-29 12:27:24,683 (trainer:737) INFO: 15epoch:train:7801-7900batch: iter_time=8.650e-05, forward_time=0.318, loss_ctc=51.106, loss_att=52.990, acc=0.711, loss=52.425, backward_time=0.430, grad_norm=38.309, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.624e-04, train_time=1.502
+[gpub010:0/16] 2024-01-29 12:29:53,805 (trainer:737) INFO: 15epoch:train:7901-8000batch: iter_time=8.440e-05, forward_time=0.313, loss_ctc=59.330, loss_att=57.276, acc=0.725, loss=57.892, backward_time=0.404, grad_norm=33.734, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.623e-04, train_time=1.492
+[gpub010:0/16] 2024-01-29 12:32:37,035 (trainer:737) INFO: 15epoch:train:8001-8100batch: iter_time=8.622e-05, forward_time=0.383, loss_ctc=50.891, loss_att=47.252, acc=0.745, loss=48.344, backward_time=0.440, grad_norm=29.970, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.623e-04, train_time=1.632
+[gpub010:0/16] 2024-01-29 12:35:30,299 (trainer:737) INFO: 15epoch:train:8101-8200batch: iter_time=8.443e-05, forward_time=0.293, loss_ctc=58.618, loss_att=62.710, acc=0.712, loss=61.482, backward_time=0.406, grad_norm=35.274, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.622e-04, train_time=1.732
+[gpub010:0/16] 2024-01-29 12:37:58,289 (trainer:737) INFO: 15epoch:train:8201-8300batch: iter_time=8.314e-05, forward_time=0.359, loss_ctc=54.506, loss_att=49.828, acc=0.731, loss=51.232, backward_time=0.417, grad_norm=32.522, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.622e-04, train_time=1.480
+[gpub010:0/16] 2024-01-29 12:39:48,661 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 12:40:29,238 (trainer:737) INFO: 15epoch:train:8301-8400batch: iter_time=8.201e-05, forward_time=0.370, loss_ctc=53.388, loss_att=46.118, acc=0.770, loss=48.299, backward_time=0.421, grad_norm=29.295, clip=100.000, loss_scale=8.811e+33, optim_step_time=0.094, optim0_lr0=2.621e-04, train_time=1.510
+[gpub010:0/16] 2024-01-29 12:43:13,517 (trainer:737) INFO: 15epoch:train:8401-8500batch: iter_time=8.631e-05, forward_time=0.290, loss_ctc=52.471, loss_att=46.320, acc=0.746, loss=48.165, backward_time=0.402, grad_norm=30.939, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.620e-04, train_time=1.642
+[gpub010:0/16] 2024-01-29 12:46:05,465 (trainer:737) INFO: 15epoch:train:8501-8600batch: iter_time=6.578e-04, forward_time=0.409, loss_ctc=51.087, loss_att=52.756, acc=0.727, loss=52.255, backward_time=0.437, grad_norm=29.531, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.620e-04, train_time=1.720
+[gpub010:0/16] 2024-01-29 12:48:25,807 (trainer:737) INFO: 15epoch:train:8601-8700batch: iter_time=9.042e-05, forward_time=0.330, loss_ctc=53.698, loss_att=56.417, acc=0.727, loss=55.601, backward_time=0.409, grad_norm=31.229, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.619e-04, train_time=1.404
+[gpub010:0/16] 2024-01-29 12:50:28,479 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-01-29 12:50:47,822 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 12:50:51,479 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb120defd60>)
+[gpub010:0/16] 2024-01-29 12:50:51,479 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-01-29 12:50:51,482 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 12:57:08,037 (trainer:737) INFO: 15epoch:train:8701-8800batch: iter_time=3.371, forward_time=0.371, loss_ctc=54.111, loss_att=54.988, acc=0.736, loss=54.725, backward_time=0.413, grad_norm=33.418, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.619e-04, train_time=5.221
+[gpub010:0/16] 2024-01-29 12:59:28,733 (trainer:737) INFO: 15epoch:train:8801-8900batch: iter_time=8.122e-05, forward_time=0.289, loss_ctc=49.874, loss_att=43.697, acc=0.728, loss=45.550, backward_time=0.402, grad_norm=32.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.618e-04, train_time=1.408
+[gpub010:0/16] 2024-01-29 13:02:14,371 (trainer:737) INFO: 15epoch:train:8901-9000batch: iter_time=8.476e-05, forward_time=0.294, loss_ctc=51.317, loss_att=51.933, acc=0.729, loss=51.748, backward_time=0.403, grad_norm=31.517, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.617e-04, train_time=1.656
+[gpub010:0/16] 2024-01-29 13:04:45,633 (trainer:737) INFO: 15epoch:train:9001-9100batch: iter_time=8.804e-05, forward_time=0.367, loss_ctc=51.889, loss_att=50.655, acc=0.717, loss=51.025, backward_time=0.449, grad_norm=38.551, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.617e-04, train_time=1.512
+[gpub010:0/16] 2024-01-29 13:09:22,579 (trainer:737) INFO: 15epoch:train:9101-9200batch: iter_time=0.093, forward_time=0.289, loss_ctc=52.721, loss_att=52.996, acc=0.721, loss=52.913, backward_time=0.402, grad_norm=35.140, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.616e-04, train_time=2.769
+[gpub010:0/16] 2024-01-29 13:12:18,996 (trainer:737) INFO: 15epoch:train:9201-9300batch: iter_time=8.763e-05, forward_time=0.400, loss_ctc=56.660, loss_att=53.305, acc=0.738, loss=54.311, backward_time=0.437, grad_norm=30.989, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.616e-04, train_time=1.764
+[gpub010:0/16] 2024-01-29 13:15:01,238 (trainer:737) INFO: 15epoch:train:9301-9400batch: iter_time=8.105e-05, forward_time=0.292, loss_ctc=56.220, loss_att=54.807, acc=0.725, loss=55.231, backward_time=0.404, grad_norm=32.055, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.615e-04, train_time=1.621
+[gpub010:0/16] 2024-01-29 13:17:27,473 (trainer:737) INFO: 15epoch:train:9401-9500batch: iter_time=8.556e-05, forward_time=0.293, loss_ctc=55.047, loss_att=55.845, acc=0.725, loss=55.606, backward_time=0.406, grad_norm=30.155, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.614e-04, train_time=1.463
+[gpub010:0/16] 2024-01-29 13:20:37,156 (trainer:737) INFO: 15epoch:train:9501-9600batch: iter_time=4.592e-04, forward_time=0.459, loss_ctc=54.435, loss_att=49.019, acc=0.751, loss=50.644, backward_time=0.438, grad_norm=31.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.614e-04, train_time=1.896
+[gpub010:0/16] 2024-01-29 13:23:07,525 (trainer:737) INFO: 15epoch:train:9601-9700batch: iter_time=8.049e-05, forward_time=0.291, loss_ctc=49.226, loss_att=38.675, acc=0.773, loss=41.840, backward_time=0.402, grad_norm=26.188, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.613e-04, train_time=1.504
+[gpub010:0/16] 2024-01-29 13:25:56,990 (trainer:737) INFO: 15epoch:train:9701-9800batch: iter_time=3.040e-04, forward_time=0.360, loss_ctc=52.202, loss_att=51.554, acc=0.731, loss=51.748, backward_time=0.479, grad_norm=31.616, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.613e-04, train_time=1.695
+[gpub010:0/16] 2024-01-29 13:28:50,572 (trainer:737) INFO: 15epoch:train:9801-9900batch: iter_time=8.247e-05, forward_time=0.296, loss_ctc=53.709, loss_att=57.051, acc=0.731, loss=56.048, backward_time=0.405, grad_norm=30.182, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.612e-04, train_time=1.735
+[gpub010:0/16] 2024-01-29 13:31:14,208 (trainer:737) INFO: 15epoch:train:9901-10000batch: iter_time=8.179e-05, forward_time=0.293, loss_ctc=59.427, loss_att=59.480, acc=0.719, loss=59.464, backward_time=0.406, grad_norm=35.147, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.612e-04, train_time=1.437
+[gpub010:0/16] 2024-01-29 13:31:34,396 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-01-29 13:31:53,544 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 13:31:57,162 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae7af9d810>)
+[gpub010:0/16] 2024-01-29 13:31:57,162 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-01-29 13:31:57,165 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 13:40:28,693 (trainer:737) INFO: 15epoch:train:10001-10100batch: iter_time=3.986, forward_time=0.421, loss_ctc=49.490, loss_att=48.371, acc=0.729, loss=48.706, backward_time=0.423, grad_norm=30.268, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.611e-04, train_time=5.545
+[gpub010:0/16] 2024-01-29 13:43:13,052 (trainer:737) INFO: 15epoch:train:10101-10200batch: iter_time=8.139e-05, forward_time=0.291, loss_ctc=48.379, loss_att=40.382, acc=0.744, loss=42.781, backward_time=0.398, grad_norm=34.670, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.610e-04, train_time=1.643
+[gpub010:0/16] 2024-01-29 13:46:09,626 (trainer:737) INFO: 15epoch:train:10201-10300batch: iter_time=8.325e-05, forward_time=0.388, loss_ctc=51.254, loss_att=52.194, acc=0.728, loss=51.912, backward_time=0.445, grad_norm=30.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.610e-04, train_time=1.766
+[gpub010:0/16] 2024-01-29 13:48:21,643 (trainer:737) INFO: 15epoch:train:10301-10400batch: iter_time=8.620e-05, forward_time=0.290, loss_ctc=50.679, loss_att=52.805, acc=0.713, loss=52.167, backward_time=0.399, grad_norm=35.898, clip=100.000, loss_scale=6.750e+33, optim_step_time=0.092, optim0_lr0=2.609e-04, train_time=1.320
+[gpub010:0/16] 2024-01-29 13:51:06,840 (trainer:737) INFO: 15epoch:train:10401-10500batch: iter_time=8.790e-05, forward_time=0.293, loss_ctc=59.737, loss_att=56.416, acc=0.728, loss=57.412, backward_time=0.405, grad_norm=33.777, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.609e-04, train_time=1.650
+[gpub010:0/16] 2024-01-29 13:51:37,679 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 13:54:22,503 (trainer:737) INFO: 15epoch:train:10501-10600batch: iter_time=2.191e-04, forward_time=0.402, loss_ctc=50.680, loss_att=46.321, acc=0.746, loss=47.629, backward_time=0.464, grad_norm=30.798, clip=100.000, loss_scale=6.031e+33, optim_step_time=0.101, optim0_lr0=2.608e-04, train_time=1.958
+[gpub010:0/16] 2024-01-29 13:56:37,541 (trainer:737) INFO: 15epoch:train:10601-10700batch: iter_time=8.468e-05, forward_time=0.295, loss_ctc=58.172, loss_att=61.706, acc=0.714, loss=60.646, backward_time=0.408, grad_norm=34.958, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.607e-04, train_time=1.351
+[gpub010:0/16] 2024-01-29 13:59:30,969 (trainer:737) INFO: 15epoch:train:10701-10800batch: iter_time=8.979e-05, forward_time=0.389, loss_ctc=54.257, loss_att=49.328, acc=0.733, loss=50.807, backward_time=0.451, grad_norm=32.258, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.607e-04, train_time=1.734
+[gpub010:0/16] 2024-01-29 14:02:17,067 (trainer:737) INFO: 15epoch:train:10801-10900batch: iter_time=8.230e-05, forward_time=0.293, loss_ctc=52.612, loss_att=45.550, acc=0.772, loss=47.669, backward_time=0.405, grad_norm=28.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.606e-04, train_time=1.660
+[gpub010:0/16] 2024-01-29 14:04:42,039 (trainer:737) INFO: 15epoch:train:10901-11000batch: iter_time=8.531e-05, forward_time=0.292, loss_ctc=51.972, loss_att=45.854, acc=0.748, loss=47.689, backward_time=0.404, grad_norm=29.946, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.606e-04, train_time=1.450
+[gpub010:0/16] 2024-01-29 14:08:08,965 (trainer:737) INFO: 15epoch:train:11001-11100batch: iter_time=5.735e-04, forward_time=0.476, loss_ctc=50.318, loss_att=52.137, acc=0.731, loss=51.591, backward_time=0.444, grad_norm=28.915, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.116, optim0_lr0=2.605e-04, train_time=2.069
+[gpub010:0/16] 2024-01-29 14:10:21,656 (trainer:737) INFO: 15epoch:train:11101-11200batch: iter_time=8.806e-05, forward_time=0.294, loss_ctc=53.677, loss_att=55.724, acc=0.729, loss=55.110, backward_time=0.408, grad_norm=30.443, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.604e-04, train_time=1.326
+[gpub010:0/16] 2024-01-29 14:11:55,562 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-01-29 14:12:14,926 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 14:12:18,505 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae8704f190>)
+[gpub010:0/16] 2024-01-29 14:12:18,505 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-01-29 14:12:18,509 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 14:19:24,518 (trainer:737) INFO: 15epoch:train:11201-11300batch: iter_time=3.751, forward_time=0.403, loss_ctc=53.865, loss_att=57.807, acc=0.718, loss=56.624, backward_time=0.420, grad_norm=35.476, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.604e-04, train_time=5.429
+[gpub010:0/16] 2024-01-29 14:21:35,995 (trainer:737) INFO: 15epoch:train:11301-11400batch: iter_time=8.618e-05, forward_time=0.289, loss_ctc=49.825, loss_att=44.798, acc=0.719, loss=46.306, backward_time=0.400, grad_norm=32.754, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.603e-04, train_time=1.315
+[gpub010:0/16] 2024-01-29 14:23:49,791 (trainer:737) INFO: 15epoch:train:11401-11500batch: iter_time=7.744e-05, forward_time=0.291, loss_ctc=50.485, loss_att=52.760, acc=0.714, loss=52.078, backward_time=0.402, grad_norm=32.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.603e-04, train_time=1.337
+[gpub010:0/16] 2024-01-29 14:27:33,441 (trainer:737) INFO: 15epoch:train:11501-11600batch: iter_time=9.481e-05, forward_time=0.393, loss_ctc=50.639, loss_att=49.916, acc=0.714, loss=50.133, backward_time=0.435, grad_norm=36.800, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.602e-04, train_time=2.237
+[gpub010:0/16] 2024-01-29 14:30:09,372 (trainer:737) INFO: 15epoch:train:11601-11700batch: iter_time=8.297e-05, forward_time=0.291, loss_ctc=52.271, loss_att=52.059, acc=0.717, loss=52.122, backward_time=0.401, grad_norm=33.811, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.601e-04, train_time=1.559
+[gpub010:0/16] 2024-01-29 14:33:22,042 (trainer:737) INFO: 15epoch:train:11701-11800batch: iter_time=8.445e-05, forward_time=0.436, loss_ctc=57.245, loss_att=52.933, acc=0.731, loss=54.227, backward_time=0.435, grad_norm=34.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.601e-04, train_time=1.926
+[gpub010:0/16] 2024-01-29 14:36:08,024 (trainer:737) INFO: 15epoch:train:11801-11900batch: iter_time=8.592e-05, forward_time=0.290, loss_ctc=55.793, loss_att=52.168, acc=0.726, loss=53.255, backward_time=0.404, grad_norm=34.009, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.600e-04, train_time=1.660
+[gpub010:0/16] 2024-01-29 14:38:52,635 (trainer:737) INFO: 15epoch:train:11901-12000batch: iter_time=8.818e-05, forward_time=0.345, loss_ctc=54.354, loss_att=55.230, acc=0.716, loss=54.967, backward_time=0.435, grad_norm=29.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.600e-04, train_time=1.646
+[gpub010:0/16] 2024-01-29 14:41:47,463 (trainer:737) INFO: 15epoch:train:12001-12100batch: iter_time=9.068e-05, forward_time=0.328, loss_ctc=54.204, loss_att=48.442, acc=0.744, loss=50.171, backward_time=0.420, grad_norm=32.392, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.599e-04, train_time=1.748
+[gpub010:0/16] 2024-01-29 14:44:38,133 (trainer:737) INFO: 15epoch:train:12101-12200batch: iter_time=8.562e-05, forward_time=0.291, loss_ctc=48.349, loss_att=37.310, acc=0.772, loss=40.621, backward_time=0.400, grad_norm=26.231, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.599e-04, train_time=1.707
+[gpub010:0/16] 2024-01-29 14:47:51,125 (trainer:737) INFO: 15epoch:train:12201-12300batch: iter_time=4.173e-04, forward_time=0.376, loss_ctc=52.121, loss_att=51.184, acc=0.726, loss=51.465, backward_time=0.473, grad_norm=32.047, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.598e-04, train_time=1.929
+[gpub010:0/16] 2024-01-29 14:50:22,860 (trainer:737) INFO: 15epoch:train:12301-12400batch: iter_time=8.540e-05, forward_time=0.292, loss_ctc=52.808, loss_att=56.396, acc=0.723, loss=55.319, backward_time=0.402, grad_norm=31.254, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.597e-04, train_time=1.517
+[gpub010:0/16] 2024-01-29 14:53:13,653 (trainer:737) INFO: 15epoch:train:12401-12500batch: iter_time=9.690e-05, forward_time=0.304, loss_ctc=58.703, loss_att=58.944, acc=0.713, loss=58.872, backward_time=0.417, grad_norm=37.899, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.597e-04, train_time=1.707
+[gpub010:0/16] 2024-01-29 14:53:33,802 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-01-29 14:53:53,286 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 14:53:56,901 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae5ed52d40>)
+[gpub010:0/16] 2024-01-29 14:53:56,902 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-01-29 14:53:56,905 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 15:01:27,572 (trainer:737) INFO: 15epoch:train:12501-12600batch: iter_time=3.139, forward_time=0.449, loss_ctc=48.689, loss_att=48.809, acc=0.711, loss=48.773, backward_time=0.430, grad_norm=31.028, clip=100.000, loss_scale=9.502e+33, optim_step_time=0.097, optim0_lr0=2.596e-04, train_time=4.940
+[gpub010:0/16] 2024-01-29 15:04:23,839 (trainer:737) INFO: 15epoch:train:12601-12700batch: iter_time=8.859e-05, forward_time=0.395, loss_ctc=47.512, loss_att=40.086, acc=0.734, loss=42.314, backward_time=0.456, grad_norm=34.827, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.596e-04, train_time=1.763
+[gpub010:0/16] 2024-01-29 15:07:13,592 (trainer:737) INFO: 15epoch:train:12701-12800batch: iter_time=9.020e-05, forward_time=0.290, loss_ctc=51.129, loss_att=52.715, acc=0.715, loss=52.240, backward_time=0.401, grad_norm=31.167, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.595e-04, train_time=1.697
+[gpub010:0/16] 2024-01-29 15:10:14,084 (trainer:737) INFO: 15epoch:train:12801-12900batch: iter_time=9.034e-05, forward_time=0.388, loss_ctc=50.066, loss_att=52.229, acc=0.713, loss=51.580, backward_time=0.492, grad_norm=37.619, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=2.594e-04, train_time=1.805
+[gpub010:0/16] 2024-01-29 15:12:09,633 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 15:12:48,530 (trainer:737) INFO: 15epoch:train:12901-13000batch: iter_time=9.407e-05, forward_time=0.291, loss_ctc=58.887, loss_att=56.673, acc=0.715, loss=57.337, backward_time=0.402, grad_norm=32.504, clip=100.000, loss_scale=9.283e+33, optim_step_time=0.092, optim0_lr0=2.594e-04, train_time=1.544
+[gpub010:0/16] 2024-01-29 15:16:16,790 (trainer:737) INFO: 15epoch:train:13001-13100batch: iter_time=9.826e-05, forward_time=0.416, loss_ctc=50.255, loss_att=45.960, acc=0.747, loss=47.249, backward_time=0.444, grad_norm=31.062, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.121, optim0_lr0=2.593e-04, train_time=2.083
+[gpub010:0/16] 2024-01-29 15:18:51,743 (trainer:737) INFO: 15epoch:train:13101-13200batch: iter_time=9.718e-05, forward_time=0.294, loss_ctc=58.447, loss_att=59.181, acc=0.708, loss=58.961, backward_time=0.406, grad_norm=34.359, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.593e-04, train_time=1.549
+[gpub010:0/16] 2024-01-29 15:22:25,592 (trainer:737) INFO: 15epoch:train:13201-13300batch: iter_time=9.715e-05, forward_time=0.469, loss_ctc=53.663, loss_att=47.475, acc=0.730, loss=49.332, backward_time=0.444, grad_norm=31.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.592e-04, train_time=2.138
+[gpub010:0/16] 2024-01-29 15:25:49,856 (trainer:737) INFO: 15epoch:train:13301-13400batch: iter_time=1.006e-04, forward_time=0.447, loss_ctc=52.691, loss_att=43.945, acc=0.774, loss=46.569, backward_time=0.476, grad_norm=28.070, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.592e-04, train_time=2.041
+[gpub010:0/16] 2024-01-29 15:28:15,930 (trainer:737) INFO: 15epoch:train:13401-13500batch: iter_time=8.576e-05, forward_time=0.291, loss_ctc=51.237, loss_att=45.936, acc=0.742, loss=47.526, backward_time=0.403, grad_norm=32.010, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.591e-04, train_time=1.461
+[gpub010:0/16] 2024-01-29 15:31:16,682 (trainer:737) INFO: 15epoch:train:13501-13600batch: iter_time=2.008e-04, forward_time=0.372, loss_ctc=50.405, loss_att=52.119, acc=0.722, loss=51.605, backward_time=0.453, grad_norm=29.699, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.590e-04, train_time=1.807
+[gpub010:0/16] 2024-01-29 15:34:20,660 (trainer:737) INFO: 15epoch:train:13601-13700batch: iter_time=1.016e-04, forward_time=0.291, loss_ctc=53.407, loss_att=55.148, acc=0.721, loss=54.625, backward_time=0.402, grad_norm=30.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.590e-04, train_time=1.840
+[gpub010:0/16] 2024-01-29 15:35:46,239 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-01-29 15:36:05,859 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 15:36:09,531 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae5bbed510>)
+[gpub010:0/16] 2024-01-29 15:36:09,531 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-01-29 15:36:09,535 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 15:43:01,852 (trainer:737) INFO: 15epoch:train:13701-13800batch: iter_time=3.531, forward_time=0.363, loss_ctc=53.925, loss_att=57.277, acc=0.730, loss=56.272, backward_time=0.417, grad_norm=36.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.589e-04, train_time=5.212
+[gpub010:0/16] 2024-01-29 15:45:43,100 (trainer:737) INFO: 15epoch:train:13801-13900batch: iter_time=8.605e-05, forward_time=0.287, loss_ctc=49.175, loss_att=44.422, acc=0.728, loss=45.848, backward_time=0.399, grad_norm=33.821, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.589e-04, train_time=1.612
+[gpub010:0/16] 2024-01-29 15:48:30,028 (trainer:737) INFO: 15epoch:train:13901-14000batch: iter_time=8.286e-05, forward_time=0.421, loss_ctc=51.187, loss_att=52.816, acc=0.729, loss=52.327, backward_time=0.446, grad_norm=31.168, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.588e-04, train_time=1.669
+[gpub010:0/16] 2024-01-29 15:51:21,308 (trainer:737) INFO: 15epoch:train:14001-14100batch: iter_time=8.446e-05, forward_time=0.289, loss_ctc=50.263, loss_att=49.312, acc=0.720, loss=49.597, backward_time=0.400, grad_norm=36.224, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.588e-04, train_time=1.712
+[gpub010:0/16] 2024-01-29 15:54:24,061 (trainer:737) INFO: 15epoch:train:14101-14200batch: iter_time=8.911e-05, forward_time=0.439, loss_ctc=51.486, loss_att=52.235, acc=0.725, loss=52.010, backward_time=0.439, grad_norm=32.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.117, optim0_lr0=2.587e-04, train_time=1.826
+[gpub010:0/16] 2024-01-29 15:56:55,778 (trainer:737) INFO: 15epoch:train:14201-14300batch: iter_time=9.120e-05, forward_time=0.292, loss_ctc=55.862, loss_att=53.308, acc=0.738, loss=54.075, backward_time=0.403, grad_norm=32.542, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.586e-04, train_time=1.518
+[gpub010:0/16] 2024-01-29 16:00:19,596 (trainer:737) INFO: 15epoch:train:14301-14400batch: iter_time=8.931e-05, forward_time=0.481, loss_ctc=55.092, loss_att=54.110, acc=0.727, loss=54.405, backward_time=0.436, grad_norm=33.855, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.586e-04, train_time=2.037
+[gpub010:0/16] 2024-01-29 16:02:55,557 (trainer:737) INFO: 15epoch:train:14401-14500batch: iter_time=9.292e-05, forward_time=0.294, loss_ctc=53.659, loss_att=54.739, acc=0.728, loss=54.415, backward_time=0.406, grad_norm=29.723, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.585e-04, train_time=1.560
+[gpub010:0/16] 2024-01-29 16:05:57,883 (trainer:737) INFO: 15epoch:train:14501-14600batch: iter_time=9.434e-05, forward_time=0.441, loss_ctc=53.374, loss_att=48.567, acc=0.753, loss=50.009, backward_time=0.431, grad_norm=32.232, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.585e-04, train_time=1.822
+[gpub010:0/16] 2024-01-29 16:09:11,179 (trainer:737) INFO: 15epoch:train:14601-14700batch: iter_time=8.968e-05, forward_time=0.471, loss_ctc=48.752, loss_att=38.493, acc=0.775, loss=41.571, backward_time=0.430, grad_norm=27.370, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.584e-04, train_time=1.933
+[gpub010:0/16] 2024-01-29 16:11:29,211 (trainer:737) INFO: 15epoch:train:14701-14800batch: iter_time=9.207e-05, forward_time=0.291, loss_ctc=51.392, loss_att=51.216, acc=0.732, loss=51.269, backward_time=0.406, grad_norm=31.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.583e-04, train_time=1.379
+[gpub010:0/16] 2024-01-29 16:15:00,233 (trainer:737) INFO: 15epoch:train:14801-14900batch: iter_time=9.204e-05, forward_time=0.438, loss_ctc=52.760, loss_att=56.815, acc=0.732, loss=55.599, backward_time=0.430, grad_norm=30.846, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.116, optim0_lr0=2.583e-04, train_time=2.111
+[gpub010:0/16] 2024-01-29 16:17:44,626 (trainer:737) INFO: 15epoch:train:14901-15000batch: iter_time=8.669e-05, forward_time=0.291, loss_ctc=57.904, loss_att=58.602, acc=0.722, loss=58.392, backward_time=0.405, grad_norm=41.731, clip=100.000, loss_scale=6.283e+33, optim_step_time=0.092, optim0_lr0=2.582e-04, train_time=1.643
+[gpub010:0/16] 2024-01-29 16:57:12,750 (trainer:343) INFO: 15epoch results: [train] iter_time=0.275, forward_time=0.337, loss_ctc=53.778, loss_att=51.619, acc=0.727, loss=52.267, backward_time=0.420, grad_norm=32.798, clip=100.000, loss_scale=5.030e+33, optim_step_time=0.096, optim0_lr0=2.627e-04, train_time=1.912, time=7 hours, 58 minutes and 24.71 seconds, total_count=255000, gpu_max_cached_mem_GB=42.072, [valid] loss_ctc=41.359, cer_ctc=0.217, loss_att=44.311, acc=0.641, cer=0.329, wer=0.998, loss=43.425, time=39 minutes and 3.54 seconds, total_count=79407, gpu_max_cached_mem_GB=42.072
+[gpub010:0/16] 2024-01-29 16:57:29,466 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-01-29 16:57:29,653 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/10epoch.pth
+[gpub010:0/16] 2024-01-29 16:57:29,874 (trainer:272) INFO: 16/45epoch started. Estimated time to finish: 1 week, 3 days and 18 hours
+[gpub010:0/16] 2024-01-29 16:57:30,411 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-29 16:57:49,060 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 16:57:52,558 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa78b131f60>)
+[gpub010:0/16] 2024-01-29 16:57:52,558 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-01-29 16:57:52,561 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 17:10:30,162 (trainer:737) INFO: 16epoch:train:1-100batch: iter_time=3.592, forward_time=0.360, loss_ctc=49.086, loss_att=45.905, acc=0.733, loss=46.859, backward_time=0.411, grad_norm=29.922, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.582e-04, train_time=7.800
+[gpub010:0/16] 2024-01-29 17:12:44,670 (trainer:737) INFO: 16epoch:train:101-200batch: iter_time=8.275e-05, forward_time=0.291, loss_ctc=57.996, loss_att=55.841, acc=0.713, loss=56.487, backward_time=0.403, grad_norm=42.880, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.581e-04, train_time=1.345
+[gpub010:0/16] 2024-01-29 17:15:19,735 (trainer:737) INFO: 16epoch:train:201-300batch: iter_time=2.335e-04, forward_time=0.295, loss_ctc=58.848, loss_att=55.518, acc=0.711, loss=56.517, backward_time=0.407, grad_norm=42.403, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.581e-04, train_time=1.550
+[gpub010:0/16] 2024-01-29 17:18:32,282 (trainer:737) INFO: 16epoch:train:301-400batch: iter_time=2.286e-04, forward_time=0.400, loss_ctc=48.977, loss_att=51.671, acc=0.728, loss=50.863, backward_time=0.448, grad_norm=31.162, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.113, optim0_lr0=2.580e-04, train_time=1.926
+[gpub010:0/16] 2024-01-29 17:20:49,132 (trainer:737) INFO: 16epoch:train:401-500batch: iter_time=8.374e-05, forward_time=0.293, loss_ctc=57.241, loss_att=56.687, acc=0.735, loss=56.853, backward_time=0.407, grad_norm=34.400, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.579e-04, train_time=1.368
+[gpub010:0/16] 2024-01-29 17:23:24,432 (trainer:737) INFO: 16epoch:train:501-600batch: iter_time=5.541e-04, forward_time=0.319, loss_ctc=47.604, loss_att=48.391, acc=0.737, loss=48.155, backward_time=0.413, grad_norm=29.347, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.579e-04, train_time=1.552
+[gpub010:0/16] 2024-01-29 17:26:04,081 (trainer:737) INFO: 16epoch:train:601-700batch: iter_time=8.448e-05, forward_time=0.408, loss_ctc=56.093, loss_att=54.736, acc=0.738, loss=55.143, backward_time=0.449, grad_norm=32.338, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.578e-04, train_time=1.597
+[gpub010:0/16] 2024-01-29 17:28:52,702 (trainer:737) INFO: 16epoch:train:701-800batch: iter_time=9.548e-05, forward_time=0.289, loss_ctc=50.803, loss_att=46.168, acc=0.730, loss=47.558, backward_time=0.401, grad_norm=33.644, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.578e-04, train_time=1.686
+[gpub010:0/16] 2024-01-29 17:31:36,105 (trainer:737) INFO: 16epoch:train:801-900batch: iter_time=7.407e-04, forward_time=0.372, loss_ctc=48.795, loss_att=46.036, acc=0.727, loss=46.864, backward_time=0.417, grad_norm=32.136, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.118, optim0_lr0=2.577e-04, train_time=1.633
+[gpub010:0/16] 2024-01-29 17:34:02,237 (trainer:737) INFO: 16epoch:train:901-1000batch: iter_time=8.307e-05, forward_time=0.330, loss_ctc=47.238, loss_att=45.668, acc=0.722, loss=46.139, backward_time=0.412, grad_norm=31.657, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.577e-04, train_time=1.462
+[gpub010:0/16] 2024-01-29 17:36:27,244 (trainer:737) INFO: 16epoch:train:1001-1100batch: iter_time=8.825e-05, forward_time=0.290, loss_ctc=51.191, loss_att=46.191, acc=0.744, loss=47.691, backward_time=0.405, grad_norm=29.476, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.576e-04, train_time=1.450
+[gpub010:0/16] 2024-01-29 17:39:35,055 (trainer:737) INFO: 16epoch:train:1101-1200batch: iter_time=0.002, forward_time=0.300, loss_ctc=55.971, loss_att=50.606, acc=0.733, loss=52.216, backward_time=0.413, grad_norm=33.161, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.575e-04, train_time=1.878
+[gpub010:0/16] 2024-01-29 17:41:15,293 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-01-29 17:41:34,218 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 17:41:37,861 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa716015f60>)
+[gpub010:0/16] 2024-01-29 17:41:37,862 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-01-29 17:41:37,865 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 17:48:15,206 (trainer:737) INFO: 16epoch:train:1201-1300batch: iter_time=3.620, forward_time=0.340, loss_ctc=57.363, loss_att=57.998, acc=0.723, loss=57.807, backward_time=0.477, grad_norm=41.831, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.575e-04, train_time=5.201
+[gpub010:0/16] 2024-01-29 17:50:31,653 (trainer:737) INFO: 16epoch:train:1301-1400batch: iter_time=1.317e-04, forward_time=0.294, loss_ctc=55.478, loss_att=51.447, acc=0.726, loss=52.657, backward_time=0.400, grad_norm=38.997, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.574e-04, train_time=1.364
+[gpub010:0/16] 2024-01-29 17:53:26,684 (trainer:737) INFO: 16epoch:train:1401-1500batch: iter_time=3.461e-04, forward_time=0.400, loss_ctc=50.544, loss_att=51.428, acc=0.714, loss=51.163, backward_time=0.461, grad_norm=36.767, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.574e-04, train_time=1.749
+[gpub010:0/16] 2024-01-29 17:56:05,992 (trainer:737) INFO: 16epoch:train:1501-1600batch: iter_time=9.148e-05, forward_time=0.293, loss_ctc=54.820, loss_att=55.069, acc=0.736, loss=54.994, backward_time=0.405, grad_norm=38.639, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.573e-04, train_time=1.593
+[gpub010:0/16] 2024-01-29 17:58:28,278 (trainer:737) INFO: 16epoch:train:1601-1700batch: iter_time=4.694e-04, forward_time=0.299, loss_ctc=52.289, loss_att=52.865, acc=0.737, loss=52.692, backward_time=0.408, grad_norm=33.902, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.573e-04, train_time=1.424
+[gpub010:0/16] 2024-01-29 18:01:19,957 (trainer:737) INFO: 16epoch:train:1701-1800batch: iter_time=8.684e-05, forward_time=0.366, loss_ctc=55.624, loss_att=59.013, acc=0.740, loss=57.996, backward_time=0.468, grad_norm=31.900, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.572e-04, train_time=1.715
+[gpub010:0/16] 2024-01-29 18:03:44,801 (trainer:737) INFO: 16epoch:train:1801-1900batch: iter_time=2.302e-04, forward_time=0.300, loss_ctc=51.542, loss_att=52.377, acc=0.742, loss=52.127, backward_time=0.408, grad_norm=31.244, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.571e-04, train_time=1.448
+[gpub010:0/16] 2024-01-29 18:06:22,172 (trainer:737) INFO: 16epoch:train:1901-2000batch: iter_time=9.707e-05, forward_time=0.300, loss_ctc=50.786, loss_att=50.192, acc=0.746, loss=50.370, backward_time=0.415, grad_norm=31.752, clip=100.000, loss_scale=1.257e+34, optim_step_time=0.092, optim0_lr0=2.571e-04, train_time=1.575
+[gpub010:0/16] 2024-01-29 18:09:06,539 (trainer:737) INFO: 16epoch:train:2001-2100batch: iter_time=9.194e-05, forward_time=0.367, loss_ctc=46.842, loss_att=42.770, acc=0.742, loss=43.991, backward_time=0.439, grad_norm=29.627, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.100, optim0_lr0=2.570e-04, train_time=1.643
+[gpub010:0/16] 2024-01-29 18:09:34,750 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 18:11:26,279 (trainer:737) INFO: 16epoch:train:2101-2200batch: iter_time=9.950e-05, forward_time=0.299, loss_ctc=49.162, loss_att=50.034, acc=0.724, loss=49.772, backward_time=0.403, grad_norm=32.569, clip=100.000, loss_scale=1.238e+34, optim_step_time=0.092, optim0_lr0=2.570e-04, train_time=1.398
+[gpub010:0/16] 2024-01-29 18:13:57,476 (trainer:737) INFO: 16epoch:train:2201-2300batch: iter_time=9.376e-05, forward_time=0.296, loss_ctc=50.123, loss_att=45.654, acc=0.752, loss=46.995, backward_time=0.411, grad_norm=29.606, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.569e-04, train_time=1.511
+[gpub010:0/16] 2024-01-29 18:17:11,437 (trainer:737) INFO: 16epoch:train:2301-2400batch: iter_time=8.238e-05, forward_time=0.411, loss_ctc=53.430, loss_att=50.295, acc=0.743, loss=51.235, backward_time=0.458, grad_norm=33.380, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.569e-04, train_time=1.939
+[gpub010:0/16] 2024-01-29 18:19:33,722 (trainer:737) INFO: 16epoch:train:2401-2500batch: iter_time=9.199e-04, forward_time=0.298, loss_ctc=58.438, loss_att=52.310, acc=0.734, loss=54.148, backward_time=0.405, grad_norm=39.413, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.568e-04, train_time=1.423
+[gpub010:0/16] 2024-01-29 18:19:53,780 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-01-29 18:20:12,545 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 18:20:16,389 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae54b5c550>)
+[gpub010:0/16] 2024-01-29 18:20:16,389 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-01-29 18:20:16,392 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 18:27:21,397 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 18:28:05,473 (trainer:737) INFO: 16epoch:train:2501-2600batch: iter_time=3.451, forward_time=0.351, loss_ctc=48.108, loss_att=44.935, acc=0.745, loss=45.887, backward_time=0.459, grad_norm=28.544, clip=100.000, loss_scale=9.178e+33, optim_step_time=0.102, optim0_lr0=2.568e-04, train_time=5.117
+[gpub010:0/16] 2024-01-29 18:30:43,572 (trainer:737) INFO: 16epoch:train:2601-2700batch: iter_time=8.533e-05, forward_time=0.325, loss_ctc=54.781, loss_att=56.214, acc=0.715, loss=55.784, backward_time=0.406, grad_norm=42.474, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.567e-04, train_time=1.580
+[gpub010:0/16] 2024-01-29 18:33:05,350 (trainer:737) INFO: 16epoch:train:2701-2800batch: iter_time=8.414e-05, forward_time=0.297, loss_ctc=55.926, loss_att=55.130, acc=0.715, loss=55.369, backward_time=0.410, grad_norm=41.404, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.566e-04, train_time=1.418
+[gpub010:0/16] 2024-01-29 18:35:52,865 (trainer:737) INFO: 16epoch:train:2801-2900batch: iter_time=8.189e-05, forward_time=0.415, loss_ctc=48.496, loss_att=52.274, acc=0.734, loss=51.141, backward_time=0.437, grad_norm=30.830, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.566e-04, train_time=1.675
+[gpub010:0/16] 2024-01-29 18:38:24,764 (trainer:737) INFO: 16epoch:train:2901-3000batch: iter_time=8.425e-05, forward_time=0.294, loss_ctc=56.327, loss_att=55.973, acc=0.749, loss=56.079, backward_time=0.406, grad_norm=31.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.565e-04, train_time=1.518
+[gpub010:0/16] 2024-01-29 18:40:59,634 (trainer:737) INFO: 16epoch:train:3001-3100batch: iter_time=9.236e-05, forward_time=0.298, loss_ctc=46.873, loss_att=50.569, acc=0.743, loss=49.460, backward_time=0.411, grad_norm=29.230, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.565e-04, train_time=1.550
+[gpub010:0/16] 2024-01-29 18:43:44,448 (trainer:737) INFO: 16epoch:train:3101-3200batch: iter_time=8.158e-05, forward_time=0.357, loss_ctc=54.650, loss_att=54.115, acc=0.747, loss=54.276, backward_time=0.459, grad_norm=30.938, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.564e-04, train_time=1.648
+[gpub010:0/16] 2024-01-29 18:46:12,200 (trainer:737) INFO: 16epoch:train:3201-3300batch: iter_time=8.354e-05, forward_time=0.290, loss_ctc=49.921, loss_att=45.806, acc=0.739, loss=47.041, backward_time=0.401, grad_norm=31.596, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.564e-04, train_time=1.477
+[gpub010:0/16] 2024-01-29 18:48:50,657 (trainer:737) INFO: 16epoch:train:3301-3400batch: iter_time=1.470e-04, forward_time=0.298, loss_ctc=48.207, loss_att=47.023, acc=0.740, loss=47.378, backward_time=0.408, grad_norm=30.389, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.563e-04, train_time=1.584
+[gpub010:0/16] 2024-01-29 18:52:10,958 (trainer:737) INFO: 16epoch:train:3401-3500batch: iter_time=8.233e-05, forward_time=0.400, loss_ctc=46.305, loss_att=44.505, acc=0.742, loss=45.045, backward_time=0.444, grad_norm=29.429, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.562e-04, train_time=2.003
+[gpub010:0/16] 2024-01-29 18:54:29,836 (trainer:737) INFO: 16epoch:train:3501-3600batch: iter_time=5.447e-04, forward_time=0.304, loss_ctc=49.996, loss_att=44.594, acc=0.763, loss=46.215, backward_time=0.405, grad_norm=29.227, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.562e-04, train_time=1.389
+[gpub010:0/16] 2024-01-29 18:57:19,415 (trainer:737) INFO: 16epoch:train:3601-3700batch: iter_time=8.308e-05, forward_time=0.371, loss_ctc=54.823, loss_att=49.988, acc=0.738, loss=51.439, backward_time=0.445, grad_norm=32.849, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.561e-04, train_time=1.696
+[gpub010:0/16] 2024-01-29 18:59:09,357 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-01-29 18:59:28,575 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 18:59:32,200 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae105240a0>)
+[gpub010:0/16] 2024-01-29 18:59:32,200 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-01-29 18:59:32,203 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 19:06:46,517 (trainer:737) INFO: 16epoch:train:3701-3800batch: iter_time=3.900, forward_time=0.320, loss_ctc=55.641, loss_att=55.402, acc=0.731, loss=55.474, backward_time=0.401, grad_norm=38.193, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.561e-04, train_time=5.671
+[gpub010:0/16] 2024-01-29 19:09:16,505 (trainer:737) INFO: 16epoch:train:3801-3900batch: iter_time=8.625e-05, forward_time=0.401, loss_ctc=55.384, loss_att=50.235, acc=0.731, loss=51.779, backward_time=0.433, grad_norm=41.589, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.560e-04, train_time=1.499
+[gpub010:0/16] 2024-01-29 19:12:07,710 (trainer:737) INFO: 16epoch:train:3901-4000batch: iter_time=9.422e-05, forward_time=0.323, loss_ctc=49.049, loss_att=49.638, acc=0.718, loss=49.462, backward_time=0.404, grad_norm=34.778, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.560e-04, train_time=1.712
+[gpub010:0/16] 2024-01-29 19:14:51,411 (trainer:737) INFO: 16epoch:train:4001-4100batch: iter_time=9.192e-05, forward_time=0.292, loss_ctc=53.104, loss_att=54.104, acc=0.739, loss=53.804, backward_time=0.406, grad_norm=36.389, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.559e-04, train_time=1.637
+[gpub010:0/16] 2024-01-29 19:17:46,428 (trainer:737) INFO: 16epoch:train:4101-4200batch: iter_time=3.808e-04, forward_time=0.380, loss_ctc=51.314, loss_att=51.957, acc=0.740, loss=51.764, backward_time=0.448, grad_norm=30.869, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.559e-04, train_time=1.748
+[gpub010:0/16] 2024-01-29 19:20:15,182 (trainer:737) INFO: 16epoch:train:4201-4300batch: iter_time=9.637e-05, forward_time=0.307, loss_ctc=54.986, loss_att=58.284, acc=0.742, loss=57.295, backward_time=0.419, grad_norm=31.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.558e-04, train_time=1.489
+[gpub010:0/16] 2024-01-29 19:23:30,008 (trainer:737) INFO: 16epoch:train:4301-4400batch: iter_time=0.001, forward_time=0.299, loss_ctc=51.375, loss_att=51.619, acc=0.745, loss=51.546, backward_time=0.407, grad_norm=30.039, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.557e-04, train_time=1.948
+[gpub010:0/16] 2024-01-29 19:25:56,739 (trainer:737) INFO: 16epoch:train:4401-4500batch: iter_time=9.333e-05, forward_time=0.368, loss_ctc=49.810, loss_att=49.257, acc=0.749, loss=49.423, backward_time=0.457, grad_norm=30.309, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.557e-04, train_time=1.466
+[gpub010:0/16] 2024-01-29 19:28:08,334 (trainer:737) INFO: 16epoch:train:4501-4600batch: iter_time=8.744e-05, forward_time=0.291, loss_ctc=46.949, loss_att=42.704, acc=0.745, loss=43.978, backward_time=0.404, grad_norm=31.639, clip=100.000, loss_scale=6.387e+33, optim_step_time=0.092, optim0_lr0=2.556e-04, train_time=1.316
+[gpub010:0/16] 2024-01-29 19:30:58,744 (trainer:737) INFO: 16epoch:train:4601-4700batch: iter_time=4.068e-04, forward_time=0.299, loss_ctc=48.293, loss_att=48.836, acc=0.727, loss=48.673, backward_time=0.407, grad_norm=30.268, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.556e-04, train_time=1.704
+[gpub010:0/16] 2024-01-29 19:33:37,013 (trainer:737) INFO: 16epoch:train:4701-4800batch: iter_time=9.664e-05, forward_time=0.379, loss_ctc=49.302, loss_att=44.988, acc=0.754, loss=46.282, backward_time=0.451, grad_norm=28.366, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.555e-04, train_time=1.581
+[gpub010:0/16] 2024-01-29 19:36:19,861 (trainer:737) INFO: 16epoch:train:4801-4900batch: iter_time=8.944e-05, forward_time=0.293, loss_ctc=52.806, loss_att=50.037, acc=0.746, loss=50.868, backward_time=0.405, grad_norm=32.145, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.555e-04, train_time=1.630
+[gpub010:0/16] 2024-01-29 19:38:39,165 (trainer:737) INFO: 16epoch:train:4901-5000batch: iter_time=5.707e-04, forward_time=0.304, loss_ctc=56.873, loss_att=52.239, acc=0.736, loss=53.629, backward_time=0.407, grad_norm=40.738, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.554e-04, train_time=1.393
+[gpub010:0/16] 2024-01-29 19:38:59,440 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-01-29 19:39:18,299 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 19:39:21,914 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae680f76d0>)
+[gpub010:0/16] 2024-01-29 19:39:21,914 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-01-29 19:39:21,917 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 19:47:10,749 (trainer:737) INFO: 16epoch:train:5001-5100batch: iter_time=3.400, forward_time=0.365, loss_ctc=47.738, loss_att=44.602, acc=0.747, loss=45.543, backward_time=0.415, grad_norm=28.575, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.554e-04, train_time=5.114
+[gpub010:0/16] 2024-01-29 19:49:44,947 (trainer:737) INFO: 16epoch:train:5101-5200batch: iter_time=8.242e-05, forward_time=0.294, loss_ctc=53.170, loss_att=54.627, acc=0.721, loss=54.190, backward_time=0.408, grad_norm=40.798, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.553e-04, train_time=1.543
+[gpub010:0/16] 2024-01-29 19:52:51,553 (trainer:737) INFO: 16epoch:train:5201-5300batch: iter_time=7.968e-05, forward_time=0.437, loss_ctc=55.619, loss_att=54.023, acc=0.722, loss=54.502, backward_time=0.427, grad_norm=37.703, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.552e-04, train_time=1.866
+[gpub010:0/16] 2024-01-29 19:55:35,648 (trainer:737) INFO: 16epoch:train:5301-5400batch: iter_time=7.931e-05, forward_time=0.289, loss_ctc=47.722, loss_att=51.512, acc=0.738, loss=50.375, backward_time=0.406, grad_norm=30.093, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.552e-04, train_time=1.641
+[gpub010:0/16] 2024-01-29 19:58:19,681 (trainer:737) INFO: 16epoch:train:5401-5500batch: iter_time=8.376e-05, forward_time=0.304, loss_ctc=55.472, loss_att=55.855, acc=0.753, loss=55.740, backward_time=0.413, grad_norm=31.320, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.551e-04, train_time=1.640
+[gpub010:0/16] 2024-01-29 20:01:18,237 (trainer:737) INFO: 16epoch:train:5501-5600batch: iter_time=8.414e-05, forward_time=0.389, loss_ctc=46.230, loss_att=49.519, acc=0.745, loss=48.532, backward_time=0.422, grad_norm=29.128, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.551e-04, train_time=1.785
+[gpub010:0/16] 2024-01-29 20:04:09,705 (trainer:737) INFO: 16epoch:train:5601-5700batch: iter_time=8.400e-05, forward_time=0.299, loss_ctc=54.203, loss_att=53.700, acc=0.750, loss=53.851, backward_time=0.410, grad_norm=30.556, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.550e-04, train_time=1.715
+[gpub010:0/16] 2024-01-29 20:07:10,144 (trainer:737) INFO: 16epoch:train:5701-5800batch: iter_time=8.570e-05, forward_time=0.373, loss_ctc=49.250, loss_att=45.188, acc=0.742, loss=46.406, backward_time=0.433, grad_norm=31.229, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.550e-04, train_time=1.804
+[gpub010:0/16] 2024-01-29 20:09:45,069 (trainer:737) INFO: 16epoch:train:5801-5900batch: iter_time=8.121e-05, forward_time=0.300, loss_ctc=47.412, loss_att=46.093, acc=0.744, loss=46.489, backward_time=0.402, grad_norm=29.635, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.549e-04, train_time=1.548
+[gpub010:0/16] 2024-01-29 20:12:18,369 (trainer:737) INFO: 16epoch:train:5901-6000batch: iter_time=8.299e-05, forward_time=0.295, loss_ctc=45.238, loss_att=43.699, acc=0.746, loss=44.161, backward_time=0.410, grad_norm=28.078, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.549e-04, train_time=1.534
+[gpub010:0/16] 2024-01-29 20:15:40,629 (trainer:737) INFO: 16epoch:train:6001-6100batch: iter_time=8.475e-05, forward_time=0.403, loss_ctc=49.672, loss_att=44.671, acc=0.763, loss=46.171, backward_time=0.423, grad_norm=27.959, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.548e-04, train_time=2.022
+[gpub010:0/16] 2024-01-29 20:18:14,452 (trainer:737) INFO: 16epoch:train:6101-6200batch: iter_time=8.261e-05, forward_time=0.293, loss_ctc=54.728, loss_att=50.622, acc=0.735, loss=51.854, backward_time=0.405, grad_norm=33.644, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.547e-04, train_time=1.537
+[gpub010:0/16] 2024-01-29 20:19:45,151 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-01-29 20:20:04,477 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 20:20:08,133 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae5889fdc0>)
+[gpub010:0/16] 2024-01-29 20:20:08,133 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-01-29 20:20:08,136 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 20:26:13,679 (trainer:737) INFO: 16epoch:train:6201-6300batch: iter_time=3.282, forward_time=0.295, loss_ctc=54.603, loss_att=56.273, acc=0.726, loss=55.772, backward_time=0.414, grad_norm=40.663, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.547e-04, train_time=4.793
+[gpub010:0/16] 2024-01-29 20:29:38,913 (trainer:737) INFO: 16epoch:train:6301-6400batch: iter_time=7.849e-05, forward_time=0.370, loss_ctc=55.019, loss_att=51.220, acc=0.728, loss=52.360, backward_time=0.449, grad_norm=42.235, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.546e-04, train_time=2.052
+[gpub010:0/16] 2024-01-29 20:31:59,422 (trainer:737) INFO: 16epoch:train:6401-6500batch: iter_time=7.851e-05, forward_time=0.289, loss_ctc=48.841, loss_att=50.681, acc=0.713, loss=50.129, backward_time=0.403, grad_norm=36.157, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.546e-04, train_time=1.405
+[gpub010:0/16] 2024-01-29 20:34:33,042 (trainer:737) INFO: 16epoch:train:6501-6600batch: iter_time=9.163e-05, forward_time=0.320, loss_ctc=52.712, loss_att=54.029, acc=0.740, loss=53.634, backward_time=0.442, grad_norm=36.586, clip=100.000, loss_scale=1.277e+34, optim_step_time=0.101, optim0_lr0=2.545e-04, train_time=1.536
+[gpub010:0/16] 2024-01-29 20:37:19,380 (trainer:737) INFO: 16epoch:train:6601-6700batch: iter_time=8.352e-05, forward_time=0.370, loss_ctc=51.093, loss_att=52.408, acc=0.731, loss=52.013, backward_time=0.414, grad_norm=30.817, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.092, optim0_lr0=2.545e-04, train_time=1.663
+[gpub010:0/16] 2024-01-29 20:39:57,075 (trainer:737) INFO: 16epoch:train:6701-6800batch: iter_time=4.201e-04, forward_time=0.299, loss_ctc=54.554, loss_att=58.051, acc=0.737, loss=57.002, backward_time=0.416, grad_norm=31.445, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.093, optim0_lr0=2.544e-04, train_time=1.577
+[gpub010:0/16] 2024-01-29 20:42:32,298 (trainer:737) INFO: 16epoch:train:6801-6900batch: iter_time=9.091e-05, forward_time=0.419, loss_ctc=50.258, loss_att=49.639, acc=0.741, loss=49.825, backward_time=0.422, grad_norm=30.554, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.095, optim0_lr0=2.544e-04, train_time=1.552
+[gpub010:0/16] 2024-01-29 20:44:03,621 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 20:45:06,755 (trainer:737) INFO: 16epoch:train:6901-7000batch: iter_time=8.529e-05, forward_time=0.292, loss_ctc=49.286, loss_att=48.574, acc=0.745, loss=48.788, backward_time=0.403, grad_norm=30.501, clip=100.000, loss_scale=1.647e+34, optim_step_time=0.092, optim0_lr0=2.543e-04, train_time=1.544
+[gpub010:0/16] 2024-01-29 20:47:54,412 (trainer:737) INFO: 16epoch:train:7001-7100batch: iter_time=4.744e-04, forward_time=0.298, loss_ctc=46.610, loss_att=40.851, acc=0.743, loss=42.579, backward_time=0.404, grad_norm=30.499, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.543e-04, train_time=1.675
+[gpub010:0/16] 2024-01-29 20:50:41,704 (trainer:737) INFO: 16epoch:train:7101-7200batch: iter_time=9.138e-05, forward_time=0.398, loss_ctc=48.176, loss_att=50.513, acc=0.706, loss=49.812, backward_time=0.420, grad_norm=33.994, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.110, optim0_lr0=2.542e-04, train_time=1.674
+[gpub010:0/16] 2024-01-29 20:53:01,697 (trainer:737) INFO: 16epoch:train:7201-7300batch: iter_time=2.076e-04, forward_time=0.295, loss_ctc=48.934, loss_att=45.495, acc=0.745, loss=46.527, backward_time=0.415, grad_norm=29.934, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.541e-04, train_time=1.399
+[gpub010:0/16] 2024-01-29 20:56:13,019 (trainer:737) INFO: 16epoch:train:7301-7400batch: iter_time=9.573e-05, forward_time=0.390, loss_ctc=52.620, loss_att=50.293, acc=0.739, loss=50.991, backward_time=0.457, grad_norm=34.209, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.541e-04, train_time=1.913
+[gpub010:0/16] 2024-01-29 20:58:30,168 (trainer:737) INFO: 16epoch:train:7401-7500batch: iter_time=9.911e-05, forward_time=0.291, loss_ctc=56.776, loss_att=51.941, acc=0.736, loss=53.392, backward_time=0.408, grad_norm=40.844, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.540e-04, train_time=1.372
+[gpub010:0/16] 2024-01-29 20:58:50,395 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-01-29 20:59:09,424 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 20:59:12,963 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae680b03d0>)
+[gpub010:0/16] 2024-01-29 20:59:12,963 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-01-29 20:59:12,966 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 21:08:09,495 (trainer:737) INFO: 16epoch:train:7501-7600batch: iter_time=3.997, forward_time=0.361, loss_ctc=47.449, loss_att=45.531, acc=0.748, loss=46.107, backward_time=0.502, grad_norm=28.794, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.540e-04, train_time=5.793
+[gpub010:0/16] 2024-01-29 21:11:26,871 (trainer:737) INFO: 16epoch:train:7601-7700batch: iter_time=7.809e-05, forward_time=0.291, loss_ctc=53.139, loss_att=54.642, acc=0.723, loss=54.191, backward_time=0.401, grad_norm=40.190, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.539e-04, train_time=1.974
+[gpub010:0/16] 2024-01-29 21:14:13,240 (trainer:737) INFO: 16epoch:train:7701-7800batch: iter_time=7.809e-05, forward_time=0.326, loss_ctc=54.551, loss_att=53.832, acc=0.724, loss=54.048, backward_time=0.459, grad_norm=40.382, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.539e-04, train_time=1.663
+[gpub010:0/16] 2024-01-29 21:16:46,054 (trainer:737) INFO: 16epoch:train:7801-7900batch: iter_time=7.948e-05, forward_time=0.363, loss_ctc=47.495, loss_att=51.625, acc=0.738, loss=50.386, backward_time=0.413, grad_norm=30.144, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=2.538e-04, train_time=1.527
+[gpub010:0/16] 2024-01-29 21:20:06,080 (trainer:737) INFO: 16epoch:train:7901-8000batch: iter_time=9.704e-04, forward_time=0.428, loss_ctc=55.357, loss_att=55.412, acc=0.754, loss=55.396, backward_time=0.428, grad_norm=30.789, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.109, optim0_lr0=2.538e-04, train_time=2.000
+[gpub010:0/16] 2024-01-29 21:22:51,788 (trainer:737) INFO: 16epoch:train:8001-8100batch: iter_time=8.208e-05, forward_time=0.290, loss_ctc=46.156, loss_att=49.413, acc=0.748, loss=48.436, backward_time=0.403, grad_norm=31.923, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.537e-04, train_time=1.658
+[gpub010:0/16] 2024-01-29 21:25:28,540 (trainer:737) INFO: 16epoch:train:8101-8200batch: iter_time=8.352e-05, forward_time=0.433, loss_ctc=54.302, loss_att=53.580, acc=0.752, loss=53.796, backward_time=0.436, grad_norm=31.357, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.537e-04, train_time=1.567
+[gpub010:0/16] 2024-01-29 21:26:45,220 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 21:28:09,238 (trainer:737) INFO: 16epoch:train:8201-8300batch: iter_time=8.237e-05, forward_time=0.289, loss_ctc=48.641, loss_att=44.864, acc=0.743, loss=45.997, backward_time=0.401, grad_norm=30.655, clip=100.000, loss_scale=7.552e+33, optim_step_time=0.092, optim0_lr0=2.536e-04, train_time=1.607
+[gpub010:0/16] 2024-01-29 21:30:37,343 (trainer:737) INFO: 16epoch:train:8301-8400batch: iter_time=8.008e-05, forward_time=0.290, loss_ctc=47.011, loss_att=45.782, acc=0.745, loss=46.151, backward_time=0.402, grad_norm=31.188, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.535e-04, train_time=1.481
+[gpub010:0/16] 2024-01-29 21:33:24,694 (trainer:737) INFO: 16epoch:train:8401-8500batch: iter_time=0.001, forward_time=0.419, loss_ctc=45.529, loss_att=44.056, acc=0.746, loss=44.498, backward_time=0.424, grad_norm=28.402, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.535e-04, train_time=1.672
+[gpub010:0/16] 2024-01-29 21:36:09,741 (trainer:737) INFO: 16epoch:train:8501-8600batch: iter_time=8.124e-05, forward_time=0.291, loss_ctc=49.409, loss_att=44.823, acc=0.764, loss=46.199, backward_time=0.403, grad_norm=28.640, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.534e-04, train_time=1.651
+[gpub010:0/16] 2024-01-29 21:39:16,983 (trainer:737) INFO: 16epoch:train:8601-8700batch: iter_time=1.336e-04, forward_time=0.398, loss_ctc=54.257, loss_att=50.473, acc=0.737, loss=51.608, backward_time=0.444, grad_norm=32.886, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.534e-04, train_time=1.871
+[gpub010:0/16] 2024-01-29 21:40:43,358 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-01-29 21:41:02,624 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 21:41:06,178 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb12950b3a0>)
+[gpub010:0/16] 2024-01-29 21:41:06,178 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-01-29 21:41:06,181 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 21:47:35,298 (trainer:737) INFO: 16epoch:train:8701-8800batch: iter_time=3.328, forward_time=0.373, loss_ctc=53.095, loss_att=55.628, acc=0.726, loss=54.868, backward_time=0.476, grad_norm=39.205, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.533e-04, train_time=4.983
+[gpub010:0/16] 2024-01-29 21:50:11,398 (trainer:737) INFO: 16epoch:train:8801-8900batch: iter_time=8.056e-05, forward_time=0.288, loss_ctc=53.299, loss_att=49.152, acc=0.733, loss=50.396, backward_time=0.399, grad_norm=35.795, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.533e-04, train_time=1.560
+[gpub010:0/16] 2024-01-29 21:52:50,881 (trainer:737) INFO: 16epoch:train:8901-9000batch: iter_time=8.503e-05, forward_time=0.289, loss_ctc=48.643, loss_att=49.880, acc=0.716, loss=49.509, backward_time=0.401, grad_norm=34.522, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.532e-04, train_time=1.596
+[gpub010:0/16] 2024-01-29 21:55:37,431 (trainer:737) INFO: 16epoch:train:9001-9100batch: iter_time=8.064e-05, forward_time=0.414, loss_ctc=51.816, loss_att=53.235, acc=0.743, loss=52.809, backward_time=0.436, grad_norm=38.618, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.135, optim0_lr0=2.532e-04, train_time=1.665
+[gpub010:0/16] 2024-01-29 21:58:29,641 (trainer:737) INFO: 16epoch:train:9101-9200batch: iter_time=8.483e-05, forward_time=0.289, loss_ctc=50.763, loss_att=52.131, acc=0.731, loss=51.721, backward_time=0.402, grad_norm=31.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.531e-04, train_time=1.720
+[gpub010:0/16] 2024-01-29 22:01:13,373 (trainer:737) INFO: 16epoch:train:9201-9300batch: iter_time=8.146e-05, forward_time=0.430, loss_ctc=54.743, loss_att=58.177, acc=0.740, loss=57.147, backward_time=0.430, grad_norm=32.610, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=2.531e-04, train_time=1.638
+[gpub010:0/16] 2024-01-29 22:03:35,087 (trainer:737) INFO: 16epoch:train:9301-9400batch: iter_time=8.368e-05, forward_time=0.291, loss_ctc=50.227, loss_att=49.791, acc=0.741, loss=49.922, backward_time=0.403, grad_norm=31.983, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.530e-04, train_time=1.418
+[gpub010:0/16] 2024-01-29 22:06:12,649 (trainer:737) INFO: 16epoch:train:9401-9500batch: iter_time=8.535e-05, forward_time=0.290, loss_ctc=48.941, loss_att=48.318, acc=0.746, loss=48.505, backward_time=0.402, grad_norm=30.522, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.529e-04, train_time=1.574
+[gpub010:0/16] 2024-01-29 22:09:04,007 (trainer:737) INFO: 16epoch:train:9501-9600batch: iter_time=8.942e-05, forward_time=0.411, loss_ctc=46.481, loss_att=40.818, acc=0.742, loss=42.517, backward_time=0.422, grad_norm=29.707, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=2.529e-04, train_time=1.714
+[gpub010:0/16] 2024-01-29 22:11:24,728 (trainer:737) INFO: 16epoch:train:9601-9700batch: iter_time=8.993e-05, forward_time=0.287, loss_ctc=47.621, loss_att=49.362, acc=0.710, loss=48.839, backward_time=0.400, grad_norm=30.931, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.528e-04, train_time=1.407
+[gpub010:0/16] 2024-01-29 22:14:28,763 (trainer:737) INFO: 16epoch:train:9701-9800batch: iter_time=9.064e-05, forward_time=0.375, loss_ctc=48.306, loss_att=44.875, acc=0.748, loss=45.904, backward_time=0.485, grad_norm=28.930, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.528e-04, train_time=1.840
+[gpub010:0/16] 2024-01-29 22:17:07,208 (trainer:737) INFO: 16epoch:train:9801-9900batch: iter_time=8.785e-05, forward_time=0.291, loss_ctc=52.345, loss_att=49.909, acc=0.737, loss=50.640, backward_time=0.402, grad_norm=34.093, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.527e-04, train_time=1.584
+[gpub010:0/16] 2024-01-29 22:19:33,473 (trainer:737) INFO: 16epoch:train:9901-10000batch: iter_time=8.496e-05, forward_time=0.290, loss_ctc=55.634, loss_att=52.928, acc=0.734, loss=53.740, backward_time=0.401, grad_norm=38.083, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.527e-04, train_time=1.463
+[gpub010:0/16] 2024-01-29 22:19:53,502 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-01-29 22:20:13,320 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 22:20:17,228 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae380d9480>)
+[gpub010:0/16] 2024-01-29 22:20:17,228 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-01-29 22:20:17,231 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 22:29:13,439 (trainer:737) INFO: 16epoch:train:10001-10100batch: iter_time=3.739, forward_time=0.435, loss_ctc=46.907, loss_att=43.524, acc=0.745, loss=44.539, backward_time=0.424, grad_norm=28.177, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.526e-04, train_time=5.799
+[gpub010:0/16] 2024-01-29 22:32:22,889 (trainer:737) INFO: 16epoch:train:10101-10200batch: iter_time=8.453e-05, forward_time=0.380, loss_ctc=52.600, loss_att=52.857, acc=0.722, loss=52.780, backward_time=0.460, grad_norm=42.632, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.526e-04, train_time=1.895
+[gpub010:0/16] 2024-01-29 22:34:52,684 (trainer:737) INFO: 16epoch:train:10201-10300batch: iter_time=9.037e-05, forward_time=0.291, loss_ctc=54.443, loss_att=53.517, acc=0.721, loss=53.795, backward_time=0.402, grad_norm=40.752, clip=100.000, loss_scale=7.996e+33, optim_step_time=0.092, optim0_lr0=2.525e-04, train_time=1.498
+[gpub010:0/16] 2024-01-29 22:38:10,177 (trainer:737) INFO: 16epoch:train:10301-10400batch: iter_time=9.344e-05, forward_time=0.287, loss_ctc=47.034, loss_att=50.711, acc=0.734, loss=49.608, backward_time=0.399, grad_norm=30.796, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.525e-04, train_time=1.975
+[gpub010:0/16] 2024-01-29 22:41:23,548 (trainer:737) INFO: 16epoch:train:10401-10500batch: iter_time=9.843e-05, forward_time=0.451, loss_ctc=55.171, loss_att=55.468, acc=0.743, loss=55.379, backward_time=0.427, grad_norm=31.355, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=2.524e-04, train_time=1.934
+[gpub010:0/16] 2024-01-29 22:44:03,535 (trainer:737) INFO: 16epoch:train:10501-10600batch: iter_time=1.008e-04, forward_time=0.289, loss_ctc=45.826, loss_att=47.347, acc=0.743, loss=46.890, backward_time=0.401, grad_norm=29.435, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.524e-04, train_time=1.600
+[gpub010:0/16] 2024-01-29 22:47:49,603 (trainer:737) INFO: 16epoch:train:10601-10700batch: iter_time=8.881e-05, forward_time=0.444, loss_ctc=53.276, loss_att=52.948, acc=0.748, loss=53.047, backward_time=0.443, grad_norm=29.982, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.523e-04, train_time=2.260
+[gpub010:0/16] 2024-01-29 22:50:01,356 (trainer:737) INFO: 16epoch:train:10701-10800batch: iter_time=8.356e-05, forward_time=0.288, loss_ctc=48.160, loss_att=44.472, acc=0.740, loss=45.579, backward_time=0.401, grad_norm=30.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.523e-04, train_time=1.318
+[gpub010:0/16] 2024-01-29 22:53:02,329 (trainer:737) INFO: 16epoch:train:10801-10900batch: iter_time=8.627e-05, forward_time=0.287, loss_ctc=47.109, loss_att=44.709, acc=0.734, loss=45.429, backward_time=0.401, grad_norm=30.974, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.522e-04, train_time=1.810
+[gpub010:0/16] 2024-01-29 22:55:33,584 (trainer:737) INFO: 16epoch:train:10901-11000batch: iter_time=2.453e-04, forward_time=0.336, loss_ctc=45.288, loss_att=44.327, acc=0.731, loss=44.616, backward_time=0.485, grad_norm=29.441, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.521e-04, train_time=1.512
+[gpub010:0/16] 2024-01-29 22:58:06,095 (trainer:737) INFO: 16epoch:train:11001-11100batch: iter_time=8.849e-05, forward_time=0.289, loss_ctc=49.201, loss_att=44.931, acc=0.751, loss=46.212, backward_time=0.402, grad_norm=30.325, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.521e-04, train_time=1.525
+[gpub010:0/16] 2024-01-29 23:01:20,705 (trainer:737) INFO: 16epoch:train:11101-11200batch: iter_time=0.001, forward_time=0.341, loss_ctc=54.698, loss_att=49.778, acc=0.740, loss=51.254, backward_time=0.482, grad_norm=34.150, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.520e-04, train_time=1.945
+[gpub010:0/16] 2024-01-29 23:02:53,268 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-01-29 23:03:12,718 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 23:03:16,718 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb12a2cf610>)
+[gpub010:0/16] 2024-01-29 23:03:16,718 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-01-29 23:03:16,721 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 23:09:18,297 (trainer:737) INFO: 16epoch:train:11201-11300batch: iter_time=3.325, forward_time=0.288, loss_ctc=53.014, loss_att=54.433, acc=0.729, loss=54.007, backward_time=0.402, grad_norm=36.584, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.520e-04, train_time=4.777
+[gpub010:0/16] 2024-01-29 23:12:11,198 (trainer:737) INFO: 16epoch:train:11301-11400batch: iter_time=8.368e-05, forward_time=0.410, loss_ctc=53.467, loss_att=49.088, acc=0.735, loss=50.402, backward_time=0.432, grad_norm=41.107, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.110, optim0_lr0=2.519e-04, train_time=1.728
+[gpub010:0/16] 2024-01-29 23:15:02,906 (trainer:737) INFO: 16epoch:train:11401-11500batch: iter_time=8.295e-05, forward_time=0.289, loss_ctc=47.932, loss_att=48.552, acc=0.719, loss=48.366, backward_time=0.401, grad_norm=34.818, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.519e-04, train_time=1.717
+[gpub010:0/16] 2024-01-29 23:17:15,253 (trainer:737) INFO: 16epoch:train:11501-11600batch: iter_time=8.508e-05, forward_time=0.290, loss_ctc=51.122, loss_att=52.099, acc=0.745, loss=51.806, backward_time=0.403, grad_norm=37.257, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.518e-04, train_time=1.323
+[gpub010:0/16] 2024-01-29 23:20:00,030 (trainer:737) INFO: 16epoch:train:11601-11700batch: iter_time=8.443e-05, forward_time=0.412, loss_ctc=50.998, loss_att=51.522, acc=0.732, loss=51.364, backward_time=0.430, grad_norm=31.096, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.518e-04, train_time=1.647
+[gpub010:0/16] 2024-01-29 23:22:41,486 (trainer:737) INFO: 16epoch:train:11701-11800batch: iter_time=8.333e-05, forward_time=0.293, loss_ctc=54.313, loss_att=57.490, acc=0.742, loss=56.537, backward_time=0.405, grad_norm=31.385, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.517e-04, train_time=1.615
+[gpub010:0/16] 2024-01-29 23:26:24,339 (trainer:737) INFO: 16epoch:train:11801-11900batch: iter_time=0.342, forward_time=0.426, loss_ctc=50.007, loss_att=48.585, acc=0.744, loss=49.012, backward_time=0.447, grad_norm=30.763, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.517e-04, train_time=2.228
+[gpub010:0/16] 2024-01-29 23:28:47,854 (trainer:737) INFO: 16epoch:train:11901-12000batch: iter_time=8.720e-05, forward_time=0.291, loss_ctc=49.137, loss_att=48.360, acc=0.747, loss=48.593, backward_time=0.402, grad_norm=30.717, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.516e-04, train_time=1.434
+[gpub010:0/16] 2024-01-29 23:31:24,417 (trainer:737) INFO: 16epoch:train:12001-12100batch: iter_time=8.656e-05, forward_time=0.288, loss_ctc=45.944, loss_att=40.258, acc=0.746, loss=41.964, backward_time=0.399, grad_norm=34.933, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.516e-04, train_time=1.566
+[gpub010:0/16] 2024-01-29 23:33:58,490 (trainer:737) INFO: 16epoch:train:12101-12200batch: iter_time=8.616e-05, forward_time=0.439, loss_ctc=47.828, loss_att=48.636, acc=0.714, loss=48.394, backward_time=0.433, grad_norm=31.993, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.515e-04, train_time=1.541
+[gpub010:0/16] 2024-01-29 23:36:25,903 (trainer:737) INFO: 16epoch:train:12201-12300batch: iter_time=8.346e-05, forward_time=0.290, loss_ctc=48.906, loss_att=44.983, acc=0.747, loss=46.160, backward_time=0.403, grad_norm=29.765, clip=100.000, loss_scale=1.599e+34, optim_step_time=0.092, optim0_lr0=2.515e-04, train_time=1.472
+[gpub010:0/16] 2024-01-29 23:37:42,119 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-29 23:39:29,731 (trainer:737) INFO: 16epoch:train:12301-12400batch: iter_time=8.505e-05, forward_time=0.291, loss_ctc=52.220, loss_att=49.441, acc=0.742, loss=50.275, backward_time=0.402, grad_norm=32.620, clip=100.000, loss_scale=1.406e+34, optim_step_time=0.092, optim0_lr0=2.514e-04, train_time=1.840
+[gpub010:0/16] 2024-01-29 23:41:56,477 (trainer:737) INFO: 16epoch:train:12401-12500batch: iter_time=8.252e-05, forward_time=0.362, loss_ctc=55.432, loss_att=52.267, acc=0.736, loss=53.216, backward_time=0.461, grad_norm=39.465, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.513e-04, train_time=1.467
+[gpub010:0/16] 2024-01-29 23:42:16,506 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-01-29 23:42:35,879 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-29 23:42:39,836 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae8b1535e0>)
+[gpub010:0/16] 2024-01-29 23:42:39,836 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-29 23:42:39,840 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-29 23:49:54,618 (trainer:737) INFO: 16epoch:train:12501-12600batch: iter_time=3.126, forward_time=0.288, loss_ctc=46.795, loss_att=46.791, acc=0.748, loss=46.792, backward_time=0.401, grad_norm=28.116, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.513e-04, train_time=4.781
+[gpub010:0/16] 2024-01-29 23:53:21,776 (trainer:737) INFO: 16epoch:train:12601-12700batch: iter_time=8.689e-05, forward_time=0.352, loss_ctc=52.383, loss_att=54.511, acc=0.726, loss=53.873, backward_time=0.487, grad_norm=41.848, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.512e-04, train_time=2.071
+[gpub010:0/16] 2024-01-29 23:56:27,844 (trainer:737) INFO: 16epoch:train:12701-12800batch: iter_time=8.679e-05, forward_time=0.290, loss_ctc=53.033, loss_att=53.451, acc=0.725, loss=53.326, backward_time=0.402, grad_norm=38.651, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.512e-04, train_time=1.861
+[gpub010:0/16] 2024-01-29 23:59:18,579 (trainer:737) INFO: 16epoch:train:12801-12900batch: iter_time=9.451e-05, forward_time=0.399, loss_ctc=47.532, loss_att=52.311, acc=0.738, loss=50.877, backward_time=0.438, grad_norm=30.444, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.511e-04, train_time=1.706
+[gpub010:0/16] 2024-01-29 23:59:31,660 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 00:02:08,542 (trainer:737) INFO: 16epoch:train:12901-13000batch: iter_time=1.087e-04, forward_time=0.292, loss_ctc=54.975, loss_att=55.637, acc=0.754, loss=55.439, backward_time=0.406, grad_norm=30.450, clip=100.000, loss_scale=5.664e+33, optim_step_time=0.092, optim0_lr0=2.511e-04, train_time=1.699
+[gpub010:0/16] 2024-01-30 00:04:59,090 (trainer:737) INFO: 16epoch:train:13001-13100batch: iter_time=1.032e-04, forward_time=0.369, loss_ctc=45.656, loss_att=49.372, acc=0.747, loss=48.257, backward_time=0.463, grad_norm=28.681, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.510e-04, train_time=1.706
+[gpub010:0/16] 2024-01-30 00:08:13,501 (trainer:737) INFO: 16epoch:train:13101-13200batch: iter_time=1.048e-04, forward_time=0.293, loss_ctc=53.295, loss_att=53.374, acc=0.753, loss=53.350, backward_time=0.404, grad_norm=30.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.510e-04, train_time=1.944
+[gpub010:0/16] 2024-01-30 00:11:13,492 (trainer:737) INFO: 16epoch:train:13201-13300batch: iter_time=1.055e-04, forward_time=0.287, loss_ctc=48.404, loss_att=45.007, acc=0.744, loss=46.026, backward_time=0.398, grad_norm=30.449, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.509e-04, train_time=1.799
+[gpub010:0/16] 2024-01-30 00:14:06,935 (trainer:737) INFO: 16epoch:train:13301-13400batch: iter_time=1.025e-04, forward_time=0.354, loss_ctc=47.959, loss_att=46.781, acc=0.745, loss=47.134, backward_time=0.483, grad_norm=31.015, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=2.509e-04, train_time=1.735
+[gpub010:0/16] 2024-01-30 00:17:19,323 (trainer:737) INFO: 16epoch:train:13401-13500batch: iter_time=9.805e-05, forward_time=0.289, loss_ctc=45.111, loss_att=43.949, acc=0.748, loss=44.298, backward_time=0.399, grad_norm=29.410, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.508e-04, train_time=1.923
+[gpub010:0/16] 2024-01-30 00:20:41,961 (trainer:737) INFO: 16epoch:train:13501-13600batch: iter_time=9.660e-05, forward_time=0.384, loss_ctc=49.071, loss_att=44.723, acc=0.765, loss=46.028, backward_time=0.444, grad_norm=28.912, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.508e-04, train_time=2.026
+[gpub010:0/16] 2024-01-30 00:23:49,532 (trainer:737) INFO: 16epoch:train:13601-13700batch: iter_time=9.176e-05, forward_time=0.290, loss_ctc=54.003, loss_att=49.898, acc=0.739, loss=51.130, backward_time=0.403, grad_norm=32.868, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.507e-04, train_time=1.876
+[gpub010:0/16] 2024-01-30 00:25:47,597 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-01-30 00:26:07,030 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 00:26:10,635 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae5a429900>)
+[gpub010:0/16] 2024-01-30 00:26:10,635 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-01-30 00:26:10,638 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 00:33:31,796 (trainer:737) INFO: 16epoch:train:13701-13800batch: iter_time=3.267, forward_time=0.377, loss_ctc=52.375, loss_att=55.601, acc=0.729, loss=54.633, backward_time=0.415, grad_norm=41.127, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.507e-04, train_time=5.822
+[gpub010:0/16] 2024-01-30 00:37:03,995 (trainer:737) INFO: 16epoch:train:13801-13900batch: iter_time=1.064e-04, forward_time=0.287, loss_ctc=53.440, loss_att=48.645, acc=0.738, loss=50.084, backward_time=0.398, grad_norm=39.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.506e-04, train_time=2.122
+[gpub010:0/16] 2024-01-30 00:39:58,362 (trainer:737) INFO: 16epoch:train:13901-14000batch: iter_time=8.864e-05, forward_time=0.352, loss_ctc=47.691, loss_att=48.993, acc=0.719, loss=48.602, backward_time=0.430, grad_norm=34.728, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.506e-04, train_time=1.743
+[gpub010:0/16] 2024-01-30 00:43:14,160 (trainer:737) INFO: 16epoch:train:14001-14100batch: iter_time=8.869e-05, forward_time=0.289, loss_ctc=51.673, loss_att=52.853, acc=0.744, loss=52.499, backward_time=0.399, grad_norm=35.763, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.505e-04, train_time=1.958
+[gpub010:0/16] 2024-01-30 00:46:42,661 (trainer:737) INFO: 16epoch:train:14101-14200batch: iter_time=9.764e-05, forward_time=0.289, loss_ctc=50.513, loss_att=51.693, acc=0.734, loss=51.339, backward_time=0.398, grad_norm=31.823, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.505e-04, train_time=2.085
+[gpub010:0/16] 2024-01-30 00:49:57,320 (trainer:737) INFO: 16epoch:train:14201-14300batch: iter_time=8.734e-05, forward_time=0.359, loss_ctc=53.693, loss_att=57.218, acc=0.743, loss=56.161, backward_time=0.442, grad_norm=31.530, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.504e-04, train_time=1.946
+[gpub010:0/16] 2024-01-30 00:52:53,550 (trainer:737) INFO: 16epoch:train:14301-14400batch: iter_time=8.506e-05, forward_time=0.290, loss_ctc=49.679, loss_att=49.119, acc=0.742, loss=49.287, backward_time=0.402, grad_norm=52.419, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.503e-04, train_time=1.762
+[gpub010:0/16] 2024-01-30 00:54:04,042 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 00:56:17,344 (trainer:737) INFO: 16epoch:train:14401-14500batch: iter_time=9.525e-05, forward_time=0.359, loss_ctc=48.331, loss_att=48.176, acc=0.748, loss=48.223, backward_time=0.454, grad_norm=30.196, clip=100.000, loss_scale=3.383e+33, optim_step_time=0.098, optim0_lr0=2.503e-04, train_time=2.038
+[gpub010:0/16] 2024-01-30 00:59:19,919 (trainer:737) INFO: 16epoch:train:14501-14600batch: iter_time=9.155e-05, forward_time=0.287, loss_ctc=46.724, loss_att=40.865, acc=0.744, loss=42.623, backward_time=0.396, grad_norm=31.414, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.502e-04, train_time=1.825
+[gpub010:0/16] 2024-01-30 01:02:13,352 (trainer:737) INFO: 16epoch:train:14601-14700batch: iter_time=1.037e-04, forward_time=0.308, loss_ctc=47.306, loss_att=48.730, acc=0.714, loss=48.303, backward_time=0.412, grad_norm=31.042, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.502e-04, train_time=1.734
+[gpub010:0/16] 2024-01-30 01:05:54,023 (trainer:737) INFO: 16epoch:train:14701-14800batch: iter_time=9.509e-05, forward_time=0.351, loss_ctc=47.964, loss_att=44.530, acc=0.749, loss=45.560, backward_time=0.417, grad_norm=29.420, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.501e-04, train_time=2.207
+[gpub010:0/16] 2024-01-30 01:09:24,018 (trainer:737) INFO: 16epoch:train:14801-14900batch: iter_time=8.918e-05, forward_time=0.289, loss_ctc=51.662, loss_att=48.524, acc=0.743, loss=49.465, backward_time=0.401, grad_norm=32.825, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.501e-04, train_time=2.100
+[gpub010:0/16] 2024-01-30 01:12:28,880 (trainer:737) INFO: 16epoch:train:14901-15000batch: iter_time=1.027e-04, forward_time=0.358, loss_ctc=55.337, loss_att=52.730, acc=0.735, loss=53.512, backward_time=0.427, grad_norm=38.426, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.500e-04, train_time=1.848
+[gpub010:0/16] 2024-01-30 01:49:57,446 (trainer:343) INFO: 16epoch results: [train] iter_time=0.283, forward_time=0.333, loss_ctc=51.076, loss_att=50.076, acc=0.738, loss=50.376, backward_time=0.421, grad_norm=33.323, clip=100.000, loss_scale=8.600e+33, optim_step_time=0.096, optim0_lr0=2.540e-04, train_time=1.980, time=8 hours, 15 minutes and 23.83 seconds, total_count=270000, gpu_max_cached_mem_GB=42.072, [valid] loss_ctc=43.657, cer_ctc=0.222, loss_att=46.324, acc=0.640, cer=0.314, wer=0.997, loss=45.524, time=37 minutes and 3.45 seconds, total_count=84078, gpu_max_cached_mem_GB=42.072
+[gpub010:0/16] 2024-01-30 01:50:07,991 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-01-30 01:50:08,023 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/11epoch.pth
+[gpub010:0/16] 2024-01-30 01:50:08,024 (trainer:272) INFO: 17/45epoch started. Estimated time to finish: 1 week, 3 days and 11 hours
+[gpub010:0/16] 2024-01-30 01:50:08,034 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-30 01:50:26,414 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 01:50:29,892 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa712b093c0>)
+[gpub010:0/16] 2024-01-30 01:50:29,892 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-01-30 01:50:29,895 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 01:57:22,221 (trainer:737) INFO: 17epoch:train:1-100batch: iter_time=2.807, forward_time=0.323, loss_ctc=50.340, loss_att=51.072, acc=0.719, loss=50.852, backward_time=0.416, grad_norm=37.879, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.500e-04, train_time=4.342
+[gpub010:0/16] 2024-01-30 01:59:44,522 (trainer:737) INFO: 17epoch:train:101-200batch: iter_time=9.544e-05, forward_time=0.315, loss_ctc=56.409, loss_att=51.014, acc=0.719, loss=52.633, backward_time=0.415, grad_norm=36.064, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=2.499e-04, train_time=1.421
+[gpub010:0/16] 2024-01-30 02:02:10,566 (trainer:737) INFO: 17epoch:train:201-300batch: iter_time=1.097e-04, forward_time=0.291, loss_ctc=52.957, loss_att=56.336, acc=0.732, loss=55.322, backward_time=0.407, grad_norm=36.214, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.499e-04, train_time=1.461
+[gpub010:0/16] 2024-01-30 02:04:47,850 (trainer:737) INFO: 17epoch:train:301-400batch: iter_time=1.022e-04, forward_time=0.326, loss_ctc=56.906, loss_att=55.198, acc=0.728, loss=55.710, backward_time=0.423, grad_norm=34.708, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.498e-04, train_time=1.573
+[gpub010:0/16] 2024-01-30 02:07:07,569 (trainer:737) INFO: 17epoch:train:401-500batch: iter_time=9.814e-05, forward_time=0.324, loss_ctc=49.707, loss_att=45.252, acc=0.722, loss=46.589, backward_time=0.420, grad_norm=32.758, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=2.498e-04, train_time=1.396
+[gpub010:0/16] 2024-01-30 02:09:28,743 (trainer:737) INFO: 17epoch:train:501-600batch: iter_time=1.038e-04, forward_time=0.291, loss_ctc=56.214, loss_att=49.748, acc=0.752, loss=51.687, backward_time=0.408, grad_norm=38.655, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.497e-04, train_time=1.412
+[gpub010:0/16] 2024-01-30 02:11:59,091 (trainer:737) INFO: 17epoch:train:601-700batch: iter_time=1.021e-04, forward_time=0.328, loss_ctc=53.762, loss_att=51.113, acc=0.729, loss=51.907, backward_time=0.432, grad_norm=34.161, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.497e-04, train_time=1.504
+[gpub010:0/16] 2024-01-30 02:14:31,159 (trainer:737) INFO: 17epoch:train:701-800batch: iter_time=1.038e-04, forward_time=0.337, loss_ctc=50.855, loss_att=42.557, acc=0.748, loss=45.046, backward_time=0.417, grad_norm=33.762, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=2.496e-04, train_time=1.520
+[gpub010:0/16] 2024-01-30 02:16:53,317 (trainer:737) INFO: 17epoch:train:801-900batch: iter_time=8.468e-05, forward_time=0.297, loss_ctc=52.382, loss_att=54.872, acc=0.742, loss=54.125, backward_time=0.422, grad_norm=31.091, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=2.496e-04, train_time=1.422
+[gpub010:0/16] 2024-01-30 02:19:20,784 (trainer:737) INFO: 17epoch:train:901-1000batch: iter_time=1.011e-04, forward_time=0.353, loss_ctc=47.520, loss_att=43.415, acc=0.749, loss=44.646, backward_time=0.423, grad_norm=32.135, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=2.495e-04, train_time=1.475
+[gpub010:0/16] 2024-01-30 02:21:59,424 (trainer:737) INFO: 17epoch:train:1001-1100batch: iter_time=1.103e-04, forward_time=0.302, loss_ctc=51.207, loss_att=49.146, acc=0.733, loss=49.764, backward_time=0.400, grad_norm=32.992, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=2.495e-04, train_time=1.585
+[gpub010:0/16] 2024-01-30 02:24:27,956 (trainer:737) INFO: 17epoch:train:1101-1200batch: iter_time=1.026e-04, forward_time=0.318, loss_ctc=62.181, loss_att=58.742, acc=0.727, loss=59.774, backward_time=0.443, grad_norm=45.228, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=2.494e-04, train_time=1.486
+[gpub010:0/16] 2024-01-30 02:25:56,663 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-01-30 02:26:15,697 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 02:26:19,294 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa751be75b0>)
+[gpub010:0/16] 2024-01-30 02:26:19,294 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-01-30 02:26:19,298 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 02:32:00,356 (trainer:737) INFO: 17epoch:train:1201-1300batch: iter_time=3.051, forward_time=0.343, loss_ctc=51.328, loss_att=47.665, acc=0.744, loss=48.764, backward_time=0.410, grad_norm=34.876, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=2.494e-04, train_time=4.524
+[gpub010:0/16] 2024-01-30 02:34:28,802 (trainer:737) INFO: 17epoch:train:1301-1400batch: iter_time=7.988e-05, forward_time=0.286, loss_ctc=51.717, loss_att=50.374, acc=0.698, loss=50.776, backward_time=0.398, grad_norm=36.452, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=2.493e-04, train_time=1.483
+[gpub010:0/16] 2024-01-30 02:37:12,765 (trainer:737) INFO: 17epoch:train:1401-1500batch: iter_time=7.879e-05, forward_time=0.316, loss_ctc=56.072, loss_att=60.231, acc=0.713, loss=58.983, backward_time=0.438, grad_norm=38.603, clip=100.000, loss_scale=4.387e+33, optim_step_time=0.095, optim0_lr0=2.493e-04, train_time=1.640
+[gpub010:0/16] 2024-01-30 02:39:33,633 (trainer:737) INFO: 17epoch:train:1501-1600batch: iter_time=8.438e-05, forward_time=0.341, loss_ctc=50.738, loss_att=48.791, acc=0.743, loss=49.375, backward_time=0.420, grad_norm=32.022, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.492e-04, train_time=1.409
+[gpub010:0/16] 2024-01-30 02:42:07,633 (trainer:737) INFO: 17epoch:train:1601-1700batch: iter_time=8.906e-05, forward_time=0.289, loss_ctc=51.227, loss_att=48.508, acc=0.713, loss=49.324, backward_time=0.406, grad_norm=34.176, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.492e-04, train_time=1.539
+[gpub010:0/16] 2024-01-30 02:44:41,331 (trainer:737) INFO: 17epoch:train:1701-1800batch: iter_time=8.050e-05, forward_time=0.331, loss_ctc=50.563, loss_att=43.805, acc=0.750, loss=45.832, backward_time=0.420, grad_norm=31.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.491e-04, train_time=1.537
+[gpub010:0/16] 2024-01-30 02:47:20,872 (trainer:737) INFO: 17epoch:train:1801-1900batch: iter_time=8.203e-05, forward_time=0.336, loss_ctc=55.728, loss_att=48.591, acc=0.732, loss=50.732, backward_time=0.413, grad_norm=45.122, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.491e-04, train_time=1.596
+[gpub010:0/16] 2024-01-30 02:49:55,282 (trainer:737) INFO: 17epoch:train:1901-2000batch: iter_time=4.852e-04, forward_time=0.301, loss_ctc=49.132, loss_att=48.246, acc=0.718, loss=48.512, backward_time=0.406, grad_norm=33.375, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.490e-04, train_time=1.543
+[gpub010:0/16] 2024-01-30 02:52:20,027 (trainer:737) INFO: 17epoch:train:2001-2100batch: iter_time=8.497e-05, forward_time=0.331, loss_ctc=52.272, loss_att=45.368, acc=0.754, loss=47.439, backward_time=0.434, grad_norm=30.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.489e-04, train_time=1.447
+[gpub010:0/16] 2024-01-30 02:55:04,271 (trainer:737) INFO: 17epoch:train:2101-2200batch: iter_time=8.137e-05, forward_time=0.316, loss_ctc=49.356, loss_att=52.533, acc=0.723, loss=51.580, backward_time=0.411, grad_norm=32.220, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.489e-04, train_time=1.643
+[gpub010:0/16] 2024-01-30 02:57:25,630 (trainer:737) INFO: 17epoch:train:2201-2300batch: iter_time=4.086e-04, forward_time=0.321, loss_ctc=45.318, loss_att=40.536, acc=0.753, loss=41.971, backward_time=0.426, grad_norm=29.660, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.488e-04, train_time=1.413
+[gpub010:0/16] 2024-01-30 02:59:59,549 (trainer:737) INFO: 17epoch:train:2301-2400batch: iter_time=8.888e-05, forward_time=0.345, loss_ctc=64.090, loss_att=64.417, acc=0.689, loss=64.319, backward_time=0.416, grad_norm=45.686, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.488e-04, train_time=1.539
+[gpub010:0/16] 2024-01-30 03:02:38,384 (trainer:737) INFO: 17epoch:train:2401-2500batch: iter_time=8.973e-05, forward_time=0.291, loss_ctc=52.552, loss_att=42.806, acc=0.760, loss=45.730, backward_time=0.407, grad_norm=32.253, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.487e-04, train_time=1.589
+[gpub010:0/16] 2024-01-30 03:02:58,542 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-01-30 03:03:17,653 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 03:03:21,163 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb00c427670>)
+[gpub010:0/16] 2024-01-30 03:03:21,163 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-01-30 03:03:21,167 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 03:13:10,182 (trainer:737) INFO: 17epoch:train:2501-2600batch: iter_time=3.053, forward_time=0.342, loss_ctc=47.470, loss_att=48.278, acc=0.712, loss=48.036, backward_time=0.405, grad_norm=33.773, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.487e-04, train_time=6.317
+[gpub010:0/16] 2024-01-30 03:15:58,792 (trainer:737) INFO: 17epoch:train:2601-2700batch: iter_time=8.546e-05, forward_time=0.321, loss_ctc=53.639, loss_att=49.658, acc=0.721, loss=50.852, backward_time=0.403, grad_norm=34.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.486e-04, train_time=1.687
+[gpub010:0/16] 2024-01-30 03:18:39,143 (trainer:737) INFO: 17epoch:train:2701-2800batch: iter_time=8.819e-05, forward_time=0.291, loss_ctc=50.927, loss_att=54.248, acc=0.733, loss=53.252, backward_time=0.407, grad_norm=31.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.486e-04, train_time=1.603
+[gpub010:0/16] 2024-01-30 03:21:33,326 (trainer:737) INFO: 17epoch:train:2801-2900batch: iter_time=8.574e-05, forward_time=0.350, loss_ctc=55.032, loss_att=54.038, acc=0.725, loss=54.336, backward_time=0.442, grad_norm=35.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.485e-04, train_time=1.741
+[gpub010:0/16] 2024-01-30 03:24:07,305 (trainer:737) INFO: 17epoch:train:2901-3000batch: iter_time=9.069e-05, forward_time=0.306, loss_ctc=47.875, loss_att=44.703, acc=0.717, loss=45.655, backward_time=0.406, grad_norm=33.477, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.485e-04, train_time=1.540
+[gpub010:0/16] 2024-01-30 03:26:48,793 (trainer:737) INFO: 17epoch:train:3001-3100batch: iter_time=8.428e-05, forward_time=0.338, loss_ctc=52.657, loss_att=47.334, acc=0.752, loss=48.931, backward_time=0.455, grad_norm=36.935, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.484e-04, train_time=1.615
+[gpub010:0/16] 2024-01-30 03:29:40,917 (trainer:737) INFO: 17epoch:train:3101-3200batch: iter_time=8.992e-05, forward_time=0.290, loss_ctc=52.479, loss_att=48.513, acc=0.736, loss=49.703, backward_time=0.401, grad_norm=32.149, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.484e-04, train_time=1.720
+[gpub010:0/16] 2024-01-30 03:32:16,602 (trainer:737) INFO: 17epoch:train:3201-3300batch: iter_time=9.040e-05, forward_time=0.313, loss_ctc=49.183, loss_att=42.384, acc=0.743, loss=44.423, backward_time=0.404, grad_norm=34.835, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.483e-04, train_time=1.557
+[gpub010:0/16] 2024-01-30 03:35:01,339 (trainer:737) INFO: 17epoch:train:3301-3400batch: iter_time=2.826e-04, forward_time=0.351, loss_ctc=51.735, loss_att=54.236, acc=0.733, loss=53.486, backward_time=0.458, grad_norm=30.636, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.483e-04, train_time=1.648
+[gpub010:0/16] 2024-01-30 03:37:54,316 (trainer:737) INFO: 17epoch:train:3401-3500batch: iter_time=9.129e-05, forward_time=0.303, loss_ctc=46.444, loss_att=42.331, acc=0.746, loss=43.565, backward_time=0.411, grad_norm=31.211, clip=100.000, loss_scale=8.775e+33, optim_step_time=0.093, optim0_lr0=2.482e-04, train_time=1.729
+[gpub010:0/16] 2024-01-30 03:40:31,855 (trainer:737) INFO: 17epoch:train:3501-3600batch: iter_time=8.487e-05, forward_time=0.287, loss_ctc=50.250, loss_att=48.475, acc=0.731, loss=49.008, backward_time=0.399, grad_norm=33.129, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.482e-04, train_time=1.576
+[gpub010:0/16] 2024-01-30 03:43:17,453 (trainer:737) INFO: 17epoch:train:3601-3700batch: iter_time=4.530e-04, forward_time=0.361, loss_ctc=60.595, loss_att=57.329, acc=0.714, loss=58.309, backward_time=0.463, grad_norm=43.628, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.481e-04, train_time=1.655
+[gpub010:0/16] 2024-01-30 03:44:56,802 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-01-30 03:45:15,504 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 03:45:19,019 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb0d1088e50>)
+[gpub010:0/16] 2024-01-30 03:45:19,019 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-01-30 03:45:19,022 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 03:51:09,069 (trainer:737) INFO: 17epoch:train:3701-3800batch: iter_time=3.064, forward_time=0.318, loss_ctc=48.852, loss_att=45.236, acc=0.755, loss=46.321, backward_time=0.404, grad_norm=33.157, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.481e-04, train_time=4.716
+[gpub010:0/16] 2024-01-30 03:53:36,205 (trainer:737) INFO: 17epoch:train:3801-3900batch: iter_time=8.472e-05, forward_time=0.289, loss_ctc=51.057, loss_att=50.364, acc=0.712, loss=50.572, backward_time=0.406, grad_norm=35.005, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.480e-04, train_time=1.471
+[gpub010:0/16] 2024-01-30 03:56:21,187 (trainer:737) INFO: 17epoch:train:3901-4000batch: iter_time=5.486e-04, forward_time=0.408, loss_ctc=54.336, loss_att=60.517, acc=0.722, loss=58.663, backward_time=0.433, grad_norm=36.156, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=2.480e-04, train_time=1.649
+[gpub010:0/16] 2024-01-30 03:59:02,217 (trainer:737) INFO: 17epoch:train:4001-4100batch: iter_time=8.329e-05, forward_time=0.291, loss_ctc=49.315, loss_att=47.468, acc=0.757, loss=48.022, backward_time=0.401, grad_norm=29.695, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.479e-04, train_time=1.610
+[gpub010:0/16] 2024-01-30 04:01:29,617 (trainer:737) INFO: 17epoch:train:4101-4200batch: iter_time=9.096e-05, forward_time=0.289, loss_ctc=50.188, loss_att=48.277, acc=0.723, loss=48.850, backward_time=0.407, grad_norm=31.474, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.479e-04, train_time=1.474
+[gpub010:0/16] 2024-01-30 04:04:21,163 (trainer:737) INFO: 17epoch:train:4201-4300batch: iter_time=3.848e-04, forward_time=0.354, loss_ctc=49.742, loss_att=42.961, acc=0.764, loss=44.996, backward_time=0.463, grad_norm=30.241, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.478e-04, train_time=1.715
+[gpub010:0/16] 2024-01-30 04:06:37,557 (trainer:737) INFO: 17epoch:train:4301-4400batch: iter_time=8.600e-05, forward_time=0.290, loss_ctc=53.105, loss_att=51.809, acc=0.730, loss=52.197, backward_time=0.401, grad_norm=38.962, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.478e-04, train_time=1.363
+[gpub010:0/16] 2024-01-30 04:09:21,713 (trainer:737) INFO: 17epoch:train:4401-4500batch: iter_time=8.855e-05, forward_time=0.317, loss_ctc=48.449, loss_att=46.780, acc=0.732, loss=47.281, backward_time=0.418, grad_norm=32.035, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.477e-04, train_time=1.642
+[gpub010:0/16] 2024-01-30 04:11:57,018 (trainer:737) INFO: 17epoch:train:4501-4600batch: iter_time=3.885e-04, forward_time=0.348, loss_ctc=51.421, loss_att=44.695, acc=0.767, loss=46.713, backward_time=0.435, grad_norm=30.809, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=2.477e-04, train_time=1.552
+[gpub010:0/16] 2024-01-30 04:14:26,964 (trainer:737) INFO: 17epoch:train:4601-4700batch: iter_time=8.019e-05, forward_time=0.294, loss_ctc=48.732, loss_att=52.391, acc=0.739, loss=51.293, backward_time=0.412, grad_norm=30.653, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.476e-04, train_time=1.499
+[gpub010:0/16] 2024-01-30 04:17:33,992 (trainer:737) INFO: 17epoch:train:4701-4800batch: iter_time=3.358e-04, forward_time=0.389, loss_ctc=44.670, loss_att=39.883, acc=0.764, loss=41.319, backward_time=0.418, grad_norm=29.438, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=2.476e-04, train_time=1.871
+[gpub010:0/16] 2024-01-30 04:19:53,340 (trainer:737) INFO: 17epoch:train:4801-4900batch: iter_time=9.112e-05, forward_time=0.292, loss_ctc=62.875, loss_att=65.411, acc=0.709, loss=64.650, backward_time=0.406, grad_norm=49.083, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.475e-04, train_time=1.393
+[gpub010:0/16] 2024-01-30 04:22:40,036 (trainer:737) INFO: 17epoch:train:4901-5000batch: iter_time=8.803e-05, forward_time=0.398, loss_ctc=51.763, loss_att=43.537, acc=0.769, loss=46.005, backward_time=0.424, grad_norm=30.649, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.126, optim0_lr0=2.475e-04, train_time=1.667
+[gpub010:0/16] 2024-01-30 04:23:00,170 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-01-30 04:23:18,993 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 04:23:22,556 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1096975b0>)
+[gpub010:0/16] 2024-01-30 04:23:22,556 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-01-30 04:23:22,559 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 04:33:29,753 (trainer:737) INFO: 17epoch:train:5001-5100batch: iter_time=3.062, forward_time=0.287, loss_ctc=46.950, loss_att=47.563, acc=0.727, loss=47.379, backward_time=0.399, grad_norm=33.112, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.474e-04, train_time=6.497
+[gpub010:0/16] 2024-01-30 04:36:06,799 (trainer:737) INFO: 17epoch:train:5101-5200batch: iter_time=7.984e-05, forward_time=0.352, loss_ctc=53.175, loss_att=49.437, acc=0.727, loss=50.558, backward_time=0.443, grad_norm=33.871, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.474e-04, train_time=1.570
+[gpub010:0/16] 2024-01-30 04:38:42,934 (trainer:737) INFO: 17epoch:train:5201-5300batch: iter_time=8.706e-05, forward_time=0.317, loss_ctc=49.950, loss_att=53.183, acc=0.744, loss=52.213, backward_time=0.415, grad_norm=30.667, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.473e-04, train_time=1.561
+[gpub010:0/16] 2024-01-30 04:41:30,577 (trainer:737) INFO: 17epoch:train:5301-5400batch: iter_time=8.556e-05, forward_time=0.291, loss_ctc=53.494, loss_att=53.270, acc=0.737, loss=53.337, backward_time=0.403, grad_norm=32.581, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.473e-04, train_time=1.676
+[gpub010:0/16] 2024-01-30 04:43:21,516 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 04:43:57,171 (trainer:737) INFO: 17epoch:train:5401-5500batch: iter_time=8.569e-05, forward_time=0.355, loss_ctc=46.924, loss_att=43.719, acc=0.731, loss=44.681, backward_time=0.465, grad_norm=31.249, clip=100.000, loss_scale=1.469e+34, optim_step_time=0.103, optim0_lr0=2.472e-04, train_time=1.466
+[gpub010:0/16] 2024-01-30 04:46:36,110 (trainer:737) INFO: 17epoch:train:5501-5600batch: iter_time=8.246e-05, forward_time=0.291, loss_ctc=52.564, loss_att=49.067, acc=0.758, loss=50.116, backward_time=0.405, grad_norm=39.183, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.472e-04, train_time=1.589
+[gpub010:0/16] 2024-01-30 04:49:23,588 (trainer:737) INFO: 17epoch:train:5601-5700batch: iter_time=8.678e-05, forward_time=0.288, loss_ctc=52.258, loss_att=50.010, acc=0.736, loss=50.685, backward_time=0.400, grad_norm=31.866, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.471e-04, train_time=1.675
+[gpub010:0/16] 2024-01-30 04:52:00,361 (trainer:737) INFO: 17epoch:train:5701-5800batch: iter_time=9.119e-05, forward_time=0.382, loss_ctc=48.555, loss_att=41.684, acc=0.753, loss=43.746, backward_time=0.438, grad_norm=32.444, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.471e-04, train_time=1.567
+[gpub010:0/16] 2024-01-30 04:54:18,546 (trainer:737) INFO: 17epoch:train:5801-5900batch: iter_time=8.625e-05, forward_time=0.294, loss_ctc=51.256, loss_att=54.823, acc=0.748, loss=53.753, backward_time=0.408, grad_norm=30.044, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.470e-04, train_time=1.382
+[gpub010:0/16] 2024-01-30 04:57:21,374 (trainer:737) INFO: 17epoch:train:5901-6000batch: iter_time=8.748e-05, forward_time=0.368, loss_ctc=45.968, loss_att=42.315, acc=0.757, loss=43.411, backward_time=0.452, grad_norm=29.723, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.470e-04, train_time=1.828
+[gpub010:0/16] 2024-01-30 04:59:51,782 (trainer:737) INFO: 17epoch:train:6001-6100batch: iter_time=8.804e-05, forward_time=0.288, loss_ctc=48.693, loss_att=47.555, acc=0.741, loss=47.897, backward_time=0.402, grad_norm=31.975, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.469e-04, train_time=1.503
+[gpub010:0/16] 2024-01-30 05:02:06,153 (trainer:737) INFO: 17epoch:train:6101-6200batch: iter_time=9.449e-05, forward_time=0.293, loss_ctc=59.736, loss_att=57.162, acc=0.733, loss=57.935, backward_time=0.405, grad_norm=38.896, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.469e-04, train_time=1.344
+[gpub010:0/16] 2024-01-30 05:04:06,823 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-01-30 05:04:26,061 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 05:04:29,633 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae5b243700>)
+[gpub010:0/16] 2024-01-30 05:04:29,633 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-01-30 05:04:29,637 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 05:09:58,439 (trainer:737) INFO: 17epoch:train:6201-6300batch: iter_time=2.929, forward_time=0.385, loss_ctc=47.853, loss_att=45.289, acc=0.754, loss=46.058, backward_time=0.448, grad_norm=31.188, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.468e-04, train_time=4.722
+[gpub010:0/16] 2024-01-30 05:12:08,580 (trainer:737) INFO: 17epoch:train:6301-6400batch: iter_time=8.746e-05, forward_time=0.288, loss_ctc=50.162, loss_att=48.055, acc=0.719, loss=48.687, backward_time=0.400, grad_norm=33.785, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.468e-04, train_time=1.302
+[gpub010:0/16] 2024-01-30 05:15:11,929 (trainer:737) INFO: 17epoch:train:6401-6500batch: iter_time=9.630e-05, forward_time=0.351, loss_ctc=53.806, loss_att=59.169, acc=0.726, loss=57.560, backward_time=0.477, grad_norm=34.770, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.467e-04, train_time=1.833
+[gpub010:0/16] 2024-01-30 05:17:45,226 (trainer:737) INFO: 17epoch:train:6501-6600batch: iter_time=8.734e-05, forward_time=0.312, loss_ctc=49.167, loss_att=46.959, acc=0.759, loss=47.621, backward_time=0.405, grad_norm=29.022, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.467e-04, train_time=1.532
+[gpub010:0/16] 2024-01-30 05:20:18,175 (trainer:737) INFO: 17epoch:train:6601-6700batch: iter_time=9.118e-05, forward_time=0.398, loss_ctc=49.409, loss_att=47.571, acc=0.726, loss=48.123, backward_time=0.432, grad_norm=33.404, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.111, optim0_lr0=2.466e-04, train_time=1.530
+[gpub010:0/16] 2024-01-30 05:22:59,359 (trainer:737) INFO: 17epoch:train:6701-6800batch: iter_time=1.020e-04, forward_time=0.289, loss_ctc=48.645, loss_att=42.090, acc=0.767, loss=44.056, backward_time=0.400, grad_norm=29.492, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.466e-04, train_time=1.612
+[gpub010:0/16] 2024-01-30 05:25:46,450 (trainer:737) INFO: 17epoch:train:6801-6900batch: iter_time=8.745e-05, forward_time=0.404, loss_ctc=52.126, loss_att=50.653, acc=0.733, loss=51.095, backward_time=0.426, grad_norm=41.329, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.136, optim0_lr0=2.465e-04, train_time=1.671
+[gpub010:0/16] 2024-01-30 05:28:01,714 (trainer:737) INFO: 17epoch:train:6901-7000batch: iter_time=9.352e-05, forward_time=0.301, loss_ctc=47.961, loss_att=46.526, acc=0.733, loss=46.957, backward_time=0.403, grad_norm=33.431, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.465e-04, train_time=1.353
+[gpub010:0/16] 2024-01-30 05:28:27,764 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 05:30:36,300 (trainer:737) INFO: 17epoch:train:7001-7100batch: iter_time=8.916e-05, forward_time=0.291, loss_ctc=50.969, loss_att=43.813, acc=0.771, loss=45.960, backward_time=0.403, grad_norm=31.476, clip=100.000, loss_scale=6.189e+33, optim_step_time=0.092, optim0_lr0=2.464e-04, train_time=1.546
+[gpub010:0/16] 2024-01-30 05:33:32,323 (trainer:737) INFO: 17epoch:train:7101-7200batch: iter_time=6.149e-04, forward_time=0.401, loss_ctc=48.434, loss_att=51.490, acc=0.741, loss=50.573, backward_time=0.424, grad_norm=30.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=2.464e-04, train_time=1.760
+[gpub010:0/16] 2024-01-30 05:35:47,909 (trainer:737) INFO: 17epoch:train:7201-7300batch: iter_time=8.637e-05, forward_time=0.290, loss_ctc=44.591, loss_att=39.465, acc=0.766, loss=41.002, backward_time=0.402, grad_norm=28.016, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.463e-04, train_time=1.356
+[gpub010:0/16] 2024-01-30 05:38:33,718 (trainer:737) INFO: 17epoch:train:7301-7400batch: iter_time=1.010e-04, forward_time=0.371, loss_ctc=60.988, loss_att=63.547, acc=0.712, loss=62.780, backward_time=0.487, grad_norm=42.464, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=2.463e-04, train_time=1.658
+[gpub010:0/16] 2024-01-30 05:41:20,016 (trainer:737) INFO: 17epoch:train:7401-7500batch: iter_time=8.519e-05, forward_time=0.290, loss_ctc=51.134, loss_att=42.928, acc=0.770, loss=45.390, backward_time=0.402, grad_norm=29.602, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.462e-04, train_time=1.663
+[gpub010:0/16] 2024-01-30 05:41:40,043 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-01-30 05:41:59,477 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 05:42:03,088 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae2d8f3bb0>)
+[gpub010:0/16] 2024-01-30 05:42:03,088 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-01-30 05:42:03,091 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 05:48:52,552 (trainer:737) INFO: 17epoch:train:7501-7600batch: iter_time=3.053, forward_time=0.340, loss_ctc=46.790, loss_att=51.009, acc=0.711, loss=49.744, backward_time=0.468, grad_norm=35.464, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.462e-04, train_time=4.525
+[gpub010:0/16] 2024-01-30 05:51:25,143 (trainer:737) INFO: 17epoch:train:7601-7700batch: iter_time=8.334e-05, forward_time=0.289, loss_ctc=52.616, loss_att=49.745, acc=0.723, loss=50.607, backward_time=0.402, grad_norm=35.005, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.461e-04, train_time=1.526
+[gpub010:0/16] 2024-01-30 05:54:08,700 (trainer:737) INFO: 17epoch:train:7701-7800batch: iter_time=8.235e-05, forward_time=0.314, loss_ctc=50.404, loss_att=54.392, acc=0.736, loss=53.196, backward_time=0.457, grad_norm=31.962, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.461e-04, train_time=1.635
+[gpub010:0/16] 2024-01-30 05:56:30,501 (trainer:737) INFO: 17epoch:train:7801-7900batch: iter_time=8.625e-05, forward_time=0.341, loss_ctc=53.384, loss_att=53.487, acc=0.727, loss=53.456, backward_time=0.409, grad_norm=33.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.460e-04, train_time=1.418
+[gpub010:0/16] 2024-01-30 05:58:41,423 (trainer:737) INFO: 17epoch:train:7901-8000batch: iter_time=8.697e-05, forward_time=0.287, loss_ctc=46.767, loss_att=44.416, acc=0.718, loss=45.121, backward_time=0.399, grad_norm=32.300, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.460e-04, train_time=1.309
+[gpub010:0/16] 2024-01-30 06:01:41,229 (trainer:737) INFO: 17epoch:train:8001-8100batch: iter_time=2.087e-04, forward_time=0.353, loss_ctc=50.321, loss_att=46.873, acc=0.754, loss=47.907, backward_time=0.493, grad_norm=35.649, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.459e-04, train_time=1.797
+[gpub010:0/16] 2024-01-30 06:04:14,128 (trainer:737) INFO: 17epoch:train:8101-8200batch: iter_time=8.759e-05, forward_time=0.288, loss_ctc=51.721, loss_att=48.422, acc=0.739, loss=49.412, backward_time=0.400, grad_norm=32.019, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.459e-04, train_time=1.528
+[gpub010:0/16] 2024-01-30 06:06:44,438 (trainer:737) INFO: 17epoch:train:8201-8300batch: iter_time=8.452e-05, forward_time=0.411, loss_ctc=47.785, loss_att=42.311, acc=0.744, loss=43.953, backward_time=0.446, grad_norm=40.768, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.458e-04, train_time=1.504
+[gpub010:0/16] 2024-01-30 06:09:08,984 (trainer:737) INFO: 17epoch:train:8301-8400batch: iter_time=8.171e-05, forward_time=0.292, loss_ctc=51.161, loss_att=54.424, acc=0.733, loss=53.445, backward_time=0.404, grad_norm=31.302, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.458e-04, train_time=1.445
+[gpub010:0/16] 2024-01-30 06:12:00,218 (trainer:737) INFO: 17epoch:train:8401-8500batch: iter_time=8.532e-05, forward_time=0.305, loss_ctc=45.591, loss_att=42.267, acc=0.749, loss=43.264, backward_time=0.418, grad_norm=29.736, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=2.457e-04, train_time=1.713
+[gpub010:0/16] 2024-01-30 06:14:22,676 (trainer:737) INFO: 17epoch:train:8501-8600batch: iter_time=8.883e-05, forward_time=0.344, loss_ctc=48.648, loss_att=48.194, acc=0.733, loss=48.330, backward_time=0.448, grad_norm=32.024, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.457e-04, train_time=1.424
+[gpub010:0/16] 2024-01-30 06:16:52,885 (trainer:737) INFO: 17epoch:train:8601-8700batch: iter_time=9.290e-05, forward_time=0.292, loss_ctc=60.449, loss_att=57.372, acc=0.717, loss=58.295, backward_time=0.404, grad_norm=43.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.456e-04, train_time=1.501
+[gpub010:0/16] 2024-01-30 06:18:20,138 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-01-30 06:18:39,047 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 06:18:42,641 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae59c975b0>)
+[gpub010:0/16] 2024-01-30 06:18:42,641 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-01-30 06:18:42,645 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 06:24:21,983 (trainer:737) INFO: 17epoch:train:8701-8800batch: iter_time=3.063, forward_time=0.366, loss_ctc=48.018, loss_att=44.052, acc=0.759, loss=45.242, backward_time=0.438, grad_norm=32.494, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=2.456e-04, train_time=4.491
+[gpub010:0/16] 2024-01-30 06:26:31,693 (trainer:737) INFO: 17epoch:train:8801-8900batch: iter_time=8.119e-05, forward_time=0.288, loss_ctc=49.829, loss_att=48.423, acc=0.718, loss=48.845, backward_time=0.400, grad_norm=34.730, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.455e-04, train_time=1.297
+[gpub010:0/16] 2024-01-30 06:28:59,759 (trainer:737) INFO: 17epoch:train:8901-9000batch: iter_time=8.704e-05, forward_time=0.297, loss_ctc=53.263, loss_att=60.196, acc=0.724, loss=58.116, backward_time=0.411, grad_norm=34.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.455e-04, train_time=1.480
+[gpub010:0/16] 2024-01-30 06:31:59,089 (trainer:737) INFO: 17epoch:train:9001-9100batch: iter_time=8.563e-05, forward_time=0.397, loss_ctc=49.114, loss_att=47.138, acc=0.758, loss=47.731, backward_time=0.433, grad_norm=30.292, clip=100.000, loss_scale=9.346e+33, optim_step_time=0.099, optim0_lr0=2.454e-04, train_time=1.794
+[gpub010:0/16] 2024-01-30 06:34:09,424 (trainer:737) INFO: 17epoch:train:9101-9200batch: iter_time=8.526e-05, forward_time=0.291, loss_ctc=48.879, loss_att=47.437, acc=0.728, loss=47.870, backward_time=0.403, grad_norm=31.994, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.454e-04, train_time=1.303
+[gpub010:0/16] 2024-01-30 06:34:18,539 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 06:37:07,051 (trainer:737) INFO: 17epoch:train:9201-9300batch: iter_time=8.243e-05, forward_time=0.364, loss_ctc=49.035, loss_att=42.139, acc=0.766, loss=44.207, backward_time=0.450, grad_norm=33.210, clip=100.000, loss_scale=5.507e+33, optim_step_time=0.102, optim0_lr0=2.453e-04, train_time=1.776
+[gpub010:0/16] 2024-01-30 06:39:18,112 (trainer:737) INFO: 17epoch:train:9301-9400batch: iter_time=8.286e-05, forward_time=0.290, loss_ctc=51.809, loss_att=50.254, acc=0.737, loss=50.720, backward_time=0.404, grad_norm=42.075, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.453e-04, train_time=1.311
+[gpub010:0/16] 2024-01-30 06:41:56,762 (trainer:737) INFO: 17epoch:train:9401-9500batch: iter_time=8.214e-05, forward_time=0.346, loss_ctc=47.421, loss_att=45.906, acc=0.736, loss=46.360, backward_time=0.452, grad_norm=32.360, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=2.452e-04, train_time=1.586
+[gpub010:0/16] 2024-01-30 06:44:33,077 (trainer:737) INFO: 17epoch:train:9501-9600batch: iter_time=9.182e-05, forward_time=0.291, loss_ctc=50.588, loss_att=43.841, acc=0.770, loss=45.865, backward_time=0.402, grad_norm=30.378, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.452e-04, train_time=1.563
+[gpub010:0/16] 2024-01-30 06:46:46,763 (trainer:737) INFO: 17epoch:train:9601-9700batch: iter_time=8.082e-05, forward_time=0.293, loss_ctc=48.084, loss_att=51.301, acc=0.741, loss=50.336, backward_time=0.408, grad_norm=30.804, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.451e-04, train_time=1.337
+[gpub010:0/16] 2024-01-30 06:49:50,601 (trainer:737) INFO: 17epoch:train:9701-9800batch: iter_time=9.727e-05, forward_time=0.386, loss_ctc=44.279, loss_att=39.678, acc=0.766, loss=41.059, backward_time=0.438, grad_norm=29.651, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=2.451e-04, train_time=1.837
+[gpub010:0/16] 2024-01-30 06:52:05,828 (trainer:737) INFO: 17epoch:train:9801-9900batch: iter_time=8.820e-05, forward_time=0.293, loss_ctc=60.891, loss_att=63.871, acc=0.713, loss=62.977, backward_time=0.406, grad_norm=46.643, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.450e-04, train_time=1.353
+[gpub010:0/16] 2024-01-30 06:54:34,231 (trainer:737) INFO: 17epoch:train:9901-10000batch: iter_time=8.748e-05, forward_time=0.375, loss_ctc=50.294, loss_att=43.255, acc=0.769, loss=45.367, backward_time=0.444, grad_norm=30.529, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.450e-04, train_time=1.483
+[gpub010:0/16] 2024-01-30 06:54:54,335 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-01-30 06:55:13,672 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 06:55:17,273 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae2f86f460>)
+[gpub010:0/16] 2024-01-30 06:55:17,274 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-30 06:55:17,277 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 07:02:29,202 (trainer:737) INFO: 17epoch:train:10001-10100batch: iter_time=3.221, forward_time=0.297, loss_ctc=46.435, loss_att=47.307, acc=0.729, loss=47.046, backward_time=0.399, grad_norm=32.798, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.449e-04, train_time=4.750
+[gpub010:0/16] 2024-01-30 07:04:40,864 (trainer:737) INFO: 17epoch:train:10101-10200batch: iter_time=8.024e-05, forward_time=0.289, loss_ctc=51.972, loss_att=48.238, acc=0.732, loss=49.358, backward_time=0.401, grad_norm=34.389, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=2.449e-04, train_time=1.316
+[gpub010:0/16] 2024-01-30 07:07:37,067 (trainer:737) INFO: 17epoch:train:10201-10300batch: iter_time=8.770e-05, forward_time=0.424, loss_ctc=49.813, loss_att=52.767, acc=0.747, loss=51.881, backward_time=0.423, grad_norm=31.003, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=2.448e-04, train_time=1.762
+[gpub010:0/16] 2024-01-30 07:10:08,190 (trainer:737) INFO: 17epoch:train:10301-10400batch: iter_time=8.234e-05, forward_time=0.292, loss_ctc=52.702, loss_att=52.378, acc=0.741, loss=52.475, backward_time=0.403, grad_norm=31.518, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.448e-04, train_time=1.511
+[gpub010:0/16] 2024-01-30 07:12:35,528 (trainer:737) INFO: 17epoch:train:10401-10500batch: iter_time=8.534e-05, forward_time=0.356, loss_ctc=46.331, loss_att=43.224, acc=0.731, loss=44.156, backward_time=0.443, grad_norm=31.294, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=2.447e-04, train_time=1.473
+[gpub010:0/16] 2024-01-30 07:14:55,622 (trainer:737) INFO: 17epoch:train:10501-10600batch: iter_time=8.167e-05, forward_time=0.294, loss_ctc=49.825, loss_att=48.913, acc=0.755, loss=49.187, backward_time=0.401, grad_norm=35.973, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.447e-04, train_time=1.400
+[gpub010:0/16] 2024-01-30 07:17:40,595 (trainer:737) INFO: 17epoch:train:10601-10700batch: iter_time=8.352e-05, forward_time=0.289, loss_ctc=51.509, loss_att=49.043, acc=0.741, loss=49.783, backward_time=0.399, grad_norm=30.707, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.446e-04, train_time=1.650
+[gpub010:0/16] 2024-01-30 07:20:11,022 (trainer:737) INFO: 17epoch:train:10701-10800batch: iter_time=8.665e-05, forward_time=0.356, loss_ctc=47.140, loss_att=40.998, acc=0.758, loss=42.841, backward_time=0.481, grad_norm=32.030, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=2.446e-04, train_time=1.504
+[gpub010:0/16] 2024-01-30 07:22:32,188 (trainer:737) INFO: 17epoch:train:10801-10900batch: iter_time=8.657e-05, forward_time=0.292, loss_ctc=50.975, loss_att=53.840, acc=0.749, loss=52.980, backward_time=0.406, grad_norm=30.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.445e-04, train_time=1.410
+[gpub010:0/16] 2024-01-30 07:25:35,301 (trainer:737) INFO: 17epoch:train:10901-11000batch: iter_time=8.804e-05, forward_time=0.363, loss_ctc=45.244, loss_att=41.125, acc=0.762, loss=42.361, backward_time=0.463, grad_norm=29.660, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.445e-04, train_time=1.832
+[gpub010:0/16] 2024-01-30 07:27:47,908 (trainer:737) INFO: 17epoch:train:11001-11100batch: iter_time=8.480e-05, forward_time=0.305, loss_ctc=48.011, loss_att=47.533, acc=0.742, loss=47.677, backward_time=0.404, grad_norm=31.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.444e-04, train_time=1.326
+[gpub010:0/16] 2024-01-30 07:30:06,302 (trainer:737) INFO: 17epoch:train:11101-11200batch: iter_time=8.486e-05, forward_time=0.293, loss_ctc=58.793, loss_att=56.657, acc=0.735, loss=57.298, backward_time=0.406, grad_norm=38.237, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.444e-04, train_time=1.382
+[gpub010:0/16] 2024-01-30 07:31:56,155 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-01-30 07:32:15,700 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 07:32:19,283 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb11d625c60>)
+[gpub010:0/16] 2024-01-30 07:32:19,284 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-01-30 07:32:19,287 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 07:38:09,003 (trainer:737) INFO: 17epoch:train:11201-11300batch: iter_time=3.117, forward_time=0.387, loss_ctc=47.369, loss_att=45.117, acc=0.755, loss=45.793, backward_time=0.434, grad_norm=31.420, clip=100.000, loss_scale=1.002e+34, optim_step_time=0.097, optim0_lr0=2.443e-04, train_time=4.828
+[gpub010:0/16] 2024-01-30 07:40:18,328 (trainer:737) INFO: 17epoch:train:11301-11400batch: iter_time=8.134e-05, forward_time=0.286, loss_ctc=49.583, loss_att=49.735, acc=0.707, loss=49.690, backward_time=0.398, grad_norm=37.865, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.443e-04, train_time=1.293
+[gpub010:0/16] 2024-01-30 07:43:09,730 (trainer:737) INFO: 17epoch:train:11401-11500batch: iter_time=8.101e-05, forward_time=0.382, loss_ctc=53.368, loss_att=59.295, acc=0.722, loss=57.517, backward_time=0.480, grad_norm=36.883, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.443e-04, train_time=1.714
+[gpub010:0/16] 2024-01-30 07:46:03,777 (trainer:737) INFO: 17epoch:train:11501-11600batch: iter_time=8.121e-05, forward_time=0.404, loss_ctc=48.969, loss_att=48.523, acc=0.749, loss=48.657, backward_time=0.438, grad_norm=29.895, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.141, optim0_lr0=2.442e-04, train_time=1.740
+[gpub010:0/16] 2024-01-30 07:48:24,138 (trainer:737) INFO: 17epoch:train:11601-11700batch: iter_time=8.805e-05, forward_time=0.288, loss_ctc=48.679, loss_att=47.302, acc=0.720, loss=47.715, backward_time=0.400, grad_norm=31.873, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.442e-04, train_time=1.404
+[gpub010:0/16] 2024-01-30 07:51:11,307 (trainer:737) INFO: 17epoch:train:11701-11800batch: iter_time=8.434e-05, forward_time=0.348, loss_ctc=48.472, loss_att=42.777, acc=0.755, loss=44.485, backward_time=0.452, grad_norm=30.628, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=2.441e-04, train_time=1.672
+[gpub010:0/16] 2024-01-30 07:53:33,755 (trainer:737) INFO: 17epoch:train:11801-11900batch: iter_time=8.265e-05, forward_time=0.290, loss_ctc=51.464, loss_att=47.477, acc=0.741, loss=48.673, backward_time=0.401, grad_norm=37.359, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.441e-04, train_time=1.424
+[gpub010:0/16] 2024-01-30 07:55:59,250 (trainer:737) INFO: 17epoch:train:11901-12000batch: iter_time=8.259e-05, forward_time=0.288, loss_ctc=47.189, loss_att=47.508, acc=0.725, loss=47.412, backward_time=0.404, grad_norm=33.256, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=2.440e-04, train_time=1.455
+[gpub010:0/16] 2024-01-30 07:58:53,706 (trainer:737) INFO: 17epoch:train:12001-12100batch: iter_time=8.210e-05, forward_time=0.402, loss_ctc=50.390, loss_att=44.063, acc=0.761, loss=45.961, backward_time=0.419, grad_norm=30.309, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.440e-04, train_time=1.744
+[gpub010:0/16] 2024-01-30 08:01:20,086 (trainer:737) INFO: 17epoch:train:12101-12200batch: iter_time=8.386e-05, forward_time=0.290, loss_ctc=47.725, loss_att=51.234, acc=0.732, loss=50.181, backward_time=0.403, grad_norm=30.976, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.439e-04, train_time=1.464
+[gpub010:0/16] 2024-01-30 08:03:58,320 (trainer:737) INFO: 17epoch:train:12201-12300batch: iter_time=3.601e-04, forward_time=0.360, loss_ctc=44.387, loss_att=39.958, acc=0.757, loss=41.287, backward_time=0.477, grad_norm=29.197, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=2.439e-04, train_time=1.582
+[gpub010:0/16] 2024-01-30 08:06:43,312 (trainer:737) INFO: 17epoch:train:12301-12400batch: iter_time=8.791e-05, forward_time=0.291, loss_ctc=61.350, loss_att=63.557, acc=0.696, loss=62.895, backward_time=0.403, grad_norm=48.063, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.438e-04, train_time=1.650
+[gpub010:0/16] 2024-01-30 08:09:17,575 (trainer:737) INFO: 17epoch:train:12401-12500batch: iter_time=3.419e-04, forward_time=0.403, loss_ctc=50.386, loss_att=42.000, acc=0.766, loss=44.516, backward_time=0.430, grad_norm=30.980, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=2.438e-04, train_time=1.542
+[gpub010:0/16] 2024-01-30 08:09:37,701 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-01-30 08:09:57,075 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 08:10:00,669 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae8c017790>)
+[gpub010:0/16] 2024-01-30 08:10:00,670 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-01-30 08:10:00,673 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 08:17:05,791 (trainer:737) INFO: 17epoch:train:12501-12600batch: iter_time=3.232, forward_time=0.289, loss_ctc=45.755, loss_att=46.983, acc=0.719, loss=46.615, backward_time=0.399, grad_norm=32.561, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.437e-04, train_time=4.683
+[gpub010:0/16] 2024-01-30 08:19:44,948 (trainer:737) INFO: 17epoch:train:12601-12700batch: iter_time=8.354e-05, forward_time=0.394, loss_ctc=52.148, loss_att=48.528, acc=0.727, loss=49.614, backward_time=0.423, grad_norm=35.016, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=2.437e-04, train_time=1.591
+[gpub010:0/16] 2024-01-30 08:22:42,708 (trainer:737) INFO: 17epoch:train:12701-12800batch: iter_time=8.293e-05, forward_time=0.414, loss_ctc=49.339, loss_att=52.905, acc=0.740, loss=51.835, backward_time=0.426, grad_norm=31.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.436e-04, train_time=1.778
+[gpub010:0/16] 2024-01-30 08:25:12,482 (trainer:737) INFO: 17epoch:train:12801-12900batch: iter_time=9.131e-05, forward_time=0.291, loss_ctc=52.191, loss_att=52.214, acc=0.730, loss=52.207, backward_time=0.402, grad_norm=33.916, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.436e-04, train_time=1.497
+[gpub010:0/16] 2024-01-30 08:27:39,470 (trainer:737) INFO: 17epoch:train:12901-13000batch: iter_time=2.871e-04, forward_time=0.363, loss_ctc=45.823, loss_att=43.426, acc=0.721, loss=44.145, backward_time=0.458, grad_norm=32.330, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=2.435e-04, train_time=1.470
+[gpub010:0/16] 2024-01-30 08:30:34,264 (trainer:737) INFO: 17epoch:train:13001-13100batch: iter_time=8.595e-05, forward_time=0.289, loss_ctc=50.063, loss_att=47.131, acc=0.756, loss=48.011, backward_time=0.400, grad_norm=36.933, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.435e-04, train_time=1.747
+[gpub010:0/16] 2024-01-30 08:33:00,883 (trainer:737) INFO: 17epoch:train:13101-13200batch: iter_time=8.945e-05, forward_time=0.350, loss_ctc=51.462, loss_att=47.897, acc=0.740, loss=48.966, backward_time=0.497, grad_norm=33.054, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=2.434e-04, train_time=1.467
+[gpub010:0/16] 2024-01-30 08:33:19,098 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 08:35:26,216 (trainer:737) INFO: 17epoch:train:13201-13300batch: iter_time=8.766e-05, forward_time=0.377, loss_ctc=47.396, loss_att=41.684, acc=0.746, loss=43.397, backward_time=0.460, grad_norm=33.813, clip=100.000, loss_scale=1.101e+34, optim_step_time=0.100, optim0_lr0=2.434e-04, train_time=1.452
+[gpub010:0/16] 2024-01-30 08:38:19,660 (trainer:737) INFO: 17epoch:train:13301-13400batch: iter_time=9.236e-05, forward_time=0.291, loss_ctc=50.805, loss_att=53.764, acc=0.736, loss=52.877, backward_time=0.404, grad_norm=30.888, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.433e-04, train_time=1.735
+[gpub010:0/16] 2024-01-30 08:40:47,866 (trainer:737) INFO: 17epoch:train:13401-13500batch: iter_time=9.404e-05, forward_time=0.414, loss_ctc=45.024, loss_att=41.431, acc=0.753, loss=42.509, backward_time=0.446, grad_norm=29.553, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=2.433e-04, train_time=1.481
+[gpub010:0/16] 2024-01-30 08:43:14,291 (trainer:737) INFO: 17epoch:train:13501-13600batch: iter_time=5.582e-04, forward_time=0.394, loss_ctc=47.793, loss_att=47.163, acc=0.736, loss=47.352, backward_time=0.453, grad_norm=31.529, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=2.432e-04, train_time=1.465
+[gpub010:0/16] 2024-01-30 08:46:05,577 (trainer:737) INFO: 17epoch:train:13601-13700batch: iter_time=9.214e-05, forward_time=0.298, loss_ctc=58.587, loss_att=56.905, acc=0.719, loss=57.410, backward_time=0.405, grad_norm=43.462, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=2.432e-04, train_time=1.711
+[gpub010:0/16] 2024-01-30 08:47:35,641 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-01-30 08:47:55,191 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 08:47:58,809 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fae893e9ba0>)
+[gpub010:0/16] 2024-01-30 08:47:58,809 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-01-30 08:47:58,812 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-01-30 08:53:42,878 (trainer:737) INFO: 17epoch:train:13701-13800batch: iter_time=3.086, forward_time=0.385, loss_ctc=47.259, loss_att=43.060, acc=0.757, loss=44.320, backward_time=0.418, grad_norm=33.787, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=2.431e-04, train_time=4.574
+[gpub010:0/16] 2024-01-30 08:54:11,379 (trainer:668) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-01-30 08:56:08,900 (trainer:737) INFO: 17epoch:train:13801-13900batch: iter_time=8.068e-05, forward_time=0.289, loss_ctc=49.681, loss_att=48.018, acc=0.711, loss=48.517, backward_time=0.398, grad_norm=34.009, clip=100.000, loss_scale=6.294e+33, optim_step_time=0.092, optim0_lr0=2.431e-04, train_time=1.459
+[gpub010:0/16] 2024-01-30 08:58:48,261 (trainer:737) INFO: 17epoch:train:13901-14000batch: iter_time=8.264e-05, forward_time=0.342, loss_ctc=52.943, loss_att=57.802, acc=0.724, loss=56.345, backward_time=0.473, grad_norm=36.478, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=2.430e-04, train_time=1.594
+[gpub010:0/16] 2024-01-30 09:01:29,481 (trainer:737) INFO: 17epoch:train:14001-14100batch: iter_time=8.159e-05, forward_time=0.298, loss_ctc=48.296, loss_att=46.931, acc=0.754, loss=47.340, backward_time=0.404, grad_norm=30.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.430e-04, train_time=1.612
+[gpub010:0/16] 2024-01-30 09:03:43,679 (trainer:737) INFO: 17epoch:train:14101-14200batch: iter_time=8.440e-05, forward_time=0.288, loss_ctc=47.995, loss_att=46.646, acc=0.722, loss=47.051, backward_time=0.404, grad_norm=31.843, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.430e-04, train_time=1.341
+[gpub010:0/16] 2024-01-30 09:06:33,908 (trainer:737) INFO: 17epoch:train:14201-14300batch: iter_time=5.380e-04, forward_time=0.359, loss_ctc=48.540, loss_att=42.555, acc=0.756, loss=44.350, backward_time=0.460, grad_norm=31.517, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.429e-04, train_time=1.703
+[gpub010:0/16] 2024-01-30 09:09:05,134 (trainer:737) INFO: 17epoch:train:14301-14400batch: iter_time=8.407e-05, forward_time=0.295, loss_ctc=51.395, loss_att=46.818, acc=0.741, loss=48.191, backward_time=0.401, grad_norm=40.156, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.429e-04, train_time=1.512
+[gpub010:0/16] 2024-01-30 09:11:46,727 (trainer:737) INFO: 17epoch:train:14401-14500batch: iter_time=2.201e-04, forward_time=0.375, loss_ctc=47.642, loss_att=47.018, acc=0.726, loss=47.206, backward_time=0.429, grad_norm=33.009, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.428e-04, train_time=1.616
+[gpub010:0/16] 2024-01-30 09:14:14,173 (trainer:737) INFO: 17epoch:train:14501-14600batch: iter_time=8.159e-05, forward_time=0.308, loss_ctc=50.106, loss_att=43.909, acc=0.761, loss=45.768, backward_time=0.403, grad_norm=29.904, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=2.428e-04, train_time=1.475
+[gpub010:0/16] 2024-01-30 09:16:57,484 (trainer:737) INFO: 17epoch:train:14601-14700batch: iter_time=8.467e-05, forward_time=0.291, loss_ctc=47.165, loss_att=50.381, acc=0.733, loss=49.416, backward_time=0.428, grad_norm=31.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.427e-04, train_time=1.633
+[gpub010:0/16] 2024-01-30 09:19:29,908 (trainer:737) INFO: 17epoch:train:14701-14800batch: iter_time=8.566e-05, forward_time=0.387, loss_ctc=43.923, loss_att=39.562, acc=0.759, loss=40.870, backward_time=0.422, grad_norm=28.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=2.427e-04, train_time=1.524
+[gpub010:0/16] 2024-01-30 09:22:11,490 (trainer:737) INFO: 17epoch:train:14801-14900batch: iter_time=8.169e-05, forward_time=0.292, loss_ctc=60.219, loss_att=63.252, acc=0.701, loss=62.342, backward_time=0.402, grad_norm=44.449, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=2.426e-04, train_time=1.616
+[gpub010:0/16] 2024-01-30 09:25:09,514 (trainer:737) INFO: 17epoch:train:14901-15000batch: iter_time=2.362e-04, forward_time=0.391, loss_ctc=50.278, loss_att=41.496, acc=0.767, loss=44.130, backward_time=0.437, grad_norm=31.385, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=2.426e-04, train_time=1.780
+[gpub010:0/16] 2024-01-30 10:02:59,919 (trainer:343) INFO: 17epoch results: [train] iter_time=0.245, forward_time=0.328, loss_ctc=50.768, loss_att=48.847, acc=0.738, loss=49.423, backward_time=0.422, grad_norm=33.867, clip=100.000, loss_scale=7.188e+33, optim_step_time=0.097, optim0_lr0=2.462e-04, train_time=1.820, time=7 hours, 35 minutes and 25.16 seconds, total_count=285000, gpu_max_cached_mem_GB=42.072, [valid] loss_ctc=42.916, cer_ctc=0.222, loss_att=45.533, acc=0.641, cer=0.311, wer=1.000, loss=44.748, time=37 minutes and 26.56 seconds, total_count=88749, gpu_max_cached_mem_GB=42.072
+[gpub010:0/16] 2024-01-30 10:03:10,583 (trainer:391) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-01-30 10:03:10,588 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/12epoch.pth
+[gpub010:0/16] 2024-01-30 10:03:10,588 (trainer:272) INFO: 18/45epoch started. Estimated time to finish: 1 week, 3 days and 31 minutes
+[gpub010:0/16] 2024-01-30 10:03:10,597 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-01-30 10:03:29,028 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-01-30 10:03:32,497 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa771839f00>)
+[gpub010:0/16] 2024-01-30 10:03:32,497 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-01-30 10:03:32,500 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2904202.0 ON gpub010 CANCELLED AT 2024-01-30T10:04:03 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.13.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.13.log
new file mode 100644
index 0000000000000000000000000000000000000000..f2d76c33bbeaa911dabde5b263be3f23bb869d12
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.13.log
@@ -0,0 +1,2549 @@
+# Running on gpub078.delta.ncsa.illinois.edu
+# Started at Wed Jan 24 17:44:19 CST 2024
+# SLURMD_NODENAME=gpub078
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2891733
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1706312636
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2891733
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[078,083-084,086]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1706139836
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[078,083-084,086]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=57932
+# SLURM_TOPOLOGY_ADDR=ss00.ss12.gpub078
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_901f8055-bf9e-4e3f-b1e1-05051b41b9c5 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_901f8055-bf9e-4e3f-b1e1-05051b41b9c5
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_901f8055-bf9e-4e3f-b1e1-05051b41b9c5
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessi_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_901f8055-bf9e-4e3f-b1e1-05051b41b9c5
+ng_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_901f8055-bf9e-4e3f-b1e1-05051b41b9c5
+[gpub078:0/16] 2024-01-24 17:48:13,475 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub078:0/16] 2024-01-24 17:48:14,485 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub078:0/16] 2024-01-24 17:48:14,572 (s2t:464) INFO: Vocabulary size: 50002
+[gpub078:0/16] 2024-01-24 17:48:28,276 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub078:0/16] 2024-01-24 17:48:28,283 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub078:0/16] 2024-01-24 17:48:28,283 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub078:0/16] 2024-01-24 17:48:28,283 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub078:0/16] 2024-01-24 17:48:28,313 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub078:0/16] 2024-01-24 17:48:33,708 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 17:48:34,626 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa31986a380>)
+[gpub078:0/16] 2024-01-24 17:48:34,626 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub078:0/16] 2024-01-24 17:48:34,627 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 17:48:52,512 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub078:57986:57986 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.178<0>
+gpub078:57986:57986 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub078:57986:57986 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub078:0/16] 2024-01-24 17:48:58,623 (trainer:284) INFO: 8/45epoch started
+[gpub078:0/16] 2024-01-24 17:48:58,667 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-24 17:49:16,343 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 17:49:19,722 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa132eb1480>)
+[gpub078:0/16] 2024-01-24 17:49:19,722 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub078:0/16] 2024-01-24 17:49:19,725 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub083:56788:56788 [1] NCCL INFO cudaDriverVersion 12020
+gpub083:56788:56788 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.183<0>
+gpub083:56788:56788 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub083:56788:56840 [1] NCCL INFO NET/IB : No device found.
+gpub083:56788:56840 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.183<0> [1]hsn0:141.142.145.183<0>
+gpub083:56788:56840 [1] NCCL INFO Using network Socket
+gpub083:56788:56840 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff,ffffffff
+gpub083:56788:56840 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub083:56788:56840 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub083:56788:56840 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub083:56788:56840 [1] NCCL INFO Connected all rings
+gpub083:56788:56840 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/Socket/1
+gpub083:56788:56840 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/Socket/1
+gpub083:56788:56840 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub083:56788:56840 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub083:56788:56840 [1] NCCL INFO Connected all trees
+gpub083:56788:56840 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub083:56788:56840 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub083:56788:56840 [1] NCCL INFO comm 0x11e29d60 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub083:56787:56787 [0] NCCL INFO cudaDriverVersion 12020
+gpub083:56787:56787 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.183<0>
+gpub083:56787:56787 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub083:56787:56841 [0] NCCL INFO NET/IB : No device found.
+gpub083:56787:56841 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.183<0> [1]hsn0:141.142.145.183<0>
+gpub083:56787:56841 [0] NCCL INFO Using network Socket
+gpub083:56787:56841 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff,ffffffff
+gpub083:56787:56841 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub083:56787:56841 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub083:56787:56841 [0] NCCL INFO Connected all rings
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/Socket/1
+gpub083:56787:56841 [0] NCCL INFO Connected all trees
+gpub083:56787:56841 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub083:56787:56841 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub083:56787:56841 [0] NCCL INFO comm 0x19d03810 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub084:57372:57372 [2] NCCL INFO cudaDriverVersion 12020
+gpub084:57372:57372 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.184<0>
+gpub084:57372:57372 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub084:57372:57431 [2] NCCL INFO NET/IB : No device found.
+gpub084:57372:57431 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.184<0> [1]hsn0:141.142.145.184<0> [2]eth0:fe80::4845:b018:cee:4be4%eth0<0>
+gpub084:57372:57431 [2] NCCL INFO Using network Socket
+gpub084:57372:57431 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub084:57372:57431 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub084:57372:57431 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub084:57372:57431 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub084:57372:57431 [2] NCCL INFO Connected all rings
+gpub084:57372:57431 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub084:57372:57431 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub084:57372:57431 [2] NCCL INFO Connected all trees
+gpub084:57372:57431 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub084:57372:57431 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub084:57372:57431 [2] NCCL INFO comm 0x1d369d00 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub083:56789:56789 [2] NCCL INFO cudaDriverVersion 12020
+gpub083:56789:56789 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.183<0>
+gpub083:56789:56789 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub083:56789:56839 [2] NCCL INFO NET/IB : No device found.
+gpub083:56789:56839 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.183<0> [1]hsn0:141.142.145.183<0>
+gpub083:56789:56839 [2] NCCL INFO Using network Socket
+gpub083:56789:56839 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff,ffffffff
+gpub083:56789:56839 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub083:56789:56839 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub083:56789:56839 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub083:56789:56839 [2] NCCL INFO Connected all rings
+gpub083:56789:56839 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub083:56789:56839 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub083:56789:56839 [2] NCCL INFO Connected all trees
+gpub083:56789:56839 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub083:56789:56839 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub083:56789:56839 [2] NCCL INFO comm 0x1448f650 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub083:56790:56790 [3] NCCL INFO cudaDriverVersion 12020
+gpub083:56790:56790 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.183<0>
+gpub083:56790:56790 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub083:56790:56842 [3] NCCL INFO NET/IB : No device found.
+gpub083:56790:56842 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.183<0> [1]hsn0:141.142.145.183<0>
+gpub083:56790:56842 [3] NCCL INFO Using network Socket
+gpub083:56790:56842 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff,ffffffff
+gpub083:56790:56842 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub083:56790:56842 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub083:56790:56842 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub083:56790:56842 [3] NCCL INFO Connected all rings
+gpub083:56790:56842 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub083:56790:56842 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub083:56790:56842 [3] NCCL INFO Connected all trees
+gpub083:56790:56842 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub083:56790:56842 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub083:56790:56842 [3] NCCL INFO comm 0x163af330 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub078:57988:57988 [2] NCCL INFO cudaDriverVersion 12020
+gpub078:57988:57988 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.178<0>
+gpub078:57988:57988 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub078:57988:58046 [2] NCCL INFO NET/IB : No device found.
+gpub078:57988:58046 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.178<0> [1]hsn0:141.142.145.178<0> [2]eth0:fe80::3d2a:d8d2:47cc:f2ae%eth0<0>
+gpub078:57988:58046 [2] NCCL INFO Using network Socket
+gpub078:57988:58046 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub078:57988:58046 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub078:57988:58046 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub078:57988:58046 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub078:57988:58046 [2] NCCL INFO Connected all rings
+gpub078:57988:58046 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub078:57988:58046 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub078:57988:58046 [2] NCCL INFO Connected all trees
+gpub078:57988:58046 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub078:57988:58046 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub078:57988:58046 [2] NCCL INFO comm 0x1b77cc60 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub078:57989:57989 [3] NCCL INFO cudaDriverVersion 12020
+gpub078:57989:57989 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.178<0>
+gpub078:57989:57989 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub078:57989:58045 [3] NCCL INFO NET/IB : No device found.
+gpub078:57989:58045 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.178<0> [1]hsn0:141.142.145.178<0> [2]eth0:fe80::3d2a:d8d2:47cc:f2ae%eth0<0>
+gpub078:57989:58045 [3] NCCL INFO Using network Socket
+gpub078:57989:58045 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub078:57989:58045 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub078:57989:58045 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub078:57989:58045 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub078:57989:58045 [3] NCCL INFO Connected all rings
+gpub078:57989:58045 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub078:57989:58045 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub078:57989:58045 [3] NCCL INFO Connected all trees
+gpub078:57989:58045 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub078:57989:58045 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub078:57989:58045 [3] NCCL INFO comm 0x14e17e60 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub084:57373:57373 [3] NCCL INFO cudaDriverVersion 12020
+gpub084:57373:57373 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.184<0>
+gpub084:57373:57373 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub084:57373:57434 [3] NCCL INFO NET/IB : No device found.
+gpub084:57373:57434 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.184<0> [1]hsn0:141.142.145.184<0> [2]eth0:fe80::4845:b018:cee:4be4%eth0<0>
+gpub084:57373:57434 [3] NCCL INFO Using network Socket
+gpub084:57373:57434 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub084:57373:57434 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub084:57373:57434 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub084:57373:57434 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub084:57373:57434 [3] NCCL INFO Connected all rings
+gpub084:57373:57434 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub084:57373:57434 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub084:57373:57434 [3] NCCL INFO Connected all trees
+gpub084:57373:57434 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub084:57373:57434 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub084:57373:57434 [3] NCCL INFO comm 0x18903bb0 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub084:57370:57370 [0] NCCL INFO cudaDriverVersion 12020
+gpub084:57370:57370 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.184<0>
+gpub084:57370:57370 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub084:57370:57433 [0] NCCL INFO NET/IB : No device found.
+gpub084:57370:57433 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.184<0> [1]hsn0:141.142.145.184<0> [2]eth0:fe80::4845:b018:cee:4be4%eth0<0>
+gpub084:57370:57433 [0] NCCL INFO Using network Socket
+gpub084:57370:57433 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub084:57370:57433 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub084:57370:57433 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub084:57370:57433 [0] NCCL INFO Connected all rings
+gpub084:57370:57433 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/Socket/1
+gpub084:57370:57433 [0] NCCL INFO Connected all trees
+gpub084:57370:57433 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub084:57370:57433 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub084:57370:57433 [0] NCCL INFO comm 0x15f25ec0 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub078:57986:58043 [0] NCCL INFO NET/IB : No device found.
+gpub078:57986:58043 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.178<0> [1]hsn0:141.142.145.178<0> [2]eth0:fe80::3d2a:d8d2:47cc:f2ae%eth0<0>
+gpub078:57986:58043 [0] NCCL INFO Using network Socket
+gpub078:57986:58043 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub078:57986:58043 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub078:57986:58043 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub078:57986:58043 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub078:57986:58043 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub078:57986:58043 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub078:57986:58043 [0] NCCL INFO Connected all rings
+gpub078:57986:58043 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub078:57986:58043 [0] NCCL INFO Connected all trees
+gpub078:57986:58043 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub078:57986:58043 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub078:57986:58043 [0] NCCL INFO comm 0x1751a9a0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub086:138843:138843 [2] NCCL INFO cudaDriverVersion 12020
+gpub086:138843:138843 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.186<0>
+gpub086:138843:138843 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub086:138843:138896 [2] NCCL INFO NET/IB : No device found.
+gpub086:138843:138896 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.186<0> [1]hsn0:141.142.145.186<0> [2]eth0:fe80::6790:58e4:2640:df66%eth0<0>
+gpub086:138843:138896 [2] NCCL INFO Using network Socket
+gpub086:138843:138896 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub086:138843:138896 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub086:138843:138896 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub086:138843:138896 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub086:138843:138896 [2] NCCL INFO Connected all rings
+gpub086:138843:138896 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub086:138843:138896 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub086:138843:138896 [2] NCCL INFO Connected all trees
+gpub086:138843:138896 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub086:138843:138896 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub086:138843:138896 [2] NCCL INFO comm 0xf2b9ab0 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub086:138841:138841 [0] NCCL INFO cudaDriverVersion 12020
+gpub086:138841:138841 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.186<0>
+gpub086:138841:138841 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub086:138841:138894 [0] NCCL INFO NET/IB : No device found.
+gpub086:138841:138894 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.186<0> [1]hsn0:141.142.145.186<0> [2]eth0:fe80::6790:58e4:2640:df66%eth0<0>
+gpub086:138841:138894 [0] NCCL INFO Using network Socket
+gpub086:138841:138894 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub086:138841:138894 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub086:138841:138894 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub086:138841:138894 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub086:138841:138894 [0] NCCL INFO Connected all rings
+gpub086:138841:138894 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/Socket/1
+gpub086:138841:138894 [0] NCCL INFO Connected all trees
+gpub086:138841:138894 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub086:138841:138894 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub086:138841:138894 [0] NCCL INFO comm 0xc14e4f0 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub086:138844:138844 [3] NCCL INFO cudaDriverVersion 12020
+gpub086:138844:138844 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.186<0>
+gpub086:138844:138844 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub086:138844:138897 [3] NCCL INFO NET/IB : No device found.
+gpub086:138844:138897 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.186<0> [1]hsn0:141.142.145.186<0> [2]eth0:fe80::6790:58e4:2640:df66%eth0<0>
+gpub086:138844:138897 [3] NCCL INFO Using network Socket
+gpub086:138844:138897 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub086:138844:138897 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub086:138844:138897 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub086:138844:138897 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub086:138844:138897 [3] NCCL INFO Connected all rings
+gpub086:138844:138897 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub086:138844:138897 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub086:138844:138897 [3] NCCL INFO Connected all trees
+gpub086:138844:138897 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub086:138844:138897 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub086:138844:138897 [3] NCCL INFO comm 0x1ae3bae0 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub078:57987:57987 [1] NCCL INFO cudaDriverVersion 12020
+gpub078:57987:57987 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.178<0>
+gpub078:57987:57987 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub078:57987:58044 [1] NCCL INFO NET/IB : No device found.
+gpub078:57987:58044 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.178<0> [1]hsn0:141.142.145.178<0> [2]eth0:fe80::3d2a:d8d2:47cc:f2ae%eth0<0>
+gpub078:57987:58044 [1] NCCL INFO Using network Socket
+gpub078:57987:58044 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub078:57987:58044 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub078:57987:58044 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub078:57987:58044 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub078:57987:58044 [1] NCCL INFO Connected all rings
+gpub078:57987:58044 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub078:57987:58044 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub078:57987:58044 [1] NCCL INFO Connected all trees
+gpub078:57987:58044 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub078:57987:58044 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub078:57987:58044 [1] NCCL INFO comm 0x1675cb20 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub086:138842:138842 [1] NCCL INFO cudaDriverVersion 12020
+gpub086:138842:138842 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.186<0>
+gpub086:138842:138842 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub086:138842:138895 [1] NCCL INFO NET/IB : No device found.
+gpub086:138842:138895 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.186<0> [1]hsn0:141.142.145.186<0> [2]eth0:fe80::6790:58e4:2640:df66%eth0<0>
+gpub086:138842:138895 [1] NCCL INFO Using network Socket
+gpub086:138842:138895 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub086:138842:138895 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub086:138842:138895 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub086:138842:138895 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub086:138842:138895 [1] NCCL INFO Connected all rings
+gpub086:138842:138895 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub086:138842:138895 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub086:138842:138895 [1] NCCL INFO Connected all trees
+gpub086:138842:138895 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub086:138842:138895 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub086:138842:138895 [1] NCCL INFO comm 0x1cffd390 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub084:57371:57371 [1] NCCL INFO cudaDriverVersion 12020
+gpub084:57371:57371 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.184<0>
+gpub084:57371:57371 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub084:57371:57432 [1] NCCL INFO NET/IB : No device found.
+gpub084:57371:57432 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.184<0> [1]hsn0:141.142.145.184<0> [2]eth0:fe80::4845:b018:cee:4be4%eth0<0>
+gpub084:57371:57432 [1] NCCL INFO Using network Socket
+gpub084:57371:57432 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub084:57371:57432 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub084:57371:57432 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub084:57371:57432 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub084:57371:57432 [1] NCCL INFO Connected all rings
+gpub084:57371:57432 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/Socket/1
+gpub084:57371:57432 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/Socket/1
+gpub084:57371:57432 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub084:57371:57432 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub084:57371:57432 [1] NCCL INFO Connected all trees
+gpub084:57371:57432 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub084:57371:57432 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub084:57371:57432 [1] NCCL INFO comm 0x1ac587c0 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+[gpub078:0/16] 2024-01-24 17:59:16,203 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub078:0/16] 2024-01-24 18:02:07,389 (trainer:737) INFO: 8epoch:train:1-100batch: iter_time=1.477, forward_time=0.521, loss_ctc=67.287, loss_att=64.664, acc=0.677, loss=65.451, backward_time=0.496, grad_norm=34.142, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.108, optim0_lr0=3.779e-04, train_time=7.868
+[gpub078:0/16] 2024-01-24 18:04:53,435 (trainer:737) INFO: 8epoch:train:101-200batch: iter_time=1.054e-04, forward_time=0.484, loss_ctc=55.465, loss_att=50.498, acc=0.711, loss=51.988, backward_time=0.524, grad_norm=48.514, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.109, optim0_lr0=3.777e-04, train_time=1.679
+[gpub078:0/16] 2024-01-24 18:08:13,440 (trainer:737) INFO: 8epoch:train:201-300batch: iter_time=2.230e-04, forward_time=0.359, loss_ctc=64.640, loss_att=74.201, acc=0.652, loss=71.333, backward_time=0.477, grad_norm=33.516, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.099, optim0_lr0=3.775e-04, train_time=2.000
+[gpub078:0/16] 2024-01-24 18:13:09,466 (trainer:737) INFO: 8epoch:train:301-400batch: iter_time=1.038e-04, forward_time=0.519, loss_ctc=57.852, loss_att=58.957, acc=0.672, loss=58.626, backward_time=0.541, grad_norm=29.836, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.109, optim0_lr0=3.773e-04, train_time=2.960
+[gpub078:0/16] 2024-01-24 18:16:06,508 (trainer:737) INFO: 8epoch:train:401-500batch: iter_time=1.108e-04, forward_time=0.427, loss_ctc=64.955, loss_att=58.926, acc=0.696, loss=60.735, backward_time=0.479, grad_norm=30.472, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.097, optim0_lr0=3.772e-04, train_time=1.771
+[gpub078:0/16] 2024-01-24 18:19:43,214 (trainer:737) INFO: 8epoch:train:501-600batch: iter_time=1.000e-04, forward_time=0.492, loss_ctc=58.232, loss_att=56.385, acc=0.694, loss=56.939, backward_time=0.460, grad_norm=30.055, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.109, optim0_lr0=3.770e-04, train_time=2.167
+[gpub078:0/16] 2024-01-24 18:23:34,149 (trainer:737) INFO: 8epoch:train:601-700batch: iter_time=9.105e-05, forward_time=0.378, loss_ctc=66.231, loss_att=65.444, acc=0.661, loss=65.680, backward_time=0.463, grad_norm=31.659, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.108, optim0_lr0=3.768e-04, train_time=2.309
+[gpub078:0/16] 2024-01-24 18:27:40,469 (trainer:737) INFO: 8epoch:train:701-800batch: iter_time=8.991e-05, forward_time=0.461, loss_ctc=77.988, loss_att=77.499, acc=0.670, loss=77.646, backward_time=0.506, grad_norm=39.467, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.106, optim0_lr0=3.766e-04, train_time=2.463
+[gpub078:0/16] 2024-01-24 18:31:43,331 (trainer:737) INFO: 8epoch:train:801-900batch: iter_time=3.867e-04, forward_time=0.391, loss_ctc=68.415, loss_att=74.714, acc=0.689, loss=72.824, backward_time=0.468, grad_norm=32.131, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.099, optim0_lr0=3.764e-04, train_time=2.429
+[gpub078:0/16] 2024-01-24 18:35:22,493 (trainer:737) INFO: 8epoch:train:901-1000batch: iter_time=9.297e-05, forward_time=0.483, loss_ctc=69.085, loss_att=66.656, acc=0.664, loss=67.385, backward_time=0.445, grad_norm=36.685, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.103, optim0_lr0=3.763e-04, train_time=2.191
+[gpub078:0/16] 2024-01-24 18:38:56,291 (trainer:737) INFO: 8epoch:train:1001-1100batch: iter_time=5.010e-04, forward_time=0.434, loss_ctc=60.066, loss_att=55.432, acc=0.691, loss=56.822, backward_time=0.500, grad_norm=27.324, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.107, optim0_lr0=3.761e-04, train_time=2.136
+[gpub078:0/16] 2024-01-24 18:43:15,704 (trainer:737) INFO: 8epoch:train:1101-1200batch: iter_time=9.201e-04, forward_time=0.533, loss_ctc=67.036, loss_att=60.037, acc=0.688, loss=62.137, backward_time=0.538, grad_norm=31.309, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.106, optim0_lr0=3.759e-04, train_time=2.595
+[gpub078:0/16] 2024-01-24 18:45:40,635 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub078:0/16] 2024-01-24 18:45:59,497 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 18:46:03,141 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa132eb02b0>)
+[gpub078:0/16] 2024-01-24 18:46:03,141 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub078:0/16] 2024-01-24 18:46:03,145 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 18:59:16,301 (trainer:737) INFO: 8epoch:train:1201-1300batch: iter_time=6.808, forward_time=0.683, loss_ctc=64.050, loss_att=72.417, acc=0.652, loss=69.907, backward_time=0.459, grad_norm=33.530, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.099, optim0_lr0=3.757e-04, train_time=9.606
+[gpub078:0/16] 2024-01-24 19:02:37,385 (trainer:737) INFO: 8epoch:train:1301-1400batch: iter_time=8.733e-05, forward_time=0.289, loss_ctc=59.245, loss_att=53.995, acc=0.690, loss=55.570, backward_time=0.399, grad_norm=31.908, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.092, optim0_lr0=3.756e-04, train_time=2.011
+[gpub078:0/16] 2024-01-24 19:07:56,243 (trainer:737) INFO: 8epoch:train:1401-1500batch: iter_time=8.794e-05, forward_time=0.662, loss_ctc=63.039, loss_att=69.598, acc=0.660, loss=67.630, backward_time=0.541, grad_norm=28.646, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.114, optim0_lr0=3.754e-04, train_time=3.188
+[gpub078:0/16] 2024-01-24 19:11:19,262 (trainer:737) INFO: 8epoch:train:1501-1600batch: iter_time=9.274e-05, forward_time=0.287, loss_ctc=55.803, loss_att=57.941, acc=0.679, loss=57.300, backward_time=0.418, grad_norm=26.454, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.102, optim0_lr0=3.752e-04, train_time=2.030
+[gpub078:0/16] 2024-01-24 19:15:42,578 (trainer:737) INFO: 8epoch:train:1601-1700batch: iter_time=1.016e-04, forward_time=0.570, loss_ctc=58.775, loss_att=58.463, acc=0.672, loss=58.556, backward_time=0.500, grad_norm=28.391, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.107, optim0_lr0=3.750e-04, train_time=2.632
+[gpub078:0/16] 2024-01-24 19:19:58,941 (trainer:737) INFO: 8epoch:train:1701-1800batch: iter_time=8.886e-05, forward_time=0.561, loss_ctc=67.197, loss_att=62.810, acc=0.688, loss=64.126, backward_time=0.627, grad_norm=32.377, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.134, optim0_lr0=3.749e-04, train_time=2.564
+[gpub078:0/16] 2024-01-24 19:23:54,548 (trainer:737) INFO: 8epoch:train:1801-1900batch: iter_time=0.001, forward_time=0.492, loss_ctc=54.172, loss_att=55.433, acc=0.671, loss=55.054, backward_time=0.525, grad_norm=27.285, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.137, optim0_lr0=3.747e-04, train_time=2.355
+[gpub078:0/16] 2024-01-24 19:28:55,528 (trainer:737) INFO: 8epoch:train:1901-2000batch: iter_time=0.001, forward_time=0.644, loss_ctc=72.970, loss_att=71.812, acc=0.658, loss=72.160, backward_time=0.519, grad_norm=34.859, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.147, optim0_lr0=3.745e-04, train_time=3.009
+[gpub078:0/16] 2024-01-24 19:33:14,512 (trainer:737) INFO: 8epoch:train:2001-2100batch: iter_time=5.994e-04, forward_time=0.476, loss_ctc=70.665, loss_att=83.147, acc=0.652, loss=79.402, backward_time=0.526, grad_norm=34.060, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.123, optim0_lr0=3.743e-04, train_time=2.592
+[gpub078:0/16] 2024-01-24 19:37:39,961 (trainer:737) INFO: 8epoch:train:2101-2200batch: iter_time=5.882e-04, forward_time=0.490, loss_ctc=66.969, loss_att=67.657, acc=0.671, loss=67.451, backward_time=0.443, grad_norm=30.973, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.136, optim0_lr0=3.742e-04, train_time=2.652
+[gpub078:0/16] 2024-01-24 19:41:46,567 (trainer:737) INFO: 8epoch:train:2201-2300batch: iter_time=0.001, forward_time=0.350, loss_ctc=65.029, loss_att=60.496, acc=0.666, loss=61.856, backward_time=0.453, grad_norm=34.476, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.101, optim0_lr0=3.740e-04, train_time=2.468
+[gpub078:0/16] 2024-01-24 19:45:47,438 (trainer:737) INFO: 8epoch:train:2301-2400batch: iter_time=8.167e-05, forward_time=0.535, loss_ctc=61.928, loss_att=53.898, acc=0.679, loss=56.307, backward_time=0.559, grad_norm=29.532, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.121, optim0_lr0=3.738e-04, train_time=2.408
+[gpub078:0/16] 2024-01-24 19:49:26,629 (trainer:737) INFO: 8epoch:train:2401-2500batch: iter_time=1.275e-04, forward_time=0.605, loss_ctc=63.961, loss_att=63.608, acc=0.677, loss=63.714, backward_time=0.484, grad_norm=31.163, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.119, optim0_lr0=3.736e-04, train_time=2.192
+[gpub078:0/16] 2024-01-24 19:49:46,694 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub078:0/16] 2024-01-24 19:50:05,368 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 19:50:08,894 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa121943e50>)
+[gpub078:0/16] 2024-01-24 19:50:08,894 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub078:0/16] 2024-01-24 19:50:08,898 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 20:03:03,411 (trainer:737) INFO: 8epoch:train:2501-2600batch: iter_time=6.147, forward_time=0.291, loss_ctc=65.172, loss_att=65.587, acc=0.673, loss=65.462, backward_time=0.405, grad_norm=34.929, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.092, optim0_lr0=3.735e-04, train_time=8.168
+[gpub078:0/16] 2024-01-24 20:06:36,862 (trainer:737) INFO: 8epoch:train:2601-2700batch: iter_time=7.638e-05, forward_time=0.289, loss_ctc=54.311, loss_att=50.899, acc=0.714, loss=51.923, backward_time=0.401, grad_norm=26.252, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.092, optim0_lr0=3.733e-04, train_time=2.134
+[gpub078:0/16] 2024-01-24 20:09:33,896 (trainer:737) INFO: 8epoch:train:2701-2800batch: iter_time=7.971e-05, forward_time=0.431, loss_ctc=62.439, loss_att=72.873, acc=0.658, loss=69.743, backward_time=0.499, grad_norm=30.887, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.109, optim0_lr0=3.731e-04, train_time=1.769
+[gpub078:0/16] 2024-01-24 20:12:03,897 (trainer:737) INFO: 8epoch:train:2801-2900batch: iter_time=7.948e-05, forward_time=0.426, loss_ctc=56.383, loss_att=58.784, acc=0.677, loss=58.064, backward_time=0.426, grad_norm=28.181, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.093, optim0_lr0=3.729e-04, train_time=1.501
+[gpub078:0/16] 2024-01-24 20:15:55,893 (trainer:737) INFO: 8epoch:train:2901-3000batch: iter_time=7.822e-05, forward_time=0.290, loss_ctc=64.191, loss_att=58.181, acc=0.701, loss=59.984, backward_time=0.402, grad_norm=28.721, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.092, optim0_lr0=3.728e-04, train_time=2.320
+[gpub078:0/16] 2024-01-24 20:18:46,715 (trainer:737) INFO: 8epoch:train:3001-3100batch: iter_time=8.864e-05, forward_time=0.288, loss_ctc=55.397, loss_att=55.235, acc=0.700, loss=55.284, backward_time=0.399, grad_norm=29.011, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.726e-04, train_time=1.708
+[gpub078:0/16] 2024-01-24 20:21:12,541 (trainer:737) INFO: 8epoch:train:3101-3200batch: iter_time=7.855e-05, forward_time=0.296, loss_ctc=63.677, loss_att=64.910, acc=0.668, loss=64.540, backward_time=0.408, grad_norm=30.250, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.102, optim0_lr0=3.724e-04, train_time=1.458
+[gpub078:0/16] 2024-01-24 20:24:16,315 (trainer:737) INFO: 8epoch:train:3201-3300batch: iter_time=7.550e-05, forward_time=0.499, loss_ctc=75.495, loss_att=77.240, acc=0.673, loss=76.717, backward_time=0.474, grad_norm=36.985, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.109, optim0_lr0=3.722e-04, train_time=1.836
+[gpub078:0/16] 2024-01-24 20:27:21,444 (trainer:737) INFO: 8epoch:train:3301-3400batch: iter_time=8.200e-05, forward_time=0.293, loss_ctc=66.213, loss_att=73.181, acc=0.697, loss=71.091, backward_time=0.407, grad_norm=30.014, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.091, optim0_lr0=3.721e-04, train_time=1.852
+[gpub078:0/16] 2024-01-24 20:29:55,186 (trainer:737) INFO: 8epoch:train:3401-3500batch: iter_time=8.005e-05, forward_time=0.288, loss_ctc=67.050, loss_att=64.978, acc=0.673, loss=65.599, backward_time=0.403, grad_norm=35.110, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.091, optim0_lr0=3.719e-04, train_time=1.537
+[gpub078:0/16] 2024-01-24 20:32:42,044 (trainer:737) INFO: 8epoch:train:3501-3600batch: iter_time=2.349e-04, forward_time=0.303, loss_ctc=58.765, loss_att=54.776, acc=0.698, loss=55.973, backward_time=0.412, grad_norm=29.936, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.097, optim0_lr0=3.717e-04, train_time=1.667
+[gpub078:0/16] 2024-01-24 20:36:02,894 (trainer:737) INFO: 8epoch:train:3601-3700batch: iter_time=8.695e-05, forward_time=0.563, loss_ctc=64.268, loss_att=58.070, acc=0.695, loss=59.930, backward_time=0.543, grad_norm=29.843, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.104, optim0_lr0=3.716e-04, train_time=2.010
+[gpub078:0/16] 2024-01-24 20:37:29,566 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub078:0/16] 2024-01-24 20:37:48,529 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 20:37:52,045 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fa120e1fd90>)
+[gpub078:0/16] 2024-01-24 20:37:52,045 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub078:0/16] 2024-01-24 20:37:52,259 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 20:51:39,541 (trainer:737) INFO: 8epoch:train:3701-3800batch: iter_time=7.204, forward_time=0.289, loss_ctc=62.626, loss_att=69.753, acc=0.658, loss=67.615, backward_time=0.405, grad_norm=32.949, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.714e-04, train_time=9.366
+[gpub078:0/16] 2024-01-24 20:54:25,932 (trainer:737) INFO: 8epoch:train:3801-3900batch: iter_time=8.709e-05, forward_time=0.288, loss_ctc=56.942, loss_att=52.250, acc=0.693, loss=53.658, backward_time=0.402, grad_norm=28.850, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.712e-04, train_time=1.664
+[gpub078:0/16] 2024-01-24 20:58:23,326 (trainer:737) INFO: 8epoch:train:3901-4000batch: iter_time=8.319e-05, forward_time=0.551, loss_ctc=62.313, loss_att=67.699, acc=0.668, loss=66.083, backward_time=0.518, grad_norm=29.697, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.121, optim0_lr0=3.710e-04, train_time=2.374
+[gpub078:0/16] 2024-01-24 21:00:47,095 (trainer:737) INFO: 8epoch:train:4001-4100batch: iter_time=8.071e-05, forward_time=0.338, loss_ctc=54.975, loss_att=57.356, acc=0.680, loss=56.641, backward_time=0.419, grad_norm=26.714, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.095, optim0_lr0=3.709e-04, train_time=1.437
+[gpub078:0/16] 2024-01-24 21:03:57,527 (trainer:737) INFO: 8epoch:train:4101-4200batch: iter_time=8.280e-05, forward_time=0.287, loss_ctc=58.126, loss_att=57.197, acc=0.676, loss=57.476, backward_time=0.400, grad_norm=29.570, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.707e-04, train_time=1.904
+[gpub078:0/16] 2024-01-24 21:06:50,470 (trainer:737) INFO: 8epoch:train:4201-4300batch: iter_time=8.077e-05, forward_time=0.290, loss_ctc=67.802, loss_att=62.084, acc=0.691, loss=63.800, backward_time=0.403, grad_norm=31.941, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.705e-04, train_time=1.730
+[gpub078:0/16] 2024-01-24 21:10:53,608 (trainer:737) INFO: 8epoch:train:4301-4400batch: iter_time=9.156e-05, forward_time=0.510, loss_ctc=53.108, loss_att=53.714, acc=0.678, loss=53.532, backward_time=0.583, grad_norm=26.781, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.109, optim0_lr0=3.704e-04, train_time=2.431
+[gpub078:0/16] 2024-01-24 21:13:25,227 (trainer:737) INFO: 8epoch:train:4401-4500batch: iter_time=8.395e-05, forward_time=0.317, loss_ctc=71.335, loss_att=70.949, acc=0.663, loss=71.065, backward_time=0.429, grad_norm=34.456, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.093, optim0_lr0=3.702e-04, train_time=1.516
+[gpub078:0/16] 2024-01-24 21:16:48,330 (trainer:737) INFO: 8epoch:train:4501-4600batch: iter_time=8.822e-05, forward_time=0.290, loss_ctc=69.839, loss_att=81.505, acc=0.657, loss=78.005, backward_time=0.405, grad_norm=36.561, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.700e-04, train_time=2.029
+[gpub078:0/16] 2024-01-24 21:19:43,164 (trainer:737) INFO: 8epoch:train:4601-4700batch: iter_time=1.006e-04, forward_time=0.288, loss_ctc=66.742, loss_att=66.865, acc=0.675, loss=66.828, backward_time=0.403, grad_norm=31.087, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.699e-04, train_time=1.749
+[gpub078:0/16] 2024-01-24 21:23:16,349 (trainer:737) INFO: 8epoch:train:4701-4800batch: iter_time=9.956e-05, forward_time=0.479, loss_ctc=62.532, loss_att=59.713, acc=0.669, loss=60.559, backward_time=0.462, grad_norm=31.152, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.122, optim0_lr0=3.697e-04, train_time=2.132
+[gpub078:0/16] 2024-01-24 21:25:40,300 (trainer:737) INFO: 8epoch:train:4801-4900batch: iter_time=9.257e-05, forward_time=0.303, loss_ctc=60.327, loss_att=53.209, acc=0.685, loss=55.345, backward_time=0.413, grad_norm=30.315, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.100, optim0_lr0=3.695e-04, train_time=1.439
+[gpub078:0/16] 2024-01-24 21:28:12,148 (trainer:737) INFO: 8epoch:train:4901-5000batch: iter_time=7.991e-05, forward_time=0.291, loss_ctc=62.789, loss_att=62.919, acc=0.682, loss=62.880, backward_time=0.403, grad_norm=30.866, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.092, optim0_lr0=3.694e-04, train_time=1.518
+[gpub078:0/16] 2024-01-24 21:28:32,495 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub078:0/16] 2024-01-24 21:28:50,764 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 21:28:54,281 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f922a04bf70>)
+[gpub078:0/16] 2024-01-24 21:28:54,281 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub078:0/16] 2024-01-24 21:28:54,367 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 21:40:30,517 (trainer:737) INFO: 8epoch:train:5001-5100batch: iter_time=5.588, forward_time=0.292, loss_ctc=63.033, loss_att=64.484, acc=0.678, loss=64.049, backward_time=0.405, grad_norm=32.837, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.093, optim0_lr0=3.692e-04, train_time=7.384
+[gpub078:0/16] 2024-01-24 21:43:33,166 (trainer:737) INFO: 8epoch:train:5101-5200batch: iter_time=7.702e-05, forward_time=0.640, loss_ctc=53.142, loss_att=49.888, acc=0.719, loss=50.864, backward_time=0.501, grad_norm=25.332, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.125, optim0_lr0=3.690e-04, train_time=1.826
+[gpub078:0/16] 2024-01-24 21:46:57,140 (trainer:737) INFO: 8epoch:train:5201-5300batch: iter_time=8.310e-05, forward_time=0.289, loss_ctc=61.762, loss_att=73.108, acc=0.661, loss=69.704, backward_time=0.406, grad_norm=30.767, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.689e-04, train_time=2.040
+[gpub078:0/16] 2024-01-24 21:49:53,787 (trainer:737) INFO: 8epoch:train:5301-5400batch: iter_time=8.817e-05, forward_time=0.290, loss_ctc=55.343, loss_att=57.299, acc=0.682, loss=56.712, backward_time=0.412, grad_norm=26.913, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.102, optim0_lr0=3.687e-04, train_time=1.766
+[gpub078:0/16] 2024-01-24 21:53:19,031 (trainer:737) INFO: 8epoch:train:5401-5500batch: iter_time=8.647e-05, forward_time=0.291, loss_ctc=62.724, loss_att=57.497, acc=0.705, loss=59.065, backward_time=0.404, grad_norm=28.239, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.685e-04, train_time=2.052
+[gpub078:0/16] 2024-01-24 21:57:18,559 (trainer:737) INFO: 8epoch:train:5501-5600batch: iter_time=0.060, forward_time=0.715, loss_ctc=54.590, loss_att=54.446, acc=0.703, loss=54.489, backward_time=0.632, grad_norm=27.886, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.128, optim0_lr0=3.684e-04, train_time=2.395
+[gpub078:0/16] 2024-01-24 22:01:22,134 (trainer:737) INFO: 8epoch:train:5601-5700batch: iter_time=8.489e-05, forward_time=0.290, loss_ctc=63.190, loss_att=63.838, acc=0.671, loss=63.644, backward_time=0.404, grad_norm=29.174, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.682e-04, train_time=2.436
+[gpub078:0/16] 2024-01-24 22:03:50,158 (trainer:737) INFO: 8epoch:train:5701-5800batch: iter_time=8.385e-05, forward_time=0.325, loss_ctc=74.278, loss_att=75.846, acc=0.679, loss=75.376, backward_time=0.409, grad_norm=34.467, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.093, optim0_lr0=3.680e-04, train_time=1.480
+[gpub078:0/16] 2024-01-24 22:06:51,253 (trainer:737) INFO: 8epoch:train:5801-5900batch: iter_time=8.697e-05, forward_time=0.316, loss_ctc=64.992, loss_att=72.950, acc=0.697, loss=70.562, backward_time=0.411, grad_norm=30.582, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.093, optim0_lr0=3.679e-04, train_time=1.811
+[gpub078:0/16] 2024-01-24 22:10:49,115 (trainer:737) INFO: 8epoch:train:5901-6000batch: iter_time=7.957e-05, forward_time=0.554, loss_ctc=65.010, loss_att=64.843, acc=0.673, loss=64.893, backward_time=0.492, grad_norm=30.853, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.114, optim0_lr0=3.677e-04, train_time=2.378
+[gpub078:0/16] 2024-01-24 22:13:53,593 (trainer:737) INFO: 8epoch:train:6001-6100batch: iter_time=7.793e-05, forward_time=0.290, loss_ctc=58.423, loss_att=55.180, acc=0.698, loss=56.153, backward_time=0.401, grad_norm=27.700, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.675e-04, train_time=1.845
+[gpub078:0/16] 2024-01-24 22:16:15,516 (trainer:737) INFO: 8epoch:train:6101-6200batch: iter_time=7.967e-05, forward_time=0.291, loss_ctc=63.547, loss_att=57.439, acc=0.698, loss=59.272, backward_time=0.429, grad_norm=29.217, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.105, optim0_lr0=3.674e-04, train_time=1.419
+[gpub078:0/16] 2024-01-24 22:18:16,718 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub078:0/16] 2024-01-24 22:18:35,575 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 22:18:39,464 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9280ed7fd0>)
+[gpub078:0/16] 2024-01-24 22:18:39,464 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub078:0/16] 2024-01-24 22:18:39,467 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 22:30:41,898 (trainer:737) INFO: 8epoch:train:6201-6300batch: iter_time=6.486, forward_time=0.580, loss_ctc=61.941, loss_att=68.623, acc=0.662, loss=66.618, backward_time=0.455, grad_norm=33.007, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.101, optim0_lr0=3.672e-04, train_time=8.664
+[gpub078:0/16] 2024-01-24 22:33:12,010 (trainer:737) INFO: 8epoch:train:6301-6400batch: iter_time=7.192e-05, forward_time=0.287, loss_ctc=55.918, loss_att=51.508, acc=0.696, loss=52.831, backward_time=0.399, grad_norm=29.252, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.670e-04, train_time=1.501
+[gpub078:0/16] 2024-01-24 22:36:09,848 (trainer:737) INFO: 8epoch:train:6401-6500batch: iter_time=8.064e-05, forward_time=0.288, loss_ctc=61.422, loss_att=66.618, acc=0.672, loss=65.060, backward_time=0.401, grad_norm=28.147, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.669e-04, train_time=1.778
+[gpub078:0/16] 2024-01-24 22:39:53,381 (trainer:737) INFO: 8epoch:train:6501-6600batch: iter_time=7.778e-05, forward_time=0.295, loss_ctc=53.722, loss_att=56.005, acc=0.688, loss=55.320, backward_time=0.410, grad_norm=26.435, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.097, optim0_lr0=3.667e-04, train_time=2.234
+[gpub078:0/16] 2024-01-24 22:43:08,762 (trainer:737) INFO: 8epoch:train:6601-6700batch: iter_time=8.299e-05, forward_time=0.598, loss_ctc=57.296, loss_att=56.583, acc=0.681, loss=56.797, backward_time=0.476, grad_norm=28.002, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.122, optim0_lr0=3.665e-04, train_time=1.954
+[gpub078:0/16] 2024-01-24 22:46:03,883 (trainer:737) INFO: 8epoch:train:6701-6800batch: iter_time=8.820e-05, forward_time=0.289, loss_ctc=65.013, loss_att=61.416, acc=0.694, loss=62.495, backward_time=0.404, grad_norm=32.354, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.664e-04, train_time=1.751
+[gpub078:0/16] 2024-01-24 22:48:58,011 (trainer:737) INFO: 8epoch:train:6801-6900batch: iter_time=8.057e-05, forward_time=0.295, loss_ctc=52.577, loss_att=53.208, acc=0.681, loss=53.019, backward_time=0.398, grad_norm=26.422, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.092, optim0_lr0=3.662e-04, train_time=1.741
+[gpub078:0/16] 2024-01-24 22:51:51,233 (trainer:737) INFO: 8epoch:train:6901-7000batch: iter_time=7.895e-05, forward_time=0.292, loss_ctc=70.545, loss_att=69.654, acc=0.668, loss=69.922, backward_time=0.433, grad_norm=32.569, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.098, optim0_lr0=3.660e-04, train_time=1.732
+[gpub078:0/16] 2024-01-24 22:55:02,608 (trainer:737) INFO: 8epoch:train:7001-7100batch: iter_time=7.890e-05, forward_time=0.541, loss_ctc=69.239, loss_att=80.677, acc=0.662, loss=77.246, backward_time=0.486, grad_norm=31.885, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.103, optim0_lr0=3.659e-04, train_time=1.912
+[gpub078:0/16] 2024-01-24 22:58:12,161 (trainer:737) INFO: 8epoch:train:7101-7200batch: iter_time=8.230e-05, forward_time=0.289, loss_ctc=65.159, loss_att=65.849, acc=0.680, loss=65.642, backward_time=0.402, grad_norm=30.445, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.657e-04, train_time=1.897
+[gpub078:0/16] 2024-01-24 23:00:30,605 (trainer:737) INFO: 8epoch:train:7201-7300batch: iter_time=8.119e-05, forward_time=0.292, loss_ctc=63.188, loss_att=59.080, acc=0.672, loss=60.312, backward_time=0.417, grad_norm=35.145, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.097, optim0_lr0=3.656e-04, train_time=1.384
+[gpub078:0/16] 2024-01-24 23:03:38,737 (trainer:737) INFO: 8epoch:train:7301-7400batch: iter_time=3.951e-04, forward_time=0.505, loss_ctc=58.899, loss_att=52.156, acc=0.688, loss=54.179, backward_time=0.466, grad_norm=29.318, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.106, optim0_lr0=3.654e-04, train_time=1.880
+[gpub078:0/16] 2024-01-24 23:06:57,923 (trainer:737) INFO: 8epoch:train:7401-7500batch: iter_time=7.726e-05, forward_time=0.291, loss_ctc=62.314, loss_att=61.999, acc=0.684, loss=62.093, backward_time=0.398, grad_norm=31.437, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.652e-04, train_time=1.992
+[gpub078:0/16] 2024-01-24 23:07:18,449 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub078:0/16] 2024-01-24 23:07:36,937 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 23:07:40,334 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f922a0c0400>)
+[gpub078:0/16] 2024-01-24 23:07:40,334 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub078:0/16] 2024-01-24 23:07:40,337 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-24 23:20:40,977 (trainer:737) INFO: 8epoch:train:7501-7600batch: iter_time=6.699, forward_time=0.304, loss_ctc=62.811, loss_att=60.300, acc=0.676, loss=61.053, backward_time=0.407, grad_norm=33.122, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.095, optim0_lr0=3.651e-04, train_time=8.231
+[gpub078:0/16] 2024-01-24 23:24:07,346 (trainer:737) INFO: 8epoch:train:7601-7700batch: iter_time=0.016, forward_time=0.597, loss_ctc=52.814, loss_att=48.227, acc=0.715, loss=49.603, backward_time=0.513, grad_norm=25.456, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.115, optim0_lr0=3.649e-04, train_time=2.063
+[gpub078:0/16] 2024-01-24 23:28:50,579 (trainer:737) INFO: 8epoch:train:7701-7800batch: iter_time=8.963e-05, forward_time=0.288, loss_ctc=61.499, loss_att=71.298, acc=0.655, loss=68.359, backward_time=0.402, grad_norm=30.173, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.647e-04, train_time=2.831
+[gpub078:0/16] 2024-01-24 23:31:09,546 (trainer:737) INFO: 8epoch:train:7801-7900batch: iter_time=9.935e-05, forward_time=0.287, loss_ctc=54.465, loss_att=54.471, acc=0.684, loss=54.469, backward_time=0.404, grad_norm=27.970, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.103, optim0_lr0=3.646e-04, train_time=1.391
+[gpub078:0/16] 2024-01-24 23:34:46,284 (trainer:737) INFO: 8epoch:train:7901-8000batch: iter_time=1.000e-04, forward_time=0.570, loss_ctc=62.655, loss_att=56.134, acc=0.700, loss=58.090, backward_time=0.522, grad_norm=30.269, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.110, optim0_lr0=3.644e-04, train_time=2.167
+[gpub078:0/16] 2024-01-24 23:37:56,352 (trainer:737) INFO: 8epoch:train:8001-8100batch: iter_time=1.062e-04, forward_time=0.288, loss_ctc=53.950, loss_att=54.649, acc=0.690, loss=54.439, backward_time=0.403, grad_norm=29.826, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.643e-04, train_time=1.897
+[gpub078:0/16] 2024-01-24 23:41:10,528 (trainer:737) INFO: 8epoch:train:8101-8200batch: iter_time=9.878e-05, forward_time=0.290, loss_ctc=62.726, loss_att=63.608, acc=0.670, loss=63.343, backward_time=0.399, grad_norm=30.855, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.641e-04, train_time=1.941
+[gpub078:0/16] 2024-01-24 23:44:20,599 (trainer:737) INFO: 8epoch:train:8201-8300batch: iter_time=9.591e-05, forward_time=0.294, loss_ctc=73.559, loss_att=73.909, acc=0.671, loss=73.804, backward_time=0.426, grad_norm=35.185, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.098, optim0_lr0=3.639e-04, train_time=1.904
+[gpub078:0/16] 2024-01-24 23:47:40,342 (trainer:737) INFO: 8epoch:train:8301-8400batch: iter_time=9.456e-05, forward_time=0.674, loss_ctc=65.220, loss_att=72.429, acc=0.684, loss=70.266, backward_time=0.508, grad_norm=31.104, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.115, optim0_lr0=3.638e-04, train_time=1.997
+[gpub078:0/16] 2024-01-24 23:50:35,964 (trainer:737) INFO: 8epoch:train:8401-8500batch: iter_time=9.248e-05, forward_time=0.288, loss_ctc=65.308, loss_att=63.566, acc=0.666, loss=64.089, backward_time=0.403, grad_norm=33.469, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.636e-04, train_time=1.753
+[gpub078:0/16] 2024-01-24 23:54:29,864 (trainer:737) INFO: 8epoch:train:8501-8600batch: iter_time=9.965e-05, forward_time=0.289, loss_ctc=57.362, loss_att=53.344, acc=0.697, loss=54.549, backward_time=0.396, grad_norm=28.308, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.635e-04, train_time=2.342
+[gpub078:0/16] 2024-01-24 23:57:00,369 (trainer:737) INFO: 8epoch:train:8601-8700batch: iter_time=3.252e-04, forward_time=0.310, loss_ctc=62.829, loss_att=57.121, acc=0.693, loss=58.833, backward_time=0.408, grad_norm=29.109, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.098, optim0_lr0=3.633e-04, train_time=1.505
+[gpub078:0/16] 2024-01-24 23:59:09,977 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub078:0/16] 2024-01-24 23:59:28,651 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-24 23:59:32,300 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9281cdf7c0>)
+[gpub078:0/16] 2024-01-24 23:59:32,300 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-24 23:59:32,340 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 00:11:19,438 (trainer:737) INFO: 8epoch:train:8701-8800batch: iter_time=6.534, forward_time=0.548, loss_ctc=60.834, loss_att=68.592, acc=0.670, loss=66.265, backward_time=0.470, grad_norm=33.352, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.105, optim0_lr0=3.631e-04, train_time=8.590
+[gpub078:0/16] 2024-01-25 00:14:10,540 (trainer:737) INFO: 8epoch:train:8801-8900batch: iter_time=8.277e-05, forward_time=0.288, loss_ctc=55.472, loss_att=51.518, acc=0.710, loss=52.704, backward_time=0.401, grad_norm=32.390, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.092, optim0_lr0=3.630e-04, train_time=1.711
+[gpub078:0/16] 2024-01-25 00:17:25,268 (trainer:737) INFO: 8epoch:train:8901-9000batch: iter_time=8.526e-05, forward_time=0.292, loss_ctc=60.573, loss_att=67.499, acc=0.679, loss=65.421, backward_time=0.414, grad_norm=28.087, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.100, optim0_lr0=3.628e-04, train_time=1.947
+[gpub078:0/16] 2024-01-25 00:20:54,382 (trainer:737) INFO: 8epoch:train:9001-9100batch: iter_time=8.638e-05, forward_time=0.560, loss_ctc=53.441, loss_att=56.378, acc=0.692, loss=55.497, backward_time=0.484, grad_norm=26.422, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.109, optim0_lr0=3.627e-04, train_time=2.089
+[gpub078:0/16] 2024-01-25 00:23:37,900 (trainer:737) INFO: 8epoch:train:9101-9200batch: iter_time=8.396e-05, forward_time=0.290, loss_ctc=56.064, loss_att=57.980, acc=0.689, loss=57.405, backward_time=0.401, grad_norm=26.989, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.092, optim0_lr0=3.625e-04, train_time=1.637
+[gpub078:0/16] 2024-01-25 00:26:59,187 (trainer:737) INFO: 8epoch:train:9201-9300batch: iter_time=8.910e-05, forward_time=0.303, loss_ctc=63.853, loss_att=60.470, acc=0.710, loss=61.485, backward_time=0.438, grad_norm=33.214, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.097, optim0_lr0=3.623e-04, train_time=2.013
+[gpub078:0/16] 2024-01-25 00:30:56,390 (trainer:737) INFO: 8epoch:train:9301-9400batch: iter_time=8.871e-05, forward_time=0.569, loss_ctc=51.494, loss_att=52.538, acc=0.692, loss=52.225, backward_time=0.516, grad_norm=25.411, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.110, optim0_lr0=3.622e-04, train_time=2.371
+[gpub078:0/16] 2024-01-25 00:33:33,453 (trainer:737) INFO: 8epoch:train:9401-9500batch: iter_time=9.037e-05, forward_time=0.292, loss_ctc=69.245, loss_att=69.552, acc=0.680, loss=69.460, backward_time=0.406, grad_norm=31.566, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.092, optim0_lr0=3.620e-04, train_time=1.570
+[gpub078:0/16] 2024-01-25 00:36:28,235 (trainer:737) INFO: 8epoch:train:9501-9600batch: iter_time=9.372e-05, forward_time=0.297, loss_ctc=67.776, loss_att=79.661, acc=0.685, loss=76.095, backward_time=0.411, grad_norm=32.796, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.105, optim0_lr0=3.619e-04, train_time=1.749
+[gpub078:0/16] 2024-01-25 00:40:38,944 (trainer:737) INFO: 8epoch:train:9601-9700batch: iter_time=8.592e-05, forward_time=0.529, loss_ctc=64.299, loss_att=66.376, acc=0.690, loss=65.753, backward_time=0.496, grad_norm=29.311, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.109, optim0_lr0=3.617e-04, train_time=2.507
+[gpub078:0/16] 2024-01-25 00:42:55,191 (trainer:737) INFO: 8epoch:train:9701-9800batch: iter_time=7.615e-05, forward_time=0.290, loss_ctc=61.367, loss_att=58.931, acc=0.684, loss=59.662, backward_time=0.403, grad_norm=30.355, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.092, optim0_lr0=3.615e-04, train_time=1.361
+[gpub078:0/16] 2024-01-25 00:46:23,639 (trainer:737) INFO: 8epoch:train:9801-9900batch: iter_time=7.927e-05, forward_time=0.288, loss_ctc=58.096, loss_att=52.534, acc=0.695, loss=54.202, backward_time=0.398, grad_norm=30.514, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.095, optim0_lr0=3.614e-04, train_time=2.082
+[gpub078:0/16] 2024-01-25 00:49:58,496 (trainer:737) INFO: 8epoch:train:9901-10000batch: iter_time=4.377e-04, forward_time=0.614, loss_ctc=61.839, loss_att=61.462, acc=0.701, loss=61.575, backward_time=0.509, grad_norm=29.643, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.125, optim0_lr0=3.612e-04, train_time=2.150
+[gpub078:0/16] 2024-01-25 00:50:18,600 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub078:0/16] 2024-01-25 00:50:37,534 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 00:50:40,961 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f92865ba500>)
+[gpub078:0/16] 2024-01-25 00:50:40,961 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub078:0/16] 2024-01-25 00:50:40,964 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 01:01:49,466 (trainer:737) INFO: 8epoch:train:10001-10100batch: iter_time=5.544, forward_time=0.339, loss_ctc=62.134, loss_att=61.416, acc=0.689, loss=61.631, backward_time=0.416, grad_norm=32.341, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.094, optim0_lr0=3.611e-04, train_time=7.110
+[gpub078:0/16] 2024-01-25 01:05:06,503 (trainer:737) INFO: 8epoch:train:10101-10200batch: iter_time=8.378e-05, forward_time=0.339, loss_ctc=52.402, loss_att=48.582, acc=0.725, loss=49.728, backward_time=0.405, grad_norm=25.259, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.100, optim0_lr0=3.609e-04, train_time=1.970
+[gpub078:0/16] 2024-01-25 01:07:54,067 (trainer:737) INFO: 8epoch:train:10201-10300batch: iter_time=1.270e-04, forward_time=0.421, loss_ctc=61.047, loss_att=71.359, acc=0.667, loss=68.265, backward_time=0.433, grad_norm=29.622, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.104, optim0_lr0=3.608e-04, train_time=1.675
+[gpub078:0/16] 2024-01-25 01:11:18,011 (trainer:737) INFO: 8epoch:train:10301-10400batch: iter_time=1.020e-04, forward_time=0.349, loss_ctc=54.472, loss_att=57.000, acc=0.685, loss=56.242, backward_time=0.426, grad_norm=30.589, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.097, optim0_lr0=3.606e-04, train_time=2.039
+[gpub078:0/16] 2024-01-25 01:14:58,948 (trainer:737) INFO: 8epoch:train:10401-10500batch: iter_time=1.055e-04, forward_time=0.336, loss_ctc=61.660, loss_att=56.253, acc=0.710, loss=57.875, backward_time=0.403, grad_norm=28.424, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.092, optim0_lr0=3.605e-04, train_time=2.208
+[gpub078:0/16] 2024-01-25 01:17:45,482 (trainer:737) INFO: 8epoch:train:10501-10600batch: iter_time=1.056e-04, forward_time=0.488, loss_ctc=56.365, loss_att=53.712, acc=0.709, loss=54.508, backward_time=0.454, grad_norm=29.520, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.111, optim0_lr0=3.603e-04, train_time=1.666
+[gpub078:0/16] 2024-01-25 01:20:18,081 (trainer:737) INFO: 8epoch:train:10601-10700batch: iter_time=1.074e-04, forward_time=0.312, loss_ctc=62.052, loss_att=63.588, acc=0.677, loss=63.127, backward_time=0.414, grad_norm=29.683, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.099, optim0_lr0=3.601e-04, train_time=1.526
+[gpub078:0/16] 2024-01-25 01:24:17,462 (trainer:737) INFO: 8epoch:train:10701-10800batch: iter_time=1.049e-04, forward_time=0.324, loss_ctc=72.880, loss_att=74.173, acc=0.684, loss=73.785, backward_time=0.423, grad_norm=33.507, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.097, optim0_lr0=3.600e-04, train_time=2.394
+[gpub078:0/16] 2024-01-25 01:26:38,082 (trainer:737) INFO: 8epoch:train:10801-10900batch: iter_time=1.043e-04, forward_time=0.304, loss_ctc=63.771, loss_att=71.333, acc=0.706, loss=69.064, backward_time=0.408, grad_norm=28.817, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.092, optim0_lr0=3.598e-04, train_time=1.405
+[gpub078:0/16] 2024-01-25 01:30:18,879 (trainer:737) INFO: 8epoch:train:10901-11000batch: iter_time=1.129e-04, forward_time=0.565, loss_ctc=64.777, loss_att=63.710, acc=0.681, loss=64.030, backward_time=0.496, grad_norm=33.086, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.098, optim0_lr0=3.597e-04, train_time=2.208
+[gpub078:0/16] 2024-01-25 01:33:28,182 (trainer:737) INFO: 8epoch:train:11001-11100batch: iter_time=6.660e-04, forward_time=0.327, loss_ctc=56.778, loss_att=53.678, acc=0.706, loss=54.608, backward_time=0.435, grad_norm=26.646, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.104, optim0_lr0=3.595e-04, train_time=1.891
+[gpub078:0/16] 2024-01-25 01:37:00,712 (trainer:737) INFO: 8epoch:train:11101-11200batch: iter_time=1.210e-04, forward_time=0.287, loss_ctc=61.886, loss_att=56.577, acc=0.702, loss=58.170, backward_time=0.400, grad_norm=29.821, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.094, optim0_lr0=3.594e-04, train_time=2.127
+[gpub078:0/16] 2024-01-25 01:38:52,617 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub078:0/16] 2024-01-25 01:39:11,416 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 01:39:14,948 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f92517b1630>)
+[gpub078:0/16] 2024-01-25 01:39:14,948 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub078:0/16] 2024-01-25 01:39:14,971 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 01:48:05,115 (trainer:737) INFO: 8epoch:train:11201-11300batch: iter_time=4.831, forward_time=0.442, loss_ctc=60.038, loss_att=67.901, acc=0.668, loss=65.542, backward_time=0.425, grad_norm=31.619, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.095, optim0_lr0=3.592e-04, train_time=6.644
+[gpub078:0/16] 2024-01-25 01:51:26,815 (trainer:737) INFO: 8epoch:train:11301-11400batch: iter_time=2.415e-04, forward_time=0.335, loss_ctc=54.436, loss_att=50.228, acc=0.705, loss=51.491, backward_time=0.411, grad_norm=28.537, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.096, optim0_lr0=3.591e-04, train_time=2.013
+[gpub078:0/16] 2024-01-25 01:54:18,742 (trainer:737) INFO: 8epoch:train:11401-11500batch: iter_time=8.561e-05, forward_time=0.289, loss_ctc=59.821, loss_att=65.393, acc=0.678, loss=63.721, backward_time=0.400, grad_norm=28.882, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.091, optim0_lr0=3.589e-04, train_time=1.723
+[gpub078:0/16] 2024-01-25 01:56:42,279 (trainer:737) INFO: 8epoch:train:11501-11600batch: iter_time=8.373e-05, forward_time=0.348, loss_ctc=53.037, loss_att=55.442, acc=0.691, loss=54.721, backward_time=0.432, grad_norm=26.054, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.097, optim0_lr0=3.587e-04, train_time=1.435
+[gpub078:0/16] 2024-01-25 01:59:16,772 (trainer:737) INFO: 8epoch:train:11601-11700batch: iter_time=3.943e-04, forward_time=0.307, loss_ctc=55.823, loss_att=54.977, acc=0.687, loss=55.231, backward_time=0.413, grad_norm=28.236, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.095, optim0_lr0=3.586e-04, train_time=1.545
+[gpub078:0/16] 2024-01-25 02:01:39,723 (trainer:737) INFO: 8epoch:train:11701-11800batch: iter_time=8.632e-05, forward_time=0.295, loss_ctc=63.336, loss_att=60.078, acc=0.702, loss=61.055, backward_time=0.406, grad_norm=33.716, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.093, optim0_lr0=3.584e-04, train_time=1.428
+[gpub078:0/16] 2024-01-25 02:04:10,951 (trainer:737) INFO: 8epoch:train:11801-11900batch: iter_time=8.785e-05, forward_time=0.290, loss_ctc=51.488, loss_att=53.311, acc=0.685, loss=52.764, backward_time=0.399, grad_norm=26.423, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.092, optim0_lr0=3.583e-04, train_time=1.513
+[gpub078:0/16] 2024-01-25 02:07:33,284 (trainer:737) INFO: 8epoch:train:11901-12000batch: iter_time=0.059, forward_time=0.415, loss_ctc=68.773, loss_att=68.366, acc=0.674, loss=68.488, backward_time=0.430, grad_norm=32.145, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.093, optim0_lr0=3.581e-04, train_time=2.023
+[gpub078:0/16] 2024-01-25 02:10:04,405 (trainer:737) INFO: 8epoch:train:12001-12100batch: iter_time=5.194e-04, forward_time=0.310, loss_ctc=67.047, loss_att=79.097, acc=0.667, loss=75.482, backward_time=0.424, grad_norm=31.817, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.099, optim0_lr0=3.580e-04, train_time=1.511
+[gpub078:0/16] 2024-01-25 02:12:22,196 (trainer:737) INFO: 8epoch:train:12101-12200batch: iter_time=8.166e-05, forward_time=0.289, loss_ctc=63.841, loss_att=64.768, acc=0.687, loss=64.490, backward_time=0.402, grad_norm=30.122, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.092, optim0_lr0=3.578e-04, train_time=1.377
+[gpub078:0/16] 2024-01-25 02:15:11,231 (trainer:737) INFO: 8epoch:train:12201-12300batch: iter_time=8.638e-05, forward_time=0.288, loss_ctc=60.222, loss_att=57.120, acc=0.680, loss=58.050, backward_time=0.400, grad_norm=31.627, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.092, optim0_lr0=3.577e-04, train_time=1.691
+[gpub078:0/16] 2024-01-25 02:17:53,493 (trainer:737) INFO: 8epoch:train:12301-12400batch: iter_time=8.241e-05, forward_time=0.357, loss_ctc=58.536, loss_att=51.930, acc=0.693, loss=53.912, backward_time=0.470, grad_norm=29.847, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.103, optim0_lr0=3.575e-04, train_time=1.622
+[gpub078:0/16] 2024-01-25 02:20:19,380 (trainer:737) INFO: 8epoch:train:12401-12500batch: iter_time=8.710e-05, forward_time=0.310, loss_ctc=60.917, loss_att=61.058, acc=0.690, loss=61.016, backward_time=0.410, grad_norm=30.314, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.094, optim0_lr0=3.574e-04, train_time=1.459
+[gpub078:0/16] 2024-01-25 02:20:39,617 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub078:0/16] 2024-01-25 02:20:58,806 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 02:21:02,459 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9225172bc0>)
+[gpub078:0/16] 2024-01-25 02:21:02,459 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub078:0/16] 2024-01-25 02:21:02,463 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 02:28:52,268 (trainer:737) INFO: 8epoch:train:12501-12600batch: iter_time=3.539, forward_time=0.292, loss_ctc=60.714, loss_att=59.880, acc=0.677, loss=60.130, backward_time=0.402, grad_norm=33.364, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.092, optim0_lr0=3.572e-04, train_time=5.128
+[gpub078:0/16] 2024-01-25 02:31:49,481 (trainer:737) INFO: 8epoch:train:12601-12700batch: iter_time=2.577e-04, forward_time=0.409, loss_ctc=51.817, loss_att=47.262, acc=0.722, loss=48.628, backward_time=0.412, grad_norm=25.554, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.103, optim0_lr0=3.571e-04, train_time=1.773
+[gpub078:0/16] 2024-01-25 02:34:03,809 (trainer:737) INFO: 8epoch:train:12701-12800batch: iter_time=8.729e-05, forward_time=0.303, loss_ctc=60.320, loss_att=70.076, acc=0.661, loss=67.149, backward_time=0.407, grad_norm=28.213, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.093, optim0_lr0=3.569e-04, train_time=1.342
+[gpub078:0/16] 2024-01-25 02:36:52,143 (trainer:737) INFO: 8epoch:train:12801-12900batch: iter_time=9.992e-05, forward_time=0.286, loss_ctc=54.227, loss_att=54.101, acc=0.689, loss=54.139, backward_time=0.397, grad_norm=28.533, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.092, optim0_lr0=3.568e-04, train_time=1.684
+[gpub078:0/16] 2024-01-25 02:39:38,067 (trainer:737) INFO: 8epoch:train:12901-13000batch: iter_time=9.090e-05, forward_time=0.380, loss_ctc=61.217, loss_att=55.058, acc=0.706, loss=56.906, backward_time=0.432, grad_norm=27.487, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.097, optim0_lr0=3.566e-04, train_time=1.658
+[gpub078:0/16] 2024-01-25 02:42:07,142 (trainer:737) INFO: 8epoch:train:13001-13100batch: iter_time=8.999e-05, forward_time=0.322, loss_ctc=52.556, loss_att=53.660, acc=0.697, loss=53.329, backward_time=0.410, grad_norm=29.676, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.094, optim0_lr0=3.565e-04, train_time=1.491
+[gpub078:0/16] 2024-01-25 02:44:40,586 (trainer:737) INFO: 8epoch:train:13101-13200batch: iter_time=9.359e-05, forward_time=0.291, loss_ctc=60.911, loss_att=62.241, acc=0.677, loss=61.842, backward_time=0.404, grad_norm=29.782, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.092, optim0_lr0=3.563e-04, train_time=1.533
+[gpub078:0/16] 2024-01-25 02:47:49,508 (trainer:737) INFO: 8epoch:train:13201-13300batch: iter_time=9.399e-05, forward_time=0.379, loss_ctc=72.104, loss_att=73.335, acc=0.675, loss=72.966, backward_time=0.440, grad_norm=33.135, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.098, optim0_lr0=3.562e-04, train_time=1.890
+[gpub078:0/16] 2024-01-25 02:50:18,412 (trainer:737) INFO: 8epoch:train:13301-13400batch: iter_time=9.763e-05, forward_time=0.316, loss_ctc=63.212, loss_att=70.237, acc=0.692, loss=68.130, backward_time=0.413, grad_norm=30.153, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.098, optim0_lr0=3.560e-04, train_time=1.489
+[gpub078:0/16] 2024-01-25 02:52:57,054 (trainer:737) INFO: 8epoch:train:13401-13500batch: iter_time=1.063e-04, forward_time=0.288, loss_ctc=63.805, loss_att=62.508, acc=0.671, loss=62.897, backward_time=0.399, grad_norm=33.353, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.092, optim0_lr0=3.559e-04, train_time=1.586
+[gpub078:0/16] 2024-01-25 02:55:35,952 (trainer:737) INFO: 8epoch:train:13501-13600batch: iter_time=5.124e-04, forward_time=0.394, loss_ctc=55.754, loss_att=53.019, acc=0.700, loss=53.839, backward_time=0.421, grad_norm=27.365, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.101, optim0_lr0=3.557e-04, train_time=1.587
+[gpub078:0/16] 2024-01-25 02:58:26,453 (trainer:737) INFO: 8epoch:train:13601-13700batch: iter_time=1.019e-04, forward_time=0.310, loss_ctc=61.690, loss_att=56.335, acc=0.699, loss=57.942, backward_time=0.409, grad_norm=31.652, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.102, optim0_lr0=3.556e-04, train_time=1.706
+[gpub078:0/16] 2024-01-25 03:00:10,614 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub078:0/16] 2024-01-25 03:00:29,167 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 03:00:32,644 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f924d6eb7c0>)
+[gpub078:0/16] 2024-01-25 03:00:32,644 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub078:0/16] 2024-01-25 03:00:32,721 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 03:07:08,485 (trainer:737) INFO: 8epoch:train:13701-13800batch: iter_time=3.544, forward_time=0.350, loss_ctc=59.593, loss_att=67.475, acc=0.674, loss=65.111, backward_time=0.416, grad_norm=31.910, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.092, optim0_lr0=3.554e-04, train_time=5.220
+[gpub078:0/16] 2024-01-25 03:09:28,655 (trainer:737) INFO: 8epoch:train:13801-13900batch: iter_time=8.069e-05, forward_time=0.326, loss_ctc=54.046, loss_att=50.474, acc=0.714, loss=51.546, backward_time=0.410, grad_norm=30.288, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.095, optim0_lr0=3.553e-04, train_time=1.401
+[gpub078:0/16] 2024-01-25 03:12:23,906 (trainer:737) INFO: 8epoch:train:13901-14000batch: iter_time=7.989e-05, forward_time=0.320, loss_ctc=59.267, loss_att=65.722, acc=0.687, loss=63.786, backward_time=0.411, grad_norm=28.137, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.098, optim0_lr0=3.551e-04, train_time=1.753
+[gpub078:0/16] 2024-01-25 03:15:07,094 (trainer:737) INFO: 8epoch:train:14001-14100batch: iter_time=9.019e-05, forward_time=0.336, loss_ctc=51.967, loss_att=54.981, acc=0.700, loss=54.077, backward_time=0.411, grad_norm=26.587, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.097, optim0_lr0=3.550e-04, train_time=1.632
+[gpub078:0/16] 2024-01-25 03:17:30,243 (trainer:737) INFO: 8epoch:train:14101-14200batch: iter_time=9.332e-05, forward_time=0.338, loss_ctc=55.440, loss_att=57.432, acc=0.690, loss=56.835, backward_time=0.409, grad_norm=28.564, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.096, optim0_lr0=3.548e-04, train_time=1.431
+[gpub078:0/16] 2024-01-25 03:20:12,200 (trainer:737) INFO: 8epoch:train:14201-14300batch: iter_time=8.946e-05, forward_time=0.292, loss_ctc=63.439, loss_att=60.127, acc=0.713, loss=61.121, backward_time=0.406, grad_norm=30.423, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.093, optim0_lr0=3.547e-04, train_time=1.619
+[gpub078:0/16] 2024-01-25 03:22:56,515 (trainer:737) INFO: 8epoch:train:14301-14400batch: iter_time=8.708e-05, forward_time=0.317, loss_ctc=50.858, loss_att=52.369, acc=0.698, loss=51.916, backward_time=0.410, grad_norm=26.054, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.098, optim0_lr0=3.545e-04, train_time=1.643
+[gpub078:0/16] 2024-01-25 03:25:19,537 (trainer:737) INFO: 8epoch:train:14401-14500batch: iter_time=8.892e-05, forward_time=0.344, loss_ctc=67.853, loss_att=68.817, acc=0.684, loss=68.528, backward_time=0.423, grad_norm=32.149, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.100, optim0_lr0=3.544e-04, train_time=1.430
+[gpub078:0/16] 2024-01-25 03:27:50,621 (trainer:737) INFO: 8epoch:train:14501-14600batch: iter_time=8.837e-05, forward_time=0.331, loss_ctc=67.369, loss_att=79.393, acc=0.689, loss=75.786, backward_time=0.421, grad_norm=31.812, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.092, optim0_lr0=3.542e-04, train_time=1.511
+[gpub078:0/16] 2024-01-25 03:30:41,635 (trainer:737) INFO: 8epoch:train:14601-14700batch: iter_time=8.326e-05, forward_time=0.312, loss_ctc=63.033, loss_att=64.642, acc=0.698, loss=64.159, backward_time=0.428, grad_norm=28.081, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.099, optim0_lr0=3.541e-04, train_time=1.710
+[gpub078:0/16] 2024-01-25 03:33:13,087 (trainer:737) INFO: 8epoch:train:14701-14800batch: iter_time=1.414e-04, forward_time=0.378, loss_ctc=60.311, loss_att=56.914, acc=0.692, loss=57.933, backward_time=0.426, grad_norm=31.694, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.102, optim0_lr0=3.539e-04, train_time=1.514
+[gpub078:0/16] 2024-01-25 03:35:30,330 (trainer:737) INFO: 8epoch:train:14801-14900batch: iter_time=9.876e-05, forward_time=0.306, loss_ctc=57.260, loss_att=51.434, acc=0.703, loss=53.182, backward_time=0.413, grad_norm=28.882, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.093, optim0_lr0=3.538e-04, train_time=1.373
+[gpub078:0/16] 2024-01-25 03:38:09,076 (trainer:737) INFO: 8epoch:train:14901-15000batch: iter_time=1.037e-04, forward_time=0.311, loss_ctc=59.925, loss_att=60.367, acc=0.706, loss=60.234, backward_time=0.418, grad_norm=28.534, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.096, optim0_lr0=3.536e-04, train_time=1.587
+[gpub078:0/16] 2024-01-25 04:16:44,859 (trainer:343) INFO: 8epoch results: [train] iter_time=0.430, forward_time=0.379, loss_ctc=61.446, loss_att=61.842, acc=0.685, loss=61.723, backward_time=0.439, grad_norm=30.474, clip=100.000, loss_scale=1.002e+22, optim_step_time=0.100, optim0_lr0=3.653e-04, train_time=2.356, time=9 hours, 49 minutes and 36.77 seconds, total_count=150000, gpu_max_cached_mem_GB=40.762, [valid] loss_ctc=55.679, cer_ctc=0.283, loss_att=52.799, acc=0.586, cer=0.406, wer=0.999, loss=53.663, time=38 minutes and 9.16 seconds, total_count=46710, gpu_max_cached_mem_GB=40.762
+[gpub078:0/16] 2024-01-25 04:17:02,355 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub078:0/16] 2024-01-25 04:17:02,433 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/3epoch.pth
+[gpub078:0/16] 2024-01-25 04:17:02,434 (trainer:272) INFO: 9/45epoch started. Estimated time to finish: 2 weeks, 2 days and 3 hours
+[gpub078:0/16] 2024-01-25 04:17:02,445 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-25 04:17:20,318 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 04:17:23,634 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f922137a2c0>)
+[gpub078:0/16] 2024-01-25 04:17:23,634 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub078:0/16] 2024-01-25 04:17:23,637 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 04:25:03,710 (trainer:737) INFO: 9epoch:train:1-100batch: iter_time=3.284, forward_time=0.373, loss_ctc=59.540, loss_att=54.501, acc=0.684, loss=56.013, backward_time=0.412, grad_norm=33.090, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.097, optim0_lr0=3.535e-04, train_time=4.812
+[gpub078:0/16] 2024-01-25 04:27:43,057 (trainer:737) INFO: 9epoch:train:101-200batch: iter_time=9.255e-05, forward_time=0.369, loss_ctc=59.005, loss_att=55.471, acc=0.672, loss=56.531, backward_time=0.443, grad_norm=32.600, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.102, optim0_lr0=3.533e-04, train_time=1.593
+[gpub078:0/16] 2024-01-25 04:30:04,906 (trainer:737) INFO: 9epoch:train:201-300batch: iter_time=1.003e-04, forward_time=0.287, loss_ctc=54.760, loss_att=53.822, acc=0.661, loss=54.103, backward_time=0.402, grad_norm=31.011, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.532e-04, train_time=1.418
+[gpub078:0/16] 2024-01-25 04:32:54,617 (trainer:737) INFO: 9epoch:train:301-400batch: iter_time=1.029e-04, forward_time=0.382, loss_ctc=61.752, loss_att=61.927, acc=0.676, loss=61.874, backward_time=0.441, grad_norm=31.582, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.101, optim0_lr0=3.530e-04, train_time=1.697
+[gpub078:0/16] 2024-01-25 04:35:26,603 (trainer:737) INFO: 9epoch:train:401-500batch: iter_time=2.467e-04, forward_time=0.311, loss_ctc=55.618, loss_att=50.012, acc=0.707, loss=51.694, backward_time=0.402, grad_norm=28.972, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.529e-04, train_time=1.520
+[gpub078:0/16] 2024-01-25 04:38:13,245 (trainer:737) INFO: 9epoch:train:501-600batch: iter_time=1.049e-04, forward_time=0.375, loss_ctc=64.141, loss_att=60.389, acc=0.678, loss=61.514, backward_time=0.420, grad_norm=32.275, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.101, optim0_lr0=3.527e-04, train_time=1.665
+[gpub078:0/16] 2024-01-25 04:40:39,823 (trainer:737) INFO: 9epoch:train:601-700batch: iter_time=9.987e-05, forward_time=0.296, loss_ctc=58.381, loss_att=56.918, acc=0.661, loss=57.357, backward_time=0.403, grad_norm=31.078, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.095, optim0_lr0=3.526e-04, train_time=1.466
+[gpub078:0/16] 2024-01-25 04:43:13,060 (trainer:737) INFO: 9epoch:train:701-800batch: iter_time=6.269e-04, forward_time=0.310, loss_ctc=68.866, loss_att=56.667, acc=0.679, loss=60.326, backward_time=0.412, grad_norm=33.344, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.097, optim0_lr0=3.525e-04, train_time=1.531
+[gpub078:0/16] 2024-01-25 04:45:57,510 (trainer:737) INFO: 9epoch:train:801-900batch: iter_time=9.241e-05, forward_time=0.372, loss_ctc=56.151, loss_att=54.649, acc=0.679, loss=55.100, backward_time=0.421, grad_norm=29.886, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.102, optim0_lr0=3.523e-04, train_time=1.645
+[gpub078:0/16] 2024-01-25 04:48:35,102 (trainer:737) INFO: 9epoch:train:901-1000batch: iter_time=1.942e-04, forward_time=0.345, loss_ctc=61.206, loss_att=59.848, acc=0.669, loss=60.256, backward_time=0.402, grad_norm=33.220, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.522e-04, train_time=1.576
+[gpub078:0/16] 2024-01-25 04:51:13,567 (trainer:737) INFO: 9epoch:train:1001-1100batch: iter_time=9.167e-05, forward_time=0.342, loss_ctc=64.709, loss_att=53.551, acc=0.677, loss=56.899, backward_time=0.446, grad_norm=33.499, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.098, optim0_lr0=3.520e-04, train_time=1.584
+[gpub078:0/16] 2024-01-25 04:53:53,492 (trainer:737) INFO: 9epoch:train:1101-1200batch: iter_time=7.517e-04, forward_time=0.349, loss_ctc=66.426, loss_att=66.681, acc=0.670, loss=66.604, backward_time=0.412, grad_norm=33.902, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.095, optim0_lr0=3.519e-04, train_time=1.598
+[gpub078:0/16] 2024-01-25 04:55:30,277 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub078:0/16] 2024-01-25 04:55:49,016 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 04:55:52,655 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f92257c7d60>)
+[gpub078:0/16] 2024-01-25 04:55:52,655 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub078:0/16] 2024-01-25 04:55:52,658 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 05:02:21,735 (trainer:737) INFO: 9epoch:train:1201-1300batch: iter_time=3.529, forward_time=0.298, loss_ctc=53.417, loss_att=58.993, acc=0.667, loss=57.320, backward_time=0.412, grad_norm=27.582, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.517e-04, train_time=5.083
+[gpub078:0/16] 2024-01-25 05:04:53,791 (trainer:737) INFO: 9epoch:train:1301-1400batch: iter_time=9.129e-05, forward_time=0.288, loss_ctc=63.420, loss_att=59.033, acc=0.675, loss=60.349, backward_time=0.401, grad_norm=33.927, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.093, optim0_lr0=3.516e-04, train_time=1.520
+[gpub078:0/16] 2024-01-25 05:07:19,992 (trainer:737) INFO: 9epoch:train:1401-1500batch: iter_time=9.284e-05, forward_time=0.341, loss_ctc=49.896, loss_att=47.035, acc=0.685, loss=47.894, backward_time=0.427, grad_norm=27.144, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.106, optim0_lr0=3.514e-04, train_time=1.461
+[gpub078:0/16] 2024-01-25 05:10:02,051 (trainer:737) INFO: 9epoch:train:1501-1600batch: iter_time=1.014e-04, forward_time=0.310, loss_ctc=61.557, loss_att=59.105, acc=0.678, loss=59.841, backward_time=0.418, grad_norm=34.578, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.513e-04, train_time=1.621
+[gpub078:0/16] 2024-01-25 05:12:43,589 (trainer:737) INFO: 9epoch:train:1601-1700batch: iter_time=8.843e-05, forward_time=0.293, loss_ctc=55.529, loss_att=49.111, acc=0.707, loss=51.036, backward_time=0.406, grad_norm=28.926, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.094, optim0_lr0=3.511e-04, train_time=1.615
+[gpub078:0/16] 2024-01-25 05:15:05,017 (trainer:737) INFO: 9epoch:train:1701-1800batch: iter_time=8.935e-05, forward_time=0.288, loss_ctc=62.456, loss_att=57.150, acc=0.692, loss=58.742, backward_time=0.401, grad_norm=31.292, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.093, optim0_lr0=3.510e-04, train_time=1.414
+[gpub078:0/16] 2024-01-25 05:17:40,631 (trainer:737) INFO: 9epoch:train:1801-1900batch: iter_time=8.920e-05, forward_time=0.335, loss_ctc=57.350, loss_att=59.875, acc=0.671, loss=59.117, backward_time=0.435, grad_norm=29.049, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.097, optim0_lr0=3.509e-04, train_time=1.555
+[gpub078:0/16] 2024-01-25 05:20:19,028 (trainer:737) INFO: 9epoch:train:1901-2000batch: iter_time=8.694e-05, forward_time=0.318, loss_ctc=62.641, loss_att=57.904, acc=0.663, loss=59.325, backward_time=0.415, grad_norm=32.176, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.095, optim0_lr0=3.507e-04, train_time=1.584
+[gpub078:0/16] 2024-01-25 05:22:45,502 (trainer:737) INFO: 9epoch:train:2001-2100batch: iter_time=8.912e-05, forward_time=0.302, loss_ctc=61.589, loss_att=48.002, acc=0.703, loss=52.078, backward_time=0.402, grad_norm=31.476, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.094, optim0_lr0=3.506e-04, train_time=1.464
+[gpub078:0/16] 2024-01-25 05:25:06,060 (trainer:737) INFO: 9epoch:train:2101-2200batch: iter_time=9.821e-05, forward_time=0.289, loss_ctc=59.195, loss_att=58.164, acc=0.677, loss=58.473, backward_time=0.401, grad_norm=31.306, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.093, optim0_lr0=3.504e-04, train_time=1.405
+[gpub078:0/16] 2024-01-25 05:27:47,132 (trainer:737) INFO: 9epoch:train:2201-2300batch: iter_time=9.426e-05, forward_time=0.355, loss_ctc=61.097, loss_att=59.837, acc=0.655, loss=60.215, backward_time=0.425, grad_norm=33.082, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.099, optim0_lr0=3.503e-04, train_time=1.610
+[gpub078:0/16] 2024-01-25 05:30:20,514 (trainer:737) INFO: 9epoch:train:2301-2400batch: iter_time=8.975e-05, forward_time=0.318, loss_ctc=57.810, loss_att=53.481, acc=0.710, loss=54.780, backward_time=0.414, grad_norm=28.770, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.097, optim0_lr0=3.501e-04, train_time=1.534
+[gpub078:0/16] 2024-01-25 05:32:55,568 (trainer:737) INFO: 9epoch:train:2401-2500batch: iter_time=1.024e-04, forward_time=0.294, loss_ctc=65.875, loss_att=70.161, acc=0.648, loss=68.875, backward_time=0.415, grad_norm=33.568, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.094, optim0_lr0=3.500e-04, train_time=1.550
+[gpub078:0/16] 2024-01-25 05:33:15,749 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub078:0/16] 2024-01-25 05:33:34,259 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 05:33:38,084 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f938d4239d0>)
+[gpub078:0/16] 2024-01-25 05:33:38,084 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub078:0/16] 2024-01-25 05:33:38,087 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 05:41:22,060 (trainer:737) INFO: 9epoch:train:2501-2600batch: iter_time=3.544, forward_time=0.310, loss_ctc=56.410, loss_att=53.982, acc=0.705, loss=54.710, backward_time=0.423, grad_norm=29.857, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.097, optim0_lr0=3.499e-04, train_time=5.065
+[gpub078:0/16] 2024-01-25 05:43:46,744 (trainer:737) INFO: 9epoch:train:2601-2700batch: iter_time=8.820e-05, forward_time=0.320, loss_ctc=55.353, loss_att=53.945, acc=0.681, loss=54.367, backward_time=0.419, grad_norm=30.859, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.099, optim0_lr0=3.497e-04, train_time=1.446
+[gpub078:0/16] 2024-01-25 05:46:31,959 (trainer:737) INFO: 9epoch:train:2701-2800batch: iter_time=8.713e-05, forward_time=0.334, loss_ctc=52.599, loss_att=54.481, acc=0.671, loss=53.917, backward_time=0.409, grad_norm=28.744, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.095, optim0_lr0=3.496e-04, train_time=1.652
+[gpub078:0/16] 2024-01-25 05:49:10,313 (trainer:737) INFO: 9epoch:train:2801-2900batch: iter_time=9.402e-05, forward_time=0.290, loss_ctc=59.565, loss_att=60.900, acc=0.687, loss=60.500, backward_time=0.402, grad_norm=29.283, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.093, optim0_lr0=3.494e-04, train_time=1.583
+[gpub078:0/16] 2024-01-25 05:51:35,879 (trainer:737) INFO: 9epoch:train:2901-3000batch: iter_time=8.970e-05, forward_time=0.334, loss_ctc=54.097, loss_att=51.993, acc=0.709, loss=52.625, backward_time=0.409, grad_norm=28.436, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.096, optim0_lr0=3.493e-04, train_time=1.456
+[gpub078:0/16] 2024-01-25 05:54:36,079 (trainer:737) INFO: 9epoch:train:3001-3100batch: iter_time=8.920e-05, forward_time=0.314, loss_ctc=61.978, loss_att=61.376, acc=0.685, loss=61.557, backward_time=0.447, grad_norm=31.657, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.097, optim0_lr0=3.491e-04, train_time=1.801
+[gpub078:0/16] 2024-01-25 05:57:07,983 (trainer:737) INFO: 9epoch:train:3101-3200batch: iter_time=9.091e-05, forward_time=0.316, loss_ctc=56.404, loss_att=58.187, acc=0.675, loss=57.652, backward_time=0.424, grad_norm=30.671, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.096, optim0_lr0=3.490e-04, train_time=1.518
+[gpub078:0/16] 2024-01-25 05:59:35,628 (trainer:737) INFO: 9epoch:train:3201-3300batch: iter_time=9.890e-05, forward_time=0.292, loss_ctc=65.147, loss_att=56.550, acc=0.697, loss=59.129, backward_time=0.402, grad_norm=31.757, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.093, optim0_lr0=3.489e-04, train_time=1.477
+[gpub078:0/16] 2024-01-25 06:02:29,588 (trainer:737) INFO: 9epoch:train:3301-3400batch: iter_time=1.079e-04, forward_time=0.311, loss_ctc=54.830, loss_att=53.576, acc=0.694, loss=53.952, backward_time=0.414, grad_norm=33.398, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.094, optim0_lr0=3.487e-04, train_time=1.740
+[gpub078:0/16] 2024-01-25 06:04:58,315 (trainer:737) INFO: 9epoch:train:3401-3500batch: iter_time=9.897e-05, forward_time=0.332, loss_ctc=58.909, loss_att=56.927, acc=0.688, loss=57.521, backward_time=0.424, grad_norm=31.943, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.098, optim0_lr0=3.486e-04, train_time=1.486
+[gpub078:0/16] 2024-01-25 06:07:53,515 (trainer:737) INFO: 9epoch:train:3501-3600batch: iter_time=9.266e-05, forward_time=0.309, loss_ctc=60.721, loss_att=52.786, acc=0.690, loss=55.166, backward_time=0.445, grad_norm=48.163, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.098, optim0_lr0=3.484e-04, train_time=1.751
+[gpub078:0/16] 2024-01-25 06:10:17,720 (trainer:737) INFO: 9epoch:train:3601-3700batch: iter_time=9.478e-05, forward_time=0.293, loss_ctc=64.251, loss_att=66.144, acc=0.685, loss=65.576, backward_time=0.407, grad_norm=31.757, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.093, optim0_lr0=3.483e-04, train_time=1.443
+[gpub078:0/16] 2024-01-25 06:11:54,072 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub078:0/16] 2024-01-25 06:12:12,892 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 06:12:16,491 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b61e6b7c0>)
+[gpub078:0/16] 2024-01-25 06:12:16,491 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-25 06:12:16,494 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 06:18:49,582 (trainer:737) INFO: 9epoch:train:3701-3800batch: iter_time=3.559, forward_time=0.349, loss_ctc=52.426, loss_att=59.834, acc=0.680, loss=57.612, backward_time=0.411, grad_norm=26.897, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.094, optim0_lr0=3.482e-04, train_time=5.119
+[gpub078:0/16] 2024-01-25 06:21:14,919 (trainer:737) INFO: 9epoch:train:3801-3900batch: iter_time=7.853e-05, forward_time=0.307, loss_ctc=62.954, loss_att=59.195, acc=0.687, loss=60.323, backward_time=0.408, grad_norm=32.846, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.095, optim0_lr0=3.480e-04, train_time=1.453
+[gpub078:0/16] 2024-01-25 06:23:44,466 (trainer:737) INFO: 9epoch:train:3901-4000batch: iter_time=8.467e-05, forward_time=0.307, loss_ctc=49.089, loss_att=46.298, acc=0.697, loss=47.135, backward_time=0.412, grad_norm=25.726, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.094, optim0_lr0=3.479e-04, train_time=1.494
+[gpub078:0/16] 2024-01-25 06:26:27,197 (trainer:737) INFO: 9epoch:train:4001-4100batch: iter_time=9.564e-05, forward_time=0.289, loss_ctc=59.873, loss_att=60.570, acc=0.683, loss=60.361, backward_time=0.401, grad_norm=31.724, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.477e-04, train_time=1.628
+[gpub078:0/16] 2024-01-25 06:29:09,660 (trainer:737) INFO: 9epoch:train:4101-4200batch: iter_time=9.721e-05, forward_time=0.335, loss_ctc=54.122, loss_att=49.700, acc=0.708, loss=51.026, backward_time=0.450, grad_norm=27.359, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.098, optim0_lr0=3.476e-04, train_time=1.624
+[gpub078:0/16] 2024-01-25 06:31:37,093 (trainer:737) INFO: 9epoch:train:4201-4300batch: iter_time=8.924e-05, forward_time=0.295, loss_ctc=60.628, loss_att=59.329, acc=0.694, loss=59.719, backward_time=0.410, grad_norm=28.586, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.094, optim0_lr0=3.475e-04, train_time=1.474
+[gpub078:0/16] 2024-01-25 06:34:38,793 (trainer:737) INFO: 9epoch:train:4301-4400batch: iter_time=8.784e-05, forward_time=0.315, loss_ctc=56.065, loss_att=59.942, acc=0.687, loss=58.779, backward_time=0.426, grad_norm=29.100, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.095, optim0_lr0=3.473e-04, train_time=1.817
+[gpub078:0/16] 2024-01-25 06:36:56,767 (trainer:737) INFO: 9epoch:train:4401-4500batch: iter_time=8.555e-05, forward_time=0.288, loss_ctc=61.403, loss_att=58.853, acc=0.678, loss=59.618, backward_time=0.402, grad_norm=30.535, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.472e-04, train_time=1.377
+[gpub078:0/16] 2024-01-25 06:39:36,916 (trainer:737) INFO: 9epoch:train:4501-4600batch: iter_time=8.295e-05, forward_time=0.377, loss_ctc=59.081, loss_att=49.538, acc=0.708, loss=52.401, backward_time=0.426, grad_norm=29.866, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.096, optim0_lr0=3.470e-04, train_time=1.603
+[gpub078:0/16] 2024-01-25 06:42:22,272 (trainer:737) INFO: 9epoch:train:4601-4700batch: iter_time=9.822e-05, forward_time=0.292, loss_ctc=57.820, loss_att=56.568, acc=0.691, loss=56.943, backward_time=0.412, grad_norm=29.059, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.094, optim0_lr0=3.469e-04, train_time=1.653
+[gpub078:0/16] 2024-01-25 06:44:53,529 (trainer:737) INFO: 9epoch:train:4701-4800batch: iter_time=9.653e-05, forward_time=0.314, loss_ctc=59.238, loss_att=59.146, acc=0.670, loss=59.174, backward_time=0.412, grad_norm=31.450, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.096, optim0_lr0=3.468e-04, train_time=1.512
+[gpub078:0/16] 2024-01-25 06:47:24,344 (trainer:737) INFO: 9epoch:train:4801-4900batch: iter_time=9.248e-05, forward_time=0.295, loss_ctc=56.783, loss_att=52.678, acc=0.726, loss=53.909, backward_time=0.404, grad_norm=27.603, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.466e-04, train_time=1.508
+[gpub078:0/16] 2024-01-25 06:50:28,642 (trainer:737) INFO: 9epoch:train:4901-5000batch: iter_time=1.020e-04, forward_time=0.347, loss_ctc=65.423, loss_att=72.750, acc=0.659, loss=70.552, backward_time=0.457, grad_norm=34.871, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.102, optim0_lr0=3.465e-04, train_time=1.841
+[gpub078:0/16] 2024-01-25 06:50:48,745 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub078:0/16] 2024-01-25 06:51:07,553 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 06:51:11,121 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f998850b910>)
+[gpub078:0/16] 2024-01-25 06:51:11,121 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub078:0/16] 2024-01-25 06:51:11,125 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 06:58:48,024 (trainer:737) INFO: 9epoch:train:5001-5100batch: iter_time=3.552, forward_time=0.320, loss_ctc=55.574, loss_att=52.878, acc=0.694, loss=53.687, backward_time=0.406, grad_norm=29.521, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.095, optim0_lr0=3.463e-04, train_time=4.995
+[gpub078:0/16] 2024-01-25 07:01:15,201 (trainer:737) INFO: 9epoch:train:5101-5200batch: iter_time=7.973e-05, forward_time=0.292, loss_ctc=54.804, loss_att=52.383, acc=0.683, loss=53.109, backward_time=0.400, grad_norm=30.085, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.462e-04, train_time=1.472
+[gpub078:0/16] 2024-01-25 07:03:59,502 (trainer:737) INFO: 9epoch:train:5201-5300batch: iter_time=9.243e-05, forward_time=0.286, loss_ctc=51.425, loss_att=52.654, acc=0.667, loss=52.285, backward_time=0.397, grad_norm=30.294, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.461e-04, train_time=1.643
+[gpub078:0/16] 2024-01-25 07:06:38,585 (trainer:737) INFO: 9epoch:train:5301-5400batch: iter_time=6.859e-04, forward_time=0.435, loss_ctc=58.313, loss_att=58.866, acc=0.690, loss=58.700, backward_time=0.423, grad_norm=30.139, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.102, optim0_lr0=3.459e-04, train_time=1.589
+[gpub078:0/16] 2024-01-25 07:09:01,086 (trainer:737) INFO: 9epoch:train:5401-5500batch: iter_time=9.500e-05, forward_time=0.295, loss_ctc=53.114, loss_att=48.565, acc=0.717, loss=49.930, backward_time=0.410, grad_norm=28.123, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.458e-04, train_time=1.427
+[gpub078:0/16] 2024-01-25 07:11:50,390 (trainer:737) INFO: 9epoch:train:5501-5600batch: iter_time=9.089e-05, forward_time=0.291, loss_ctc=60.083, loss_att=58.062, acc=0.690, loss=58.668, backward_time=0.404, grad_norm=27.781, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.456e-04, train_time=1.693
+[gpub078:0/16] 2024-01-25 07:14:36,266 (trainer:737) INFO: 9epoch:train:5601-5700batch: iter_time=9.102e-05, forward_time=0.388, loss_ctc=55.847, loss_att=55.396, acc=0.674, loss=55.531, backward_time=0.431, grad_norm=29.605, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.123, optim0_lr0=3.455e-04, train_time=1.658
+[gpub078:0/16] 2024-01-25 07:17:06,128 (trainer:737) INFO: 9epoch:train:5701-5800batch: iter_time=8.549e-05, forward_time=0.315, loss_ctc=64.222, loss_att=54.231, acc=0.689, loss=57.228, backward_time=0.405, grad_norm=33.498, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.454e-04, train_time=1.497
+[gpub078:0/16] 2024-01-25 07:19:25,634 (trainer:737) INFO: 9epoch:train:5801-5900batch: iter_time=8.067e-05, forward_time=0.293, loss_ctc=53.268, loss_att=51.851, acc=0.692, loss=52.276, backward_time=0.404, grad_norm=29.280, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.452e-04, train_time=1.396
+[gpub078:0/16] 2024-01-25 07:22:04,603 (trainer:737) INFO: 9epoch:train:5901-6000batch: iter_time=9.022e-05, forward_time=0.290, loss_ctc=58.856, loss_att=57.611, acc=0.682, loss=57.985, backward_time=0.402, grad_norm=33.590, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.093, optim0_lr0=3.451e-04, train_time=1.589
+[gpub078:0/16] 2024-01-25 07:24:40,723 (trainer:737) INFO: 9epoch:train:6001-6100batch: iter_time=8.056e-05, forward_time=0.372, loss_ctc=59.550, loss_att=51.507, acc=0.687, loss=53.920, backward_time=0.427, grad_norm=31.001, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.101, optim0_lr0=3.450e-04, train_time=1.561
+[gpub078:0/16] 2024-01-25 07:27:07,481 (trainer:737) INFO: 9epoch:train:6101-6200batch: iter_time=9.125e-05, forward_time=0.307, loss_ctc=63.535, loss_att=64.133, acc=0.680, loss=63.954, backward_time=0.417, grad_norm=34.189, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.448e-04, train_time=1.467
+[gpub078:0/16] 2024-01-25 07:29:09,607 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub078:0/16] 2024-01-25 07:29:28,441 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 07:29:31,994 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f998b835990>)
+[gpub078:0/16] 2024-01-25 07:29:31,995 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub078:0/16] 2024-01-25 07:29:31,998 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 07:36:51,410 (trainer:737) INFO: 9epoch:train:6201-6300batch: iter_time=4.016, forward_time=0.292, loss_ctc=52.397, loss_att=59.314, acc=0.677, loss=57.239, backward_time=0.401, grad_norm=29.134, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.447e-04, train_time=5.839
+[gpub078:0/16] 2024-01-25 07:39:22,118 (trainer:737) INFO: 9epoch:train:6301-6400batch: iter_time=9.030e-05, forward_time=0.363, loss_ctc=60.903, loss_att=58.552, acc=0.689, loss=59.258, backward_time=0.442, grad_norm=32.438, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.103, optim0_lr0=3.446e-04, train_time=1.507
+[gpub078:0/16] 2024-01-25 07:42:18,883 (trainer:737) INFO: 9epoch:train:6401-6500batch: iter_time=9.034e-05, forward_time=0.302, loss_ctc=48.304, loss_att=45.989, acc=0.701, loss=46.684, backward_time=0.406, grad_norm=27.836, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.094, optim0_lr0=3.444e-04, train_time=1.767
+[gpub078:0/16] 2024-01-25 07:44:37,152 (trainer:737) INFO: 9epoch:train:6501-6600batch: iter_time=9.243e-05, forward_time=0.296, loss_ctc=58.945, loss_att=60.052, acc=0.688, loss=59.720, backward_time=0.406, grad_norm=30.464, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.095, optim0_lr0=3.443e-04, train_time=1.382
+[gpub078:0/16] 2024-01-25 07:47:29,677 (trainer:737) INFO: 9epoch:train:6601-6700batch: iter_time=1.857e-04, forward_time=0.437, loss_ctc=53.644, loss_att=48.697, acc=0.711, loss=50.181, backward_time=0.415, grad_norm=26.605, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.099, optim0_lr0=3.441e-04, train_time=1.726
+[gpub078:0/16] 2024-01-25 07:49:59,196 (trainer:737) INFO: 9epoch:train:6701-6800batch: iter_time=1.005e-04, forward_time=0.311, loss_ctc=59.800, loss_att=58.200, acc=0.698, loss=58.680, backward_time=0.406, grad_norm=29.499, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.094, optim0_lr0=3.440e-04, train_time=1.495
+[gpub078:0/16] 2024-01-25 07:52:49,543 (trainer:737) INFO: 9epoch:train:6801-6900batch: iter_time=9.041e-05, forward_time=0.296, loss_ctc=55.968, loss_att=60.875, acc=0.685, loss=59.403, backward_time=0.406, grad_norm=28.881, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.094, optim0_lr0=3.439e-04, train_time=1.703
+[gpub078:0/16] 2024-01-25 07:55:23,751 (trainer:737) INFO: 9epoch:train:6901-7000batch: iter_time=9.819e-05, forward_time=0.442, loss_ctc=60.561, loss_att=58.217, acc=0.681, loss=58.920, backward_time=0.419, grad_norm=29.750, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.097, optim0_lr0=3.437e-04, train_time=1.543
+[gpub078:0/16] 2024-01-25 07:58:13,687 (trainer:737) INFO: 9epoch:train:7001-7100batch: iter_time=9.000e-05, forward_time=0.312, loss_ctc=58.145, loss_att=48.770, acc=0.710, loss=51.582, backward_time=0.408, grad_norm=29.698, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.436e-04, train_time=1.699
+[gpub078:0/16] 2024-01-25 08:00:42,911 (trainer:737) INFO: 9epoch:train:7101-7200batch: iter_time=1.008e-04, forward_time=0.292, loss_ctc=57.323, loss_att=57.149, acc=0.690, loss=57.201, backward_time=0.406, grad_norm=29.957, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.435e-04, train_time=1.492
+[gpub078:0/16] 2024-01-25 08:03:08,577 (trainer:737) INFO: 9epoch:train:7201-7300batch: iter_time=9.678e-05, forward_time=0.288, loss_ctc=58.025, loss_att=59.428, acc=0.667, loss=59.007, backward_time=0.400, grad_norm=30.398, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.433e-04, train_time=1.456
+[gpub078:0/16] 2024-01-25 08:05:56,796 (trainer:737) INFO: 9epoch:train:7301-7400batch: iter_time=7.855e-04, forward_time=0.385, loss_ctc=56.342, loss_att=51.666, acc=0.732, loss=53.069, backward_time=0.435, grad_norm=26.421, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.105, optim0_lr0=3.432e-04, train_time=1.683
+[gpub078:0/16] 2024-01-25 08:08:30,315 (trainer:737) INFO: 9epoch:train:7401-7500batch: iter_time=9.247e-05, forward_time=0.292, loss_ctc=65.134, loss_att=72.387, acc=0.661, loss=70.211, backward_time=0.405, grad_norm=35.113, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.431e-04, train_time=1.535
+[gpub078:0/16] 2024-01-25 08:08:50,343 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub078:0/16] 2024-01-25 08:09:09,415 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 08:09:13,089 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98fb325480>)
+[gpub078:0/16] 2024-01-25 08:09:13,089 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub078:0/16] 2024-01-25 08:09:13,092 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 08:16:48,904 (trainer:737) INFO: 9epoch:train:7501-7600batch: iter_time=3.533, forward_time=0.304, loss_ctc=55.867, loss_att=50.522, acc=0.715, loss=52.125, backward_time=0.404, grad_norm=29.311, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.429e-04, train_time=4.986
+[gpub078:0/16] 2024-01-25 08:19:29,513 (trainer:737) INFO: 9epoch:train:7601-7700batch: iter_time=9.815e-05, forward_time=0.359, loss_ctc=54.489, loss_att=52.500, acc=0.687, loss=53.097, backward_time=0.458, grad_norm=29.702, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.099, optim0_lr0=3.428e-04, train_time=1.606
+[gpub078:0/16] 2024-01-25 08:22:07,875 (trainer:737) INFO: 9epoch:train:7701-7800batch: iter_time=9.448e-05, forward_time=0.310, loss_ctc=51.120, loss_att=52.516, acc=0.682, loss=52.097, backward_time=0.402, grad_norm=27.800, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.094, optim0_lr0=3.427e-04, train_time=1.584
+[gpub078:0/16] 2024-01-25 08:24:54,551 (trainer:737) INFO: 9epoch:train:7801-7900batch: iter_time=9.512e-05, forward_time=0.296, loss_ctc=57.338, loss_att=58.736, acc=0.697, loss=58.316, backward_time=0.405, grad_norm=29.190, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.094, optim0_lr0=3.425e-04, train_time=1.666
+[gpub078:0/16] 2024-01-25 08:27:21,336 (trainer:737) INFO: 9epoch:train:7901-8000batch: iter_time=1.038e-04, forward_time=0.288, loss_ctc=52.796, loss_att=50.168, acc=0.718, loss=50.957, backward_time=0.401, grad_norm=27.586, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.093, optim0_lr0=3.424e-04, train_time=1.468
+[gpub078:0/16] 2024-01-25 08:30:14,223 (trainer:737) INFO: 9epoch:train:8001-8100batch: iter_time=9.110e-05, forward_time=0.412, loss_ctc=58.918, loss_att=59.714, acc=0.692, loss=59.475, backward_time=0.435, grad_norm=28.761, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.102, optim0_lr0=3.423e-04, train_time=1.728
+[gpub078:0/16] 2024-01-25 08:32:47,495 (trainer:737) INFO: 9epoch:train:8101-8200batch: iter_time=9.009e-05, forward_time=0.289, loss_ctc=55.440, loss_att=56.430, acc=0.684, loss=56.133, backward_time=0.401, grad_norm=30.503, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.421e-04, train_time=1.533
+[gpub078:0/16] 2024-01-25 08:35:18,211 (trainer:737) INFO: 9epoch:train:8201-8300batch: iter_time=9.655e-05, forward_time=0.295, loss_ctc=63.674, loss_att=54.506, acc=0.705, loss=57.256, backward_time=0.407, grad_norm=30.085, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.094, optim0_lr0=3.420e-04, train_time=1.506
+[gpub078:0/16] 2024-01-25 08:37:54,935 (trainer:737) INFO: 9epoch:train:8301-8400batch: iter_time=1.012e-04, forward_time=0.288, loss_ctc=53.127, loss_att=52.840, acc=0.698, loss=52.926, backward_time=0.400, grad_norm=27.930, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.419e-04, train_time=1.567
+[gpub078:0/16] 2024-01-25 08:40:38,872 (trainer:737) INFO: 9epoch:train:8401-8500batch: iter_time=9.835e-05, forward_time=0.370, loss_ctc=57.331, loss_att=56.834, acc=0.690, loss=56.983, backward_time=0.446, grad_norm=29.509, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.104, optim0_lr0=3.417e-04, train_time=1.639
+[gpub078:0/16] 2024-01-25 08:43:25,879 (trainer:737) INFO: 9epoch:train:8501-8600batch: iter_time=9.299e-05, forward_time=0.288, loss_ctc=58.336, loss_att=51.286, acc=0.699, loss=53.401, backward_time=0.399, grad_norm=30.280, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.416e-04, train_time=1.671
+[gpub078:0/16] 2024-01-25 08:45:56,030 (trainer:737) INFO: 9epoch:train:8601-8700batch: iter_time=1.099e-04, forward_time=0.296, loss_ctc=63.325, loss_att=63.844, acc=0.696, loss=63.688, backward_time=0.413, grad_norm=32.539, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.094, optim0_lr0=3.415e-04, train_time=1.501
+[gpub078:0/16] 2024-01-25 08:47:32,082 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub078:0/16] 2024-01-25 08:47:51,405 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 08:47:55,012 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98fbadafe0>)
+[gpub078:0/16] 2024-01-25 08:47:55,013 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub078:0/16] 2024-01-25 08:47:55,016 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 08:54:37,109 (trainer:737) INFO: 9epoch:train:8701-8800batch: iter_time=3.663, forward_time=0.395, loss_ctc=51.234, loss_att=60.741, acc=0.677, loss=57.889, backward_time=0.409, grad_norm=28.071, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.095, optim0_lr0=3.413e-04, train_time=5.210
+[gpub078:0/16] 2024-01-25 08:56:56,453 (trainer:737) INFO: 9epoch:train:8801-8900batch: iter_time=9.458e-05, forward_time=0.289, loss_ctc=59.989, loss_att=58.662, acc=0.681, loss=59.060, backward_time=0.403, grad_norm=33.972, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.094, optim0_lr0=3.412e-04, train_time=1.394
+[gpub078:0/16] 2024-01-25 08:59:43,610 (trainer:737) INFO: 9epoch:train:8901-9000batch: iter_time=9.364e-05, forward_time=0.294, loss_ctc=47.967, loss_att=46.540, acc=0.695, loss=46.968, backward_time=0.405, grad_norm=26.939, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.094, optim0_lr0=3.411e-04, train_time=1.671
+[gpub078:0/16] 2024-01-25 09:02:22,208 (trainer:737) INFO: 9epoch:train:9001-9100batch: iter_time=1.036e-04, forward_time=0.313, loss_ctc=57.761, loss_att=58.091, acc=0.686, loss=57.992, backward_time=0.405, grad_norm=31.757, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.099, optim0_lr0=3.409e-04, train_time=1.586
+[gpub078:0/16] 2024-01-25 09:04:54,677 (trainer:737) INFO: 9epoch:train:9101-9200batch: iter_time=1.019e-04, forward_time=0.377, loss_ctc=53.565, loss_att=48.506, acc=0.714, loss=50.024, backward_time=0.428, grad_norm=28.040, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.102, optim0_lr0=3.408e-04, train_time=1.524
+[gpub078:0/16] 2024-01-25 09:07:49,822 (trainer:737) INFO: 9epoch:train:9201-9300batch: iter_time=1.034e-04, forward_time=0.293, loss_ctc=58.735, loss_att=55.932, acc=0.701, loss=56.773, backward_time=0.406, grad_norm=29.157, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.407e-04, train_time=1.752
+[gpub078:0/16] 2024-01-25 09:10:20,185 (trainer:737) INFO: 9epoch:train:9301-9400batch: iter_time=9.511e-05, forward_time=0.295, loss_ctc=55.248, loss_att=58.771, acc=0.681, loss=57.714, backward_time=0.402, grad_norm=29.515, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.094, optim0_lr0=3.405e-04, train_time=1.503
+[gpub078:0/16] 2024-01-25 09:13:20,284 (trainer:737) INFO: 9epoch:train:9401-9500batch: iter_time=9.523e-05, forward_time=0.342, loss_ctc=59.189, loss_att=55.995, acc=0.672, loss=56.953, backward_time=0.475, grad_norm=29.006, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.101, optim0_lr0=3.404e-04, train_time=1.801
+[gpub078:0/16] 2024-01-25 09:15:43,267 (trainer:737) INFO: 9epoch:train:9501-9600batch: iter_time=8.492e-05, forward_time=0.288, loss_ctc=56.312, loss_att=46.468, acc=0.713, loss=49.421, backward_time=0.401, grad_norm=27.981, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.403e-04, train_time=1.429
+[gpub078:0/16] 2024-01-25 09:18:19,872 (trainer:737) INFO: 9epoch:train:9601-9700batch: iter_time=9.191e-05, forward_time=0.298, loss_ctc=57.082, loss_att=56.971, acc=0.686, loss=57.004, backward_time=0.406, grad_norm=29.876, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.401e-04, train_time=1.566
+[gpub078:0/16] 2024-01-25 09:20:51,760 (trainer:737) INFO: 9epoch:train:9701-9800batch: iter_time=9.288e-05, forward_time=0.288, loss_ctc=58.290, loss_att=58.109, acc=0.666, loss=58.163, backward_time=0.402, grad_norm=32.255, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.400e-04, train_time=1.519
+[gpub078:0/16] 2024-01-25 09:23:55,841 (trainer:737) INFO: 9epoch:train:9801-9900batch: iter_time=0.001, forward_time=0.385, loss_ctc=55.443, loss_att=52.497, acc=0.717, loss=53.381, backward_time=0.432, grad_norm=27.596, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.104, optim0_lr0=3.399e-04, train_time=1.839
+[gpub078:0/16] 2024-01-25 09:26:16,073 (trainer:737) INFO: 9epoch:train:9901-10000batch: iter_time=9.743e-05, forward_time=0.291, loss_ctc=64.281, loss_att=69.744, acc=0.655, loss=68.106, backward_time=0.404, grad_norm=32.296, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.093, optim0_lr0=3.397e-04, train_time=1.403
+[gpub078:0/16] 2024-01-25 09:26:36,373 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub078:0/16] 2024-01-25 09:26:55,276 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 09:26:59,114 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f999682feb0>)
+[gpub078:0/16] 2024-01-25 09:26:59,114 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub078:0/16] 2024-01-25 09:26:59,117 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 09:35:11,427 (trainer:737) INFO: 9epoch:train:10001-10100batch: iter_time=3.516, forward_time=0.299, loss_ctc=54.895, loss_att=51.114, acc=0.717, loss=52.248, backward_time=0.402, grad_norm=29.228, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.093, optim0_lr0=3.396e-04, train_time=5.354
+[gpub078:0/16] 2024-01-25 09:38:17,380 (trainer:737) INFO: 9epoch:train:10101-10200batch: iter_time=9.036e-05, forward_time=0.402, loss_ctc=53.419, loss_att=51.897, acc=0.691, loss=52.353, backward_time=0.454, grad_norm=30.228, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.110, optim0_lr0=3.395e-04, train_time=1.859
+[gpub078:0/16] 2024-01-25 09:41:01,951 (trainer:737) INFO: 9epoch:train:10201-10300batch: iter_time=9.266e-05, forward_time=0.286, loss_ctc=50.236, loss_att=52.560, acc=0.683, loss=51.863, backward_time=0.398, grad_norm=28.575, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.394e-04, train_time=1.645
+[gpub078:0/16] 2024-01-25 09:43:42,443 (trainer:737) INFO: 9epoch:train:10301-10400batch: iter_time=9.666e-05, forward_time=0.297, loss_ctc=57.287, loss_att=58.842, acc=0.697, loss=58.376, backward_time=0.407, grad_norm=29.781, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.392e-04, train_time=1.605
+[gpub078:0/16] 2024-01-25 09:46:01,701 (trainer:737) INFO: 9epoch:train:10401-10500batch: iter_time=8.632e-05, forward_time=0.318, loss_ctc=52.574, loss_att=50.178, acc=0.719, loss=50.897, backward_time=0.401, grad_norm=27.266, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.102, optim0_lr0=3.391e-04, train_time=1.393
+[gpub078:0/16] 2024-01-25 09:49:17,034 (trainer:737) INFO: 9epoch:train:10501-10600batch: iter_time=1.000e-04, forward_time=0.402, loss_ctc=59.107, loss_att=58.614, acc=0.695, loss=58.762, backward_time=0.418, grad_norm=29.523, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.095, optim0_lr0=3.390e-04, train_time=1.953
+[gpub078:0/16] 2024-01-25 09:51:48,583 (trainer:737) INFO: 9epoch:train:10601-10700batch: iter_time=9.121e-05, forward_time=0.294, loss_ctc=54.220, loss_att=55.691, acc=0.687, loss=55.250, backward_time=0.406, grad_norm=29.257, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.388e-04, train_time=1.514
+[gpub078:0/16] 2024-01-25 09:54:01,691 (trainer:737) INFO: 9epoch:train:10701-10800batch: iter_time=9.352e-05, forward_time=0.292, loss_ctc=62.157, loss_att=54.379, acc=0.707, loss=56.713, backward_time=0.405, grad_norm=29.567, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.387e-04, train_time=1.332
+[gpub078:0/16] 2024-01-25 09:57:16,626 (trainer:737) INFO: 9epoch:train:10801-10900batch: iter_time=5.044e-04, forward_time=0.381, loss_ctc=51.617, loss_att=51.606, acc=0.704, loss=51.610, backward_time=0.436, grad_norm=27.720, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.106, optim0_lr0=3.386e-04, train_time=1.950
+[gpub078:0/16] 2024-01-25 09:59:50,469 (trainer:737) INFO: 9epoch:train:10901-11000batch: iter_time=8.967e-05, forward_time=0.291, loss_ctc=57.103, loss_att=55.539, acc=0.697, loss=56.008, backward_time=0.409, grad_norm=31.074, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.384e-04, train_time=1.538
+[gpub078:0/16] 2024-01-25 10:02:30,884 (trainer:737) INFO: 9epoch:train:11001-11100batch: iter_time=9.167e-05, forward_time=0.291, loss_ctc=58.106, loss_att=50.797, acc=0.701, loss=52.990, backward_time=0.400, grad_norm=30.815, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.093, optim0_lr0=3.383e-04, train_time=1.602
+[gpub078:0/16] 2024-01-25 10:05:15,914 (trainer:737) INFO: 9epoch:train:11101-11200batch: iter_time=0.001, forward_time=0.363, loss_ctc=62.652, loss_att=63.654, acc=0.698, loss=63.353, backward_time=0.443, grad_norm=30.691, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.101, optim0_lr0=3.382e-04, train_time=1.652
+[gpub078:0/16] 2024-01-25 10:06:49,318 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub078:0/16] 2024-01-25 10:07:08,499 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 10:07:11,989 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f995723b9a0>)
+[gpub078:0/16] 2024-01-25 10:07:11,989 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub078:0/16] 2024-01-25 10:07:11,993 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 10:14:10,831 (trainer:737) INFO: 9epoch:train:11201-11300batch: iter_time=3.729, forward_time=0.295, loss_ctc=50.740, loss_att=58.832, acc=0.687, loss=56.404, backward_time=0.403, grad_norm=27.059, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.381e-04, train_time=5.349
+[gpub078:0/16] 2024-01-25 10:16:39,767 (trainer:737) INFO: 9epoch:train:11301-11400batch: iter_time=8.083e-05, forward_time=0.369, loss_ctc=59.512, loss_att=56.295, acc=0.698, loss=57.260, backward_time=0.420, grad_norm=32.421, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.111, optim0_lr0=3.379e-04, train_time=1.489
+[gpub078:0/16] 2024-01-25 10:19:32,015 (trainer:737) INFO: 9epoch:train:11401-11500batch: iter_time=8.274e-05, forward_time=0.338, loss_ctc=47.187, loss_att=45.014, acc=0.705, loss=45.666, backward_time=0.404, grad_norm=25.910, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.094, optim0_lr0=3.378e-04, train_time=1.722
+[gpub078:0/16] 2024-01-25 10:22:02,271 (trainer:737) INFO: 9epoch:train:11501-11600batch: iter_time=2.416e-04, forward_time=0.359, loss_ctc=57.435, loss_att=58.194, acc=0.695, loss=57.966, backward_time=0.415, grad_norm=30.598, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.111, optim0_lr0=3.377e-04, train_time=1.503
+[gpub078:0/16] 2024-01-25 10:25:09,172 (trainer:737) INFO: 9epoch:train:11601-11700batch: iter_time=8.059e-05, forward_time=0.287, loss_ctc=52.553, loss_att=48.165, acc=0.716, loss=49.481, backward_time=0.398, grad_norm=27.122, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.093, optim0_lr0=3.375e-04, train_time=1.869
+[gpub078:0/16] 2024-01-25 10:27:33,983 (trainer:737) INFO: 9epoch:train:11701-11800batch: iter_time=8.470e-05, forward_time=0.340, loss_ctc=58.152, loss_att=57.493, acc=0.703, loss=57.690, backward_time=0.419, grad_norm=29.018, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.093, optim0_lr0=3.374e-04, train_time=1.448
+[gpub078:0/16] 2024-01-25 10:30:32,761 (trainer:737) INFO: 9epoch:train:11801-11900batch: iter_time=8.653e-05, forward_time=0.383, loss_ctc=54.853, loss_att=59.214, acc=0.691, loss=57.905, backward_time=0.416, grad_norm=28.930, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.111, optim0_lr0=3.373e-04, train_time=1.788
+[gpub078:0/16] 2024-01-25 10:33:35,101 (trainer:737) INFO: 9epoch:train:11901-12000batch: iter_time=8.581e-05, forward_time=0.320, loss_ctc=59.538, loss_att=57.115, acc=0.687, loss=57.842, backward_time=0.427, grad_norm=29.791, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.093, optim0_lr0=3.372e-04, train_time=1.823
+[gpub078:0/16] 2024-01-25 10:36:26,087 (trainer:737) INFO: 9epoch:train:12001-12100batch: iter_time=8.798e-05, forward_time=0.423, loss_ctc=55.999, loss_att=48.103, acc=0.716, loss=50.472, backward_time=0.423, grad_norm=29.659, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.105, optim0_lr0=3.370e-04, train_time=1.709
+[gpub078:0/16] 2024-01-25 10:39:00,352 (trainer:737) INFO: 9epoch:train:12101-12200batch: iter_time=8.126e-05, forward_time=0.289, loss_ctc=55.876, loss_att=55.290, acc=0.697, loss=55.466, backward_time=0.401, grad_norm=28.971, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.369e-04, train_time=1.543
+[gpub078:0/16] 2024-01-25 10:41:58,556 (trainer:737) INFO: 9epoch:train:12201-12300batch: iter_time=4.306e-04, forward_time=0.416, loss_ctc=56.790, loss_att=58.385, acc=0.673, loss=57.907, backward_time=0.429, grad_norm=29.538, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.116, optim0_lr0=3.368e-04, train_time=1.782
+[gpub078:0/16] 2024-01-25 10:45:01,343 (trainer:737) INFO: 9epoch:train:12301-12400batch: iter_time=8.531e-05, forward_time=0.289, loss_ctc=55.117, loss_att=50.474, acc=0.735, loss=51.866, backward_time=0.402, grad_norm=27.987, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.367e-04, train_time=1.827
+[gpub078:0/16] 2024-01-25 10:47:14,470 (trainer:737) INFO: 9epoch:train:12401-12500batch: iter_time=8.443e-05, forward_time=0.293, loss_ctc=63.604, loss_att=70.626, acc=0.667, loss=68.520, backward_time=0.407, grad_norm=33.475, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.365e-04, train_time=1.331
+[gpub078:0/16] 2024-01-25 10:47:34,497 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub078:0/16] 2024-01-25 10:47:54,331 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 10:47:58,049 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9996ed5480>)
+[gpub078:0/16] 2024-01-25 10:47:58,050 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub078:0/16] 2024-01-25 10:47:58,053 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 10:55:29,655 (trainer:737) INFO: 9epoch:train:12501-12600batch: iter_time=3.283, forward_time=0.390, loss_ctc=54.564, loss_att=52.655, acc=0.701, loss=53.228, backward_time=0.458, grad_norm=30.055, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.103, optim0_lr0=3.364e-04, train_time=4.951
+[gpub078:0/16] 2024-01-25 10:58:20,977 (trainer:737) INFO: 9epoch:train:12601-12700batch: iter_time=8.853e-05, forward_time=0.287, loss_ctc=53.200, loss_att=51.321, acc=0.691, loss=51.885, backward_time=0.399, grad_norm=30.310, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.363e-04, train_time=1.713
+[gpub078:0/16] 2024-01-25 11:00:50,260 (trainer:737) INFO: 9epoch:train:12701-12800batch: iter_time=1.068e-04, forward_time=0.288, loss_ctc=49.358, loss_att=51.372, acc=0.675, loss=50.768, backward_time=0.398, grad_norm=28.307, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.094, optim0_lr0=3.361e-04, train_time=1.493
+[gpub078:0/16] 2024-01-25 11:03:11,521 (trainer:737) INFO: 9epoch:train:12801-12900batch: iter_time=1.118e-04, forward_time=0.358, loss_ctc=56.689, loss_att=58.434, acc=0.695, loss=57.910, backward_time=0.424, grad_norm=29.067, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.097, optim0_lr0=3.360e-04, train_time=1.412
+[gpub078:0/16] 2024-01-25 11:05:53,141 (trainer:737) INFO: 9epoch:train:12901-13000batch: iter_time=1.018e-04, forward_time=0.358, loss_ctc=51.818, loss_att=47.390, acc=0.726, loss=48.719, backward_time=0.419, grad_norm=26.886, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.095, optim0_lr0=3.359e-04, train_time=1.616
+[gpub078:0/16] 2024-01-25 11:08:49,293 (trainer:737) INFO: 9epoch:train:13001-13100batch: iter_time=1.121e-04, forward_time=0.289, loss_ctc=58.388, loss_att=57.017, acc=0.696, loss=57.428, backward_time=0.402, grad_norm=28.188, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.358e-04, train_time=1.761
+[gpub078:0/16] 2024-01-25 11:11:08,823 (trainer:737) INFO: 9epoch:train:13101-13200batch: iter_time=1.107e-04, forward_time=0.358, loss_ctc=53.807, loss_att=53.468, acc=0.683, loss=53.570, backward_time=0.409, grad_norm=29.695, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.356e-04, train_time=1.395
+[gpub078:0/16] 2024-01-25 11:13:51,919 (trainer:737) INFO: 9epoch:train:13201-13300batch: iter_time=9.248e-05, forward_time=0.366, loss_ctc=61.383, loss_att=52.710, acc=0.696, loss=55.312, backward_time=0.432, grad_norm=29.135, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.103, optim0_lr0=3.355e-04, train_time=1.631
+[gpub078:0/16] 2024-01-25 11:16:38,492 (trainer:737) INFO: 9epoch:train:13301-13400batch: iter_time=8.456e-05, forward_time=0.287, loss_ctc=51.004, loss_att=50.948, acc=0.701, loss=50.965, backward_time=0.398, grad_norm=28.868, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.354e-04, train_time=1.665
+[gpub078:0/16] 2024-01-25 11:18:52,708 (trainer:737) INFO: 9epoch:train:13401-13500batch: iter_time=8.326e-05, forward_time=0.290, loss_ctc=57.385, loss_att=56.473, acc=0.689, loss=56.747, backward_time=0.401, grad_norm=29.898, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.093, optim0_lr0=3.353e-04, train_time=1.342
+[gpub078:0/16] 2024-01-25 11:21:34,296 (trainer:737) INFO: 9epoch:train:13501-13600batch: iter_time=4.202e-04, forward_time=0.351, loss_ctc=57.018, loss_att=49.954, acc=0.695, loss=52.073, backward_time=0.416, grad_norm=29.647, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.099, optim0_lr0=3.351e-04, train_time=1.616
+[gpub078:0/16] 2024-01-25 11:24:19,365 (trainer:737) INFO: 9epoch:train:13601-13700batch: iter_time=8.228e-05, forward_time=0.341, loss_ctc=61.638, loss_att=63.083, acc=0.687, loss=62.650, backward_time=0.422, grad_norm=32.480, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.097, optim0_lr0=3.350e-04, train_time=1.651
+[gpub078:0/16] 2024-01-25 11:26:01,978 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub078:0/16] 2024-01-25 11:26:21,396 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 11:26:24,956 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99009b1630>)
+[gpub078:0/16] 2024-01-25 11:26:24,956 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub078:0/16] 2024-01-25 11:26:24,960 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 11:32:50,469 (trainer:737) INFO: 9epoch:train:13701-13800batch: iter_time=3.436, forward_time=0.379, loss_ctc=51.148, loss_att=56.724, acc=0.683, loss=55.052, backward_time=0.411, grad_norm=26.890, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.096, optim0_lr0=3.349e-04, train_time=5.110
+[gpub078:0/16] 2024-01-25 11:35:16,380 (trainer:737) INFO: 9epoch:train:13801-13900batch: iter_time=8.232e-05, forward_time=0.349, loss_ctc=59.063, loss_att=56.510, acc=0.688, loss=57.276, backward_time=0.443, grad_norm=33.352, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.113, optim0_lr0=3.348e-04, train_time=1.459
+[gpub078:0/16] 2024-01-25 11:38:34,495 (trainer:737) INFO: 9epoch:train:13901-14000batch: iter_time=8.523e-05, forward_time=0.338, loss_ctc=47.114, loss_att=45.355, acc=0.700, loss=45.883, backward_time=0.466, grad_norm=25.710, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.096, optim0_lr0=3.346e-04, train_time=1.981
+[gpub078:0/16] 2024-01-25 11:41:09,467 (trainer:737) INFO: 9epoch:train:14001-14100batch: iter_time=9.229e-05, forward_time=0.290, loss_ctc=57.197, loss_att=56.727, acc=0.691, loss=56.868, backward_time=0.403, grad_norm=30.139, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.345e-04, train_time=1.549
+[gpub078:0/16] 2024-01-25 11:43:59,061 (trainer:737) INFO: 9epoch:train:14101-14200batch: iter_time=4.948e-04, forward_time=0.380, loss_ctc=51.924, loss_att=46.092, acc=0.723, loss=47.841, backward_time=0.475, grad_norm=28.079, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.103, optim0_lr0=3.344e-04, train_time=1.696
+[gpub078:0/16] 2024-01-25 11:46:34,395 (trainer:737) INFO: 9epoch:train:14201-14300batch: iter_time=9.066e-05, forward_time=0.290, loss_ctc=57.630, loss_att=55.143, acc=0.705, loss=55.889, backward_time=0.400, grad_norm=28.793, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.343e-04, train_time=1.553
+[gpub078:0/16] 2024-01-25 11:49:07,864 (trainer:737) INFO: 9epoch:train:14301-14400batch: iter_time=9.952e-05, forward_time=0.307, loss_ctc=54.512, loss_att=57.459, acc=0.685, loss=56.575, backward_time=0.401, grad_norm=30.301, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.341e-04, train_time=1.534
+[gpub078:0/16] 2024-01-25 11:52:04,291 (trainer:737) INFO: 9epoch:train:14401-14500batch: iter_time=9.332e-05, forward_time=0.423, loss_ctc=59.007, loss_att=55.309, acc=0.675, loss=56.418, backward_time=0.428, grad_norm=32.139, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.110, optim0_lr0=3.340e-04, train_time=1.764
+[gpub078:0/16] 2024-01-25 11:55:00,955 (trainer:737) INFO: 9epoch:train:14501-14600batch: iter_time=1.013e-04, forward_time=0.290, loss_ctc=55.826, loss_att=45.322, acc=0.717, loss=48.474, backward_time=0.399, grad_norm=27.729, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.339e-04, train_time=1.767
+[gpub078:0/16] 2024-01-25 11:57:20,738 (trainer:737) INFO: 9epoch:train:14601-14700batch: iter_time=9.664e-05, forward_time=0.289, loss_ctc=55.780, loss_att=54.998, acc=0.693, loss=55.232, backward_time=0.402, grad_norm=29.989, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.094, optim0_lr0=3.338e-04, train_time=1.398
+[gpub078:0/16] 2024-01-25 12:00:29,635 (trainer:737) INFO: 9epoch:train:14701-14800batch: iter_time=4.398e-04, forward_time=0.356, loss_ctc=55.934, loss_att=55.828, acc=0.673, loss=55.860, backward_time=0.515, grad_norm=75.187, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.104, optim0_lr0=3.336e-04, train_time=1.889
+[gpub078:0/16] 2024-01-25 12:03:27,691 (trainer:737) INFO: 9epoch:train:14801-14900batch: iter_time=9.342e-05, forward_time=0.289, loss_ctc=54.574, loss_att=52.167, acc=0.720, loss=52.889, backward_time=0.402, grad_norm=27.618, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.335e-04, train_time=1.780
+[gpub078:0/16] 2024-01-25 12:05:40,695 (trainer:737) INFO: 9epoch:train:14901-15000batch: iter_time=8.960e-05, forward_time=0.291, loss_ctc=62.961, loss_att=67.609, acc=0.662, loss=66.215, backward_time=0.406, grad_norm=33.629, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.334e-04, train_time=1.330
+[gpub078:0/16] 2024-01-25 12:44:00,505 (trainer:343) INFO: 9epoch results: [train] iter_time=0.284, forward_time=0.325, loss_ctc=57.321, loss_att=55.734, acc=0.690, loss=56.210, backward_time=0.416, grad_norm=30.458, clip=100.000, loss_scale=1.924e+24, optim_step_time=0.097, optim0_lr0=3.431e-04, train_time=1.874, time=7 hours, 49 minutes and 4.66 seconds, total_count=165000, gpu_max_cached_mem_GB=40.762, [valid] loss_ctc=52.523, cer_ctc=0.260, loss_att=49.861, acc=0.606, cer=0.363, wer=0.998, loss=50.659, time=37 minutes and 53.12 seconds, total_count=51381, gpu_max_cached_mem_GB=40.762
+[gpub078:0/16] 2024-01-25 12:44:11,030 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub078:0/16] 2024-01-25 12:44:11,085 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/4epoch.pth
+[gpub078:0/16] 2024-01-25 12:44:11,085 (trainer:272) INFO: 10/45epoch started. Estimated time to finish: 2 weeks, 4 hours and 33 minutes
+[gpub078:0/16] 2024-01-25 12:44:11,095 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-25 12:44:29,038 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 12:44:32,480 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f924ad0f0a0>)
+[gpub078:0/16] 2024-01-25 12:44:32,480 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub078:0/16] 2024-01-25 12:44:32,483 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 12:52:27,300 (trainer:737) INFO: 10epoch:train:1-100batch: iter_time=3.326, forward_time=0.347, loss_ctc=61.164, loss_att=60.284, acc=0.701, loss=60.548, backward_time=0.426, grad_norm=33.879, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.096, optim0_lr0=3.333e-04, train_time=4.962
+[gpub078:0/16] 2024-01-25 12:54:46,473 (trainer:737) INFO: 10epoch:train:101-200batch: iter_time=9.899e-05, forward_time=0.303, loss_ctc=60.816, loss_att=57.712, acc=0.711, loss=58.643, backward_time=0.408, grad_norm=30.739, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.095, optim0_lr0=3.331e-04, train_time=1.392
+[gpub078:0/16] 2024-01-25 12:57:11,818 (trainer:737) INFO: 10epoch:train:201-300batch: iter_time=9.762e-05, forward_time=0.342, loss_ctc=46.049, loss_att=43.950, acc=0.705, loss=44.580, backward_time=0.424, grad_norm=24.308, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.096, optim0_lr0=3.330e-04, train_time=1.453
+[gpub078:0/16] 2024-01-25 13:00:13,974 (trainer:737) INFO: 10epoch:train:301-400batch: iter_time=1.807e-04, forward_time=0.325, loss_ctc=65.874, loss_att=69.942, acc=0.689, loss=68.721, backward_time=0.463, grad_norm=31.780, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.101, optim0_lr0=3.329e-04, train_time=1.822
+[gpub078:0/16] 2024-01-25 13:02:53,561 (trainer:737) INFO: 10epoch:train:401-500batch: iter_time=1.076e-04, forward_time=0.303, loss_ctc=53.708, loss_att=56.163, acc=0.697, loss=55.427, backward_time=0.401, grad_norm=29.509, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.328e-04, train_time=1.596
+[gpub078:0/16] 2024-01-25 13:05:44,460 (trainer:737) INFO: 10epoch:train:501-600batch: iter_time=9.996e-05, forward_time=0.373, loss_ctc=58.052, loss_att=55.332, acc=0.726, loss=56.148, backward_time=0.419, grad_norm=27.823, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.097, optim0_lr0=3.327e-04, train_time=1.709
+[gpub078:0/16] 2024-01-25 13:08:46,222 (trainer:737) INFO: 10epoch:train:601-700batch: iter_time=9.820e-05, forward_time=0.364, loss_ctc=60.374, loss_att=56.752, acc=0.699, loss=57.838, backward_time=0.455, grad_norm=30.758, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.111, optim0_lr0=3.325e-04, train_time=1.816
+[gpub078:0/16] 2024-01-25 13:11:27,669 (trainer:737) INFO: 10epoch:train:701-800batch: iter_time=9.730e-05, forward_time=0.307, loss_ctc=53.848, loss_att=56.729, acc=0.692, loss=55.864, backward_time=0.417, grad_norm=28.549, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.093, optim0_lr0=3.324e-04, train_time=1.616
+[gpub078:0/16] 2024-01-25 13:14:09,543 (trainer:737) INFO: 10epoch:train:801-900batch: iter_time=1.025e-04, forward_time=0.298, loss_ctc=71.660, loss_att=64.594, acc=0.689, loss=66.714, backward_time=0.403, grad_norm=39.998, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.094, optim0_lr0=3.323e-04, train_time=1.619
+[gpub078:0/16] 2024-01-25 13:17:17,690 (trainer:737) INFO: 10epoch:train:901-1000batch: iter_time=1.049e-04, forward_time=0.378, loss_ctc=72.047, loss_att=71.813, acc=0.682, loss=71.884, backward_time=0.441, grad_norm=37.312, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.102, optim0_lr0=3.322e-04, train_time=1.881
+[gpub078:0/16] 2024-01-25 13:19:44,181 (trainer:737) INFO: 10epoch:train:1001-1100batch: iter_time=9.584e-05, forward_time=0.314, loss_ctc=53.015, loss_att=42.278, acc=0.733, loss=45.499, backward_time=0.420, grad_norm=24.804, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.098, optim0_lr0=3.320e-04, train_time=1.463
+[gpub078:0/16] 2024-01-25 13:22:11,413 (trainer:737) INFO: 10epoch:train:1101-1200batch: iter_time=8.353e-05, forward_time=0.302, loss_ctc=64.718, loss_att=68.031, acc=0.683, loss=67.037, backward_time=0.403, grad_norm=33.193, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.319e-04, train_time=1.473
+[gpub078:0/16] 2024-01-25 13:23:59,653 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub078:0/16] 2024-01-25 13:24:18,873 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 13:24:22,362 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9221526350>)
+[gpub078:0/16] 2024-01-25 13:24:22,362 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub078:0/16] 2024-01-25 13:24:22,365 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 13:31:00,387 (trainer:737) INFO: 10epoch:train:1201-1300batch: iter_time=3.556, forward_time=0.386, loss_ctc=57.726, loss_att=53.951, acc=0.715, loss=55.084, backward_time=0.407, grad_norm=29.230, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.095, optim0_lr0=3.318e-04, train_time=5.290
+[gpub078:0/16] 2024-01-25 13:33:13,835 (trainer:737) INFO: 10epoch:train:1301-1400batch: iter_time=7.824e-05, forward_time=0.290, loss_ctc=58.052, loss_att=55.153, acc=0.690, loss=56.023, backward_time=0.402, grad_norm=31.450, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.317e-04, train_time=1.334
+[gpub078:0/16] 2024-01-25 13:35:40,626 (trainer:737) INFO: 10epoch:train:1401-1500batch: iter_time=8.151e-05, forward_time=0.311, loss_ctc=51.892, loss_att=50.922, acc=0.705, loss=51.213, backward_time=0.440, grad_norm=28.033, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.098, optim0_lr0=3.316e-04, train_time=1.467
+[gpub078:0/16] 2024-01-25 13:38:38,421 (trainer:737) INFO: 10epoch:train:1501-1600batch: iter_time=9.115e-05, forward_time=0.298, loss_ctc=50.366, loss_att=51.516, acc=0.690, loss=51.171, backward_time=0.400, grad_norm=26.154, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.094, optim0_lr0=3.314e-04, train_time=1.778
+[gpub078:0/16] 2024-01-25 13:41:37,348 (trainer:737) INFO: 10epoch:train:1601-1700batch: iter_time=9.326e-05, forward_time=0.402, loss_ctc=58.370, loss_att=63.438, acc=0.691, loss=61.918, backward_time=0.424, grad_norm=30.360, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.098, optim0_lr0=3.313e-04, train_time=1.789
+[gpub078:0/16] 2024-01-25 13:44:16,704 (trainer:737) INFO: 10epoch:train:1701-1800batch: iter_time=9.065e-05, forward_time=0.337, loss_ctc=59.927, loss_att=59.963, acc=0.697, loss=59.952, backward_time=0.423, grad_norm=30.719, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.106, optim0_lr0=3.312e-04, train_time=1.593
+[gpub078:0/16] 2024-01-25 13:47:16,069 (trainer:737) INFO: 10epoch:train:1801-1900batch: iter_time=9.025e-05, forward_time=0.289, loss_ctc=58.170, loss_att=51.648, acc=0.701, loss=53.605, backward_time=0.405, grad_norm=30.432, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.311e-04, train_time=1.794
+[gpub078:0/16] 2024-01-25 13:49:30,103 (trainer:737) INFO: 10epoch:train:1901-2000batch: iter_time=8.646e-05, forward_time=0.302, loss_ctc=51.604, loss_att=50.309, acc=0.710, loss=50.698, backward_time=0.401, grad_norm=27.550, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.094, optim0_lr0=3.309e-04, train_time=1.341
+[gpub078:0/16] 2024-01-25 13:52:27,170 (trainer:737) INFO: 10epoch:train:2001-2100batch: iter_time=9.998e-05, forward_time=0.368, loss_ctc=60.219, loss_att=61.824, acc=0.679, loss=61.342, backward_time=0.410, grad_norm=30.773, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.100, optim0_lr0=3.308e-04, train_time=1.770
+[gpub078:0/16] 2024-01-25 13:55:04,386 (trainer:737) INFO: 10epoch:train:2101-2200batch: iter_time=9.362e-05, forward_time=0.326, loss_ctc=73.195, loss_att=63.205, acc=0.688, loss=66.202, backward_time=0.432, grad_norm=40.934, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.097, optim0_lr0=3.307e-04, train_time=1.572
+[gpub078:0/16] 2024-01-25 13:57:36,631 (trainer:737) INFO: 10epoch:train:2201-2300batch: iter_time=9.318e-05, forward_time=0.289, loss_ctc=58.035, loss_att=59.891, acc=0.699, loss=59.334, backward_time=0.401, grad_norm=28.538, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.306e-04, train_time=1.521
+[gpub078:0/16] 2024-01-25 13:59:56,316 (trainer:737) INFO: 10epoch:train:2301-2400batch: iter_time=8.984e-05, forward_time=0.297, loss_ctc=59.864, loss_att=52.565, acc=0.702, loss=54.754, backward_time=0.401, grad_norm=30.831, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.305e-04, train_time=1.398
+[gpub078:0/16] 2024-01-25 14:02:31,532 (trainer:737) INFO: 10epoch:train:2401-2500batch: iter_time=9.576e-05, forward_time=0.333, loss_ctc=54.866, loss_att=54.102, acc=0.699, loss=54.331, backward_time=0.447, grad_norm=29.434, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.099, optim0_lr0=3.303e-04, train_time=1.552
+[gpub078:0/16] 2024-01-25 14:02:51,565 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub078:0/16] 2024-01-25 14:03:10,474 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 14:03:14,104 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f92495637c0>)
+[gpub078:0/16] 2024-01-25 14:03:14,104 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub078:0/16] 2024-01-25 14:03:14,107 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 14:11:21,755 (trainer:737) INFO: 10epoch:train:2501-2600batch: iter_time=3.802, forward_time=0.326, loss_ctc=59.893, loss_att=57.260, acc=0.698, loss=58.050, backward_time=0.416, grad_norm=29.737, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.095, optim0_lr0=3.302e-04, train_time=5.302
+[gpub078:0/16] 2024-01-25 14:13:39,872 (trainer:737) INFO: 10epoch:train:2601-2700batch: iter_time=7.692e-05, forward_time=0.291, loss_ctc=57.962, loss_att=55.679, acc=0.706, loss=56.364, backward_time=0.403, grad_norm=31.098, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.301e-04, train_time=1.381
+[gpub078:0/16] 2024-01-25 14:16:15,004 (trainer:737) INFO: 10epoch:train:2701-2800batch: iter_time=7.921e-05, forward_time=0.297, loss_ctc=44.274, loss_att=42.423, acc=0.702, loss=42.978, backward_time=0.396, grad_norm=23.932, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.094, optim0_lr0=3.300e-04, train_time=1.552
+[gpub078:0/16] 2024-01-25 14:19:22,280 (trainer:737) INFO: 10epoch:train:2801-2900batch: iter_time=8.719e-05, forward_time=0.385, loss_ctc=63.750, loss_att=67.663, acc=0.684, loss=66.490, backward_time=0.453, grad_norm=31.926, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.101, optim0_lr0=3.299e-04, train_time=1.872
+[gpub078:0/16] 2024-01-25 14:21:50,661 (trainer:737) INFO: 10epoch:train:2901-3000batch: iter_time=8.508e-05, forward_time=0.288, loss_ctc=52.986, loss_att=52.874, acc=0.698, loss=52.908, backward_time=0.401, grad_norm=31.178, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.093, optim0_lr0=3.297e-04, train_time=1.484
+[gpub078:0/16] 2024-01-25 14:24:08,549 (trainer:737) INFO: 10epoch:train:3001-3100batch: iter_time=8.173e-05, forward_time=0.297, loss_ctc=55.249, loss_att=52.877, acc=0.725, loss=53.589, backward_time=0.404, grad_norm=27.283, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.296e-04, train_time=1.378
+[gpub078:0/16] 2024-01-25 14:26:49,807 (trainer:737) INFO: 10epoch:train:3101-3200batch: iter_time=9.990e-05, forward_time=0.349, loss_ctc=58.435, loss_att=54.958, acc=0.698, loss=56.001, backward_time=0.408, grad_norm=30.684, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.096, optim0_lr0=3.295e-04, train_time=1.613
+[gpub078:0/16] 2024-01-25 14:29:43,299 (trainer:737) INFO: 10epoch:train:3201-3300batch: iter_time=1.905e-04, forward_time=0.327, loss_ctc=52.467, loss_att=54.359, acc=0.693, loss=53.791, backward_time=0.442, grad_norm=26.966, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.096, optim0_lr0=3.294e-04, train_time=1.734
+[gpub078:0/16] 2024-01-25 14:32:12,501 (trainer:737) INFO: 10epoch:train:3301-3400batch: iter_time=8.749e-05, forward_time=0.290, loss_ctc=65.804, loss_att=62.351, acc=0.685, loss=63.387, backward_time=0.404, grad_norm=36.947, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.293e-04, train_time=1.492
+[gpub078:0/16] 2024-01-25 14:34:27,332 (trainer:737) INFO: 10epoch:train:3401-3500batch: iter_time=1.139e-04, forward_time=0.299, loss_ctc=68.702, loss_att=69.961, acc=0.673, loss=69.583, backward_time=0.405, grad_norm=37.000, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.292e-04, train_time=1.348
+[gpub078:0/16] 2024-01-25 14:37:10,432 (trainer:737) INFO: 10epoch:train:3501-3600batch: iter_time=1.007e-04, forward_time=0.382, loss_ctc=52.384, loss_att=41.427, acc=0.735, loss=44.714, backward_time=0.415, grad_norm=25.069, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.098, optim0_lr0=3.290e-04, train_time=1.631
+[gpub078:0/16] 2024-01-25 14:40:10,712 (trainer:737) INFO: 10epoch:train:3601-3700batch: iter_time=8.486e-05, forward_time=0.326, loss_ctc=61.927, loss_att=64.510, acc=0.686, loss=63.735, backward_time=0.424, grad_norm=34.030, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.098, optim0_lr0=3.289e-04, train_time=1.803
+[gpub078:0/16] 2024-01-25 14:41:36,722 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub078:0/16] 2024-01-25 14:41:55,728 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 14:41:59,266 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b1a2ed8d0>)
+[gpub078:0/16] 2024-01-25 14:41:59,266 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub078:0/16] 2024-01-25 14:41:59,269 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 14:48:23,395 (trainer:737) INFO: 10epoch:train:3701-3800batch: iter_time=3.523, forward_time=0.293, loss_ctc=56.582, loss_att=54.478, acc=0.717, loss=55.109, backward_time=0.402, grad_norm=27.827, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.288e-04, train_time=4.926
+[gpub078:0/16] 2024-01-25 14:51:01,235 (trainer:737) INFO: 10epoch:train:3801-3900batch: iter_time=8.743e-05, forward_time=0.287, loss_ctc=57.131, loss_att=54.384, acc=0.707, loss=55.208, backward_time=0.400, grad_norm=29.491, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.287e-04, train_time=1.578
+[gpub078:0/16] 2024-01-25 14:53:41,254 (trainer:737) INFO: 10epoch:train:3901-4000batch: iter_time=6.340e-04, forward_time=0.420, loss_ctc=50.925, loss_att=50.353, acc=0.717, loss=50.525, backward_time=0.447, grad_norm=27.148, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.099, optim0_lr0=3.286e-04, train_time=1.601
+[gpub078:0/16] 2024-01-25 14:56:05,821 (trainer:737) INFO: 10epoch:train:4001-4100batch: iter_time=8.734e-05, forward_time=0.289, loss_ctc=49.802, loss_att=51.380, acc=0.706, loss=50.906, backward_time=0.402, grad_norm=26.233, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.284e-04, train_time=1.445
+[gpub078:0/16] 2024-01-25 14:58:54,985 (trainer:737) INFO: 10epoch:train:4101-4200batch: iter_time=8.404e-05, forward_time=0.296, loss_ctc=56.742, loss_att=64.792, acc=0.700, loss=62.377, backward_time=0.403, grad_norm=31.620, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.093, optim0_lr0=3.283e-04, train_time=1.691
+[gpub078:0/16] 2024-01-25 15:01:40,139 (trainer:737) INFO: 10epoch:train:4201-4300batch: iter_time=8.835e-05, forward_time=0.404, loss_ctc=58.461, loss_att=59.609, acc=0.708, loss=59.264, backward_time=0.417, grad_norm=30.181, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.097, optim0_lr0=3.282e-04, train_time=1.652
+[gpub078:0/16] 2024-01-25 15:04:17,817 (trainer:737) INFO: 10epoch:train:4301-4400batch: iter_time=8.586e-05, forward_time=0.375, loss_ctc=57.453, loss_att=51.547, acc=0.714, loss=53.319, backward_time=0.441, grad_norm=29.176, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.095, optim0_lr0=3.281e-04, train_time=1.577
+[gpub078:0/16] 2024-01-25 15:07:15,881 (trainer:737) INFO: 10epoch:train:4401-4500batch: iter_time=9.994e-05, forward_time=0.295, loss_ctc=51.351, loss_att=51.542, acc=0.714, loss=51.484, backward_time=0.400, grad_norm=27.321, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.094, optim0_lr0=3.280e-04, train_time=1.780
+[gpub078:0/16] 2024-01-25 15:09:44,934 (trainer:737) INFO: 10epoch:train:4501-4600batch: iter_time=9.178e-05, forward_time=0.366, loss_ctc=59.152, loss_att=61.081, acc=0.695, loss=60.502, backward_time=0.450, grad_norm=31.683, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.099, optim0_lr0=3.279e-04, train_time=1.490
+[gpub078:0/16] 2024-01-25 15:12:03,114 (trainer:737) INFO: 10epoch:train:4601-4700batch: iter_time=9.028e-05, forward_time=0.310, loss_ctc=69.131, loss_att=62.155, acc=0.701, loss=64.248, backward_time=0.429, grad_norm=40.100, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.096, optim0_lr0=3.277e-04, train_time=1.382
+[gpub078:0/16] 2024-01-25 15:15:19,017 (trainer:737) INFO: 10epoch:train:4701-4800batch: iter_time=9.465e-05, forward_time=0.303, loss_ctc=57.065, loss_att=60.333, acc=0.710, loss=59.352, backward_time=0.412, grad_norm=28.948, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.094, optim0_lr0=3.276e-04, train_time=1.959
+[gpub078:0/16] 2024-01-25 15:17:52,339 (trainer:737) INFO: 10epoch:train:4801-4900batch: iter_time=8.924e-05, forward_time=0.397, loss_ctc=59.768, loss_att=54.309, acc=0.707, loss=55.947, backward_time=0.429, grad_norm=32.311, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.102, optim0_lr0=3.275e-04, train_time=1.533
+[gpub078:0/16] 2024-01-25 15:20:37,474 (trainer:737) INFO: 10epoch:train:4901-5000batch: iter_time=9.122e-05, forward_time=0.325, loss_ctc=54.179, loss_att=53.790, acc=0.710, loss=53.907, backward_time=0.405, grad_norm=30.023, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.097, optim0_lr0=3.274e-04, train_time=1.651
+[gpub078:0/16] 2024-01-25 15:20:57,644 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub078:0/16] 2024-01-25 15:21:16,469 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 15:21:20,229 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c4ca8f5b0>)
+[gpub078:0/16] 2024-01-25 15:21:20,230 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub078:0/16] 2024-01-25 15:21:20,235 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 15:29:12,537 (trainer:737) INFO: 10epoch:train:5001-5100batch: iter_time=3.494, forward_time=0.337, loss_ctc=58.630, loss_att=58.305, acc=0.699, loss=58.403, backward_time=0.428, grad_norm=29.512, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.094, optim0_lr0=3.273e-04, train_time=5.151
+[gpub078:0/16] 2024-01-25 15:32:00,511 (trainer:737) INFO: 10epoch:train:5101-5200batch: iter_time=9.257e-05, forward_time=0.289, loss_ctc=57.155, loss_att=55.631, acc=0.708, loss=56.088, backward_time=0.404, grad_norm=31.581, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.271e-04, train_time=1.680
+[gpub078:0/16] 2024-01-25 15:34:45,572 (trainer:737) INFO: 10epoch:train:5201-5300batch: iter_time=1.024e-04, forward_time=0.356, loss_ctc=43.598, loss_att=42.672, acc=0.702, loss=42.950, backward_time=0.433, grad_norm=24.737, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.106, optim0_lr0=3.270e-04, train_time=1.650
+[gpub078:0/16] 2024-01-25 15:37:28,001 (trainer:737) INFO: 10epoch:train:5301-5400batch: iter_time=1.002e-04, forward_time=0.324, loss_ctc=62.766, loss_att=67.227, acc=0.688, loss=65.889, backward_time=0.432, grad_norm=30.619, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.095, optim0_lr0=3.269e-04, train_time=1.624
+[gpub078:0/16] 2024-01-25 15:39:48,077 (trainer:737) INFO: 10epoch:train:5401-5500batch: iter_time=9.045e-05, forward_time=0.292, loss_ctc=52.315, loss_att=51.894, acc=0.704, loss=52.020, backward_time=0.401, grad_norm=27.525, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.268e-04, train_time=1.401
+[gpub078:0/16] 2024-01-25 15:43:08,424 (trainer:737) INFO: 10epoch:train:5501-5600batch: iter_time=8.302e-05, forward_time=0.368, loss_ctc=54.987, loss_att=52.710, acc=0.729, loss=53.393, backward_time=0.412, grad_norm=27.354, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.094, optim0_lr0=3.267e-04, train_time=2.003
+[gpub078:0/16] 2024-01-25 15:45:21,686 (trainer:737) INFO: 10epoch:train:5601-5700batch: iter_time=9.173e-05, forward_time=0.290, loss_ctc=57.089, loss_att=53.887, acc=0.705, loss=54.848, backward_time=0.402, grad_norm=29.115, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.266e-04, train_time=1.332
+[gpub078:0/16] 2024-01-25 15:48:14,333 (trainer:737) INFO: 10epoch:train:5701-5800batch: iter_time=9.057e-05, forward_time=0.328, loss_ctc=52.520, loss_att=54.565, acc=0.693, loss=53.951, backward_time=0.420, grad_norm=28.059, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.099, optim0_lr0=3.265e-04, train_time=1.726
+[gpub078:0/16] 2024-01-25 15:50:34,756 (trainer:737) INFO: 10epoch:train:5801-5900batch: iter_time=8.857e-05, forward_time=0.313, loss_ctc=65.293, loss_att=62.814, acc=0.683, loss=63.558, backward_time=0.401, grad_norm=35.714, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.263e-04, train_time=1.404
+[gpub078:0/16] 2024-01-25 15:53:18,986 (trainer:737) INFO: 10epoch:train:5901-6000batch: iter_time=9.542e-05, forward_time=0.345, loss_ctc=66.342, loss_att=69.279, acc=0.676, loss=68.398, backward_time=0.486, grad_norm=36.251, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.100, optim0_lr0=3.262e-04, train_time=1.642
+[gpub078:0/16] 2024-01-25 15:56:16,240 (trainer:737) INFO: 10epoch:train:6001-6100batch: iter_time=9.344e-05, forward_time=0.289, loss_ctc=51.954, loss_att=40.915, acc=0.738, loss=44.227, backward_time=0.399, grad_norm=26.116, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.261e-04, train_time=1.772
+[gpub078:0/16] 2024-01-25 15:58:46,515 (trainer:737) INFO: 10epoch:train:6101-6200batch: iter_time=0.002, forward_time=0.323, loss_ctc=61.354, loss_att=63.455, acc=0.691, loss=62.825, backward_time=0.425, grad_norm=31.971, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.097, optim0_lr0=3.260e-04, train_time=1.501
+[gpub078:0/16] 2024-01-25 16:00:15,844 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub078:0/16] 2024-01-25 16:00:35,122 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 16:00:38,801 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99b2e25090>)
+[gpub078:0/16] 2024-01-25 16:00:38,802 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub078:0/16] 2024-01-25 16:00:38,805 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 16:07:45,157 (trainer:737) INFO: 10epoch:train:6201-6300batch: iter_time=3.827, forward_time=0.382, loss_ctc=56.655, loss_att=53.957, acc=0.718, loss=54.767, backward_time=0.413, grad_norm=28.984, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.095, optim0_lr0=3.259e-04, train_time=5.387
+[gpub078:0/16] 2024-01-25 16:10:27,638 (trainer:737) INFO: 10epoch:train:6301-6400batch: iter_time=9.327e-05, forward_time=0.288, loss_ctc=55.513, loss_att=54.091, acc=0.707, loss=54.518, backward_time=0.400, grad_norm=30.372, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.258e-04, train_time=1.625
+[gpub078:0/16] 2024-01-25 16:13:15,582 (trainer:737) INFO: 10epoch:train:6401-6500batch: iter_time=8.424e-05, forward_time=0.350, loss_ctc=49.819, loss_att=49.603, acc=0.719, loss=49.668, backward_time=0.423, grad_norm=26.540, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.095, optim0_lr0=3.256e-04, train_time=1.679
+[gpub078:0/16] 2024-01-25 16:16:13,840 (trainer:737) INFO: 10epoch:train:6501-6600batch: iter_time=8.405e-05, forward_time=0.291, loss_ctc=49.329, loss_att=50.394, acc=0.709, loss=50.074, backward_time=0.402, grad_norm=26.093, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.255e-04, train_time=1.783
+[gpub078:0/16] 2024-01-25 16:19:04,279 (trainer:737) INFO: 10epoch:train:6601-6700batch: iter_time=2.218e-04, forward_time=0.398, loss_ctc=56.519, loss_att=63.799, acc=0.702, loss=61.615, backward_time=0.423, grad_norm=31.524, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.102, optim0_lr0=3.254e-04, train_time=1.704
+[gpub078:0/16] 2024-01-25 16:21:42,767 (trainer:737) INFO: 10epoch:train:6701-6800batch: iter_time=9.247e-05, forward_time=0.320, loss_ctc=57.644, loss_att=58.663, acc=0.711, loss=58.357, backward_time=0.405, grad_norm=28.913, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.093, optim0_lr0=3.253e-04, train_time=1.585
+[gpub078:0/16] 2024-01-25 16:25:24,582 (trainer:737) INFO: 10epoch:train:6801-6900batch: iter_time=1.097e-04, forward_time=0.325, loss_ctc=56.287, loss_att=51.735, acc=0.716, loss=53.100, backward_time=0.430, grad_norm=28.213, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.096, optim0_lr0=3.252e-04, train_time=2.218
+[gpub078:0/16] 2024-01-25 16:28:06,068 (trainer:737) INFO: 10epoch:train:6901-7000batch: iter_time=9.316e-05, forward_time=0.409, loss_ctc=50.260, loss_att=51.194, acc=0.713, loss=50.914, backward_time=0.412, grad_norm=26.742, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.104, optim0_lr0=3.251e-04, train_time=1.615
+[gpub078:0/16] 2024-01-25 16:31:08,407 (trainer:737) INFO: 10epoch:train:7001-7100batch: iter_time=9.668e-05, forward_time=0.290, loss_ctc=57.949, loss_att=60.359, acc=0.698, loss=59.636, backward_time=0.402, grad_norm=29.442, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.250e-04, train_time=1.823
+[gpub078:0/16] 2024-01-25 16:33:38,121 (trainer:737) INFO: 10epoch:train:7101-7200batch: iter_time=9.352e-05, forward_time=0.334, loss_ctc=68.503, loss_att=61.789, acc=0.705, loss=63.803, backward_time=0.418, grad_norm=37.747, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.097, optim0_lr0=3.248e-04, train_time=1.496
+[gpub078:0/16] 2024-01-25 16:36:38,183 (trainer:737) INFO: 10epoch:train:7201-7300batch: iter_time=9.451e-05, forward_time=0.347, loss_ctc=54.901, loss_att=60.080, acc=0.714, loss=58.527, backward_time=0.437, grad_norm=28.830, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.104, optim0_lr0=3.247e-04, train_time=1.801
+[gpub078:0/16] 2024-01-25 16:39:17,630 (trainer:737) INFO: 10epoch:train:7301-7400batch: iter_time=8.955e-05, forward_time=0.288, loss_ctc=57.848, loss_att=53.294, acc=0.711, loss=54.660, backward_time=0.400, grad_norm=29.874, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.246e-04, train_time=1.594
+[gpub078:0/16] 2024-01-25 16:41:49,708 (trainer:737) INFO: 10epoch:train:7401-7500batch: iter_time=9.283e-05, forward_time=0.288, loss_ctc=52.940, loss_att=52.943, acc=0.714, loss=52.942, backward_time=0.402, grad_norm=27.898, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.245e-04, train_time=1.521
+[gpub078:0/16] 2024-01-25 16:42:09,737 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub078:0/16] 2024-01-25 16:42:28,691 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 16:42:32,150 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99ccedb160>)
+[gpub078:0/16] 2024-01-25 16:42:32,150 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub078:0/16] 2024-01-25 16:42:32,153 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 16:50:11,156 (trainer:737) INFO: 10epoch:train:7501-7600batch: iter_time=3.248, forward_time=0.428, loss_ctc=58.343, loss_att=57.681, acc=0.701, loss=57.879, backward_time=0.428, grad_norm=28.598, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.098, optim0_lr0=3.244e-04, train_time=5.013
+[gpub078:0/16] 2024-01-25 16:52:45,916 (trainer:737) INFO: 10epoch:train:7601-7700batch: iter_time=8.409e-05, forward_time=0.293, loss_ctc=56.441, loss_att=55.579, acc=0.711, loss=55.837, backward_time=0.402, grad_norm=30.569, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.243e-04, train_time=1.548
+[gpub078:0/16] 2024-01-25 16:55:02,508 (trainer:737) INFO: 10epoch:train:7701-7800batch: iter_time=8.102e-05, forward_time=0.287, loss_ctc=43.519, loss_att=42.565, acc=0.702, loss=42.851, backward_time=0.400, grad_norm=24.531, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.242e-04, train_time=1.366
+[gpub078:0/16] 2024-01-25 16:58:28,520 (trainer:737) INFO: 10epoch:train:7801-7900batch: iter_time=8.736e-05, forward_time=0.383, loss_ctc=61.511, loss_att=65.834, acc=0.691, loss=64.537, backward_time=0.453, grad_norm=30.074, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.103, optim0_lr0=3.240e-04, train_time=2.060
+[gpub078:0/16] 2024-01-25 17:00:43,522 (trainer:737) INFO: 10epoch:train:7901-8000batch: iter_time=9.009e-05, forward_time=0.288, loss_ctc=51.652, loss_att=51.292, acc=0.709, loss=51.400, backward_time=0.400, grad_norm=27.482, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.239e-04, train_time=1.349
+[gpub078:0/16] 2024-01-25 17:03:18,835 (trainer:737) INFO: 10epoch:train:8001-8100batch: iter_time=8.238e-05, forward_time=0.291, loss_ctc=54.452, loss_att=52.067, acc=0.731, loss=52.782, backward_time=0.404, grad_norm=27.877, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.238e-04, train_time=1.553
+[gpub078:0/16] 2024-01-25 17:05:53,357 (trainer:737) INFO: 10epoch:train:8101-8200batch: iter_time=9.360e-05, forward_time=0.289, loss_ctc=56.752, loss_att=53.267, acc=0.707, loss=54.313, backward_time=0.401, grad_norm=30.539, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.237e-04, train_time=1.545
+[gpub078:0/16] 2024-01-25 17:09:57,153 (trainer:737) INFO: 10epoch:train:8201-8300batch: iter_time=0.022, forward_time=0.397, loss_ctc=51.272, loss_att=53.642, acc=0.696, loss=52.931, backward_time=0.448, grad_norm=28.062, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.106, optim0_lr0=3.236e-04, train_time=2.438
+[gpub078:0/16] 2024-01-25 17:12:19,393 (trainer:737) INFO: 10epoch:train:8301-8400batch: iter_time=8.818e-05, forward_time=0.294, loss_ctc=64.208, loss_att=62.517, acc=0.687, loss=63.024, backward_time=0.402, grad_norm=35.446, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.235e-04, train_time=1.422
+[gpub078:0/16] 2024-01-25 17:15:02,659 (trainer:737) INFO: 10epoch:train:8401-8500batch: iter_time=9.320e-05, forward_time=0.291, loss_ctc=64.836, loss_att=68.587, acc=0.679, loss=67.462, backward_time=0.405, grad_norm=36.963, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.234e-04, train_time=1.633
+[gpub078:0/16] 2024-01-25 17:17:39,599 (trainer:737) INFO: 10epoch:train:8501-8600batch: iter_time=9.580e-05, forward_time=0.288, loss_ctc=51.084, loss_att=40.243, acc=0.741, loss=43.495, backward_time=0.400, grad_norm=25.135, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.233e-04, train_time=1.569
+[gpub078:0/16] 2024-01-25 17:20:39,034 (trainer:737) INFO: 10epoch:train:8601-8700batch: iter_time=8.996e-05, forward_time=0.482, loss_ctc=60.526, loss_att=62.773, acc=0.692, loss=62.099, backward_time=0.480, grad_norm=33.893, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.102, optim0_lr0=3.231e-04, train_time=1.794
+[gpub078:0/16] 2024-01-25 17:22:54,315 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub078:0/16] 2024-01-25 17:23:13,352 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 17:23:16,891 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c2d4c7970>)
+[gpub078:0/16] 2024-01-25 17:23:16,891 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub078:0/16] 2024-01-25 17:23:16,916 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 17:29:34,427 (trainer:737) INFO: 10epoch:train:8701-8800batch: iter_time=3.498, forward_time=0.289, loss_ctc=56.158, loss_att=53.283, acc=0.721, loss=54.145, backward_time=0.403, grad_norm=29.152, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.230e-04, train_time=5.354
+[gpub078:0/16] 2024-01-25 17:32:09,556 (trainer:737) INFO: 10epoch:train:8801-8900batch: iter_time=9.543e-05, forward_time=0.361, loss_ctc=55.327, loss_att=52.379, acc=0.713, loss=53.263, backward_time=0.432, grad_norm=28.779, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.107, optim0_lr0=3.229e-04, train_time=1.551
+[gpub078:0/16] 2024-01-25 17:35:28,560 (trainer:737) INFO: 10epoch:train:8901-9000batch: iter_time=9.952e-05, forward_time=0.312, loss_ctc=49.431, loss_att=49.439, acc=0.721, loss=49.436, backward_time=0.410, grad_norm=26.886, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.093, optim0_lr0=3.228e-04, train_time=1.990
+[gpub078:0/16] 2024-01-25 17:37:54,185 (trainer:737) INFO: 10epoch:train:9001-9100batch: iter_time=9.668e-05, forward_time=0.341, loss_ctc=49.025, loss_att=50.203, acc=0.710, loss=49.850, backward_time=0.422, grad_norm=25.343, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.100, optim0_lr0=3.227e-04, train_time=1.456
+[gpub078:0/16] 2024-01-25 17:40:37,857 (trainer:737) INFO: 10epoch:train:9101-9200batch: iter_time=8.916e-05, forward_time=0.347, loss_ctc=55.831, loss_att=63.301, acc=0.704, loss=61.060, backward_time=0.429, grad_norm=30.719, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.096, optim0_lr0=3.226e-04, train_time=1.636
+[gpub078:0/16] 2024-01-25 17:43:16,706 (trainer:737) INFO: 10epoch:train:9201-9300batch: iter_time=8.450e-05, forward_time=0.294, loss_ctc=57.240, loss_att=58.784, acc=0.711, loss=58.321, backward_time=0.406, grad_norm=28.788, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.225e-04, train_time=1.588
+[gpub078:0/16] 2024-01-25 17:46:16,290 (trainer:737) INFO: 10epoch:train:9301-9400batch: iter_time=9.380e-05, forward_time=0.367, loss_ctc=55.382, loss_att=50.140, acc=0.720, loss=51.713, backward_time=0.459, grad_norm=27.118, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.105, optim0_lr0=3.224e-04, train_time=1.796
+[gpub078:0/16] 2024-01-25 17:48:29,360 (trainer:737) INFO: 10epoch:train:9401-9500batch: iter_time=8.899e-05, forward_time=0.289, loss_ctc=49.947, loss_att=50.319, acc=0.719, loss=50.208, backward_time=0.403, grad_norm=26.938, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.222e-04, train_time=1.330
+[gpub078:0/16] 2024-01-25 17:51:03,363 (trainer:737) INFO: 10epoch:train:9501-9600batch: iter_time=1.001e-04, forward_time=0.292, loss_ctc=57.571, loss_att=59.913, acc=0.701, loss=59.211, backward_time=0.405, grad_norm=30.205, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.221e-04, train_time=1.540
+[gpub078:0/16] 2024-01-25 17:54:23,957 (trainer:737) INFO: 10epoch:train:9601-9700batch: iter_time=8.929e-05, forward_time=0.404, loss_ctc=66.085, loss_att=61.361, acc=0.704, loss=62.778, backward_time=0.459, grad_norm=37.852, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.103, optim0_lr0=3.220e-04, train_time=2.006
+[gpub078:0/16] 2024-01-25 17:56:52,625 (trainer:737) INFO: 10epoch:train:9701-9800batch: iter_time=9.116e-05, forward_time=0.291, loss_ctc=54.976, loss_att=59.368, acc=0.715, loss=58.051, backward_time=0.404, grad_norm=29.550, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.219e-04, train_time=1.486
+[gpub078:0/16] 2024-01-25 17:59:22,169 (trainer:737) INFO: 10epoch:train:9801-9900batch: iter_time=9.309e-05, forward_time=0.403, loss_ctc=57.203, loss_att=52.772, acc=0.714, loss=54.102, backward_time=0.442, grad_norm=30.211, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.100, optim0_lr0=3.218e-04, train_time=1.495
+[gpub078:0/16] 2024-01-25 18:02:08,528 (trainer:737) INFO: 10epoch:train:9901-10000batch: iter_time=9.984e-05, forward_time=0.291, loss_ctc=53.291, loss_att=52.363, acc=0.716, loss=52.641, backward_time=0.402, grad_norm=30.148, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.217e-04, train_time=1.664
+[gpub078:0/16] 2024-01-25 18:02:28,556 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub078:0/16] 2024-01-25 18:02:47,774 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 18:02:51,688 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9952127f10>)
+[gpub078:0/16] 2024-01-25 18:02:51,688 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub078:0/16] 2024-01-25 18:02:51,691 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 18:11:08,109 (trainer:737) INFO: 10epoch:train:10001-10100batch: iter_time=3.626, forward_time=0.287, loss_ctc=58.620, loss_att=56.903, acc=0.712, loss=57.418, backward_time=0.400, grad_norm=28.386, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.216e-04, train_time=5.396
+[gpub078:0/16] 2024-01-25 18:14:30,222 (trainer:737) INFO: 10epoch:train:10101-10200batch: iter_time=8.812e-05, forward_time=0.376, loss_ctc=55.597, loss_att=54.459, acc=0.723, loss=54.800, backward_time=0.486, grad_norm=31.624, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.101, optim0_lr0=3.215e-04, train_time=2.021
+[gpub078:0/16] 2024-01-25 18:16:46,510 (trainer:737) INFO: 10epoch:train:10201-10300batch: iter_time=8.826e-05, forward_time=0.288, loss_ctc=43.104, loss_att=41.287, acc=0.720, loss=41.832, backward_time=0.400, grad_norm=24.783, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.214e-04, train_time=1.363
+[gpub078:0/16] 2024-01-25 18:20:09,441 (trainer:737) INFO: 10epoch:train:10301-10400batch: iter_time=9.505e-05, forward_time=0.293, loss_ctc=61.207, loss_att=66.058, acc=0.701, loss=64.603, backward_time=0.406, grad_norm=30.952, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.212e-04, train_time=2.029
+[gpub078:0/16] 2024-01-25 18:23:08,355 (trainer:737) INFO: 10epoch:train:10401-10500batch: iter_time=9.221e-05, forward_time=0.408, loss_ctc=51.053, loss_att=53.925, acc=0.707, loss=53.063, backward_time=0.477, grad_norm=27.722, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.104, optim0_lr0=3.211e-04, train_time=1.789
+[gpub078:0/16] 2024-01-25 18:25:47,529 (trainer:737) INFO: 10epoch:train:10501-10600batch: iter_time=8.568e-05, forward_time=0.294, loss_ctc=53.839, loss_att=52.421, acc=0.737, loss=52.847, backward_time=0.406, grad_norm=26.369, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.210e-04, train_time=1.592
+[gpub078:0/16] 2024-01-25 18:28:48,392 (trainer:737) INFO: 10epoch:train:10601-10700batch: iter_time=8.747e-05, forward_time=0.394, loss_ctc=55.782, loss_att=54.146, acc=0.711, loss=54.637, backward_time=0.438, grad_norm=30.130, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.117, optim0_lr0=3.209e-04, train_time=1.808
+[gpub078:0/16] 2024-01-25 18:31:17,539 (trainer:737) INFO: 10epoch:train:10701-10800batch: iter_time=9.314e-05, forward_time=0.288, loss_ctc=51.363, loss_att=55.020, acc=0.701, loss=53.923, backward_time=0.401, grad_norm=28.133, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.208e-04, train_time=1.490
+[gpub078:0/16] 2024-01-25 18:33:55,461 (trainer:737) INFO: 10epoch:train:10801-10900batch: iter_time=8.961e-05, forward_time=0.288, loss_ctc=62.513, loss_att=60.580, acc=0.703, loss=61.159, backward_time=0.402, grad_norm=32.759, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.093, optim0_lr0=3.207e-04, train_time=1.580
+[gpub078:0/16] 2024-01-25 18:36:51,894 (trainer:737) INFO: 10epoch:train:10901-11000batch: iter_time=8.912e-05, forward_time=0.406, loss_ctc=63.814, loss_att=67.533, acc=0.695, loss=66.417, backward_time=0.433, grad_norm=35.415, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.106, optim0_lr0=3.206e-04, train_time=1.764
+[gpub078:0/16] 2024-01-25 18:40:04,881 (trainer:737) INFO: 10epoch:train:11001-11100batch: iter_time=8.794e-05, forward_time=0.287, loss_ctc=50.831, loss_att=40.877, acc=0.739, loss=43.863, backward_time=0.398, grad_norm=24.770, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.205e-04, train_time=1.929
+[gpub078:0/16] 2024-01-25 18:42:35,361 (trainer:737) INFO: 10epoch:train:11101-11200batch: iter_time=8.879e-05, forward_time=0.292, loss_ctc=59.387, loss_att=64.809, acc=0.697, loss=63.182, backward_time=0.402, grad_norm=31.943, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.204e-04, train_time=1.505
+[gpub078:0/16] 2024-01-25 18:44:56,772 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub078:0/16] 2024-01-25 18:45:15,727 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 18:45:19,667 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c2d470220>)
+[gpub078:0/16] 2024-01-25 18:45:19,667 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-25 18:45:19,670 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 18:52:35,033 (trainer:737) INFO: 10epoch:train:11201-11300batch: iter_time=3.725, forward_time=0.430, loss_ctc=55.915, loss_att=52.446, acc=0.729, loss=53.487, backward_time=0.433, grad_norm=26.860, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.106, optim0_lr0=3.203e-04, train_time=5.996
+[gpub078:0/16] 2024-01-25 18:56:50,970 (trainer:737) INFO: 10epoch:train:11301-11400batch: iter_time=8.831e-05, forward_time=0.384, loss_ctc=55.008, loss_att=51.207, acc=0.714, loss=52.347, backward_time=0.492, grad_norm=28.151, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.102, optim0_lr0=3.201e-04, train_time=2.559
+[gpub078:0/16] 2024-01-25 19:00:19,739 (trainer:737) INFO: 10epoch:train:11401-11500batch: iter_time=8.567e-05, forward_time=0.288, loss_ctc=49.077, loss_att=48.249, acc=0.723, loss=48.498, backward_time=0.398, grad_norm=26.647, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.200e-04, train_time=2.088
+[gpub078:0/16] 2024-01-25 19:03:07,458 (trainer:737) INFO: 10epoch:train:11501-11600batch: iter_time=9.608e-05, forward_time=0.289, loss_ctc=48.918, loss_att=49.018, acc=0.714, loss=48.988, backward_time=0.399, grad_norm=24.814, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.199e-04, train_time=1.677
+[gpub078:0/16] 2024-01-25 19:06:36,385 (trainer:737) INFO: 10epoch:train:11601-11700batch: iter_time=9.144e-05, forward_time=0.355, loss_ctc=55.913, loss_att=63.019, acc=0.707, loss=60.887, backward_time=0.508, grad_norm=30.360, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.104, optim0_lr0=3.198e-04, train_time=2.089
+[gpub078:0/16] 2024-01-25 19:09:47,809 (trainer:737) INFO: 10epoch:train:11701-11800batch: iter_time=8.681e-05, forward_time=0.292, loss_ctc=57.431, loss_att=57.751, acc=0.715, loss=57.655, backward_time=0.401, grad_norm=29.153, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.197e-04, train_time=1.914
+[gpub078:0/16] 2024-01-25 19:12:54,023 (trainer:737) INFO: 10epoch:train:11801-11900batch: iter_time=8.619e-05, forward_time=0.342, loss_ctc=55.142, loss_att=50.221, acc=0.721, loss=51.697, backward_time=0.491, grad_norm=26.983, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.099, optim0_lr0=3.196e-04, train_time=1.862
+[gpub078:0/16] 2024-01-25 19:15:51,710 (trainer:737) INFO: 10epoch:train:11901-12000batch: iter_time=7.958e-05, forward_time=0.289, loss_ctc=50.200, loss_att=50.622, acc=0.717, loss=50.495, backward_time=0.398, grad_norm=27.627, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.195e-04, train_time=1.777
+[gpub078:0/16] 2024-01-25 19:20:00,176 (trainer:737) INFO: 10epoch:train:12001-12100batch: iter_time=8.990e-05, forward_time=0.359, loss_ctc=57.100, loss_att=58.939, acc=0.702, loss=58.387, backward_time=0.492, grad_norm=29.696, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.100, optim0_lr0=3.194e-04, train_time=2.484
+[gpub078:0/16] 2024-01-25 19:23:18,870 (trainer:737) INFO: 10epoch:train:12101-12200batch: iter_time=8.582e-05, forward_time=0.292, loss_ctc=64.869, loss_att=60.269, acc=0.708, loss=61.649, backward_time=0.484, grad_norm=33.772, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.099, optim0_lr0=3.193e-04, train_time=1.986
+[gpub078:0/16] 2024-01-25 19:26:20,930 (trainer:737) INFO: 10epoch:train:12201-12300batch: iter_time=9.559e-05, forward_time=0.331, loss_ctc=54.617, loss_att=58.474, acc=0.716, loss=57.317, backward_time=0.400, grad_norm=29.911, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.192e-04, train_time=1.821
+[gpub078:0/16] 2024-01-25 19:29:41,259 (trainer:737) INFO: 10epoch:train:12301-12400batch: iter_time=9.026e-05, forward_time=0.288, loss_ctc=57.727, loss_att=52.989, acc=0.714, loss=54.410, backward_time=0.397, grad_norm=29.265, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.191e-04, train_time=2.003
+[gpub078:0/16] 2024-01-25 19:33:04,796 (trainer:737) INFO: 10epoch:train:12401-12500batch: iter_time=9.026e-05, forward_time=0.426, loss_ctc=52.636, loss_att=52.950, acc=0.717, loss=52.856, backward_time=0.469, grad_norm=28.595, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.104, optim0_lr0=3.189e-04, train_time=2.035
+[gpub078:0/16] 2024-01-25 19:33:24,976 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub078:0/16] 2024-01-25 19:33:44,032 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 19:33:47,976 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99b84f2f50>)
+[gpub078:0/16] 2024-01-25 19:33:47,977 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub078:0/16] 2024-01-25 19:33:47,980 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 19:40:31,594 (trainer:737) INFO: 10epoch:train:12501-12600batch: iter_time=2.860, forward_time=0.288, loss_ctc=58.211, loss_att=58.177, acc=0.703, loss=58.187, backward_time=0.400, grad_norm=30.701, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.188e-04, train_time=4.468
+[gpub078:0/16] 2024-01-25 19:43:30,204 (trainer:737) INFO: 10epoch:train:12601-12700batch: iter_time=8.952e-05, forward_time=0.346, loss_ctc=54.955, loss_att=55.722, acc=0.713, loss=55.492, backward_time=0.460, grad_norm=30.940, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.099, optim0_lr0=3.187e-04, train_time=1.785
+[gpub078:0/16] 2024-01-25 19:46:17,099 (trainer:737) INFO: 10epoch:train:12701-12800batch: iter_time=8.599e-05, forward_time=0.286, loss_ctc=42.784, loss_att=42.645, acc=0.706, loss=42.687, backward_time=0.396, grad_norm=24.979, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.186e-04, train_time=1.669
+[gpub078:0/16] 2024-01-25 19:49:56,967 (trainer:737) INFO: 10epoch:train:12801-12900batch: iter_time=8.962e-05, forward_time=0.553, loss_ctc=60.743, loss_att=66.713, acc=0.692, loss=64.922, backward_time=0.447, grad_norm=31.826, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.113, optim0_lr0=3.185e-04, train_time=2.198
+[gpub078:0/16] 2024-01-25 19:53:02,647 (trainer:737) INFO: 10epoch:train:12901-13000batch: iter_time=9.129e-05, forward_time=0.287, loss_ctc=50.612, loss_att=51.411, acc=0.710, loss=51.171, backward_time=0.398, grad_norm=27.752, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.093, optim0_lr0=3.184e-04, train_time=1.857
+[gpub078:0/16] 2024-01-25 19:56:39,312 (trainer:737) INFO: 10epoch:train:13001-13100batch: iter_time=8.667e-05, forward_time=0.312, loss_ctc=53.202, loss_att=51.313, acc=0.733, loss=51.880, backward_time=0.438, grad_norm=27.191, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.099, optim0_lr0=3.183e-04, train_time=2.166
+[gpub078:0/16] 2024-01-25 19:59:55,919 (trainer:737) INFO: 10epoch:train:13101-13200batch: iter_time=1.023e-04, forward_time=0.439, loss_ctc=55.637, loss_att=53.112, acc=0.710, loss=53.869, backward_time=0.441, grad_norm=29.735, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.119, optim0_lr0=3.182e-04, train_time=1.966
+[gpub078:0/16] 2024-01-25 20:03:19,448 (trainer:737) INFO: 10epoch:train:13201-13300batch: iter_time=1.034e-04, forward_time=0.319, loss_ctc=50.940, loss_att=54.038, acc=0.698, loss=53.109, backward_time=0.433, grad_norm=29.736, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.097, optim0_lr0=3.181e-04, train_time=2.035
+[gpub078:0/16] 2024-01-25 20:06:45,252 (trainer:737) INFO: 10epoch:train:13301-13400batch: iter_time=8.540e-05, forward_time=0.421, loss_ctc=63.442, loss_att=61.859, acc=0.689, loss=62.334, backward_time=0.427, grad_norm=37.237, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.143, optim0_lr0=3.180e-04, train_time=2.058
+[gpub078:0/16] 2024-01-25 20:10:33,823 (trainer:737) INFO: 10epoch:train:13401-13500batch: iter_time=9.664e-05, forward_time=0.361, loss_ctc=62.729, loss_att=68.007, acc=0.680, loss=66.424, backward_time=0.422, grad_norm=36.040, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.111, optim0_lr0=3.179e-04, train_time=2.285
+[gpub078:0/16] 2024-01-25 20:14:19,436 (trainer:737) INFO: 10epoch:train:13501-13600batch: iter_time=9.868e-05, forward_time=0.445, loss_ctc=50.600, loss_att=40.083, acc=0.743, loss=43.238, backward_time=0.428, grad_norm=25.286, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.106, optim0_lr0=3.178e-04, train_time=2.256
+[gpub078:0/16] 2024-01-25 20:16:48,852 (trainer:737) INFO: 10epoch:train:13601-13700batch: iter_time=1.058e-04, forward_time=0.290, loss_ctc=59.491, loss_att=62.588, acc=0.693, loss=61.659, backward_time=0.401, grad_norm=32.938, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.177e-04, train_time=1.494
+[gpub078:0/16] 2024-01-25 20:18:53,488 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub078:0/16] 2024-01-25 20:19:13,486 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 20:19:17,487 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c4b7944c0>)
+[gpub078:0/16] 2024-01-25 20:19:17,487 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub078:0/16] 2024-01-25 20:19:17,490 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 20:27:46,565 (trainer:737) INFO: 10epoch:train:13701-13800batch: iter_time=4.512, forward_time=0.484, loss_ctc=55.194, loss_att=52.058, acc=0.721, loss=52.999, backward_time=0.469, grad_norm=29.315, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.098, optim0_lr0=3.176e-04, train_time=6.577
+[gpub078:0/16] 2024-01-25 20:30:21,954 (trainer:737) INFO: 10epoch:train:13801-13900batch: iter_time=8.455e-05, forward_time=0.286, loss_ctc=54.247, loss_att=51.280, acc=0.706, loss=52.170, backward_time=0.400, grad_norm=29.485, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.174e-04, train_time=1.553
+[gpub078:0/16] 2024-01-25 20:33:14,297 (trainer:737) INFO: 10epoch:train:13901-14000batch: iter_time=8.329e-05, forward_time=0.289, loss_ctc=49.262, loss_att=48.281, acc=0.716, loss=48.575, backward_time=0.399, grad_norm=27.793, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.173e-04, train_time=1.723
+[gpub078:0/16] 2024-01-25 20:36:45,822 (trainer:737) INFO: 10epoch:train:14001-14100batch: iter_time=8.861e-05, forward_time=0.489, loss_ctc=48.363, loss_att=49.344, acc=0.701, loss=49.050, backward_time=0.441, grad_norm=25.756, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.112, optim0_lr0=3.172e-04, train_time=2.115
+[gpub078:0/16] 2024-01-25 20:39:32,203 (trainer:737) INFO: 10epoch:train:14101-14200batch: iter_time=8.126e-05, forward_time=0.289, loss_ctc=54.895, loss_att=60.632, acc=0.703, loss=58.911, backward_time=0.403, grad_norm=30.389, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.171e-04, train_time=1.664
+[gpub078:0/16] 2024-01-25 20:42:30,829 (trainer:737) INFO: 10epoch:train:14201-14300batch: iter_time=8.280e-05, forward_time=0.289, loss_ctc=56.393, loss_att=57.286, acc=0.709, loss=57.018, backward_time=0.402, grad_norm=28.098, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.170e-04, train_time=1.785
+[gpub078:0/16] 2024-01-25 20:46:31,645 (trainer:737) INFO: 10epoch:train:14301-14400batch: iter_time=8.460e-05, forward_time=0.461, loss_ctc=55.192, loss_att=48.850, acc=0.715, loss=50.752, backward_time=0.471, grad_norm=29.273, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.110, optim0_lr0=3.169e-04, train_time=2.408
+[gpub078:0/16] 2024-01-25 20:48:51,906 (trainer:737) INFO: 10epoch:train:14401-14500batch: iter_time=8.933e-05, forward_time=0.286, loss_ctc=49.744, loss_att=48.827, acc=0.720, loss=49.102, backward_time=0.401, grad_norm=27.472, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.168e-04, train_time=1.403
+[gpub078:0/16] 2024-01-25 20:52:10,476 (trainer:737) INFO: 10epoch:train:14501-14600batch: iter_time=8.454e-05, forward_time=0.288, loss_ctc=56.770, loss_att=58.770, acc=0.691, loss=58.170, backward_time=0.401, grad_norm=29.349, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.167e-04, train_time=1.986
+[gpub078:0/16] 2024-01-25 20:54:37,903 (trainer:737) INFO: 10epoch:train:14601-14700batch: iter_time=8.977e-05, forward_time=0.314, loss_ctc=64.402, loss_att=60.027, acc=0.699, loss=61.340, backward_time=0.434, grad_norm=34.816, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.097, optim0_lr0=3.166e-04, train_time=1.473
+[gpub078:0/16] 2024-01-25 20:57:52,265 (trainer:737) INFO: 10epoch:train:14701-14800batch: iter_time=9.391e-05, forward_time=0.378, loss_ctc=53.359, loss_att=57.618, acc=0.709, loss=56.341, backward_time=0.436, grad_norm=27.621, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.099, optim0_lr0=3.165e-04, train_time=1.944
+[gpub078:0/16] 2024-01-25 21:00:39,632 (trainer:737) INFO: 10epoch:train:14801-14900batch: iter_time=8.847e-05, forward_time=0.287, loss_ctc=56.753, loss_att=50.479, acc=0.713, loss=52.361, backward_time=0.399, grad_norm=31.144, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.164e-04, train_time=1.673
+[gpub078:0/16] 2024-01-25 21:03:21,412 (trainer:737) INFO: 10epoch:train:14901-15000batch: iter_time=8.427e-05, forward_time=0.320, loss_ctc=52.249, loss_att=51.862, acc=0.711, loss=51.978, backward_time=0.398, grad_norm=30.534, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.093, optim0_lr0=3.163e-04, train_time=1.618
+[gpub078:0/16] 2024-01-25 21:44:17,385 (trainer:343) INFO: 10epoch results: [train] iter_time=0.287, forward_time=0.333, loss_ctc=56.382, loss_att=55.655, acc=0.706, loss=55.873, backward_time=0.421, grad_norm=29.839, clip=100.000, loss_scale=3.282e+26, optim_step_time=0.097, optim0_lr0=3.246e-04, train_time=1.996, time=8 hours, 19 minutes and 36.89 seconds, total_count=180000, gpu_max_cached_mem_GB=42.420, [valid] loss_ctc=52.152, cer_ctc=0.263, loss_att=51.216, acc=0.607, cer=0.381, wer=0.998, loss=51.497, time=40 minutes and 29.1 seconds, total_count=56052, gpu_max_cached_mem_GB=42.420
+[gpub078:0/16] 2024-01-25 21:44:28,241 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub078:0/16] 2024-01-25 21:44:28,350 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/5epoch.pth
+[gpub078:0/16] 2024-01-25 21:44:28,350 (trainer:272) INFO: 11/45epoch started. Estimated time to finish: 1 week, 6 days and 13 hours
+[gpub078:0/16] 2024-01-25 21:44:28,361 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-25 21:44:46,981 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 21:44:50,759 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f922139eef0>)
+[gpub078:0/16] 2024-01-25 21:44:50,759 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub078:0/16] 2024-01-25 21:44:50,762 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 21:54:16,991 (trainer:737) INFO: 11epoch:train:1-100batch: iter_time=4.162, forward_time=0.297, loss_ctc=63.999, loss_att=55.517, acc=0.704, loss=58.061, backward_time=0.429, grad_norm=32.087, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.096, optim0_lr0=3.162e-04, train_time=5.886
+[gpub078:0/16] 2024-01-25 21:56:51,589 (trainer:737) INFO: 11epoch:train:101-200batch: iter_time=8.385e-05, forward_time=0.390, loss_ctc=73.361, loss_att=74.568, acc=0.650, loss=74.206, backward_time=0.410, grad_norm=54.219, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.100, optim0_lr0=3.161e-04, train_time=1.546
+[gpub078:0/16] 2024-01-25 21:59:14,978 (trainer:737) INFO: 11epoch:train:201-300batch: iter_time=8.818e-05, forward_time=0.353, loss_ctc=58.990, loss_att=52.724, acc=0.695, loss=54.604, backward_time=0.426, grad_norm=34.010, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.100, optim0_lr0=3.160e-04, train_time=1.433
+[gpub078:0/16] 2024-01-25 22:02:19,301 (trainer:737) INFO: 11epoch:train:301-400batch: iter_time=9.060e-05, forward_time=0.303, loss_ctc=60.356, loss_att=58.335, acc=0.700, loss=58.941, backward_time=0.425, grad_norm=32.827, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.094, optim0_lr0=3.159e-04, train_time=1.843
+[gpub078:0/16] 2024-01-25 22:04:58,437 (trainer:737) INFO: 11epoch:train:401-500batch: iter_time=8.918e-05, forward_time=0.289, loss_ctc=52.096, loss_att=51.648, acc=0.698, loss=51.783, backward_time=0.402, grad_norm=29.034, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.093, optim0_lr0=3.158e-04, train_time=1.591
+[gpub078:0/16] 2024-01-25 22:07:51,868 (trainer:737) INFO: 11epoch:train:501-600batch: iter_time=9.884e-05, forward_time=0.426, loss_ctc=58.693, loss_att=54.223, acc=0.715, loss=55.564, backward_time=0.429, grad_norm=30.631, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.110, optim0_lr0=3.156e-04, train_time=1.734
+[gpub078:0/16] 2024-01-25 22:10:32,652 (trainer:737) INFO: 11epoch:train:601-700batch: iter_time=9.766e-05, forward_time=0.301, loss_ctc=55.643, loss_att=56.293, acc=0.698, loss=56.098, backward_time=0.426, grad_norm=30.538, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.096, optim0_lr0=3.155e-04, train_time=1.608
+[gpub078:0/16] 2024-01-25 22:12:55,630 (trainer:737) INFO: 11epoch:train:701-800batch: iter_time=9.368e-05, forward_time=0.292, loss_ctc=55.735, loss_att=52.888, acc=0.685, loss=53.742, backward_time=0.405, grad_norm=30.136, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.095, optim0_lr0=3.154e-04, train_time=1.430
+[gpub078:0/16] 2024-01-25 22:16:26,355 (trainer:737) INFO: 11epoch:train:801-900batch: iter_time=9.576e-05, forward_time=0.418, loss_ctc=63.970, loss_att=63.140, acc=0.701, loss=63.389, backward_time=0.455, grad_norm=34.882, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.103, optim0_lr0=3.153e-04, train_time=2.106
+[gpub078:0/16] 2024-01-25 22:18:52,160 (trainer:737) INFO: 11epoch:train:901-1000batch: iter_time=9.383e-05, forward_time=0.307, loss_ctc=67.886, loss_att=60.803, acc=0.708, loss=62.928, backward_time=0.431, grad_norm=35.862, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.095, optim0_lr0=3.152e-04, train_time=1.458
+[gpub078:0/16] 2024-01-25 22:21:47,881 (trainer:737) INFO: 11epoch:train:1001-1100batch: iter_time=9.946e-05, forward_time=0.295, loss_ctc=62.384, loss_att=60.018, acc=0.687, loss=60.728, backward_time=0.404, grad_norm=32.927, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.095, optim0_lr0=3.151e-04, train_time=1.758
+[gpub078:0/16] 2024-01-25 22:24:37,401 (trainer:737) INFO: 11epoch:train:1101-1200batch: iter_time=1.044e-04, forward_time=0.442, loss_ctc=61.521, loss_att=55.006, acc=0.695, loss=56.961, backward_time=0.437, grad_norm=32.427, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.118, optim0_lr0=3.150e-04, train_time=1.695
+[gpub078:0/16] 2024-01-25 22:26:05,930 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub078:0/16] 2024-01-25 22:26:25,007 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 22:26:28,600 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b5ba3b700>)
+[gpub078:0/16] 2024-01-25 22:26:28,600 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub078:0/16] 2024-01-25 22:26:28,603 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 22:35:01,446 (trainer:737) INFO: 11epoch:train:1201-1300batch: iter_time=4.608, forward_time=0.290, loss_ctc=57.822, loss_att=53.920, acc=0.689, loss=55.091, backward_time=0.403, grad_norm=31.545, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.093, optim0_lr0=3.149e-04, train_time=6.241
+[gpub078:0/16] 2024-01-25 22:37:37,446 (trainer:737) INFO: 11epoch:train:1301-1400batch: iter_time=7.834e-05, forward_time=0.296, loss_ctc=61.497, loss_att=61.371, acc=0.680, loss=61.409, backward_time=0.409, grad_norm=34.636, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.095, optim0_lr0=3.148e-04, train_time=1.560
+[gpub078:0/16] 2024-01-25 22:40:15,162 (trainer:737) INFO: 11epoch:train:1401-1500batch: iter_time=8.535e-05, forward_time=0.367, loss_ctc=61.433, loss_att=62.319, acc=0.680, loss=62.053, backward_time=0.407, grad_norm=44.408, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.099, optim0_lr0=3.147e-04, train_time=1.577
+[gpub078:0/16] 2024-01-25 22:42:35,929 (trainer:737) INFO: 11epoch:train:1501-1600batch: iter_time=8.750e-05, forward_time=0.303, loss_ctc=64.336, loss_att=56.830, acc=0.689, loss=59.082, backward_time=0.421, grad_norm=34.100, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.097, optim0_lr0=3.146e-04, train_time=1.407
+[gpub078:0/16] 2024-01-25 22:45:47,339 (trainer:737) INFO: 11epoch:train:1601-1700batch: iter_time=8.340e-05, forward_time=0.305, loss_ctc=55.738, loss_att=53.968, acc=0.703, loss=54.499, backward_time=0.414, grad_norm=31.481, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.094, optim0_lr0=3.145e-04, train_time=1.914
+[gpub078:0/16] 2024-01-25 22:48:50,097 (trainer:737) INFO: 11epoch:train:1701-1800batch: iter_time=8.276e-05, forward_time=0.294, loss_ctc=56.537, loss_att=54.917, acc=0.711, loss=55.403, backward_time=0.407, grad_norm=29.294, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.094, optim0_lr0=3.144e-04, train_time=1.827
+[gpub078:0/16] 2024-01-25 22:51:36,509 (trainer:737) INFO: 11epoch:train:1801-1900batch: iter_time=0.002, forward_time=0.511, loss_ctc=54.292, loss_att=52.673, acc=0.708, loss=53.158, backward_time=0.425, grad_norm=28.916, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.104, optim0_lr0=3.143e-04, train_time=1.664
+[gpub078:0/16] 2024-01-25 22:53:59,669 (trainer:737) INFO: 11epoch:train:1901-2000batch: iter_time=8.069e-05, forward_time=0.291, loss_ctc=51.416, loss_att=51.593, acc=0.698, loss=51.540, backward_time=0.403, grad_norm=29.749, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.093, optim0_lr0=3.142e-04, train_time=1.431
+[gpub078:0/16] 2024-01-25 22:57:19,330 (trainer:737) INFO: 11epoch:train:2001-2100batch: iter_time=8.858e-05, forward_time=0.300, loss_ctc=63.999, loss_att=64.377, acc=0.697, loss=64.263, backward_time=0.419, grad_norm=32.170, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.095, optim0_lr0=3.141e-04, train_time=1.997
+[gpub078:0/16] 2024-01-25 23:00:12,157 (trainer:737) INFO: 11epoch:train:2101-2200batch: iter_time=8.521e-05, forward_time=0.525, loss_ctc=62.340, loss_att=53.368, acc=0.717, loss=56.060, backward_time=0.438, grad_norm=33.726, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.103, optim0_lr0=3.140e-04, train_time=1.727
+[gpub078:0/16] 2024-01-25 23:02:41,480 (trainer:737) INFO: 11epoch:train:2201-2300batch: iter_time=8.394e-05, forward_time=0.290, loss_ctc=63.179, loss_att=59.321, acc=0.694, loss=60.478, backward_time=0.401, grad_norm=33.750, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.139e-04, train_time=1.494
+[gpub078:0/16] 2024-01-25 23:05:03,201 (trainer:737) INFO: 11epoch:train:2301-2400batch: iter_time=9.304e-05, forward_time=0.292, loss_ctc=61.924, loss_att=56.124, acc=0.704, loss=57.864, backward_time=0.407, grad_norm=32.652, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.094, optim0_lr0=3.138e-04, train_time=1.416
+[gpub078:0/16] 2024-01-25 23:08:17,669 (trainer:737) INFO: 11epoch:train:2401-2500batch: iter_time=9.067e-05, forward_time=0.470, loss_ctc=54.896, loss_att=55.258, acc=0.672, loss=55.149, backward_time=0.424, grad_norm=33.291, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.099, optim0_lr0=3.137e-04, train_time=1.945
+[gpub078:0/16] 2024-01-25 23:08:37,870 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub078:0/16] 2024-01-25 23:08:56,893 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 23:09:00,425 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f92816c7c40>)
+[gpub078:0/16] 2024-01-25 23:09:00,425 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub078:0/16] 2024-01-25 23:09:00,428 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 23:17:58,019 (trainer:737) INFO: 11epoch:train:2501-2600batch: iter_time=4.294, forward_time=0.289, loss_ctc=62.113, loss_att=53.936, acc=0.711, loss=56.389, backward_time=0.402, grad_norm=31.873, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.136e-04, train_time=5.803
+[gpub078:0/16] 2024-01-25 23:20:23,467 (trainer:737) INFO: 11epoch:train:2601-2700batch: iter_time=8.754e-05, forward_time=0.291, loss_ctc=66.836, loss_att=71.534, acc=0.659, loss=70.125, backward_time=0.403, grad_norm=44.182, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.135e-04, train_time=1.454
+[gpub078:0/16] 2024-01-25 23:23:30,063 (trainer:737) INFO: 11epoch:train:2701-2800batch: iter_time=8.864e-05, forward_time=0.393, loss_ctc=56.932, loss_att=51.147, acc=0.703, loss=52.882, backward_time=0.496, grad_norm=31.944, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.104, optim0_lr0=3.134e-04, train_time=1.866
+[gpub078:0/16] 2024-01-25 23:26:11,337 (trainer:737) INFO: 11epoch:train:2801-2900batch: iter_time=8.871e-05, forward_time=0.291, loss_ctc=57.993, loss_att=56.603, acc=0.708, loss=57.020, backward_time=0.404, grad_norm=30.670, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.133e-04, train_time=1.613
+[gpub078:0/16] 2024-01-25 23:28:45,710 (trainer:737) INFO: 11epoch:train:2901-3000batch: iter_time=9.462e-05, forward_time=0.291, loss_ctc=50.987, loss_att=50.990, acc=0.702, loss=50.989, backward_time=0.403, grad_norm=26.786, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.132e-04, train_time=1.544
+[gpub078:0/16] 2024-01-25 23:31:17,917 (trainer:737) INFO: 11epoch:train:3001-3100batch: iter_time=9.358e-05, forward_time=0.289, loss_ctc=56.546, loss_att=52.196, acc=0.722, loss=53.501, backward_time=0.402, grad_norm=29.612, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.131e-04, train_time=1.522
+[gpub078:0/16] 2024-01-25 23:34:20,961 (trainer:737) INFO: 11epoch:train:3101-3200batch: iter_time=1.018e-04, forward_time=0.466, loss_ctc=53.846, loss_att=55.026, acc=0.702, loss=54.672, backward_time=0.430, grad_norm=29.283, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.146, optim0_lr0=3.130e-04, train_time=1.830
+[gpub078:0/16] 2024-01-25 23:37:14,228 (trainer:737) INFO: 11epoch:train:3201-3300batch: iter_time=9.710e-05, forward_time=0.290, loss_ctc=53.347, loss_att=49.845, acc=0.696, loss=50.896, backward_time=0.399, grad_norm=29.303, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.129e-04, train_time=1.733
+[gpub078:0/16] 2024-01-25 23:39:50,771 (trainer:737) INFO: 11epoch:train:3301-3400batch: iter_time=9.308e-05, forward_time=0.293, loss_ctc=61.590, loss_att=60.899, acc=0.714, loss=61.106, backward_time=0.405, grad_norm=31.206, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.094, optim0_lr0=3.128e-04, train_time=1.565
+[gpub078:0/16] 2024-01-25 23:42:54,113 (trainer:737) INFO: 11epoch:train:3401-3500batch: iter_time=2.345e-04, forward_time=0.468, loss_ctc=64.494, loss_att=58.688, acc=0.715, loss=60.430, backward_time=0.488, grad_norm=32.136, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.102, optim0_lr0=3.127e-04, train_time=1.833
+[gpub078:0/16] 2024-01-25 23:45:28,801 (trainer:737) INFO: 11epoch:train:3501-3600batch: iter_time=9.321e-05, forward_time=0.289, loss_ctc=59.928, loss_att=58.090, acc=0.692, loss=58.641, backward_time=0.401, grad_norm=32.367, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.125e-04, train_time=1.546
+[gpub078:0/16] 2024-01-25 23:48:22,931 (trainer:737) INFO: 11epoch:train:3601-3700batch: iter_time=9.172e-05, forward_time=0.292, loss_ctc=58.371, loss_att=52.794, acc=0.703, loss=54.467, backward_time=0.399, grad_norm=31.589, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.124e-04, train_time=1.742
+[gpub078:0/16] 2024-01-25 23:49:52,815 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub078:0/16] 2024-01-25 23:50:12,025 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-25 23:50:15,677 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b5af7a560>)
+[gpub078:0/16] 2024-01-25 23:50:15,677 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub078:0/16] 2024-01-25 23:50:15,680 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-25 23:58:36,581 (trainer:737) INFO: 11epoch:train:3701-3800batch: iter_time=4.332, forward_time=0.290, loss_ctc=56.669, loss_att=53.127, acc=0.692, loss=54.190, backward_time=0.401, grad_norm=31.388, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.123e-04, train_time=6.136
+[gpub078:0/16] 2024-01-26 00:01:36,519 (trainer:737) INFO: 11epoch:train:3801-3900batch: iter_time=8.955e-05, forward_time=0.493, loss_ctc=59.983, loss_att=60.154, acc=0.683, loss=60.103, backward_time=0.457, grad_norm=32.578, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.124, optim0_lr0=3.122e-04, train_time=1.798
+[gpub078:0/16] 2024-01-26 00:04:00,279 (trainer:737) INFO: 11epoch:train:3901-4000batch: iter_time=8.289e-05, forward_time=0.289, loss_ctc=61.288, loss_att=61.371, acc=0.685, loss=61.346, backward_time=0.403, grad_norm=45.070, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.093, optim0_lr0=3.121e-04, train_time=1.438
+[gpub078:0/16] 2024-01-26 00:07:32,287 (trainer:737) INFO: 11epoch:train:4001-4100batch: iter_time=9.700e-05, forward_time=0.291, loss_ctc=62.440, loss_att=55.714, acc=0.693, loss=57.732, backward_time=0.402, grad_norm=31.508, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.120e-04, train_time=2.120
+[gpub078:0/16] 2024-01-26 00:09:46,747 (trainer:737) INFO: 11epoch:train:4101-4200batch: iter_time=1.015e-04, forward_time=0.290, loss_ctc=54.856, loss_att=53.282, acc=0.708, loss=53.754, backward_time=0.404, grad_norm=29.758, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.119e-04, train_time=1.344
+[gpub078:0/16] 2024-01-26 00:12:52,630 (trainer:737) INFO: 11epoch:train:4201-4300batch: iter_time=9.463e-05, forward_time=0.420, loss_ctc=54.590, loss_att=52.897, acc=0.717, loss=53.405, backward_time=0.486, grad_norm=27.128, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.108, optim0_lr0=3.118e-04, train_time=1.858
+[gpub078:0/16] 2024-01-26 00:15:40,720 (trainer:737) INFO: 11epoch:train:4301-4400batch: iter_time=8.625e-05, forward_time=0.290, loss_ctc=53.278, loss_att=52.732, acc=0.710, loss=52.896, backward_time=0.402, grad_norm=28.127, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.117e-04, train_time=1.681
+[gpub078:0/16] 2024-01-26 00:18:19,815 (trainer:737) INFO: 11epoch:train:4401-4500batch: iter_time=7.939e-05, forward_time=0.323, loss_ctc=50.281, loss_att=50.985, acc=0.700, loss=50.774, backward_time=0.401, grad_norm=30.158, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.116e-04, train_time=1.591
+[gpub078:0/16] 2024-01-26 00:22:01,195 (trainer:737) INFO: 11epoch:train:4501-4600batch: iter_time=8.462e-05, forward_time=0.567, loss_ctc=63.029, loss_att=64.001, acc=0.701, loss=63.709, backward_time=0.446, grad_norm=31.619, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.123, optim0_lr0=3.115e-04, train_time=2.214
+[gpub078:0/16] 2024-01-26 00:24:19,790 (trainer:737) INFO: 11epoch:train:4601-4700batch: iter_time=8.948e-05, forward_time=0.290, loss_ctc=61.180, loss_att=52.502, acc=0.721, loss=55.105, backward_time=0.405, grad_norm=34.266, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.114e-04, train_time=1.385
+[gpub078:0/16] 2024-01-26 00:28:28,545 (trainer:737) INFO: 11epoch:train:4701-4800batch: iter_time=1.042e-04, forward_time=0.291, loss_ctc=60.939, loss_att=58.469, acc=0.697, loss=59.210, backward_time=0.404, grad_norm=34.370, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.113e-04, train_time=2.487
+[gpub078:0/16] 2024-01-26 00:30:46,234 (trainer:737) INFO: 11epoch:train:4801-4900batch: iter_time=9.618e-05, forward_time=0.292, loss_ctc=60.108, loss_att=54.804, acc=0.707, loss=56.396, backward_time=0.404, grad_norm=31.107, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.112e-04, train_time=1.378
+[gpub078:0/16] 2024-01-26 00:33:36,560 (trainer:737) INFO: 11epoch:train:4901-5000batch: iter_time=9.102e-05, forward_time=0.460, loss_ctc=54.315, loss_att=54.426, acc=0.675, loss=54.392, backward_time=0.539, grad_norm=31.947, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.103, optim0_lr0=3.111e-04, train_time=1.703
+[gpub078:0/16] 2024-01-26 00:33:56,588 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub078:0/16] 2024-01-26 00:34:15,732 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 00:34:19,278 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f924cb95270>)
+[gpub078:0/16] 2024-01-26 00:34:19,278 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub078:0/16] 2024-01-26 00:34:19,281 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 00:46:11,947 (trainer:737) INFO: 11epoch:train:5001-5100batch: iter_time=5.492, forward_time=0.459, loss_ctc=61.287, loss_att=55.473, acc=0.722, loss=57.217, backward_time=0.529, grad_norm=33.431, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.111, optim0_lr0=3.110e-04, train_time=7.554
+[gpub078:0/16] 2024-01-26 00:49:04,577 (trainer:737) INFO: 11epoch:train:5101-5200batch: iter_time=8.455e-05, forward_time=0.292, loss_ctc=63.932, loss_att=74.844, acc=0.662, loss=71.570, backward_time=0.405, grad_norm=41.003, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.109e-04, train_time=1.725
+[gpub078:0/16] 2024-01-26 00:52:40,335 (trainer:737) INFO: 11epoch:train:5201-5300batch: iter_time=8.361e-05, forward_time=0.521, loss_ctc=55.875, loss_att=53.128, acc=0.709, loss=53.952, backward_time=0.444, grad_norm=32.203, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.146, optim0_lr0=3.108e-04, train_time=2.158
+[gpub078:0/16] 2024-01-26 00:55:34,954 (trainer:737) INFO: 11epoch:train:5301-5400batch: iter_time=8.676e-05, forward_time=0.340, loss_ctc=56.750, loss_att=58.482, acc=0.710, loss=57.963, backward_time=0.405, grad_norm=31.576, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.093, optim0_lr0=3.107e-04, train_time=1.746
+[gpub078:0/16] 2024-01-26 00:59:07,215 (trainer:737) INFO: 11epoch:train:5401-5500batch: iter_time=6.003e-04, forward_time=0.479, loss_ctc=50.228, loss_att=53.094, acc=0.713, loss=52.234, backward_time=0.450, grad_norm=26.553, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.106, optim0_lr0=3.106e-04, train_time=2.123
+[gpub078:0/16] 2024-01-26 01:02:01,684 (trainer:737) INFO: 11epoch:train:5501-5600batch: iter_time=9.155e-05, forward_time=0.292, loss_ctc=56.230, loss_att=56.216, acc=0.726, loss=56.220, backward_time=0.414, grad_norm=28.865, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.096, optim0_lr0=3.105e-04, train_time=1.744
+[gpub078:0/16] 2024-01-26 01:06:28,862 (trainer:737) INFO: 11epoch:train:5601-5700batch: iter_time=8.756e-05, forward_time=0.582, loss_ctc=52.930, loss_att=54.960, acc=0.706, loss=54.351, backward_time=0.453, grad_norm=28.857, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.098, optim0_lr0=3.104e-04, train_time=2.672
+[gpub078:0/16] 2024-01-26 01:08:52,416 (trainer:737) INFO: 11epoch:train:5701-5800batch: iter_time=6.102e-04, forward_time=0.345, loss_ctc=52.309, loss_att=50.099, acc=0.703, loss=50.762, backward_time=0.413, grad_norm=28.851, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.096, optim0_lr0=3.103e-04, train_time=1.434
+[gpub078:0/16] 2024-01-26 01:13:05,684 (trainer:737) INFO: 11epoch:train:5801-5900batch: iter_time=0.023, forward_time=0.522, loss_ctc=60.238, loss_att=61.913, acc=0.716, loss=61.411, backward_time=0.435, grad_norm=31.501, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.095, optim0_lr0=3.102e-04, train_time=2.532
+[gpub078:0/16] 2024-01-26 01:16:19,055 (trainer:737) INFO: 11epoch:train:5901-6000batch: iter_time=3.281e-04, forward_time=0.472, loss_ctc=62.998, loss_att=59.026, acc=0.720, loss=60.217, backward_time=0.479, grad_norm=33.204, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.116, optim0_lr0=3.101e-04, train_time=1.934
+[gpub078:0/16] 2024-01-26 01:19:20,138 (trainer:737) INFO: 11epoch:train:6001-6100batch: iter_time=8.685e-05, forward_time=0.307, loss_ctc=58.904, loss_att=59.876, acc=0.698, loss=59.584, backward_time=0.403, grad_norm=32.602, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.095, optim0_lr0=3.100e-04, train_time=1.811
+[gpub078:0/16] 2024-01-26 01:22:39,344 (trainer:737) INFO: 11epoch:train:6101-6200batch: iter_time=2.155e-04, forward_time=0.548, loss_ctc=57.966, loss_att=53.135, acc=0.717, loss=54.584, backward_time=0.443, grad_norm=31.845, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.097, optim0_lr0=3.099e-04, train_time=1.992
+[gpub078:0/16] 2024-01-26 01:24:55,036 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub078:0/16] 2024-01-26 01:25:13,891 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 01:25:17,835 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99033468c0>)
+[gpub078:0/16] 2024-01-26 01:25:17,835 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub078:0/16] 2024-01-26 01:25:18,064 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 01:35:22,225 (trainer:737) INFO: 11epoch:train:6201-6300batch: iter_time=5.653, forward_time=0.385, loss_ctc=55.155, loss_att=53.000, acc=0.709, loss=53.647, backward_time=0.420, grad_norm=29.917, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.097, optim0_lr0=3.098e-04, train_time=7.629
+[gpub078:0/16] 2024-01-26 01:37:58,460 (trainer:737) INFO: 11epoch:train:6301-6400batch: iter_time=1.032e-04, forward_time=0.313, loss_ctc=59.904, loss_att=61.556, acc=0.692, loss=61.060, backward_time=0.407, grad_norm=32.064, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.094, optim0_lr0=3.097e-04, train_time=1.562
+[gpub078:0/16] 2024-01-26 01:40:44,883 (trainer:737) INFO: 11epoch:train:6401-6500batch: iter_time=9.958e-04, forward_time=0.443, loss_ctc=57.734, loss_att=60.156, acc=0.694, loss=59.430, backward_time=0.475, grad_norm=44.002, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.100, optim0_lr0=3.096e-04, train_time=1.664
+[gpub078:0/16] 2024-01-26 01:43:52,058 (trainer:737) INFO: 11epoch:train:6501-6600batch: iter_time=9.971e-05, forward_time=0.300, loss_ctc=61.967, loss_att=57.632, acc=0.703, loss=58.933, backward_time=0.418, grad_norm=32.218, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.095e-04, train_time=1.872
+[gpub078:0/16] 2024-01-26 01:46:27,360 (trainer:737) INFO: 11epoch:train:6601-6700batch: iter_time=1.040e-04, forward_time=0.290, loss_ctc=54.439, loss_att=53.746, acc=0.716, loss=53.954, backward_time=0.405, grad_norm=28.697, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.094, optim0_lr0=3.094e-04, train_time=1.553
+[gpub078:0/16] 2024-01-26 01:49:14,686 (trainer:737) INFO: 11epoch:train:6701-6800batch: iter_time=2.093e-04, forward_time=0.488, loss_ctc=53.870, loss_att=53.746, acc=0.729, loss=53.783, backward_time=0.440, grad_norm=26.484, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.138, optim0_lr0=3.093e-04, train_time=1.673
+[gpub078:0/16] 2024-01-26 01:52:17,015 (trainer:737) INFO: 11epoch:train:6801-6900batch: iter_time=9.776e-05, forward_time=0.287, loss_ctc=52.117, loss_att=54.879, acc=0.715, loss=54.050, backward_time=0.402, grad_norm=28.529, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.092e-04, train_time=1.823
+[gpub078:0/16] 2024-01-26 01:55:01,951 (trainer:737) INFO: 11epoch:train:6901-7000batch: iter_time=9.816e-05, forward_time=0.309, loss_ctc=50.115, loss_att=51.221, acc=0.704, loss=50.889, backward_time=0.407, grad_norm=30.102, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.091e-04, train_time=1.649
+[gpub078:0/16] 2024-01-26 01:58:27,364 (trainer:737) INFO: 11epoch:train:7001-7100batch: iter_time=0.001, forward_time=0.440, loss_ctc=62.268, loss_att=63.217, acc=0.712, loss=62.932, backward_time=0.440, grad_norm=29.874, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.121, optim0_lr0=3.090e-04, train_time=2.054
+[gpub078:0/16] 2024-01-26 02:00:47,046 (trainer:737) INFO: 11epoch:train:7101-7200batch: iter_time=8.859e-05, forward_time=0.312, loss_ctc=60.067, loss_att=53.705, acc=0.726, loss=55.614, backward_time=0.407, grad_norm=32.219, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.089e-04, train_time=1.396
+[gpub078:0/16] 2024-01-26 02:03:02,510 (trainer:737) INFO: 11epoch:train:7201-7300batch: iter_time=9.517e-05, forward_time=0.293, loss_ctc=59.461, loss_att=58.507, acc=0.704, loss=58.793, backward_time=0.405, grad_norm=33.501, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.089e-04, train_time=1.355
+[gpub078:0/16] 2024-01-26 02:06:38,118 (trainer:737) INFO: 11epoch:train:7301-7400batch: iter_time=1.033e-04, forward_time=0.373, loss_ctc=59.761, loss_att=56.362, acc=0.716, loss=57.382, backward_time=0.498, grad_norm=32.260, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.101, optim0_lr0=3.088e-04, train_time=2.156
+[gpub078:0/16] 2024-01-26 02:08:50,019 (trainer:737) INFO: 11epoch:train:7401-7500batch: iter_time=8.462e-05, forward_time=0.289, loss_ctc=52.803, loss_att=54.994, acc=0.689, loss=54.337, backward_time=0.402, grad_norm=30.225, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.087e-04, train_time=1.319
+[gpub078:0/16] 2024-01-26 02:09:10,047 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub078:0/16] 2024-01-26 02:09:28,783 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 02:09:32,628 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97e54b7fd0>)
+[gpub078:0/16] 2024-01-26 02:09:32,628 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub078:0/16] 2024-01-26 02:09:32,631 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 02:18:07,798 (trainer:737) INFO: 11epoch:train:7501-7600batch: iter_time=4.133, forward_time=0.308, loss_ctc=60.826, loss_att=52.968, acc=0.726, loss=55.325, backward_time=0.408, grad_norm=30.856, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.086e-04, train_time=5.577
+[gpub078:0/16] 2024-01-26 02:21:55,929 (trainer:737) INFO: 11epoch:train:7601-7700batch: iter_time=8.243e-05, forward_time=0.354, loss_ctc=65.314, loss_att=71.765, acc=0.669, loss=69.830, backward_time=0.515, grad_norm=41.282, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.103, optim0_lr0=3.085e-04, train_time=2.281
+[gpub078:0/16] 2024-01-26 02:24:10,999 (trainer:737) INFO: 11epoch:train:7701-7800batch: iter_time=9.068e-05, forward_time=0.310, loss_ctc=54.961, loss_att=51.265, acc=0.713, loss=52.374, backward_time=0.405, grad_norm=29.746, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.093, optim0_lr0=3.084e-04, train_time=1.351
+[gpub078:0/16] 2024-01-26 02:26:34,562 (trainer:737) INFO: 11epoch:train:7801-7900batch: iter_time=9.499e-05, forward_time=0.309, loss_ctc=55.901, loss_att=55.730, acc=0.721, loss=55.781, backward_time=0.413, grad_norm=30.192, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.103, optim0_lr0=3.083e-04, train_time=1.435
+[gpub078:0/16] 2024-01-26 02:29:32,364 (trainer:737) INFO: 11epoch:train:7901-8000batch: iter_time=9.721e-05, forward_time=0.344, loss_ctc=50.062, loss_att=50.725, acc=0.721, loss=50.526, backward_time=0.466, grad_norm=27.078, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.096, optim0_lr0=3.082e-04, train_time=1.778
+[gpub078:0/16] 2024-01-26 02:32:30,499 (trainer:737) INFO: 11epoch:train:8001-8100batch: iter_time=9.297e-05, forward_time=0.313, loss_ctc=55.852, loss_att=55.511, acc=0.726, loss=55.614, backward_time=0.407, grad_norm=29.642, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.081e-04, train_time=1.782
+[gpub078:0/16] 2024-01-26 02:34:43,018 (trainer:737) INFO: 11epoch:train:8101-8200batch: iter_time=9.266e-05, forward_time=0.290, loss_ctc=52.582, loss_att=54.557, acc=0.708, loss=53.965, backward_time=0.403, grad_norm=29.325, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.080e-04, train_time=1.325
+[gpub078:0/16] 2024-01-26 02:37:57,053 (trainer:737) INFO: 11epoch:train:8201-8300batch: iter_time=9.039e-05, forward_time=0.360, loss_ctc=52.426, loss_att=49.462, acc=0.707, loss=50.351, backward_time=0.541, grad_norm=30.153, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.106, optim0_lr0=3.079e-04, train_time=1.940
+[gpub078:0/16] 2024-01-26 02:41:06,305 (trainer:737) INFO: 11epoch:train:8301-8400batch: iter_time=9.295e-05, forward_time=0.308, loss_ctc=59.650, loss_att=61.414, acc=0.719, loss=60.885, backward_time=0.413, grad_norm=31.910, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.078e-04, train_time=1.892
+[gpub078:0/16] 2024-01-26 02:43:45,539 (trainer:737) INFO: 11epoch:train:8401-8500batch: iter_time=9.564e-05, forward_time=0.478, loss_ctc=62.648, loss_att=57.748, acc=0.726, loss=59.218, backward_time=0.434, grad_norm=31.731, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.098, optim0_lr0=3.077e-04, train_time=1.593
+[gpub078:0/16] 2024-01-26 02:46:26,625 (trainer:737) INFO: 11epoch:train:8501-8600batch: iter_time=8.512e-05, forward_time=0.294, loss_ctc=58.216, loss_att=59.085, acc=0.700, loss=58.824, backward_time=0.403, grad_norm=32.924, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.076e-04, train_time=1.609
+[gpub078:0/16] 2024-01-26 02:49:11,055 (trainer:737) INFO: 11epoch:train:8601-8700batch: iter_time=9.348e-05, forward_time=0.312, loss_ctc=56.984, loss_att=52.107, acc=0.720, loss=53.570, backward_time=0.414, grad_norm=32.754, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.075e-04, train_time=1.645
+[gpub078:0/16] 2024-01-26 02:50:55,326 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub078:0/16] 2024-01-26 02:51:14,692 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 02:51:18,359 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f995ec63e20>)
+[gpub078:0/16] 2024-01-26 02:51:18,359 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub078:0/16] 2024-01-26 02:51:18,362 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 02:59:00,306 (trainer:737) INFO: 11epoch:train:8701-8800batch: iter_time=4.261, forward_time=0.378, loss_ctc=54.978, loss_att=52.571, acc=0.709, loss=53.293, backward_time=0.421, grad_norm=28.401, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.094, optim0_lr0=3.074e-04, train_time=5.893
+[gpub078:0/16] 2024-01-26 03:01:24,892 (trainer:737) INFO: 11epoch:train:8801-8900batch: iter_time=7.798e-05, forward_time=0.290, loss_ctc=58.909, loss_att=60.049, acc=0.696, loss=59.707, backward_time=0.405, grad_norm=32.924, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.073e-04, train_time=1.445
+[gpub078:0/16] 2024-01-26 03:04:17,780 (trainer:737) INFO: 11epoch:train:8901-9000batch: iter_time=8.804e-05, forward_time=0.302, loss_ctc=56.280, loss_att=59.451, acc=0.698, loss=58.500, backward_time=0.407, grad_norm=39.910, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.072e-04, train_time=1.729
+[gpub078:0/16] 2024-01-26 03:07:22,059 (trainer:737) INFO: 11epoch:train:9001-9100batch: iter_time=0.003, forward_time=0.391, loss_ctc=60.868, loss_att=56.688, acc=0.706, loss=57.942, backward_time=0.447, grad_norm=31.654, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.103, optim0_lr0=3.071e-04, train_time=1.843
+[gpub078:0/16] 2024-01-26 03:09:36,588 (trainer:737) INFO: 11epoch:train:9101-9200batch: iter_time=9.581e-05, forward_time=0.291, loss_ctc=54.348, loss_att=53.165, acc=0.717, loss=53.520, backward_time=0.404, grad_norm=36.381, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.070e-04, train_time=1.345
+[gpub078:0/16] 2024-01-26 03:12:26,590 (trainer:737) INFO: 11epoch:train:9201-9300batch: iter_time=1.182e-04, forward_time=0.311, loss_ctc=53.932, loss_att=53.080, acc=0.731, loss=53.336, backward_time=0.409, grad_norm=26.450, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.069e-04, train_time=1.699
+[gpub078:0/16] 2024-01-26 03:15:16,984 (trainer:737) INFO: 11epoch:train:9301-9400batch: iter_time=1.075e-04, forward_time=0.393, loss_ctc=52.391, loss_att=55.155, acc=0.716, loss=54.326, backward_time=0.413, grad_norm=28.547, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.107, optim0_lr0=3.068e-04, train_time=1.704
+[gpub078:0/16] 2024-01-26 03:17:37,004 (trainer:737) INFO: 11epoch:train:9401-9500batch: iter_time=9.730e-05, forward_time=0.331, loss_ctc=49.447, loss_att=50.369, acc=0.709, loss=50.092, backward_time=0.415, grad_norm=28.386, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.067e-04, train_time=1.400
+[gpub078:0/16] 2024-01-26 03:20:21,279 (trainer:737) INFO: 11epoch:train:9501-9600batch: iter_time=9.181e-05, forward_time=0.308, loss_ctc=61.857, loss_att=62.623, acc=0.716, loss=62.393, backward_time=0.413, grad_norm=31.695, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.094, optim0_lr0=3.066e-04, train_time=1.642
+[gpub078:0/16] 2024-01-26 03:23:31,658 (trainer:737) INFO: 11epoch:train:9601-9700batch: iter_time=9.876e-05, forward_time=0.333, loss_ctc=59.273, loss_att=52.677, acc=0.729, loss=54.656, backward_time=0.463, grad_norm=33.061, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.106, optim0_lr0=3.065e-04, train_time=1.904
+[gpub078:0/16] 2024-01-26 03:25:50,567 (trainer:737) INFO: 11epoch:train:9701-9800batch: iter_time=9.816e-05, forward_time=0.331, loss_ctc=59.278, loss_att=57.788, acc=0.705, loss=58.235, backward_time=0.417, grad_norm=33.424, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.064e-04, train_time=1.389
+[gpub078:0/16] 2024-01-26 03:28:20,543 (trainer:737) INFO: 11epoch:train:9801-9900batch: iter_time=9.027e-05, forward_time=0.316, loss_ctc=59.065, loss_att=55.901, acc=0.720, loss=56.850, backward_time=0.411, grad_norm=29.954, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.093, optim0_lr0=3.063e-04, train_time=1.499
+[gpub078:0/16] 2024-01-26 03:31:08,072 (trainer:737) INFO: 11epoch:train:9901-10000batch: iter_time=8.729e-05, forward_time=0.340, loss_ctc=52.719, loss_att=54.411, acc=0.691, loss=53.904, backward_time=0.411, grad_norm=29.552, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.095, optim0_lr0=3.062e-04, train_time=1.675
+[gpub078:0/16] 2024-01-26 03:31:28,433 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub078:0/16] 2024-01-26 03:31:47,710 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 03:31:51,347 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f97df99f7c0>)
+[gpub078:0/16] 2024-01-26 03:31:51,347 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-26 03:31:51,350 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 03:40:35,607 (trainer:737) INFO: 11epoch:train:10001-10100batch: iter_time=3.967, forward_time=0.369, loss_ctc=60.760, loss_att=52.385, acc=0.730, loss=54.897, backward_time=0.418, grad_norm=30.634, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.096, optim0_lr0=3.061e-04, train_time=5.675
+[gpub078:0/16] 2024-01-26 03:43:09,827 (trainer:737) INFO: 11epoch:train:10101-10200batch: iter_time=8.664e-05, forward_time=0.324, loss_ctc=61.808, loss_att=70.731, acc=0.674, loss=68.054, backward_time=0.417, grad_norm=40.193, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.101, optim0_lr0=3.060e-04, train_time=1.542
+[gpub078:0/16] 2024-01-26 03:45:46,408 (trainer:737) INFO: 11epoch:train:10201-10300batch: iter_time=1.146e-04, forward_time=0.386, loss_ctc=54.946, loss_att=51.598, acc=0.713, loss=52.603, backward_time=0.413, grad_norm=31.191, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.097, optim0_lr0=3.059e-04, train_time=1.565
+[gpub078:0/16] 2024-01-26 03:48:31,675 (trainer:737) INFO: 11epoch:train:10301-10400batch: iter_time=9.296e-05, forward_time=0.315, loss_ctc=55.932, loss_att=56.122, acc=0.720, loss=56.065, backward_time=0.409, grad_norm=29.567, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.059e-04, train_time=1.652
+[gpub078:0/16] 2024-01-26 03:51:11,800 (trainer:737) INFO: 11epoch:train:10401-10500batch: iter_time=9.385e-05, forward_time=0.362, loss_ctc=49.719, loss_att=50.587, acc=0.721, loss=50.326, backward_time=0.482, grad_norm=27.469, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.102, optim0_lr0=3.058e-04, train_time=1.601
+[gpub078:0/16] 2024-01-26 03:54:04,442 (trainer:737) INFO: 11epoch:train:10501-10600batch: iter_time=9.022e-05, forward_time=0.292, loss_ctc=55.620, loss_att=54.630, acc=0.732, loss=54.927, backward_time=0.403, grad_norm=29.133, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.092, optim0_lr0=3.057e-04, train_time=1.727
+[gpub078:0/16] 2024-01-26 03:57:16,773 (trainer:737) INFO: 11epoch:train:10601-10700batch: iter_time=9.613e-05, forward_time=0.463, loss_ctc=52.514, loss_att=54.136, acc=0.710, loss=53.649, backward_time=0.456, grad_norm=29.087, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.106, optim0_lr0=3.056e-04, train_time=1.923
+[gpub078:0/16] 2024-01-26 03:59:45,263 (trainer:737) INFO: 11epoch:train:10701-10800batch: iter_time=1.157e-04, forward_time=0.289, loss_ctc=51.562, loss_att=48.929, acc=0.709, loss=49.719, backward_time=0.402, grad_norm=29.927, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.055e-04, train_time=1.485
+[gpub078:0/16] 2024-01-26 04:02:35,967 (trainer:737) INFO: 11epoch:train:10801-10900batch: iter_time=1.145e-04, forward_time=0.292, loss_ctc=59.433, loss_att=61.093, acc=0.720, loss=60.595, backward_time=0.404, grad_norm=31.037, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.054e-04, train_time=1.707
+[gpub078:0/16] 2024-01-26 04:05:39,542 (trainer:737) INFO: 11epoch:train:10901-11000batch: iter_time=1.009e-04, forward_time=0.430, loss_ctc=62.158, loss_att=57.542, acc=0.726, loss=58.927, backward_time=0.442, grad_norm=30.836, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.109, optim0_lr0=3.053e-04, train_time=1.835
+[gpub078:0/16] 2024-01-26 04:08:05,093 (trainer:737) INFO: 11epoch:train:11001-11100batch: iter_time=1.034e-04, forward_time=0.290, loss_ctc=57.336, loss_att=57.630, acc=0.703, loss=57.542, backward_time=0.403, grad_norm=31.775, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.052e-04, train_time=1.455
+[gpub078:0/16] 2024-01-26 04:10:51,626 (trainer:737) INFO: 11epoch:train:11101-11200batch: iter_time=1.019e-04, forward_time=0.289, loss_ctc=56.156, loss_att=51.479, acc=0.723, loss=52.882, backward_time=0.401, grad_norm=29.889, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.051e-04, train_time=1.665
+[gpub078:0/16] 2024-01-26 04:12:50,422 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub078:0/16] 2024-01-26 04:13:09,259 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 04:13:12,883 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99685b8880>)
+[gpub078:0/16] 2024-01-26 04:13:12,883 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub078:0/16] 2024-01-26 04:13:12,908 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 04:20:45,921 (trainer:737) INFO: 11epoch:train:11201-11300batch: iter_time=4.074, forward_time=0.396, loss_ctc=54.981, loss_att=51.566, acc=0.718, loss=52.590, backward_time=0.563, grad_norm=30.201, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.107, optim0_lr0=3.050e-04, train_time=5.943
+[gpub078:0/16] 2024-01-26 04:23:26,871 (trainer:737) INFO: 11epoch:train:11301-11400batch: iter_time=1.136e-04, forward_time=0.292, loss_ctc=58.254, loss_att=59.869, acc=0.699, loss=59.384, backward_time=0.403, grad_norm=31.111, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.049e-04, train_time=1.610
+[gpub078:0/16] 2024-01-26 04:26:54,146 (trainer:737) INFO: 11epoch:train:11401-11500batch: iter_time=8.994e-05, forward_time=0.381, loss_ctc=58.221, loss_att=58.789, acc=0.700, loss=58.618, backward_time=0.597, grad_norm=44.655, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.102, optim0_lr0=3.048e-04, train_time=2.072
+[gpub078:0/16] 2024-01-26 04:29:13,590 (trainer:737) INFO: 11epoch:train:11501-11600batch: iter_time=8.390e-05, forward_time=0.294, loss_ctc=60.804, loss_att=57.331, acc=0.703, loss=58.373, backward_time=0.405, grad_norm=31.671, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.047e-04, train_time=1.394
+[gpub078:0/16] 2024-01-26 04:31:49,570 (trainer:737) INFO: 11epoch:train:11601-11700batch: iter_time=8.259e-05, forward_time=0.291, loss_ctc=53.755, loss_att=53.111, acc=0.716, loss=53.304, backward_time=0.404, grad_norm=28.279, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.046e-04, train_time=1.560
+[gpub078:0/16] 2024-01-26 04:35:05,943 (trainer:737) INFO: 11epoch:train:11701-11800batch: iter_time=8.797e-05, forward_time=0.445, loss_ctc=53.524, loss_att=52.704, acc=0.733, loss=52.950, backward_time=0.448, grad_norm=26.358, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.099, optim0_lr0=3.045e-04, train_time=1.963
+[gpub078:0/16] 2024-01-26 04:37:20,932 (trainer:737) INFO: 11epoch:train:11801-11900batch: iter_time=8.405e-05, forward_time=0.291, loss_ctc=52.202, loss_att=53.942, acc=0.721, loss=53.420, backward_time=0.404, grad_norm=28.868, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.093, optim0_lr0=3.044e-04, train_time=1.350
+[gpub078:0/16] 2024-01-26 04:40:34,864 (trainer:737) INFO: 11epoch:train:11901-12000batch: iter_time=0.004, forward_time=0.468, loss_ctc=48.682, loss_att=50.291, acc=0.708, loss=49.808, backward_time=0.428, grad_norm=28.748, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.104, optim0_lr0=3.043e-04, train_time=1.937
+[gpub078:0/16] 2024-01-26 04:43:22,267 (trainer:737) INFO: 11epoch:train:12001-12100batch: iter_time=8.957e-05, forward_time=0.292, loss_ctc=60.967, loss_att=61.901, acc=0.716, loss=61.621, backward_time=0.405, grad_norm=30.037, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.042e-04, train_time=1.674
+[gpub078:0/16] 2024-01-26 04:45:38,041 (trainer:737) INFO: 11epoch:train:12101-12200batch: iter_time=8.445e-05, forward_time=0.291, loss_ctc=58.304, loss_att=51.618, acc=0.733, loss=53.624, backward_time=0.404, grad_norm=31.813, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.041e-04, train_time=1.358
+[gpub078:0/16] 2024-01-26 04:48:57,810 (trainer:737) INFO: 11epoch:train:12201-12300batch: iter_time=9.048e-05, forward_time=0.481, loss_ctc=57.871, loss_att=57.169, acc=0.707, loss=57.379, backward_time=0.438, grad_norm=31.113, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.100, optim0_lr0=3.041e-04, train_time=1.997
+[gpub078:0/16] 2024-01-26 04:51:42,585 (trainer:737) INFO: 11epoch:train:12301-12400batch: iter_time=8.880e-05, forward_time=0.291, loss_ctc=57.979, loss_att=55.383, acc=0.722, loss=56.161, backward_time=0.403, grad_norm=30.513, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.040e-04, train_time=1.648
+[gpub078:0/16] 2024-01-26 04:54:25,571 (trainer:737) INFO: 11epoch:train:12401-12500batch: iter_time=8.603e-05, forward_time=0.414, loss_ctc=51.946, loss_att=53.606, acc=0.698, loss=53.108, backward_time=0.437, grad_norm=29.724, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.109, optim0_lr0=3.039e-04, train_time=1.630
+[gpub078:0/16] 2024-01-26 04:54:45,599 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub078:0/16] 2024-01-26 04:55:04,892 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 04:55:08,445 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9947bf7a00>)
+[gpub078:0/16] 2024-01-26 04:55:08,445 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub078:0/16] 2024-01-26 04:55:08,449 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 05:07:49,522 (trainer:737) INFO: 11epoch:train:12501-12600batch: iter_time=3.838, forward_time=0.296, loss_ctc=60.396, loss_att=55.045, acc=0.713, loss=56.651, backward_time=0.403, grad_norm=30.854, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.038e-04, train_time=8.039
+[gpub078:0/16] 2024-01-26 05:10:52,609 (trainer:737) INFO: 11epoch:train:12601-12700batch: iter_time=8.000e-05, forward_time=0.335, loss_ctc=62.120, loss_att=72.446, acc=0.663, loss=69.349, backward_time=0.424, grad_norm=39.670, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.098, optim0_lr0=3.037e-04, train_time=1.831
+[gpub078:0/16] 2024-01-26 05:13:45,934 (trainer:737) INFO: 11epoch:train:12701-12800batch: iter_time=8.376e-05, forward_time=0.397, loss_ctc=54.386, loss_att=50.516, acc=0.709, loss=51.677, backward_time=0.424, grad_norm=31.440, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.097, optim0_lr0=3.036e-04, train_time=1.733
+[gpub078:0/16] 2024-01-26 05:16:33,420 (trainer:737) INFO: 11epoch:train:12801-12900batch: iter_time=8.769e-05, forward_time=0.292, loss_ctc=55.657, loss_att=55.897, acc=0.714, loss=55.825, backward_time=0.403, grad_norm=32.036, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.035e-04, train_time=1.674
+[gpub078:0/16] 2024-01-26 05:19:13,738 (trainer:737) INFO: 11epoch:train:12901-13000batch: iter_time=9.309e-05, forward_time=0.343, loss_ctc=49.526, loss_att=51.684, acc=0.704, loss=51.037, backward_time=0.432, grad_norm=27.943, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.099, optim0_lr0=3.034e-04, train_time=1.603
+[gpub078:0/16] 2024-01-26 05:22:14,153 (trainer:737) INFO: 11epoch:train:13001-13100batch: iter_time=9.512e-05, forward_time=0.360, loss_ctc=55.198, loss_att=51.755, acc=0.726, loss=52.788, backward_time=0.408, grad_norm=30.590, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.097, optim0_lr0=3.033e-04, train_time=1.804
+[gpub078:0/16] 2024-01-26 05:24:40,383 (trainer:737) INFO: 11epoch:train:13101-13200batch: iter_time=9.680e-05, forward_time=0.332, loss_ctc=51.946, loss_att=54.443, acc=0.708, loss=53.694, backward_time=0.418, grad_norm=28.905, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.106, optim0_lr0=3.032e-04, train_time=1.462
+[gpub078:0/16] 2024-01-26 05:27:36,125 (trainer:737) INFO: 11epoch:train:13201-13300batch: iter_time=9.444e-05, forward_time=0.359, loss_ctc=50.640, loss_att=48.869, acc=0.703, loss=49.400, backward_time=0.409, grad_norm=29.031, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.102, optim0_lr0=3.031e-04, train_time=1.758
+[gpub078:0/16] 2024-01-26 05:30:25,569 (trainer:737) INFO: 11epoch:train:13301-13400batch: iter_time=9.187e-05, forward_time=0.359, loss_ctc=59.076, loss_att=59.274, acc=0.720, loss=59.214, backward_time=0.419, grad_norm=32.819, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.094, optim0_lr0=3.030e-04, train_time=1.694
+[gpub078:0/16] 2024-01-26 05:33:20,907 (trainer:737) INFO: 11epoch:train:13401-13500batch: iter_time=9.541e-05, forward_time=0.352, loss_ctc=61.486, loss_att=57.644, acc=0.720, loss=58.796, backward_time=0.431, grad_norm=33.415, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.102, optim0_lr0=3.029e-04, train_time=1.753
+[gpub078:0/16] 2024-01-26 05:35:55,690 (trainer:737) INFO: 11epoch:train:13501-13600batch: iter_time=9.726e-05, forward_time=0.290, loss_ctc=57.428, loss_att=56.598, acc=0.700, loss=56.847, backward_time=0.402, grad_norm=31.307, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.028e-04, train_time=1.548
+[gpub078:0/16] 2024-01-26 05:38:55,703 (trainer:737) INFO: 11epoch:train:13601-13700batch: iter_time=9.121e-05, forward_time=0.422, loss_ctc=56.426, loss_att=52.236, acc=0.710, loss=53.493, backward_time=0.451, grad_norm=34.049, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.100, optim0_lr0=3.028e-04, train_time=1.800
+[gpub078:0/16] 2024-01-26 05:40:38,332 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub078:0/16] 2024-01-26 05:40:57,726 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 05:41:01,264 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f994bcf4e20>)
+[gpub078:0/16] 2024-01-26 05:41:01,265 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub078:0/16] 2024-01-26 05:41:01,301 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 05:47:48,367 (trainer:737) INFO: 11epoch:train:13701-13800batch: iter_time=3.641, forward_time=0.291, loss_ctc=54.890, loss_att=52.731, acc=0.696, loss=53.379, backward_time=0.401, grad_norm=31.493, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.027e-04, train_time=5.327
+[gpub078:0/16] 2024-01-26 05:50:22,508 (trainer:737) INFO: 11epoch:train:13801-13900batch: iter_time=8.476e-05, forward_time=0.386, loss_ctc=58.152, loss_att=59.986, acc=0.690, loss=59.436, backward_time=0.481, grad_norm=32.497, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.099, optim0_lr0=3.026e-04, train_time=1.541
+[gpub078:0/16] 2024-01-26 05:53:10,562 (trainer:737) INFO: 11epoch:train:13901-14000batch: iter_time=8.257e-05, forward_time=0.288, loss_ctc=56.755, loss_att=58.343, acc=0.696, loss=57.867, backward_time=0.399, grad_norm=39.840, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.093, optim0_lr0=3.025e-04, train_time=1.681
+[gpub078:0/16] 2024-01-26 05:56:06,774 (trainer:737) INFO: 11epoch:train:14001-14100batch: iter_time=8.452e-05, forward_time=0.397, loss_ctc=60.091, loss_att=54.858, acc=0.700, loss=56.428, backward_time=0.419, grad_norm=31.977, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.119, optim0_lr0=3.024e-04, train_time=1.761
+[gpub078:0/16] 2024-01-26 05:58:22,354 (trainer:737) INFO: 11epoch:train:14101-14200batch: iter_time=8.658e-05, forward_time=0.290, loss_ctc=53.252, loss_att=51.976, acc=0.715, loss=52.359, backward_time=0.404, grad_norm=29.589, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.093, optim0_lr0=3.023e-04, train_time=1.356
+[gpub078:0/16] 2024-01-26 06:01:37,732 (trainer:737) INFO: 11epoch:train:14201-14300batch: iter_time=8.579e-05, forward_time=0.411, loss_ctc=52.903, loss_att=53.059, acc=0.718, loss=53.012, backward_time=0.427, grad_norm=27.506, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.118, optim0_lr0=3.022e-04, train_time=1.953
+[gpub078:0/16] 2024-01-26 06:04:10,568 (trainer:737) INFO: 11epoch:train:14301-14400batch: iter_time=9.395e-05, forward_time=0.289, loss_ctc=51.863, loss_att=51.707, acc=0.715, loss=51.754, backward_time=0.402, grad_norm=31.232, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.095, optim0_lr0=3.021e-04, train_time=1.528
+[gpub078:0/16] 2024-01-26 06:07:03,368 (trainer:737) INFO: 11epoch:train:14401-14500batch: iter_time=9.590e-05, forward_time=0.369, loss_ctc=48.776, loss_att=49.622, acc=0.709, loss=49.368, backward_time=0.471, grad_norm=28.401, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.100, optim0_lr0=3.020e-04, train_time=1.729
+[gpub078:0/16] 2024-01-26 06:09:53,274 (trainer:737) INFO: 11epoch:train:14501-14600batch: iter_time=8.917e-05, forward_time=0.351, loss_ctc=60.410, loss_att=62.151, acc=0.710, loss=61.629, backward_time=0.512, grad_norm=32.128, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.102, optim0_lr0=3.019e-04, train_time=1.698
+[gpub078:0/16] 2024-01-26 06:12:28,519 (trainer:737) INFO: 11epoch:train:14601-14700batch: iter_time=9.019e-05, forward_time=0.292, loss_ctc=58.820, loss_att=50.543, acc=0.731, loss=53.026, backward_time=0.409, grad_norm=32.236, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.093, optim0_lr0=3.018e-04, train_time=1.553
+[gpub078:0/16] 2024-01-26 06:15:08,951 (trainer:737) INFO: 11epoch:train:14701-14800batch: iter_time=9.367e-05, forward_time=0.288, loss_ctc=57.923, loss_att=57.201, acc=0.703, loss=57.418, backward_time=0.401, grad_norm=33.675, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.106, optim0_lr0=3.017e-04, train_time=1.599
+[gpub078:0/16] 2024-01-26 06:17:58,782 (trainer:737) INFO: 11epoch:train:14801-14900batch: iter_time=9.469e-05, forward_time=0.406, loss_ctc=58.451, loss_att=53.947, acc=0.712, loss=55.298, backward_time=0.439, grad_norm=31.856, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.095, optim0_lr0=3.016e-04, train_time=1.703
+[gpub078:0/16] 2024-01-26 06:20:35,043 (trainer:737) INFO: 11epoch:train:14901-15000batch: iter_time=8.957e-05, forward_time=0.289, loss_ctc=52.117, loss_att=53.757, acc=0.681, loss=53.265, backward_time=0.413, grad_norm=33.013, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.098, optim0_lr0=3.016e-04, train_time=1.562
+[gpub078:0/16] 2024-01-26 06:59:04,165 (trainer:343) INFO: 11epoch results: [train] iter_time=0.350, forward_time=0.349, loss_ctc=57.415, loss_att=56.047, acc=0.706, loss=56.457, backward_time=0.426, grad_norm=31.927, clip=100.000, loss_scale=6.305e+28, optim_step_time=0.099, optim0_lr0=3.087e-04, train_time=2.064, time=8 hours, 36 minutes and 33.3 seconds, total_count=195000, gpu_max_cached_mem_GB=42.420, [valid] loss_ctc=48.419, cer_ctc=0.247, loss_att=46.736, acc=0.624, cer=0.373, wer=0.997, loss=47.241, time=38 minutes and 2.32 seconds, total_count=60723, gpu_max_cached_mem_GB=42.420
+[gpub078:0/16] 2024-01-26 06:59:14,438 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub078:0/16] 2024-01-26 06:59:14,508 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/6epoch.pth
+[gpub078:0/16] 2024-01-26 06:59:14,508 (trainer:272) INFO: 12/45epoch started. Estimated time to finish: 1 week, 6 days and 3 hours
+[gpub078:0/16] 2024-01-26 06:59:14,517 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-26 06:59:32,708 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 06:59:36,245 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9233f90eb0>)
+[gpub078:0/16] 2024-01-26 06:59:36,245 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-26 06:59:36,249 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 07:07:12,416 (trainer:737) INFO: 12epoch:train:1-100batch: iter_time=3.204, forward_time=0.299, loss_ctc=53.331, loss_att=61.368, acc=0.708, loss=58.957, backward_time=0.413, grad_norm=28.168, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.095, optim0_lr0=3.015e-04, train_time=4.779
+[gpub078:0/16] 2024-01-26 07:09:45,820 (trainer:737) INFO: 12epoch:train:101-200batch: iter_time=8.544e-05, forward_time=0.329, loss_ctc=55.144, loss_att=51.336, acc=0.723, loss=52.478, backward_time=0.409, grad_norm=31.208, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.094, optim0_lr0=3.014e-04, train_time=1.534
+[gpub078:0/16] 2024-01-26 07:12:32,802 (trainer:737) INFO: 12epoch:train:201-300batch: iter_time=9.399e-05, forward_time=0.331, loss_ctc=51.746, loss_att=48.490, acc=0.715, loss=49.467, backward_time=0.401, grad_norm=30.767, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.094, optim0_lr0=3.013e-04, train_time=1.669
+[gpub078:0/16] 2024-01-26 07:14:56,754 (trainer:737) INFO: 12epoch:train:301-400batch: iter_time=1.733e-04, forward_time=0.298, loss_ctc=67.569, loss_att=52.145, acc=0.728, loss=56.772, backward_time=0.426, grad_norm=37.904, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.098, optim0_lr0=3.012e-04, train_time=1.440
+[gpub078:0/16] 2024-01-26 07:17:27,377 (trainer:737) INFO: 12epoch:train:401-500batch: iter_time=9.153e-05, forward_time=0.313, loss_ctc=55.223, loss_att=54.649, acc=0.688, loss=54.821, backward_time=0.404, grad_norm=34.450, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.095, optim0_lr0=3.011e-04, train_time=1.506
+[gpub078:0/16] 2024-01-26 07:19:54,470 (trainer:737) INFO: 12epoch:train:501-600batch: iter_time=8.570e-05, forward_time=0.317, loss_ctc=59.771, loss_att=57.725, acc=0.701, loss=58.339, backward_time=0.404, grad_norm=34.496, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.093, optim0_lr0=3.010e-04, train_time=1.471
+[gpub078:0/16] 2024-01-26 07:22:26,476 (trainer:737) INFO: 12epoch:train:601-700batch: iter_time=9.706e-05, forward_time=0.324, loss_ctc=62.374, loss_att=57.657, acc=0.711, loss=59.072, backward_time=0.409, grad_norm=32.825, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.093, optim0_lr0=3.009e-04, train_time=1.520
+[gpub078:0/16] 2024-01-26 07:25:12,140 (trainer:737) INFO: 12epoch:train:701-800batch: iter_time=9.615e-05, forward_time=0.328, loss_ctc=58.983, loss_att=55.575, acc=0.714, loss=56.597, backward_time=0.438, grad_norm=30.274, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.107, optim0_lr0=3.008e-04, train_time=1.657
+[gpub078:0/16] 2024-01-26 07:27:38,914 (trainer:737) INFO: 12epoch:train:801-900batch: iter_time=1.020e-04, forward_time=0.313, loss_ctc=58.673, loss_att=54.602, acc=0.722, loss=55.823, backward_time=0.409, grad_norm=31.548, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.095, optim0_lr0=3.007e-04, train_time=1.468
+[gpub078:0/16] 2024-01-26 07:30:13,652 (trainer:737) INFO: 12epoch:train:901-1000batch: iter_time=1.052e-04, forward_time=0.307, loss_ctc=66.442, loss_att=62.626, acc=0.702, loss=63.771, backward_time=0.410, grad_norm=38.467, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.093, optim0_lr0=3.006e-04, train_time=1.547
+[gpub078:0/16] 2024-01-26 07:32:55,792 (trainer:737) INFO: 12epoch:train:1001-1100batch: iter_time=1.082e-04, forward_time=0.340, loss_ctc=51.416, loss_att=52.654, acc=0.719, loss=52.283, backward_time=0.413, grad_norm=30.217, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.094, optim0_lr0=3.006e-04, train_time=1.620
+[gpub078:0/16] 2024-01-26 07:35:23,846 (trainer:737) INFO: 12epoch:train:1101-1200batch: iter_time=0.001, forward_time=0.315, loss_ctc=69.585, loss_att=55.490, acc=0.737, loss=59.719, backward_time=0.446, grad_norm=37.330, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.103, optim0_lr0=3.005e-04, train_time=1.481
+[gpub078:0/16] 2024-01-26 07:37:14,162 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub078:0/16] 2024-01-26 07:37:33,386 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 07:37:37,044 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f99694805b0>)
+[gpub078:0/16] 2024-01-26 07:37:37,044 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub078:0/16] 2024-01-26 07:37:37,047 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 07:44:10,499 (trainer:737) INFO: 12epoch:train:1201-1300batch: iter_time=3.536, forward_time=0.333, loss_ctc=52.175, loss_att=53.217, acc=0.716, loss=52.905, backward_time=0.402, grad_norm=28.810, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=3.004e-04, train_time=5.266
+[gpub078:0/16] 2024-01-26 07:46:44,568 (trainer:737) INFO: 12epoch:train:1301-1400batch: iter_time=8.194e-05, forward_time=0.322, loss_ctc=52.014, loss_att=51.849, acc=0.722, loss=51.899, backward_time=0.418, grad_norm=29.443, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.097, optim0_lr0=3.003e-04, train_time=1.540
+[gpub078:0/16] 2024-01-26 07:49:18,770 (trainer:737) INFO: 12epoch:train:1401-1500batch: iter_time=8.174e-05, forward_time=0.332, loss_ctc=53.890, loss_att=50.906, acc=0.710, loss=51.801, backward_time=0.431, grad_norm=30.600, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.100, optim0_lr0=3.002e-04, train_time=1.542
+[gpub078:0/16] 2024-01-26 07:51:58,601 (trainer:737) INFO: 12epoch:train:1501-1600batch: iter_time=8.795e-05, forward_time=0.294, loss_ctc=60.646, loss_att=56.766, acc=0.709, loss=57.930, backward_time=0.401, grad_norm=33.051, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=3.001e-04, train_time=1.598
+[gpub078:0/16] 2024-01-26 07:54:32,513 (trainer:737) INFO: 12epoch:train:1601-1700batch: iter_time=8.716e-05, forward_time=0.318, loss_ctc=61.928, loss_att=49.732, acc=0.699, loss=53.391, backward_time=0.411, grad_norm=38.976, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.095, optim0_lr0=3.000e-04, train_time=1.539
+[gpub078:0/16] 2024-01-26 07:57:06,316 (trainer:737) INFO: 12epoch:train:1701-1800batch: iter_time=8.892e-05, forward_time=0.345, loss_ctc=48.180, loss_att=51.707, acc=0.693, loss=50.649, backward_time=0.414, grad_norm=35.255, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.094, optim0_lr0=2.999e-04, train_time=1.538
+[gpub078:0/16] 2024-01-26 07:59:54,796 (trainer:737) INFO: 12epoch:train:1801-1900batch: iter_time=9.202e-05, forward_time=0.326, loss_ctc=59.552, loss_att=56.837, acc=0.712, loss=57.651, backward_time=0.435, grad_norm=33.585, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.098, optim0_lr0=2.998e-04, train_time=1.684
+[gpub078:0/16] 2024-01-26 08:02:21,821 (trainer:737) INFO: 12epoch:train:1901-2000batch: iter_time=9.336e-05, forward_time=0.338, loss_ctc=67.114, loss_att=58.869, acc=0.701, loss=61.342, backward_time=0.409, grad_norm=33.932, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.997e-04, train_time=1.471
+[gpub078:0/16] 2024-01-26 08:04:50,977 (trainer:737) INFO: 12epoch:train:2001-2100batch: iter_time=8.860e-05, forward_time=0.289, loss_ctc=53.564, loss_att=54.566, acc=0.709, loss=54.266, backward_time=0.401, grad_norm=29.240, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.997e-04, train_time=1.491
+[gpub078:0/16] 2024-01-26 08:07:24,441 (trainer:737) INFO: 12epoch:train:2101-2200batch: iter_time=9.242e-05, forward_time=0.332, loss_ctc=68.483, loss_att=57.937, acc=0.709, loss=61.100, backward_time=0.433, grad_norm=42.573, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.095, optim0_lr0=2.996e-04, train_time=1.534
+[gpub078:0/16] 2024-01-26 08:10:11,706 (trainer:737) INFO: 12epoch:train:2201-2300batch: iter_time=8.536e-05, forward_time=0.335, loss_ctc=51.721, loss_att=56.900, acc=0.696, loss=55.347, backward_time=0.443, grad_norm=32.307, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.101, optim0_lr0=2.995e-04, train_time=1.672
+[gpub078:0/16] 2024-01-26 08:12:29,917 (trainer:737) INFO: 12epoch:train:2301-2400batch: iter_time=8.626e-05, forward_time=0.293, loss_ctc=54.703, loss_att=44.278, acc=0.733, loss=47.405, backward_time=0.401, grad_norm=29.755, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.994e-04, train_time=1.382
+[gpub078:0/16] 2024-01-26 08:15:00,368 (trainer:737) INFO: 12epoch:train:2401-2500batch: iter_time=9.040e-05, forward_time=0.288, loss_ctc=62.992, loss_att=53.488, acc=0.724, loss=56.339, backward_time=0.399, grad_norm=39.986, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.993e-04, train_time=1.504
+[gpub078:0/16] 2024-01-26 08:15:20,396 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub078:0/16] 2024-01-26 08:15:39,597 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 08:15:43,342 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f922573c3d0>)
+[gpub078:0/16] 2024-01-26 08:15:43,342 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub078:0/16] 2024-01-26 08:15:43,347 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 08:24:02,327 (trainer:737) INFO: 12epoch:train:2501-2600batch: iter_time=3.635, forward_time=0.335, loss_ctc=52.104, loss_att=60.390, acc=0.714, loss=57.904, backward_time=0.413, grad_norm=28.822, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.095, optim0_lr0=2.992e-04, train_time=5.419
+[gpub078:0/16] 2024-01-26 08:27:11,468 (trainer:737) INFO: 12epoch:train:2601-2700batch: iter_time=9.007e-05, forward_time=0.344, loss_ctc=53.457, loss_att=50.422, acc=0.729, loss=51.332, backward_time=0.404, grad_norm=29.572, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.094, optim0_lr0=2.991e-04, train_time=1.891
+[gpub078:0/16] 2024-01-26 08:30:27,032 (trainer:737) INFO: 12epoch:train:2701-2800batch: iter_time=8.530e-05, forward_time=0.287, loss_ctc=51.157, loss_att=47.934, acc=0.721, loss=48.901, backward_time=0.396, grad_norm=30.145, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.990e-04, train_time=1.956
+[gpub078:0/16] 2024-01-26 08:33:35,434 (trainer:737) INFO: 12epoch:train:2801-2900batch: iter_time=8.674e-05, forward_time=0.327, loss_ctc=63.967, loss_att=51.493, acc=0.731, loss=55.235, backward_time=0.447, grad_norm=35.531, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.100, optim0_lr0=2.989e-04, train_time=1.884
+[gpub078:0/16] 2024-01-26 08:36:38,590 (trainer:737) INFO: 12epoch:train:2901-3000batch: iter_time=8.864e-05, forward_time=0.292, loss_ctc=53.944, loss_att=53.675, acc=0.694, loss=53.756, backward_time=0.405, grad_norm=33.176, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.093, optim0_lr0=2.989e-04, train_time=1.831
+[gpub078:0/16] 2024-01-26 08:39:55,595 (trainer:737) INFO: 12epoch:train:3001-3100batch: iter_time=8.864e-05, forward_time=0.323, loss_ctc=57.814, loss_att=55.814, acc=0.710, loss=56.414, backward_time=0.406, grad_norm=42.540, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.988e-04, train_time=1.969
+[gpub078:0/16] 2024-01-26 08:43:01,442 (trainer:737) INFO: 12epoch:train:3101-3200batch: iter_time=9.445e-05, forward_time=0.391, loss_ctc=59.001, loss_att=56.546, acc=0.718, loss=57.282, backward_time=0.430, grad_norm=31.786, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.107, optim0_lr0=2.987e-04, train_time=1.859
+[gpub078:0/16] 2024-01-26 08:46:02,044 (trainer:737) INFO: 12epoch:train:3201-3300batch: iter_time=9.027e-05, forward_time=0.290, loss_ctc=57.871, loss_att=55.121, acc=0.717, loss=55.946, backward_time=0.403, grad_norm=30.027, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.986e-04, train_time=1.806
+[gpub078:0/16] 2024-01-26 08:49:10,620 (trainer:737) INFO: 12epoch:train:3301-3400batch: iter_time=8.732e-05, forward_time=0.295, loss_ctc=56.818, loss_att=53.598, acc=0.726, loss=54.564, backward_time=0.404, grad_norm=30.741, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.985e-04, train_time=1.886
+[gpub078:0/16] 2024-01-26 08:51:41,431 (trainer:737) INFO: 12epoch:train:3401-3500batch: iter_time=8.495e-05, forward_time=0.330, loss_ctc=63.421, loss_att=62.215, acc=0.707, loss=62.577, backward_time=0.404, grad_norm=42.191, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.094, optim0_lr0=2.984e-04, train_time=1.507
+[gpub078:0/16] 2024-01-26 08:55:02,493 (trainer:737) INFO: 12epoch:train:3501-3600batch: iter_time=7.896e-04, forward_time=0.362, loss_ctc=50.447, loss_att=52.044, acc=0.723, loss=51.565, backward_time=0.438, grad_norm=28.875, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.098, optim0_lr0=2.983e-04, train_time=2.011
+[gpub078:0/16] 2024-01-26 08:57:52,879 (trainer:737) INFO: 12epoch:train:3601-3700batch: iter_time=8.106e-05, forward_time=0.290, loss_ctc=66.681, loss_att=53.440, acc=0.740, loss=57.412, backward_time=0.402, grad_norm=38.346, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.092, optim0_lr0=2.982e-04, train_time=1.704
+[gpub078:0/16] 2024-01-26 08:59:20,950 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub078:0/16] 2024-01-26 08:59:40,157 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 08:59:44,078 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b5e5e28f0>)
+[gpub078:0/16] 2024-01-26 08:59:44,078 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub078:0/16] 2024-01-26 08:59:44,081 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 09:06:34,108 (trainer:737) INFO: 12epoch:train:3701-3800batch: iter_time=3.738, forward_time=0.344, loss_ctc=51.665, loss_att=50.962, acc=0.728, loss=51.172, backward_time=0.406, grad_norm=28.314, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.981e-04, train_time=5.212
+[gpub078:0/16] 2024-01-26 09:09:47,645 (trainer:737) INFO: 12epoch:train:3801-3900batch: iter_time=8.370e-05, forward_time=0.347, loss_ctc=51.359, loss_att=53.431, acc=0.723, loss=52.809, backward_time=0.457, grad_norm=29.814, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.102, optim0_lr0=2.981e-04, train_time=1.935
+[gpub078:0/16] 2024-01-26 09:12:09,677 (trainer:737) INFO: 12epoch:train:3901-4000batch: iter_time=7.762e-05, forward_time=0.290, loss_ctc=52.863, loss_att=49.864, acc=0.722, loss=50.763, backward_time=0.402, grad_norm=30.121, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.980e-04, train_time=1.420
+[gpub078:0/16] 2024-01-26 09:14:58,047 (trainer:737) INFO: 12epoch:train:4001-4100batch: iter_time=8.426e-05, forward_time=0.297, loss_ctc=60.532, loss_att=55.234, acc=0.724, loss=56.824, backward_time=0.407, grad_norm=33.958, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.095, optim0_lr0=2.979e-04, train_time=1.683
+[gpub078:0/16] 2024-01-26 09:17:42,913 (trainer:737) INFO: 12epoch:train:4101-4200batch: iter_time=8.001e-05, forward_time=0.334, loss_ctc=60.553, loss_att=48.083, acc=0.718, loss=51.824, backward_time=0.401, grad_norm=37.653, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.094, optim0_lr0=2.978e-04, train_time=1.648
+[gpub078:0/16] 2024-01-26 09:20:19,581 (trainer:737) INFO: 12epoch:train:4201-4300batch: iter_time=8.478e-05, forward_time=0.405, loss_ctc=47.207, loss_att=50.945, acc=0.701, loss=49.824, backward_time=0.420, grad_norm=28.948, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.096, optim0_lr0=2.977e-04, train_time=1.566
+[gpub078:0/16] 2024-01-26 09:23:00,380 (trainer:737) INFO: 12epoch:train:4301-4400batch: iter_time=9.102e-05, forward_time=0.291, loss_ctc=58.193, loss_att=55.906, acc=0.723, loss=56.592, backward_time=0.401, grad_norm=31.519, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.976e-04, train_time=1.608
+[gpub078:0/16] 2024-01-26 09:25:51,402 (trainer:737) INFO: 12epoch:train:4401-4500batch: iter_time=8.673e-05, forward_time=0.344, loss_ctc=66.091, loss_att=57.195, acc=0.720, loss=59.864, backward_time=0.415, grad_norm=35.351, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.094, optim0_lr0=2.975e-04, train_time=1.710
+[gpub078:0/16] 2024-01-26 09:28:29,343 (trainer:737) INFO: 12epoch:train:4501-4600batch: iter_time=8.755e-05, forward_time=0.359, loss_ctc=53.445, loss_att=54.831, acc=0.722, loss=54.415, backward_time=0.427, grad_norm=29.155, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.105, optim0_lr0=2.974e-04, train_time=1.579
+[gpub078:0/16] 2024-01-26 09:31:09,434 (trainer:737) INFO: 12epoch:train:4601-4700batch: iter_time=2.753e-04, forward_time=0.299, loss_ctc=65.011, loss_att=57.179, acc=0.715, loss=59.529, backward_time=0.402, grad_norm=40.094, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.973e-04, train_time=1.600
+[gpub078:0/16] 2024-01-26 09:33:56,906 (trainer:737) INFO: 12epoch:train:4701-4800batch: iter_time=9.264e-05, forward_time=0.298, loss_ctc=51.113, loss_att=56.685, acc=0.713, loss=55.014, backward_time=0.411, grad_norm=30.816, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.094, optim0_lr0=2.973e-04, train_time=1.676
+[gpub078:0/16] 2024-01-26 09:36:23,013 (trainer:737) INFO: 12epoch:train:4801-4900batch: iter_time=9.800e-05, forward_time=0.315, loss_ctc=54.174, loss_att=44.172, acc=0.742, loss=47.173, backward_time=0.414, grad_norm=30.436, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.093, optim0_lr0=2.972e-04, train_time=1.461
+[gpub078:0/16] 2024-01-26 09:38:50,168 (trainer:737) INFO: 12epoch:train:4901-5000batch: iter_time=1.480e-04, forward_time=0.306, loss_ctc=60.832, loss_att=50.799, acc=0.741, loss=53.809, backward_time=0.433, grad_norm=37.057, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.100, optim0_lr0=2.971e-04, train_time=1.471
+[gpub078:0/16] 2024-01-26 09:39:10,197 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub078:0/16] 2024-01-26 09:39:29,247 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 09:39:32,918 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9b5b1387c0>)
+[gpub078:0/16] 2024-01-26 09:39:32,918 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub078:0/16] 2024-01-26 09:39:32,933 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 09:48:03,082 (trainer:737) INFO: 12epoch:train:5001-5100batch: iter_time=3.723, forward_time=0.326, loss_ctc=51.572, loss_att=60.335, acc=0.713, loss=57.706, backward_time=0.406, grad_norm=29.829, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.094, optim0_lr0=2.970e-04, train_time=5.529
+[gpub078:0/16] 2024-01-26 09:50:48,142 (trainer:737) INFO: 12epoch:train:5101-5200batch: iter_time=8.810e-05, forward_time=0.334, loss_ctc=52.868, loss_att=48.984, acc=0.728, loss=50.149, backward_time=0.412, grad_norm=29.699, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.094, optim0_lr0=2.969e-04, train_time=1.650
+[gpub078:0/16] 2024-01-26 09:53:25,089 (trainer:737) INFO: 12epoch:train:5201-5300batch: iter_time=9.242e-05, forward_time=0.288, loss_ctc=51.069, loss_att=48.134, acc=0.714, loss=49.015, backward_time=0.400, grad_norm=28.612, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.968e-04, train_time=1.569
+[gpub078:0/16] 2024-01-26 09:56:17,347 (trainer:737) INFO: 12epoch:train:5301-5400batch: iter_time=1.059e-04, forward_time=0.350, loss_ctc=62.377, loss_att=51.039, acc=0.728, loss=54.440, backward_time=0.470, grad_norm=34.854, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.101, optim0_lr0=2.967e-04, train_time=1.722
+[gpub078:0/16] 2024-01-26 09:58:51,356 (trainer:737) INFO: 12epoch:train:5401-5500batch: iter_time=8.940e-05, forward_time=0.288, loss_ctc=53.540, loss_att=53.571, acc=0.682, loss=53.562, backward_time=0.400, grad_norm=33.609, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.967e-04, train_time=1.540
+[gpub078:0/16] 2024-01-26 10:01:27,005 (trainer:737) INFO: 12epoch:train:5501-5600batch: iter_time=9.129e-05, forward_time=0.359, loss_ctc=56.786, loss_att=56.184, acc=0.702, loss=56.365, backward_time=0.410, grad_norm=33.364, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.094, optim0_lr0=2.966e-04, train_time=1.556
+[gpub078:0/16] 2024-01-26 10:04:19,584 (trainer:737) INFO: 12epoch:train:5601-5700batch: iter_time=9.263e-05, forward_time=0.316, loss_ctc=58.503, loss_att=55.815, acc=0.715, loss=56.621, backward_time=0.409, grad_norm=32.673, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.095, optim0_lr0=2.965e-04, train_time=1.726
+[gpub078:0/16] 2024-01-26 10:06:54,587 (trainer:737) INFO: 12epoch:train:5701-5800batch: iter_time=9.703e-05, forward_time=0.328, loss_ctc=56.814, loss_att=53.968, acc=0.708, loss=54.822, backward_time=0.436, grad_norm=31.983, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.098, optim0_lr0=2.964e-04, train_time=1.550
+[gpub078:0/16] 2024-01-26 10:09:12,295 (trainer:737) INFO: 12epoch:train:5801-5900batch: iter_time=8.748e-05, forward_time=0.302, loss_ctc=55.654, loss_att=53.288, acc=0.719, loss=53.998, backward_time=0.408, grad_norm=32.377, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.963e-04, train_time=1.377
+[gpub078:0/16] 2024-01-26 10:12:14,651 (trainer:737) INFO: 12epoch:train:5901-6000batch: iter_time=9.573e-05, forward_time=0.318, loss_ctc=59.662, loss_att=57.743, acc=0.703, loss=58.319, backward_time=0.410, grad_norm=46.076, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.962e-04, train_time=1.824
+[gpub078:0/16] 2024-01-26 10:14:43,031 (trainer:737) INFO: 12epoch:train:6001-6100batch: iter_time=5.799e-04, forward_time=0.372, loss_ctc=49.517, loss_att=50.676, acc=0.719, loss=50.328, backward_time=0.448, grad_norm=52.637, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.110, optim0_lr0=2.961e-04, train_time=1.482
+[gpub078:0/16] 2024-01-26 10:17:27,679 (trainer:737) INFO: 12epoch:train:6101-6200batch: iter_time=1.025e-04, forward_time=0.291, loss_ctc=66.589, loss_att=54.770, acc=0.737, loss=58.316, backward_time=0.403, grad_norm=38.511, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.960e-04, train_time=1.648
+[gpub078:0/16] 2024-01-26 10:18:55,808 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub078:0/16] 2024-01-26 10:19:14,944 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 10:19:18,550 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9972783490>)
+[gpub078:0/16] 2024-01-26 10:19:18,550 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub078:0/16] 2024-01-26 10:19:18,582 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 10:25:42,540 (trainer:737) INFO: 12epoch:train:6201-6300batch: iter_time=3.350, forward_time=0.333, loss_ctc=51.215, loss_att=52.107, acc=0.718, loss=51.839, backward_time=0.410, grad_norm=29.553, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.960e-04, train_time=4.949
+[gpub078:0/16] 2024-01-26 10:28:12,984 (trainer:737) INFO: 12epoch:train:6301-6400batch: iter_time=8.324e-05, forward_time=0.292, loss_ctc=51.031, loss_att=53.162, acc=0.727, loss=52.522, backward_time=0.406, grad_norm=30.052, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.959e-04, train_time=1.504
+[gpub078:0/16] 2024-01-26 10:31:01,923 (trainer:737) INFO: 12epoch:train:6401-6500batch: iter_time=8.110e-05, forward_time=0.335, loss_ctc=52.869, loss_att=50.188, acc=0.724, loss=50.992, backward_time=0.467, grad_norm=29.024, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.100, optim0_lr0=2.958e-04, train_time=1.688
+[gpub078:0/16] 2024-01-26 10:34:03,997 (trainer:737) INFO: 12epoch:train:6501-6600batch: iter_time=9.642e-05, forward_time=0.312, loss_ctc=60.650, loss_att=55.563, acc=0.725, loss=57.089, backward_time=0.411, grad_norm=35.610, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.095, optim0_lr0=2.957e-04, train_time=1.821
+[gpub078:0/16] 2024-01-26 10:36:19,246 (trainer:737) INFO: 12epoch:train:6601-6700batch: iter_time=9.753e-05, forward_time=0.313, loss_ctc=59.260, loss_att=47.966, acc=0.718, loss=51.354, backward_time=0.405, grad_norm=36.313, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.956e-04, train_time=1.352
+[gpub078:0/16] 2024-01-26 10:39:02,392 (trainer:737) INFO: 12epoch:train:6701-6800batch: iter_time=9.421e-05, forward_time=0.349, loss_ctc=46.493, loss_att=51.190, acc=0.703, loss=49.781, backward_time=0.446, grad_norm=29.567, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.103, optim0_lr0=2.955e-04, train_time=1.631
+[gpub078:0/16] 2024-01-26 10:42:07,288 (trainer:737) INFO: 12epoch:train:6801-6900batch: iter_time=9.424e-05, forward_time=0.296, loss_ctc=58.445, loss_att=56.434, acc=0.725, loss=57.037, backward_time=0.408, grad_norm=31.681, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.093, optim0_lr0=2.954e-04, train_time=1.847
+[gpub078:0/16] 2024-01-26 10:44:41,722 (trainer:737) INFO: 12epoch:train:6901-7000batch: iter_time=1.094e-04, forward_time=0.339, loss_ctc=65.676, loss_att=57.218, acc=0.721, loss=59.755, backward_time=0.407, grad_norm=36.273, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.095, optim0_lr0=2.954e-04, train_time=1.545
+[gpub078:0/16] 2024-01-26 10:47:13,701 (trainer:737) INFO: 12epoch:train:7001-7100batch: iter_time=1.135e-04, forward_time=0.295, loss_ctc=52.947, loss_att=54.890, acc=0.722, loss=54.307, backward_time=0.429, grad_norm=29.560, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.095, optim0_lr0=2.953e-04, train_time=1.521
+[gpub078:0/16] 2024-01-26 10:50:02,386 (trainer:737) INFO: 12epoch:train:7101-7200batch: iter_time=3.471e-04, forward_time=0.326, loss_ctc=63.773, loss_att=56.373, acc=0.721, loss=58.593, backward_time=0.428, grad_norm=43.023, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.096, optim0_lr0=2.952e-04, train_time=1.687
+[gpub078:0/16] 2024-01-26 10:52:45,256 (trainer:737) INFO: 12epoch:train:7201-7300batch: iter_time=9.825e-05, forward_time=0.330, loss_ctc=50.489, loss_att=56.559, acc=0.713, loss=54.738, backward_time=0.412, grad_norm=30.536, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.094, optim0_lr0=2.951e-04, train_time=1.627
+[gpub078:0/16] 2024-01-26 10:55:03,872 (trainer:737) INFO: 12epoch:train:7301-7400batch: iter_time=9.421e-05, forward_time=0.289, loss_ctc=53.630, loss_att=43.898, acc=0.743, loss=46.817, backward_time=0.402, grad_norm=30.678, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.950e-04, train_time=1.387
+[gpub078:0/16] 2024-01-26 10:57:53,625 (trainer:737) INFO: 12epoch:train:7401-7500batch: iter_time=8.586e-05, forward_time=0.366, loss_ctc=60.308, loss_att=51.266, acc=0.741, loss=53.978, backward_time=0.431, grad_norm=35.625, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.100, optim0_lr0=2.949e-04, train_time=1.697
+[gpub078:0/16] 2024-01-26 10:58:13,666 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub078:0/16] 2024-01-26 10:58:32,733 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 10:58:36,348 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9972174d00>)
+[gpub078:0/16] 2024-01-26 10:58:36,348 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub078:0/16] 2024-01-26 10:58:36,352 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 11:07:03,011 (trainer:737) INFO: 12epoch:train:7501-7600batch: iter_time=3.540, forward_time=0.331, loss_ctc=51.261, loss_att=59.570, acc=0.712, loss=57.077, backward_time=0.405, grad_norm=29.533, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.094, optim0_lr0=2.948e-04, train_time=5.494
+[gpub078:0/16] 2024-01-26 11:10:01,166 (trainer:737) INFO: 12epoch:train:7601-7700batch: iter_time=9.282e-05, forward_time=0.288, loss_ctc=51.956, loss_att=48.383, acc=0.729, loss=49.455, backward_time=0.399, grad_norm=29.155, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.948e-04, train_time=1.781
+[gpub078:0/16] 2024-01-26 11:12:53,344 (trainer:737) INFO: 12epoch:train:7701-7800batch: iter_time=8.331e-04, forward_time=0.372, loss_ctc=50.074, loss_att=47.217, acc=0.719, loss=48.074, backward_time=0.425, grad_norm=28.071, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.098, optim0_lr0=2.947e-04, train_time=1.722
+[gpub078:0/16] 2024-01-26 11:15:55,898 (trainer:737) INFO: 12epoch:train:7801-7900batch: iter_time=8.495e-05, forward_time=0.290, loss_ctc=61.538, loss_att=49.756, acc=0.733, loss=53.291, backward_time=0.403, grad_norm=34.681, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.946e-04, train_time=1.825
+[gpub078:0/16] 2024-01-26 11:18:48,072 (trainer:737) INFO: 12epoch:train:7901-8000batch: iter_time=8.532e-05, forward_time=0.331, loss_ctc=53.485, loss_att=53.206, acc=0.685, loss=53.290, backward_time=0.406, grad_norm=36.083, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.094, optim0_lr0=2.945e-04, train_time=1.721
+[gpub078:0/16] 2024-01-26 11:21:50,015 (trainer:737) INFO: 12epoch:train:8001-8100batch: iter_time=7.961e-05, forward_time=0.289, loss_ctc=56.453, loss_att=55.813, acc=0.704, loss=56.005, backward_time=0.402, grad_norm=35.625, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.095, optim0_lr0=2.944e-04, train_time=1.820
+[gpub078:0/16] 2024-01-26 11:24:37,976 (trainer:737) INFO: 12epoch:train:8101-8200batch: iter_time=8.381e-05, forward_time=0.342, loss_ctc=58.678, loss_att=55.900, acc=0.714, loss=56.734, backward_time=0.455, grad_norm=34.044, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.100, optim0_lr0=2.943e-04, train_time=1.679
+[gpub078:0/16] 2024-01-26 11:27:52,862 (trainer:737) INFO: 12epoch:train:8201-8300batch: iter_time=1.097e-04, forward_time=0.361, loss_ctc=56.595, loss_att=53.555, acc=0.709, loss=54.467, backward_time=0.409, grad_norm=30.321, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.095, optim0_lr0=2.942e-04, train_time=1.949
+[gpub078:0/16] 2024-01-26 11:30:43,656 (trainer:737) INFO: 12epoch:train:8301-8400batch: iter_time=8.766e-05, forward_time=0.289, loss_ctc=55.239, loss_att=52.726, acc=0.721, loss=53.480, backward_time=0.399, grad_norm=30.355, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.942e-04, train_time=1.707
+[gpub078:0/16] 2024-01-26 11:33:57,117 (trainer:737) INFO: 12epoch:train:8401-8500batch: iter_time=8.962e-05, forward_time=0.358, loss_ctc=59.327, loss_att=59.053, acc=0.702, loss=59.135, backward_time=0.454, grad_norm=41.825, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.105, optim0_lr0=2.941e-04, train_time=1.935
+[gpub078:0/16] 2024-01-26 11:36:51,792 (trainer:737) INFO: 12epoch:train:8501-8600batch: iter_time=9.830e-05, forward_time=0.330, loss_ctc=48.908, loss_att=50.434, acc=0.718, loss=49.976, backward_time=0.417, grad_norm=27.865, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.095, optim0_lr0=2.940e-04, train_time=1.746
+[gpub078:0/16] 2024-01-26 11:39:54,422 (trainer:737) INFO: 12epoch:train:8601-8700batch: iter_time=9.456e-05, forward_time=0.290, loss_ctc=64.639, loss_att=52.131, acc=0.739, loss=55.884, backward_time=0.401, grad_norm=38.545, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.939e-04, train_time=1.826
+[gpub078:0/16] 2024-01-26 11:41:30,535 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub078:0/16] 2024-01-26 11:41:50,094 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 11:41:53,682 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c097d7430>)
+[gpub078:0/16] 2024-01-26 11:41:53,682 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub078:0/16] 2024-01-26 11:41:53,686 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 11:48:25,273 (trainer:737) INFO: 12epoch:train:8701-8800batch: iter_time=3.563, forward_time=0.345, loss_ctc=50.323, loss_att=51.178, acc=0.720, loss=50.921, backward_time=0.416, grad_norm=28.623, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.095, optim0_lr0=2.938e-04, train_time=5.108
+[gpub078:0/16] 2024-01-26 11:51:24,462 (trainer:737) INFO: 12epoch:train:8801-8900batch: iter_time=8.528e-05, forward_time=0.343, loss_ctc=50.273, loss_att=53.050, acc=0.728, loss=52.217, backward_time=0.405, grad_norm=28.061, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.096, optim0_lr0=2.937e-04, train_time=1.792
+[gpub078:0/16] 2024-01-26 11:53:58,219 (trainer:737) INFO: 12epoch:train:8901-9000batch: iter_time=9.348e-05, forward_time=0.290, loss_ctc=52.389, loss_att=49.879, acc=0.725, loss=50.632, backward_time=0.402, grad_norm=58.615, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.093, optim0_lr0=2.937e-04, train_time=1.537
+[gpub078:0/16] 2024-01-26 11:57:03,623 (trainer:737) INFO: 12epoch:train:9001-9100batch: iter_time=5.523e-04, forward_time=0.351, loss_ctc=59.104, loss_att=54.599, acc=0.727, loss=55.950, backward_time=0.429, grad_norm=33.255, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.106, optim0_lr0=2.936e-04, train_time=1.854
+[gpub078:0/16] 2024-01-26 11:59:49,201 (trainer:737) INFO: 12epoch:train:9101-9200batch: iter_time=8.338e-05, forward_time=0.324, loss_ctc=58.477, loss_att=47.467, acc=0.722, loss=50.770, backward_time=0.410, grad_norm=34.726, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.096, optim0_lr0=2.935e-04, train_time=1.655
+[gpub078:0/16] 2024-01-26 12:02:30,375 (trainer:737) INFO: 12epoch:train:9201-9300batch: iter_time=8.411e-05, forward_time=0.288, loss_ctc=46.148, loss_att=50.934, acc=0.706, loss=49.499, backward_time=0.401, grad_norm=29.209, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.934e-04, train_time=1.611
+[gpub078:0/16] 2024-01-26 12:05:33,468 (trainer:737) INFO: 12epoch:train:9301-9400batch: iter_time=8.274e-05, forward_time=0.350, loss_ctc=57.769, loss_att=55.770, acc=0.725, loss=56.370, backward_time=0.473, grad_norm=31.585, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.108, optim0_lr0=2.933e-04, train_time=1.831
+[gpub078:0/16] 2024-01-26 12:08:29,319 (trainer:737) INFO: 12epoch:train:9401-9500batch: iter_time=8.514e-05, forward_time=0.315, loss_ctc=64.899, loss_att=56.548, acc=0.722, loss=59.053, backward_time=0.423, grad_norm=33.839, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.094, optim0_lr0=2.932e-04, train_time=1.758
+[gpub078:0/16] 2024-01-26 12:10:54,630 (trainer:737) INFO: 12epoch:train:9501-9600batch: iter_time=8.995e-05, forward_time=0.291, loss_ctc=51.747, loss_att=54.048, acc=0.726, loss=53.358, backward_time=0.404, grad_norm=28.927, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.094, optim0_lr0=2.931e-04, train_time=1.452
+[gpub078:0/16] 2024-01-26 12:13:58,004 (trainer:737) INFO: 12epoch:train:9601-9700batch: iter_time=9.067e-05, forward_time=0.369, loss_ctc=63.440, loss_att=57.662, acc=0.722, loss=59.395, backward_time=0.422, grad_norm=45.497, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.103, optim0_lr0=2.931e-04, train_time=1.834
+[gpub078:0/16] 2024-01-26 12:16:31,624 (trainer:737) INFO: 12epoch:train:9701-9800batch: iter_time=8.855e-05, forward_time=0.352, loss_ctc=49.732, loss_att=55.963, acc=0.715, loss=54.094, backward_time=0.406, grad_norm=29.961, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.094, optim0_lr0=2.930e-04, train_time=1.536
+[gpub078:0/16] 2024-01-26 12:19:32,971 (trainer:737) INFO: 12epoch:train:9801-9900batch: iter_time=8.635e-05, forward_time=0.288, loss_ctc=53.002, loss_att=43.555, acc=0.745, loss=46.389, backward_time=0.399, grad_norm=31.873, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.929e-04, train_time=1.813
+[gpub078:0/16] 2024-01-26 12:22:26,250 (trainer:737) INFO: 12epoch:train:9901-10000batch: iter_time=8.298e-05, forward_time=0.404, loss_ctc=61.168, loss_att=50.784, acc=0.746, loss=53.899, backward_time=0.447, grad_norm=34.900, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.114, optim0_lr0=2.928e-04, train_time=1.733
+[gpub078:0/16] 2024-01-26 12:22:46,401 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub078:0/16] 2024-01-26 12:23:05,379 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 12:23:09,313 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f98fb860c40>)
+[gpub078:0/16] 2024-01-26 12:23:09,314 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub078:0/16] 2024-01-26 12:23:09,317 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 12:31:55,689 (trainer:737) INFO: 12epoch:train:10001-10100batch: iter_time=4.099, forward_time=0.290, loss_ctc=50.812, loss_att=59.527, acc=0.712, loss=56.913, backward_time=0.403, grad_norm=30.802, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.927e-04, train_time=5.694
+[gpub078:0/16] 2024-01-26 12:34:58,376 (trainer:737) INFO: 12epoch:train:10101-10200batch: iter_time=8.516e-05, forward_time=0.288, loss_ctc=51.796, loss_att=48.110, acc=0.729, loss=49.216, backward_time=0.400, grad_norm=30.337, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.926e-04, train_time=1.827
+[gpub078:0/16] 2024-01-26 12:37:29,345 (trainer:737) INFO: 12epoch:train:10201-10300batch: iter_time=7.859e-05, forward_time=0.354, loss_ctc=49.847, loss_att=47.090, acc=0.719, loss=47.917, backward_time=0.442, grad_norm=30.793, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.105, optim0_lr0=2.926e-04, train_time=1.509
+[gpub078:0/16] 2024-01-26 12:40:44,455 (trainer:737) INFO: 12epoch:train:10301-10400batch: iter_time=9.236e-05, forward_time=0.329, loss_ctc=60.760, loss_att=49.584, acc=0.734, loss=52.937, backward_time=0.408, grad_norm=35.351, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.094, optim0_lr0=2.925e-04, train_time=1.951
+[gpub078:0/16] 2024-01-26 12:43:08,276 (trainer:737) INFO: 12epoch:train:10401-10500batch: iter_time=8.505e-05, forward_time=0.290, loss_ctc=53.414, loss_att=52.763, acc=0.687, loss=52.958, backward_time=0.401, grad_norm=36.153, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.924e-04, train_time=1.438
+[gpub078:0/16] 2024-01-26 12:46:00,479 (trainer:737) INFO: 12epoch:train:10501-10600batch: iter_time=8.742e-05, forward_time=0.289, loss_ctc=56.640, loss_att=55.532, acc=0.706, loss=55.865, backward_time=0.400, grad_norm=33.976, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.923e-04, train_time=1.722
+[gpub078:0/16] 2024-01-26 12:48:54,479 (trainer:737) INFO: 12epoch:train:10601-10700batch: iter_time=8.416e-05, forward_time=0.386, loss_ctc=58.903, loss_att=55.258, acc=0.716, loss=56.351, backward_time=0.472, grad_norm=33.426, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.107, optim0_lr0=2.922e-04, train_time=1.739
+[gpub078:0/16] 2024-01-26 12:51:53,029 (trainer:737) INFO: 12epoch:train:10701-10800batch: iter_time=8.086e-05, forward_time=0.288, loss_ctc=56.136, loss_att=53.320, acc=0.711, loss=54.165, backward_time=0.399, grad_norm=31.357, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.921e-04, train_time=1.786
+[gpub078:0/16] 2024-01-26 12:54:10,855 (trainer:737) INFO: 12epoch:train:10801-10900batch: iter_time=8.162e-05, forward_time=0.291, loss_ctc=54.971, loss_att=52.660, acc=0.723, loss=53.354, backward_time=0.402, grad_norm=33.035, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.093, optim0_lr0=2.921e-04, train_time=1.378
+[gpub078:0/16] 2024-01-26 12:57:16,845 (trainer:737) INFO: 12epoch:train:10901-11000batch: iter_time=8.605e-05, forward_time=0.452, loss_ctc=58.246, loss_att=58.362, acc=0.703, loss=58.327, backward_time=0.425, grad_norm=45.544, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.099, optim0_lr0=2.920e-04, train_time=1.859
+[gpub078:0/16] 2024-01-26 13:00:17,494 (trainer:737) INFO: 12epoch:train:11001-11100batch: iter_time=8.798e-05, forward_time=0.288, loss_ctc=48.687, loss_att=49.926, acc=0.719, loss=49.554, backward_time=0.399, grad_norm=28.333, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.919e-04, train_time=1.806
+[gpub078:0/16] 2024-01-26 13:02:54,354 (trainer:737) INFO: 12epoch:train:11101-11200batch: iter_time=8.180e-05, forward_time=0.289, loss_ctc=65.758, loss_att=54.181, acc=0.741, loss=57.654, backward_time=0.402, grad_norm=41.728, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.918e-04, train_time=1.569
+[gpub078:0/16] 2024-01-26 13:04:25,161 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub078:0/16] 2024-01-26 13:04:44,845 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 13:04:48,603 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9c743bb4f0>)
+[gpub078:0/16] 2024-01-26 13:04:48,603 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub078:0/16] 2024-01-26 13:04:48,629 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 13:11:51,780 (trainer:737) INFO: 12epoch:train:11201-11300batch: iter_time=3.825, forward_time=0.397, loss_ctc=50.753, loss_att=49.602, acc=0.720, loss=49.948, backward_time=0.417, grad_norm=29.809, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.096, optim0_lr0=2.917e-04, train_time=5.374
+[gpub078:0/16] 2024-01-26 13:14:32,328 (trainer:737) INFO: 12epoch:train:11301-11400batch: iter_time=8.722e-05, forward_time=0.288, loss_ctc=50.430, loss_att=50.334, acc=0.732, loss=50.363, backward_time=0.402, grad_norm=28.546, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.916e-04, train_time=1.605
+[gpub078:0/16] 2024-01-26 13:17:16,595 (trainer:737) INFO: 12epoch:train:11401-11500batch: iter_time=8.121e-05, forward_time=0.289, loss_ctc=52.498, loss_att=49.468, acc=0.720, loss=50.377, backward_time=0.400, grad_norm=29.495, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.916e-04, train_time=1.643
+[gpub078:0/16] 2024-01-26 13:20:02,382 (trainer:737) INFO: 12epoch:train:11501-11600batch: iter_time=7.931e-05, forward_time=0.418, loss_ctc=59.560, loss_att=54.959, acc=0.720, loss=56.340, backward_time=0.436, grad_norm=34.875, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.118, optim0_lr0=2.915e-04, train_time=1.658
+[gpub078:0/16] 2024-01-26 13:23:06,807 (trainer:737) INFO: 12epoch:train:11601-11700batch: iter_time=9.757e-05, forward_time=0.288, loss_ctc=57.787, loss_att=47.620, acc=0.710, loss=50.670, backward_time=0.398, grad_norm=34.726, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.914e-04, train_time=1.844
+[gpub078:0/16] 2024-01-26 13:25:29,638 (trainer:737) INFO: 12epoch:train:11701-11800batch: iter_time=8.064e-05, forward_time=0.287, loss_ctc=46.028, loss_att=49.632, acc=0.703, loss=48.550, backward_time=0.398, grad_norm=29.634, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.913e-04, train_time=1.427
+[gpub078:0/16] 2024-01-26 13:28:11,468 (trainer:737) INFO: 12epoch:train:11801-11900batch: iter_time=9.301e-05, forward_time=0.371, loss_ctc=57.689, loss_att=55.303, acc=0.723, loss=56.018, backward_time=0.432, grad_norm=32.539, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.105, optim0_lr0=2.912e-04, train_time=1.619
+[gpub078:0/16] 2024-01-26 13:30:52,958 (trainer:737) INFO: 12epoch:train:11901-12000batch: iter_time=9.173e-05, forward_time=0.308, loss_ctc=65.161, loss_att=57.173, acc=0.711, loss=59.569, backward_time=0.406, grad_norm=33.799, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.096, optim0_lr0=2.912e-04, train_time=1.614
+[gpub078:0/16] 2024-01-26 13:33:48,338 (trainer:737) INFO: 12epoch:train:12001-12100batch: iter_time=8.988e-05, forward_time=0.290, loss_ctc=51.784, loss_att=53.177, acc=0.718, loss=52.759, backward_time=0.400, grad_norm=45.289, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.911e-04, train_time=1.754
+[gpub078:0/16] 2024-01-26 13:36:13,328 (trainer:737) INFO: 12epoch:train:12101-12200batch: iter_time=8.956e-05, forward_time=0.289, loss_ctc=63.259, loss_att=56.030, acc=0.719, loss=58.199, backward_time=0.402, grad_norm=42.092, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.910e-04, train_time=1.448
+[gpub078:0/16] 2024-01-26 13:39:22,878 (trainer:737) INFO: 12epoch:train:12201-12300batch: iter_time=8.939e-05, forward_time=0.399, loss_ctc=49.741, loss_att=55.131, acc=0.705, loss=53.514, backward_time=0.478, grad_norm=31.556, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.102, optim0_lr0=2.909e-04, train_time=1.897
+[gpub078:0/16] 2024-01-26 13:42:19,641 (trainer:737) INFO: 12epoch:train:12301-12400batch: iter_time=7.951e-05, forward_time=0.291, loss_ctc=52.435, loss_att=42.608, acc=0.741, loss=45.556, backward_time=0.400, grad_norm=29.466, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.094, optim0_lr0=2.908e-04, train_time=1.767
+[gpub078:0/16] 2024-01-26 13:45:14,038 (trainer:737) INFO: 12epoch:train:12401-12500batch: iter_time=8.073e-05, forward_time=0.289, loss_ctc=59.903, loss_att=50.878, acc=0.733, loss=53.585, backward_time=0.399, grad_norm=36.609, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.094, optim0_lr0=2.907e-04, train_time=1.744
+[gpub078:0/16] 2024-01-26 13:45:34,066 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub078:0/16] 2024-01-26 13:45:53,360 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 13:45:57,033 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f981cd97550>)
+[gpub078:0/16] 2024-01-26 13:45:57,033 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub078:0/16] 2024-01-26 13:45:57,036 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 13:54:15,606 (trainer:737) INFO: 12epoch:train:12501-12600batch: iter_time=3.679, forward_time=0.395, loss_ctc=51.019, loss_att=60.090, acc=0.720, loss=57.369, backward_time=0.429, grad_norm=517.471, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.097, optim0_lr0=2.907e-04, train_time=5.415
+[gpub078:0/16] 2024-01-26 13:57:43,184 (trainer:737) INFO: 12epoch:train:12601-12700batch: iter_time=8.701e-05, forward_time=0.288, loss_ctc=51.643, loss_att=49.611, acc=0.736, loss=50.221, backward_time=0.400, grad_norm=28.229, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.906e-04, train_time=2.076
+[gpub078:0/16] 2024-01-26 14:00:27,421 (trainer:737) INFO: 12epoch:train:12701-12800batch: iter_time=7.555e-05, forward_time=0.287, loss_ctc=50.157, loss_att=47.065, acc=0.728, loss=47.993, backward_time=0.400, grad_norm=28.894, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.905e-04, train_time=1.642
+[gpub078:0/16] 2024-01-26 14:04:15,551 (trainer:737) INFO: 12epoch:train:12801-12900batch: iter_time=9.134e-05, forward_time=0.436, loss_ctc=60.141, loss_att=49.424, acc=0.740, loss=52.639, backward_time=0.431, grad_norm=31.986, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.101, optim0_lr0=2.904e-04, train_time=2.281
+[gpub078:0/16] 2024-01-26 14:07:04,085 (trainer:737) INFO: 12epoch:train:12901-13000batch: iter_time=7.885e-05, forward_time=0.289, loss_ctc=52.806, loss_att=52.563, acc=0.704, loss=52.636, backward_time=0.400, grad_norm=32.403, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.093, optim0_lr0=2.903e-04, train_time=1.685
+[gpub078:0/16] 2024-01-26 14:10:06,840 (trainer:737) INFO: 12epoch:train:13001-13100batch: iter_time=9.349e-05, forward_time=0.290, loss_ctc=56.150, loss_att=55.643, acc=0.714, loss=55.795, backward_time=0.403, grad_norm=34.166, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.903e-04, train_time=1.827
+[gpub078:0/16] 2024-01-26 14:14:07,631 (trainer:737) INFO: 12epoch:train:13101-13200batch: iter_time=9.477e-05, forward_time=0.428, loss_ctc=57.583, loss_att=55.284, acc=0.723, loss=55.974, backward_time=0.500, grad_norm=32.869, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.103, optim0_lr0=2.902e-04, train_time=2.407
+[gpub078:0/16] 2024-01-26 14:17:20,577 (trainer:737) INFO: 12epoch:train:13201-13300batch: iter_time=1.019e-04, forward_time=0.290, loss_ctc=55.635, loss_att=53.220, acc=0.726, loss=53.945, backward_time=0.401, grad_norm=30.519, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.094, optim0_lr0=2.901e-04, train_time=1.930
+[gpub078:0/16] 2024-01-26 14:20:11,697 (trainer:737) INFO: 12epoch:train:13301-13400batch: iter_time=9.275e-05, forward_time=0.290, loss_ctc=54.293, loss_att=52.120, acc=0.735, loss=52.772, backward_time=0.403, grad_norm=28.917, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.900e-04, train_time=1.711
+[gpub078:0/16] 2024-01-26 14:23:55,706 (trainer:737) INFO: 12epoch:train:13401-13500batch: iter_time=8.997e-05, forward_time=0.411, loss_ctc=58.583, loss_att=59.303, acc=0.712, loss=59.087, backward_time=0.486, grad_norm=39.159, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.096, optim0_lr0=2.899e-04, train_time=2.239
+[gpub078:0/16] 2024-01-26 14:26:51,613 (trainer:737) INFO: 12epoch:train:13501-13600batch: iter_time=9.400e-05, forward_time=0.289, loss_ctc=48.463, loss_att=50.526, acc=0.731, loss=49.907, backward_time=0.400, grad_norm=27.832, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.898e-04, train_time=1.759
+[gpub078:0/16] 2024-01-26 14:30:32,904 (trainer:737) INFO: 12epoch:train:13601-13700batch: iter_time=9.253e-05, forward_time=0.289, loss_ctc=64.096, loss_att=51.397, acc=0.747, loss=55.206, backward_time=0.401, grad_norm=38.758, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.898e-04, train_time=2.213
+[gpub078:0/16] 2024-01-26 14:32:17,501 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub078:0/16] 2024-01-26 14:32:37,325 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 14:32:41,013 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f981fafa0e0>)
+[gpub078:0/16] 2024-01-26 14:32:41,013 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub078:0/16] 2024-01-26 14:32:41,016 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 14:39:40,032 (trainer:737) INFO: 12epoch:train:13701-13800batch: iter_time=3.742, forward_time=0.372, loss_ctc=50.619, loss_att=50.629, acc=0.725, loss=50.626, backward_time=0.414, grad_norm=29.315, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.096, optim0_lr0=2.897e-04, train_time=5.471
+[gpub078:0/16] 2024-01-26 14:42:29,190 (trainer:737) INFO: 12epoch:train:13801-13900batch: iter_time=8.509e-05, forward_time=0.288, loss_ctc=49.996, loss_att=50.631, acc=0.730, loss=50.441, backward_time=0.398, grad_norm=29.281, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.896e-04, train_time=1.692
+[gpub078:0/16] 2024-01-26 14:44:52,673 (trainer:737) INFO: 12epoch:train:13901-14000batch: iter_time=8.569e-05, forward_time=0.289, loss_ctc=52.168, loss_att=50.024, acc=0.720, loss=50.667, backward_time=0.401, grad_norm=29.376, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.895e-04, train_time=1.435
+[gpub078:0/16] 2024-01-26 14:48:08,946 (trainer:737) INFO: 12epoch:train:14001-14100batch: iter_time=6.637e-04, forward_time=0.358, loss_ctc=58.119, loss_att=53.458, acc=0.724, loss=54.856, backward_time=0.443, grad_norm=33.676, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.113, optim0_lr0=2.894e-04, train_time=1.963
+[gpub078:0/16] 2024-01-26 14:51:04,715 (trainer:737) INFO: 12epoch:train:14101-14200batch: iter_time=9.371e-05, forward_time=0.285, loss_ctc=56.159, loss_att=46.548, acc=0.713, loss=49.432, backward_time=0.397, grad_norm=32.321, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.094, optim0_lr0=2.894e-04, train_time=1.757
+[gpub078:0/16] 2024-01-26 14:53:42,242 (trainer:737) INFO: 12epoch:train:14201-14300batch: iter_time=9.933e-05, forward_time=0.287, loss_ctc=45.837, loss_att=50.048, acc=0.705, loss=48.785, backward_time=0.399, grad_norm=28.143, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.094, optim0_lr0=2.893e-04, train_time=1.575
+[gpub078:0/16] 2024-01-26 14:56:51,805 (trainer:737) INFO: 12epoch:train:14301-14400batch: iter_time=2.470e-04, forward_time=0.429, loss_ctc=56.727, loss_att=55.549, acc=0.721, loss=55.902, backward_time=0.426, grad_norm=31.616, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.116, optim0_lr0=2.892e-04, train_time=1.895
+[gpub078:0/16] 2024-01-26 14:59:50,693 (trainer:737) INFO: 12epoch:train:14401-14500batch: iter_time=9.321e-05, forward_time=0.290, loss_ctc=64.085, loss_att=56.254, acc=0.712, loss=58.604, backward_time=0.402, grad_norm=34.292, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.891e-04, train_time=1.789
+[gpub078:0/16] 2024-01-26 15:02:24,531 (trainer:737) INFO: 12epoch:train:14501-14600batch: iter_time=8.428e-05, forward_time=0.289, loss_ctc=51.388, loss_att=53.070, acc=0.719, loss=52.566, backward_time=0.402, grad_norm=28.597, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.890e-04, train_time=1.538
+[gpub078:0/16] 2024-01-26 15:05:13,128 (trainer:737) INFO: 12epoch:train:14601-14700batch: iter_time=8.574e-05, forward_time=0.294, loss_ctc=62.571, loss_att=55.033, acc=0.716, loss=57.295, backward_time=0.410, grad_norm=44.812, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.105, optim0_lr0=2.890e-04, train_time=1.686
+[gpub078:0/16] 2024-01-26 15:08:01,886 (trainer:737) INFO: 12epoch:train:14701-14800batch: iter_time=9.305e-05, forward_time=0.407, loss_ctc=49.174, loss_att=55.072, acc=0.704, loss=53.302, backward_time=0.438, grad_norm=29.330, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.100, optim0_lr0=2.889e-04, train_time=1.687
+[gpub078:0/16] 2024-01-26 15:11:14,608 (trainer:737) INFO: 12epoch:train:14801-14900batch: iter_time=9.187e-05, forward_time=0.288, loss_ctc=52.636, loss_att=42.563, acc=0.742, loss=45.585, backward_time=0.400, grad_norm=31.123, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.888e-04, train_time=1.927
+[gpub078:0/16] 2024-01-26 15:13:57,763 (trainer:737) INFO: 12epoch:train:14901-15000batch: iter_time=9.244e-05, forward_time=0.288, loss_ctc=58.712, loss_att=50.330, acc=0.735, loss=52.845, backward_time=0.398, grad_norm=35.206, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.093, optim0_lr0=2.887e-04, train_time=1.631
+[gpub078:0/16] 2024-01-26 15:53:38,776 (trainer:343) INFO: 12epoch results: [train] iter_time=0.291, forward_time=0.323, loss_ctc=56.114, loss_att=53.090, acc=0.719, loss=53.997, backward_time=0.415, grad_norm=36.631, clip=100.000, loss_scale=1.075e+31, optim_step_time=0.096, optim0_lr0=2.950e-04, train_time=1.979, time=8 hours, 15 minutes and 10.02 seconds, total_count=210000, gpu_max_cached_mem_GB=42.420, [valid] loss_ctc=49.367, cer_ctc=0.243, loss_att=49.346, acc=0.625, cer=0.375, wer=0.997, loss=49.352, time=39 minutes and 14.04 seconds, total_count=65394, gpu_max_cached_mem_GB=42.420
+[gpub078:0/16] 2024-01-26 15:53:49,168 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub078:0/16] 2024-01-26 15:53:49,237 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/7epoch.pth
+[gpub078:0/16] 2024-01-26 15:53:49,237 (trainer:272) INFO: 13/45epoch started. Estimated time to finish: 1 week, 5 days and 16 hours
+[gpub078:0/16] 2024-01-26 15:53:49,256 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub078:0/16] 2024-01-26 15:54:07,557 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub078:0/16] 2024-01-26 15:54:11,010 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f9261188550>)
+[gpub078:0/16] 2024-01-26 15:54:11,010 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub078:0/16] 2024-01-26 15:54:11,013 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub078:0/16] 2024-01-26 16:02:03,494 (trainer:737) INFO: 13epoch:train:1-100batch: iter_time=3.405, forward_time=0.293, loss_ctc=66.003, loss_att=62.670, acc=0.690, loss=63.670, backward_time=0.415, grad_norm=38.380, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.096, optim0_lr0=2.886e-04, train_time=4.942
+[gpub078:0/16] 2024-01-26 16:04:18,969 (trainer:737) INFO: 13epoch:train:101-200batch: iter_time=8.850e-05, forward_time=0.289, loss_ctc=51.104, loss_att=54.558, acc=0.704, loss=53.522, backward_time=0.407, grad_norm=31.184, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.093, optim0_lr0=2.886e-04, train_time=1.355
+[gpub078:0/16] 2024-01-26 16:07:02,665 (trainer:737) INFO: 13epoch:train:201-300batch: iter_time=1.018e-04, forward_time=0.348, loss_ctc=69.088, loss_att=61.634, acc=0.704, loss=63.870, backward_time=0.424, grad_norm=40.764, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.098, optim0_lr0=2.885e-04, train_time=1.637
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2891733.0 ON gpub078 CANCELLED AT 2024-01-26T16:09:25 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.14.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.14.log
new file mode 100644
index 0000000000000000000000000000000000000000..05d9245d751810a37e5ee5caa051f410d00c1ca5
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.14.log
@@ -0,0 +1,1979 @@
+# Running on gpuc01.delta.ncsa.illinois.edu
+# Started at Tue Jan 23 14:54:20 CST 2024
+# SLURMD_NODENAME=gpuc01
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=128
+# SLURM_CPUS_PER_TASK=128
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=8
+# SLURM_GTIDS=0
+# SLURM_JOBID=2886934
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='128(x2)'
+# SLURM_JOB_END_TIME=1706216051
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3,4,5,6,7
+# SLURM_JOB_ID=2886934
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpuc[01-02]'
+# SLURM_JOB_NUM_NODES=2
+# SLURM_JOB_PARTITION=gpuA100x8
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1706043251
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=2000000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=2
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpuc[01-02]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x2)'
+# SLURM_TASK_PID=65608
+# SLURM_TOPOLOGY_ADDR=ss00.ss05.gpuc01
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_bba9bc3d-96cd-4319-939a-daad2e59a370 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_bba9bc3d-96cd-4319-939a-daad2e59a370
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 8 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_bba9bc3d-96cd-4319-939a-daad2e59a370
+[gpuc01:0/16] 2024-01-23 14:54:34,411 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpuc01:0/16] 2024-01-23 14:54:34,885 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpuc01:0/16] 2024-01-23 14:54:34,918 (s2t:464) INFO: Vocabulary size: 50002
+[gpuc01:0/16] 2024-01-23 14:54:39,305 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpuc01:0/16] 2024-01-23 14:54:39,311 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpuc01:0/16] 2024-01-23 14:54:39,311 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpuc01:0/16] 2024-01-23 14:54:39,311 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpuc01:0/16] 2024-01-23 14:54:39,313 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpuc01:0/16] 2024-01-23 14:54:45,088 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 14:54:46,403 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7376442410>)
+[gpuc01:0/16] 2024-01-23 14:54:46,403 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=9342, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpuc01:0/16] 2024-01-23 14:54:46,405 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=9342, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 14:54:49,193 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpuc01:65673:65673 [0] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65673:65673 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65673:65673 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpuc01:0/16] 2024-01-23 14:54:52,960 (trainer:284) INFO: 6/45epoch started
+[gpuc01:0/16] 2024-01-23 14:54:53,011 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpuc01:0/16] 2024-01-23 14:55:13,407 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 14:55:17,334 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7185ac1360>)
+[gpuc01:0/16] 2024-01-23 14:55:17,334 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpuc01:0/16] 2024-01-23 14:55:17,341 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+gpuc02:64034:64034 [3] NCCL INFO cudaDriverVersion 12020
+gpuc02:64034:64034 [3] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64034:64034 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64034:64100 [3] NCCL INFO NET/IB : No device found.
+gpuc02:64034:64100 [3] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64034:64100 [3] NCCL INFO Using network Socket
+gpuc02:64034:64100 [3] NCCL INFO Setting affinity for GPU 3 to ffff0000
+gpuc02:64034:64100 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10
+gpuc02:64034:64100 [3] NCCL INFO Channel 00/0 : 11[4c000] -> 12[88000] via P2P/IPC/read
+gpuc02:64034:64100 [3] NCCL INFO Channel 01/0 : 11[4c000] -> 12[88000] via P2P/IPC/read
+gpuc02:64034:64100 [3] NCCL INFO Connected all rings
+gpuc02:64034:64100 [3] NCCL INFO Channel 00/0 : 11[4c000] -> 10[48000] via P2P/IPC/read
+gpuc02:64034:64100 [3] NCCL INFO Channel 01/0 : 11[4c000] -> 10[48000] via P2P/IPC/read
+gpuc02:64034:64100 [3] NCCL INFO Connected all trees
+gpuc02:64034:64100 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64034:64100 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64034:64100 [3] NCCL INFO comm 0x1986bdd0 rank 11 nranks 16 cudaDev 3 busId 4c000 - Init COMPLETE
+gpuc02:64032:64032 [1] NCCL INFO cudaDriverVersion 12020
+gpuc02:64032:64032 [1] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64032:64032 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64032:64101 [1] NCCL INFO NET/IB : No device found.
+gpuc02:64032:64101 [1] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64032:64101 [1] NCCL INFO Using network Socket
+gpuc02:64032:64101 [1] NCCL INFO Setting affinity for GPU 1 to ffff0000,00000000
+gpuc02:64032:64101 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->8
+gpuc02:64032:64101 [1] NCCL INFO Channel 00/0 : 9[b000] -> 10[48000] via P2P/IPC/read
+gpuc02:64032:64101 [1] NCCL INFO Channel 01/0 : 9[b000] -> 10[48000] via P2P/IPC/read
+gpuc02:64032:64101 [1] NCCL INFO Connected all rings
+gpuc02:64032:64101 [1] NCCL INFO Channel 00/0 : 9[b000] -> 8[7000] via P2P/IPC/read
+gpuc02:64032:64101 [1] NCCL INFO Channel 01/0 : 9[b000] -> 8[7000] via P2P/IPC/read
+gpuc02:64032:64101 [1] NCCL INFO Connected all trees
+gpuc02:64032:64101 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64032:64101 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64032:64101 [1] NCCL INFO comm 0x16499180 rank 9 nranks 16 cudaDev 1 busId b000 - Init COMPLETE
+gpuc02:64038:64038 [7] NCCL INFO cudaDriverVersion 12020
+gpuc02:64038:64038 [7] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64038:64038 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64038:64097 [7] NCCL INFO NET/IB : No device found.
+gpuc02:64038:64097 [7] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64038:64097 [7] NCCL INFO Using network Socket
+gpuc02:64038:64097 [7] NCCL INFO Setting affinity for GPU 7 to ffff0000,00000000,00000000
+gpuc02:64038:64097 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpuc02:64038:64097 [7] NCCL INFO Channel 00/0 : 15[cb000] -> 0[7000] [send] via NET/Socket/1
+gpuc02:64038:64097 [7] NCCL INFO Channel 01/0 : 15[cb000] -> 0[7000] [send] via NET/Socket/1
+gpuc02:64038:64097 [7] NCCL INFO Connected all rings
+gpuc02:64038:64097 [7] NCCL INFO Channel 00/0 : 15[cb000] -> 14[c8000] via P2P/IPC/read
+gpuc02:64038:64097 [7] NCCL INFO Channel 01/0 : 15[cb000] -> 14[c8000] via P2P/IPC/read
+gpuc02:64038:64097 [7] NCCL INFO Connected all trees
+gpuc02:64038:64097 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64038:64097 [7] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64038:64097 [7] NCCL INFO comm 0x7533b180 rank 15 nranks 16 cudaDev 7 busId cb000 - Init COMPLETE
+gpuc02:64036:64036 [5] NCCL INFO cudaDriverVersion 12020
+gpuc02:64036:64036 [5] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64036:64036 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64036:64102 [5] NCCL INFO NET/IB : No device found.
+gpuc02:64036:64102 [5] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64036:64102 [5] NCCL INFO Using network Socket
+gpuc02:64036:64102 [5] NCCL INFO Setting affinity for GPU 5 to ffff0000,00000000,00000000,00000000
+gpuc02:64036:64102 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpuc02:64036:64102 [5] NCCL INFO Channel 00/0 : 13[8b000] -> 14[c8000] via P2P/IPC/read
+gpuc02:64036:64102 [5] NCCL INFO Channel 01/0 : 13[8b000] -> 14[c8000] via P2P/IPC/read
+gpuc02:64036:64102 [5] NCCL INFO Connected all rings
+gpuc02:64036:64102 [5] NCCL INFO Channel 00/0 : 13[8b000] -> 12[88000] via P2P/IPC/read
+gpuc02:64036:64102 [5] NCCL INFO Channel 01/0 : 13[8b000] -> 12[88000] via P2P/IPC/read
+gpuc02:64036:64102 [5] NCCL INFO Connected all trees
+gpuc02:64036:64102 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64036:64102 [5] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64036:64102 [5] NCCL INFO comm 0x146eabe0 rank 13 nranks 16 cudaDev 5 busId 8b000 - Init COMPLETE
+gpuc02:64035:64035 [4] NCCL INFO cudaDriverVersion 12020
+gpuc02:64035:64035 [4] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64035:64035 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64035:64098 [4] NCCL INFO NET/IB : No device found.
+gpuc02:64035:64098 [4] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64035:64098 [4] NCCL INFO Using network Socket
+gpuc02:64035:64098 [4] NCCL INFO Setting affinity for GPU 4 to ffff0000,00000000,00000000,00000000
+gpuc02:64035:64098 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11
+gpuc02:64035:64098 [4] NCCL INFO Channel 00/0 : 12[88000] -> 13[8b000] via P2P/IPC/read
+gpuc02:64035:64098 [4] NCCL INFO Channel 01/0 : 12[88000] -> 13[8b000] via P2P/IPC/read
+gpuc02:64035:64098 [4] NCCL INFO Connected all rings
+gpuc02:64035:64098 [4] NCCL INFO Channel 00/0 : 12[88000] -> 11[4c000] via P2P/IPC/read
+gpuc02:64035:64098 [4] NCCL INFO Channel 01/0 : 12[88000] -> 11[4c000] via P2P/IPC/read
+gpuc02:64035:64098 [4] NCCL INFO Connected all trees
+gpuc02:64035:64098 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64035:64098 [4] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64035:64098 [4] NCCL INFO comm 0x7841b240 rank 12 nranks 16 cudaDev 4 busId 88000 - Init COMPLETE
+gpuc01:65675:65675 [2] NCCL INFO cudaDriverVersion 12020
+gpuc01:65675:65675 [2] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65675:65675 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65675:65743 [2] NCCL INFO NET/IB : No device found.
+gpuc01:65675:65743 [2] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65675:65743 [2] NCCL INFO Using network Socket
+gpuc01:65675:65743 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpuc01:65675:65743 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpuc01:65675:65743 [2] NCCL INFO Channel 00/0 : 2[48000] -> 3[4c000] via P2P/IPC/read
+gpuc01:65675:65743 [2] NCCL INFO Channel 01/0 : 2[48000] -> 3[4c000] via P2P/IPC/read
+gpuc01:65675:65743 [2] NCCL INFO Connected all rings
+gpuc01:65675:65743 [2] NCCL INFO Channel 00/0 : 2[48000] -> 1[b000] via P2P/IPC/read
+gpuc01:65675:65743 [2] NCCL INFO Channel 01/0 : 2[48000] -> 1[b000] via P2P/IPC/read
+gpuc01:65675:65743 [2] NCCL INFO Connected all trees
+gpuc01:65675:65743 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65675:65743 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65675:65743 [2] NCCL INFO comm 0x607c5180 rank 2 nranks 16 cudaDev 2 busId 48000 - Init COMPLETE
+gpuc02:64031:64031 [0] NCCL INFO cudaDriverVersion 12020
+gpuc02:64031:64031 [0] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64031:64031 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64031:64103 [0] NCCL INFO NET/IB : No device found.
+gpuc02:64031:64103 [0] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64031:64103 [0] NCCL INFO Using network Socket
+gpuc02:64031:64103 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpuc02:64031:64103 [0] NCCL INFO Trees [0] 9/-1/-1->8->0 [1] 9/0/-1->8->-1
+gpuc02:64031:64103 [0] NCCL INFO Channel 00/0 : 7[cb000] -> 8[7000] [receive] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Channel 01/0 : 7[cb000] -> 8[7000] [receive] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[b000] via P2P/IPC/read
+gpuc02:64031:64103 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[b000] via P2P/IPC/read
+gpuc02:64031:64103 [0] NCCL INFO Connected all rings
+gpuc02:64031:64103 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Channel 01/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Channel 01/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpuc02:64031:64103 [0] NCCL INFO Connected all trees
+gpuc02:64031:64103 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64031:64103 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64031:64103 [0] NCCL INFO comm 0x78f18190 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpuc01:65680:65680 [7] NCCL INFO cudaDriverVersion 12020
+gpuc01:65680:65680 [7] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65680:65680 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65680:65744 [7] NCCL INFO NET/IB : No device found.
+gpuc01:65680:65744 [7] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65680:65744 [7] NCCL INFO Using network Socket
+gpuc01:65680:65744 [7] NCCL INFO Setting affinity for GPU 7 to ffff0000,00000000,00000000
+gpuc01:65680:65744 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpuc01:65680:65744 [7] NCCL INFO Channel 00/0 : 7[cb000] -> 8[7000] [send] via NET/Socket/1
+gpuc01:65680:65744 [7] NCCL INFO Channel 01/0 : 7[cb000] -> 8[7000] [send] via NET/Socket/1
+gpuc01:65680:65744 [7] NCCL INFO Connected all rings
+gpuc01:65680:65744 [7] NCCL INFO Channel 00/0 : 7[cb000] -> 6[c8000] via P2P/IPC/read
+gpuc01:65680:65744 [7] NCCL INFO Channel 01/0 : 7[cb000] -> 6[c8000] via P2P/IPC/read
+gpuc01:65680:65744 [7] NCCL INFO Connected all trees
+gpuc01:65680:65744 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65680:65744 [7] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65680:65744 [7] NCCL INFO comm 0x13fe39b0 rank 7 nranks 16 cudaDev 7 busId cb000 - Init COMPLETE
+gpuc01:65674:65674 [1] NCCL INFO cudaDriverVersion 12020
+gpuc01:65674:65674 [1] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65674:65674 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65674:65747 [1] NCCL INFO NET/IB : No device found.
+gpuc01:65674:65747 [1] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65674:65747 [1] NCCL INFO Using network Socket
+gpuc01:65674:65747 [1] NCCL INFO Setting affinity for GPU 1 to ffff0000,00000000
+gpuc01:65674:65747 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpuc01:65674:65747 [1] NCCL INFO Channel 00/0 : 1[b000] -> 2[48000] via P2P/IPC/read
+gpuc01:65674:65747 [1] NCCL INFO Channel 01/0 : 1[b000] -> 2[48000] via P2P/IPC/read
+gpuc01:65674:65747 [1] NCCL INFO Connected all rings
+gpuc01:65674:65747 [1] NCCL INFO Channel 00/0 : 1[b000] -> 0[7000] via P2P/IPC/read
+gpuc01:65674:65747 [1] NCCL INFO Channel 01/0 : 1[b000] -> 0[7000] via P2P/IPC/read
+gpuc01:65674:65747 [1] NCCL INFO Connected all trees
+gpuc01:65674:65747 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65674:65747 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65674:65747 [1] NCCL INFO comm 0x178ef6d0 rank 1 nranks 16 cudaDev 1 busId b000 - Init COMPLETE
+gpuc02:64037:64037 [6] NCCL INFO cudaDriverVersion 12020
+gpuc02:64037:64037 [6] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64037:64037 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64037:64104 [6] NCCL INFO NET/IB : No device found.
+gpuc02:64037:64104 [6] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64037:64104 [6] NCCL INFO Using network Socket
+gpuc02:64037:64104 [6] NCCL INFO Setting affinity for GPU 6 to ffff0000,00000000,00000000
+gpuc02:64037:64104 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpuc02:64037:64104 [6] NCCL INFO Channel 00/0 : 14[c8000] -> 15[cb000] via P2P/IPC/read
+gpuc02:64037:64104 [6] NCCL INFO Channel 01/0 : 14[c8000] -> 15[cb000] via P2P/IPC/read
+gpuc02:64037:64104 [6] NCCL INFO Connected all rings
+gpuc02:64037:64104 [6] NCCL INFO Channel 00/0 : 14[c8000] -> 13[8b000] via P2P/IPC/read
+gpuc02:64037:64104 [6] NCCL INFO Channel 01/0 : 14[c8000] -> 13[8b000] via P2P/IPC/read
+gpuc02:64037:64104 [6] NCCL INFO Connected all trees
+gpuc02:64037:64104 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64037:64104 [6] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64037:64104 [6] NCCL INFO comm 0x1169bcc0 rank 14 nranks 16 cudaDev 6 busId c8000 - Init COMPLETE
+gpuc01:65673:65741 [0] NCCL INFO NET/IB : No device found.
+gpuc01:65673:65741 [0] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65673:65741 [0] NCCL INFO Using network Socket
+gpuc01:65673:65741 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpuc01:65673:65741 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpuc01:65673:65741 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpuc01:65673:65741 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->8
+gpuc01:65673:65741 [0] NCCL INFO Channel 00/0 : 15[cb000] -> 0[7000] [receive] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Channel 01/0 : 15[cb000] -> 0[7000] [receive] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[b000] via P2P/IPC/read
+gpuc01:65673:65741 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[b000] via P2P/IPC/read
+gpuc01:65673:65741 [0] NCCL INFO Connected all rings
+gpuc01:65673:65741 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Channel 01/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Channel 01/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpuc01:65673:65741 [0] NCCL INFO Connected all trees
+gpuc01:65673:65741 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65673:65741 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65673:65741 [0] NCCL INFO comm 0x1bd5cad0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpuc01:65678:65678 [5] NCCL INFO cudaDriverVersion 12020
+gpuc01:65678:65678 [5] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65678:65678 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65678:65746 [5] NCCL INFO NET/IB : No device found.
+gpuc01:65678:65746 [5] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65678:65746 [5] NCCL INFO Using network Socket
+gpuc01:65678:65746 [5] NCCL INFO Setting affinity for GPU 5 to ffff0000,00000000,00000000,00000000
+gpuc01:65678:65746 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4
+gpuc01:65678:65746 [5] NCCL INFO Channel 00/0 : 5[8b000] -> 6[c8000] via P2P/IPC/read
+gpuc01:65678:65746 [5] NCCL INFO Channel 01/0 : 5[8b000] -> 6[c8000] via P2P/IPC/read
+gpuc01:65678:65746 [5] NCCL INFO Connected all rings
+gpuc01:65678:65746 [5] NCCL INFO Channel 00/0 : 5[8b000] -> 4[88000] via P2P/IPC/read
+gpuc01:65678:65746 [5] NCCL INFO Channel 01/0 : 5[8b000] -> 4[88000] via P2P/IPC/read
+gpuc01:65678:65746 [5] NCCL INFO Connected all trees
+gpuc01:65678:65746 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65678:65746 [5] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65678:65746 [5] NCCL INFO comm 0x5dbeb010 rank 5 nranks 16 cudaDev 5 busId 8b000 - Init COMPLETE
+gpuc01:65679:65679 [6] NCCL INFO cudaDriverVersion 12020
+gpuc01:65679:65679 [6] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65679:65679 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65679:65742 [6] NCCL INFO NET/IB : No device found.
+gpuc01:65679:65742 [6] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65679:65742 [6] NCCL INFO Using network Socket
+gpuc01:65679:65742 [6] NCCL INFO Setting affinity for GPU 6 to ffff0000,00000000,00000000
+gpuc01:65679:65742 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpuc01:65679:65742 [6] NCCL INFO Channel 00/0 : 6[c8000] -> 7[cb000] via P2P/IPC/read
+gpuc01:65679:65742 [6] NCCL INFO Channel 01/0 : 6[c8000] -> 7[cb000] via P2P/IPC/read
+gpuc01:65679:65742 [6] NCCL INFO Connected all rings
+gpuc01:65679:65742 [6] NCCL INFO Channel 00/0 : 6[c8000] -> 5[8b000] via P2P/IPC/read
+gpuc01:65679:65742 [6] NCCL INFO Channel 01/0 : 6[c8000] -> 5[8b000] via P2P/IPC/read
+gpuc01:65679:65742 [6] NCCL INFO Connected all trees
+gpuc01:65679:65742 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65679:65742 [6] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65679:65742 [6] NCCL INFO comm 0x5ef0c670 rank 6 nranks 16 cudaDev 6 busId c8000 - Init COMPLETE
+gpuc01:65676:65676 [3] NCCL INFO cudaDriverVersion 12020
+gpuc01:65676:65676 [3] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65676:65676 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65676:65748 [3] NCCL INFO NET/IB : No device found.
+gpuc01:65676:65748 [3] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65676:65748 [3] NCCL INFO Using network Socket
+gpuc01:65676:65748 [3] NCCL INFO Setting affinity for GPU 3 to ffff0000
+gpuc01:65676:65748 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2
+gpuc01:65676:65748 [3] NCCL INFO Channel 00/0 : 3[4c000] -> 4[88000] via P2P/IPC/read
+gpuc01:65676:65748 [3] NCCL INFO Channel 01/0 : 3[4c000] -> 4[88000] via P2P/IPC/read
+gpuc01:65676:65748 [3] NCCL INFO Connected all rings
+gpuc01:65676:65748 [3] NCCL INFO Channel 00/0 : 3[4c000] -> 2[48000] via P2P/IPC/read
+gpuc01:65676:65748 [3] NCCL INFO Channel 01/0 : 3[4c000] -> 2[48000] via P2P/IPC/read
+gpuc01:65676:65748 [3] NCCL INFO Connected all trees
+gpuc01:65676:65748 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65676:65748 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65676:65748 [3] NCCL INFO comm 0x77358280 rank 3 nranks 16 cudaDev 3 busId 4c000 - Init COMPLETE
+gpuc01:65677:65677 [4] NCCL INFO cudaDriverVersion 12020
+gpuc01:65677:65677 [4] NCCL INFO Bootstrap : Using eth0:172.28.23.201<0>
+gpuc01:65677:65677 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc01:65677:65745 [4] NCCL INFO NET/IB : No device found.
+gpuc01:65677:65745 [4] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.201<0> [1]hsn0:141.142.145.201<0>
+gpuc01:65677:65745 [4] NCCL INFO Using network Socket
+gpuc01:65677:65745 [4] NCCL INFO Setting affinity for GPU 4 to ffff0000,00000000,00000000,00000000
+gpuc01:65677:65745 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3
+gpuc01:65677:65745 [4] NCCL INFO Channel 00/0 : 4[88000] -> 5[8b000] via P2P/IPC/read
+gpuc01:65677:65745 [4] NCCL INFO Channel 01/0 : 4[88000] -> 5[8b000] via P2P/IPC/read
+gpuc01:65677:65745 [4] NCCL INFO Connected all rings
+gpuc01:65677:65745 [4] NCCL INFO Channel 00/0 : 4[88000] -> 3[4c000] via P2P/IPC/read
+gpuc01:65677:65745 [4] NCCL INFO Channel 01/0 : 4[88000] -> 3[4c000] via P2P/IPC/read
+gpuc01:65677:65745 [4] NCCL INFO Connected all trees
+gpuc01:65677:65745 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc01:65677:65745 [4] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc01:65677:65745 [4] NCCL INFO comm 0x60f79500 rank 4 nranks 16 cudaDev 4 busId 88000 - Init COMPLETE
+gpuc02:64033:64033 [2] NCCL INFO cudaDriverVersion 12020
+gpuc02:64033:64033 [2] NCCL INFO Bootstrap : Using eth0:172.28.23.202<0>
+gpuc02:64033:64033 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpuc02:64033:64099 [2] NCCL INFO NET/IB : No device found.
+gpuc02:64033:64099 [2] NCCL INFO NET/Socket : Using [0]eth0:172.28.23.202<0> [1]hsn0:141.142.145.202<0>
+gpuc02:64033:64099 [2] NCCL INFO Using network Socket
+gpuc02:64033:64099 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpuc02:64033:64099 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpuc02:64033:64099 [2] NCCL INFO Channel 00/0 : 10[48000] -> 11[4c000] via P2P/IPC/read
+gpuc02:64033:64099 [2] NCCL INFO Channel 01/0 : 10[48000] -> 11[4c000] via P2P/IPC/read
+gpuc02:64033:64099 [2] NCCL INFO Connected all rings
+gpuc02:64033:64099 [2] NCCL INFO Channel 00/0 : 10[48000] -> 9[b000] via P2P/IPC/read
+gpuc02:64033:64099 [2] NCCL INFO Channel 01/0 : 10[48000] -> 9[b000] via P2P/IPC/read
+gpuc02:64033:64099 [2] NCCL INFO Connected all trees
+gpuc02:64033:64099 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpuc02:64033:64099 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpuc02:64033:64099 [2] NCCL INFO comm 0x78ed87d0 rank 10 nranks 16 cudaDev 2 busId 48000 - Init COMPLETE
+[gpuc01:0/16] 2024-01-23 14:57:23,789 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpuc01:0/16] 2024-01-23 14:59:49,879 (trainer:737) INFO: 6epoch:train:1-100batch: iter_time=1.383, forward_time=0.123, loss_ctc=74.653, loss_att=70.242, acc=0.635, loss=71.565, backward_time=0.143, grad_norm=30.045, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.471e-04, train_time=5.936
+[gpuc01:0/16] 2024-01-23 15:02:16,338 (trainer:737) INFO: 6epoch:train:101-200batch: iter_time=1.713e-04, forward_time=0.115, loss_ctc=72.379, loss_att=72.675, acc=0.618, loss=72.586, backward_time=0.137, grad_norm=27.565, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.470e-04, train_time=2.930
+[gpuc01:0/16] 2024-01-23 15:04:42,758 (trainer:737) INFO: 6epoch:train:201-300batch: iter_time=1.720e-04, forward_time=0.116, loss_ctc=78.465, loss_att=81.089, acc=0.606, loss=80.302, backward_time=0.138, grad_norm=29.067, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.468e-04, train_time=2.928
+[gpuc01:0/16] 2024-01-23 15:07:08,423 (trainer:737) INFO: 6epoch:train:301-400batch: iter_time=1.680e-04, forward_time=0.116, loss_ctc=75.745, loss_att=67.448, acc=0.622, loss=69.937, backward_time=0.135, grad_norm=26.802, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.467e-04, train_time=2.913
+[gpuc01:0/16] 2024-01-23 15:09:34,602 (trainer:737) INFO: 6epoch:train:401-500batch: iter_time=1.786e-04, forward_time=0.115, loss_ctc=77.316, loss_att=70.605, acc=0.627, loss=72.619, backward_time=0.135, grad_norm=29.122, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.465e-04, train_time=2.923
+[gpuc01:0/16] 2024-01-23 15:12:01,748 (trainer:737) INFO: 6epoch:train:501-600batch: iter_time=1.706e-04, forward_time=0.115, loss_ctc=68.813, loss_att=70.225, acc=0.625, loss=69.802, backward_time=0.135, grad_norm=26.969, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.464e-04, train_time=2.943
+[gpuc01:0/16] 2024-01-23 15:14:27,895 (trainer:737) INFO: 6epoch:train:601-700batch: iter_time=1.771e-04, forward_time=0.115, loss_ctc=68.369, loss_att=69.981, acc=0.605, loss=69.497, backward_time=0.135, grad_norm=24.312, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.462e-04, train_time=2.923
+[gpuc01:0/16] 2024-01-23 15:16:53,082 (trainer:737) INFO: 6epoch:train:701-800batch: iter_time=1.895e-04, forward_time=0.115, loss_ctc=69.461, loss_att=67.086, acc=0.628, loss=67.799, backward_time=0.135, grad_norm=27.404, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.461e-04, train_time=2.904
+[gpuc01:0/16] 2024-01-23 15:19:20,282 (trainer:737) INFO: 6epoch:train:801-900batch: iter_time=1.770e-04, forward_time=0.115, loss_ctc=86.466, loss_att=72.725, acc=0.627, loss=76.847, backward_time=0.136, grad_norm=33.478, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.459e-04, train_time=2.944
+[gpuc01:0/16] 2024-01-23 15:21:48,062 (trainer:737) INFO: 6epoch:train:901-1000batch: iter_time=1.699e-04, forward_time=0.115, loss_ctc=72.813, loss_att=65.988, acc=0.630, loss=68.035, backward_time=0.135, grad_norm=27.229, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.065, optim0_lr0=4.458e-04, train_time=2.955
+[gpuc01:0/16] 2024-01-23 15:24:15,401 (trainer:737) INFO: 6epoch:train:1001-1100batch: iter_time=1.882e-04, forward_time=0.115, loss_ctc=74.061, loss_att=66.214, acc=0.636, loss=68.568, backward_time=0.135, grad_norm=33.650, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.457e-04, train_time=2.947
+[gpuc01:0/16] 2024-01-23 15:26:42,699 (trainer:737) INFO: 6epoch:train:1101-1200batch: iter_time=1.613e-04, forward_time=0.115, loss_ctc=75.201, loss_att=67.062, acc=0.624, loss=69.503, backward_time=0.135, grad_norm=29.480, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.455e-04, train_time=2.946
+[gpuc01:0/16] 2024-01-23 15:29:10,795 (trainer:737) INFO: 6epoch:train:1201-1300batch: iter_time=1.847e-04, forward_time=0.115, loss_ctc=75.204, loss_att=67.473, acc=0.620, loss=69.793, backward_time=0.137, grad_norm=31.284, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.454e-04, train_time=2.962
+[gpuc01:0/16] 2024-01-23 15:31:38,951 (trainer:737) INFO: 6epoch:train:1301-1400batch: iter_time=1.740e-04, forward_time=0.116, loss_ctc=71.496, loss_att=68.384, acc=0.642, loss=69.317, backward_time=0.135, grad_norm=28.104, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.452e-04, train_time=2.963
+[gpuc01:0/16] 2024-01-23 15:34:06,622 (trainer:737) INFO: 6epoch:train:1401-1500batch: iter_time=1.652e-04, forward_time=0.114, loss_ctc=57.174, loss_att=52.343, acc=0.628, loss=53.792, backward_time=0.135, grad_norm=24.659, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.065, optim0_lr0=4.451e-04, train_time=2.953
+[gpuc01:0/16] 2024-01-23 15:36:34,695 (trainer:737) INFO: 6epoch:train:1501-1600batch: iter_time=1.629e-04, forward_time=0.118, loss_ctc=94.054, loss_att=79.139, acc=0.616, loss=83.613, backward_time=0.136, grad_norm=34.040, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.065, optim0_lr0=4.449e-04, train_time=2.961
+[gpuc01:0/16] 2024-01-23 15:39:01,288 (trainer:737) INFO: 6epoch:train:1601-1700batch: iter_time=1.740e-04, forward_time=0.116, loss_ctc=73.266, loss_att=78.532, acc=0.616, loss=76.952, backward_time=0.136, grad_norm=31.987, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.065, optim0_lr0=4.448e-04, train_time=2.932
+[gpuc01:0/16] 2024-01-23 15:41:28,084 (trainer:737) INFO: 6epoch:train:1701-1800batch: iter_time=1.792e-04, forward_time=0.115, loss_ctc=71.174, loss_att=69.177, acc=0.617, loss=69.776, backward_time=0.136, grad_norm=25.165, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.446e-04, train_time=2.936
+[gpuc01:0/16] 2024-01-23 15:43:56,227 (trainer:737) INFO: 6epoch:train:1801-1900batch: iter_time=1.925e-04, forward_time=0.116, loss_ctc=76.690, loss_att=78.679, acc=0.618, loss=78.082, backward_time=0.136, grad_norm=28.880, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.445e-04, train_time=2.963
+[gpuc01:0/16] 2024-01-23 15:46:25,348 (trainer:737) INFO: 6epoch:train:1901-2000batch: iter_time=1.642e-04, forward_time=0.116, loss_ctc=91.833, loss_att=83.458, acc=0.631, loss=85.971, backward_time=0.136, grad_norm=33.296, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.066, optim0_lr0=4.443e-04, train_time=2.982
+[gpuc01:0/16] 2024-01-23 15:48:54,208 (trainer:737) INFO: 6epoch:train:2001-2100batch: iter_time=1.726e-04, forward_time=0.116, loss_ctc=65.789, loss_att=60.476, acc=0.646, loss=62.070, backward_time=0.135, grad_norm=24.606, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.442e-04, train_time=2.977
+[gpuc01:0/16] 2024-01-23 15:51:22,822 (trainer:737) INFO: 6epoch:train:2101-2200batch: iter_time=2.012e-04, forward_time=0.116, loss_ctc=74.956, loss_att=78.386, acc=0.608, loss=77.357, backward_time=0.136, grad_norm=29.327, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.440e-04, train_time=2.972
+[gpuc01:0/16] 2024-01-23 15:53:48,367 (trainer:737) INFO: 6epoch:train:2201-2300batch: iter_time=1.613e-04, forward_time=0.115, loss_ctc=69.821, loss_att=69.313, acc=0.625, loss=69.465, backward_time=0.135, grad_norm=29.655, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.439e-04, train_time=2.911
+[gpuc01:0/16] 2024-01-23 15:56:14,694 (trainer:737) INFO: 6epoch:train:2301-2400batch: iter_time=1.675e-04, forward_time=0.116, loss_ctc=66.843, loss_att=58.541, acc=0.634, loss=61.032, backward_time=0.135, grad_norm=26.631, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.437e-04, train_time=2.926
+[gpuc01:0/16] 2024-01-23 15:58:42,524 (trainer:737) INFO: 6epoch:train:2401-2500batch: iter_time=1.576e-04, forward_time=0.117, loss_ctc=81.396, loss_att=64.416, acc=0.637, loss=69.510, backward_time=0.135, grad_norm=34.065, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.436e-04, train_time=2.956
+[gpuc01:0/16] 2024-01-23 15:58:45,249 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpuc01:0/16] 2024-01-23 15:59:04,162 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 15:59:07,816 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7185443ee0>)
+[gpuc01:0/16] 2024-01-23 15:59:07,816 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpuc01:0/16] 2024-01-23 15:59:07,822 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 16:03:34,924 (trainer:737) INFO: 6epoch:train:2501-2600batch: iter_time=1.435, forward_time=0.119, loss_ctc=71.458, loss_att=70.294, acc=0.645, loss=70.643, backward_time=0.137, grad_norm=28.610, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.069, optim0_lr0=4.435e-04, train_time=5.848
+[gpuc01:0/16] 2024-01-23 16:06:02,291 (trainer:737) INFO: 6epoch:train:2601-2700batch: iter_time=2.047e-04, forward_time=0.117, loss_ctc=69.242, loss_att=73.015, acc=0.633, loss=71.883, backward_time=0.135, grad_norm=25.418, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.433e-04, train_time=2.947
+[gpuc01:0/16] 2024-01-23 16:08:30,278 (trainer:737) INFO: 6epoch:train:2701-2800batch: iter_time=2.227e-04, forward_time=0.118, loss_ctc=76.335, loss_att=82.696, acc=0.613, loss=80.788, backward_time=0.136, grad_norm=28.460, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.432e-04, train_time=2.960
+[gpuc01:0/16] 2024-01-23 16:10:57,719 (trainer:737) INFO: 6epoch:train:2801-2900batch: iter_time=2.494e-04, forward_time=0.117, loss_ctc=73.980, loss_att=73.561, acc=0.621, loss=73.686, backward_time=0.135, grad_norm=29.260, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.430e-04, train_time=2.949
+[gpuc01:0/16] 2024-01-23 16:13:25,077 (trainer:737) INFO: 6epoch:train:2901-3000batch: iter_time=2.202e-04, forward_time=0.116, loss_ctc=74.396, loss_att=69.403, acc=0.634, loss=70.901, backward_time=0.135, grad_norm=28.721, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.429e-04, train_time=2.947
+[gpuc01:0/16] 2024-01-23 16:15:53,687 (trainer:737) INFO: 6epoch:train:3001-3100batch: iter_time=2.313e-04, forward_time=0.116, loss_ctc=67.601, loss_att=72.208, acc=0.632, loss=70.826, backward_time=0.136, grad_norm=25.799, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.427e-04, train_time=2.972
+[gpuc01:0/16] 2024-01-23 16:18:24,129 (trainer:737) INFO: 6epoch:train:3101-3200batch: iter_time=2.464e-04, forward_time=0.115, loss_ctc=65.864, loss_att=67.530, acc=0.627, loss=67.031, backward_time=0.135, grad_norm=24.290, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.426e-04, train_time=3.009
+[gpuc01:0/16] 2024-01-23 16:20:54,396 (trainer:737) INFO: 6epoch:train:3201-3300batch: iter_time=2.427e-04, forward_time=0.116, loss_ctc=65.993, loss_att=65.479, acc=0.638, loss=65.633, backward_time=0.135, grad_norm=28.206, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.424e-04, train_time=3.005
+[gpuc01:0/16] 2024-01-23 16:23:23,434 (trainer:737) INFO: 6epoch:train:3301-3400batch: iter_time=2.369e-04, forward_time=0.117, loss_ctc=83.603, loss_att=73.199, acc=0.633, loss=76.320, backward_time=0.136, grad_norm=32.179, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.423e-04, train_time=2.981
+[gpuc01:0/16] 2024-01-23 16:25:51,795 (trainer:737) INFO: 6epoch:train:3401-3500batch: iter_time=2.196e-04, forward_time=0.117, loss_ctc=71.827, loss_att=65.457, acc=0.649, loss=67.368, backward_time=0.136, grad_norm=27.118, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.422e-04, train_time=2.967
+[gpuc01:0/16] 2024-01-23 16:28:20,715 (trainer:737) INFO: 6epoch:train:3501-3600batch: iter_time=2.406e-04, forward_time=0.118, loss_ctc=72.381, loss_att=66.318, acc=0.646, loss=68.137, backward_time=0.136, grad_norm=31.576, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.420e-04, train_time=2.978
+[gpuc01:0/16] 2024-01-23 16:30:49,104 (trainer:737) INFO: 6epoch:train:3601-3700batch: iter_time=2.289e-04, forward_time=0.117, loss_ctc=72.575, loss_att=65.559, acc=0.644, loss=67.663, backward_time=0.136, grad_norm=28.286, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.419e-04, train_time=2.968
+[gpuc01:0/16] 2024-01-23 16:33:17,783 (trainer:737) INFO: 6epoch:train:3701-3800batch: iter_time=2.186e-04, forward_time=0.117, loss_ctc=71.924, loss_att=66.156, acc=0.636, loss=67.887, backward_time=0.136, grad_norm=27.407, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.417e-04, train_time=2.973
+[gpuc01:0/16] 2024-01-23 16:35:46,042 (trainer:737) INFO: 6epoch:train:3801-3900batch: iter_time=1.967e-04, forward_time=0.117, loss_ctc=70.588, loss_att=68.497, acc=0.652, loss=69.125, backward_time=0.136, grad_norm=28.114, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.416e-04, train_time=2.965
+[gpuc01:0/16] 2024-01-23 16:38:15,662 (trainer:737) INFO: 6epoch:train:3901-4000batch: iter_time=2.179e-04, forward_time=0.116, loss_ctc=55.168, loss_att=50.960, acc=0.641, loss=52.222, backward_time=0.135, grad_norm=22.197, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.414e-04, train_time=2.992
+[gpuc01:0/16] 2024-01-23 16:40:44,557 (trainer:737) INFO: 6epoch:train:4001-4100batch: iter_time=2.201e-04, forward_time=0.117, loss_ctc=90.259, loss_att=77.413, acc=0.627, loss=81.267, backward_time=0.136, grad_norm=33.404, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.413e-04, train_time=2.978
+[gpuc01:0/16] 2024-01-23 16:43:13,908 (trainer:737) INFO: 6epoch:train:4101-4200batch: iter_time=2.225e-04, forward_time=0.124, loss_ctc=69.846, loss_att=76.457, acc=0.631, loss=74.474, backward_time=0.136, grad_norm=27.533, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.411e-04, train_time=2.987
+[gpuc01:0/16] 2024-01-23 16:45:42,061 (trainer:737) INFO: 6epoch:train:4201-4300batch: iter_time=1.881e-04, forward_time=0.117, loss_ctc=69.164, loss_att=68.089, acc=0.639, loss=68.412, backward_time=0.137, grad_norm=25.316, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.410e-04, train_time=2.963
+[gpuc01:0/16] 2024-01-23 16:48:12,398 (trainer:737) INFO: 6epoch:train:4301-4400batch: iter_time=1.763e-04, forward_time=0.117, loss_ctc=74.526, loss_att=79.521, acc=0.629, loss=78.023, backward_time=0.136, grad_norm=28.993, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.409e-04, train_time=3.007
+[gpuc01:0/16] 2024-01-23 16:50:42,383 (trainer:737) INFO: 6epoch:train:4401-4500batch: iter_time=2.065e-04, forward_time=0.119, loss_ctc=90.095, loss_att=86.794, acc=0.632, loss=87.784, backward_time=0.136, grad_norm=33.667, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.407e-04, train_time=2.999
+[gpuc01:0/16] 2024-01-23 16:53:12,180 (trainer:737) INFO: 6epoch:train:4501-4600batch: iter_time=1.961e-04, forward_time=0.117, loss_ctc=64.873, loss_att=60.529, acc=0.657, loss=61.832, backward_time=0.135, grad_norm=24.319, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.406e-04, train_time=2.996
+[gpuc01:0/16] 2024-01-23 16:55:41,725 (trainer:737) INFO: 6epoch:train:4601-4700batch: iter_time=1.999e-04, forward_time=0.118, loss_ctc=72.953, loss_att=78.886, acc=0.622, loss=77.106, backward_time=0.137, grad_norm=28.267, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.404e-04, train_time=2.991
+[gpuc01:0/16] 2024-01-23 16:58:10,195 (trainer:737) INFO: 6epoch:train:4701-4800batch: iter_time=1.766e-04, forward_time=0.117, loss_ctc=67.949, loss_att=64.983, acc=0.643, loss=65.873, backward_time=0.136, grad_norm=26.353, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.403e-04, train_time=2.969
+[gpuc01:0/16] 2024-01-23 17:00:38,954 (trainer:737) INFO: 6epoch:train:4801-4900batch: iter_time=1.823e-04, forward_time=0.117, loss_ctc=65.472, loss_att=57.452, acc=0.640, loss=59.858, backward_time=0.135, grad_norm=25.391, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.402e-04, train_time=2.975
+[gpuc01:0/16] 2024-01-23 17:03:10,671 (trainer:737) INFO: 6epoch:train:4901-5000batch: iter_time=1.820e-04, forward_time=0.117, loss_ctc=77.971, loss_att=63.545, acc=0.644, loss=67.873, backward_time=0.136, grad_norm=36.471, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.400e-04, train_time=3.034
+[gpuc01:0/16] 2024-01-23 17:03:13,862 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpuc01:0/16] 2024-01-23 17:03:32,685 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 17:03:36,221 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71850b4520>)
+[gpuc01:0/16] 2024-01-23 17:03:36,222 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpuc01:0/16] 2024-01-23 17:03:36,228 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 17:07:59,427 (trainer:737) INFO: 6epoch:train:5001-5100batch: iter_time=1.419, forward_time=0.117, loss_ctc=70.875, loss_att=67.986, acc=0.644, loss=68.852, backward_time=0.135, grad_norm=29.054, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.399e-04, train_time=5.775
+[gpuc01:0/16] 2024-01-23 17:10:27,650 (trainer:737) INFO: 6epoch:train:5101-5200batch: iter_time=2.101e-04, forward_time=0.116, loss_ctc=68.212, loss_att=69.807, acc=0.638, loss=69.328, backward_time=0.136, grad_norm=27.528, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.397e-04, train_time=2.964
+[gpuc01:0/16] 2024-01-23 17:12:54,393 (trainer:737) INFO: 6epoch:train:5201-5300batch: iter_time=2.030e-04, forward_time=0.118, loss_ctc=74.777, loss_att=79.448, acc=0.616, loss=78.047, backward_time=0.136, grad_norm=29.070, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.396e-04, train_time=2.935
+[gpuc01:0/16] 2024-01-23 17:15:21,307 (trainer:737) INFO: 6epoch:train:5301-5400batch: iter_time=1.961e-04, forward_time=0.117, loss_ctc=72.200, loss_att=66.364, acc=0.627, loss=68.115, backward_time=0.135, grad_norm=29.377, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.394e-04, train_time=2.938
+[gpuc01:0/16] 2024-01-23 17:17:48,942 (trainer:737) INFO: 6epoch:train:5401-5500batch: iter_time=1.847e-04, forward_time=0.117, loss_ctc=73.759, loss_att=68.213, acc=0.636, loss=69.877, backward_time=0.135, grad_norm=29.575, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.393e-04, train_time=2.952
+[gpuc01:0/16] 2024-01-23 17:20:16,382 (trainer:737) INFO: 6epoch:train:5501-5600batch: iter_time=1.721e-04, forward_time=0.117, loss_ctc=66.342, loss_att=67.419, acc=0.634, loss=67.096, backward_time=0.135, grad_norm=25.997, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.392e-04, train_time=2.949
+[gpuc01:0/16] 2024-01-23 17:22:45,716 (trainer:737) INFO: 6epoch:train:5601-5700batch: iter_time=1.773e-04, forward_time=0.116, loss_ctc=65.241, loss_att=66.963, acc=0.615, loss=66.446, backward_time=0.135, grad_norm=23.002, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.066, optim0_lr0=4.390e-04, train_time=2.986
+[gpuc01:0/16] 2024-01-23 17:25:12,939 (trainer:737) INFO: 6epoch:train:5701-5800batch: iter_time=1.860e-04, forward_time=0.116, loss_ctc=64.746, loss_att=63.643, acc=0.639, loss=63.974, backward_time=0.135, grad_norm=26.970, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.389e-04, train_time=2.944
+[gpuc01:0/16] 2024-01-23 17:27:41,326 (trainer:737) INFO: 6epoch:train:5801-5900batch: iter_time=1.928e-04, forward_time=0.117, loss_ctc=82.646, loss_att=68.726, acc=0.638, loss=72.902, backward_time=0.135, grad_norm=33.145, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.387e-04, train_time=2.968
+[gpuc01:0/16] 2024-01-23 17:30:08,690 (trainer:737) INFO: 6epoch:train:5901-6000batch: iter_time=2.103e-04, forward_time=0.116, loss_ctc=71.723, loss_att=64.160, acc=0.639, loss=66.429, backward_time=0.135, grad_norm=29.370, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.067, optim0_lr0=4.386e-04, train_time=2.947
+[gpuc01:0/16] 2024-01-23 17:32:34,616 (trainer:737) INFO: 6epoch:train:6001-6100batch: iter_time=2.045e-04, forward_time=0.117, loss_ctc=70.042, loss_att=63.635, acc=0.645, loss=65.557, backward_time=0.135, grad_norm=29.497, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.385e-04, train_time=2.918
+[gpuc01:0/16] 2024-01-23 17:35:00,797 (trainer:737) INFO: 6epoch:train:6101-6200batch: iter_time=2.069e-04, forward_time=0.117, loss_ctc=70.094, loss_att=64.049, acc=0.640, loss=65.862, backward_time=0.135, grad_norm=30.107, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.383e-04, train_time=2.923
+[gpuc01:0/16] 2024-01-23 17:37:26,784 (trainer:737) INFO: 6epoch:train:6201-6300batch: iter_time=2.188e-04, forward_time=0.117, loss_ctc=71.029, loss_att=65.074, acc=0.628, loss=66.860, backward_time=0.135, grad_norm=27.618, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.382e-04, train_time=2.920
+[gpuc01:0/16] 2024-01-23 17:39:54,411 (trainer:737) INFO: 6epoch:train:6301-6400batch: iter_time=1.996e-04, forward_time=0.117, loss_ctc=70.174, loss_att=66.533, acc=0.649, loss=67.625, backward_time=0.135, grad_norm=28.313, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.380e-04, train_time=2.952
+[gpuc01:0/16] 2024-01-23 17:42:21,484 (trainer:737) INFO: 6epoch:train:6401-6500batch: iter_time=2.024e-04, forward_time=0.116, loss_ctc=54.584, loss_att=49.842, acc=0.639, loss=51.265, backward_time=0.135, grad_norm=23.620, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.379e-04, train_time=2.941
+[gpuc01:0/16] 2024-01-23 17:44:48,741 (trainer:737) INFO: 6epoch:train:6501-6600batch: iter_time=2.064e-04, forward_time=0.117, loss_ctc=87.995, loss_att=75.092, acc=0.628, loss=78.963, backward_time=0.136, grad_norm=32.235, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.378e-04, train_time=2.945
+[gpuc01:0/16] 2024-01-23 17:47:16,017 (trainer:737) INFO: 6epoch:train:6601-6700batch: iter_time=2.067e-04, forward_time=0.117, loss_ctc=69.703, loss_att=76.084, acc=0.624, loss=74.170, backward_time=0.136, grad_norm=27.008, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.376e-04, train_time=2.945
+[gpuc01:0/16] 2024-01-23 17:49:42,716 (trainer:737) INFO: 6epoch:train:6701-6800batch: iter_time=2.062e-04, forward_time=0.116, loss_ctc=69.201, loss_att=67.111, acc=0.627, loss=67.738, backward_time=0.135, grad_norm=25.841, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.375e-04, train_time=2.934
+[gpuc01:0/16] 2024-01-23 17:52:10,190 (trainer:737) INFO: 6epoch:train:6801-6900batch: iter_time=1.943e-04, forward_time=0.118, loss_ctc=73.851, loss_att=77.740, acc=0.624, loss=76.573, backward_time=0.136, grad_norm=28.743, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.373e-04, train_time=2.949
+[gpuc01:0/16] 2024-01-23 17:54:37,533 (trainer:737) INFO: 6epoch:train:6901-7000batch: iter_time=1.773e-04, forward_time=0.117, loss_ctc=87.676, loss_att=81.523, acc=0.640, loss=83.369, backward_time=0.136, grad_norm=31.334, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.372e-04, train_time=2.947
+[gpuc01:0/16] 2024-01-23 17:57:05,604 (trainer:737) INFO: 6epoch:train:7001-7100batch: iter_time=1.738e-04, forward_time=0.119, loss_ctc=63.948, loss_att=58.702, acc=0.653, loss=60.276, backward_time=0.135, grad_norm=23.895, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.371e-04, train_time=2.961
+[gpuc01:0/16] 2024-01-23 17:59:32,877 (trainer:737) INFO: 6epoch:train:7101-7200batch: iter_time=1.697e-04, forward_time=0.117, loss_ctc=72.348, loss_att=76.123, acc=0.613, loss=74.991, backward_time=0.136, grad_norm=27.808, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.369e-04, train_time=2.945
+[gpuc01:0/16] 2024-01-23 18:02:00,182 (trainer:737) INFO: 6epoch:train:7201-7300batch: iter_time=1.772e-04, forward_time=0.116, loss_ctc=66.820, loss_att=65.272, acc=0.635, loss=65.736, backward_time=0.135, grad_norm=27.603, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.368e-04, train_time=2.946
+[gpuc01:0/16] 2024-01-23 18:04:28,531 (trainer:737) INFO: 6epoch:train:7301-7400batch: iter_time=1.643e-04, forward_time=0.115, loss_ctc=63.822, loss_att=56.930, acc=0.644, loss=58.997, backward_time=0.134, grad_norm=25.535, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.366e-04, train_time=2.967
+[gpuc01:0/16] 2024-01-23 18:06:56,828 (trainer:737) INFO: 6epoch:train:7401-7500batch: iter_time=1.658e-04, forward_time=0.115, loss_ctc=75.893, loss_att=62.017, acc=0.645, loss=66.180, backward_time=0.135, grad_norm=32.122, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.365e-04, train_time=2.966
+[gpuc01:0/16] 2024-01-23 18:06:59,485 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpuc01:0/16] 2024-01-23 18:07:18,562 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 18:07:22,112 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6c094a3b50>)
+[gpuc01:0/16] 2024-01-23 18:07:22,112 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpuc01:0/16] 2024-01-23 18:07:22,118 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 18:11:50,561 (trainer:737) INFO: 6epoch:train:7501-7600batch: iter_time=1.401, forward_time=0.114, loss_ctc=69.514, loss_att=65.800, acc=0.650, loss=66.914, backward_time=0.135, grad_norm=29.205, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.364e-04, train_time=5.874
+[gpuc01:0/16] 2024-01-23 18:14:23,981 (trainer:737) INFO: 6epoch:train:7601-7700batch: iter_time=1.693e-04, forward_time=0.115, loss_ctc=67.245, loss_att=67.187, acc=0.640, loss=67.204, backward_time=0.136, grad_norm=26.211, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.362e-04, train_time=3.068
+[gpuc01:0/16] 2024-01-23 18:16:58,413 (trainer:737) INFO: 6epoch:train:7701-7800batch: iter_time=1.592e-04, forward_time=0.117, loss_ctc=74.147, loss_att=76.790, acc=0.619, loss=75.997, backward_time=0.136, grad_norm=26.864, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.361e-04, train_time=3.088
+[gpuc01:0/16] 2024-01-23 18:19:31,625 (trainer:737) INFO: 6epoch:train:7801-7900batch: iter_time=1.644e-04, forward_time=0.114, loss_ctc=70.880, loss_att=63.745, acc=0.636, loss=65.886, backward_time=0.135, grad_norm=25.699, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.359e-04, train_time=3.064
+[gpuc01:0/16] 2024-01-23 18:22:04,633 (trainer:737) INFO: 6epoch:train:7901-8000batch: iter_time=1.603e-04, forward_time=0.114, loss_ctc=71.959, loss_att=65.805, acc=0.643, loss=67.652, backward_time=0.135, grad_norm=27.725, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.358e-04, train_time=3.060
+[gpuc01:0/16] 2024-01-23 18:24:37,397 (trainer:737) INFO: 6epoch:train:8001-8100batch: iter_time=1.508e-04, forward_time=0.114, loss_ctc=66.314, loss_att=67.382, acc=0.636, loss=67.062, backward_time=0.135, grad_norm=24.624, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.357e-04, train_time=3.055
+[gpuc01:0/16] 2024-01-23 18:27:09,935 (trainer:737) INFO: 6epoch:train:8101-8200batch: iter_time=1.710e-04, forward_time=0.114, loss_ctc=64.812, loss_att=66.557, acc=0.616, loss=66.033, backward_time=0.134, grad_norm=23.642, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.355e-04, train_time=3.051
+[gpuc01:0/16] 2024-01-23 18:29:43,159 (trainer:737) INFO: 6epoch:train:8201-8300batch: iter_time=1.555e-04, forward_time=0.114, loss_ctc=63.993, loss_att=61.027, acc=0.645, loss=61.916, backward_time=0.135, grad_norm=26.180, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.354e-04, train_time=3.064
+[gpuc01:0/16] 2024-01-23 18:32:14,389 (trainer:737) INFO: 6epoch:train:8301-8400batch: iter_time=1.525e-04, forward_time=0.114, loss_ctc=82.546, loss_att=68.479, acc=0.637, loss=72.699, backward_time=0.135, grad_norm=32.336, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.353e-04, train_time=3.024
+[gpuc01:0/16] 2024-01-23 18:34:47,963 (trainer:737) INFO: 6epoch:train:8401-8500batch: iter_time=1.681e-04, forward_time=0.114, loss_ctc=69.669, loss_att=62.311, acc=0.646, loss=64.518, backward_time=0.135, grad_norm=25.831, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.351e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-23 18:37:22,952 (trainer:737) INFO: 6epoch:train:8501-8600batch: iter_time=1.493e-04, forward_time=0.114, loss_ctc=69.931, loss_att=62.032, acc=0.651, loss=64.401, backward_time=0.135, grad_norm=32.689, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.350e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-23 18:39:56,495 (trainer:737) INFO: 6epoch:train:8601-8700batch: iter_time=1.480e-04, forward_time=0.114, loss_ctc=69.126, loss_att=62.979, acc=0.641, loss=64.823, backward_time=0.134, grad_norm=26.346, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.348e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-23 18:42:30,193 (trainer:737) INFO: 6epoch:train:8701-8800batch: iter_time=1.509e-04, forward_time=0.114, loss_ctc=70.979, loss_att=63.349, acc=0.633, loss=65.638, backward_time=0.134, grad_norm=29.282, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.347e-04, train_time=3.074
+[gpuc01:0/16] 2024-01-23 18:45:05,521 (trainer:737) INFO: 6epoch:train:8801-8900batch: iter_time=1.463e-04, forward_time=0.115, loss_ctc=67.974, loss_att=65.104, acc=0.653, loss=65.965, backward_time=0.135, grad_norm=27.893, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.346e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-23 18:47:38,990 (trainer:737) INFO: 6epoch:train:8901-9000batch: iter_time=1.514e-04, forward_time=0.114, loss_ctc=53.519, loss_att=49.467, acc=0.641, loss=50.682, backward_time=0.134, grad_norm=22.065, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.344e-04, train_time=3.069
+[gpuc01:0/16] 2024-01-23 18:50:12,987 (trainer:737) INFO: 6epoch:train:9001-9100batch: iter_time=1.381e-04, forward_time=0.116, loss_ctc=87.031, loss_att=73.110, acc=0.633, loss=77.286, backward_time=0.135, grad_norm=32.947, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.343e-04, train_time=3.080
+[gpuc01:0/16] 2024-01-23 18:52:46,865 (trainer:737) INFO: 6epoch:train:9101-9200batch: iter_time=1.502e-04, forward_time=0.116, loss_ctc=68.452, loss_att=75.273, acc=0.629, loss=73.227, backward_time=0.135, grad_norm=27.029, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.342e-04, train_time=3.077
+[gpuc01:0/16] 2024-01-23 18:55:20,488 (trainer:737) INFO: 6epoch:train:9201-9300batch: iter_time=1.496e-04, forward_time=0.115, loss_ctc=68.908, loss_att=66.250, acc=0.629, loss=67.047, backward_time=0.134, grad_norm=27.121, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.340e-04, train_time=3.072
+[gpuc01:0/16] 2024-01-23 18:57:53,805 (trainer:737) INFO: 6epoch:train:9301-9400batch: iter_time=1.365e-04, forward_time=0.117, loss_ctc=72.152, loss_att=74.899, acc=0.633, loss=74.075, backward_time=0.135, grad_norm=29.181, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.339e-04, train_time=3.066
+[gpuc01:0/16] 2024-01-23 19:00:27,663 (trainer:737) INFO: 6epoch:train:9401-9500batch: iter_time=1.372e-04, forward_time=0.117, loss_ctc=86.591, loss_att=78.977, acc=0.647, loss=81.261, backward_time=0.136, grad_norm=31.523, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.338e-04, train_time=3.077
+[gpuc01:0/16] 2024-01-23 19:03:01,536 (trainer:737) INFO: 6epoch:train:9501-9600batch: iter_time=1.340e-04, forward_time=0.115, loss_ctc=63.471, loss_att=57.919, acc=0.657, loss=59.585, backward_time=0.135, grad_norm=23.437, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.336e-04, train_time=3.077
+[gpuc01:0/16] 2024-01-23 19:05:35,775 (trainer:737) INFO: 6epoch:train:9601-9700batch: iter_time=1.399e-04, forward_time=0.119, loss_ctc=71.601, loss_att=74.820, acc=0.620, loss=73.854, backward_time=0.135, grad_norm=29.582, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.067, optim0_lr0=4.335e-04, train_time=3.085
+[gpuc01:0/16] 2024-01-23 19:08:09,294 (trainer:737) INFO: 6epoch:train:9701-9800batch: iter_time=1.444e-04, forward_time=0.117, loss_ctc=66.200, loss_att=63.899, acc=0.641, loss=64.589, backward_time=0.135, grad_norm=26.219, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.333e-04, train_time=3.070
+[gpuc01:0/16] 2024-01-23 19:10:43,995 (trainer:737) INFO: 6epoch:train:9801-9900batch: iter_time=1.361e-04, forward_time=0.119, loss_ctc=63.048, loss_att=56.287, acc=0.649, loss=58.315, backward_time=0.134, grad_norm=25.170, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.332e-04, train_time=3.094
+[gpuc01:0/16] 2024-01-23 19:13:17,226 (trainer:737) INFO: 6epoch:train:9901-10000batch: iter_time=3.187e-04, forward_time=0.117, loss_ctc=73.969, loss_att=60.377, acc=0.650, loss=64.454, backward_time=0.135, grad_norm=37.831, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.066, optim0_lr0=4.331e-04, train_time=3.064
+[gpuc01:0/16] 2024-01-23 19:13:19,890 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpuc01:0/16] 2024-01-23 19:13:39,735 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 19:13:43,338 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7172cc7520>)
+[gpuc01:0/16] 2024-01-23 19:13:43,338 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpuc01:0/16] 2024-01-23 19:13:43,344 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 19:18:12,612 (trainer:737) INFO: 6epoch:train:10001-10100batch: iter_time=1.392, forward_time=0.116, loss_ctc=68.750, loss_att=67.302, acc=0.659, loss=67.737, backward_time=0.135, grad_norm=31.158, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.329e-04, train_time=5.907
+[gpuc01:0/16] 2024-01-23 19:20:47,320 (trainer:737) INFO: 6epoch:train:10101-10200batch: iter_time=1.705e-04, forward_time=0.116, loss_ctc=67.532, loss_att=70.880, acc=0.642, loss=69.876, backward_time=0.135, grad_norm=26.065, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.328e-04, train_time=3.094
+[gpuc01:0/16] 2024-01-23 19:23:21,960 (trainer:737) INFO: 6epoch:train:10201-10300batch: iter_time=1.694e-04, forward_time=0.116, loss_ctc=72.672, loss_att=78.612, acc=0.627, loss=76.830, backward_time=0.135, grad_norm=27.436, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.327e-04, train_time=3.093
+[gpuc01:0/16] 2024-01-23 19:25:56,416 (trainer:737) INFO: 6epoch:train:10301-10400batch: iter_time=1.621e-04, forward_time=0.115, loss_ctc=70.345, loss_att=70.247, acc=0.634, loss=70.277, backward_time=0.134, grad_norm=28.640, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.325e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-23 19:28:29,934 (trainer:737) INFO: 6epoch:train:10401-10500batch: iter_time=1.648e-04, forward_time=0.115, loss_ctc=70.841, loss_att=65.623, acc=0.652, loss=67.189, backward_time=0.135, grad_norm=26.826, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.324e-04, train_time=3.070
+[gpuc01:0/16] 2024-01-23 19:31:04,979 (trainer:737) INFO: 6epoch:train:10501-10600batch: iter_time=1.536e-04, forward_time=0.116, loss_ctc=64.431, loss_att=69.794, acc=0.644, loss=68.185, backward_time=0.135, grad_norm=25.719, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.323e-04, train_time=3.101
+[gpuc01:0/16] 2024-01-23 19:33:39,728 (trainer:737) INFO: 6epoch:train:10601-10700batch: iter_time=1.642e-04, forward_time=0.115, loss_ctc=64.570, loss_att=66.330, acc=0.632, loss=65.802, backward_time=0.135, grad_norm=24.967, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.321e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 19:36:14,075 (trainer:737) INFO: 6epoch:train:10701-10800batch: iter_time=1.641e-04, forward_time=0.118, loss_ctc=62.847, loss_att=62.781, acc=0.651, loss=62.800, backward_time=0.134, grad_norm=26.109, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.320e-04, train_time=3.087
+[gpuc01:0/16] 2024-01-23 19:38:48,540 (trainer:737) INFO: 6epoch:train:10801-10900batch: iter_time=1.558e-04, forward_time=0.116, loss_ctc=80.497, loss_att=68.637, acc=0.651, loss=72.195, backward_time=0.135, grad_norm=30.292, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.319e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-23 19:41:22,426 (trainer:737) INFO: 6epoch:train:10901-11000batch: iter_time=1.552e-04, forward_time=0.116, loss_ctc=68.990, loss_att=63.077, acc=0.658, loss=64.851, backward_time=0.135, grad_norm=25.825, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.317e-04, train_time=3.078
+[gpuc01:0/16] 2024-01-23 19:43:57,020 (trainer:737) INFO: 6epoch:train:11001-11100batch: iter_time=1.507e-04, forward_time=0.116, loss_ctc=68.767, loss_att=62.718, acc=0.658, loss=64.533, backward_time=0.134, grad_norm=28.878, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.316e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-23 19:46:31,146 (trainer:737) INFO: 6epoch:train:11101-11200batch: iter_time=1.596e-04, forward_time=0.115, loss_ctc=68.856, loss_att=63.708, acc=0.653, loss=65.252, backward_time=0.135, grad_norm=28.358, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.315e-04, train_time=3.082
+[gpuc01:0/16] 2024-01-23 19:49:06,359 (trainer:737) INFO: 6epoch:train:11201-11300batch: iter_time=1.667e-04, forward_time=0.116, loss_ctc=69.760, loss_att=63.714, acc=0.644, loss=65.528, backward_time=0.135, grad_norm=28.401, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.313e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-23 19:51:41,182 (trainer:737) INFO: 6epoch:train:11301-11400batch: iter_time=1.642e-04, forward_time=0.115, loss_ctc=67.051, loss_att=65.617, acc=0.663, loss=66.047, backward_time=0.135, grad_norm=27.318, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.312e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-23 19:54:15,785 (trainer:737) INFO: 6epoch:train:11401-11500batch: iter_time=1.604e-04, forward_time=0.115, loss_ctc=52.710, loss_att=48.353, acc=0.654, loss=49.660, backward_time=0.134, grad_norm=22.518, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.311e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-23 19:56:50,555 (trainer:737) INFO: 6epoch:train:11501-11600batch: iter_time=1.650e-04, forward_time=0.116, loss_ctc=86.920, loss_att=75.005, acc=0.639, loss=78.580, backward_time=0.135, grad_norm=32.593, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.309e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 19:59:26,579 (trainer:737) INFO: 6epoch:train:11601-11700batch: iter_time=1.713e-04, forward_time=0.116, loss_ctc=68.121, loss_att=74.579, acc=0.641, loss=72.641, backward_time=0.135, grad_norm=29.025, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.308e-04, train_time=3.120
+[gpuc01:0/16] 2024-01-23 20:02:01,326 (trainer:737) INFO: 6epoch:train:11701-11800batch: iter_time=1.569e-04, forward_time=0.116, loss_ctc=67.359, loss_att=65.966, acc=0.647, loss=66.384, backward_time=0.135, grad_norm=24.960, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.307e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 20:04:35,910 (trainer:737) INFO: 6epoch:train:11801-11900batch: iter_time=1.588e-04, forward_time=0.116, loss_ctc=72.055, loss_att=77.505, acc=0.639, loss=75.870, backward_time=0.135, grad_norm=33.565, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.305e-04, train_time=3.091
+[gpuc01:0/16] 2024-01-23 20:07:10,515 (trainer:737) INFO: 6epoch:train:11901-12000batch: iter_time=1.586e-04, forward_time=0.116, loss_ctc=86.497, loss_att=84.227, acc=0.642, loss=84.908, backward_time=0.135, grad_norm=31.563, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.304e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-23 20:09:45,093 (trainer:737) INFO: 6epoch:train:12001-12100batch: iter_time=1.624e-04, forward_time=0.115, loss_ctc=62.885, loss_att=58.211, acc=0.672, loss=59.613, backward_time=0.134, grad_norm=24.622, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.303e-04, train_time=3.091
+[gpuc01:0/16] 2024-01-23 20:12:20,709 (trainer:737) INFO: 6epoch:train:12101-12200batch: iter_time=1.503e-04, forward_time=0.116, loss_ctc=72.182, loss_att=77.616, acc=0.632, loss=75.986, backward_time=0.136, grad_norm=28.879, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.301e-04, train_time=3.112
+[gpuc01:0/16] 2024-01-23 20:14:56,625 (trainer:737) INFO: 6epoch:train:12201-12300batch: iter_time=1.503e-04, forward_time=0.115, loss_ctc=65.588, loss_att=63.151, acc=0.653, loss=63.882, backward_time=0.134, grad_norm=25.898, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.300e-04, train_time=3.118
+[gpuc01:0/16] 2024-01-23 20:17:31,375 (trainer:737) INFO: 6epoch:train:12301-12400batch: iter_time=1.581e-04, forward_time=0.115, loss_ctc=62.684, loss_att=55.332, acc=0.653, loss=57.538, backward_time=0.134, grad_norm=24.064, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.067, optim0_lr0=4.299e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 20:20:06,532 (trainer:737) INFO: 6epoch:train:12401-12500batch: iter_time=1.572e-04, forward_time=0.115, loss_ctc=73.083, loss_att=60.830, acc=0.654, loss=64.506, backward_time=0.134, grad_norm=32.767, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.297e-04, train_time=3.103
+[gpuc01:0/16] 2024-01-23 20:20:09,142 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpuc01:0/16] 2024-01-23 20:20:27,878 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 20:20:33,762 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6cc11895a0>)
+[gpuc01:0/16] 2024-01-23 20:20:33,762 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpuc01:0/16] 2024-01-23 20:20:33,768 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 20:27:20,466 (trainer:737) INFO: 6epoch:train:12501-12600batch: iter_time=2.714, forward_time=0.178, loss_ctc=68.077, loss_att=64.747, acc=0.664, loss=65.746, backward_time=0.144, grad_norm=28.904, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.072, optim0_lr0=4.296e-04, train_time=8.678
+[gpuc01:0/16] 2024-01-23 20:29:56,210 (trainer:737) INFO: 6epoch:train:12601-12700batch: iter_time=1.511e-04, forward_time=0.116, loss_ctc=66.258, loss_att=67.272, acc=0.651, loss=66.968, backward_time=0.135, grad_norm=26.462, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.295e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-23 20:32:31,999 (trainer:737) INFO: 6epoch:train:12701-12800batch: iter_time=1.532e-04, forward_time=0.117, loss_ctc=72.889, loss_att=76.489, acc=0.633, loss=75.409, backward_time=0.135, grad_norm=28.381, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.293e-04, train_time=3.116
+[gpuc01:0/16] 2024-01-23 20:35:07,320 (trainer:737) INFO: 6epoch:train:12801-12900batch: iter_time=1.463e-04, forward_time=0.118, loss_ctc=69.419, loss_att=67.805, acc=0.641, loss=68.290, backward_time=0.134, grad_norm=27.583, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.292e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-23 20:37:41,268 (trainer:737) INFO: 6epoch:train:12901-13000batch: iter_time=1.468e-04, forward_time=0.118, loss_ctc=70.421, loss_att=64.820, acc=0.656, loss=66.500, backward_time=0.135, grad_norm=29.032, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.291e-04, train_time=3.079
+[gpuc01:0/16] 2024-01-23 20:40:15,042 (trainer:737) INFO: 6epoch:train:13001-13100batch: iter_time=1.501e-04, forward_time=0.117, loss_ctc=63.901, loss_att=66.939, acc=0.651, loss=66.028, backward_time=0.135, grad_norm=25.580, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.289e-04, train_time=3.075
+[gpuc01:0/16] 2024-01-23 20:42:50,007 (trainer:737) INFO: 6epoch:train:13101-13200batch: iter_time=1.468e-04, forward_time=0.116, loss_ctc=63.011, loss_att=63.885, acc=0.636, loss=63.623, backward_time=0.135, grad_norm=24.577, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.288e-04, train_time=3.099
+[gpuc01:0/16] 2024-01-23 20:45:25,885 (trainer:737) INFO: 6epoch:train:13201-13300batch: iter_time=1.566e-04, forward_time=0.115, loss_ctc=62.372, loss_att=61.825, acc=0.653, loss=61.989, backward_time=0.134, grad_norm=26.327, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.287e-04, train_time=3.117
+[gpuc01:0/16] 2024-01-23 20:48:01,363 (trainer:737) INFO: 6epoch:train:13301-13400batch: iter_time=1.479e-04, forward_time=0.116, loss_ctc=79.097, loss_att=67.727, acc=0.655, loss=71.138, backward_time=0.135, grad_norm=31.131, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.285e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-23 20:50:36,420 (trainer:737) INFO: 6epoch:train:13401-13500batch: iter_time=1.422e-04, forward_time=0.116, loss_ctc=68.189, loss_att=61.574, acc=0.663, loss=63.558, backward_time=0.135, grad_norm=27.185, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.284e-04, train_time=3.101
+[gpuc01:0/16] 2024-01-23 20:53:11,383 (trainer:737) INFO: 6epoch:train:13501-13600batch: iter_time=1.457e-04, forward_time=0.116, loss_ctc=67.951, loss_att=60.924, acc=0.664, loss=63.032, backward_time=0.135, grad_norm=27.629, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.283e-04, train_time=3.099
+[gpuc01:0/16] 2024-01-23 20:55:46,410 (trainer:737) INFO: 6epoch:train:13601-13700batch: iter_time=1.418e-04, forward_time=0.116, loss_ctc=67.844, loss_att=61.687, acc=0.659, loss=63.534, backward_time=0.135, grad_norm=26.553, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.282e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-23 20:58:20,880 (trainer:737) INFO: 6epoch:train:13701-13800batch: iter_time=1.464e-04, forward_time=0.116, loss_ctc=69.107, loss_att=62.284, acc=0.648, loss=64.331, backward_time=0.135, grad_norm=28.121, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.280e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-23 21:00:54,528 (trainer:737) INFO: 6epoch:train:13801-13900batch: iter_time=1.428e-04, forward_time=0.116, loss_ctc=66.614, loss_att=64.510, acc=0.665, loss=65.141, backward_time=0.135, grad_norm=26.163, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.279e-04, train_time=3.073
+[gpuc01:0/16] 2024-01-23 21:03:28,038 (trainer:737) INFO: 6epoch:train:13901-14000batch: iter_time=1.513e-04, forward_time=0.116, loss_ctc=52.487, loss_att=47.889, acc=0.658, loss=49.269, backward_time=0.134, grad_norm=21.708, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.066, optim0_lr0=4.278e-04, train_time=3.070
+[gpuc01:0/16] 2024-01-23 21:06:02,920 (trainer:737) INFO: 6epoch:train:14001-14100batch: iter_time=1.542e-04, forward_time=0.117, loss_ctc=86.462, loss_att=73.367, acc=0.642, loss=77.296, backward_time=0.135, grad_norm=31.941, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.276e-04, train_time=3.097
+[gpuc01:0/16] 2024-01-23 21:08:37,842 (trainer:737) INFO: 6epoch:train:14101-14200batch: iter_time=1.377e-04, forward_time=0.119, loss_ctc=67.859, loss_att=73.445, acc=0.645, loss=71.769, backward_time=0.135, grad_norm=28.147, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.275e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-23 21:11:11,736 (trainer:737) INFO: 6epoch:train:14201-14300batch: iter_time=1.420e-04, forward_time=0.117, loss_ctc=66.329, loss_att=64.205, acc=0.654, loss=64.842, backward_time=0.135, grad_norm=26.004, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.274e-04, train_time=3.078
+[gpuc01:0/16] 2024-01-23 21:13:45,282 (trainer:737) INFO: 6epoch:train:14301-14400batch: iter_time=1.447e-04, forward_time=0.117, loss_ctc=70.256, loss_att=74.891, acc=0.645, loss=73.501, backward_time=0.136, grad_norm=27.569, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.272e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-23 21:16:20,379 (trainer:737) INFO: 6epoch:train:14401-14500batch: iter_time=1.501e-04, forward_time=0.117, loss_ctc=84.405, loss_att=81.970, acc=0.648, loss=82.700, backward_time=0.135, grad_norm=29.439, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.271e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-23 21:18:54,974 (trainer:737) INFO: 6epoch:train:14501-14600batch: iter_time=1.484e-04, forward_time=0.116, loss_ctc=62.361, loss_att=56.708, acc=0.675, loss=58.404, backward_time=0.135, grad_norm=22.697, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.065, optim0_lr0=4.270e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-23 21:21:30,785 (trainer:737) INFO: 6epoch:train:14601-14700batch: iter_time=1.467e-04, forward_time=0.117, loss_ctc=69.739, loss_att=74.513, acc=0.639, loss=73.081, backward_time=0.136, grad_norm=27.319, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.269e-04, train_time=3.116
+[gpuc01:0/16] 2024-01-23 21:24:05,560 (trainer:737) INFO: 6epoch:train:14701-14800batch: iter_time=1.496e-04, forward_time=0.116, loss_ctc=65.151, loss_att=63.425, acc=0.654, loss=63.942, backward_time=0.135, grad_norm=25.838, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.065, optim0_lr0=4.267e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 21:26:40,532 (trainer:737) INFO: 6epoch:train:14801-14900batch: iter_time=1.609e-04, forward_time=0.116, loss_ctc=61.964, loss_att=54.608, acc=0.658, loss=56.815, backward_time=0.134, grad_norm=25.168, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.065, optim0_lr0=4.266e-04, train_time=3.099
+[gpuc01:0/16] 2024-01-23 21:29:17,012 (trainer:737) INFO: 6epoch:train:14901-15000batch: iter_time=1.416e-04, forward_time=0.116, loss_ctc=71.844, loss_att=59.679, acc=0.657, loss=63.329, backward_time=0.135, grad_norm=32.814, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.265e-04, train_time=3.129
+[gpuc01:0/16] 2024-01-23 21:29:31,846 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpuc01:0/16] 2024-01-23 21:29:52,203 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 21:29:55,901 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f67480af640>)
+[gpuc01:0/16] 2024-01-23 21:29:55,901 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpuc01:0/16] 2024-01-23 21:29:55,908 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 21:34:51,866 (trainer:737) INFO: 6epoch:train:15001-15100batch: iter_time=1.765, forward_time=0.116, loss_ctc=67.329, loss_att=65.756, acc=0.656, loss=66.228, backward_time=0.135, grad_norm=27.209, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.263e-04, train_time=6.697
+[gpuc01:0/16] 2024-01-23 21:37:26,053 (trainer:737) INFO: 6epoch:train:15101-15200batch: iter_time=1.554e-04, forward_time=0.116, loss_ctc=66.238, loss_att=68.201, acc=0.649, loss=67.612, backward_time=0.135, grad_norm=25.883, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.262e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-23 21:40:00,779 (trainer:737) INFO: 6epoch:train:15201-15300batch: iter_time=1.385e-04, forward_time=0.118, loss_ctc=71.612, loss_att=75.534, acc=0.627, loss=74.358, backward_time=0.136, grad_norm=27.656, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.261e-04, train_time=3.094
+[gpuc01:0/16] 2024-01-23 21:42:34,740 (trainer:737) INFO: 6epoch:train:15301-15400batch: iter_time=1.588e-04, forward_time=0.119, loss_ctc=68.629, loss_att=63.842, acc=0.641, loss=65.278, backward_time=0.135, grad_norm=28.482, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.067, optim0_lr0=4.259e-04, train_time=3.079
+[gpuc01:0/16] 2024-01-23 21:45:08,659 (trainer:737) INFO: 6epoch:train:15401-15500batch: iter_time=1.511e-04, forward_time=0.116, loss_ctc=69.983, loss_att=65.667, acc=0.650, loss=66.962, backward_time=0.135, grad_norm=28.133, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.258e-04, train_time=3.078
+[gpuc01:0/16] 2024-01-23 21:47:42,273 (trainer:737) INFO: 6epoch:train:15501-15600batch: iter_time=1.549e-04, forward_time=0.115, loss_ctc=63.467, loss_att=64.650, acc=0.650, loss=64.296, backward_time=0.135, grad_norm=26.181, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.257e-04, train_time=3.072
+[gpuc01:0/16] 2024-01-23 21:50:15,824 (trainer:737) INFO: 6epoch:train:15601-15700batch: iter_time=1.613e-04, forward_time=0.115, loss_ctc=62.967, loss_att=64.599, acc=0.628, loss=64.109, backward_time=0.134, grad_norm=24.418, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.256e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-23 21:52:50,014 (trainer:737) INFO: 6epoch:train:15701-15800batch: iter_time=1.615e-04, forward_time=0.115, loss_ctc=62.285, loss_att=61.124, acc=0.652, loss=61.472, backward_time=0.135, grad_norm=26.219, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.067, optim0_lr0=4.254e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-23 21:55:24,059 (trainer:737) INFO: 6epoch:train:15801-15900batch: iter_time=1.572e-04, forward_time=0.115, loss_ctc=79.641, loss_att=65.672, acc=0.649, loss=69.863, backward_time=0.135, grad_norm=32.109, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.253e-04, train_time=3.081
+[gpuc01:0/16] 2024-01-23 21:57:58,290 (trainer:737) INFO: 6epoch:train:15901-16000batch: iter_time=1.430e-04, forward_time=0.115, loss_ctc=67.729, loss_att=61.593, acc=0.654, loss=63.434, backward_time=0.135, grad_norm=24.977, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.252e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-23 22:00:33,217 (trainer:737) INFO: 6epoch:train:16001-16100batch: iter_time=1.504e-04, forward_time=0.115, loss_ctc=68.179, loss_att=61.653, acc=0.656, loss=63.611, backward_time=0.135, grad_norm=30.443, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.067, optim0_lr0=4.250e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-23 22:03:08,453 (trainer:737) INFO: 6epoch:train:16101-16200batch: iter_time=1.471e-04, forward_time=0.115, loss_ctc=68.069, loss_att=61.624, acc=0.651, loss=63.557, backward_time=0.135, grad_norm=28.171, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.249e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-23 22:05:43,314 (trainer:737) INFO: 6epoch:train:16201-16300batch: iter_time=1.322e-04, forward_time=0.116, loss_ctc=68.510, loss_att=62.632, acc=0.638, loss=64.396, backward_time=0.135, grad_norm=27.372, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.248e-04, train_time=3.097
+[gpuc01:0/16] 2024-01-23 22:08:19,336 (trainer:737) INFO: 6epoch:train:16301-16400batch: iter_time=1.461e-04, forward_time=0.117, loss_ctc=66.301, loss_att=63.582, acc=0.664, loss=64.398, backward_time=0.135, grad_norm=27.368, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.065, optim0_lr0=4.247e-04, train_time=3.120
+[gpuc01:0/16] 2024-01-23 22:10:54,471 (trainer:737) INFO: 6epoch:train:16401-16500batch: iter_time=1.311e-04, forward_time=0.115, loss_ctc=51.814, loss_att=47.661, acc=0.653, loss=48.907, backward_time=0.134, grad_norm=21.766, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.245e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-23 22:13:30,566 (trainer:737) INFO: 6epoch:train:16501-16600batch: iter_time=1.367e-04, forward_time=0.116, loss_ctc=84.134, loss_att=71.381, acc=0.640, loss=75.207, backward_time=0.135, grad_norm=32.514, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.244e-04, train_time=3.122
+[gpuc01:0/16] 2024-01-23 22:16:04,775 (trainer:737) INFO: 6epoch:train:16601-16700batch: iter_time=1.400e-04, forward_time=0.116, loss_ctc=68.406, loss_att=74.330, acc=0.633, loss=72.553, backward_time=0.135, grad_norm=29.449, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.065, optim0_lr0=4.243e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-23 22:18:38,753 (trainer:737) INFO: 6epoch:train:16701-16800batch: iter_time=1.398e-04, forward_time=0.115, loss_ctc=65.770, loss_att=63.974, acc=0.637, loss=64.513, backward_time=0.134, grad_norm=26.443, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.242e-04, train_time=3.079
+[gpuc01:0/16] 2024-01-23 22:21:13,908 (trainer:737) INFO: 6epoch:train:16801-16900batch: iter_time=1.277e-04, forward_time=0.116, loss_ctc=70.890, loss_att=73.960, acc=0.636, loss=73.039, backward_time=0.135, grad_norm=27.777, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.240e-04, train_time=3.103
+[gpuc01:0/16] 2024-01-23 22:23:49,232 (trainer:737) INFO: 6epoch:train:16901-17000batch: iter_time=1.326e-04, forward_time=0.116, loss_ctc=84.056, loss_att=77.569, acc=0.653, loss=79.515, backward_time=0.135, grad_norm=29.960, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.239e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-23 22:26:23,850 (trainer:737) INFO: 6epoch:train:17001-17100batch: iter_time=1.304e-04, forward_time=0.116, loss_ctc=61.085, loss_att=56.134, acc=0.667, loss=57.619, backward_time=0.134, grad_norm=25.399, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.238e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-23 22:28:58,733 (trainer:737) INFO: 6epoch:train:17101-17200batch: iter_time=1.272e-04, forward_time=0.116, loss_ctc=68.903, loss_att=73.077, acc=0.628, loss=71.825, backward_time=0.135, grad_norm=29.096, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.236e-04, train_time=3.097
+[gpuc01:0/16] 2024-01-23 22:31:33,765 (trainer:737) INFO: 6epoch:train:17201-17300batch: iter_time=1.280e-04, forward_time=0.117, loss_ctc=64.566, loss_att=62.015, acc=0.648, loss=62.780, backward_time=0.135, grad_norm=26.473, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.235e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-23 22:34:09,068 (trainer:737) INFO: 6epoch:train:17301-17400batch: iter_time=1.248e-04, forward_time=0.115, loss_ctc=60.808, loss_att=53.315, acc=0.659, loss=55.563, backward_time=0.134, grad_norm=24.605, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.234e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-23 22:36:43,996 (trainer:737) INFO: 6epoch:train:17401-17500batch: iter_time=1.258e-04, forward_time=0.115, loss_ctc=70.227, loss_att=59.180, acc=0.653, loss=62.494, backward_time=0.134, grad_norm=31.451, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.233e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-23 22:36:46,986 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpuc01:0/16] 2024-01-23 22:37:06,264 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 22:37:10,317 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f66f46bde10>)
+[gpuc01:0/16] 2024-01-23 22:37:10,317 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpuc01:0/16] 2024-01-23 22:37:10,323 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 22:43:33,317 (trainer:737) INFO: 6epoch:train:17501-17600batch: iter_time=2.511, forward_time=0.117, loss_ctc=66.238, loss_att=64.760, acc=0.666, loss=65.203, backward_time=0.135, grad_norm=28.366, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.231e-04, train_time=8.186
+[gpuc01:0/16] 2024-01-23 22:46:09,133 (trainer:737) INFO: 6epoch:train:17601-17700batch: iter_time=1.516e-04, forward_time=0.115, loss_ctc=65.755, loss_att=67.714, acc=0.652, loss=67.126, backward_time=0.135, grad_norm=26.797, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.230e-04, train_time=3.116
+[gpuc01:0/16] 2024-01-23 22:48:44,483 (trainer:737) INFO: 6epoch:train:17701-17800batch: iter_time=1.414e-04, forward_time=0.115, loss_ctc=70.496, loss_att=75.865, acc=0.638, loss=74.254, backward_time=0.136, grad_norm=32.155, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.229e-04, train_time=3.107
+[gpuc01:0/16] 2024-01-23 22:51:19,266 (trainer:737) INFO: 6epoch:train:17801-17900batch: iter_time=1.381e-04, forward_time=0.115, loss_ctc=67.573, loss_att=67.733, acc=0.643, loss=67.685, backward_time=0.135, grad_norm=27.879, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.228e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 22:53:53,737 (trainer:737) INFO: 6epoch:train:17901-18000batch: iter_time=1.548e-04, forward_time=0.115, loss_ctc=69.172, loss_att=63.396, acc=0.662, loss=65.129, backward_time=0.135, grad_norm=27.753, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.066, optim0_lr0=4.226e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-23 22:56:28,254 (trainer:737) INFO: 6epoch:train:18001-18100batch: iter_time=1.451e-04, forward_time=0.116, loss_ctc=62.958, loss_att=67.139, acc=0.655, loss=65.885, backward_time=0.135, grad_norm=24.761, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.225e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-23 22:59:03,626 (trainer:737) INFO: 6epoch:train:18101-18200batch: iter_time=1.406e-04, forward_time=0.114, loss_ctc=61.898, loss_att=63.055, acc=0.642, loss=62.708, backward_time=0.135, grad_norm=23.362, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.224e-04, train_time=3.107
+[gpuc01:0/16] 2024-01-23 23:01:38,571 (trainer:737) INFO: 6epoch:train:18201-18300batch: iter_time=1.427e-04, forward_time=0.114, loss_ctc=60.961, loss_att=60.648, acc=0.657, loss=60.742, backward_time=0.134, grad_norm=25.605, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.223e-04, train_time=3.099
+[gpuc01:0/16] 2024-01-23 23:04:13,743 (trainer:737) INFO: 6epoch:train:18301-18400batch: iter_time=1.379e-04, forward_time=0.115, loss_ctc=78.378, loss_att=66.653, acc=0.660, loss=70.170, backward_time=0.135, grad_norm=30.985, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.221e-04, train_time=3.103
+[gpuc01:0/16] 2024-01-23 23:06:47,945 (trainer:737) INFO: 6epoch:train:18401-18500batch: iter_time=1.433e-04, forward_time=0.115, loss_ctc=67.892, loss_att=60.929, acc=0.667, loss=63.018, backward_time=0.135, grad_norm=26.561, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.220e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-23 23:09:23,947 (trainer:737) INFO: 6epoch:train:18501-18600batch: iter_time=1.607e-04, forward_time=0.115, loss_ctc=67.331, loss_att=60.477, acc=0.671, loss=62.533, backward_time=0.135, grad_norm=30.229, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.219e-04, train_time=3.120
+[gpuc01:0/16] 2024-01-23 23:11:59,390 (trainer:737) INFO: 6epoch:train:18601-18700batch: iter_time=1.566e-04, forward_time=0.115, loss_ctc=66.255, loss_att=60.334, acc=0.662, loss=62.110, backward_time=0.135, grad_norm=26.932, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.218e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-23 23:14:34,605 (trainer:737) INFO: 6epoch:train:18701-18800batch: iter_time=1.994e-04, forward_time=0.115, loss_ctc=68.269, loss_att=61.099, acc=0.655, loss=63.250, backward_time=0.136, grad_norm=28.894, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.216e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-23 23:17:10,347 (trainer:737) INFO: 6epoch:train:18801-18900batch: iter_time=1.460e-04, forward_time=0.116, loss_ctc=64.771, loss_att=63.540, acc=0.672, loss=63.910, backward_time=0.135, grad_norm=28.294, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.215e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-23 23:19:45,096 (trainer:737) INFO: 6epoch:train:18901-19000batch: iter_time=1.559e-04, forward_time=0.115, loss_ctc=51.350, loss_att=47.284, acc=0.661, loss=48.503, backward_time=0.134, grad_norm=22.770, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.214e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 23:22:20,502 (trainer:737) INFO: 6epoch:train:19001-19100batch: iter_time=1.610e-04, forward_time=0.118, loss_ctc=84.042, loss_att=71.979, acc=0.648, loss=75.598, backward_time=0.136, grad_norm=33.749, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.213e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-23 23:24:55,243 (trainer:737) INFO: 6epoch:train:19101-19200batch: iter_time=1.530e-04, forward_time=0.116, loss_ctc=66.201, loss_att=72.105, acc=0.650, loss=70.334, backward_time=0.136, grad_norm=26.325, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.211e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 23:27:29,549 (trainer:737) INFO: 6epoch:train:19201-19300batch: iter_time=1.634e-04, forward_time=0.115, loss_ctc=64.971, loss_att=62.742, acc=0.658, loss=63.411, backward_time=0.135, grad_norm=24.590, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.210e-04, train_time=3.086
+[gpuc01:0/16] 2024-01-23 23:30:04,950 (trainer:737) INFO: 6epoch:train:19301-19400batch: iter_time=1.446e-04, forward_time=0.116, loss_ctc=69.989, loss_att=74.619, acc=0.648, loss=73.230, backward_time=0.136, grad_norm=25.994, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.209e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-23 23:32:39,422 (trainer:737) INFO: 6epoch:train:19401-19500batch: iter_time=1.377e-04, forward_time=0.117, loss_ctc=82.612, loss_att=79.880, acc=0.653, loss=80.699, backward_time=0.135, grad_norm=28.814, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.208e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-23 23:35:14,087 (trainer:737) INFO: 6epoch:train:19501-19600batch: iter_time=1.421e-04, forward_time=0.116, loss_ctc=60.458, loss_att=55.427, acc=0.682, loss=56.937, backward_time=0.135, grad_norm=23.208, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.206e-04, train_time=3.093
+[gpuc01:0/16] 2024-01-23 23:37:49,383 (trainer:737) INFO: 6epoch:train:19601-19700batch: iter_time=1.323e-04, forward_time=0.117, loss_ctc=69.010, loss_att=72.959, acc=0.644, loss=71.774, backward_time=0.136, grad_norm=29.163, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.205e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-23 23:40:24,140 (trainer:737) INFO: 6epoch:train:19701-19800batch: iter_time=1.386e-04, forward_time=0.116, loss_ctc=64.066, loss_att=61.927, acc=0.659, loss=62.569, backward_time=0.135, grad_norm=25.638, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.204e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 23:42:58,470 (trainer:737) INFO: 6epoch:train:19801-19900batch: iter_time=1.311e-04, forward_time=0.116, loss_ctc=60.158, loss_att=53.761, acc=0.662, loss=55.680, backward_time=0.134, grad_norm=24.077, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.203e-04, train_time=3.086
+[gpuc01:0/16] 2024-01-23 23:45:33,603 (trainer:737) INFO: 6epoch:train:19901-20000batch: iter_time=1.291e-04, forward_time=0.116, loss_ctc=70.576, loss_att=59.007, acc=0.660, loss=62.478, backward_time=0.134, grad_norm=31.583, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.067, optim0_lr0=4.201e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-23 23:45:37,473 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpuc01:0/16] 2024-01-23 23:45:57,559 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-23 23:46:01,378 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f671a333bb0>)
+[gpuc01:0/16] 2024-01-23 23:46:01,378 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpuc01:0/16] 2024-01-23 23:46:01,384 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-23 23:51:52,898 (trainer:737) INFO: 6epoch:train:20001-20100batch: iter_time=2.183, forward_time=0.161, loss_ctc=65.878, loss_att=61.662, acc=0.675, loss=62.927, backward_time=0.156, grad_norm=28.015, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.072, optim0_lr0=4.200e-04, train_time=7.586
+[gpuc01:0/16] 2024-01-23 23:54:27,676 (trainer:737) INFO: 6epoch:train:20101-20200batch: iter_time=1.575e-04, forward_time=0.116, loss_ctc=64.183, loss_att=65.780, acc=0.660, loss=65.301, backward_time=0.135, grad_norm=25.463, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.199e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-23 23:57:03,375 (trainer:737) INFO: 6epoch:train:20201-20300batch: iter_time=1.418e-04, forward_time=0.116, loss_ctc=70.496, loss_att=74.045, acc=0.641, loss=72.981, backward_time=0.135, grad_norm=27.360, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.198e-04, train_time=3.114
+[gpuc01:0/16] 2024-01-23 23:59:37,340 (trainer:737) INFO: 6epoch:train:20301-20400batch: iter_time=1.423e-04, forward_time=0.116, loss_ctc=67.291, loss_att=65.697, acc=0.648, loss=66.175, backward_time=0.134, grad_norm=27.834, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.196e-04, train_time=3.079
+[gpuc01:0/16] 2024-01-24 00:02:12,268 (trainer:737) INFO: 6epoch:train:20401-20500batch: iter_time=1.348e-04, forward_time=0.119, loss_ctc=67.922, loss_att=62.952, acc=0.662, loss=64.443, backward_time=0.135, grad_norm=28.496, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.195e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-24 00:04:47,529 (trainer:737) INFO: 6epoch:train:20501-20600batch: iter_time=1.385e-04, forward_time=0.116, loss_ctc=62.475, loss_att=65.585, acc=0.658, loss=64.652, backward_time=0.135, grad_norm=25.922, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.194e-04, train_time=3.105
+[gpuc01:0/16] 2024-01-24 00:07:22,962 (trainer:737) INFO: 6epoch:train:20601-20700batch: iter_time=1.372e-04, forward_time=0.115, loss_ctc=61.077, loss_att=62.011, acc=0.649, loss=61.730, backward_time=0.134, grad_norm=22.312, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.193e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-24 00:09:58,006 (trainer:737) INFO: 6epoch:train:20701-20800batch: iter_time=1.403e-04, forward_time=0.115, loss_ctc=60.824, loss_att=60.055, acc=0.659, loss=60.286, backward_time=0.134, grad_norm=27.484, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.192e-04, train_time=3.101
+[gpuc01:0/16] 2024-01-24 00:12:33,691 (trainer:737) INFO: 6epoch:train:20801-20900batch: iter_time=1.364e-04, forward_time=0.115, loss_ctc=78.767, loss_att=66.285, acc=0.659, loss=70.030, backward_time=0.135, grad_norm=33.503, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.190e-04, train_time=3.113
+[gpuc01:0/16] 2024-01-24 00:15:09,060 (trainer:737) INFO: 6epoch:train:20901-21000batch: iter_time=1.327e-04, forward_time=0.115, loss_ctc=66.299, loss_att=58.908, acc=0.672, loss=61.125, backward_time=0.135, grad_norm=23.786, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.189e-04, train_time=3.107
+[gpuc01:0/16] 2024-01-24 00:17:44,092 (trainer:737) INFO: 6epoch:train:21001-21100batch: iter_time=1.257e-04, forward_time=0.115, loss_ctc=66.187, loss_att=59.289, acc=0.671, loss=61.358, backward_time=0.134, grad_norm=27.112, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.188e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 00:20:19,365 (trainer:737) INFO: 6epoch:train:21101-21200batch: iter_time=1.362e-04, forward_time=0.116, loss_ctc=65.860, loss_att=59.707, acc=0.667, loss=61.553, backward_time=0.134, grad_norm=26.890, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.187e-04, train_time=3.105
+[gpuc01:0/16] 2024-01-24 00:22:54,881 (trainer:737) INFO: 6epoch:train:21201-21300batch: iter_time=1.284e-04, forward_time=0.115, loss_ctc=67.972, loss_att=60.282, acc=0.657, loss=62.589, backward_time=0.134, grad_norm=29.197, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.185e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 00:25:31,069 (trainer:737) INFO: 6epoch:train:21301-21400batch: iter_time=1.287e-04, forward_time=0.116, loss_ctc=64.565, loss_att=62.549, acc=0.676, loss=63.153, backward_time=0.135, grad_norm=27.114, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.184e-04, train_time=3.124
+[gpuc01:0/16] 2024-01-24 00:28:07,121 (trainer:737) INFO: 6epoch:train:21401-21500batch: iter_time=1.314e-04, forward_time=0.115, loss_ctc=50.729, loss_att=47.251, acc=0.663, loss=48.294, backward_time=0.134, grad_norm=22.153, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.183e-04, train_time=3.121
+[gpuc01:0/16] 2024-01-24 00:30:42,853 (trainer:737) INFO: 6epoch:train:21501-21600batch: iter_time=1.373e-04, forward_time=0.115, loss_ctc=82.844, loss_att=70.413, acc=0.655, loss=74.142, backward_time=0.135, grad_norm=31.786, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.182e-04, train_time=3.114
+[gpuc01:0/16] 2024-01-24 00:33:18,296 (trainer:737) INFO: 6epoch:train:21601-21700batch: iter_time=1.318e-04, forward_time=0.116, loss_ctc=65.284, loss_att=72.001, acc=0.652, loss=69.986, backward_time=0.135, grad_norm=26.827, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.181e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 00:35:55,062 (trainer:737) INFO: 6epoch:train:21701-21800batch: iter_time=1.283e-04, forward_time=0.115, loss_ctc=65.321, loss_att=62.566, acc=0.660, loss=63.393, backward_time=0.135, grad_norm=25.146, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.179e-04, train_time=3.135
+[gpuc01:0/16] 2024-01-24 00:38:31,114 (trainer:737) INFO: 6epoch:train:21801-21900batch: iter_time=1.263e-04, forward_time=0.116, loss_ctc=69.262, loss_att=72.946, acc=0.654, loss=71.841, backward_time=0.135, grad_norm=27.390, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.178e-04, train_time=3.121
+[gpuc01:0/16] 2024-01-24 00:41:07,752 (trainer:737) INFO: 6epoch:train:21901-22000batch: iter_time=1.253e-04, forward_time=0.116, loss_ctc=82.920, loss_att=79.995, acc=0.655, loss=80.873, backward_time=0.136, grad_norm=29.769, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.066, optim0_lr0=4.177e-04, train_time=3.133
+[gpuc01:0/16] 2024-01-24 00:43:43,442 (trainer:737) INFO: 6epoch:train:22001-22100batch: iter_time=1.331e-04, forward_time=0.116, loss_ctc=60.327, loss_att=55.320, acc=0.682, loss=56.822, backward_time=0.135, grad_norm=23.394, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.176e-04, train_time=3.114
+[gpuc01:0/16] 2024-01-24 00:46:19,033 (trainer:737) INFO: 6epoch:train:22101-22200batch: iter_time=1.229e-04, forward_time=0.117, loss_ctc=68.023, loss_att=71.627, acc=0.648, loss=70.546, backward_time=0.136, grad_norm=27.268, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.174e-04, train_time=3.112
+[gpuc01:0/16] 2024-01-24 00:48:54,560 (trainer:737) INFO: 6epoch:train:22201-22300batch: iter_time=1.208e-04, forward_time=0.116, loss_ctc=63.101, loss_att=60.153, acc=0.669, loss=61.038, backward_time=0.135, grad_norm=25.710, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.173e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 00:51:29,187 (trainer:737) INFO: 6epoch:train:22301-22400batch: iter_time=1.332e-04, forward_time=0.115, loss_ctc=60.221, loss_att=53.565, acc=0.663, loss=55.562, backward_time=0.134, grad_norm=25.203, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.172e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-24 00:54:22,618 (trainer:737) INFO: 6epoch:train:22401-22500batch: iter_time=1.173e-04, forward_time=0.126, loss_ctc=70.649, loss_att=57.735, acc=0.668, loss=61.609, backward_time=0.137, grad_norm=29.854, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.171e-04, train_time=3.468
+[gpuc01:0/16] 2024-01-24 00:54:42,831 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpuc01:0/16] 2024-01-24 00:55:02,173 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 00:55:05,889 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f671d4674f0>)
+[gpuc01:0/16] 2024-01-24 00:55:05,889 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpuc01:0/16] 2024-01-24 00:55:05,895 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 01:01:59,395 (trainer:737) INFO: 6epoch:train:22501-22600batch: iter_time=2.812, forward_time=0.116, loss_ctc=66.171, loss_att=64.628, acc=0.661, loss=65.091, backward_time=0.135, grad_norm=30.821, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.170e-04, train_time=9.135
+[gpuc01:0/16] 2024-01-24 01:04:35,655 (trainer:737) INFO: 6epoch:train:22601-22700batch: iter_time=1.135e-04, forward_time=0.118, loss_ctc=64.447, loss_att=67.321, acc=0.651, loss=66.459, backward_time=0.135, grad_norm=26.598, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.168e-04, train_time=3.125
+[gpuc01:0/16] 2024-01-24 01:07:11,556 (trainer:737) INFO: 6epoch:train:22701-22800batch: iter_time=1.266e-04, forward_time=0.116, loss_ctc=70.253, loss_att=74.791, acc=0.634, loss=73.430, backward_time=0.135, grad_norm=27.533, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.167e-04, train_time=3.118
+[gpuc01:0/16] 2024-01-24 01:09:45,778 (trainer:737) INFO: 6epoch:train:22801-22900batch: iter_time=1.401e-04, forward_time=0.115, loss_ctc=66.476, loss_att=61.494, acc=0.650, loss=62.989, backward_time=0.134, grad_norm=28.590, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.166e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-24 01:12:21,434 (trainer:737) INFO: 6epoch:train:22901-23000batch: iter_time=1.400e-04, forward_time=0.115, loss_ctc=67.536, loss_att=63.096, acc=0.657, loss=64.428, backward_time=0.135, grad_norm=27.430, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.067, optim0_lr0=4.165e-04, train_time=3.113
+[gpuc01:0/16] 2024-01-24 01:14:56,933 (trainer:737) INFO: 6epoch:train:23001-23100batch: iter_time=1.323e-04, forward_time=0.115, loss_ctc=61.181, loss_att=63.662, acc=0.654, loss=62.918, backward_time=0.135, grad_norm=24.921, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.164e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 01:17:32,443 (trainer:737) INFO: 6epoch:train:23101-23200batch: iter_time=1.357e-04, forward_time=0.115, loss_ctc=60.405, loss_att=63.556, acc=0.629, loss=62.611, backward_time=0.134, grad_norm=24.305, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.162e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 01:20:08,895 (trainer:737) INFO: 6epoch:train:23201-23300batch: iter_time=1.350e-04, forward_time=0.115, loss_ctc=60.593, loss_att=59.410, acc=0.658, loss=59.765, backward_time=0.134, grad_norm=25.408, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.161e-04, train_time=3.129
+[gpuc01:0/16] 2024-01-24 01:22:44,729 (trainer:737) INFO: 6epoch:train:23301-23400batch: iter_time=1.392e-04, forward_time=0.115, loss_ctc=76.887, loss_att=64.086, acc=0.657, loss=67.927, backward_time=0.135, grad_norm=32.022, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.160e-04, train_time=3.116
+[gpuc01:0/16] 2024-01-24 01:25:20,179 (trainer:737) INFO: 6epoch:train:23401-23500batch: iter_time=1.336e-04, forward_time=0.116, loss_ctc=66.955, loss_att=60.789, acc=0.660, loss=62.639, backward_time=0.135, grad_norm=26.782, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.159e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 01:27:55,585 (trainer:737) INFO: 6epoch:train:23501-23600batch: iter_time=1.347e-04, forward_time=0.115, loss_ctc=66.896, loss_att=59.918, acc=0.664, loss=62.011, backward_time=0.134, grad_norm=27.328, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.158e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-24 01:30:30,899 (trainer:737) INFO: 6epoch:train:23601-23700batch: iter_time=1.248e-04, forward_time=0.115, loss_ctc=65.074, loss_att=60.229, acc=0.659, loss=61.682, backward_time=0.135, grad_norm=26.348, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.156e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-24 01:33:05,101 (trainer:737) INFO: 6epoch:train:23701-23800batch: iter_time=1.335e-04, forward_time=0.115, loss_ctc=67.362, loss_att=60.363, acc=0.650, loss=62.463, backward_time=0.134, grad_norm=29.290, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.155e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-24 01:35:41,213 (trainer:737) INFO: 6epoch:train:23801-23900batch: iter_time=1.247e-04, forward_time=0.115, loss_ctc=63.651, loss_att=62.260, acc=0.667, loss=62.677, backward_time=0.135, grad_norm=26.042, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.154e-04, train_time=3.122
+[gpuc01:0/16] 2024-01-24 01:38:17,174 (trainer:737) INFO: 6epoch:train:23901-24000batch: iter_time=1.301e-04, forward_time=0.115, loss_ctc=49.928, loss_att=46.702, acc=0.659, loss=47.670, backward_time=0.134, grad_norm=21.655, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.153e-04, train_time=3.119
+[gpuc01:0/16] 2024-01-24 01:40:53,004 (trainer:737) INFO: 6epoch:train:24001-24100batch: iter_time=1.312e-04, forward_time=0.115, loss_ctc=82.737, loss_att=70.259, acc=0.646, loss=74.002, backward_time=0.135, grad_norm=31.477, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.152e-04, train_time=3.116
+[gpuc01:0/16] 2024-01-24 01:43:29,255 (trainer:737) INFO: 6epoch:train:24101-24200batch: iter_time=1.350e-04, forward_time=0.115, loss_ctc=64.696, loss_att=71.255, acc=0.643, loss=69.287, backward_time=0.135, grad_norm=27.576, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.150e-04, train_time=3.125
+[gpuc01:0/16] 2024-01-24 01:46:05,024 (trainer:737) INFO: 6epoch:train:24201-24300batch: iter_time=1.284e-04, forward_time=0.115, loss_ctc=64.682, loss_att=62.498, acc=0.646, loss=63.153, backward_time=0.134, grad_norm=25.652, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.149e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-24 01:48:41,232 (trainer:737) INFO: 6epoch:train:24301-24400batch: iter_time=1.282e-04, forward_time=0.116, loss_ctc=68.243, loss_att=71.299, acc=0.644, loss=70.382, backward_time=0.135, grad_norm=27.879, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.148e-04, train_time=3.124
+[gpuc01:0/16] 2024-01-24 01:51:17,483 (trainer:737) INFO: 6epoch:train:24401-24500batch: iter_time=1.346e-04, forward_time=0.116, loss_ctc=82.720, loss_att=75.653, acc=0.662, loss=77.773, backward_time=0.135, grad_norm=31.123, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.147e-04, train_time=3.125
+[gpuc01:0/16] 2024-01-24 01:53:53,060 (trainer:737) INFO: 6epoch:train:24501-24600batch: iter_time=1.306e-04, forward_time=0.115, loss_ctc=59.742, loss_att=54.806, acc=0.675, loss=56.286, backward_time=0.134, grad_norm=24.838, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.146e-04, train_time=3.111
+[gpuc01:0/16] 2024-01-24 01:56:29,015 (trainer:737) INFO: 6epoch:train:24601-24700batch: iter_time=1.308e-04, forward_time=0.116, loss_ctc=67.227, loss_att=70.291, acc=0.635, loss=69.372, backward_time=0.136, grad_norm=27.587, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.145e-04, train_time=3.119
+[gpuc01:0/16] 2024-01-24 01:59:04,463 (trainer:737) INFO: 6epoch:train:24701-24800batch: iter_time=1.287e-04, forward_time=0.116, loss_ctc=62.685, loss_att=61.261, acc=0.656, loss=61.688, backward_time=0.135, grad_norm=25.407, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.143e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 02:01:41,128 (trainer:737) INFO: 6epoch:train:24801-24900batch: iter_time=1.292e-04, forward_time=0.119, loss_ctc=59.049, loss_att=52.485, acc=0.662, loss=54.454, backward_time=0.134, grad_norm=24.186, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.142e-04, train_time=3.133
+[gpuc01:0/16] 2024-01-24 02:04:45,117 (trainer:737) INFO: 6epoch:train:24901-25000batch: iter_time=1.220e-04, forward_time=0.242, loss_ctc=69.694, loss_att=56.976, acc=0.666, loss=60.791, backward_time=0.149, grad_norm=30.595, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.074, optim0_lr0=4.141e-04, train_time=3.666
+[gpuc01:0/16] 2024-01-24 02:05:03,619 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpuc01:0/16] 2024-01-24 02:05:22,860 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 02:05:26,480 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f671d10a920>)
+[gpuc01:0/16] 2024-01-24 02:05:26,480 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpuc01:0/16] 2024-01-24 02:05:26,487 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 02:11:21,462 (trainer:737) INFO: 6epoch:train:25001-25100batch: iter_time=2.390, forward_time=0.116, loss_ctc=65.598, loss_att=61.703, acc=0.672, loss=62.871, backward_time=0.135, grad_norm=28.508, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.140e-04, train_time=7.941
+[gpuc01:0/16] 2024-01-24 02:13:57,571 (trainer:737) INFO: 6epoch:train:25101-25200batch: iter_time=1.453e-04, forward_time=0.115, loss_ctc=63.577, loss_att=62.314, acc=0.660, loss=62.693, backward_time=0.135, grad_norm=25.073, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.139e-04, train_time=3.122
+[gpuc01:0/16] 2024-01-24 02:16:32,605 (trainer:737) INFO: 6epoch:train:25201-25300batch: iter_time=1.466e-04, forward_time=0.115, loss_ctc=69.201, loss_att=72.547, acc=0.639, loss=71.543, backward_time=0.135, grad_norm=26.459, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.137e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 02:19:07,393 (trainer:737) INFO: 6epoch:train:25301-25400batch: iter_time=1.305e-04, forward_time=0.115, loss_ctc=66.136, loss_att=60.739, acc=0.651, loss=62.358, backward_time=0.134, grad_norm=26.979, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.136e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-24 02:21:42,934 (trainer:737) INFO: 6epoch:train:25401-25500batch: iter_time=1.425e-04, forward_time=0.115, loss_ctc=67.232, loss_att=62.070, acc=0.663, loss=63.618, backward_time=0.135, grad_norm=27.644, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.135e-04, train_time=3.111
+[gpuc01:0/16] 2024-01-24 02:24:17,443 (trainer:737) INFO: 6epoch:train:25501-25600batch: iter_time=1.432e-04, forward_time=0.115, loss_ctc=61.244, loss_att=61.649, acc=0.662, loss=61.527, backward_time=0.135, grad_norm=24.424, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.065, optim0_lr0=4.134e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 02:26:52,136 (trainer:737) INFO: 6epoch:train:25601-25700batch: iter_time=1.456e-04, forward_time=0.114, loss_ctc=61.160, loss_att=62.341, acc=0.636, loss=61.987, backward_time=0.134, grad_norm=23.442, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.133e-04, train_time=3.094
+[gpuc01:0/16] 2024-01-24 02:29:28,398 (trainer:737) INFO: 6epoch:train:25701-25800batch: iter_time=1.417e-04, forward_time=0.114, loss_ctc=59.173, loss_att=58.436, acc=0.663, loss=58.657, backward_time=0.134, grad_norm=26.189, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.132e-04, train_time=3.125
+[gpuc01:0/16] 2024-01-24 02:32:04,778 (trainer:737) INFO: 6epoch:train:25801-25900batch: iter_time=1.510e-04, forward_time=0.115, loss_ctc=77.218, loss_att=64.276, acc=0.658, loss=68.159, backward_time=0.135, grad_norm=34.663, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.130e-04, train_time=3.127
+[gpuc01:0/16] 2024-01-24 02:34:39,527 (trainer:737) INFO: 6epoch:train:25901-26000batch: iter_time=1.468e-04, forward_time=0.114, loss_ctc=64.943, loss_att=58.116, acc=0.666, loss=60.164, backward_time=0.134, grad_norm=27.747, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.066, optim0_lr0=4.129e-04, train_time=3.095
+[gpuc01:0/16] 2024-01-24 02:37:15,813 (trainer:737) INFO: 6epoch:train:26001-26100batch: iter_time=1.388e-04, forward_time=0.115, loss_ctc=65.795, loss_att=59.074, acc=0.666, loss=61.090, backward_time=0.135, grad_norm=26.591, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.128e-04, train_time=3.125
+[gpuc01:0/16] 2024-01-24 02:39:52,206 (trainer:737) INFO: 6epoch:train:26101-26200batch: iter_time=1.459e-04, forward_time=0.114, loss_ctc=64.570, loss_att=58.821, acc=0.661, loss=60.546, backward_time=0.135, grad_norm=26.530, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.127e-04, train_time=3.128
+[gpuc01:0/16] 2024-01-24 02:42:28,810 (trainer:737) INFO: 6epoch:train:26201-26300batch: iter_time=1.577e-04, forward_time=0.114, loss_ctc=66.696, loss_att=60.263, acc=0.647, loss=62.193, backward_time=0.135, grad_norm=27.388, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.126e-04, train_time=3.132
+[gpuc01:0/16] 2024-01-24 02:45:05,964 (trainer:737) INFO: 6epoch:train:26301-26400batch: iter_time=1.635e-04, forward_time=0.114, loss_ctc=63.553, loss_att=61.698, acc=0.672, loss=62.255, backward_time=0.135, grad_norm=28.458, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.124e-04, train_time=3.143
+[gpuc01:0/16] 2024-01-24 02:47:46,599 (trainer:737) INFO: 6epoch:train:26401-26500batch: iter_time=1.523e-04, forward_time=0.113, loss_ctc=49.790, loss_att=45.864, acc=0.663, loss=47.041, backward_time=0.134, grad_norm=24.912, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.123e-04, train_time=3.212
+[gpuc01:0/16] 2024-01-24 02:50:27,045 (trainer:737) INFO: 6epoch:train:26501-26600batch: iter_time=1.576e-04, forward_time=0.119, loss_ctc=80.292, loss_att=68.202, acc=0.651, loss=71.829, backward_time=0.136, grad_norm=31.810, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.122e-04, train_time=3.209
+[gpuc01:0/16] 2024-01-24 02:53:03,054 (trainer:737) INFO: 6epoch:train:26601-26700batch: iter_time=1.502e-04, forward_time=0.114, loss_ctc=64.490, loss_att=71.497, acc=0.645, loss=69.395, backward_time=0.135, grad_norm=27.443, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.121e-04, train_time=3.120
+[gpuc01:0/16] 2024-01-24 02:55:39,126 (trainer:737) INFO: 6epoch:train:26701-26800batch: iter_time=1.593e-04, forward_time=0.115, loss_ctc=64.295, loss_att=62.138, acc=0.649, loss=62.785, backward_time=0.134, grad_norm=25.779, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.120e-04, train_time=3.121
+[gpuc01:0/16] 2024-01-24 02:58:15,278 (trainer:737) INFO: 6epoch:train:26801-26900batch: iter_time=1.470e-04, forward_time=0.116, loss_ctc=67.996, loss_att=71.486, acc=0.646, loss=70.439, backward_time=0.135, grad_norm=28.717, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.119e-04, train_time=3.123
+[gpuc01:0/16] 2024-01-24 03:00:50,308 (trainer:737) INFO: 6epoch:train:26901-27000batch: iter_time=1.494e-04, forward_time=0.115, loss_ctc=81.268, loss_att=74.051, acc=0.665, loss=76.216, backward_time=0.136, grad_norm=30.372, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.117e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 03:03:25,918 (trainer:737) INFO: 6epoch:train:27001-27100batch: iter_time=1.544e-04, forward_time=0.115, loss_ctc=59.550, loss_att=54.472, acc=0.677, loss=55.996, backward_time=0.134, grad_norm=23.646, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.116e-04, train_time=3.112
+[gpuc01:0/16] 2024-01-24 03:06:01,756 (trainer:737) INFO: 6epoch:train:27101-27200batch: iter_time=1.536e-04, forward_time=0.116, loss_ctc=67.319, loss_att=70.525, acc=0.638, loss=69.563, backward_time=0.136, grad_norm=28.593, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.115e-04, train_time=3.117
+[gpuc01:0/16] 2024-01-24 03:08:37,472 (trainer:737) INFO: 6epoch:train:27201-27300batch: iter_time=1.474e-04, forward_time=0.115, loss_ctc=62.476, loss_att=60.427, acc=0.659, loss=61.042, backward_time=0.135, grad_norm=25.748, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.114e-04, train_time=3.114
+[gpuc01:0/16] 2024-01-24 03:11:11,610 (trainer:737) INFO: 6epoch:train:27301-27400batch: iter_time=1.516e-04, forward_time=0.114, loss_ctc=59.349, loss_att=52.159, acc=0.667, loss=54.316, backward_time=0.134, grad_norm=23.893, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.113e-04, train_time=3.083
+[gpuc01:0/16] 2024-01-24 03:13:52,330 (trainer:737) INFO: 6epoch:train:27401-27500batch: iter_time=5.021e-04, forward_time=0.116, loss_ctc=68.750, loss_att=56.793, acc=0.668, loss=60.380, backward_time=0.134, grad_norm=32.537, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.112e-04, train_time=3.214
+[gpuc01:0/16] 2024-01-24 03:14:05,427 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpuc01:0/16] 2024-01-24 03:14:23,908 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 03:14:27,636 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6716624820>)
+[gpuc01:0/16] 2024-01-24 03:14:27,636 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpuc01:0/16] 2024-01-24 03:14:27,642 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 03:20:40,355 (trainer:737) INFO: 6epoch:train:27501-27600batch: iter_time=2.518, forward_time=0.115, loss_ctc=64.816, loss_att=63.361, acc=0.677, loss=63.797, backward_time=0.135, grad_norm=29.929, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.111e-04, train_time=8.160
+[gpuc01:0/16] 2024-01-24 03:23:15,142 (trainer:737) INFO: 6epoch:train:27601-27700batch: iter_time=1.483e-04, forward_time=0.115, loss_ctc=63.203, loss_att=65.798, acc=0.665, loss=65.019, backward_time=0.136, grad_norm=25.611, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.109e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-24 03:25:50,162 (trainer:737) INFO: 6epoch:train:27701-27800batch: iter_time=1.535e-04, forward_time=0.115, loss_ctc=69.370, loss_att=75.157, acc=0.643, loss=73.421, backward_time=0.136, grad_norm=27.343, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.108e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 03:28:23,668 (trainer:737) INFO: 6epoch:train:27801-27900batch: iter_time=1.526e-04, forward_time=0.114, loss_ctc=66.034, loss_att=66.428, acc=0.654, loss=66.310, backward_time=0.135, grad_norm=27.624, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.107e-04, train_time=3.070
+[gpuc01:0/16] 2024-01-24 03:30:58,733 (trainer:737) INFO: 6epoch:train:27901-28000batch: iter_time=1.595e-04, forward_time=0.115, loss_ctc=66.876, loss_att=62.213, acc=0.667, loss=63.612, backward_time=0.135, grad_norm=27.073, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.106e-04, train_time=3.101
+[gpuc01:0/16] 2024-01-24 03:33:34,147 (trainer:737) INFO: 6epoch:train:28001-28100batch: iter_time=1.562e-04, forward_time=0.115, loss_ctc=60.327, loss_att=64.780, acc=0.665, loss=63.444, backward_time=0.136, grad_norm=25.420, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.105e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-24 03:36:08,970 (trainer:737) INFO: 6epoch:train:28101-28200batch: iter_time=1.538e-04, forward_time=0.114, loss_ctc=60.097, loss_att=61.213, acc=0.655, loss=60.878, backward_time=0.135, grad_norm=22.684, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.104e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-24 03:38:44,488 (trainer:737) INFO: 6epoch:train:28201-28300batch: iter_time=1.395e-04, forward_time=0.118, loss_ctc=59.318, loss_att=58.796, acc=0.665, loss=58.952, backward_time=0.135, grad_norm=24.942, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.102e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 03:41:18,992 (trainer:737) INFO: 6epoch:train:28301-28400batch: iter_time=1.390e-04, forward_time=0.115, loss_ctc=75.873, loss_att=65.373, acc=0.661, loss=68.523, backward_time=0.135, grad_norm=31.450, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.101e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 03:43:53,511 (trainer:737) INFO: 6epoch:train:28401-28500batch: iter_time=1.486e-04, forward_time=0.114, loss_ctc=65.044, loss_att=58.744, acc=0.678, loss=60.634, backward_time=0.135, grad_norm=24.576, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.100e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 03:46:28,539 (trainer:737) INFO: 6epoch:train:28501-28600batch: iter_time=1.432e-04, forward_time=0.115, loss_ctc=65.228, loss_att=59.062, acc=0.677, loss=60.911, backward_time=0.135, grad_norm=27.653, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.065, optim0_lr0=4.099e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 03:49:03,071 (trainer:737) INFO: 6epoch:train:28601-28700batch: iter_time=1.529e-04, forward_time=0.115, loss_ctc=64.507, loss_att=59.240, acc=0.674, loss=60.820, backward_time=0.135, grad_norm=27.762, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.098e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 03:51:37,959 (trainer:737) INFO: 6epoch:train:28701-28800batch: iter_time=1.527e-04, forward_time=0.114, loss_ctc=65.716, loss_att=58.876, acc=0.665, loss=60.928, backward_time=0.135, grad_norm=27.210, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.097e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-24 03:54:12,869 (trainer:737) INFO: 6epoch:train:28801-28900batch: iter_time=1.446e-04, forward_time=0.114, loss_ctc=62.308, loss_att=60.868, acc=0.680, loss=61.300, backward_time=0.136, grad_norm=24.739, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.096e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-24 03:56:48,003 (trainer:737) INFO: 6epoch:train:28901-29000batch: iter_time=1.536e-04, forward_time=0.114, loss_ctc=49.070, loss_att=45.735, acc=0.670, loss=46.736, backward_time=0.134, grad_norm=21.079, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.094e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-24 03:59:23,466 (trainer:737) INFO: 6epoch:train:29001-29100batch: iter_time=1.652e-04, forward_time=0.116, loss_ctc=80.262, loss_att=69.024, acc=0.659, loss=72.396, backward_time=0.136, grad_norm=32.730, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.093e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 04:01:58,559 (trainer:737) INFO: 6epoch:train:29101-29200batch: iter_time=1.539e-04, forward_time=0.116, loss_ctc=63.860, loss_att=69.779, acc=0.659, loss=68.003, backward_time=0.136, grad_norm=27.544, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.092e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-24 04:04:33,215 (trainer:737) INFO: 6epoch:train:29201-29300batch: iter_time=1.549e-04, forward_time=0.115, loss_ctc=63.313, loss_att=61.076, acc=0.668, loss=61.747, backward_time=0.135, grad_norm=23.747, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.091e-04, train_time=3.093
+[gpuc01:0/16] 2024-01-24 04:07:07,930 (trainer:737) INFO: 6epoch:train:29301-29400batch: iter_time=1.496e-04, forward_time=0.116, loss_ctc=67.853, loss_att=71.335, acc=0.658, loss=70.291, backward_time=0.136, grad_norm=29.734, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.090e-04, train_time=3.094
+[gpuc01:0/16] 2024-01-24 04:09:43,005 (trainer:737) INFO: 6epoch:train:29401-29500batch: iter_time=1.472e-04, forward_time=0.116, loss_ctc=81.134, loss_att=77.662, acc=0.664, loss=78.704, backward_time=0.136, grad_norm=29.072, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.089e-04, train_time=3.101
+[gpuc01:0/16] 2024-01-24 04:12:18,526 (trainer:737) INFO: 6epoch:train:29501-29600batch: iter_time=1.443e-04, forward_time=0.117, loss_ctc=59.054, loss_att=54.377, acc=0.689, loss=55.780, backward_time=0.135, grad_norm=23.222, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.088e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 04:14:54,996 (trainer:737) INFO: 6epoch:train:29601-29700batch: iter_time=1.500e-04, forward_time=0.116, loss_ctc=65.568, loss_att=70.464, acc=0.654, loss=68.995, backward_time=0.136, grad_norm=27.595, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.086e-04, train_time=3.129
+[gpuc01:0/16] 2024-01-24 04:17:30,680 (trainer:737) INFO: 6epoch:train:29701-29800batch: iter_time=1.468e-04, forward_time=0.115, loss_ctc=61.800, loss_att=59.293, acc=0.675, loss=60.045, backward_time=0.135, grad_norm=25.507, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.085e-04, train_time=3.113
+[gpuc01:0/16] 2024-01-24 04:20:07,093 (trainer:737) INFO: 6epoch:train:29801-29900batch: iter_time=1.518e-04, forward_time=0.115, loss_ctc=58.439, loss_att=51.834, acc=0.670, loss=53.816, backward_time=0.135, grad_norm=24.969, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.084e-04, train_time=3.128
+[gpuc01:0/16] 2024-01-24 04:22:43,514 (trainer:737) INFO: 6epoch:train:29901-30000batch: iter_time=1.572e-04, forward_time=0.116, loss_ctc=69.224, loss_att=57.503, acc=0.671, loss=61.019, backward_time=0.135, grad_norm=34.094, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.066, optim0_lr0=4.083e-04, train_time=3.128
+[gpuc01:0/16] 2024-01-24 05:01:08,781 (trainer:343) INFO: 6epoch results: [train] iter_time=0.080, forward_time=0.117, loss_ctc=68.760, loss_att=65.566, acc=0.648, loss=66.524, backward_time=0.135, grad_norm=27.664, clip=100.000, loss_scale=3.056e+17, optim_step_time=0.066, optim0_lr0=4.268e-04, train_time=3.231, time=13 hours, 27 minutes and 57.18 seconds, total_count=105000, gpu_max_cached_mem_GB=25.098, [valid] loss_ctc=60.976, cer_ctc=0.313, loss_att=55.410, acc=0.561, cer=0.389, wer=1.000, loss=57.080, time=38 minutes and 15.79 seconds, total_count=32697, gpu_max_cached_mem_GB=25.098
+[gpuc01:0/16] 2024-01-24 05:01:30,989 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpuc01:0/16] 2024-01-24 05:01:31,017 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/1epoch.pth
+[gpuc01:0/16] 2024-01-24 05:01:31,018 (trainer:272) INFO: 7/45epoch started. Estimated time to finish: 3 weeks, 1 day and 22 hours
+[gpuc01:0/16] 2024-01-24 05:01:31,038 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpuc01:0/16] 2024-01-24 05:01:49,992 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 05:01:53,545 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f671d10be50>)
+[gpuc01:0/16] 2024-01-24 05:01:53,545 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpuc01:0/16] 2024-01-24 05:01:53,551 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 05:06:14,119 (trainer:737) INFO: 7epoch:train:1-100batch: iter_time=1.257, forward_time=0.119, loss_ctc=66.859, loss_att=70.218, acc=0.641, loss=69.210, backward_time=0.146, grad_norm=32.746, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.068, optim0_lr0=4.082e-04, train_time=5.661
+[gpuc01:0/16] 2024-01-24 05:08:49,305 (trainer:737) INFO: 7epoch:train:101-200batch: iter_time=1.408e-04, forward_time=0.117, loss_ctc=65.274, loss_att=62.674, acc=0.670, loss=63.454, backward_time=0.137, grad_norm=26.045, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.081e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-24 05:11:24,492 (trainer:737) INFO: 7epoch:train:201-300batch: iter_time=1.399e-04, forward_time=0.119, loss_ctc=65.223, loss_att=62.026, acc=0.645, loss=62.985, backward_time=0.136, grad_norm=25.975, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.080e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-24 05:13:58,265 (trainer:737) INFO: 7epoch:train:301-400batch: iter_time=1.387e-04, forward_time=0.117, loss_ctc=71.611, loss_att=71.182, acc=0.663, loss=71.311, backward_time=0.137, grad_norm=30.300, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.078e-04, train_time=3.075
+[gpuc01:0/16] 2024-01-24 05:16:32,060 (trainer:737) INFO: 7epoch:train:401-500batch: iter_time=1.425e-04, forward_time=0.117, loss_ctc=80.914, loss_att=68.155, acc=0.667, loss=71.983, backward_time=0.136, grad_norm=30.921, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.077e-04, train_time=3.076
+[gpuc01:0/16] 2024-01-24 05:19:06,229 (trainer:737) INFO: 7epoch:train:501-600batch: iter_time=1.414e-04, forward_time=0.117, loss_ctc=62.229, loss_att=62.835, acc=0.677, loss=62.653, backward_time=0.136, grad_norm=23.407, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.076e-04, train_time=3.083
+[gpuc01:0/16] 2024-01-24 05:21:40,076 (trainer:737) INFO: 7epoch:train:601-700batch: iter_time=1.450e-04, forward_time=0.116, loss_ctc=61.912, loss_att=57.731, acc=0.635, loss=58.986, backward_time=0.135, grad_norm=26.029, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.075e-04, train_time=3.077
+[gpuc01:0/16] 2024-01-24 05:24:14,988 (trainer:737) INFO: 7epoch:train:701-800batch: iter_time=1.481e-04, forward_time=0.117, loss_ctc=70.174, loss_att=62.281, acc=0.654, loss=64.649, backward_time=0.136, grad_norm=27.461, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.074e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-24 05:26:49,446 (trainer:737) INFO: 7epoch:train:801-900batch: iter_time=1.372e-04, forward_time=0.117, loss_ctc=72.409, loss_att=66.252, acc=0.666, loss=68.099, backward_time=0.136, grad_norm=28.246, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.073e-04, train_time=3.089
+[gpuc01:0/16] 2024-01-24 05:29:24,456 (trainer:737) INFO: 7epoch:train:901-1000batch: iter_time=1.364e-04, forward_time=0.116, loss_ctc=58.005, loss_att=59.206, acc=0.661, loss=58.845, backward_time=0.135, grad_norm=23.707, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.072e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 05:31:58,648 (trainer:737) INFO: 7epoch:train:1001-1100batch: iter_time=1.427e-04, forward_time=0.116, loss_ctc=62.130, loss_att=66.269, acc=0.653, loss=65.027, backward_time=0.135, grad_norm=24.829, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.071e-04, train_time=3.084
+[gpuc01:0/16] 2024-01-24 05:34:35,098 (trainer:737) INFO: 7epoch:train:1101-1200batch: iter_time=1.325e-04, forward_time=0.116, loss_ctc=60.464, loss_att=60.647, acc=0.660, loss=60.592, backward_time=0.135, grad_norm=23.909, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.069e-04, train_time=3.129
+[gpuc01:0/16] 2024-01-24 05:37:07,867 (trainer:737) INFO: 7epoch:train:1201-1300batch: iter_time=1.337e-04, forward_time=0.116, loss_ctc=63.745, loss_att=58.330, acc=0.670, loss=59.955, backward_time=0.135, grad_norm=26.927, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.068e-04, train_time=3.055
+[gpuc01:0/16] 2024-01-24 05:39:41,437 (trainer:737) INFO: 7epoch:train:1301-1400batch: iter_time=1.374e-04, forward_time=0.116, loss_ctc=79.571, loss_att=74.544, acc=0.640, loss=76.052, backward_time=0.136, grad_norm=31.822, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.067e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-24 05:42:14,744 (trainer:737) INFO: 7epoch:train:1401-1500batch: iter_time=1.418e-04, forward_time=0.116, loss_ctc=63.185, loss_att=72.439, acc=0.630, loss=69.663, backward_time=0.136, grad_norm=29.250, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.066e-04, train_time=3.066
+[gpuc01:0/16] 2024-01-24 05:44:50,072 (trainer:737) INFO: 7epoch:train:1501-1600batch: iter_time=1.342e-04, forward_time=0.115, loss_ctc=64.412, loss_att=55.972, acc=0.668, loss=58.504, backward_time=0.135, grad_norm=28.477, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.065e-04, train_time=3.106
+[gpuc01:0/16] 2024-01-24 05:47:25,335 (trainer:737) INFO: 7epoch:train:1601-1700batch: iter_time=1.423e-04, forward_time=0.116, loss_ctc=70.978, loss_att=77.430, acc=0.638, loss=75.495, backward_time=0.136, grad_norm=30.908, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.068, optim0_lr0=4.064e-04, train_time=3.105
+[gpuc01:0/16] 2024-01-24 05:50:00,282 (trainer:737) INFO: 7epoch:train:1701-1800batch: iter_time=1.566e-04, forward_time=0.117, loss_ctc=69.694, loss_att=67.331, acc=0.681, loss=68.040, backward_time=0.137, grad_norm=29.623, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.063e-04, train_time=3.099
+[gpuc01:0/16] 2024-01-24 05:52:36,151 (trainer:737) INFO: 7epoch:train:1801-1900batch: iter_time=1.582e-04, forward_time=0.117, loss_ctc=70.814, loss_att=67.575, acc=0.648, loss=68.547, backward_time=0.137, grad_norm=25.900, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.062e-04, train_time=3.117
+[gpuc01:0/16] 2024-01-24 05:55:11,746 (trainer:737) INFO: 7epoch:train:1901-2000batch: iter_time=1.607e-04, forward_time=0.116, loss_ctc=63.408, loss_att=65.672, acc=0.658, loss=64.993, backward_time=0.136, grad_norm=24.525, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.061e-04, train_time=3.112
+[gpuc01:0/16] 2024-01-24 05:57:46,634 (trainer:737) INFO: 7epoch:train:2001-2100batch: iter_time=1.545e-04, forward_time=0.119, loss_ctc=63.068, loss_att=65.213, acc=0.673, loss=64.570, backward_time=0.136, grad_norm=24.620, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.059e-04, train_time=3.098
+[gpuc01:0/16] 2024-01-24 06:00:21,444 (trainer:737) INFO: 7epoch:train:2101-2200batch: iter_time=1.522e-04, forward_time=0.117, loss_ctc=63.779, loss_att=69.923, acc=0.640, loss=68.080, backward_time=0.136, grad_norm=25.762, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.058e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-24 06:02:56,250 (trainer:737) INFO: 7epoch:train:2201-2300batch: iter_time=1.492e-04, forward_time=0.116, loss_ctc=72.625, loss_att=55.665, acc=0.672, loss=60.753, backward_time=0.135, grad_norm=31.539, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.057e-04, train_time=3.096
+[gpuc01:0/16] 2024-01-24 06:05:32,262 (trainer:737) INFO: 7epoch:train:2301-2400batch: iter_time=1.602e-04, forward_time=0.116, loss_ctc=70.676, loss_att=67.781, acc=0.645, loss=68.649, backward_time=0.136, grad_norm=30.339, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.056e-04, train_time=3.120
+[gpuc01:0/16] 2024-01-24 06:08:07,851 (trainer:737) INFO: 7epoch:train:2401-2500batch: iter_time=1.454e-04, forward_time=0.117, loss_ctc=73.286, loss_att=68.834, acc=0.646, loss=70.169, backward_time=0.136, grad_norm=30.131, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.055e-04, train_time=3.112
+[gpuc01:0/16] 2024-01-24 06:08:11,722 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpuc01:0/16] 2024-01-24 06:08:31,060 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 06:08:34,744 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71867d6470>)
+[gpuc01:0/16] 2024-01-24 06:08:34,744 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpuc01:0/16] 2024-01-24 06:08:34,750 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 06:15:08,638 (trainer:737) INFO: 7epoch:train:2501-2600batch: iter_time=2.213, forward_time=0.117, loss_ctc=70.204, loss_att=70.471, acc=0.646, loss=70.391, backward_time=0.136, grad_norm=33.079, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.054e-04, train_time=8.416
+[gpuc01:0/16] 2024-01-24 06:17:44,374 (trainer:737) INFO: 7epoch:train:2601-2700batch: iter_time=1.420e-04, forward_time=0.116, loss_ctc=63.709, loss_att=61.687, acc=0.681, loss=62.294, backward_time=0.136, grad_norm=24.817, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.053e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-24 06:20:19,543 (trainer:737) INFO: 7epoch:train:2701-2800batch: iter_time=1.434e-04, forward_time=0.119, loss_ctc=63.879, loss_att=60.667, acc=0.666, loss=61.630, backward_time=0.136, grad_norm=25.489, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.052e-04, train_time=3.103
+[gpuc01:0/16] 2024-01-24 06:22:54,924 (trainer:737) INFO: 7epoch:train:2801-2900batch: iter_time=1.425e-04, forward_time=0.117, loss_ctc=71.520, loss_att=70.086, acc=0.673, loss=70.516, backward_time=0.137, grad_norm=28.783, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.051e-04, train_time=3.107
+[gpuc01:0/16] 2024-01-24 06:25:30,666 (trainer:737) INFO: 7epoch:train:2901-3000batch: iter_time=1.516e-04, forward_time=0.117, loss_ctc=80.856, loss_att=67.841, acc=0.672, loss=71.746, backward_time=0.136, grad_norm=30.734, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.049e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-24 06:28:05,268 (trainer:737) INFO: 7epoch:train:3001-3100batch: iter_time=1.664e-04, forward_time=0.117, loss_ctc=61.360, loss_att=63.069, acc=0.678, loss=62.556, backward_time=0.136, grad_norm=24.238, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.048e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-24 06:30:40,359 (trainer:737) INFO: 7epoch:train:3101-3200batch: iter_time=1.620e-04, forward_time=0.116, loss_ctc=60.353, loss_att=56.416, acc=0.650, loss=57.597, backward_time=0.135, grad_norm=25.396, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.047e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-24 06:33:16,232 (trainer:737) INFO: 7epoch:train:3201-3300batch: iter_time=1.554e-04, forward_time=0.117, loss_ctc=69.128, loss_att=63.963, acc=0.656, loss=65.513, backward_time=0.136, grad_norm=28.424, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.046e-04, train_time=3.117
+[gpuc01:0/16] 2024-01-24 06:35:51,689 (trainer:737) INFO: 7epoch:train:3301-3400batch: iter_time=1.580e-04, forward_time=0.117, loss_ctc=72.123, loss_att=65.656, acc=0.674, loss=67.596, backward_time=0.136, grad_norm=27.878, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.045e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 06:38:25,975 (trainer:737) INFO: 7epoch:train:3401-3500batch: iter_time=1.651e-04, forward_time=0.117, loss_ctc=56.657, loss_att=59.324, acc=0.669, loss=58.524, backward_time=0.136, grad_norm=21.756, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.044e-04, train_time=3.086
+[gpuc01:0/16] 2024-01-24 06:40:59,281 (trainer:737) INFO: 7epoch:train:3501-3600batch: iter_time=1.749e-04, forward_time=0.117, loss_ctc=60.442, loss_att=65.259, acc=0.659, loss=63.814, backward_time=0.137, grad_norm=23.213, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.043e-04, train_time=3.066
+[gpuc01:0/16] 2024-01-24 06:43:32,439 (trainer:737) INFO: 7epoch:train:3601-3700batch: iter_time=1.549e-04, forward_time=0.116, loss_ctc=59.150, loss_att=58.908, acc=0.673, loss=58.981, backward_time=0.136, grad_norm=23.519, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.042e-04, train_time=3.063
+[gpuc01:0/16] 2024-01-24 06:46:06,035 (trainer:737) INFO: 7epoch:train:3701-3800batch: iter_time=1.514e-04, forward_time=0.117, loss_ctc=62.233, loss_att=56.806, acc=0.687, loss=58.434, backward_time=0.136, grad_norm=25.090, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.041e-04, train_time=3.072
+[gpuc01:0/16] 2024-01-24 06:48:39,267 (trainer:737) INFO: 7epoch:train:3801-3900batch: iter_time=1.598e-04, forward_time=0.117, loss_ctc=78.488, loss_att=75.118, acc=0.649, loss=76.129, backward_time=0.137, grad_norm=31.391, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.067, optim0_lr0=4.039e-04, train_time=3.064
+[gpuc01:0/16] 2024-01-24 06:51:11,401 (trainer:737) INFO: 7epoch:train:3901-4000batch: iter_time=1.575e-04, forward_time=0.117, loss_ctc=61.627, loss_att=70.308, acc=0.653, loss=67.704, backward_time=0.136, grad_norm=26.086, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.066, optim0_lr0=4.038e-04, train_time=3.042
+[gpuc01:0/16] 2024-01-24 06:53:45,007 (trainer:737) INFO: 7epoch:train:4001-4100batch: iter_time=1.597e-04, forward_time=0.116, loss_ctc=63.630, loss_att=54.345, acc=0.683, loss=57.130, backward_time=0.136, grad_norm=28.165, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.037e-04, train_time=3.072
+[gpuc01:0/16] 2024-01-24 06:56:18,570 (trainer:737) INFO: 7epoch:train:4101-4200batch: iter_time=1.525e-04, forward_time=0.117, loss_ctc=70.332, loss_att=82.562, acc=0.642, loss=78.893, backward_time=0.137, grad_norm=29.150, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.036e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-24 06:58:52,107 (trainer:737) INFO: 7epoch:train:4201-4300batch: iter_time=1.568e-04, forward_time=0.117, loss_ctc=68.965, loss_att=67.807, acc=0.686, loss=68.154, backward_time=0.136, grad_norm=27.087, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.035e-04, train_time=3.071
+[gpuc01:0/16] 2024-01-24 07:01:26,226 (trainer:737) INFO: 7epoch:train:4301-4400batch: iter_time=1.424e-04, forward_time=0.117, loss_ctc=69.025, loss_att=67.169, acc=0.660, loss=67.726, backward_time=0.136, grad_norm=24.927, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.034e-04, train_time=3.082
+[gpuc01:0/16] 2024-01-24 07:03:59,235 (trainer:737) INFO: 7epoch:train:4401-4500batch: iter_time=1.578e-04, forward_time=0.117, loss_ctc=62.761, loss_att=64.797, acc=0.676, loss=64.186, backward_time=0.136, grad_norm=23.821, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.033e-04, train_time=3.060
+[gpuc01:0/16] 2024-01-24 07:06:33,005 (trainer:737) INFO: 7epoch:train:4501-4600batch: iter_time=1.570e-04, forward_time=0.117, loss_ctc=61.265, loss_att=63.038, acc=0.680, loss=62.506, backward_time=0.136, grad_norm=25.773, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.032e-04, train_time=3.075
+[gpuc01:0/16] 2024-01-24 07:09:06,977 (trainer:737) INFO: 7epoch:train:4601-4700batch: iter_time=1.480e-04, forward_time=0.117, loss_ctc=62.021, loss_att=67.707, acc=0.660, loss=66.001, backward_time=0.136, grad_norm=24.837, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.031e-04, train_time=3.079
+[gpuc01:0/16] 2024-01-24 07:11:41,049 (trainer:737) INFO: 7epoch:train:4701-4800batch: iter_time=1.557e-04, forward_time=0.119, loss_ctc=70.365, loss_att=54.515, acc=0.687, loss=59.270, backward_time=0.136, grad_norm=29.686, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.030e-04, train_time=3.081
+[gpuc01:0/16] 2024-01-24 07:14:15,557 (trainer:737) INFO: 7epoch:train:4801-4900batch: iter_time=1.610e-04, forward_time=0.117, loss_ctc=68.620, loss_att=65.538, acc=0.655, loss=66.462, backward_time=0.136, grad_norm=28.907, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.029e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 07:17:08,889 (trainer:737) INFO: 7epoch:train:4901-5000batch: iter_time=7.131e-04, forward_time=0.231, loss_ctc=69.909, loss_att=68.076, acc=0.666, loss=68.626, backward_time=0.158, grad_norm=28.536, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.113, optim0_lr0=4.027e-04, train_time=3.466
+[gpuc01:0/16] 2024-01-24 07:17:29,065 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpuc01:0/16] 2024-01-24 07:17:48,657 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 07:17:52,298 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6a0cff0880>)
+[gpuc01:0/16] 2024-01-24 07:17:52,298 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpuc01:0/16] 2024-01-24 07:17:52,304 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 07:24:29,636 (trainer:737) INFO: 7epoch:train:5001-5100batch: iter_time=2.860, forward_time=0.117, loss_ctc=67.805, loss_att=68.484, acc=0.651, loss=68.280, backward_time=0.136, grad_norm=31.484, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.026e-04, train_time=8.815
+[gpuc01:0/16] 2024-01-24 07:27:04,745 (trainer:737) INFO: 7epoch:train:5101-5200batch: iter_time=1.497e-04, forward_time=0.116, loss_ctc=62.567, loss_att=60.723, acc=0.676, loss=61.276, backward_time=0.136, grad_norm=25.614, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.025e-04, train_time=3.102
+[gpuc01:0/16] 2024-01-24 07:29:39,254 (trainer:737) INFO: 7epoch:train:5201-5300batch: iter_time=1.609e-04, forward_time=0.116, loss_ctc=63.197, loss_att=59.798, acc=0.654, loss=60.818, backward_time=0.135, grad_norm=25.431, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.024e-04, train_time=3.090
+[gpuc01:0/16] 2024-01-24 07:32:13,414 (trainer:737) INFO: 7epoch:train:5301-5400batch: iter_time=1.589e-04, forward_time=0.117, loss_ctc=70.380, loss_att=69.471, acc=0.668, loss=69.743, backward_time=0.136, grad_norm=28.482, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.023e-04, train_time=3.083
+[gpuc01:0/16] 2024-01-24 07:34:48,082 (trainer:737) INFO: 7epoch:train:5401-5500batch: iter_time=1.600e-04, forward_time=0.117, loss_ctc=77.935, loss_att=65.698, acc=0.677, loss=69.369, backward_time=0.136, grad_norm=28.230, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.022e-04, train_time=3.093
+[gpuc01:0/16] 2024-01-24 07:37:23,108 (trainer:737) INFO: 7epoch:train:5501-5600batch: iter_time=1.502e-04, forward_time=0.119, loss_ctc=60.521, loss_att=60.925, acc=0.685, loss=60.804, backward_time=0.135, grad_norm=23.596, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.021e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 07:39:57,290 (trainer:737) INFO: 7epoch:train:5601-5700batch: iter_time=1.510e-04, forward_time=0.115, loss_ctc=59.306, loss_att=55.867, acc=0.647, loss=56.899, backward_time=0.134, grad_norm=26.364, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.020e-04, train_time=3.083
+[gpuc01:0/16] 2024-01-24 07:42:33,029 (trainer:737) INFO: 7epoch:train:5701-5800batch: iter_time=1.509e-04, forward_time=0.116, loss_ctc=67.936, loss_att=60.234, acc=0.664, loss=62.545, backward_time=0.135, grad_norm=27.708, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.019e-04, train_time=3.115
+[gpuc01:0/16] 2024-01-24 07:45:08,233 (trainer:737) INFO: 7epoch:train:5801-5900batch: iter_time=1.571e-04, forward_time=0.117, loss_ctc=71.474, loss_att=65.705, acc=0.672, loss=67.436, backward_time=0.136, grad_norm=27.020, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.018e-04, train_time=3.104
+[gpuc01:0/16] 2024-01-24 07:47:43,223 (trainer:737) INFO: 7epoch:train:5901-6000batch: iter_time=1.630e-04, forward_time=0.116, loss_ctc=56.506, loss_att=57.492, acc=0.667, loss=57.196, backward_time=0.135, grad_norm=24.860, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.017e-04, train_time=3.100
+[gpuc01:0/16] 2024-01-24 07:50:18,656 (trainer:737) INFO: 7epoch:train:6001-6100batch: iter_time=1.711e-04, forward_time=0.116, loss_ctc=60.244, loss_att=64.293, acc=0.659, loss=63.078, backward_time=0.135, grad_norm=23.326, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.016e-04, train_time=3.108
+[gpuc01:0/16] 2024-01-24 07:52:53,272 (trainer:737) INFO: 7epoch:train:6101-6200batch: iter_time=1.588e-04, forward_time=0.116, loss_ctc=58.502, loss_att=58.587, acc=0.664, loss=58.562, backward_time=0.135, grad_norm=23.509, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.014e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-24 07:55:27,440 (trainer:737) INFO: 7epoch:train:6201-6300batch: iter_time=1.584e-04, forward_time=0.116, loss_ctc=61.123, loss_att=56.094, acc=0.679, loss=57.603, backward_time=0.135, grad_norm=25.456, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=4.013e-04, train_time=3.083
+[gpuc01:0/16] 2024-01-24 07:58:02,972 (trainer:737) INFO: 7epoch:train:6301-6400batch: iter_time=1.586e-04, forward_time=0.117, loss_ctc=77.684, loss_att=73.704, acc=0.644, loss=74.898, backward_time=0.136, grad_norm=31.967, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.012e-04, train_time=3.110
+[gpuc01:0/16] 2024-01-24 08:00:38,241 (trainer:737) INFO: 7epoch:train:6401-6500batch: iter_time=1.625e-04, forward_time=0.116, loss_ctc=61.444, loss_att=69.964, acc=0.639, loss=67.408, backward_time=0.135, grad_norm=27.964, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.011e-04, train_time=3.105
+[gpuc01:0/16] 2024-01-24 08:03:12,828 (trainer:737) INFO: 7epoch:train:6501-6600batch: iter_time=1.526e-04, forward_time=0.116, loss_ctc=63.575, loss_att=54.665, acc=0.674, loss=57.338, backward_time=0.135, grad_norm=29.346, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.010e-04, train_time=3.092
+[gpuc01:0/16] 2024-01-24 08:05:46,612 (trainer:737) INFO: 7epoch:train:6601-6700batch: iter_time=1.514e-04, forward_time=0.116, loss_ctc=68.588, loss_att=75.650, acc=0.644, loss=73.532, backward_time=0.135, grad_norm=29.704, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.009e-04, train_time=3.075
+[gpuc01:0/16] 2024-01-24 08:08:20,268 (trainer:737) INFO: 7epoch:train:6701-6800batch: iter_time=1.638e-04, forward_time=0.117, loss_ctc=67.712, loss_att=65.018, acc=0.687, loss=65.826, backward_time=0.137, grad_norm=26.736, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=4.008e-04, train_time=3.073
+[gpuc01:0/16] 2024-01-24 08:10:55,819 (trainer:737) INFO: 7epoch:train:6801-6900batch: iter_time=1.586e-04, forward_time=0.116, loss_ctc=68.245, loss_att=66.227, acc=0.655, loss=66.833, backward_time=0.136, grad_norm=25.952, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.007e-04, train_time=3.111
+[gpuc01:0/16] 2024-01-24 08:13:31,284 (trainer:737) INFO: 7epoch:train:6901-7000batch: iter_time=1.577e-04, forward_time=0.117, loss_ctc=61.462, loss_att=63.623, acc=0.663, loss=62.975, backward_time=0.136, grad_norm=25.644, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.006e-04, train_time=3.109
+[gpuc01:0/16] 2024-01-24 08:18:09,166 (trainer:737) INFO: 7epoch:train:7001-7100batch: iter_time=0.184, forward_time=0.116, loss_ctc=60.095, loss_att=63.880, acc=0.676, loss=62.745, backward_time=0.135, grad_norm=26.611, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.005e-04, train_time=5.557
+[gpuc01:0/16] 2024-01-24 08:20:11,280 (trainer:737) INFO: 7epoch:train:7101-7200batch: iter_time=1.428e-04, forward_time=0.116, loss_ctc=61.806, loss_att=68.226, acc=0.648, loss=66.300, backward_time=0.137, grad_norm=26.573, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=4.004e-04, train_time=2.442
+[gpuc01:0/16] 2024-01-24 08:22:13,773 (trainer:737) INFO: 7epoch:train:7201-7300batch: iter_time=1.417e-04, forward_time=0.116, loss_ctc=70.260, loss_att=54.320, acc=0.678, loss=59.102, backward_time=0.137, grad_norm=32.850, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.003e-04, train_time=2.450
+[gpuc01:0/16] 2024-01-24 08:24:16,121 (trainer:737) INFO: 7epoch:train:7301-7400batch: iter_time=1.530e-04, forward_time=0.116, loss_ctc=67.488, loss_att=65.635, acc=0.649, loss=66.191, backward_time=0.137, grad_norm=31.229, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.002e-04, train_time=2.447
+[gpuc01:0/16] 2024-01-24 08:26:24,545 (trainer:737) INFO: 7epoch:train:7401-7500batch: iter_time=1.378e-04, forward_time=0.117, loss_ctc=68.720, loss_att=65.960, acc=0.654, loss=66.788, backward_time=0.137, grad_norm=28.983, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=4.001e-04, train_time=2.568
+[gpuc01:0/16] 2024-01-24 08:26:28,465 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpuc01:0/16] 2024-01-24 08:26:47,581 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 08:26:51,627 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6c5b410580>)
+[gpuc01:0/16] 2024-01-24 08:26:51,627 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpuc01:0/16] 2024-01-24 08:26:51,634 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 08:33:26,899 (trainer:737) INFO: 7epoch:train:7501-7600batch: iter_time=2.272, forward_time=0.125, loss_ctc=67.788, loss_att=65.837, acc=0.653, loss=66.422, backward_time=0.139, grad_norm=31.914, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=3.999e-04, train_time=8.447
+[gpuc01:0/16] 2024-01-24 08:35:33,368 (trainer:737) INFO: 7epoch:train:7601-7700batch: iter_time=1.639e-04, forward_time=0.116, loss_ctc=61.428, loss_att=59.442, acc=0.678, loss=60.038, backward_time=0.138, grad_norm=25.801, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=3.998e-04, train_time=2.529
+[gpuc01:0/16] 2024-01-24 08:37:39,527 (trainer:737) INFO: 7epoch:train:7701-7800batch: iter_time=1.649e-04, forward_time=0.116, loss_ctc=63.338, loss_att=58.405, acc=0.659, loss=59.885, backward_time=0.138, grad_norm=24.175, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=3.997e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 08:39:44,321 (trainer:737) INFO: 7epoch:train:7801-7900batch: iter_time=1.408e-04, forward_time=0.117, loss_ctc=70.166, loss_att=67.972, acc=0.670, loss=68.630, backward_time=0.139, grad_norm=28.231, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.068, optim0_lr0=3.996e-04, train_time=2.496
+[gpuc01:0/16] 2024-01-24 08:41:49,886 (trainer:737) INFO: 7epoch:train:7901-8000batch: iter_time=1.366e-04, forward_time=0.117, loss_ctc=76.838, loss_att=64.865, acc=0.678, loss=68.457, backward_time=0.138, grad_norm=29.901, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.067, optim0_lr0=3.995e-04, train_time=2.511
+[gpuc01:0/16] 2024-01-24 08:43:56,055 (trainer:737) INFO: 7epoch:train:8001-8100batch: iter_time=1.454e-04, forward_time=0.116, loss_ctc=59.828, loss_att=59.608, acc=0.686, loss=59.674, backward_time=0.138, grad_norm=24.426, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.994e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 08:46:02,353 (trainer:737) INFO: 7epoch:train:8101-8200batch: iter_time=1.525e-04, forward_time=0.116, loss_ctc=59.309, loss_att=54.184, acc=0.652, loss=55.722, backward_time=0.137, grad_norm=24.208, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.993e-04, train_time=2.526
+[gpuc01:0/16] 2024-01-24 08:48:08,371 (trainer:737) INFO: 7epoch:train:8201-8300batch: iter_time=1.774e-04, forward_time=0.117, loss_ctc=68.014, loss_att=59.403, acc=0.664, loss=61.986, backward_time=0.138, grad_norm=27.729, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.992e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 08:50:13,951 (trainer:737) INFO: 7epoch:train:8301-8400batch: iter_time=1.589e-04, forward_time=0.117, loss_ctc=71.012, loss_att=64.932, acc=0.673, loss=66.756, backward_time=0.138, grad_norm=28.703, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.991e-04, train_time=2.511
+[gpuc01:0/16] 2024-01-24 08:52:19,690 (trainer:737) INFO: 7epoch:train:8401-8500batch: iter_time=1.606e-04, forward_time=0.116, loss_ctc=55.032, loss_att=56.420, acc=0.668, loss=56.004, backward_time=0.137, grad_norm=22.534, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.990e-04, train_time=2.515
+[gpuc01:0/16] 2024-01-24 08:54:25,398 (trainer:737) INFO: 7epoch:train:8501-8600batch: iter_time=1.599e-04, forward_time=0.116, loss_ctc=59.076, loss_att=62.961, acc=0.664, loss=61.796, backward_time=0.137, grad_norm=23.731, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.989e-04, train_time=2.514
+[gpuc01:0/16] 2024-01-24 08:56:31,460 (trainer:737) INFO: 7epoch:train:8601-8700batch: iter_time=1.426e-04, forward_time=0.118, loss_ctc=58.343, loss_att=57.096, acc=0.672, loss=57.470, backward_time=0.136, grad_norm=24.074, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.988e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 08:58:37,473 (trainer:737) INFO: 7epoch:train:8701-8800batch: iter_time=1.396e-04, forward_time=0.116, loss_ctc=59.731, loss_att=54.962, acc=0.683, loss=56.393, backward_time=0.137, grad_norm=25.209, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.987e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 09:00:43,409 (trainer:737) INFO: 7epoch:train:8801-8900batch: iter_time=1.500e-04, forward_time=0.116, loss_ctc=76.773, loss_att=73.549, acc=0.643, loss=74.516, backward_time=0.138, grad_norm=33.732, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.986e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 09:02:48,916 (trainer:737) INFO: 7epoch:train:8901-9000batch: iter_time=1.400e-04, forward_time=0.116, loss_ctc=60.917, loss_att=70.105, acc=0.640, loss=67.349, backward_time=0.137, grad_norm=29.307, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.985e-04, train_time=2.510
+[gpuc01:0/16] 2024-01-24 09:04:55,160 (trainer:737) INFO: 7epoch:train:9001-9100batch: iter_time=1.429e-04, forward_time=0.115, loss_ctc=62.184, loss_att=55.456, acc=0.670, loss=57.474, backward_time=0.136, grad_norm=29.076, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.984e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 09:07:02,114 (trainer:737) INFO: 7epoch:train:9101-9200batch: iter_time=1.405e-04, forward_time=0.115, loss_ctc=67.817, loss_att=73.270, acc=0.652, loss=71.634, backward_time=0.137, grad_norm=31.837, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.982e-04, train_time=2.539
+[gpuc01:0/16] 2024-01-24 09:09:08,246 (trainer:737) INFO: 7epoch:train:9201-9300batch: iter_time=1.439e-04, forward_time=0.117, loss_ctc=66.513, loss_att=63.588, acc=0.688, loss=64.466, backward_time=0.138, grad_norm=27.322, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.981e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 09:11:14,482 (trainer:737) INFO: 7epoch:train:9301-9400batch: iter_time=1.512e-04, forward_time=0.115, loss_ctc=67.880, loss_att=64.619, acc=0.660, loss=65.597, backward_time=0.137, grad_norm=26.365, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.980e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 09:13:21,117 (trainer:737) INFO: 7epoch:train:9401-9500batch: iter_time=1.445e-04, forward_time=0.115, loss_ctc=60.517, loss_att=63.038, acc=0.666, loss=62.282, backward_time=0.137, grad_norm=23.558, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.979e-04, train_time=2.532
+[gpuc01:0/16] 2024-01-24 09:15:27,204 (trainer:737) INFO: 7epoch:train:9501-9600batch: iter_time=1.705e-04, forward_time=0.116, loss_ctc=58.713, loss_att=63.396, acc=0.681, loss=61.991, backward_time=0.138, grad_norm=24.687, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.978e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 09:17:33,541 (trainer:737) INFO: 7epoch:train:9601-9700batch: iter_time=1.460e-04, forward_time=0.118, loss_ctc=60.847, loss_att=65.947, acc=0.656, loss=64.417, backward_time=0.137, grad_norm=25.363, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.977e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 09:19:40,251 (trainer:737) INFO: 7epoch:train:9701-9800batch: iter_time=1.344e-04, forward_time=0.116, loss_ctc=69.174, loss_att=53.426, acc=0.681, loss=58.151, backward_time=0.137, grad_norm=29.721, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.976e-04, train_time=2.534
+[gpuc01:0/16] 2024-01-24 09:21:46,302 (trainer:737) INFO: 7epoch:train:9801-9900batch: iter_time=1.568e-04, forward_time=0.117, loss_ctc=66.594, loss_att=64.738, acc=0.652, loss=65.295, backward_time=0.137, grad_norm=32.321, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.066, optim0_lr0=3.975e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 09:23:52,329 (trainer:737) INFO: 7epoch:train:9901-10000batch: iter_time=2.552e-04, forward_time=0.117, loss_ctc=69.118, loss_att=65.887, acc=0.658, loss=66.856, backward_time=0.137, grad_norm=29.993, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.974e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 09:23:54,754 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpuc01:0/16] 2024-01-24 09:24:13,775 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 09:24:17,328 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6706baf460>)
+[gpuc01:0/16] 2024-01-24 09:24:17,329 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpuc01:0/16] 2024-01-24 09:24:17,335 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 09:28:20,168 (trainer:737) INFO: 7epoch:train:10001-10100batch: iter_time=1.375, forward_time=0.117, loss_ctc=66.606, loss_att=65.640, acc=0.653, loss=65.930, backward_time=0.137, grad_norm=29.259, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.973e-04, train_time=5.357
+[gpuc01:0/16] 2024-01-24 09:30:23,162 (trainer:737) INFO: 7epoch:train:10101-10200batch: iter_time=1.389e-04, forward_time=0.118, loss_ctc=60.903, loss_att=58.176, acc=0.684, loss=58.994, backward_time=0.137, grad_norm=26.093, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.972e-04, train_time=2.460
+[gpuc01:0/16] 2024-01-24 09:32:25,934 (trainer:737) INFO: 7epoch:train:10201-10300batch: iter_time=1.289e-04, forward_time=0.118, loss_ctc=62.766, loss_att=56.995, acc=0.663, loss=58.726, backward_time=0.137, grad_norm=25.103, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.971e-04, train_time=2.455
+[gpuc01:0/16] 2024-01-24 09:34:29,208 (trainer:737) INFO: 7epoch:train:10301-10400batch: iter_time=1.422e-04, forward_time=0.118, loss_ctc=69.040, loss_att=68.335, acc=0.671, loss=68.546, backward_time=0.138, grad_norm=27.570, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.970e-04, train_time=2.465
+[gpuc01:0/16] 2024-01-24 09:36:31,763 (trainer:737) INFO: 7epoch:train:10401-10500batch: iter_time=1.682e-04, forward_time=0.118, loss_ctc=76.317, loss_att=64.919, acc=0.679, loss=68.338, backward_time=0.138, grad_norm=29.495, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.969e-04, train_time=2.451
+[gpuc01:0/16] 2024-01-24 09:38:34,262 (trainer:737) INFO: 7epoch:train:10501-10600batch: iter_time=1.658e-04, forward_time=0.116, loss_ctc=59.463, loss_att=59.521, acc=0.687, loss=59.503, backward_time=0.137, grad_norm=25.978, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.968e-04, train_time=2.450
+[gpuc01:0/16] 2024-01-24 09:40:37,782 (trainer:737) INFO: 7epoch:train:10601-10700batch: iter_time=1.429e-04, forward_time=0.116, loss_ctc=59.138, loss_att=53.639, acc=0.654, loss=55.289, backward_time=0.136, grad_norm=24.849, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.967e-04, train_time=2.470
+[gpuc01:0/16] 2024-01-24 09:42:41,288 (trainer:737) INFO: 7epoch:train:10701-10800batch: iter_time=1.553e-04, forward_time=0.117, loss_ctc=66.919, loss_att=59.181, acc=0.667, loss=61.502, backward_time=0.137, grad_norm=28.170, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.966e-04, train_time=2.470
+[gpuc01:0/16] 2024-01-24 09:44:44,573 (trainer:737) INFO: 7epoch:train:10801-10900batch: iter_time=1.546e-04, forward_time=0.117, loss_ctc=70.346, loss_att=64.065, acc=0.677, loss=65.950, backward_time=0.138, grad_norm=28.163, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.965e-04, train_time=2.465
+[gpuc01:0/16] 2024-01-24 09:46:47,977 (trainer:737) INFO: 7epoch:train:10901-11000batch: iter_time=1.371e-04, forward_time=0.116, loss_ctc=55.102, loss_att=56.364, acc=0.669, loss=55.986, backward_time=0.138, grad_norm=22.380, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.964e-04, train_time=2.468
+[gpuc01:0/16] 2024-01-24 09:48:51,889 (trainer:737) INFO: 7epoch:train:11001-11100batch: iter_time=1.534e-04, forward_time=0.117, loss_ctc=58.799, loss_att=62.042, acc=0.665, loss=61.069, backward_time=0.138, grad_norm=24.498, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.963e-04, train_time=2.478
+[gpuc01:0/16] 2024-01-24 09:50:56,372 (trainer:737) INFO: 7epoch:train:11101-11200batch: iter_time=1.608e-04, forward_time=0.116, loss_ctc=58.117, loss_att=56.605, acc=0.672, loss=57.059, backward_time=0.138, grad_norm=23.242, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.962e-04, train_time=2.489
+[gpuc01:0/16] 2024-01-24 09:53:00,785 (trainer:737) INFO: 7epoch:train:11201-11300batch: iter_time=1.478e-04, forward_time=0.117, loss_ctc=59.887, loss_att=54.135, acc=0.686, loss=55.860, backward_time=0.137, grad_norm=25.127, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.961e-04, train_time=2.488
+[gpuc01:0/16] 2024-01-24 09:55:04,592 (trainer:737) INFO: 7epoch:train:11301-11400batch: iter_time=1.582e-04, forward_time=0.117, loss_ctc=76.232, loss_att=72.508, acc=0.646, loss=73.625, backward_time=0.138, grad_norm=29.735, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.960e-04, train_time=2.476
+[gpuc01:0/16] 2024-01-24 09:57:09,052 (trainer:737) INFO: 7epoch:train:11401-11500batch: iter_time=1.811e-04, forward_time=0.117, loss_ctc=60.477, loss_att=68.867, acc=0.644, loss=66.350, backward_time=0.138, grad_norm=25.836, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.958e-04, train_time=2.489
+[gpuc01:0/16] 2024-01-24 09:59:13,729 (trainer:737) INFO: 7epoch:train:11501-11600batch: iter_time=1.735e-04, forward_time=0.116, loss_ctc=61.908, loss_att=55.107, acc=0.670, loss=57.147, backward_time=0.138, grad_norm=29.349, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.068, optim0_lr0=3.957e-04, train_time=2.493
+[gpuc01:0/16] 2024-01-24 10:01:17,777 (trainer:737) INFO: 7epoch:train:11601-11700batch: iter_time=1.519e-04, forward_time=0.116, loss_ctc=67.103, loss_att=72.441, acc=0.655, loss=70.840, backward_time=0.139, grad_norm=28.487, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.956e-04, train_time=2.481
+[gpuc01:0/16] 2024-01-24 10:03:22,220 (trainer:737) INFO: 7epoch:train:11701-11800batch: iter_time=1.459e-04, forward_time=0.118, loss_ctc=66.584, loss_att=63.341, acc=0.693, loss=64.314, backward_time=0.139, grad_norm=26.305, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.955e-04, train_time=2.489
+[gpuc01:0/16] 2024-01-24 10:05:26,505 (trainer:737) INFO: 7epoch:train:11801-11900batch: iter_time=1.419e-04, forward_time=0.117, loss_ctc=68.172, loss_att=65.414, acc=0.660, loss=66.242, backward_time=0.138, grad_norm=25.568, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.954e-04, train_time=2.485
+[gpuc01:0/16] 2024-01-24 10:07:30,558 (trainer:737) INFO: 7epoch:train:11901-12000batch: iter_time=1.413e-04, forward_time=0.117, loss_ctc=61.264, loss_att=62.564, acc=0.668, loss=62.174, backward_time=0.138, grad_norm=25.017, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.067, optim0_lr0=3.953e-04, train_time=2.481
+[gpuc01:0/16] 2024-01-24 10:09:34,589 (trainer:737) INFO: 7epoch:train:12001-12100batch: iter_time=1.455e-04, forward_time=0.116, loss_ctc=58.746, loss_att=62.958, acc=0.681, loss=61.695, backward_time=0.137, grad_norm=25.761, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.952e-04, train_time=2.480
+[gpuc01:0/16] 2024-01-24 10:11:38,824 (trainer:737) INFO: 7epoch:train:12101-12200batch: iter_time=1.481e-04, forward_time=0.117, loss_ctc=59.950, loss_att=65.019, acc=0.659, loss=63.499, backward_time=0.137, grad_norm=25.001, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.951e-04, train_time=2.484
+[gpuc01:0/16] 2024-01-24 10:13:43,005 (trainer:737) INFO: 7epoch:train:12201-12300batch: iter_time=1.374e-04, forward_time=0.116, loss_ctc=68.617, loss_att=52.340, acc=0.686, loss=57.223, backward_time=0.137, grad_norm=29.531, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.950e-04, train_time=2.483
+[gpuc01:0/16] 2024-01-24 10:15:47,645 (trainer:737) INFO: 7epoch:train:12301-12400batch: iter_time=1.461e-04, forward_time=0.119, loss_ctc=65.284, loss_att=62.561, acc=0.658, loss=63.378, backward_time=0.138, grad_norm=30.815, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.949e-04, train_time=2.493
+[gpuc01:0/16] 2024-01-24 10:17:51,845 (trainer:737) INFO: 7epoch:train:12401-12500batch: iter_time=1.411e-04, forward_time=0.116, loss_ctc=67.499, loss_att=65.554, acc=0.661, loss=66.137, backward_time=0.138, grad_norm=28.868, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.948e-04, train_time=2.484
+[gpuc01:0/16] 2024-01-24 10:17:54,526 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpuc01:0/16] 2024-01-24 10:18:14,147 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 10:18:17,703 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6707fb88e0>)
+[gpuc01:0/16] 2024-01-24 10:18:17,703 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpuc01:0/16] 2024-01-24 10:18:17,709 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 10:22:17,909 (trainer:737) INFO: 7epoch:train:12501-12600batch: iter_time=1.388, forward_time=0.116, loss_ctc=65.548, loss_att=64.640, acc=0.658, loss=64.913, backward_time=0.138, grad_norm=28.598, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.068, optim0_lr0=3.947e-04, train_time=5.321
+[gpuc01:0/16] 2024-01-24 10:24:23,250 (trainer:737) INFO: 7epoch:train:12601-12700batch: iter_time=1.541e-04, forward_time=0.116, loss_ctc=61.195, loss_att=57.827, acc=0.683, loss=58.837, backward_time=0.138, grad_norm=23.610, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.068, optim0_lr0=3.946e-04, train_time=2.507
+[gpuc01:0/16] 2024-01-24 10:26:28,194 (trainer:737) INFO: 7epoch:train:12701-12800batch: iter_time=1.615e-04, forward_time=0.118, loss_ctc=62.222, loss_att=57.018, acc=0.661, loss=58.579, backward_time=0.137, grad_norm=24.219, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.945e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 10:28:33,195 (trainer:737) INFO: 7epoch:train:12801-12900batch: iter_time=1.583e-04, forward_time=0.117, loss_ctc=68.365, loss_att=67.320, acc=0.675, loss=67.634, backward_time=0.139, grad_norm=26.820, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.944e-04, train_time=2.500
+[gpuc01:0/16] 2024-01-24 10:30:38,543 (trainer:737) INFO: 7epoch:train:12901-13000batch: iter_time=1.766e-04, forward_time=0.117, loss_ctc=76.469, loss_att=65.984, acc=0.677, loss=69.129, backward_time=0.138, grad_norm=29.433, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.943e-04, train_time=2.507
+[gpuc01:0/16] 2024-01-24 10:32:43,460 (trainer:737) INFO: 7epoch:train:13001-13100batch: iter_time=1.587e-04, forward_time=0.117, loss_ctc=59.029, loss_att=58.637, acc=0.691, loss=58.755, backward_time=0.137, grad_norm=23.161, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.942e-04, train_time=2.498
+[gpuc01:0/16] 2024-01-24 10:34:48,438 (trainer:737) INFO: 7epoch:train:13101-13200batch: iter_time=1.536e-04, forward_time=0.116, loss_ctc=58.501, loss_att=53.837, acc=0.656, loss=55.236, backward_time=0.136, grad_norm=25.368, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.941e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 10:36:53,449 (trainer:737) INFO: 7epoch:train:13201-13300batch: iter_time=1.569e-04, forward_time=0.116, loss_ctc=66.433, loss_att=57.984, acc=0.672, loss=60.518, backward_time=0.137, grad_norm=27.877, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.940e-04, train_time=2.500
+[gpuc01:0/16] 2024-01-24 10:38:58,551 (trainer:737) INFO: 7epoch:train:13301-13400batch: iter_time=1.521e-04, forward_time=0.117, loss_ctc=70.265, loss_att=63.280, acc=0.678, loss=65.376, backward_time=0.137, grad_norm=28.178, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.939e-04, train_time=2.502
+[gpuc01:0/16] 2024-01-24 10:41:03,858 (trainer:737) INFO: 7epoch:train:13401-13500batch: iter_time=1.551e-04, forward_time=0.117, loss_ctc=54.476, loss_att=54.938, acc=0.678, loss=54.799, backward_time=0.137, grad_norm=22.587, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.938e-04, train_time=2.506
+[gpuc01:0/16] 2024-01-24 10:43:09,647 (trainer:737) INFO: 7epoch:train:13501-13600batch: iter_time=1.512e-04, forward_time=0.116, loss_ctc=58.779, loss_att=62.191, acc=0.666, loss=61.168, backward_time=0.137, grad_norm=24.647, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.937e-04, train_time=2.516
+[gpuc01:0/16] 2024-01-24 10:45:14,629 (trainer:737) INFO: 7epoch:train:13601-13700batch: iter_time=1.550e-04, forward_time=0.120, loss_ctc=56.958, loss_att=56.482, acc=0.673, loss=56.625, backward_time=0.136, grad_norm=24.970, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.936e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 10:47:19,317 (trainer:737) INFO: 7epoch:train:13701-13800batch: iter_time=1.419e-04, forward_time=0.117, loss_ctc=59.544, loss_att=54.172, acc=0.688, loss=55.784, backward_time=0.137, grad_norm=24.633, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.935e-04, train_time=2.494
+[gpuc01:0/16] 2024-01-24 10:49:24,234 (trainer:737) INFO: 7epoch:train:13801-13900batch: iter_time=1.312e-04, forward_time=0.118, loss_ctc=77.043, loss_att=72.220, acc=0.651, loss=73.667, backward_time=0.137, grad_norm=32.307, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.934e-04, train_time=2.498
+[gpuc01:0/16] 2024-01-24 10:51:29,256 (trainer:737) INFO: 7epoch:train:13901-14000batch: iter_time=1.453e-04, forward_time=0.118, loss_ctc=59.507, loss_att=67.330, acc=0.647, loss=64.983, backward_time=0.137, grad_norm=28.715, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.933e-04, train_time=2.500
+[gpuc01:0/16] 2024-01-24 10:53:34,231 (trainer:737) INFO: 7epoch:train:14001-14100batch: iter_time=1.446e-04, forward_time=0.118, loss_ctc=60.329, loss_att=52.377, acc=0.681, loss=54.762, backward_time=0.136, grad_norm=27.185, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.932e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 10:55:39,443 (trainer:737) INFO: 7epoch:train:14101-14200batch: iter_time=1.355e-04, forward_time=0.119, loss_ctc=68.370, loss_att=74.027, acc=0.651, loss=72.330, backward_time=0.138, grad_norm=32.023, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.066, optim0_lr0=3.931e-04, train_time=2.504
+[gpuc01:0/16] 2024-01-24 10:57:44,780 (trainer:737) INFO: 7epoch:train:14201-14300batch: iter_time=1.387e-04, forward_time=0.119, loss_ctc=65.869, loss_att=62.829, acc=0.694, loss=63.741, backward_time=0.138, grad_norm=26.681, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.930e-04, train_time=2.507
+[gpuc01:0/16] 2024-01-24 10:59:51,194 (trainer:737) INFO: 7epoch:train:14301-14400batch: iter_time=1.421e-04, forward_time=0.117, loss_ctc=66.610, loss_att=64.811, acc=0.663, loss=65.351, backward_time=0.137, grad_norm=25.093, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.929e-04, train_time=2.528
+[gpuc01:0/16] 2024-01-24 11:01:57,483 (trainer:737) INFO: 7epoch:train:14401-14500batch: iter_time=1.447e-04, forward_time=0.117, loss_ctc=61.260, loss_att=62.137, acc=0.670, loss=61.874, backward_time=0.137, grad_norm=24.396, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.928e-04, train_time=2.526
+[gpuc01:0/16] 2024-01-24 11:04:03,551 (trainer:737) INFO: 7epoch:train:14501-14600batch: iter_time=1.373e-04, forward_time=0.116, loss_ctc=57.711, loss_att=61.284, acc=0.687, loss=60.212, backward_time=0.137, grad_norm=28.591, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.066, optim0_lr0=3.927e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 11:06:09,773 (trainer:737) INFO: 7epoch:train:14601-14700batch: iter_time=1.494e-04, forward_time=0.117, loss_ctc=59.619, loss_att=65.165, acc=0.655, loss=63.501, backward_time=0.137, grad_norm=25.629, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.926e-04, train_time=2.524
+[gpuc01:0/16] 2024-01-24 11:08:16,036 (trainer:737) INFO: 7epoch:train:14701-14800batch: iter_time=1.412e-04, forward_time=0.116, loss_ctc=68.167, loss_att=52.670, acc=0.684, loss=57.319, backward_time=0.137, grad_norm=31.521, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.925e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 11:10:22,132 (trainer:737) INFO: 7epoch:train:14801-14900batch: iter_time=1.488e-04, forward_time=0.116, loss_ctc=65.326, loss_att=63.096, acc=0.659, loss=63.765, backward_time=0.137, grad_norm=31.320, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.924e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 11:12:28,370 (trainer:737) INFO: 7epoch:train:14901-15000batch: iter_time=1.435e-04, forward_time=0.117, loss_ctc=65.800, loss_att=64.416, acc=0.661, loss=64.831, backward_time=0.137, grad_norm=30.467, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.923e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 11:12:31,082 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpuc01:0/16] 2024-01-24 11:12:50,152 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 11:12:53,779 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f674585db70>)
+[gpuc01:0/16] 2024-01-24 11:12:53,780 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpuc01:0/16] 2024-01-24 11:12:53,786 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 11:16:56,226 (trainer:737) INFO: 7epoch:train:15001-15100batch: iter_time=1.390, forward_time=0.121, loss_ctc=64.804, loss_att=67.808, acc=0.659, loss=66.907, backward_time=0.137, grad_norm=31.395, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.922e-04, train_time=5.357
+[gpuc01:0/16] 2024-01-24 11:19:03,123 (trainer:737) INFO: 7epoch:train:15101-15200batch: iter_time=1.858e-04, forward_time=0.117, loss_ctc=61.259, loss_att=59.975, acc=0.689, loss=60.360, backward_time=0.137, grad_norm=25.858, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.921e-04, train_time=2.538
+[gpuc01:0/16] 2024-01-24 11:21:09,764 (trainer:737) INFO: 7epoch:train:15201-15300batch: iter_time=2.082e-04, forward_time=0.117, loss_ctc=61.553, loss_att=59.496, acc=0.675, loss=60.113, backward_time=0.137, grad_norm=24.845, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.920e-04, train_time=2.533
+[gpuc01:0/16] 2024-01-24 11:23:16,606 (trainer:737) INFO: 7epoch:train:15301-15400batch: iter_time=1.790e-04, forward_time=0.116, loss_ctc=67.340, loss_att=69.080, acc=0.675, loss=68.558, backward_time=0.138, grad_norm=28.878, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.919e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 11:25:23,142 (trainer:737) INFO: 7epoch:train:15401-15500batch: iter_time=2.063e-04, forward_time=0.117, loss_ctc=74.376, loss_att=64.904, acc=0.683, loss=67.746, backward_time=0.137, grad_norm=29.089, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.918e-04, train_time=2.530
+[gpuc01:0/16] 2024-01-24 11:27:29,679 (trainer:737) INFO: 7epoch:train:15501-15600batch: iter_time=1.840e-04, forward_time=0.116, loss_ctc=58.586, loss_att=60.571, acc=0.688, loss=59.976, backward_time=0.137, grad_norm=24.015, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.917e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 11:29:36,835 (trainer:737) INFO: 7epoch:train:15601-15700batch: iter_time=3.199e-04, forward_time=0.118, loss_ctc=57.373, loss_att=55.089, acc=0.661, loss=55.774, backward_time=0.137, grad_norm=24.868, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.067, optim0_lr0=3.916e-04, train_time=2.543
+[gpuc01:0/16] 2024-01-24 11:31:43,904 (trainer:737) INFO: 7epoch:train:15701-15800batch: iter_time=2.521e-04, forward_time=0.119, loss_ctc=65.601, loss_att=60.334, acc=0.672, loss=61.914, backward_time=0.137, grad_norm=29.647, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.068, optim0_lr0=3.915e-04, train_time=2.541
+[gpuc01:0/16] 2024-01-24 11:33:51,119 (trainer:737) INFO: 7epoch:train:15801-15900batch: iter_time=2.590e-04, forward_time=0.116, loss_ctc=69.319, loss_att=63.116, acc=0.683, loss=64.977, backward_time=0.138, grad_norm=26.411, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.068, optim0_lr0=3.914e-04, train_time=2.544
+[gpuc01:0/16] 2024-01-24 11:35:58,069 (trainer:737) INFO: 7epoch:train:15901-16000batch: iter_time=2.432e-04, forward_time=0.117, loss_ctc=53.673, loss_att=55.179, acc=0.686, loss=54.727, backward_time=0.138, grad_norm=22.711, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.068, optim0_lr0=3.913e-04, train_time=2.539
+[gpuc01:0/16] 2024-01-24 11:38:05,346 (trainer:737) INFO: 7epoch:train:16001-16100batch: iter_time=2.613e-04, forward_time=0.117, loss_ctc=58.776, loss_att=63.857, acc=0.669, loss=62.333, backward_time=0.138, grad_norm=26.039, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.912e-04, train_time=2.545
+[gpuc01:0/16] 2024-01-24 11:40:12,769 (trainer:737) INFO: 7epoch:train:16101-16200batch: iter_time=2.661e-04, forward_time=0.116, loss_ctc=56.686, loss_att=56.945, acc=0.685, loss=56.868, backward_time=0.137, grad_norm=22.830, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.911e-04, train_time=2.548
+[gpuc01:0/16] 2024-01-24 11:42:19,839 (trainer:737) INFO: 7epoch:train:16201-16300batch: iter_time=2.664e-04, forward_time=0.117, loss_ctc=59.909, loss_att=55.306, acc=0.698, loss=56.686, backward_time=0.138, grad_norm=24.935, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.910e-04, train_time=2.541
+[gpuc01:0/16] 2024-01-24 11:44:27,749 (trainer:737) INFO: 7epoch:train:16301-16400batch: iter_time=2.602e-04, forward_time=0.118, loss_ctc=75.921, loss_att=73.901, acc=0.660, loss=74.507, backward_time=0.139, grad_norm=32.910, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.909e-04, train_time=2.558
+[gpuc01:0/16] 2024-01-24 11:46:35,840 (trainer:737) INFO: 7epoch:train:16401-16500batch: iter_time=2.842e-04, forward_time=0.117, loss_ctc=59.754, loss_att=69.726, acc=0.656, loss=66.734, backward_time=0.138, grad_norm=26.782, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.908e-04, train_time=2.562
+[gpuc01:0/16] 2024-01-24 11:48:42,796 (trainer:737) INFO: 7epoch:train:16501-16600batch: iter_time=2.779e-04, forward_time=0.117, loss_ctc=60.420, loss_att=53.801, acc=0.686, loss=55.787, backward_time=0.138, grad_norm=28.075, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.907e-04, train_time=2.539
+[gpuc01:0/16] 2024-01-24 11:50:49,483 (trainer:737) INFO: 7epoch:train:16601-16700batch: iter_time=3.008e-04, forward_time=0.117, loss_ctc=67.986, loss_att=79.345, acc=0.654, loss=75.937, backward_time=0.138, grad_norm=30.741, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.906e-04, train_time=2.534
+[gpuc01:0/16] 2024-01-24 11:52:56,324 (trainer:737) INFO: 7epoch:train:16701-16800batch: iter_time=2.636e-04, forward_time=0.117, loss_ctc=66.181, loss_att=67.265, acc=0.690, loss=66.940, backward_time=0.139, grad_norm=27.789, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.905e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 11:55:03,979 (trainer:737) INFO: 7epoch:train:16801-16900batch: iter_time=2.838e-04, forward_time=0.117, loss_ctc=66.606, loss_att=64.651, acc=0.671, loss=65.237, backward_time=0.138, grad_norm=25.165, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.904e-04, train_time=2.553
+[gpuc01:0/16] 2024-01-24 11:57:11,075 (trainer:737) INFO: 7epoch:train:16901-17000batch: iter_time=2.637e-04, forward_time=0.117, loss_ctc=59.762, loss_att=64.010, acc=0.682, loss=62.735, backward_time=0.138, grad_norm=23.949, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.903e-04, train_time=2.542
+[gpuc01:0/16] 2024-01-24 11:59:18,158 (trainer:737) INFO: 7epoch:train:17001-17100batch: iter_time=2.793e-04, forward_time=0.117, loss_ctc=57.903, loss_att=62.381, acc=0.683, loss=61.038, backward_time=0.138, grad_norm=24.727, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.902e-04, train_time=2.541
+[gpuc01:0/16] 2024-01-24 12:01:25,371 (trainer:737) INFO: 7epoch:train:17101-17200batch: iter_time=2.928e-04, forward_time=0.117, loss_ctc=59.134, loss_att=66.166, acc=0.669, loss=64.057, backward_time=0.139, grad_norm=24.274, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.901e-04, train_time=2.544
+[gpuc01:0/16] 2024-01-24 12:03:32,191 (trainer:737) INFO: 7epoch:train:17201-17300batch: iter_time=2.107e-04, forward_time=0.119, loss_ctc=67.301, loss_att=53.519, acc=0.696, loss=57.653, backward_time=0.138, grad_norm=28.211, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.900e-04, train_time=2.536
+[gpuc01:0/16] 2024-01-24 12:05:39,131 (trainer:737) INFO: 7epoch:train:17301-17400batch: iter_time=1.699e-04, forward_time=0.119, loss_ctc=64.176, loss_att=63.886, acc=0.661, loss=63.973, backward_time=0.139, grad_norm=32.280, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.899e-04, train_time=2.539
+[gpuc01:0/16] 2024-01-24 12:07:47,233 (trainer:737) INFO: 7epoch:train:17401-17500batch: iter_time=1.826e-04, forward_time=0.120, loss_ctc=66.388, loss_att=65.513, acc=0.679, loss=65.776, backward_time=0.138, grad_norm=28.974, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.898e-04, train_time=2.562
+[gpuc01:0/16] 2024-01-24 12:07:50,425 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpuc01:0/16] 2024-01-24 12:08:09,749 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 12:08:13,729 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6710988af0>)
+[gpuc01:0/16] 2024-01-24 12:08:13,730 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpuc01:0/16] 2024-01-24 12:08:13,736 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 12:12:14,116 (trainer:737) INFO: 7epoch:train:17501-17600batch: iter_time=1.402, forward_time=0.117, loss_ctc=65.274, loss_att=66.017, acc=0.663, loss=65.794, backward_time=0.137, grad_norm=29.420, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.897e-04, train_time=5.337
+[gpuc01:0/16] 2024-01-24 12:14:20,687 (trainer:737) INFO: 7epoch:train:17601-17700batch: iter_time=2.312e-04, forward_time=0.117, loss_ctc=60.171, loss_att=57.839, acc=0.692, loss=58.538, backward_time=0.137, grad_norm=23.648, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.896e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 12:16:27,545 (trainer:737) INFO: 7epoch:train:17701-17800batch: iter_time=2.127e-04, forward_time=0.117, loss_ctc=61.088, loss_att=56.824, acc=0.680, loss=58.103, backward_time=0.138, grad_norm=23.759, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.895e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 12:18:33,326 (trainer:737) INFO: 7epoch:train:17801-17900batch: iter_time=1.847e-04, forward_time=0.115, loss_ctc=67.363, loss_att=66.673, acc=0.688, loss=66.880, backward_time=0.138, grad_norm=26.978, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.894e-04, train_time=2.515
+[gpuc01:0/16] 2024-01-24 12:20:39,057 (trainer:737) INFO: 7epoch:train:17901-18000batch: iter_time=1.975e-04, forward_time=0.116, loss_ctc=74.580, loss_att=64.678, acc=0.688, loss=67.649, backward_time=0.137, grad_norm=29.521, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.893e-04, train_time=2.514
+[gpuc01:0/16] 2024-01-24 12:22:44,967 (trainer:737) INFO: 7epoch:train:18001-18100batch: iter_time=2.323e-04, forward_time=0.117, loss_ctc=58.393, loss_att=59.609, acc=0.692, loss=59.244, backward_time=0.137, grad_norm=24.203, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.892e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 12:24:50,767 (trainer:737) INFO: 7epoch:train:18101-18200batch: iter_time=2.286e-04, forward_time=0.114, loss_ctc=57.775, loss_att=53.779, acc=0.665, loss=54.977, backward_time=0.136, grad_norm=23.700, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.891e-04, train_time=2.516
+[gpuc01:0/16] 2024-01-24 12:26:56,967 (trainer:737) INFO: 7epoch:train:18201-18300batch: iter_time=2.369e-04, forward_time=0.115, loss_ctc=66.086, loss_att=60.566, acc=0.671, loss=62.222, backward_time=0.137, grad_norm=27.896, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.890e-04, train_time=2.524
+[gpuc01:0/16] 2024-01-24 12:29:02,985 (trainer:737) INFO: 7epoch:train:18301-18400batch: iter_time=2.020e-04, forward_time=0.115, loss_ctc=69.201, loss_att=62.976, acc=0.683, loss=64.844, backward_time=0.138, grad_norm=26.517, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.889e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 12:31:08,570 (trainer:737) INFO: 7epoch:train:18401-18500batch: iter_time=2.357e-04, forward_time=0.114, loss_ctc=54.010, loss_att=56.024, acc=0.683, loss=55.420, backward_time=0.137, grad_norm=22.347, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.888e-04, train_time=2.511
+[gpuc01:0/16] 2024-01-24 12:33:14,038 (trainer:737) INFO: 7epoch:train:18501-18600batch: iter_time=2.448e-04, forward_time=0.114, loss_ctc=58.363, loss_att=62.397, acc=0.673, loss=61.186, backward_time=0.137, grad_norm=24.608, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.887e-04, train_time=2.509
+[gpuc01:0/16] 2024-01-24 12:35:19,844 (trainer:737) INFO: 7epoch:train:18601-18700batch: iter_time=2.471e-04, forward_time=0.115, loss_ctc=57.125, loss_att=56.612, acc=0.685, loss=56.766, backward_time=0.136, grad_norm=22.526, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.886e-04, train_time=2.516
+[gpuc01:0/16] 2024-01-24 12:37:25,784 (trainer:737) INFO: 7epoch:train:18701-18800batch: iter_time=2.236e-04, forward_time=0.115, loss_ctc=58.546, loss_att=53.851, acc=0.702, loss=55.260, backward_time=0.137, grad_norm=23.102, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.885e-04, train_time=2.519
+[gpuc01:0/16] 2024-01-24 12:39:31,964 (trainer:737) INFO: 7epoch:train:18801-18900batch: iter_time=2.100e-04, forward_time=0.116, loss_ctc=74.678, loss_att=71.611, acc=0.665, loss=72.531, backward_time=0.137, grad_norm=39.268, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.884e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 12:41:38,212 (trainer:737) INFO: 7epoch:train:18901-19000batch: iter_time=2.067e-04, forward_time=0.115, loss_ctc=58.079, loss_att=67.013, acc=0.665, loss=64.333, backward_time=0.137, grad_norm=26.047, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.883e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 12:43:44,298 (trainer:737) INFO: 7epoch:train:19001-19100batch: iter_time=1.838e-04, forward_time=0.114, loss_ctc=58.575, loss_att=51.195, acc=0.694, loss=53.409, backward_time=0.136, grad_norm=26.762, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.882e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 12:45:50,640 (trainer:737) INFO: 7epoch:train:19101-19200batch: iter_time=2.130e-04, forward_time=0.116, loss_ctc=66.804, loss_att=78.851, acc=0.655, loss=75.237, backward_time=0.137, grad_norm=29.184, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.881e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 12:47:56,570 (trainer:737) INFO: 7epoch:train:19201-19300batch: iter_time=1.774e-04, forward_time=0.115, loss_ctc=64.988, loss_att=63.641, acc=0.700, loss=64.045, backward_time=0.138, grad_norm=26.245, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.880e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 12:50:02,536 (trainer:737) INFO: 7epoch:train:19301-19400batch: iter_time=2.022e-04, forward_time=0.114, loss_ctc=65.790, loss_att=63.599, acc=0.675, loss=64.256, backward_time=0.137, grad_norm=25.529, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.879e-04, train_time=2.519
+[gpuc01:0/16] 2024-01-24 12:52:09,008 (trainer:737) INFO: 7epoch:train:19401-19500batch: iter_time=2.046e-04, forward_time=0.118, loss_ctc=58.821, loss_att=61.372, acc=0.688, loss=60.607, backward_time=0.137, grad_norm=24.842, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.878e-04, train_time=2.529
+[gpuc01:0/16] 2024-01-24 12:54:15,080 (trainer:737) INFO: 7epoch:train:19501-19600batch: iter_time=2.021e-04, forward_time=0.113, loss_ctc=56.704, loss_att=60.338, acc=0.693, loss=59.248, backward_time=0.137, grad_norm=25.359, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.877e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 12:56:21,115 (trainer:737) INFO: 7epoch:train:19601-19700batch: iter_time=1.989e-04, forward_time=0.115, loss_ctc=59.328, loss_att=64.968, acc=0.674, loss=63.276, backward_time=0.137, grad_norm=25.282, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.876e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 12:58:27,283 (trainer:737) INFO: 7epoch:train:19701-19800batch: iter_time=2.089e-04, forward_time=0.116, loss_ctc=67.283, loss_att=52.678, acc=0.697, loss=57.059, backward_time=0.137, grad_norm=32.040, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.068, optim0_lr0=3.875e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 13:00:33,402 (trainer:737) INFO: 7epoch:train:19801-19900batch: iter_time=1.751e-04, forward_time=0.115, loss_ctc=64.075, loss_att=63.414, acc=0.663, loss=63.612, backward_time=0.137, grad_norm=28.867, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.874e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 13:02:39,661 (trainer:737) INFO: 7epoch:train:19901-20000batch: iter_time=1.539e-04, forward_time=0.115, loss_ctc=64.946, loss_att=65.224, acc=0.678, loss=65.141, backward_time=0.137, grad_norm=27.913, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.067, optim0_lr0=3.873e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 13:02:41,819 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpuc01:0/16] 2024-01-24 13:03:01,562 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 13:03:05,229 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6723547c10>)
+[gpuc01:0/16] 2024-01-24 13:03:05,229 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpuc01:0/16] 2024-01-24 13:03:05,235 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 13:07:07,825 (trainer:737) INFO: 7epoch:train:20001-20100batch: iter_time=1.394, forward_time=0.117, loss_ctc=64.506, loss_att=64.507, acc=0.670, loss=64.507, backward_time=0.138, grad_norm=30.053, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.872e-04, train_time=5.363
+[gpuc01:0/16] 2024-01-24 13:09:14,192 (trainer:737) INFO: 7epoch:train:20101-20200batch: iter_time=2.320e-04, forward_time=0.117, loss_ctc=59.349, loss_att=57.603, acc=0.695, loss=58.127, backward_time=0.137, grad_norm=23.875, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.872e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 13:11:20,769 (trainer:737) INFO: 7epoch:train:20201-20300batch: iter_time=2.178e-04, forward_time=0.116, loss_ctc=60.563, loss_att=55.733, acc=0.685, loss=57.182, backward_time=0.137, grad_norm=25.244, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.871e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 13:13:27,874 (trainer:737) INFO: 7epoch:train:20301-20400batch: iter_time=2.239e-04, forward_time=0.118, loss_ctc=67.294, loss_att=67.957, acc=0.683, loss=67.758, backward_time=0.138, grad_norm=28.023, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.870e-04, train_time=2.542
+[gpuc01:0/16] 2024-01-24 13:15:34,496 (trainer:737) INFO: 7epoch:train:20401-20500batch: iter_time=1.649e-04, forward_time=0.119, loss_ctc=72.939, loss_att=63.485, acc=0.690, loss=66.321, backward_time=0.137, grad_norm=28.919, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.869e-04, train_time=2.532
+[gpuc01:0/16] 2024-01-24 13:17:41,338 (trainer:737) INFO: 7epoch:train:20501-20600batch: iter_time=2.056e-04, forward_time=0.117, loss_ctc=58.493, loss_att=59.487, acc=0.693, loss=59.189, backward_time=0.137, grad_norm=22.774, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.868e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 13:19:48,144 (trainer:737) INFO: 7epoch:train:20601-20700batch: iter_time=2.088e-04, forward_time=0.116, loss_ctc=57.245, loss_att=53.641, acc=0.667, loss=54.722, backward_time=0.137, grad_norm=22.749, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.867e-04, train_time=2.536
+[gpuc01:0/16] 2024-01-24 13:21:55,006 (trainer:737) INFO: 7epoch:train:20701-20800batch: iter_time=2.244e-04, forward_time=0.116, loss_ctc=65.049, loss_att=59.274, acc=0.678, loss=61.006, backward_time=0.137, grad_norm=26.684, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.866e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 13:24:01,675 (trainer:737) INFO: 7epoch:train:20801-20900batch: iter_time=2.273e-04, forward_time=0.117, loss_ctc=67.200, loss_att=61.063, acc=0.692, loss=62.904, backward_time=0.137, grad_norm=24.705, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.865e-04, train_time=2.533
+[gpuc01:0/16] 2024-01-24 13:26:08,230 (trainer:737) INFO: 7epoch:train:20901-21000batch: iter_time=2.399e-04, forward_time=0.117, loss_ctc=53.418, loss_att=54.551, acc=0.688, loss=54.211, backward_time=0.137, grad_norm=22.750, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.864e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 13:28:15,337 (trainer:737) INFO: 7epoch:train:21001-21100batch: iter_time=2.291e-04, forward_time=0.120, loss_ctc=57.213, loss_att=61.338, acc=0.674, loss=60.101, backward_time=0.137, grad_norm=23.252, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.863e-04, train_time=2.542
+[gpuc01:0/16] 2024-01-24 13:30:21,863 (trainer:737) INFO: 7epoch:train:21101-21200batch: iter_time=2.192e-04, forward_time=0.116, loss_ctc=56.268, loss_att=55.142, acc=0.691, loss=55.479, backward_time=0.137, grad_norm=22.084, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.862e-04, train_time=2.530
+[gpuc01:0/16] 2024-01-24 13:32:28,275 (trainer:737) INFO: 7epoch:train:21201-21300batch: iter_time=2.276e-04, forward_time=0.116, loss_ctc=58.508, loss_att=53.060, acc=0.704, loss=54.694, backward_time=0.138, grad_norm=24.009, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.861e-04, train_time=2.528
+[gpuc01:0/16] 2024-01-24 13:34:34,793 (trainer:737) INFO: 7epoch:train:21301-21400batch: iter_time=2.218e-04, forward_time=0.117, loss_ctc=74.579, loss_att=70.987, acc=0.666, loss=72.065, backward_time=0.139, grad_norm=30.705, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.860e-04, train_time=2.530
+[gpuc01:0/16] 2024-01-24 13:36:42,163 (trainer:737) INFO: 7epoch:train:21401-21500batch: iter_time=2.300e-04, forward_time=0.117, loss_ctc=57.867, loss_att=66.654, acc=0.665, loss=64.018, backward_time=0.138, grad_norm=25.543, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.859e-04, train_time=2.547
+[gpuc01:0/16] 2024-01-24 13:38:48,615 (trainer:737) INFO: 7epoch:train:21501-21600batch: iter_time=2.526e-04, forward_time=0.117, loss_ctc=59.482, loss_att=52.764, acc=0.690, loss=54.780, backward_time=0.137, grad_norm=28.446, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.858e-04, train_time=2.529
+[gpuc01:0/16] 2024-01-24 13:40:55,551 (trainer:737) INFO: 7epoch:train:21601-21700batch: iter_time=2.368e-04, forward_time=0.117, loss_ctc=66.974, loss_att=77.340, acc=0.659, loss=74.230, backward_time=0.138, grad_norm=32.327, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.857e-04, train_time=2.538
+[gpuc01:0/16] 2024-01-24 13:43:02,147 (trainer:737) INFO: 7epoch:train:21701-21800batch: iter_time=2.146e-04, forward_time=0.117, loss_ctc=65.432, loss_att=64.784, acc=0.696, loss=64.979, backward_time=0.139, grad_norm=28.003, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.856e-04, train_time=2.532
+[gpuc01:0/16] 2024-01-24 13:45:09,475 (trainer:737) INFO: 7epoch:train:21801-21900batch: iter_time=2.058e-04, forward_time=0.117, loss_ctc=64.828, loss_att=63.602, acc=0.675, loss=63.970, backward_time=0.138, grad_norm=24.272, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.855e-04, train_time=2.546
+[gpuc01:0/16] 2024-01-24 13:47:16,267 (trainer:737) INFO: 7epoch:train:21901-22000batch: iter_time=2.132e-04, forward_time=0.117, loss_ctc=58.771, loss_att=61.842, acc=0.687, loss=60.921, backward_time=0.138, grad_norm=22.822, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.854e-04, train_time=2.536
+[gpuc01:0/16] 2024-01-24 13:49:22,963 (trainer:737) INFO: 7epoch:train:22001-22100batch: iter_time=2.022e-04, forward_time=0.117, loss_ctc=56.140, loss_att=62.741, acc=0.686, loss=60.761, backward_time=0.137, grad_norm=24.601, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.853e-04, train_time=2.534
+[gpuc01:0/16] 2024-01-24 13:51:30,001 (trainer:737) INFO: 7epoch:train:22101-22200batch: iter_time=1.928e-04, forward_time=0.117, loss_ctc=58.188, loss_att=65.019, acc=0.674, loss=62.969, backward_time=0.138, grad_norm=25.435, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.852e-04, train_time=2.541
+[gpuc01:0/16] 2024-01-24 13:53:37,898 (trainer:737) INFO: 7epoch:train:22201-22300batch: iter_time=1.876e-04, forward_time=0.117, loss_ctc=67.148, loss_att=51.832, acc=0.700, loss=56.427, backward_time=0.138, grad_norm=30.177, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.851e-04, train_time=2.558
+[gpuc01:0/16] 2024-01-24 13:55:45,013 (trainer:737) INFO: 7epoch:train:22301-22400batch: iter_time=1.966e-04, forward_time=0.117, loss_ctc=62.350, loss_att=62.600, acc=0.666, loss=62.525, backward_time=0.137, grad_norm=28.947, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.850e-04, train_time=2.542
+[gpuc01:0/16] 2024-01-24 13:57:52,245 (trainer:737) INFO: 7epoch:train:22401-22500batch: iter_time=1.799e-04, forward_time=0.117, loss_ctc=65.068, loss_att=65.232, acc=0.682, loss=65.183, backward_time=0.138, grad_norm=28.815, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.849e-04, train_time=2.544
+[gpuc01:0/16] 2024-01-24 13:57:54,784 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpuc01:0/16] 2024-01-24 13:58:13,955 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 13:58:17,662 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f673fbc37f0>)
+[gpuc01:0/16] 2024-01-24 13:58:17,662 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpuc01:0/16] 2024-01-24 13:58:17,669 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 14:02:21,701 (trainer:737) INFO: 7epoch:train:22501-22600batch: iter_time=1.405, forward_time=0.116, loss_ctc=63.331, loss_att=65.301, acc=0.663, loss=64.710, backward_time=0.138, grad_norm=30.728, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.848e-04, train_time=5.389
+[gpuc01:0/16] 2024-01-24 14:04:27,879 (trainer:737) INFO: 7epoch:train:22601-22700batch: iter_time=1.956e-04, forward_time=0.116, loss_ctc=59.043, loss_att=58.100, acc=0.689, loss=58.383, backward_time=0.137, grad_norm=24.267, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.848e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 14:06:34,088 (trainer:737) INFO: 7epoch:train:22701-22800batch: iter_time=2.023e-04, forward_time=0.119, loss_ctc=60.369, loss_att=57.812, acc=0.665, loss=58.579, backward_time=0.137, grad_norm=24.004, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.847e-04, train_time=2.524
+[gpuc01:0/16] 2024-01-24 14:08:40,273 (trainer:737) INFO: 7epoch:train:22801-22900batch: iter_time=2.233e-04, forward_time=0.117, loss_ctc=66.266, loss_att=66.641, acc=0.679, loss=66.529, backward_time=0.138, grad_norm=26.488, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.846e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 14:10:46,658 (trainer:737) INFO: 7epoch:train:22901-23000batch: iter_time=2.192e-04, forward_time=0.117, loss_ctc=72.651, loss_att=62.892, acc=0.689, loss=65.820, backward_time=0.137, grad_norm=27.833, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.845e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 14:12:53,854 (trainer:737) INFO: 7epoch:train:23001-23100batch: iter_time=2.031e-04, forward_time=0.117, loss_ctc=57.082, loss_att=57.978, acc=0.694, loss=57.709, backward_time=0.138, grad_norm=22.495, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.844e-04, train_time=2.544
+[gpuc01:0/16] 2024-01-24 14:15:00,078 (trainer:737) INFO: 7epoch:train:23101-23200batch: iter_time=2.005e-04, forward_time=0.116, loss_ctc=57.480, loss_att=53.237, acc=0.662, loss=54.510, backward_time=0.136, grad_norm=24.959, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.843e-04, train_time=2.524
+[gpuc01:0/16] 2024-01-24 14:17:06,369 (trainer:737) INFO: 7epoch:train:23201-23300batch: iter_time=2.240e-04, forward_time=0.117, loss_ctc=63.963, loss_att=56.821, acc=0.677, loss=58.964, backward_time=0.137, grad_norm=26.459, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.842e-04, train_time=2.526
+[gpuc01:0/16] 2024-01-24 14:19:12,425 (trainer:737) INFO: 7epoch:train:23301-23400batch: iter_time=2.268e-04, forward_time=0.117, loss_ctc=67.169, loss_att=61.777, acc=0.684, loss=63.394, backward_time=0.138, grad_norm=28.356, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.841e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 14:21:18,906 (trainer:737) INFO: 7epoch:train:23401-23500batch: iter_time=2.423e-04, forward_time=0.116, loss_ctc=53.458, loss_att=54.939, acc=0.677, loss=54.495, backward_time=0.137, grad_norm=23.175, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.840e-04, train_time=2.529
+[gpuc01:0/16] 2024-01-24 14:23:24,904 (trainer:737) INFO: 7epoch:train:23501-23600batch: iter_time=1.856e-04, forward_time=0.117, loss_ctc=57.068, loss_att=60.439, acc=0.674, loss=59.427, backward_time=0.137, grad_norm=23.988, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.839e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 14:25:30,437 (trainer:737) INFO: 7epoch:train:23601-23700batch: iter_time=1.860e-04, forward_time=0.116, loss_ctc=55.623, loss_att=54.786, acc=0.680, loss=55.037, backward_time=0.137, grad_norm=23.740, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.838e-04, train_time=2.510
+[gpuc01:0/16] 2024-01-24 14:27:36,597 (trainer:737) INFO: 7epoch:train:23701-23800batch: iter_time=1.889e-04, forward_time=0.116, loss_ctc=57.538, loss_att=54.049, acc=0.690, loss=55.096, backward_time=0.137, grad_norm=24.816, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.837e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 14:29:42,576 (trainer:737) INFO: 7epoch:train:23801-23900batch: iter_time=1.889e-04, forward_time=0.117, loss_ctc=74.158, loss_att=71.585, acc=0.653, loss=72.357, backward_time=0.138, grad_norm=31.063, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.067, optim0_lr0=3.836e-04, train_time=2.519
+[gpuc01:0/16] 2024-01-24 14:31:49,610 (trainer:737) INFO: 7epoch:train:23901-24000batch: iter_time=1.785e-04, forward_time=0.116, loss_ctc=57.316, loss_att=67.448, acc=0.651, loss=64.408, backward_time=0.137, grad_norm=26.236, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.068, optim0_lr0=3.835e-04, train_time=2.540
+[gpuc01:0/16] 2024-01-24 14:33:55,194 (trainer:737) INFO: 7epoch:train:24001-24100batch: iter_time=1.974e-04, forward_time=0.115, loss_ctc=58.978, loss_att=53.472, acc=0.681, loss=55.124, backward_time=0.137, grad_norm=27.630, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.834e-04, train_time=2.511
+[gpuc01:0/16] 2024-01-24 14:36:01,174 (trainer:737) INFO: 7epoch:train:24101-24200batch: iter_time=1.872e-04, forward_time=0.116, loss_ctc=65.489, loss_att=71.554, acc=0.661, loss=69.734, backward_time=0.138, grad_norm=31.009, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.833e-04, train_time=2.519
+[gpuc01:0/16] 2024-01-24 14:38:07,166 (trainer:737) INFO: 7epoch:train:24201-24300batch: iter_time=1.785e-04, forward_time=0.117, loss_ctc=64.014, loss_att=62.086, acc=0.699, loss=62.664, backward_time=0.138, grad_norm=25.946, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.832e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 14:40:12,860 (trainer:737) INFO: 7epoch:train:24301-24400batch: iter_time=2.033e-04, forward_time=0.116, loss_ctc=64.981, loss_att=62.928, acc=0.669, loss=63.544, backward_time=0.137, grad_norm=25.888, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.832e-04, train_time=2.514
+[gpuc01:0/16] 2024-01-24 14:42:18,785 (trainer:737) INFO: 7epoch:train:24401-24500batch: iter_time=2.108e-04, forward_time=0.116, loss_ctc=58.363, loss_att=61.167, acc=0.675, loss=60.326, backward_time=0.137, grad_norm=23.633, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.831e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 14:44:25,043 (trainer:737) INFO: 7epoch:train:24501-24600batch: iter_time=2.148e-04, forward_time=0.116, loss_ctc=55.831, loss_att=60.711, acc=0.692, loss=59.247, backward_time=0.137, grad_norm=24.636, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.830e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 14:46:30,975 (trainer:737) INFO: 7epoch:train:24601-24700batch: iter_time=2.083e-04, forward_time=0.116, loss_ctc=57.776, loss_att=63.447, acc=0.668, loss=61.745, backward_time=0.137, grad_norm=23.621, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.829e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 14:48:36,994 (trainer:737) INFO: 7epoch:train:24701-24800batch: iter_time=2.262e-04, forward_time=0.115, loss_ctc=65.255, loss_att=50.611, acc=0.693, loss=55.005, backward_time=0.137, grad_norm=29.550, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.828e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 14:50:43,833 (trainer:737) INFO: 7epoch:train:24801-24900batch: iter_time=2.066e-04, forward_time=0.116, loss_ctc=62.463, loss_att=62.512, acc=0.663, loss=62.497, backward_time=0.138, grad_norm=29.146, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.827e-04, train_time=2.537
+[gpuc01:0/16] 2024-01-24 14:52:50,333 (trainer:737) INFO: 7epoch:train:24901-25000batch: iter_time=1.947e-04, forward_time=0.117, loss_ctc=64.281, loss_att=62.655, acc=0.668, loss=63.143, backward_time=0.137, grad_norm=28.472, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.826e-04, train_time=2.530
+[gpuc01:0/16] 2024-01-24 14:52:52,492 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpuc01:0/16] 2024-01-24 14:53:11,866 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 14:53:15,470 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f66dbd950f0>)
+[gpuc01:0/16] 2024-01-24 14:53:15,470 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpuc01:0/16] 2024-01-24 14:53:15,476 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 14:57:17,989 (trainer:737) INFO: 7epoch:train:25001-25100batch: iter_time=1.388, forward_time=0.116, loss_ctc=63.892, loss_att=65.764, acc=0.666, loss=65.202, backward_time=0.138, grad_norm=30.321, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.825e-04, train_time=5.353
+[gpuc01:0/16] 2024-01-24 14:59:23,413 (trainer:737) INFO: 7epoch:train:25101-25200batch: iter_time=2.456e-04, forward_time=0.116, loss_ctc=59.209, loss_att=57.156, acc=0.699, loss=57.772, backward_time=0.137, grad_norm=25.184, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.824e-04, train_time=2.508
+[gpuc01:0/16] 2024-01-24 15:01:28,004 (trainer:737) INFO: 7epoch:train:25201-25300batch: iter_time=2.623e-04, forward_time=0.115, loss_ctc=60.336, loss_att=56.390, acc=0.684, loss=57.574, backward_time=0.137, grad_norm=25.175, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.823e-04, train_time=2.492
+[gpuc01:0/16] 2024-01-24 15:03:32,911 (trainer:737) INFO: 7epoch:train:25301-25400batch: iter_time=2.579e-04, forward_time=0.116, loss_ctc=66.050, loss_att=64.991, acc=0.692, loss=65.308, backward_time=0.138, grad_norm=27.847, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.822e-04, train_time=2.498
+[gpuc01:0/16] 2024-01-24 15:05:37,897 (trainer:737) INFO: 7epoch:train:25401-25500batch: iter_time=2.361e-04, forward_time=0.116, loss_ctc=71.756, loss_att=61.575, acc=0.694, loss=64.630, backward_time=0.138, grad_norm=28.826, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.821e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 15:07:43,480 (trainer:737) INFO: 7epoch:train:25501-25600batch: iter_time=2.515e-04, forward_time=0.115, loss_ctc=57.415, loss_att=58.735, acc=0.697, loss=58.339, backward_time=0.137, grad_norm=23.149, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.820e-04, train_time=2.511
+[gpuc01:0/16] 2024-01-24 15:09:49,512 (trainer:737) INFO: 7epoch:train:25601-25700batch: iter_time=2.488e-04, forward_time=0.115, loss_ctc=56.323, loss_att=52.465, acc=0.669, loss=53.623, backward_time=0.136, grad_norm=27.112, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.819e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 15:11:54,485 (trainer:737) INFO: 7epoch:train:25701-25800batch: iter_time=2.131e-04, forward_time=0.116, loss_ctc=64.295, loss_att=59.453, acc=0.677, loss=60.905, backward_time=0.137, grad_norm=30.210, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.818e-04, train_time=2.499
+[gpuc01:0/16] 2024-01-24 15:13:59,799 (trainer:737) INFO: 7epoch:train:25801-25900batch: iter_time=2.267e-04, forward_time=0.116, loss_ctc=67.870, loss_att=62.519, acc=0.690, loss=64.124, backward_time=0.138, grad_norm=27.000, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.818e-04, train_time=2.506
+[gpuc01:0/16] 2024-01-24 15:16:05,110 (trainer:737) INFO: 7epoch:train:25901-26000batch: iter_time=2.259e-04, forward_time=0.115, loss_ctc=52.834, loss_att=53.878, acc=0.689, loss=53.565, backward_time=0.137, grad_norm=21.404, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.817e-04, train_time=2.506
+[gpuc01:0/16] 2024-01-24 15:18:10,105 (trainer:737) INFO: 7epoch:train:26001-26100batch: iter_time=2.301e-04, forward_time=0.116, loss_ctc=56.199, loss_att=60.674, acc=0.679, loss=59.332, backward_time=0.137, grad_norm=23.001, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.816e-04, train_time=2.500
+[gpuc01:0/16] 2024-01-24 15:20:14,995 (trainer:737) INFO: 7epoch:train:26101-26200batch: iter_time=2.394e-04, forward_time=0.115, loss_ctc=55.730, loss_att=55.015, acc=0.690, loss=55.230, backward_time=0.137, grad_norm=22.531, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.815e-04, train_time=2.498
+[gpuc01:0/16] 2024-01-24 15:22:20,197 (trainer:737) INFO: 7epoch:train:26201-26300batch: iter_time=2.536e-04, forward_time=0.115, loss_ctc=57.211, loss_att=52.355, acc=0.707, loss=53.812, backward_time=0.137, grad_norm=25.337, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.814e-04, train_time=2.504
+[gpuc01:0/16] 2024-01-24 15:24:25,257 (trainer:737) INFO: 7epoch:train:26301-26400batch: iter_time=2.439e-04, forward_time=0.116, loss_ctc=74.434, loss_att=71.113, acc=0.666, loss=72.109, backward_time=0.138, grad_norm=43.402, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.813e-04, train_time=2.501
+[gpuc01:0/16] 2024-01-24 15:26:30,629 (trainer:737) INFO: 7epoch:train:26401-26500batch: iter_time=2.255e-04, forward_time=0.115, loss_ctc=57.686, loss_att=66.091, acc=0.670, loss=63.570, backward_time=0.137, grad_norm=27.586, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.812e-04, train_time=2.507
+[gpuc01:0/16] 2024-01-24 15:28:36,295 (trainer:737) INFO: 7epoch:train:26501-26600batch: iter_time=2.371e-04, forward_time=0.116, loss_ctc=59.051, loss_att=51.088, acc=0.697, loss=53.477, backward_time=0.137, grad_norm=27.382, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.811e-04, train_time=2.513
+[gpuc01:0/16] 2024-01-24 15:30:41,483 (trainer:737) INFO: 7epoch:train:26601-26700batch: iter_time=2.313e-04, forward_time=0.115, loss_ctc=64.201, loss_att=75.840, acc=0.663, loss=72.348, backward_time=0.138, grad_norm=31.402, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.810e-04, train_time=2.504
+[gpuc01:0/16] 2024-01-24 15:32:46,763 (trainer:737) INFO: 7epoch:train:26701-26800batch: iter_time=2.167e-04, forward_time=0.115, loss_ctc=64.542, loss_att=64.325, acc=0.701, loss=64.390, backward_time=0.138, grad_norm=27.131, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.809e-04, train_time=2.505
+[gpuc01:0/16] 2024-01-24 15:34:52,090 (trainer:737) INFO: 7epoch:train:26801-26900batch: iter_time=2.584e-04, forward_time=0.115, loss_ctc=64.766, loss_att=62.343, acc=0.680, loss=63.070, backward_time=0.137, grad_norm=24.646, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.808e-04, train_time=2.506
+[gpuc01:0/16] 2024-01-24 15:36:57,002 (trainer:737) INFO: 7epoch:train:26901-27000batch: iter_time=2.546e-04, forward_time=0.116, loss_ctc=58.203, loss_att=60.411, acc=0.692, loss=59.749, backward_time=0.137, grad_norm=25.580, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.807e-04, train_time=2.498
+[gpuc01:0/16] 2024-01-24 15:39:03,352 (trainer:737) INFO: 7epoch:train:27001-27100batch: iter_time=2.332e-04, forward_time=0.115, loss_ctc=55.620, loss_att=59.438, acc=0.696, loss=58.293, backward_time=0.137, grad_norm=25.862, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.806e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 15:41:08,841 (trainer:737) INFO: 7epoch:train:27101-27200batch: iter_time=2.549e-04, forward_time=0.116, loss_ctc=57.387, loss_att=63.540, acc=0.678, loss=61.694, backward_time=0.137, grad_norm=24.405, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.806e-04, train_time=2.510
+[gpuc01:0/16] 2024-01-24 15:43:14,073 (trainer:737) INFO: 7epoch:train:27201-27300batch: iter_time=2.265e-04, forward_time=0.116, loss_ctc=66.745, loss_att=51.581, acc=0.703, loss=56.130, backward_time=0.137, grad_norm=29.988, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.805e-04, train_time=2.504
+[gpuc01:0/16] 2024-01-24 15:45:19,229 (trainer:737) INFO: 7epoch:train:27301-27400batch: iter_time=1.965e-04, forward_time=0.116, loss_ctc=61.701, loss_att=61.911, acc=0.669, loss=61.848, backward_time=0.136, grad_norm=26.492, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.804e-04, train_time=2.503
+[gpuc01:0/16] 2024-01-24 15:47:24,351 (trainer:737) INFO: 7epoch:train:27401-27500batch: iter_time=1.631e-04, forward_time=0.115, loss_ctc=63.033, loss_att=62.752, acc=0.686, loss=62.836, backward_time=0.137, grad_norm=26.671, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.803e-04, train_time=2.502
+[gpuc01:0/16] 2024-01-24 15:47:27,554 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpuc01:0/16] 2024-01-24 15:47:46,074 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 15:47:50,034 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f66c1655f90>)
+[gpuc01:0/16] 2024-01-24 15:47:50,034 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpuc01:0/16] 2024-01-24 15:47:50,041 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 15:51:53,618 (trainer:737) INFO: 7epoch:train:27501-27600batch: iter_time=1.405, forward_time=0.115, loss_ctc=63.783, loss_att=62.860, acc=0.675, loss=63.137, backward_time=0.138, grad_norm=31.575, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.802e-04, train_time=5.385
+[gpuc01:0/16] 2024-01-24 15:53:59,862 (trainer:737) INFO: 7epoch:train:27601-27700batch: iter_time=2.287e-04, forward_time=0.116, loss_ctc=58.338, loss_att=56.150, acc=0.701, loss=56.807, backward_time=0.137, grad_norm=24.387, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.801e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 15:56:06,002 (trainer:737) INFO: 7epoch:train:27701-27800batch: iter_time=2.536e-04, forward_time=0.115, loss_ctc=59.967, loss_att=54.725, acc=0.688, loss=56.298, backward_time=0.137, grad_norm=23.476, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.068, optim0_lr0=3.800e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 15:58:12,162 (trainer:737) INFO: 7epoch:train:27801-27900batch: iter_time=2.197e-04, forward_time=0.116, loss_ctc=66.015, loss_att=65.465, acc=0.689, loss=65.630, backward_time=0.138, grad_norm=27.745, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.799e-04, train_time=2.523
+[gpuc01:0/16] 2024-01-24 16:00:18,390 (trainer:737) INFO: 7epoch:train:27901-28000batch: iter_time=2.055e-04, forward_time=0.115, loss_ctc=71.140, loss_att=62.166, acc=0.695, loss=64.858, backward_time=0.139, grad_norm=28.398, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.067, optim0_lr0=3.798e-04, train_time=2.524
+[gpuc01:0/16] 2024-01-24 16:02:24,428 (trainer:737) INFO: 7epoch:train:28001-28100batch: iter_time=2.163e-04, forward_time=0.116, loss_ctc=56.700, loss_att=57.517, acc=0.700, loss=57.272, backward_time=0.138, grad_norm=23.478, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.797e-04, train_time=2.521
+[gpuc01:0/16] 2024-01-24 16:04:30,272 (trainer:737) INFO: 7epoch:train:28101-28200batch: iter_time=2.092e-04, forward_time=0.114, loss_ctc=56.311, loss_att=52.118, acc=0.673, loss=53.376, backward_time=0.138, grad_norm=22.917, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.796e-04, train_time=2.517
+[gpuc01:0/16] 2024-01-24 16:06:36,597 (trainer:737) INFO: 7epoch:train:28201-28300batch: iter_time=2.273e-04, forward_time=0.116, loss_ctc=64.664, loss_att=57.463, acc=0.681, loss=59.623, backward_time=0.138, grad_norm=27.550, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.795e-04, train_time=2.526
+[gpuc01:0/16] 2024-01-24 16:08:42,902 (trainer:737) INFO: 7epoch:train:28301-28400batch: iter_time=3.015e-04, forward_time=0.119, loss_ctc=66.226, loss_att=60.638, acc=0.696, loss=62.314, backward_time=0.138, grad_norm=27.357, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.795e-04, train_time=2.526
+[gpuc01:0/16] 2024-01-24 16:10:48,709 (trainer:737) INFO: 7epoch:train:28401-28500batch: iter_time=2.163e-04, forward_time=0.120, loss_ctc=53.333, loss_att=53.675, acc=0.694, loss=53.572, backward_time=0.137, grad_norm=22.577, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.794e-04, train_time=2.516
+[gpuc01:0/16] 2024-01-24 16:12:54,827 (trainer:737) INFO: 7epoch:train:28501-28600batch: iter_time=2.224e-04, forward_time=0.122, loss_ctc=56.555, loss_att=61.005, acc=0.677, loss=59.670, backward_time=0.137, grad_norm=22.946, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.793e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 16:15:00,817 (trainer:737) INFO: 7epoch:train:28601-28700batch: iter_time=2.218e-04, forward_time=0.119, loss_ctc=55.494, loss_att=54.062, acc=0.695, loss=54.492, backward_time=0.137, grad_norm=24.332, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.792e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 16:17:07,370 (trainer:737) INFO: 7epoch:train:28701-28800batch: iter_time=2.995e-04, forward_time=0.114, loss_ctc=56.846, loss_att=51.890, acc=0.709, loss=53.377, backward_time=0.138, grad_norm=23.617, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.791e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 16:19:13,743 (trainer:737) INFO: 7epoch:train:28801-28900batch: iter_time=2.773e-04, forward_time=0.116, loss_ctc=73.620, loss_att=69.733, acc=0.671, loss=70.899, backward_time=0.139, grad_norm=31.302, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.790e-04, train_time=2.527
+[gpuc01:0/16] 2024-01-24 16:21:20,651 (trainer:737) INFO: 7epoch:train:28901-29000batch: iter_time=2.236e-04, forward_time=0.115, loss_ctc=57.057, loss_att=65.601, acc=0.669, loss=63.038, backward_time=0.137, grad_norm=26.144, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.789e-04, train_time=2.538
+[gpuc01:0/16] 2024-01-24 16:23:27,208 (trainer:737) INFO: 7epoch:train:29001-29100batch: iter_time=2.193e-04, forward_time=0.114, loss_ctc=59.028, loss_att=51.379, acc=0.695, loss=53.674, backward_time=0.136, grad_norm=26.799, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.788e-04, train_time=2.531
+[gpuc01:0/16] 2024-01-24 16:25:33,858 (trainer:737) INFO: 7epoch:train:29101-29200batch: iter_time=2.158e-04, forward_time=0.115, loss_ctc=64.861, loss_att=75.950, acc=0.664, loss=72.624, backward_time=0.138, grad_norm=28.709, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.787e-04, train_time=2.533
+[gpuc01:0/16] 2024-01-24 16:27:40,298 (trainer:737) INFO: 7epoch:train:29201-29300batch: iter_time=2.112e-04, forward_time=0.115, loss_ctc=63.116, loss_att=62.873, acc=0.702, loss=62.946, backward_time=0.138, grad_norm=26.521, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.786e-04, train_time=2.529
+[gpuc01:0/16] 2024-01-24 16:29:46,422 (trainer:737) INFO: 7epoch:train:29301-29400batch: iter_time=2.165e-04, forward_time=0.115, loss_ctc=64.441, loss_att=62.551, acc=0.680, loss=63.118, backward_time=0.137, grad_norm=24.327, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.066, optim0_lr0=3.785e-04, train_time=2.522
+[gpuc01:0/16] 2024-01-24 16:31:52,911 (trainer:737) INFO: 7epoch:train:29401-29500batch: iter_time=2.188e-04, forward_time=0.114, loss_ctc=57.810, loss_att=60.849, acc=0.692, loss=59.938, backward_time=0.137, grad_norm=23.530, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.785e-04, train_time=2.530
+[gpuc01:0/16] 2024-01-24 16:33:59,168 (trainer:737) INFO: 7epoch:train:29501-29600batch: iter_time=2.348e-04, forward_time=0.114, loss_ctc=55.109, loss_att=60.406, acc=0.692, loss=58.817, backward_time=0.137, grad_norm=26.949, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.784e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 16:36:08,873 (trainer:737) INFO: 7epoch:train:29601-29700batch: iter_time=1.886e-04, forward_time=0.114, loss_ctc=57.684, loss_att=64.400, acc=0.677, loss=62.385, backward_time=0.137, grad_norm=27.419, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.783e-04, train_time=2.594
+[gpuc01:0/16] 2024-01-24 16:38:15,477 (trainer:737) INFO: 7epoch:train:29701-29800batch: iter_time=2.075e-04, forward_time=0.114, loss_ctc=65.076, loss_att=50.910, acc=0.707, loss=55.160, backward_time=0.137, grad_norm=31.974, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.782e-04, train_time=2.532
+[gpuc01:0/16] 2024-01-24 16:40:21,479 (trainer:737) INFO: 7epoch:train:29801-29900batch: iter_time=2.110e-04, forward_time=0.114, loss_ctc=62.075, loss_att=61.983, acc=0.668, loss=62.010, backward_time=0.137, grad_norm=30.801, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.781e-04, train_time=2.520
+[gpuc01:0/16] 2024-01-24 16:42:27,759 (trainer:737) INFO: 7epoch:train:29901-30000batch: iter_time=1.955e-04, forward_time=0.114, loss_ctc=63.348, loss_att=62.786, acc=0.688, loss=62.954, backward_time=0.137, grad_norm=28.008, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.780e-04, train_time=2.525
+[gpuc01:0/16] 2024-01-24 17:18:48,977 (trainer:343) INFO: 7epoch results: [train] iter_time=0.067, forward_time=0.117, loss_ctc=63.680, loss_att=62.237, acc=0.673, loss=62.670, backward_time=0.137, grad_norm=26.919, clip=100.000, loss_scale=5.872e+19, optim_step_time=0.067, optim0_lr0=3.925e-04, train_time=2.804, time=11 hours, 41 minutes and 4.29 seconds, total_count=135000, gpu_max_cached_mem_GB=25.098, [valid] loss_ctc=56.140, cer_ctc=0.294, loss_att=53.668, acc=0.580, cer=0.415, wer=1.000, loss=54.410, time=36 minutes and 13.51 seconds, total_count=42039, gpu_max_cached_mem_GB=25.098
+[gpuc01:0/16] 2024-01-24 17:18:58,243 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpuc01:0/16] 2024-01-24 17:18:58,261 (trainer:445) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/2epoch.pth
+[gpuc01:0/16] 2024-01-24 17:18:58,261 (trainer:272) INFO: 8/45epoch started. Estimated time to finish: 2 weeks, 6 days and 21 hours
+[gpuc01:0/16] 2024-01-24 17:18:58,278 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpuc01:0/16] 2024-01-24 17:19:16,389 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpuc01:0/16] 2024-01-24 17:19:19,781 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f66e4cba530>)
+[gpuc01:0/16] 2024-01-24 17:19:19,781 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=38055, batch_size=128, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpuc01:0/16] 2024-01-24 17:19:19,787 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=38055, mean=128.0, min=128, max=129
+[gpuc01:0/16] 2024-01-24 17:23:15,657 (trainer:737) INFO: 8epoch:train:1-100batch: iter_time=1.319, forward_time=0.118, loss_ctc=70.696, loss_att=74.686, acc=0.679, loss=73.489, backward_time=0.139, grad_norm=28.237, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.779e-04, train_time=5.147
+[gpuc01:0/16] 2024-01-24 17:25:20,661 (trainer:737) INFO: 8epoch:train:101-200batch: iter_time=1.950e-04, forward_time=0.116, loss_ctc=57.839, loss_att=55.395, acc=0.684, loss=56.128, backward_time=0.138, grad_norm=28.965, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.778e-04, train_time=2.500
+[gpuc01:0/16] 2024-01-24 17:27:25,793 (trainer:737) INFO: 8epoch:train:201-300batch: iter_time=1.899e-04, forward_time=0.117, loss_ctc=62.583, loss_att=63.256, acc=0.674, loss=63.054, backward_time=0.139, grad_norm=26.807, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.777e-04, train_time=2.502
+[gpuc01:0/16] 2024-01-24 17:29:31,712 (trainer:737) INFO: 8epoch:train:301-400batch: iter_time=1.965e-04, forward_time=0.117, loss_ctc=62.899, loss_att=69.622, acc=0.675, loss=67.605, backward_time=0.139, grad_norm=27.557, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.776e-04, train_time=2.518
+[gpuc01:0/16] 2024-01-24 17:31:36,818 (trainer:737) INFO: 8epoch:train:401-500batch: iter_time=1.917e-04, forward_time=0.117, loss_ctc=63.634, loss_att=59.670, acc=0.698, loss=60.859, backward_time=0.139, grad_norm=28.627, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.776e-04, train_time=2.502
+[gpuc01:0/16] 2024-01-24 17:33:42,016 (trainer:737) INFO: 8epoch:train:501-600batch: iter_time=1.967e-04, forward_time=0.117, loss_ctc=59.203, loss_att=50.668, acc=0.705, loss=53.228, backward_time=0.138, grad_norm=28.469, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.775e-04, train_time=2.504
+[gpuc01:0/16] 2024-01-24 17:35:47,197 (trainer:737) INFO: 8epoch:train:601-700batch: iter_time=2.039e-04, forward_time=0.116, loss_ctc=63.595, loss_att=64.318, acc=0.664, loss=64.101, backward_time=0.139, grad_norm=25.200, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.774e-04, train_time=2.503
+[gpuc01:0/16] 2024-01-24 17:37:51,926 (trainer:737) INFO: 8epoch:train:701-800batch: iter_time=2.025e-04, forward_time=0.116, loss_ctc=59.763, loss_att=57.125, acc=0.676, loss=57.916, backward_time=0.139, grad_norm=25.372, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.067, optim0_lr0=3.773e-04, train_time=2.494
+[gpuc01:0/16] 2024-01-24 17:39:57,070 (trainer:737) INFO: 8epoch:train:801-900batch: iter_time=2.039e-04, forward_time=0.117, loss_ctc=60.454, loss_att=63.483, acc=0.676, loss=62.574, backward_time=0.139, grad_norm=24.788, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.068, optim0_lr0=3.772e-04, train_time=2.503
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2886934.0 ON gpuc01 CANCELLED AT 2024-01-24T17:41:28 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.17.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.17.log
new file mode 100644
index 0000000000000000000000000000000000000000..566b493d0c3befcde7122578c68a2d8b6ba23542
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.17.log
@@ -0,0 +1,2080 @@
+# Running on gpub052.delta.ncsa.illinois.edu
+# Started at Mon Jan 22 05:52:51 CST 2024
+# SLURMD_NODENAME=gpub052
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2878933
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1706097127
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2878933
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[052,054-056]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1705924327
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[052,054-056]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login01.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2868225
+# SLURM_TOPOLOGY_ADDR=ss00.ss11.gpub052
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_318fdf1c-27cd-4320-ad3f-342d16f1467f 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_318fdf1c-27cd-4320-ad3f-342d16f1467f
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_318fdf1c-27cd-4320-ad3f-342d16f1467f
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_318fdf1c-27cd-4320-ad3f-342d16f1467f
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_318fdf1c-27cd-4320-ad3f-342d16f1467f
+[gpub052:0/16] 2024-01-22 06:04:08,611 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub052:0/16] 2024-01-22 06:04:08,706 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub052:0/16] 2024-01-22 06:04:08,824 (s2t:464) INFO: Vocabulary size: 50002
+[gpub052:0/16] 2024-01-22 06:04:32,348 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub052:0/16] 2024-01-22 06:04:32,354 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub052:0/16] 2024-01-22 06:04:32,354 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub052:0/16] 2024-01-22 06:04:32,354 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub052:0/16] 2024-01-22 06:04:32,356 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub052:0/16] 2024-01-22 06:04:38,741 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 06:04:39,911 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f624ecde350>)
+[gpub052:0/16] 2024-01-22 06:04:39,911 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub052:0/16] 2024-01-22 06:04:39,938 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 06:05:16,876 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub052:2868421:2868421 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.152<0>
+gpub052:2868421:2868421 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub052:2868421:2868421 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub052:0/16] 2024-01-22 06:05:31,064 (trainer:284) INFO: 3/45epoch started
+[gpub052:0/16] 2024-01-22 06:05:31,112 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub052:0/16] 2024-01-22 06:05:49,680 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 06:05:53,155 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6067bc5510>)
+[gpub052:0/16] 2024-01-22 06:05:53,155 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub052:0/16] 2024-01-22 06:05:53,158 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub052:2868422:2868422 [1] NCCL INFO cudaDriverVersion 12020
+gpub052:2868422:2868422 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.152<0>
+gpub052:2868422:2868422 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub052:2868422:2868495 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.152<0>
+gpub052:2868422:2868495 [1] NCCL INFO Using network IB
+gpub052:2868422:2868495 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub052:2868422:2868495 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub052:2868422:2868495 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub052:2868422:2868495 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub052:2868422:2868495 [1] NCCL INFO Connected all rings
+gpub052:2868422:2868495 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub052:2868422:2868495 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub052:2868422:2868495 [1] NCCL INFO Connected all trees
+gpub052:2868422:2868495 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub052:2868422:2868495 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub052:2868422:2868495 [1] NCCL INFO comm 0x14e813e0 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub056:538247:538247 [1] NCCL INFO cudaDriverVersion 12020
+gpub056:538247:538247 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.156<0>
+gpub056:538247:538247 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub056:538247:538322 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.156<0>
+gpub056:538247:538322 [1] NCCL INFO Using network IB
+gpub056:538247:538322 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub056:538247:538322 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub056:538247:538322 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub056:538247:538322 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub056:538247:538322 [1] NCCL INFO Connected all rings
+gpub056:538247:538322 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub056:538247:538322 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub056:538247:538322 [1] NCCL INFO Connected all trees
+gpub056:538247:538322 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub056:538247:538322 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub056:538247:538322 [1] NCCL INFO comm 0x11edd1a0 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub052:2868424:2868424 [3] NCCL INFO cudaDriverVersion 12020
+gpub052:2868424:2868424 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.152<0>
+gpub052:2868424:2868424 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub052:2868424:2868497 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.152<0>
+gpub052:2868424:2868497 [3] NCCL INFO Using network IB
+gpub052:2868424:2868497 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub052:2868424:2868497 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub052:2868424:2868497 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/IB/0
+gpub052:2868424:2868497 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/IB/0
+gpub052:2868424:2868497 [3] NCCL INFO Connected all rings
+gpub052:2868424:2868497 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub052:2868424:2868497 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub052:2868424:2868497 [3] NCCL INFO Connected all trees
+gpub052:2868424:2868497 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub052:2868424:2868497 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub052:2868424:2868497 [3] NCCL INFO comm 0x127ac190 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub056:538249:538249 [3] NCCL INFO cudaDriverVersion 12020
+gpub056:538249:538249 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.156<0>
+gpub056:538249:538249 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub056:538249:538323 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.156<0>
+gpub056:538249:538323 [3] NCCL INFO Using network IB
+gpub056:538249:538323 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub056:538249:538323 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub056:538249:538323 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/IB/0
+gpub056:538249:538323 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/IB/0
+gpub056:538249:538323 [3] NCCL INFO Connected all rings
+gpub056:538249:538323 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub056:538249:538323 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub056:538249:538323 [3] NCCL INFO Connected all trees
+gpub056:538249:538323 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub056:538249:538323 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub056:538249:538323 [3] NCCL INFO comm 0x1523bce0 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub055:132138:132138 [3] NCCL INFO cudaDriverVersion 12020
+gpub055:132138:132138 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.155<0>
+gpub055:132138:132138 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub055:132138:132200 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.155<0>
+gpub055:132138:132200 [3] NCCL INFO Using network IB
+gpub055:132138:132200 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub055:132138:132200 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub055:132138:132200 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/IB/0
+gpub055:132138:132200 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/IB/0
+gpub055:132138:132200 [3] NCCL INFO Connected all rings
+gpub055:132138:132200 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub055:132138:132200 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub055:132138:132200 [3] NCCL INFO Connected all trees
+gpub055:132138:132200 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub055:132138:132200 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub055:132138:132200 [3] NCCL INFO comm 0x751a62a0 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub055:132136:132136 [1] NCCL INFO cudaDriverVersion 12020
+gpub055:132136:132136 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.155<0>
+gpub055:132136:132136 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub055:132136:132199 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.155<0>
+gpub055:132136:132199 [1] NCCL INFO Using network IB
+gpub055:132136:132199 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub055:132136:132199 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub055:132136:132199 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub055:132136:132199 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub055:132136:132199 [1] NCCL INFO Connected all rings
+gpub055:132136:132199 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/IB/0
+gpub055:132136:132199 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/IB/0
+gpub055:132136:132199 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub055:132136:132199 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub055:132136:132199 [1] NCCL INFO Connected all trees
+gpub055:132136:132199 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub055:132136:132199 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub055:132136:132199 [1] NCCL INFO comm 0x17d720d0 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub055:132135:132135 [0] NCCL INFO cudaDriverVersion 12020
+gpub055:132135:132135 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.155<0>
+gpub055:132135:132135 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub055:132135:132197 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.155<0>
+gpub055:132135:132197 [0] NCCL INFO Using network IB
+gpub055:132135:132197 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub055:132135:132197 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub055:132135:132197 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub055:132135:132197 [0] NCCL INFO Connected all rings
+gpub055:132135:132197 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/IB/0
+gpub055:132135:132197 [0] NCCL INFO Connected all trees
+gpub055:132135:132197 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub055:132135:132197 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub055:132135:132197 [0] NCCL INFO comm 0x15efd860 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub056:538248:538248 [2] NCCL INFO cudaDriverVersion 12020
+gpub056:538248:538248 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.156<0>
+gpub056:538248:538248 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub056:538248:538324 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.156<0>
+gpub056:538248:538324 [2] NCCL INFO Using network IB
+gpub056:538248:538324 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub056:538248:538324 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub056:538248:538324 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub056:538248:538324 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub056:538248:538324 [2] NCCL INFO Connected all rings
+gpub056:538248:538324 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub056:538248:538324 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub056:538248:538324 [2] NCCL INFO Connected all trees
+gpub056:538248:538324 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub056:538248:538324 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub056:538248:538324 [2] NCCL INFO comm 0x157385a0 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub052:2868421:2868496 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.152<0>
+gpub052:2868421:2868496 [0] NCCL INFO Using network IB
+gpub052:2868421:2868496 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub052:2868421:2868496 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub052:2868421:2868496 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub052:2868421:2868496 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub052:2868421:2868496 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub052:2868421:2868496 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub052:2868421:2868496 [0] NCCL INFO Connected all rings
+gpub052:2868421:2868496 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/IB/0
+gpub052:2868421:2868496 [0] NCCL INFO Connected all trees
+gpub052:2868421:2868496 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub052:2868421:2868496 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub052:2868421:2868496 [0] NCCL INFO comm 0x16b58ec0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub055:132137:132137 [2] NCCL INFO cudaDriverVersion 12020
+gpub055:132137:132137 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.155<0>
+gpub055:132137:132137 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub055:132137:132198 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.155<0>
+gpub055:132137:132198 [2] NCCL INFO Using network IB
+gpub055:132137:132198 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub055:132137:132198 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub055:132137:132198 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub055:132137:132198 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub055:132137:132198 [2] NCCL INFO Connected all rings
+gpub055:132137:132198 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub055:132137:132198 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub055:132137:132198 [2] NCCL INFO Connected all trees
+gpub055:132137:132198 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub055:132137:132198 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub055:132137:132198 [2] NCCL INFO comm 0x1abfe340 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub056:538246:538246 [0] NCCL INFO cudaDriverVersion 12020
+gpub056:538246:538246 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.156<0>
+gpub056:538246:538246 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub056:538246:538321 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.156<0>
+gpub056:538246:538321 [0] NCCL INFO Using network IB
+gpub056:538246:538321 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub056:538246:538321 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub056:538246:538321 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub056:538246:538321 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub056:538246:538321 [0] NCCL INFO Connected all rings
+gpub056:538246:538321 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/IB/0
+gpub056:538246:538321 [0] NCCL INFO Connected all trees
+gpub056:538246:538321 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub056:538246:538321 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub056:538246:538321 [0] NCCL INFO comm 0x15674980 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub052:2868423:2868423 [2] NCCL INFO cudaDriverVersion 12020
+gpub052:2868423:2868423 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.152<0>
+gpub052:2868423:2868423 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub052:2868423:2868494 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.152<0>
+gpub052:2868423:2868494 [2] NCCL INFO Using network IB
+gpub052:2868423:2868494 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub052:2868423:2868494 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub052:2868423:2868494 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub052:2868423:2868494 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub052:2868423:2868494 [2] NCCL INFO Connected all rings
+gpub052:2868423:2868494 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub052:2868423:2868494 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub052:2868423:2868494 [2] NCCL INFO Connected all trees
+gpub052:2868423:2868494 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub052:2868423:2868494 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub052:2868423:2868494 [2] NCCL INFO comm 0x181269b0 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub054:3241515:3241515 [2] NCCL INFO cudaDriverVersion 12020
+gpub054:3241515:3241515 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.154<0>
+gpub054:3241515:3241515 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub054:3241515:3241576 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.154<0>
+gpub054:3241515:3241576 [2] NCCL INFO Using network IB
+gpub054:3241515:3241576 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub054:3241515:3241576 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub054:3241515:3241576 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub054:3241515:3241576 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub054:3241515:3241576 [2] NCCL INFO Connected all rings
+gpub054:3241515:3241576 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub054:3241515:3241576 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub054:3241515:3241576 [2] NCCL INFO Connected all trees
+gpub054:3241515:3241576 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub054:3241515:3241576 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub054:3241515:3241576 [2] NCCL INFO comm 0x10a74200 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub054:3241514:3241514 [1] NCCL INFO cudaDriverVersion 12020
+gpub054:3241514:3241514 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.154<0>
+gpub054:3241514:3241514 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub054:3241514:3241577 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.154<0>
+gpub054:3241514:3241577 [1] NCCL INFO Using network IB
+gpub054:3241514:3241577 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub054:3241514:3241577 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub054:3241514:3241577 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub054:3241514:3241577 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub054:3241514:3241577 [1] NCCL INFO Connected all rings
+gpub054:3241514:3241577 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/IB/0
+gpub054:3241514:3241577 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/IB/0
+gpub054:3241514:3241577 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub054:3241514:3241577 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub054:3241514:3241577 [1] NCCL INFO Connected all trees
+gpub054:3241514:3241577 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub054:3241514:3241577 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub054:3241514:3241577 [1] NCCL INFO comm 0x15ce2950 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub054:3241516:3241516 [3] NCCL INFO cudaDriverVersion 12020
+gpub054:3241516:3241516 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.154<0>
+gpub054:3241516:3241516 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub054:3241516:3241575 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.154<0>
+gpub054:3241516:3241575 [3] NCCL INFO Using network IB
+gpub054:3241516:3241575 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub054:3241516:3241575 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub054:3241516:3241575 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/IB/0
+gpub054:3241516:3241575 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/IB/0
+gpub054:3241516:3241575 [3] NCCL INFO Connected all rings
+gpub054:3241516:3241575 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub054:3241516:3241575 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub054:3241516:3241575 [3] NCCL INFO Connected all trees
+gpub054:3241516:3241575 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub054:3241516:3241575 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub054:3241516:3241575 [3] NCCL INFO comm 0x14911a10 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub054:3241513:3241513 [0] NCCL INFO cudaDriverVersion 12020
+gpub054:3241513:3241513 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.154<0>
+gpub054:3241513:3241513 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub054:3241513:3241574 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.154<0>
+gpub054:3241513:3241574 [0] NCCL INFO Using network IB
+gpub054:3241513:3241574 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub054:3241513:3241574 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub054:3241513:3241574 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub054:3241513:3241574 [0] NCCL INFO Connected all rings
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/IB/0
+gpub054:3241513:3241574 [0] NCCL INFO Connected all trees
+gpub054:3241513:3241574 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub054:3241513:3241574 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub054:3241513:3241574 [0] NCCL INFO comm 0x129ab3a0 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+[gpub052:0/16] 2024-01-22 06:21:48,296 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub052:0/16] 2024-01-22 06:27:01,567 (trainer:737) INFO: 3epoch:train:1-100batch: iter_time=6.494, forward_time=0.636, loss_ctc=164.980, loss_att=197.734, acc=0.226, loss=187.908, backward_time=0.471, grad_norm=67.384, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.107, optim0_lr0=5.077e-05, train_time=12.857
+[gpub052:0/16] 2024-01-22 06:33:41,819 (trainer:737) INFO: 3epoch:train:101-200batch: iter_time=1.010e-04, forward_time=0.642, loss_ctc=150.521, loss_att=196.377, acc=0.231, loss=182.620, backward_time=0.751, grad_norm=55.752, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.129, optim0_lr0=5.227e-05, train_time=4.049
+[gpub052:0/16] 2024-01-22 06:42:28,676 (trainer:737) INFO: 3epoch:train:201-300batch: iter_time=9.928e-05, forward_time=0.314, loss_ctc=156.589, loss_att=201.152, acc=0.223, loss=187.783, backward_time=0.417, grad_norm=60.450, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.100, optim0_lr0=5.377e-05, train_time=5.269
+[gpub052:0/16] 2024-01-22 06:51:06,256 (trainer:737) INFO: 3epoch:train:301-400batch: iter_time=1.048e-04, forward_time=0.460, loss_ctc=160.480, loss_att=204.048, acc=0.224, loss=190.978, backward_time=0.546, grad_norm=55.267, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.107, optim0_lr0=5.527e-05, train_time=5.174
+[gpub052:0/16] 2024-01-22 06:59:11,131 (trainer:737) INFO: 3epoch:train:401-500batch: iter_time=1.028e-04, forward_time=0.401, loss_ctc=167.439, loss_att=202.769, acc=0.228, loss=192.170, backward_time=0.400, grad_norm=70.942, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.097, optim0_lr0=5.677e-05, train_time=4.850
+[gpub052:0/16] 2024-01-22 07:06:51,276 (trainer:737) INFO: 3epoch:train:501-600batch: iter_time=0.001, forward_time=0.552, loss_ctc=150.975, loss_att=192.250, acc=0.226, loss=179.868, backward_time=0.513, grad_norm=62.260, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.105, optim0_lr0=5.827e-05, train_time=4.601
+[gpub052:0/16] 2024-01-22 07:15:37,430 (trainer:737) INFO: 3epoch:train:601-700batch: iter_time=9.891e-05, forward_time=0.298, loss_ctc=151.548, loss_att=194.596, acc=0.231, loss=181.681, backward_time=0.397, grad_norm=54.993, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.093, optim0_lr0=5.977e-05, train_time=5.261
+[gpub052:0/16] 2024-01-22 07:24:22,360 (trainer:737) INFO: 3epoch:train:701-800batch: iter_time=1.041e-04, forward_time=0.537, loss_ctc=185.442, loss_att=214.856, acc=0.220, loss=206.032, backward_time=0.482, grad_norm=79.463, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.111, optim0_lr0=6.127e-05, train_time=5.250
+[gpub052:0/16] 2024-01-22 07:32:30,100 (trainer:737) INFO: 3epoch:train:801-900batch: iter_time=1.008e-04, forward_time=0.302, loss_ctc=155.284, loss_att=210.979, acc=0.223, loss=194.271, backward_time=0.412, grad_norm=52.703, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.093, optim0_lr0=6.277e-05, train_time=4.876
+[gpub052:0/16] 2024-01-22 07:40:51,113 (trainer:737) INFO: 3epoch:train:901-1000batch: iter_time=9.960e-05, forward_time=0.306, loss_ctc=154.244, loss_att=209.748, acc=0.225, loss=193.097, backward_time=0.409, grad_norm=59.505, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.094, optim0_lr0=6.427e-05, train_time=5.010
+[gpub052:0/16] 2024-01-22 07:49:21,343 (trainer:737) INFO: 3epoch:train:1001-1100batch: iter_time=0.002, forward_time=0.438, loss_ctc=148.678, loss_att=177.528, acc=0.237, loss=168.873, backward_time=0.430, grad_norm=64.546, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.103, optim0_lr0=6.577e-05, train_time=5.103
+[gpub052:0/16] 2024-01-22 07:58:13,809 (trainer:737) INFO: 3epoch:train:1101-1200batch: iter_time=9.354e-05, forward_time=0.471, loss_ctc=161.262, loss_att=208.027, acc=0.222, loss=193.998, backward_time=0.493, grad_norm=56.999, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.110, optim0_lr0=6.727e-05, train_time=5.325
+[gpub052:0/16] 2024-01-22 08:01:02,472 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub052:0/16] 2024-01-22 08:01:22,116 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 08:01:25,612 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f6067bc42b0>)
+[gpub052:0/16] 2024-01-22 08:01:25,612 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub052:0/16] 2024-01-22 08:01:25,753 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 08:15:18,540 (trainer:737) INFO: 3epoch:train:1201-1300batch: iter_time=7.221, forward_time=0.585, loss_ctc=145.304, loss_att=176.806, acc=0.244, loss=167.355, backward_time=0.568, grad_norm=68.496, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.127, optim0_lr0=6.877e-05, train_time=10.247
+[gpub052:0/16] 2024-01-22 08:18:40,992 (trainer:737) INFO: 3epoch:train:1301-1400batch: iter_time=8.559e-05, forward_time=0.402, loss_ctc=169.056, loss_att=209.661, acc=0.218, loss=197.480, backward_time=0.465, grad_norm=71.123, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.100, optim0_lr0=7.027e-05, train_time=2.019
+[gpub052:0/16] 2024-01-22 08:23:15,208 (trainer:737) INFO: 3epoch:train:1401-1500batch: iter_time=0.001, forward_time=0.420, loss_ctc=153.125, loss_att=207.059, acc=0.232, loss=190.879, backward_time=0.514, grad_norm=56.455, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.112, optim0_lr0=7.177e-05, train_time=2.747
+[gpub052:0/16] 2024-01-22 08:25:11,247 (trainer:737) INFO: 3epoch:train:1501-1600batch: iter_time=8.523e-05, forward_time=0.303, loss_ctc=139.006, loss_att=175.714, acc=0.227, loss=164.702, backward_time=0.409, grad_norm=52.574, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.094, optim0_lr0=7.327e-05, train_time=1.159
+[gpub052:0/16] 2024-01-22 08:31:13,111 (trainer:737) INFO: 3epoch:train:1601-1700batch: iter_time=8.218e-05, forward_time=0.488, loss_ctc=173.144, loss_att=230.888, acc=0.226, loss=213.565, backward_time=0.446, grad_norm=71.737, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.119, optim0_lr0=7.477e-05, train_time=3.618
+[gpub052:0/16] 2024-01-22 08:36:01,687 (trainer:737) INFO: 3epoch:train:1701-1800batch: iter_time=8.366e-05, forward_time=0.612, loss_ctc=145.179, loss_att=186.708, acc=0.227, loss=174.250, backward_time=0.577, grad_norm=59.156, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.112, optim0_lr0=7.627e-05, train_time=2.886
+[gpub052:0/16] 2024-01-22 08:39:21,293 (trainer:737) INFO: 3epoch:train:1801-1900batch: iter_time=8.639e-05, forward_time=0.308, loss_ctc=142.561, loss_att=184.296, acc=0.234, loss=171.775, backward_time=0.399, grad_norm=55.681, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.094, optim0_lr0=7.777e-05, train_time=1.996
+[gpub052:0/16] 2024-01-22 08:42:24,519 (trainer:737) INFO: 3epoch:train:1901-2000batch: iter_time=8.347e-05, forward_time=0.483, loss_ctc=160.533, loss_att=203.524, acc=0.223, loss=190.627, backward_time=0.480, grad_norm=61.743, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.112, optim0_lr0=7.927e-05, train_time=1.832
+[gpub052:0/16] 2024-01-22 08:46:02,235 (trainer:737) INFO: 3epoch:train:2001-2100batch: iter_time=2.982e-04, forward_time=0.339, loss_ctc=177.157, loss_att=227.191, acc=0.222, loss=212.181, backward_time=0.420, grad_norm=68.869, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.094, optim0_lr0=8.077e-05, train_time=2.177
+[gpub052:0/16] 2024-01-22 08:49:45,585 (trainer:737) INFO: 3epoch:train:2101-2200batch: iter_time=8.126e-05, forward_time=0.311, loss_ctc=131.064, loss_att=188.862, acc=0.230, loss=171.522, backward_time=0.406, grad_norm=47.072, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.095, optim0_lr0=8.227e-05, train_time=2.234
+[gpub052:0/16] 2024-01-22 08:53:25,192 (trainer:737) INFO: 3epoch:train:2201-2300batch: iter_time=5.779e-04, forward_time=0.710, loss_ctc=167.004, loss_att=210.465, acc=0.231, loss=197.427, backward_time=0.517, grad_norm=69.648, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.113, optim0_lr0=8.377e-05, train_time=2.195
+[gpub052:0/16] 2024-01-22 08:55:56,045 (trainer:737) INFO: 3epoch:train:2301-2400batch: iter_time=8.277e-05, forward_time=0.291, loss_ctc=134.715, loss_att=176.637, acc=0.231, loss=164.060, backward_time=0.401, grad_norm=50.295, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.093, optim0_lr0=8.527e-05, train_time=1.508
+[gpub052:0/16] 2024-01-22 08:58:39,210 (trainer:737) INFO: 3epoch:train:2401-2500batch: iter_time=8.283e-05, forward_time=0.352, loss_ctc=157.680, loss_att=204.418, acc=0.232, loss=190.397, backward_time=0.483, grad_norm=62.509, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.097, optim0_lr0=8.677e-05, train_time=1.632
+[gpub052:0/16] 2024-01-22 08:58:59,653 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub052:0/16] 2024-01-22 08:59:27,378 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 08:59:35,647 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f617ee8ab60>)
+[gpub052:0/16] 2024-01-22 08:59:35,647 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub052:0/16] 2024-01-22 08:59:35,650 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 09:15:47,093 (trainer:737) INFO: 3epoch:train:2501-2600batch: iter_time=7.486, forward_time=0.565, loss_ctc=154.247, loss_att=194.634, acc=0.229, loss=182.518, backward_time=0.445, grad_norm=66.285, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.100, optim0_lr0=8.827e-05, train_time=10.279
+[gpub052:0/16] 2024-01-22 09:18:59,032 (trainer:737) INFO: 3epoch:train:2601-2700batch: iter_time=8.319e-05, forward_time=0.302, loss_ctc=140.913, loss_att=195.390, acc=0.233, loss=179.047, backward_time=0.405, grad_norm=58.382, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.093, optim0_lr0=8.977e-05, train_time=1.918
+[gpub052:0/16] 2024-01-22 09:22:19,384 (trainer:737) INFO: 3epoch:train:2701-2800batch: iter_time=8.127e-05, forward_time=0.346, loss_ctc=144.933, loss_att=196.660, acc=0.228, loss=181.142, backward_time=0.428, grad_norm=59.116, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.106, optim0_lr0=9.127e-05, train_time=1.995
+[gpub052:0/16] 2024-01-22 09:26:47,946 (trainer:737) INFO: 3epoch:train:2801-2900batch: iter_time=0.008, forward_time=0.682, loss_ctc=147.272, loss_att=200.281, acc=0.230, loss=184.378, backward_time=0.567, grad_norm=52.701, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.130, optim0_lr0=9.277e-05, train_time=2.695
+[gpub052:0/16] 2024-01-22 09:29:44,987 (trainer:737) INFO: 3epoch:train:2901-3000batch: iter_time=8.146e-05, forward_time=0.290, loss_ctc=153.016, loss_att=199.874, acc=0.229, loss=185.816, backward_time=0.400, grad_norm=66.976, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.093, optim0_lr0=9.427e-05, train_time=1.770
+[gpub052:0/16] 2024-01-22 09:32:18,202 (trainer:737) INFO: 3epoch:train:3001-3100batch: iter_time=7.928e-05, forward_time=0.306, loss_ctc=139.379, loss_att=189.704, acc=0.229, loss=174.607, backward_time=0.411, grad_norm=54.976, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.094, optim0_lr0=9.577e-05, train_time=1.531
+[gpub052:0/16] 2024-01-22 09:35:34,343 (trainer:737) INFO: 3epoch:train:3101-3200batch: iter_time=8.374e-05, forward_time=0.498, loss_ctc=138.664, loss_att=191.404, acc=0.235, loss=175.582, backward_time=0.555, grad_norm=47.746, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.113, optim0_lr0=9.727e-05, train_time=1.961
+[gpub052:0/16] 2024-01-22 09:38:01,704 (trainer:737) INFO: 3epoch:train:3201-3300batch: iter_time=8.584e-05, forward_time=0.334, loss_ctc=169.685, loss_att=208.269, acc=0.224, loss=196.694, backward_time=0.420, grad_norm=71.885, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.105, optim0_lr0=9.877e-05, train_time=1.475
+[gpub052:0/16] 2024-01-22 09:41:07,046 (trainer:737) INFO: 3epoch:train:3301-3400batch: iter_time=8.240e-05, forward_time=0.349, loss_ctc=140.957, loss_att=207.466, acc=0.226, loss=187.513, backward_time=0.406, grad_norm=50.400, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.096, optim0_lr0=1.003e-04, train_time=1.853
+[gpub052:0/16] 2024-01-22 09:43:56,596 (trainer:737) INFO: 3epoch:train:3401-3500batch: iter_time=8.311e-05, forward_time=0.394, loss_ctc=140.608, loss_att=205.674, acc=0.230, loss=186.155, backward_time=0.478, grad_norm=55.737, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.101, optim0_lr0=1.018e-04, train_time=1.695
+[gpub052:0/16] 2024-01-22 09:46:19,935 (trainer:737) INFO: 3epoch:train:3501-3600batch: iter_time=8.199e-05, forward_time=0.420, loss_ctc=137.426, loss_att=175.183, acc=0.239, loss=163.856, backward_time=0.431, grad_norm=60.795, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.106, optim0_lr0=1.033e-04, train_time=1.434
+[gpub052:0/16] 2024-01-22 09:48:44,130 (trainer:737) INFO: 3epoch:train:3601-3700batch: iter_time=7.406e-04, forward_time=0.303, loss_ctc=149.556, loss_att=204.519, acc=0.227, loss=188.030, backward_time=0.409, grad_norm=54.432, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.094, optim0_lr0=1.048e-04, train_time=1.442
+[gpub052:0/16] 2024-01-22 09:50:37,660 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub052:0/16] 2024-01-22 09:50:57,361 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 09:51:00,886 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5172266aa0>)
+[gpub052:0/16] 2024-01-22 09:51:00,886 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub052:0/16] 2024-01-22 09:51:00,940 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 10:05:09,069 (trainer:737) INFO: 3epoch:train:3701-3800batch: iter_time=7.668, forward_time=0.599, loss_ctc=132.951, loss_att=171.333, acc=0.249, loss=159.818, backward_time=0.458, grad_norm=58.118, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.098, optim0_lr0=1.063e-04, train_time=9.849
+[gpub052:0/16] 2024-01-22 10:08:26,145 (trainer:737) INFO: 3epoch:train:3801-3900batch: iter_time=8.715e-05, forward_time=0.378, loss_ctc=156.236, loss_att=201.892, acc=0.223, loss=188.195, backward_time=0.424, grad_norm=65.020, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.097, optim0_lr0=1.078e-04, train_time=1.971
+[gpub052:0/16] 2024-01-22 10:10:44,942 (trainer:737) INFO: 3epoch:train:3901-4000batch: iter_time=7.922e-05, forward_time=0.311, loss_ctc=140.426, loss_att=200.512, acc=0.235, loss=182.486, backward_time=0.407, grad_norm=56.636, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.093, optim0_lr0=1.093e-04, train_time=1.388
+[gpub052:0/16] 2024-01-22 10:15:28,493 (trainer:737) INFO: 3epoch:train:4001-4100batch: iter_time=9.056e-05, forward_time=0.580, loss_ctc=128.725, loss_att=168.593, acc=0.232, loss=156.632, backward_time=0.569, grad_norm=54.399, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.118, optim0_lr0=1.108e-04, train_time=2.835
+[gpub052:0/16] 2024-01-22 10:17:57,560 (trainer:737) INFO: 3epoch:train:4101-4200batch: iter_time=8.383e-05, forward_time=0.295, loss_ctc=157.820, loss_att=221.617, acc=0.228, loss=202.478, backward_time=0.413, grad_norm=63.832, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.095, optim0_lr0=1.123e-04, train_time=1.491
+[gpub052:0/16] 2024-01-22 10:22:16,660 (trainer:737) INFO: 3epoch:train:4201-4300batch: iter_time=8.587e-05, forward_time=0.552, loss_ctc=133.745, loss_att=180.381, acc=0.235, loss=166.390, backward_time=0.536, grad_norm=55.128, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.117, optim0_lr0=1.138e-04, train_time=2.591
+[gpub052:0/16] 2024-01-22 10:24:37,525 (trainer:737) INFO: 3epoch:train:4301-4400batch: iter_time=3.150e-04, forward_time=0.367, loss_ctc=131.485, loss_att=178.843, acc=0.238, loss=164.636, backward_time=0.452, grad_norm=51.009, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.106, optim0_lr0=1.153e-04, train_time=1.409
+[gpub052:0/16] 2024-01-22 10:28:14,650 (trainer:737) INFO: 3epoch:train:4401-4500batch: iter_time=8.056e-05, forward_time=0.311, loss_ctc=147.533, loss_att=197.604, acc=0.226, loss=182.582, backward_time=0.404, grad_norm=58.995, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.094, optim0_lr0=1.168e-04, train_time=2.171
+[gpub052:0/16] 2024-01-22 10:31:29,485 (trainer:737) INFO: 3epoch:train:4501-4600batch: iter_time=8.017e-05, forward_time=0.544, loss_ctc=162.988, loss_att=220.192, acc=0.228, loss=203.031, backward_time=0.483, grad_norm=66.509, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.111, optim0_lr0=1.183e-04, train_time=1.947
+[gpub052:0/16] 2024-01-22 10:34:33,158 (trainer:737) INFO: 3epoch:train:4601-4700batch: iter_time=8.218e-05, forward_time=0.352, loss_ctc=119.832, loss_att=183.299, acc=0.238, loss=164.259, backward_time=0.412, grad_norm=45.915, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.095, optim0_lr0=1.198e-04, train_time=1.837
+[gpub052:0/16] 2024-01-22 10:38:17,264 (trainer:737) INFO: 3epoch:train:4701-4800batch: iter_time=8.292e-05, forward_time=0.312, loss_ctc=150.181, loss_att=203.621, acc=0.239, loss=187.589, backward_time=0.422, grad_norm=67.959, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.095, optim0_lr0=1.213e-04, train_time=2.241
+[gpub052:0/16] 2024-01-22 10:41:17,007 (trainer:737) INFO: 3epoch:train:4801-4900batch: iter_time=0.001, forward_time=0.579, loss_ctc=122.872, loss_att=172.391, acc=0.235, loss=157.535, backward_time=0.490, grad_norm=46.840, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.110, optim0_lr0=1.228e-04, train_time=1.796
+[gpub052:0/16] 2024-01-22 10:45:07,890 (trainer:737) INFO: 3epoch:train:4901-5000batch: iter_time=8.311e-05, forward_time=0.291, loss_ctc=143.348, loss_att=197.518, acc=0.239, loss=181.267, backward_time=0.401, grad_norm=57.751, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.093, optim0_lr0=1.243e-04, train_time=2.309
+[gpub052:0/16] 2024-01-22 10:45:28,460 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub052:0/16] 2024-01-22 10:45:47,895 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 10:45:51,473 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f605614b7f0>)
+[gpub052:0/16] 2024-01-22 10:45:51,473 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub052:0/16] 2024-01-22 10:45:51,627 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 11:00:39,624 (trainer:737) INFO: 3epoch:train:5001-5100batch: iter_time=7.483, forward_time=0.520, loss_ctc=142.672, loss_att=186.315, acc=0.235, loss=173.222, backward_time=0.467, grad_norm=68.777, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.100, optim0_lr0=1.258e-04, train_time=9.318
+[gpub052:0/16] 2024-01-22 11:03:23,942 (trainer:737) INFO: 3epoch:train:5101-5200batch: iter_time=8.039e-05, forward_time=0.402, loss_ctc=129.485, loss_att=185.561, acc=0.241, loss=168.738, backward_time=0.468, grad_norm=53.740, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.098, optim0_lr0=1.273e-04, train_time=1.643
+[gpub052:0/16] 2024-01-22 11:07:01,565 (trainer:737) INFO: 3epoch:train:5201-5300batch: iter_time=9.143e-05, forward_time=0.294, loss_ctc=133.725, loss_att=189.988, acc=0.233, loss=173.109, backward_time=0.398, grad_norm=56.950, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.093, optim0_lr0=1.288e-04, train_time=2.176
+[gpub052:0/16] 2024-01-22 11:09:42,971 (trainer:737) INFO: 3epoch:train:5301-5400batch: iter_time=9.245e-05, forward_time=0.492, loss_ctc=134.919, loss_att=189.130, acc=0.236, loss=172.867, backward_time=0.493, grad_norm=51.759, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.115, optim0_lr0=1.303e-04, train_time=1.614
+[gpub052:0/16] 2024-01-22 11:12:43,536 (trainer:737) INFO: 3epoch:train:5401-5500batch: iter_time=8.873e-05, forward_time=0.347, loss_ctc=141.931, loss_att=189.810, acc=0.240, loss=175.446, backward_time=0.444, grad_norm=61.576, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.101, optim0_lr0=1.318e-04, train_time=1.806
+[gpub052:0/16] 2024-01-22 11:15:50,113 (trainer:737) INFO: 3epoch:train:5501-5600batch: iter_time=8.348e-05, forward_time=0.350, loss_ctc=128.262, loss_att=181.028, acc=0.237, loss=165.198, backward_time=0.417, grad_norm=53.868, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.096, optim0_lr0=1.333e-04, train_time=1.863
+[gpub052:0/16] 2024-01-22 11:18:29,457 (trainer:737) INFO: 3epoch:train:5601-5700batch: iter_time=8.615e-05, forward_time=0.308, loss_ctc=127.859, loss_att=186.418, acc=0.241, loss=168.850, backward_time=0.405, grad_norm=46.801, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.094, optim0_lr0=1.348e-04, train_time=1.596
+[gpub052:0/16] 2024-01-22 11:21:43,079 (trainer:737) INFO: 3epoch:train:5701-5800batch: iter_time=8.943e-05, forward_time=0.573, loss_ctc=155.028, loss_att=201.503, acc=0.232, loss=187.561, backward_time=0.500, grad_norm=64.143, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.111, optim0_lr0=1.363e-04, train_time=1.936
+[gpub052:0/16] 2024-01-22 11:24:20,335 (trainer:737) INFO: 3epoch:train:5801-5900batch: iter_time=0.002, forward_time=0.321, loss_ctc=129.999, loss_att=200.340, acc=0.237, loss=179.238, backward_time=0.425, grad_norm=55.681, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.096, optim0_lr0=1.378e-04, train_time=1.573
+[gpub052:0/16] 2024-01-22 11:27:11,640 (trainer:737) INFO: 3epoch:train:5901-6000batch: iter_time=8.782e-05, forward_time=0.305, loss_ctc=130.360, loss_att=197.616, acc=0.240, loss=177.439, backward_time=0.404, grad_norm=57.867, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.094, optim0_lr0=1.393e-04, train_time=1.709
+[gpub052:0/16] 2024-01-22 11:30:07,891 (trainer:737) INFO: 3epoch:train:6001-6100batch: iter_time=8.813e-05, forward_time=0.477, loss_ctc=128.133, loss_att=171.484, acc=0.250, loss=158.479, backward_time=0.496, grad_norm=52.130, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.102, optim0_lr0=1.408e-04, train_time=1.766
+[gpub052:0/16] 2024-01-22 11:32:50,445 (trainer:737) INFO: 3epoch:train:6101-6200batch: iter_time=5.408e-04, forward_time=0.330, loss_ctc=140.445, loss_att=196.492, acc=0.237, loss=179.678, backward_time=0.430, grad_norm=54.500, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.096, optim0_lr0=1.423e-04, train_time=1.625
+[gpub052:0/16] 2024-01-22 11:34:05,725 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub052:0/16] 2024-01-22 11:34:25,860 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 11:34:29,505 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f51f412ff10>)
+[gpub052:0/16] 2024-01-22 11:34:29,505 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub052:0/16] 2024-01-22 11:34:29,509 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 11:45:06,268 (trainer:737) INFO: 3epoch:train:6201-6300batch: iter_time=6.035, forward_time=0.427, loss_ctc=121.718, loss_att=165.145, acc=0.256, loss=152.117, backward_time=0.430, grad_norm=51.980, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.099, optim0_lr0=1.438e-04, train_time=7.358
+[gpub052:0/16] 2024-01-22 11:47:28,870 (trainer:737) INFO: 3epoch:train:6301-6400batch: iter_time=8.115e-05, forward_time=0.340, loss_ctc=146.462, loss_att=195.552, acc=0.233, loss=180.825, backward_time=0.416, grad_norm=70.060, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.096, optim0_lr0=1.453e-04, train_time=1.426
+[gpub052:0/16] 2024-01-22 11:50:35,541 (trainer:737) INFO: 3epoch:train:6401-6500batch: iter_time=8.088e-05, forward_time=0.487, loss_ctc=131.667, loss_att=194.524, acc=0.248, loss=175.667, backward_time=0.482, grad_norm=53.765, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.112, optim0_lr0=1.468e-04, train_time=1.867
+[gpub052:0/16] 2024-01-22 11:52:33,074 (trainer:737) INFO: 3epoch:train:6501-6600batch: iter_time=8.187e-05, forward_time=0.340, loss_ctc=119.028, loss_att=163.532, acc=0.244, loss=150.181, backward_time=0.417, grad_norm=53.392, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.094, optim0_lr0=1.483e-04, train_time=1.175
+[gpub052:0/16] 2024-01-22 11:55:55,662 (trainer:737) INFO: 3epoch:train:6601-6700batch: iter_time=8.567e-05, forward_time=0.396, loss_ctc=144.101, loss_att=212.186, acc=0.244, loss=191.760, backward_time=0.563, grad_norm=58.817, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.109, optim0_lr0=1.498e-04, train_time=2.026
+[gpub052:0/16] 2024-01-22 11:58:33,487 (trainer:737) INFO: 3epoch:train:6701-6800batch: iter_time=9.669e-05, forward_time=0.328, loss_ctc=123.910, loss_att=176.350, acc=0.248, loss=160.618, backward_time=0.415, grad_norm=51.722, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.095, optim0_lr0=1.513e-04, train_time=1.578
+[gpub052:0/16] 2024-01-22 12:01:09,804 (trainer:737) INFO: 3epoch:train:6801-6900batch: iter_time=8.000e-05, forward_time=0.465, loss_ctc=119.628, loss_att=173.371, acc=0.253, loss=157.248, backward_time=0.456, grad_norm=48.179, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.103, optim0_lr0=1.528e-04, train_time=1.562
+[gpub052:0/16] 2024-01-22 12:03:33,982 (trainer:737) INFO: 3epoch:train:6901-7000batch: iter_time=3.146e-04, forward_time=0.331, loss_ctc=134.801, loss_att=190.981, acc=0.241, loss=174.127, backward_time=0.413, grad_norm=52.236, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.100, optim0_lr0=1.543e-04, train_time=1.442
+[gpub052:0/16] 2024-01-22 12:06:25,646 (trainer:737) INFO: 3epoch:train:7001-7100batch: iter_time=8.525e-05, forward_time=0.401, loss_ctc=148.449, loss_att=211.927, acc=0.242, loss=192.883, backward_time=0.486, grad_norm=62.208, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.106, optim0_lr0=1.558e-04, train_time=1.717
+[gpub052:0/16] 2024-01-22 12:08:57,819 (trainer:737) INFO: 3epoch:train:7101-7200batch: iter_time=1.019e-04, forward_time=0.327, loss_ctc=110.051, loss_att=175.195, acc=0.259, loss=155.652, backward_time=0.417, grad_norm=42.062, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.096, optim0_lr0=1.573e-04, train_time=1.520
+[gpub052:0/16] 2024-01-22 12:11:33,750 (trainer:737) INFO: 3epoch:train:7201-7300batch: iter_time=8.209e-05, forward_time=0.442, loss_ctc=141.005, loss_att=196.020, acc=0.259, loss=179.516, backward_time=0.473, grad_norm=63.356, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.110, optim0_lr0=1.588e-04, train_time=1.560
+[gpub052:0/16] 2024-01-22 12:14:09,665 (trainer:737) INFO: 3epoch:train:7301-7400batch: iter_time=0.001, forward_time=0.345, loss_ctc=113.672, loss_att=164.812, acc=0.254, loss=149.470, backward_time=0.418, grad_norm=44.589, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.095, optim0_lr0=1.603e-04, train_time=1.558
+[gpub052:0/16] 2024-01-22 12:17:12,759 (trainer:737) INFO: 3epoch:train:7401-7500batch: iter_time=7.947e-05, forward_time=0.396, loss_ctc=135.145, loss_att=188.644, acc=0.262, loss=172.595, backward_time=0.522, grad_norm=52.431, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.105, optim0_lr0=1.618e-04, train_time=1.831
+[gpub052:0/16] 2024-01-22 12:17:32,789 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub052:0/16] 2024-01-22 12:17:52,499 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 12:17:56,317 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f51565379a0>)
+[gpub052:0/16] 2024-01-22 12:17:56,317 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub052:0/16] 2024-01-22 12:17:56,320 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 12:30:18,368 (trainer:737) INFO: 3epoch:train:7501-7600batch: iter_time=5.982, forward_time=0.488, loss_ctc=130.015, loss_att=185.913, acc=0.254, loss=169.144, backward_time=0.517, grad_norm=58.030, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.115, optim0_lr0=1.633e-04, train_time=7.857
+[gpub052:0/16] 2024-01-22 12:32:43,596 (trainer:737) INFO: 3epoch:train:7601-7700batch: iter_time=8.201e-05, forward_time=0.346, loss_ctc=119.657, loss_att=184.073, acc=0.262, loss=164.749, backward_time=0.414, grad_norm=50.883, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.095, optim0_lr0=1.648e-04, train_time=1.452
+[gpub052:0/16] 2024-01-22 12:36:30,952 (trainer:737) INFO: 3epoch:train:7701-7800batch: iter_time=8.248e-05, forward_time=0.521, loss_ctc=123.986, loss_att=185.099, acc=0.259, loss=166.765, backward_time=0.487, grad_norm=54.926, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.102, optim0_lr0=1.663e-04, train_time=2.274
+[gpub052:0/16] 2024-01-22 12:38:59,880 (trainer:737) INFO: 3epoch:train:7801-7900batch: iter_time=0.001, forward_time=0.345, loss_ctc=123.573, loss_att=187.208, acc=0.261, loss=168.118, backward_time=0.418, grad_norm=48.652, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.096, optim0_lr0=1.678e-04, train_time=1.488
+[gpub052:0/16] 2024-01-22 12:42:01,731 (trainer:737) INFO: 3epoch:train:7901-8000batch: iter_time=8.428e-05, forward_time=0.443, loss_ctc=131.214, loss_att=184.994, acc=0.263, loss=168.860, backward_time=0.559, grad_norm=60.744, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.111, optim0_lr0=1.693e-04, train_time=1.820
+[gpub052:0/16] 2024-01-22 12:45:26,641 (trainer:737) INFO: 3epoch:train:8001-8100batch: iter_time=3.782e-04, forward_time=0.373, loss_ctc=117.458, loss_att=175.222, acc=0.262, loss=157.893, backward_time=0.412, grad_norm=44.600, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.093, optim0_lr0=1.708e-04, train_time=2.046
+[gpub052:0/16] 2024-01-22 12:48:17,438 (trainer:737) INFO: 3epoch:train:8101-8200batch: iter_time=8.086e-05, forward_time=0.494, loss_ctc=116.429, loss_att=176.431, acc=0.275, loss=158.431, backward_time=0.475, grad_norm=44.337, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.111, optim0_lr0=1.723e-04, train_time=1.709
+[gpub052:0/16] 2024-01-22 12:51:08,995 (trainer:737) INFO: 3epoch:train:8201-8300batch: iter_time=0.001, forward_time=0.335, loss_ctc=142.590, loss_att=191.926, acc=0.257, loss=177.125, backward_time=0.408, grad_norm=60.014, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.112, optim0_lr0=1.738e-04, train_time=1.715
+[gpub052:0/16] 2024-01-22 12:54:47,469 (trainer:737) INFO: 3epoch:train:8301-8400batch: iter_time=9.014e-05, forward_time=0.477, loss_ctc=119.624, loss_att=187.519, acc=0.277, loss=167.151, backward_time=0.447, grad_norm=48.772, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.113, optim0_lr0=1.753e-04, train_time=2.186
+[gpub052:0/16] 2024-01-22 12:57:06,140 (trainer:737) INFO: 3epoch:train:8401-8500batch: iter_time=1.970e-04, forward_time=0.335, loss_ctc=119.449, loss_att=184.190, acc=0.284, loss=164.768, backward_time=0.410, grad_norm=55.010, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.096, optim0_lr0=1.768e-04, train_time=1.384
+[gpub052:0/16] 2024-01-22 13:00:05,890 (trainer:737) INFO: 3epoch:train:8501-8600batch: iter_time=8.461e-05, forward_time=0.479, loss_ctc=115.939, loss_att=157.783, acc=0.289, loss=145.230, backward_time=0.474, grad_norm=54.845, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.101, optim0_lr0=1.783e-04, train_time=1.797
+[gpub052:0/16] 2024-01-22 13:03:16,743 (trainer:737) INFO: 3epoch:train:8601-8700batch: iter_time=8.490e-04, forward_time=0.329, loss_ctc=127.795, loss_att=184.594, acc=0.277, loss=167.554, backward_time=0.408, grad_norm=54.428, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.094, optim0_lr0=1.798e-04, train_time=1.909
+[gpub052:0/16] 2024-01-22 13:05:57,370 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub052:0/16] 2024-01-22 13:06:17,191 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 13:06:21,128 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f518d32f790>)
+[gpub052:0/16] 2024-01-22 13:06:21,128 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub052:0/16] 2024-01-22 13:06:21,131 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 13:19:54,986 (trainer:737) INFO: 3epoch:train:8701-8800batch: iter_time=7.718, forward_time=0.491, loss_ctc=112.851, loss_att=156.137, acc=0.294, loss=143.151, backward_time=0.547, grad_norm=51.168, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.103, optim0_lr0=1.813e-04, train_time=9.984
+[gpub052:0/16] 2024-01-22 13:22:33,481 (trainer:737) INFO: 3epoch:train:8801-8900batch: iter_time=8.040e-05, forward_time=0.372, loss_ctc=138.122, loss_att=187.023, acc=0.271, loss=172.353, backward_time=0.416, grad_norm=66.047, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.094, optim0_lr0=1.828e-04, train_time=1.584
+[gpub052:0/16] 2024-01-22 13:26:50,187 (trainer:737) INFO: 3epoch:train:8901-9000batch: iter_time=8.113e-05, forward_time=0.470, loss_ctc=119.848, loss_att=181.943, acc=0.292, loss=163.315, backward_time=0.559, grad_norm=54.108, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.110, optim0_lr0=1.843e-04, train_time=2.568
+[gpub052:0/16] 2024-01-22 13:29:22,591 (trainer:737) INFO: 3epoch:train:9001-9100batch: iter_time=9.945e-05, forward_time=0.339, loss_ctc=110.222, loss_att=153.105, acc=0.285, loss=140.240, backward_time=0.411, grad_norm=52.979, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.098, optim0_lr0=1.858e-04, train_time=1.524
+[gpub052:0/16] 2024-01-22 13:33:39,112 (trainer:737) INFO: 3epoch:train:9101-9200batch: iter_time=9.256e-05, forward_time=0.958, loss_ctc=135.284, loss_att=199.662, acc=0.291, loss=180.349, backward_time=0.532, grad_norm=56.587, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.148, optim0_lr0=1.873e-04, train_time=2.561
+[gpub052:0/16] 2024-01-22 13:37:26,535 (trainer:737) INFO: 3epoch:train:9201-9300batch: iter_time=1.209e-04, forward_time=0.350, loss_ctc=113.539, loss_att=161.343, acc=0.292, loss=147.002, backward_time=0.409, grad_norm=53.157, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.094, optim0_lr0=1.888e-04, train_time=2.278
+[gpub052:0/16] 2024-01-22 13:40:41,653 (trainer:737) INFO: 3epoch:train:9301-9400batch: iter_time=9.915e-05, forward_time=0.530, loss_ctc=110.543, loss_att=158.745, acc=0.297, loss=144.285, backward_time=0.600, grad_norm=49.916, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.116, optim0_lr0=1.903e-04, train_time=1.950
+[gpub052:0/16] 2024-01-22 13:43:28,439 (trainer:737) INFO: 3epoch:train:9401-9500batch: iter_time=0.001, forward_time=0.339, loss_ctc=126.134, loss_att=177.453, acc=0.286, loss=162.057, backward_time=0.432, grad_norm=52.642, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.099, optim0_lr0=1.918e-04, train_time=1.667
+[gpub052:0/16] 2024-01-22 13:46:57,010 (trainer:737) INFO: 3epoch:train:9501-9600batch: iter_time=9.189e-05, forward_time=0.311, loss_ctc=139.951, loss_att=193.278, acc=0.295, loss=177.280, backward_time=0.408, grad_norm=65.420, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.097, optim0_lr0=1.933e-04, train_time=2.087
+[gpub052:0/16] 2024-01-22 13:50:31,476 (trainer:737) INFO: 3epoch:train:9601-9700batch: iter_time=9.624e-05, forward_time=0.632, loss_ctc=103.039, loss_att=154.670, acc=0.317, loss=139.180, backward_time=0.541, grad_norm=46.259, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.107, optim0_lr0=1.948e-04, train_time=2.144
+[gpub052:0/16] 2024-01-22 13:53:10,446 (trainer:737) INFO: 3epoch:train:9701-9800batch: iter_time=5.097e-04, forward_time=0.390, loss_ctc=129.038, loss_att=174.298, acc=0.323, loss=160.720, backward_time=0.441, grad_norm=63.136, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.096, optim0_lr0=1.963e-04, train_time=1.588
+[gpub052:0/16] 2024-01-22 13:56:36,443 (trainer:737) INFO: 3epoch:train:9801-9900batch: iter_time=8.932e-05, forward_time=0.330, loss_ctc=106.169, loss_att=148.551, acc=0.304, loss=135.836, backward_time=0.425, grad_norm=49.526, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.096, optim0_lr0=1.978e-04, train_time=2.062
+[gpub052:0/16] 2024-01-22 14:00:56,666 (trainer:737) INFO: 3epoch:train:9901-10000batch: iter_time=8.871e-05, forward_time=0.506, loss_ctc=124.331, loss_att=167.675, acc=0.321, loss=154.672, backward_time=0.599, grad_norm=56.682, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.124, optim0_lr0=1.993e-04, train_time=2.602
+[gpub052:0/16] 2024-01-22 14:01:16,757 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub052:0/16] 2024-01-22 14:01:36,976 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 14:01:40,701 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f51af601990>)
+[gpub052:0/16] 2024-01-22 14:01:40,701 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub052:0/16] 2024-01-22 14:01:40,704 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 14:19:31,141 (trainer:737) INFO: 3epoch:train:10001-10100batch: iter_time=8.721, forward_time=0.604, loss_ctc=125.234, loss_att=161.338, acc=0.300, loss=150.507, backward_time=0.463, grad_norm=68.037, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.102, optim0_lr0=2.008e-04, train_time=11.145
+[gpub052:0/16] 2024-01-22 14:23:27,469 (trainer:737) INFO: 3epoch:train:10101-10200batch: iter_time=9.558e-05, forward_time=0.552, loss_ctc=112.280, loss_att=159.281, acc=0.315, loss=145.180, backward_time=0.564, grad_norm=55.719, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.109, optim0_lr0=2.023e-04, train_time=2.363
+[gpub052:0/16] 2024-01-22 14:26:48,098 (trainer:737) INFO: 3epoch:train:10201-10300batch: iter_time=9.240e-05, forward_time=0.292, loss_ctc=114.348, loss_att=159.684, acc=0.316, loss=146.083, backward_time=0.400, grad_norm=57.011, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.093, optim0_lr0=2.038e-04, train_time=2.007
+[gpub052:0/16] 2024-01-22 14:30:17,741 (trainer:737) INFO: 3epoch:train:10301-10400batch: iter_time=8.996e-05, forward_time=0.689, loss_ctc=116.775, loss_att=159.996, acc=0.315, loss=147.030, backward_time=0.513, grad_norm=51.915, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.124, optim0_lr0=2.053e-04, train_time=2.096
+[gpub052:0/16] 2024-01-22 14:33:28,641 (trainer:737) INFO: 3epoch:train:10401-10500batch: iter_time=9.311e-05, forward_time=0.299, loss_ctc=122.506, loss_att=159.783, acc=0.322, loss=148.600, backward_time=0.401, grad_norm=61.199, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.094, optim0_lr0=2.068e-04, train_time=1.908
+[gpub052:0/16] 2024-01-22 14:36:53,875 (trainer:737) INFO: 3epoch:train:10501-10600batch: iter_time=9.394e-05, forward_time=0.606, loss_ctc=110.656, loss_att=151.738, acc=0.315, loss=139.413, backward_time=0.514, grad_norm=50.077, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.128, optim0_lr0=2.083e-04, train_time=2.053
+[gpub052:0/16] 2024-01-22 14:39:13,810 (trainer:737) INFO: 3epoch:train:10601-10700batch: iter_time=8.922e-05, forward_time=0.295, loss_ctc=109.807, loss_att=153.798, acc=0.334, loss=140.600, backward_time=0.404, grad_norm=52.079, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.093, optim0_lr0=2.098e-04, train_time=1.399
+[gpub052:0/16] 2024-01-22 14:43:27,117 (trainer:737) INFO: 3epoch:train:10701-10800batch: iter_time=5.819e-04, forward_time=0.296, loss_ctc=135.371, loss_att=171.852, acc=0.303, loss=160.908, backward_time=0.401, grad_norm=64.471, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.097, optim0_lr0=2.113e-04, train_time=2.532
+[gpub052:0/16] 2024-01-22 14:46:48,368 (trainer:737) INFO: 3epoch:train:10801-10900batch: iter_time=8.713e-05, forward_time=0.613, loss_ctc=113.254, loss_att=156.845, acc=0.351, loss=143.768, backward_time=0.564, grad_norm=55.352, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.119, optim0_lr0=2.128e-04, train_time=2.013
+[gpub052:0/16] 2024-01-22 14:49:52,856 (trainer:737) INFO: 3epoch:train:10901-11000batch: iter_time=8.710e-05, forward_time=0.345, loss_ctc=110.838, loss_att=152.568, acc=0.356, loss=140.049, backward_time=0.404, grad_norm=56.197, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.093, optim0_lr0=2.143e-04, train_time=1.845
+[gpub052:0/16] 2024-01-22 14:52:55,254 (trainer:737) INFO: 3epoch:train:11001-11100batch: iter_time=8.819e-05, forward_time=0.291, loss_ctc=108.345, loss_att=138.428, acc=0.347, loss=129.403, backward_time=0.399, grad_norm=56.292, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.093, optim0_lr0=2.158e-04, train_time=1.824
+[gpub052:0/16] 2024-01-22 14:56:21,143 (trainer:737) INFO: 3epoch:train:11101-11200batch: iter_time=0.007, forward_time=0.667, loss_ctc=120.528, loss_att=158.275, acc=0.338, loss=146.951, backward_time=0.527, grad_norm=53.128, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.124, optim0_lr0=2.173e-04, train_time=2.058
+[gpub052:0/16] 2024-01-22 14:58:01,940 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub052:0/16] 2024-01-22 14:58:21,856 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 14:58:25,591 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f518e151030>)
+[gpub052:0/16] 2024-01-22 14:58:25,592 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub052:0/16] 2024-01-22 14:58:25,595 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 15:14:20,721 (trainer:737) INFO: 3epoch:train:11201-11300batch: iter_time=8.698, forward_time=0.291, loss_ctc=105.264, loss_att=138.630, acc=0.352, loss=128.620, backward_time=0.404, grad_norm=50.522, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.094, optim0_lr0=2.188e-04, train_time=10.796
+[gpub052:0/16] 2024-01-22 15:20:25,152 (trainer:737) INFO: 3epoch:train:11301-11400batch: iter_time=8.440e-05, forward_time=0.532, loss_ctc=126.919, loss_att=164.101, acc=0.332, loss=152.946, backward_time=0.625, grad_norm=67.157, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.110, optim0_lr0=2.203e-04, train_time=3.644
+[gpub052:0/16] 2024-01-22 15:23:56,572 (trainer:737) INFO: 3epoch:train:11401-11500batch: iter_time=8.363e-05, forward_time=0.361, loss_ctc=111.997, loss_att=160.913, acc=0.359, loss=146.238, backward_time=0.412, grad_norm=54.079, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.111, optim0_lr0=2.218e-04, train_time=2.114
+[gpub052:0/16] 2024-01-22 15:28:02,674 (trainer:737) INFO: 3epoch:train:11501-11600batch: iter_time=8.595e-05, forward_time=0.417, loss_ctc=102.355, loss_att=132.186, acc=0.346, loss=123.237, backward_time=0.674, grad_norm=57.067, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.106, optim0_lr0=2.233e-04, train_time=2.461
+[gpub052:0/16] 2024-01-22 15:32:43,923 (trainer:737) INFO: 3epoch:train:11601-11700batch: iter_time=8.474e-05, forward_time=0.506, loss_ctc=127.209, loss_att=175.732, acc=0.362, loss=161.175, backward_time=0.661, grad_norm=63.211, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.120, optim0_lr0=2.248e-04, train_time=2.812
+[gpub052:0/16] 2024-01-22 15:36:53,480 (trainer:737) INFO: 3epoch:train:11701-11800batch: iter_time=4.109e-04, forward_time=0.290, loss_ctc=107.211, loss_att=140.249, acc=0.359, loss=130.338, backward_time=0.398, grad_norm=52.720, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.094, optim0_lr0=2.263e-04, train_time=2.495
+[gpub052:0/16] 2024-01-22 15:40:57,707 (trainer:737) INFO: 3epoch:train:11801-11900batch: iter_time=8.504e-05, forward_time=0.618, loss_ctc=102.616, loss_att=138.587, acc=0.363, loss=127.796, backward_time=0.599, grad_norm=50.760, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.118, optim0_lr0=2.278e-04, train_time=2.441
+[gpub052:0/16] 2024-01-22 15:45:14,829 (trainer:737) INFO: 3epoch:train:11901-12000batch: iter_time=3.796e-04, forward_time=0.302, loss_ctc=118.706, loss_att=158.193, acc=0.345, loss=146.347, backward_time=0.406, grad_norm=55.370, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.095, optim0_lr0=2.293e-04, train_time=2.572
+[gpub052:0/16] 2024-01-22 15:50:23,449 (trainer:737) INFO: 3epoch:train:12001-12100batch: iter_time=8.601e-05, forward_time=0.598, loss_ctc=130.337, loss_att=168.136, acc=0.365, loss=156.796, backward_time=0.604, grad_norm=71.018, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.107, optim0_lr0=2.308e-04, train_time=3.086
+[gpub052:0/16] 2024-01-22 15:53:36,235 (trainer:737) INFO: 3epoch:train:12101-12200batch: iter_time=2.975e-04, forward_time=0.561, loss_ctc=95.436, loss_att=128.651, acc=0.399, loss=118.686, backward_time=0.668, grad_norm=54.919, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.120, optim0_lr0=2.323e-04, train_time=1.924
+[gpub052:0/16] 2024-01-22 15:58:08,715 (trainer:737) INFO: 3epoch:train:12201-12300batch: iter_time=4.373e-04, forward_time=0.299, loss_ctc=121.773, loss_att=149.237, acc=0.398, loss=140.998, backward_time=0.405, grad_norm=65.711, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.094, optim0_lr0=2.338e-04, train_time=2.727
+[gpub052:0/16] 2024-01-22 16:03:23,856 (trainer:737) INFO: 3epoch:train:12301-12400batch: iter_time=0.001, forward_time=0.523, loss_ctc=99.370, loss_att=129.888, acc=0.367, loss=120.732, backward_time=0.761, grad_norm=51.583, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.130, optim0_lr0=2.353e-04, train_time=3.149
+[gpub052:0/16] 2024-01-22 16:06:00,474 (trainer:737) INFO: 3epoch:train:12401-12500batch: iter_time=0.002, forward_time=0.329, loss_ctc=118.861, loss_att=145.267, acc=0.392, loss=137.345, backward_time=0.407, grad_norm=56.123, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.094, optim0_lr0=2.368e-04, train_time=1.568
+[gpub052:0/16] 2024-01-22 16:06:20,533 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub052:0/16] 2024-01-22 16:06:40,915 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 16:06:44,565 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5164977ac0>)
+[gpub052:0/16] 2024-01-22 16:06:44,565 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub052:0/16] 2024-01-22 16:06:44,568 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 16:21:47,381 (trainer:737) INFO: 3epoch:train:12501-12600batch: iter_time=7.865, forward_time=0.384, loss_ctc=116.454, loss_att=138.577, acc=0.372, loss=131.940, backward_time=0.432, grad_norm=61.446, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.099, optim0_lr0=2.383e-04, train_time=9.469
+[gpub052:0/16] 2024-01-22 16:24:30,937 (trainer:737) INFO: 3epoch:train:12601-12700batch: iter_time=8.097e-05, forward_time=0.481, loss_ctc=104.938, loss_att=138.145, acc=0.386, loss=128.183, backward_time=0.427, grad_norm=59.748, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.097, optim0_lr0=2.398e-04, train_time=1.635
+[gpub052:0/16] 2024-01-22 16:26:46,751 (trainer:737) INFO: 3epoch:train:12701-12800batch: iter_time=0.001, forward_time=0.293, loss_ctc=109.241, loss_att=138.572, acc=0.386, loss=129.773, backward_time=0.405, grad_norm=60.643, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.094, optim0_lr0=2.413e-04, train_time=1.358
+[gpub052:0/16] 2024-01-22 16:29:16,167 (trainer:737) INFO: 3epoch:train:12801-12900batch: iter_time=8.852e-05, forward_time=0.366, loss_ctc=109.005, loss_att=138.453, acc=0.382, loss=129.619, backward_time=0.452, grad_norm=53.130, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.101, optim0_lr0=2.428e-04, train_time=1.494
+[gpub052:0/16] 2024-01-22 16:31:49,813 (trainer:737) INFO: 3epoch:train:12901-13000batch: iter_time=8.975e-05, forward_time=0.327, loss_ctc=115.447, loss_att=140.594, acc=0.389, loss=133.050, backward_time=0.431, grad_norm=59.457, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.098, optim0_lr0=2.443e-04, train_time=1.536
+[gpub052:0/16] 2024-01-22 16:34:17,744 (trainer:737) INFO: 3epoch:train:13001-13100batch: iter_time=8.774e-05, forward_time=0.396, loss_ctc=103.607, loss_att=133.343, acc=0.378, loss=124.422, backward_time=0.425, grad_norm=52.857, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.100, optim0_lr0=2.458e-04, train_time=1.479
+[gpub052:0/16] 2024-01-22 16:38:09,017 (trainer:737) INFO: 3epoch:train:13101-13200batch: iter_time=8.680e-05, forward_time=0.539, loss_ctc=103.036, loss_att=131.193, acc=0.412, loss=122.746, backward_time=0.557, grad_norm=49.190, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.118, optim0_lr0=2.473e-04, train_time=2.312
+[gpub052:0/16] 2024-01-22 16:41:22,366 (trainer:737) INFO: 3epoch:train:13201-13300batch: iter_time=8.306e-05, forward_time=0.350, loss_ctc=129.776, loss_att=149.167, acc=0.375, loss=143.350, backward_time=0.414, grad_norm=64.970, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.094, optim0_lr0=2.488e-04, train_time=1.934
+[gpub052:0/16] 2024-01-22 16:44:36,881 (trainer:737) INFO: 3epoch:train:13301-13400batch: iter_time=8.489e-05, forward_time=0.598, loss_ctc=106.115, loss_att=132.149, acc=0.427, loss=124.339, backward_time=0.495, grad_norm=55.311, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.150, optim0_lr0=2.503e-04, train_time=1.943
+[gpub052:0/16] 2024-01-22 16:48:02,996 (trainer:737) INFO: 3epoch:train:13401-13500batch: iter_time=9.492e-05, forward_time=0.323, loss_ctc=104.672, loss_att=128.553, acc=0.435, loss=121.388, backward_time=0.401, grad_norm=54.788, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.095, optim0_lr0=2.518e-04, train_time=2.061
+[gpub052:0/16] 2024-01-22 16:51:25,056 (trainer:737) INFO: 3epoch:train:13501-13600batch: iter_time=9.024e-05, forward_time=0.690, loss_ctc=103.498, loss_att=119.164, acc=0.418, loss=114.464, backward_time=0.467, grad_norm=62.291, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.101, optim0_lr0=2.533e-04, train_time=2.020
+[gpub052:0/16] 2024-01-22 16:55:21,661 (trainer:737) INFO: 3epoch:train:13601-13700batch: iter_time=0.002, forward_time=0.517, loss_ctc=112.942, loss_att=138.708, acc=0.400, loss=130.978, backward_time=0.594, grad_norm=52.201, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.113, optim0_lr0=2.548e-04, train_time=2.365
+[gpub052:0/16] 2024-01-22 16:57:09,192 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+
+gpub054:3241516:3241579 [0] misc/ibvwrap.cc:230 NCCL WARN Call to ibv_get_async_event failed
+[gpub052:0/16] 2024-01-22 16:57:29,611 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 16:57:33,196 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5152c9ff70>)
+[gpub052:0/16] 2024-01-22 16:57:33,196 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub052:0/16] 2024-01-22 16:57:33,200 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 17:10:04,841 (trainer:737) INFO: 3epoch:train:13701-13800batch: iter_time=7.071, forward_time=0.378, loss_ctc=99.347, loss_att=115.622, acc=0.425, loss=110.740, backward_time=0.424, grad_norm=54.088, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.095, optim0_lr0=2.563e-04, train_time=8.833
+[gpub052:0/16] 2024-01-22 17:13:58,620 (trainer:737) INFO: 3epoch:train:13801-13900batch: iter_time=8.173e-05, forward_time=0.507, loss_ctc=119.308, loss_att=137.076, acc=0.396, loss=131.746, backward_time=0.572, grad_norm=63.856, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.116, optim0_lr0=2.578e-04, train_time=2.337
+[gpub052:0/16] 2024-01-22 17:17:30,506 (trainer:737) INFO: 3epoch:train:13901-14000batch: iter_time=8.026e-05, forward_time=0.677, loss_ctc=105.777, loss_att=135.396, acc=0.426, loss=126.510, backward_time=0.512, grad_norm=52.613, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.111, optim0_lr0=2.593e-04, train_time=2.117
+[gpub052:0/16] 2024-01-22 17:20:13,107 (trainer:737) INFO: 3epoch:train:14001-14100batch: iter_time=8.125e-05, forward_time=0.294, loss_ctc=96.728, loss_att=111.695, acc=0.403, loss=107.205, backward_time=0.410, grad_norm=53.172, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.094, optim0_lr0=2.608e-04, train_time=1.627
+[gpub052:0/16] 2024-01-22 17:24:11,314 (trainer:737) INFO: 3epoch:train:14101-14200batch: iter_time=0.001, forward_time=0.570, loss_ctc=120.348, loss_att=149.901, acc=0.421, loss=141.035, backward_time=0.521, grad_norm=63.433, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.125, optim0_lr0=2.623e-04, train_time=2.381
+[gpub052:0/16] 2024-01-22 17:27:33,089 (trainer:737) INFO: 3epoch:train:14201-14300batch: iter_time=8.311e-05, forward_time=0.758, loss_ctc=99.757, loss_att=118.130, acc=0.427, loss=112.618, backward_time=0.489, grad_norm=51.692, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.108, optim0_lr0=2.638e-04, train_time=2.019
+[gpub052:0/16] 2024-01-22 17:30:39,765 (trainer:737) INFO: 3epoch:train:14301-14400batch: iter_time=3.432e-04, forward_time=0.451, loss_ctc=96.953, loss_att=120.370, acc=0.423, loss=113.345, backward_time=0.536, grad_norm=48.298, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.117, optim0_lr0=2.653e-04, train_time=1.865
+[gpub052:0/16] 2024-01-22 17:35:36,190 (trainer:737) INFO: 3epoch:train:14401-14500batch: iter_time=5.346e-04, forward_time=0.726, loss_ctc=112.654, loss_att=139.141, acc=0.400, loss=131.195, backward_time=0.507, grad_norm=53.530, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.141, optim0_lr0=2.668e-04, train_time=2.965
+[gpub052:0/16] 2024-01-22 17:38:10,492 (trainer:737) INFO: 3epoch:train:14501-14600batch: iter_time=2.540e-04, forward_time=0.366, loss_ctc=123.881, loss_att=145.076, acc=0.424, loss=138.717, backward_time=0.410, grad_norm=62.783, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.097, optim0_lr0=2.683e-04, train_time=1.540
+[gpub052:0/16] 2024-01-22 17:42:01,979 (trainer:737) INFO: 3epoch:train:14601-14700batch: iter_time=8.081e-05, forward_time=0.661, loss_ctc=90.493, loss_att=107.280, acc=0.468, loss=102.244, backward_time=0.490, grad_norm=50.053, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.196, optim0_lr0=2.698e-04, train_time=2.316
+[gpub052:0/16] 2024-01-22 17:46:22,057 (trainer:737) INFO: 3epoch:train:14701-14800batch: iter_time=8.209e-05, forward_time=0.633, loss_ctc=114.889, loss_att=126.375, acc=0.464, loss=122.929, backward_time=0.448, grad_norm=59.215, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.127, optim0_lr0=2.713e-04, train_time=2.597
+[gpub052:0/16] 2024-01-22 17:49:01,364 (trainer:737) INFO: 3epoch:train:14801-14900batch: iter_time=3.005e-04, forward_time=0.297, loss_ctc=95.049, loss_att=113.094, acc=0.423, loss=107.680, backward_time=0.411, grad_norm=54.421, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.095, optim0_lr0=2.728e-04, train_time=1.597
+[gpub052:0/16] 2024-01-22 17:53:25,039 (trainer:737) INFO: 3epoch:train:14901-15000batch: iter_time=3.913e-04, forward_time=0.655, loss_ctc=113.689, loss_att=125.617, acc=0.452, loss=122.039, backward_time=0.619, grad_norm=55.788, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.120, optim0_lr0=2.743e-04, train_time=2.635
+
+gpub054:3241516:3241583 [3] proxy.cc:1059 NCCL WARN [Proxy Service] Poll failed with error 1
+[gpub052:0/16] 2024-01-22 18:40:43,570 (trainer:343) INFO: 3epoch results: [train] iter_time=0.590, forward_time=0.436, loss_ctc=129.551, loss_att=173.511, acc=0.291, loss=160.323, backward_time=0.470, grad_norm=57.062, clip=100.000, loss_scale=5.469e+10, optim_step_time=0.105, optim0_lr0=1.625e-04, train_time=2.831, time=11 hours, 48 minutes and 17.87 seconds, total_count=45000, gpu_max_cached_mem_GB=40.713, [valid] loss_ctc=101.939, cer_ctc=0.487, loss_att=102.070, acc=0.327, cer=0.565, wer=1.000, loss=102.030, time=46 minutes and 53.89 seconds, total_count=14013, gpu_max_cached_mem_GB=40.713
+[gpub052:0/16] 2024-01-22 18:41:23,025 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub052:0/16] 2024-01-22 18:41:23,026 (trainer:272) INFO: 4/45epoch started. Estimated time to finish: 3 weeks, 1 day and 1 hour
+[gpub052:0/16] 2024-01-22 18:41:23,036 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub052:0/16] 2024-01-22 18:41:42,037 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 18:41:45,554 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f517bade710>)
+[gpub052:0/16] 2024-01-22 18:41:45,554 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub052:0/16] 2024-01-22 18:41:45,557 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 18:54:43,453 (trainer:737) INFO: 4epoch:train:1-100batch: iter_time=6.491, forward_time=0.431, loss_ctc=102.721, loss_att=134.870, acc=0.438, loss=125.226, backward_time=0.450, grad_norm=49.335, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.098, optim0_lr0=2.758e-04, train_time=8.004
+[gpub052:0/16] 2024-01-22 18:57:12,087 (trainer:737) INFO: 4epoch:train:101-200batch: iter_time=8.015e-05, forward_time=0.323, loss_ctc=108.679, loss_att=115.329, acc=0.455, loss=113.334, backward_time=0.407, grad_norm=58.778, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.103, optim0_lr0=2.773e-04, train_time=1.486
+[gpub052:0/16] 2024-01-22 19:00:59,519 (trainer:737) INFO: 4epoch:train:201-300batch: iter_time=7.988e-05, forward_time=0.545, loss_ctc=105.523, loss_att=129.595, acc=0.411, loss=122.373, backward_time=0.481, grad_norm=55.661, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.123, optim0_lr0=2.788e-04, train_time=2.275
+[gpub052:0/16] 2024-01-22 19:03:34,097 (trainer:737) INFO: 4epoch:train:301-400batch: iter_time=8.337e-05, forward_time=0.293, loss_ctc=103.589, loss_att=117.914, acc=0.459, loss=113.617, backward_time=0.406, grad_norm=54.214, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.095, optim0_lr0=2.803e-04, train_time=1.546
+[gpub052:0/16] 2024-01-22 19:07:21,620 (trainer:737) INFO: 4epoch:train:401-500batch: iter_time=8.076e-05, forward_time=0.509, loss_ctc=95.065, loss_att=107.264, acc=0.439, loss=103.604, backward_time=0.566, grad_norm=49.986, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.120, optim0_lr0=2.818e-04, train_time=2.274
+[gpub052:0/16] 2024-01-22 19:10:12,854 (trainer:737) INFO: 4epoch:train:501-600batch: iter_time=8.022e-05, forward_time=0.293, loss_ctc=124.618, loss_att=138.728, acc=0.427, loss=134.495, backward_time=0.403, grad_norm=62.402, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.095, optim0_lr0=2.833e-04, train_time=1.713
+[gpub052:0/16] 2024-01-22 19:13:34,209 (trainer:737) INFO: 4epoch:train:601-700batch: iter_time=8.260e-05, forward_time=0.564, loss_ctc=106.494, loss_att=113.098, acc=0.436, loss=111.117, backward_time=0.474, grad_norm=51.388, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.156, optim0_lr0=2.848e-04, train_time=2.011
+[gpub052:0/16] 2024-01-22 19:16:59,809 (trainer:737) INFO: 4epoch:train:701-800batch: iter_time=8.944e-05, forward_time=0.290, loss_ctc=104.552, loss_att=122.921, acc=0.435, loss=117.410, backward_time=0.398, grad_norm=54.619, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.094, optim0_lr0=2.863e-04, train_time=2.058
+[gpub052:0/16] 2024-01-22 19:19:56,767 (trainer:737) INFO: 4epoch:train:801-900batch: iter_time=8.282e-04, forward_time=0.294, loss_ctc=120.090, loss_att=142.078, acc=0.432, loss=135.482, backward_time=0.404, grad_norm=57.593, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.097, optim0_lr0=2.878e-04, train_time=1.768
+[gpub052:0/16] 2024-01-22 19:23:22,824 (trainer:737) INFO: 4epoch:train:901-1000batch: iter_time=8.504e-05, forward_time=0.466, loss_ctc=95.242, loss_att=114.233, acc=0.472, loss=108.536, backward_time=0.563, grad_norm=46.783, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.111, optim0_lr0=2.893e-04, train_time=2.060
+[gpub052:0/16] 2024-01-22 19:26:08,155 (trainer:737) INFO: 4epoch:train:1001-1100batch: iter_time=8.478e-05, forward_time=0.290, loss_ctc=97.166, loss_att=98.750, acc=0.479, loss=98.275, backward_time=0.400, grad_norm=53.053, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.095, optim0_lr0=2.908e-04, train_time=1.655
+[gpub052:0/16] 2024-01-22 19:29:36,790 (trainer:737) INFO: 4epoch:train:1101-1200batch: iter_time=1.015e-04, forward_time=0.493, loss_ctc=111.889, loss_att=116.813, acc=0.451, loss=115.336, backward_time=0.520, grad_norm=57.702, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.113, optim0_lr0=2.923e-04, train_time=2.086
+[gpub052:0/16] 2024-01-22 19:31:04,120 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub052:0/16] 2024-01-22 19:31:24,000 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 19:31:27,682 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5a84d3fa30>)
+[gpub052:0/16] 2024-01-22 19:31:27,682 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub052:0/16] 2024-01-22 19:31:27,686 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 19:41:15,068 (trainer:737) INFO: 4epoch:train:1201-1300batch: iter_time=5.595, forward_time=0.292, loss_ctc=96.477, loss_att=116.176, acc=0.459, loss=110.266, backward_time=0.410, grad_norm=49.422, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=2.938e-04, train_time=6.983
+[gpub052:0/16] 2024-01-22 19:43:45,285 (trainer:737) INFO: 4epoch:train:1301-1400batch: iter_time=8.387e-05, forward_time=0.292, loss_ctc=106.128, loss_att=119.030, acc=0.481, loss=115.159, backward_time=0.404, grad_norm=53.020, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=2.953e-04, train_time=1.502
+[gpub052:0/16] 2024-01-22 19:46:38,208 (trainer:737) INFO: 4epoch:train:1401-1500batch: iter_time=0.002, forward_time=0.447, loss_ctc=103.150, loss_att=127.873, acc=0.443, loss=120.456, backward_time=0.449, grad_norm=53.725, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.108, optim0_lr0=2.968e-04, train_time=1.727
+[gpub052:0/16] 2024-01-22 19:48:54,310 (trainer:737) INFO: 4epoch:train:1501-1600batch: iter_time=8.653e-05, forward_time=0.290, loss_ctc=94.525, loss_att=109.548, acc=0.450, loss=105.041, backward_time=0.402, grad_norm=46.833, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.095, optim0_lr0=2.983e-04, train_time=1.362
+[gpub052:0/16] 2024-01-22 19:51:54,339 (trainer:737) INFO: 4epoch:train:1601-1700batch: iter_time=8.408e-05, forward_time=0.292, loss_ctc=103.499, loss_att=105.138, acc=0.487, loss=104.647, backward_time=0.400, grad_norm=51.479, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=2.998e-04, train_time=1.801
+[gpub052:0/16] 2024-01-22 19:56:17,699 (trainer:737) INFO: 4epoch:train:1701-1800batch: iter_time=8.563e-05, forward_time=0.487, loss_ctc=97.788, loss_att=106.221, acc=0.454, loss=103.691, backward_time=0.527, grad_norm=52.135, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.105, optim0_lr0=3.013e-04, train_time=2.633
+[gpub052:0/16] 2024-01-22 19:58:54,810 (trainer:737) INFO: 4epoch:train:1801-1900batch: iter_time=8.223e-05, forward_time=0.415, loss_ctc=114.647, loss_att=128.663, acc=0.441, loss=124.458, backward_time=0.431, grad_norm=56.391, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=3.028e-04, train_time=1.571
+[gpub052:0/16] 2024-01-22 20:01:09,924 (trainer:737) INFO: 4epoch:train:1901-2000batch: iter_time=8.469e-05, forward_time=0.290, loss_ctc=106.921, loss_att=110.359, acc=0.464, loss=109.328, backward_time=0.401, grad_norm=52.576, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=3.043e-04, train_time=1.348
+[gpub052:0/16] 2024-01-22 20:03:57,052 (trainer:737) INFO: 4epoch:train:2001-2100batch: iter_time=9.745e-04, forward_time=0.308, loss_ctc=115.620, loss_att=138.348, acc=0.446, loss=131.529, backward_time=0.420, grad_norm=52.839, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.097, optim0_lr0=3.058e-04, train_time=1.675
+[gpub052:0/16] 2024-01-22 20:07:44,614 (trainer:737) INFO: 4epoch:train:2101-2200batch: iter_time=8.214e-05, forward_time=0.545, loss_ctc=90.686, loss_att=103.736, acc=0.469, loss=99.821, backward_time=0.518, grad_norm=46.206, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.106, optim0_lr0=3.073e-04, train_time=2.275
+[gpub052:0/16] 2024-01-22 20:09:45,536 (trainer:737) INFO: 4epoch:train:2201-2300batch: iter_time=8.183e-05, forward_time=0.291, loss_ctc=90.250, loss_att=106.626, acc=0.488, loss=101.713, backward_time=0.405, grad_norm=44.047, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.094, optim0_lr0=3.088e-04, train_time=1.209
+[gpub052:0/16] 2024-01-22 20:12:37,663 (trainer:737) INFO: 4epoch:train:2301-2400batch: iter_time=8.353e-05, forward_time=0.292, loss_ctc=109.421, loss_att=101.950, acc=0.492, loss=104.192, backward_time=0.401, grad_norm=58.166, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.096, optim0_lr0=3.103e-04, train_time=1.719
+[gpub052:0/16] 2024-01-22 20:16:15,450 (trainer:737) INFO: 4epoch:train:2401-2500batch: iter_time=0.002, forward_time=0.521, loss_ctc=96.448, loss_att=102.637, acc=0.500, loss=100.780, backward_time=0.524, grad_norm=48.423, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.120, optim0_lr0=3.118e-04, train_time=2.181
+[gpub052:0/16] 2024-01-22 20:16:35,496 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub052:0/16] 2024-01-22 20:16:55,354 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 20:16:59,339 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f51568affd0>)
+[gpub052:0/16] 2024-01-22 20:16:59,339 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub052:0/16] 2024-01-22 20:16:59,343 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 20:30:44,888 (trainer:737) INFO: 4epoch:train:2501-2600batch: iter_time=6.799, forward_time=0.292, loss_ctc=97.552, loss_att=132.380, acc=0.469, loss=121.931, backward_time=0.405, grad_norm=53.296, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.095, optim0_lr0=3.133e-04, train_time=8.694
+[gpub052:0/16] 2024-01-22 20:33:58,230 (trainer:737) INFO: 4epoch:train:2601-2700batch: iter_time=0.001, forward_time=0.551, loss_ctc=103.168, loss_att=117.520, acc=0.486, loss=113.214, backward_time=0.491, grad_norm=58.918, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.121, optim0_lr0=3.148e-04, train_time=1.931
+[gpub052:0/16] 2024-01-22 20:36:44,161 (trainer:737) INFO: 4epoch:train:2701-2800batch: iter_time=8.614e-05, forward_time=0.291, loss_ctc=100.606, loss_att=124.744, acc=0.444, loss=117.503, backward_time=0.400, grad_norm=51.030, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.095, optim0_lr0=3.163e-04, train_time=1.659
+[gpub052:0/16] 2024-01-22 20:39:49,168 (trainer:737) INFO: 4epoch:train:2801-2900batch: iter_time=2.512e-04, forward_time=0.326, loss_ctc=98.807, loss_att=110.894, acc=0.501, loss=107.268, backward_time=0.445, grad_norm=51.324, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.100, optim0_lr0=3.178e-04, train_time=1.852
+[gpub052:0/16] 2024-01-22 20:42:50,999 (trainer:737) INFO: 4epoch:train:2901-3000batch: iter_time=8.860e-05, forward_time=0.463, loss_ctc=90.170, loss_att=100.679, acc=0.489, loss=97.526, backward_time=0.545, grad_norm=47.451, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.108, optim0_lr0=3.193e-04, train_time=1.818
+[gpub052:0/16] 2024-01-22 20:45:25,244 (trainer:737) INFO: 4epoch:train:3001-3100batch: iter_time=4.813e-04, forward_time=0.332, loss_ctc=120.109, loss_att=137.089, acc=0.457, loss=131.995, backward_time=0.416, grad_norm=57.888, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.099, optim0_lr0=3.208e-04, train_time=1.540
+[gpub052:0/16] 2024-01-22 20:48:43,774 (trainer:737) INFO: 4epoch:train:3101-3200batch: iter_time=8.518e-05, forward_time=0.451, loss_ctc=100.290, loss_att=105.460, acc=0.479, loss=103.909, backward_time=0.472, grad_norm=50.996, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.106, optim0_lr0=3.223e-04, train_time=1.987
+[gpub052:0/16] 2024-01-22 20:51:15,846 (trainer:737) INFO: 4epoch:train:3201-3300batch: iter_time=3.210e-04, forward_time=0.312, loss_ctc=98.129, loss_att=120.334, acc=0.465, loss=113.672, backward_time=0.405, grad_norm=49.224, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.095, optim0_lr0=3.238e-04, train_time=1.521
+[gpub052:0/16] 2024-01-22 20:54:30,960 (trainer:737) INFO: 4epoch:train:3301-3400batch: iter_time=8.233e-05, forward_time=0.550, loss_ctc=114.734, loss_att=134.516, acc=0.477, loss=128.582, backward_time=0.504, grad_norm=55.482, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.108, optim0_lr0=3.253e-04, train_time=1.948
+[gpub052:0/16] 2024-01-22 20:57:44,121 (trainer:737) INFO: 4epoch:train:3401-3500batch: iter_time=9.744e-05, forward_time=0.292, loss_ctc=90.139, loss_att=109.294, acc=0.512, loss=103.548, backward_time=0.402, grad_norm=44.963, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.099, optim0_lr0=3.268e-04, train_time=1.934
+[gpub052:0/16] 2024-01-22 21:00:59,337 (trainer:737) INFO: 4epoch:train:3501-3600batch: iter_time=1.029e-04, forward_time=0.579, loss_ctc=92.778, loss_att=90.607, acc=0.513, loss=91.259, backward_time=0.517, grad_norm=50.277, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.130, optim0_lr0=3.283e-04, train_time=1.952
+[gpub052:0/16] 2024-01-22 21:04:21,894 (trainer:737) INFO: 4epoch:train:3601-3700batch: iter_time=8.484e-05, forward_time=0.301, loss_ctc=106.074, loss_att=107.950, acc=0.491, loss=107.387, backward_time=0.404, grad_norm=56.387, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.096, optim0_lr0=3.298e-04, train_time=2.023
+[gpub052:0/16] 2024-01-22 21:06:09,254 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub052:0/16] 2024-01-22 21:06:29,254 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 21:06:32,886 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5a25bd8ca0>)
+[gpub052:0/16] 2024-01-22 21:06:32,886 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub052:0/16] 2024-01-22 21:06:32,889 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 21:21:32,863 (trainer:737) INFO: 4epoch:train:3701-3800batch: iter_time=7.934, forward_time=0.746, loss_ctc=91.137, loss_att=111.375, acc=0.495, loss=105.304, backward_time=0.501, grad_norm=46.442, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.116, optim0_lr0=3.313e-04, train_time=10.312
+[gpub052:0/16] 2024-01-22 21:26:00,413 (trainer:737) INFO: 4epoch:train:3801-3900batch: iter_time=0.167, forward_time=0.297, loss_ctc=102.310, loss_att=115.360, acc=0.513, loss=111.445, backward_time=0.415, grad_norm=54.301, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.094, optim0_lr0=3.328e-04, train_time=2.675
+[gpub052:0/16] 2024-01-22 21:31:10,267 (trainer:737) INFO: 4epoch:train:3901-4000batch: iter_time=8.322e-05, forward_time=0.599, loss_ctc=97.461, loss_att=123.955, acc=0.471, loss=116.007, backward_time=0.519, grad_norm=55.324, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.143, optim0_lr0=3.343e-04, train_time=3.098
+[gpub052:0/16] 2024-01-22 21:33:51,262 (trainer:737) INFO: 4epoch:train:4001-4100batch: iter_time=8.562e-05, forward_time=0.289, loss_ctc=89.303, loss_att=100.770, acc=0.490, loss=97.330, backward_time=0.406, grad_norm=46.203, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.095, optim0_lr0=3.358e-04, train_time=1.606
+[gpub052:0/16] 2024-01-22 21:37:36,331 (trainer:737) INFO: 4epoch:train:4101-4200batch: iter_time=9.558e-05, forward_time=0.409, loss_ctc=98.042, loss_att=98.049, acc=0.527, loss=98.047, backward_time=0.428, grad_norm=48.356, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.104, optim0_lr0=3.373e-04, train_time=2.254
+[gpub052:0/16] 2024-01-22 21:40:39,249 (trainer:737) INFO: 4epoch:train:4201-4300batch: iter_time=8.102e-05, forward_time=0.475, loss_ctc=94.870, loss_att=98.856, acc=0.491, loss=97.660, backward_time=0.461, grad_norm=51.806, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.102, optim0_lr0=3.388e-04, train_time=1.829
+[gpub052:0/16] 2024-01-22 21:43:36,767 (trainer:737) INFO: 4epoch:train:4301-4400batch: iter_time=2.595e-04, forward_time=0.335, loss_ctc=109.948, loss_att=126.674, acc=0.464, loss=121.656, backward_time=0.517, grad_norm=55.438, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.105, optim0_lr0=3.403e-04, train_time=1.775
+[gpub052:0/16] 2024-01-22 21:48:30,837 (trainer:737) INFO: 4epoch:train:4401-4500batch: iter_time=8.195e-05, forward_time=0.560, loss_ctc=103.386, loss_att=106.288, acc=0.495, loss=105.418, backward_time=0.468, grad_norm=50.079, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.102, optim0_lr0=3.418e-04, train_time=2.939
+[gpub052:0/16] 2024-01-22 21:51:06,096 (trainer:737) INFO: 4epoch:train:4501-4600batch: iter_time=8.238e-05, forward_time=0.296, loss_ctc=110.946, loss_att=132.659, acc=0.483, loss=126.145, backward_time=0.408, grad_norm=52.862, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.095, optim0_lr0=3.433e-04, train_time=1.554
+[gpub052:0/16] 2024-01-22 21:56:17,793 (trainer:737) INFO: 4epoch:train:4601-4700batch: iter_time=3.321e-04, forward_time=0.558, loss_ctc=88.269, loss_att=99.804, acc=0.505, loss=96.344, backward_time=0.601, grad_norm=46.711, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.118, optim0_lr0=3.448e-04, train_time=3.115
+[gpub052:0/16] 2024-01-22 22:01:22,862 (trainer:737) INFO: 4epoch:train:4701-4800batch: iter_time=0.137, forward_time=0.297, loss_ctc=86.909, loss_att=100.168, acc=0.518, loss=96.190, backward_time=0.422, grad_norm=44.094, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.095, optim0_lr0=3.463e-04, train_time=3.052
+[gpub052:0/16] 2024-01-22 22:04:03,018 (trainer:737) INFO: 4epoch:train:4801-4900batch: iter_time=8.466e-05, forward_time=0.292, loss_ctc=104.761, loss_att=95.696, acc=0.520, loss=98.415, backward_time=0.405, grad_norm=55.750, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.094, optim0_lr0=3.478e-04, train_time=1.601
+[gpub052:0/16] 2024-01-22 22:08:11,052 (trainer:737) INFO: 4epoch:train:4901-5000batch: iter_time=3.270e-04, forward_time=0.554, loss_ctc=91.561, loss_att=95.842, acc=0.535, loss=94.558, backward_time=0.548, grad_norm=44.837, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.134, optim0_lr0=3.493e-04, train_time=2.478
+[gpub052:0/16] 2024-01-22 22:08:31,097 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub052:0/16] 2024-01-22 22:08:51,263 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 22:08:54,821 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f516841be20>)
+[gpub052:0/16] 2024-01-22 22:08:54,821 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub052:0/16] 2024-01-22 22:08:54,825 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 22:23:14,549 (trainer:737) INFO: 4epoch:train:5001-5100batch: iter_time=7.263, forward_time=0.385, loss_ctc=93.789, loss_att=116.610, acc=0.498, loss=109.764, backward_time=0.429, grad_norm=49.035, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.118, optim0_lr0=3.508e-04, train_time=9.037
+[gpub052:0/16] 2024-01-22 22:26:28,591 (trainer:737) INFO: 4epoch:train:5101-5200batch: iter_time=8.244e-05, forward_time=0.466, loss_ctc=99.626, loss_att=102.109, acc=0.514, loss=101.364, backward_time=0.486, grad_norm=54.983, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.101, optim0_lr0=3.523e-04, train_time=1.941
+[gpub052:0/16] 2024-01-22 22:29:28,188 (trainer:737) INFO: 4epoch:train:5201-5300batch: iter_time=8.190e-05, forward_time=0.290, loss_ctc=94.575, loss_att=109.815, acc=0.471, loss=105.243, backward_time=0.400, grad_norm=47.880, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.538e-04, train_time=1.794
+[gpub052:0/16] 2024-01-22 22:33:00,047 (trainer:737) INFO: 4epoch:train:5301-5400batch: iter_time=8.412e-05, forward_time=0.492, loss_ctc=94.123, loss_att=100.733, acc=0.525, loss=98.750, backward_time=0.546, grad_norm=47.747, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.122, optim0_lr0=3.553e-04, train_time=2.119
+[gpub052:0/16] 2024-01-22 22:35:40,629 (trainer:737) INFO: 4epoch:train:5401-5500batch: iter_time=8.338e-05, forward_time=0.288, loss_ctc=87.863, loss_att=92.652, acc=0.498, loss=91.215, backward_time=0.400, grad_norm=47.081, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.568e-04, train_time=1.607
+[gpub052:0/16] 2024-01-22 22:38:34,712 (trainer:737) INFO: 4epoch:train:5501-5600batch: iter_time=8.148e-05, forward_time=0.381, loss_ctc=115.289, loss_att=123.442, acc=0.481, loss=120.996, backward_time=0.416, grad_norm=57.669, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.097, optim0_lr0=3.583e-04, train_time=1.740
+[gpub052:0/16] 2024-01-22 22:41:09,460 (trainer:737) INFO: 4epoch:train:5601-5700batch: iter_time=8.304e-05, forward_time=0.416, loss_ctc=96.330, loss_att=94.079, acc=0.501, loss=94.754, backward_time=0.431, grad_norm=47.167, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.102, optim0_lr0=3.598e-04, train_time=1.547
+[gpub052:0/16] 2024-01-22 22:44:18,086 (trainer:737) INFO: 4epoch:train:5701-5800batch: iter_time=8.223e-05, forward_time=0.292, loss_ctc=95.855, loss_att=106.560, acc=0.492, loss=103.349, backward_time=0.400, grad_norm=48.587, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.613e-04, train_time=1.887
+[gpub052:0/16] 2024-01-22 22:47:28,037 (trainer:737) INFO: 4epoch:train:5801-5900batch: iter_time=7.072e-04, forward_time=0.323, loss_ctc=111.353, loss_att=125.036, acc=0.486, loss=120.931, backward_time=0.438, grad_norm=54.065, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.106, optim0_lr0=3.628e-04, train_time=1.897
+[gpub052:0/16] 2024-01-22 22:50:17,499 (trainer:737) INFO: 4epoch:train:5901-6000batch: iter_time=2.015e-04, forward_time=0.391, loss_ctc=86.613, loss_att=98.841, acc=0.528, loss=95.173, backward_time=0.463, grad_norm=42.601, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.103, optim0_lr0=3.643e-04, train_time=1.694
+[gpub052:0/16] 2024-01-22 22:52:50,124 (trainer:737) INFO: 4epoch:train:6001-6100batch: iter_time=8.522e-05, forward_time=0.289, loss_ctc=89.656, loss_att=83.830, acc=0.531, loss=85.578, backward_time=0.401, grad_norm=49.047, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.658e-04, train_time=1.528
+[gpub052:0/16] 2024-01-22 22:55:29,586 (trainer:737) INFO: 4epoch:train:6101-6200batch: iter_time=9.861e-05, forward_time=0.385, loss_ctc=102.780, loss_att=98.541, acc=0.512, loss=99.813, backward_time=0.445, grad_norm=54.908, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.104, optim0_lr0=3.673e-04, train_time=1.593
+[gpub052:0/16] 2024-01-22 22:57:04,857 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub052:0/16] 2024-01-22 22:57:24,535 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 22:57:28,099 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f585bc037c0>)
+[gpub052:0/16] 2024-01-22 22:57:28,099 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub052:0/16] 2024-01-22 22:57:28,103 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 23:03:57,961 (trainer:737) INFO: 4epoch:train:6201-6300batch: iter_time=3.631, forward_time=0.309, loss_ctc=88.536, loss_att=101.125, acc=0.513, loss=97.349, backward_time=0.403, grad_norm=46.211, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.688e-04, train_time=5.084
+[gpub052:0/16] 2024-01-22 23:06:19,112 (trainer:737) INFO: 4epoch:train:6301-6400batch: iter_time=8.389e-05, forward_time=0.429, loss_ctc=98.150, loss_att=102.898, acc=0.534, loss=101.474, backward_time=0.484, grad_norm=48.551, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.105, optim0_lr0=3.703e-04, train_time=1.410
+[gpub052:0/16] 2024-01-22 23:08:51,494 (trainer:737) INFO: 4epoch:train:6401-6500batch: iter_time=8.164e-05, forward_time=0.292, loss_ctc=94.739, loss_att=111.201, acc=0.493, loss=106.263, backward_time=0.404, grad_norm=49.175, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.718e-04, train_time=1.524
+[gpub052:0/16] 2024-01-22 23:11:49,346 (trainer:737) INFO: 4epoch:train:6501-6600batch: iter_time=8.837e-05, forward_time=0.441, loss_ctc=86.454, loss_att=94.895, acc=0.505, loss=92.363, backward_time=0.462, grad_norm=45.219, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.104, optim0_lr0=3.733e-04, train_time=1.778
+[gpub052:0/16] 2024-01-22 23:13:46,735 (trainer:737) INFO: 4epoch:train:6601-6700batch: iter_time=8.107e-05, forward_time=0.309, loss_ctc=95.104, loss_att=91.264, acc=0.538, loss=92.416, backward_time=0.405, grad_norm=46.611, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.748e-04, train_time=1.173
+[gpub052:0/16] 2024-01-22 23:17:08,143 (trainer:737) INFO: 4epoch:train:6701-6800batch: iter_time=8.314e-05, forward_time=0.511, loss_ctc=91.403, loss_att=92.167, acc=0.500, loss=91.938, backward_time=0.444, grad_norm=51.462, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.105, optim0_lr0=3.763e-04, train_time=2.015
+[gpub052:0/16] 2024-01-22 23:19:41,507 (trainer:737) INFO: 4epoch:train:6801-6900batch: iter_time=8.437e-05, forward_time=0.292, loss_ctc=109.369, loss_att=115.948, acc=0.488, loss=113.974, backward_time=0.406, grad_norm=54.088, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.095, optim0_lr0=3.778e-04, train_time=1.533
+[gpub052:0/16] 2024-01-22 23:22:11,030 (trainer:737) INFO: 4epoch:train:6901-7000batch: iter_time=5.844e-04, forward_time=0.478, loss_ctc=100.250, loss_att=95.949, acc=0.513, loss=97.240, backward_time=0.450, grad_norm=49.970, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.113, optim0_lr0=3.793e-04, train_time=1.495
+[gpub052:0/16] 2024-01-22 23:25:06,255 (trainer:737) INFO: 4epoch:train:7001-7100batch: iter_time=7.834e-04, forward_time=0.380, loss_ctc=107.739, loss_att=122.664, acc=0.493, loss=118.186, backward_time=0.410, grad_norm=50.642, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.097, optim0_lr0=3.808e-04, train_time=1.752
+[gpub052:0/16] 2024-01-22 23:27:34,889 (trainer:737) INFO: 4epoch:train:7101-7200batch: iter_time=8.247e-05, forward_time=0.402, loss_ctc=85.231, loss_att=91.194, acc=0.516, loss=89.405, backward_time=0.460, grad_norm=43.709, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.100, optim0_lr0=3.823e-04, train_time=1.486
+[gpub052:0/16] 2024-01-22 23:30:14,558 (trainer:737) INFO: 4epoch:train:7201-7300batch: iter_time=6.471e-04, forward_time=0.347, loss_ctc=84.599, loss_att=93.386, acc=0.536, loss=90.750, backward_time=0.408, grad_norm=41.795, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.131, optim0_lr0=3.838e-04, train_time=1.595
+[gpub052:0/16] 2024-01-22 23:32:28,236 (trainer:737) INFO: 4epoch:train:7301-7400batch: iter_time=8.220e-05, forward_time=0.409, loss_ctc=101.760, loss_att=89.221, acc=0.535, loss=92.983, backward_time=0.456, grad_norm=53.786, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.100, optim0_lr0=3.853e-04, train_time=1.338
+[gpub052:0/16] 2024-01-22 23:35:20,564 (trainer:737) INFO: 4epoch:train:7401-7500batch: iter_time=8.243e-05, forward_time=0.290, loss_ctc=89.042, loss_att=88.473, acc=0.547, loss=88.644, backward_time=0.401, grad_norm=43.756, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.096, optim0_lr0=3.868e-04, train_time=1.723
+[gpub052:0/16] 2024-01-22 23:35:41,235 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub052:0/16] 2024-01-22 23:36:00,908 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-22 23:36:04,535 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f583dcf7550>)
+[gpub052:0/16] 2024-01-22 23:36:04,535 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub052:0/16] 2024-01-22 23:36:04,687 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-22 23:44:51,086 (trainer:737) INFO: 4epoch:train:7501-7600batch: iter_time=4.421, forward_time=0.398, loss_ctc=91.693, loss_att=117.298, acc=0.512, loss=109.616, backward_time=0.423, grad_norm=49.420, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.100, optim0_lr0=3.883e-04, train_time=5.704
+[gpub052:0/16] 2024-01-22 23:47:50,424 (trainer:737) INFO: 4epoch:train:7601-7700batch: iter_time=8.357e-05, forward_time=0.448, loss_ctc=96.592, loss_att=104.234, acc=0.527, loss=101.941, backward_time=0.475, grad_norm=54.486, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.116, optim0_lr0=3.898e-04, train_time=1.792
+[gpub052:0/16] 2024-01-22 23:50:25,150 (trainer:737) INFO: 4epoch:train:7701-7800batch: iter_time=8.007e-05, forward_time=0.292, loss_ctc=92.201, loss_att=107.201, acc=0.493, loss=102.701, backward_time=0.400, grad_norm=47.784, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.096, optim0_lr0=3.913e-04, train_time=1.549
+[gpub052:0/16] 2024-01-22 23:53:43,779 (trainer:737) INFO: 4epoch:train:7801-7900batch: iter_time=8.276e-05, forward_time=0.393, loss_ctc=91.918, loss_att=97.674, acc=0.546, loss=95.947, backward_time=0.588, grad_norm=47.438, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.109, optim0_lr0=3.928e-04, train_time=1.982
+[gpub052:0/16] 2024-01-22 23:55:51,311 (trainer:737) INFO: 4epoch:train:7901-8000batch: iter_time=8.288e-05, forward_time=0.290, loss_ctc=84.223, loss_att=88.307, acc=0.530, loss=87.082, backward_time=0.403, grad_norm=44.499, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.095, optim0_lr0=3.943e-04, train_time=1.279
+[gpub052:0/16] 2024-01-23 00:00:39,060 (trainer:737) INFO: 4epoch:train:8001-8100batch: iter_time=1.849e-04, forward_time=0.464, loss_ctc=114.189, loss_att=122.838, acc=0.498, loss=120.243, backward_time=0.448, grad_norm=60.089, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.111, optim0_lr0=3.958e-04, train_time=2.874
+[gpub052:0/16] 2024-01-23 00:03:38,914 (trainer:737) INFO: 4epoch:train:8101-8200batch: iter_time=8.292e-05, forward_time=0.296, loss_ctc=93.819, loss_att=92.228, acc=0.521, loss=92.705, backward_time=0.399, grad_norm=47.032, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.095, optim0_lr0=3.973e-04, train_time=1.802
+[gpub052:0/16] 2024-01-23 00:06:44,546 (trainer:737) INFO: 4epoch:train:8201-8300batch: iter_time=8.289e-05, forward_time=0.292, loss_ctc=92.574, loss_att=105.699, acc=0.509, loss=101.761, backward_time=0.406, grad_norm=46.750, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.094, optim0_lr0=3.988e-04, train_time=1.856
+[gpub052:0/16] 2024-01-23 00:09:50,177 (trainer:737) INFO: 4epoch:train:8301-8400batch: iter_time=0.003, forward_time=0.414, loss_ctc=108.082, loss_att=120.372, acc=0.518, loss=116.685, backward_time=0.506, grad_norm=52.063, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.124, optim0_lr0=4.003e-04, train_time=1.854
+[gpub052:0/16] 2024-01-23 00:11:55,875 (trainer:737) INFO: 4epoch:train:8401-8500batch: iter_time=8.434e-05, forward_time=0.295, loss_ctc=85.123, loss_att=98.247, acc=0.550, loss=94.310, backward_time=0.408, grad_norm=42.644, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.095, optim0_lr0=4.018e-04, train_time=1.259
+[gpub052:0/16] 2024-01-23 00:15:02,191 (trainer:737) INFO: 4epoch:train:8501-8600batch: iter_time=7.658e-04, forward_time=0.493, loss_ctc=86.554, loss_att=79.550, acc=0.550, loss=81.651, backward_time=0.461, grad_norm=48.684, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.125, optim0_lr0=4.033e-04, train_time=1.863
+[gpub052:0/16] 2024-01-23 00:17:52,835 (trainer:737) INFO: 4epoch:train:8601-8700batch: iter_time=8.416e-05, forward_time=0.291, loss_ctc=100.038, loss_att=94.306, acc=0.531, loss=96.026, backward_time=0.401, grad_norm=53.271, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.094, optim0_lr0=4.048e-04, train_time=1.704
+[gpub052:0/16] 2024-01-23 00:19:11,773 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub052:0/16] 2024-01-23 00:19:31,840 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 00:19:35,652 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58cd577820>)
+[gpub052:0/16] 2024-01-23 00:19:35,652 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub052:0/16] 2024-01-23 00:19:35,655 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 00:27:46,403 (trainer:737) INFO: 4epoch:train:8701-8800batch: iter_time=4.514, forward_time=0.404, loss_ctc=86.177, loss_att=99.775, acc=0.532, loss=95.696, backward_time=0.423, grad_norm=45.331, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.098, optim0_lr0=4.063e-04, train_time=5.938
+[gpub052:0/16] 2024-01-23 00:30:09,059 (trainer:737) INFO: 4epoch:train:8801-8900batch: iter_time=8.262e-05, forward_time=0.294, loss_ctc=97.066, loss_att=103.981, acc=0.548, loss=101.907, backward_time=0.407, grad_norm=52.488, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.095, optim0_lr0=4.078e-04, train_time=1.426
+[gpub052:0/16] 2024-01-23 00:32:51,065 (trainer:737) INFO: 4epoch:train:8901-9000batch: iter_time=8.248e-05, forward_time=0.291, loss_ctc=93.543, loss_att=112.783, acc=0.506, loss=107.011, backward_time=0.405, grad_norm=51.911, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.095, optim0_lr0=4.093e-04, train_time=1.617
+[gpub052:0/16] 2024-01-23 00:35:07,460 (trainer:737) INFO: 4epoch:train:9001-9100batch: iter_time=8.220e-05, forward_time=0.463, loss_ctc=86.226, loss_att=91.458, acc=0.521, loss=89.889, backward_time=0.454, grad_norm=45.118, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.101, optim0_lr0=4.108e-04, train_time=1.366
+[gpub052:0/16] 2024-01-23 00:37:33,724 (trainer:737) INFO: 4epoch:train:9101-9200batch: iter_time=8.454e-05, forward_time=0.293, loss_ctc=92.898, loss_att=87.681, acc=0.564, loss=89.246, backward_time=0.404, grad_norm=43.941, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.123e-04, train_time=1.462
+[gpub052:0/16] 2024-01-23 00:40:05,460 (trainer:737) INFO: 4epoch:train:9201-9300batch: iter_time=8.692e-05, forward_time=0.290, loss_ctc=87.955, loss_att=88.013, acc=0.530, loss=87.996, backward_time=0.402, grad_norm=45.018, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.138e-04, train_time=1.517
+[gpub052:0/16] 2024-01-23 00:42:31,065 (trainer:737) INFO: 4epoch:train:9301-9400batch: iter_time=5.661e-04, forward_time=0.392, loss_ctc=104.866, loss_att=116.544, acc=0.500, loss=113.040, backward_time=0.429, grad_norm=52.491, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.112, optim0_lr0=4.153e-04, train_time=1.456
+[gpub052:0/16] 2024-01-23 00:44:23,172 (trainer:737) INFO: 4epoch:train:9401-9500batch: iter_time=8.434e-05, forward_time=0.292, loss_ctc=98.099, loss_att=95.608, acc=0.529, loss=96.355, backward_time=0.405, grad_norm=46.088, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.094, optim0_lr0=4.168e-04, train_time=1.121
+[gpub052:0/16] 2024-01-23 00:47:00,057 (trainer:737) INFO: 4epoch:train:9501-9600batch: iter_time=8.944e-05, forward_time=0.295, loss_ctc=104.596, loss_att=120.023, acc=0.520, loss=115.395, backward_time=0.406, grad_norm=47.808, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.094, optim0_lr0=4.183e-04, train_time=1.569
+[gpub052:0/16] 2024-01-23 00:49:11,918 (trainer:737) INFO: 4epoch:train:9601-9700batch: iter_time=2.781e-04, forward_time=0.374, loss_ctc=83.526, loss_att=88.598, acc=0.542, loss=87.076, backward_time=0.439, grad_norm=42.397, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.102, optim0_lr0=4.198e-04, train_time=1.318
+[gpub052:0/16] 2024-01-23 00:51:25,322 (trainer:737) INFO: 4epoch:train:9701-9800batch: iter_time=8.362e-05, forward_time=0.315, loss_ctc=82.011, loss_att=91.465, acc=0.550, loss=88.629, backward_time=0.410, grad_norm=41.660, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.097, optim0_lr0=4.213e-04, train_time=1.334
+[gpub052:0/16] 2024-01-23 00:53:40,762 (trainer:737) INFO: 4epoch:train:9801-9900batch: iter_time=8.695e-05, forward_time=0.292, loss_ctc=98.894, loss_att=85.480, acc=0.554, loss=89.504, backward_time=0.405, grad_norm=52.554, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.094, optim0_lr0=4.228e-04, train_time=1.354
+[gpub052:0/16] 2024-01-23 00:55:49,226 (trainer:737) INFO: 4epoch:train:9901-10000batch: iter_time=8.038e-05, forward_time=0.294, loss_ctc=87.310, loss_att=85.261, acc=0.569, loss=85.876, backward_time=0.405, grad_norm=42.480, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.243e-04, train_time=1.284
+[gpub052:0/16] 2024-01-23 00:56:09,264 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub052:0/16] 2024-01-23 00:56:29,051 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 00:56:32,665 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f583dcb03d0>)
+[gpub052:0/16] 2024-01-23 00:56:32,665 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub052:0/16] 2024-01-23 00:56:32,668 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 01:03:42,631 (trainer:737) INFO: 4epoch:train:10001-10100batch: iter_time=3.466, forward_time=0.370, loss_ctc=89.222, loss_att=107.865, acc=0.528, loss=102.272, backward_time=0.418, grad_norm=46.803, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.097, optim0_lr0=4.258e-04, train_time=4.734
+[gpub052:0/16] 2024-01-23 01:06:29,429 (trainer:737) INFO: 4epoch:train:10101-10200batch: iter_time=8.282e-05, forward_time=0.422, loss_ctc=93.040, loss_att=93.723, acc=0.541, loss=93.518, backward_time=0.442, grad_norm=54.430, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.099, optim0_lr0=4.273e-04, train_time=1.668
+[gpub052:0/16] 2024-01-23 01:08:27,150 (trainer:737) INFO: 4epoch:train:10201-10300batch: iter_time=8.045e-05, forward_time=0.291, loss_ctc=89.762, loss_att=101.653, acc=0.500, loss=98.086, backward_time=0.405, grad_norm=47.987, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.288e-04, train_time=1.177
+[gpub052:0/16] 2024-01-23 01:10:40,718 (trainer:737) INFO: 4epoch:train:10301-10400batch: iter_time=8.546e-05, forward_time=0.386, loss_ctc=90.310, loss_att=93.012, acc=0.554, loss=92.201, backward_time=0.454, grad_norm=45.617, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.101, optim0_lr0=4.303e-04, train_time=1.335
+[gpub052:0/16] 2024-01-23 01:13:08,954 (trainer:737) INFO: 4epoch:train:10401-10500batch: iter_time=8.017e-05, forward_time=0.289, loss_ctc=82.157, loss_att=84.706, acc=0.527, loss=83.941, backward_time=0.401, grad_norm=44.339, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.318e-04, train_time=1.482
+[gpub052:0/16] 2024-01-23 01:15:31,727 (trainer:737) INFO: 4epoch:train:10501-10600batch: iter_time=0.004, forward_time=0.385, loss_ctc=113.520, loss_att=115.842, acc=0.507, loss=115.145, backward_time=0.437, grad_norm=57.745, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.104, optim0_lr0=4.333e-04, train_time=1.427
+[gpub052:0/16] 2024-01-23 01:17:40,074 (trainer:737) INFO: 4epoch:train:10601-10700batch: iter_time=7.997e-05, forward_time=0.292, loss_ctc=92.596, loss_att=86.315, acc=0.532, loss=88.199, backward_time=0.409, grad_norm=46.347, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.095, optim0_lr0=4.348e-04, train_time=1.284
+[gpub052:0/16] 2024-01-23 01:19:51,948 (trainer:737) INFO: 4epoch:train:10701-10800batch: iter_time=9.494e-04, forward_time=0.392, loss_ctc=89.790, loss_att=98.552, acc=0.519, loss=95.923, backward_time=0.444, grad_norm=44.291, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.103, optim0_lr0=4.363e-04, train_time=1.318
+[gpub052:0/16] 2024-01-23 01:22:27,689 (trainer:737) INFO: 4epoch:train:10801-10900batch: iter_time=8.190e-05, forward_time=0.304, loss_ctc=106.404, loss_att=116.323, acc=0.513, loss=113.348, backward_time=0.415, grad_norm=53.409, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.100, optim0_lr0=4.378e-04, train_time=1.558
+[gpub052:0/16] 2024-01-23 01:24:32,956 (trainer:737) INFO: 4epoch:train:10901-11000batch: iter_time=4.484e-04, forward_time=0.364, loss_ctc=82.099, loss_att=91.972, acc=0.557, loss=89.010, backward_time=0.446, grad_norm=39.578, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.102, optim0_lr0=4.393e-04, train_time=1.253
+[gpub052:0/16] 2024-01-23 01:27:09,080 (trainer:737) INFO: 4epoch:train:11001-11100batch: iter_time=0.003, forward_time=0.425, loss_ctc=84.439, loss_att=75.467, acc=0.562, loss=78.159, backward_time=0.427, grad_norm=46.160, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.134, optim0_lr0=4.408e-04, train_time=1.560
+[gpub052:0/16] 2024-01-23 01:29:26,763 (trainer:737) INFO: 4epoch:train:11101-11200batch: iter_time=9.203e-05, forward_time=0.292, loss_ctc=97.454, loss_att=89.872, acc=0.536, loss=92.146, backward_time=0.403, grad_norm=49.376, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.094, optim0_lr0=4.423e-04, train_time=1.378
+[gpub052:0/16] 2024-01-23 01:30:41,365 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub052:0/16] 2024-01-23 01:31:01,012 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 01:31:04,997 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5ba42e3190>)
+[gpub052:0/16] 2024-01-23 01:31:04,997 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub052:0/16] 2024-01-23 01:31:05,000 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 01:37:07,071 (trainer:737) INFO: 4epoch:train:11201-11300batch: iter_time=3.402, forward_time=0.369, loss_ctc=83.364, loss_att=95.183, acc=0.543, loss=91.637, backward_time=0.417, grad_norm=41.113, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.096, optim0_lr0=4.438e-04, train_time=4.603
+[gpub052:0/16] 2024-01-23 01:39:21,382 (trainer:737) INFO: 4epoch:train:11301-11400batch: iter_time=2.171e-04, forward_time=0.370, loss_ctc=93.608, loss_att=101.509, acc=0.557, loss=99.139, backward_time=0.464, grad_norm=50.042, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.108, optim0_lr0=4.453e-04, train_time=1.342
+[gpub052:0/16] 2024-01-23 01:41:50,607 (trainer:737) INFO: 4epoch:train:11401-11500batch: iter_time=9.373e-05, forward_time=0.292, loss_ctc=92.097, loss_att=109.062, acc=0.519, loss=103.973, backward_time=0.402, grad_norm=51.847, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.093, optim0_lr0=4.468e-04, train_time=1.492
+[gpub052:0/16] 2024-01-23 01:43:56,402 (trainer:737) INFO: 4epoch:train:11501-11600batch: iter_time=8.740e-05, forward_time=0.319, loss_ctc=82.907, loss_att=87.809, acc=0.538, loss=86.338, backward_time=0.404, grad_norm=42.936, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.095, optim0_lr0=4.483e-04, train_time=1.258
+[gpub052:0/16] 2024-01-23 01:46:34,468 (trainer:737) INFO: 4epoch:train:11601-11700batch: iter_time=8.983e-05, forward_time=0.407, loss_ctc=91.553, loss_att=85.445, acc=0.575, loss=87.278, backward_time=0.432, grad_norm=42.382, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.098, optim0_lr0=4.498e-04, train_time=1.580
+[gpub052:0/16] 2024-01-23 01:48:45,279 (trainer:737) INFO: 4epoch:train:11701-11800batch: iter_time=8.724e-05, forward_time=0.291, loss_ctc=86.886, loss_att=85.596, acc=0.541, loss=85.983, backward_time=0.402, grad_norm=44.433, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.093, optim0_lr0=4.513e-04, train_time=1.308
+[gpub052:0/16] 2024-01-23 01:50:46,773 (trainer:737) INFO: 4epoch:train:11801-11900batch: iter_time=3.717e-04, forward_time=0.301, loss_ctc=102.803, loss_att=111.360, acc=0.515, loss=108.793, backward_time=0.408, grad_norm=55.076, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.095, optim0_lr0=4.528e-04, train_time=1.214
+[gpub052:0/16] 2024-01-23 01:53:23,494 (trainer:737) INFO: 4epoch:train:11901-12000batch: iter_time=8.558e-05, forward_time=0.345, loss_ctc=96.310, loss_att=93.305, acc=0.538, loss=94.207, backward_time=0.472, grad_norm=47.639, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.097, optim0_lr0=4.543e-04, train_time=1.567
+[gpub052:0/16] 2024-01-23 01:55:17,195 (trainer:737) INFO: 4epoch:train:12001-12100batch: iter_time=8.692e-05, forward_time=0.295, loss_ctc=103.209, loss_att=118.404, acc=0.525, loss=113.846, backward_time=0.408, grad_norm=49.674, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.094, optim0_lr0=4.558e-04, train_time=1.137
+[gpub052:0/16] 2024-01-23 01:57:53,294 (trainer:737) INFO: 4epoch:train:12101-12200batch: iter_time=8.537e-05, forward_time=0.391, loss_ctc=80.801, loss_att=84.392, acc=0.559, loss=83.315, backward_time=0.440, grad_norm=40.552, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.106, optim0_lr0=4.573e-04, train_time=1.561
+[gpub052:0/16] 2024-01-23 01:59:58,767 (trainer:737) INFO: 4epoch:train:12201-12300batch: iter_time=9.027e-05, forward_time=0.295, loss_ctc=80.373, loss_att=88.191, acc=0.562, loss=85.845, backward_time=0.409, grad_norm=40.897, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.094, optim0_lr0=4.588e-04, train_time=1.254
+[gpub052:0/16] 2024-01-23 02:02:11,235 (trainer:737) INFO: 4epoch:train:12301-12400batch: iter_time=8.305e-05, forward_time=0.292, loss_ctc=97.569, loss_att=83.281, acc=0.561, loss=87.567, backward_time=0.404, grad_norm=49.784, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.094, optim0_lr0=4.603e-04, train_time=1.325
+[gpub052:0/16] 2024-01-23 02:04:23,365 (trainer:737) INFO: 4epoch:train:12401-12500batch: iter_time=2.886e-04, forward_time=0.337, loss_ctc=84.552, loss_att=81.919, acc=0.577, loss=82.709, backward_time=0.456, grad_norm=40.301, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.101, optim0_lr0=4.618e-04, train_time=1.321
+[gpub052:0/16] 2024-01-23 02:04:43,638 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub052:0/16] 2024-01-23 02:05:03,695 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 02:05:07,835 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f591128b340>)
+[gpub052:0/16] 2024-01-23 02:05:07,835 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub052:0/16] 2024-01-23 02:05:07,838 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 02:12:17,563 (trainer:737) INFO: 4epoch:train:12501-12600batch: iter_time=3.466, forward_time=0.333, loss_ctc=87.561, loss_att=103.233, acc=0.547, loss=98.531, backward_time=0.408, grad_norm=43.080, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.094, optim0_lr0=4.633e-04, train_time=4.741
+[gpub052:0/16] 2024-01-23 02:14:31,372 (trainer:737) INFO: 4epoch:train:12601-12700batch: iter_time=8.750e-05, forward_time=0.369, loss_ctc=90.580, loss_att=92.173, acc=0.560, loss=91.695, backward_time=0.505, grad_norm=49.192, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.104, optim0_lr0=4.648e-04, train_time=1.338
+[gpub052:0/16] 2024-01-23 02:17:10,620 (trainer:737) INFO: 4epoch:train:12701-12800batch: iter_time=9.257e-05, forward_time=0.347, loss_ctc=87.738, loss_att=97.826, acc=0.519, loss=94.800, backward_time=0.460, grad_norm=42.790, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.098, optim0_lr0=4.663e-04, train_time=1.592
+[gpub052:0/16] 2024-01-23 02:19:21,029 (trainer:737) INFO: 4epoch:train:12801-12900batch: iter_time=9.124e-05, forward_time=0.362, loss_ctc=88.650, loss_att=89.781, acc=0.572, loss=89.442, backward_time=0.456, grad_norm=45.283, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.098, optim0_lr0=4.678e-04, train_time=1.304
+[gpub052:0/16] 2024-01-23 02:21:24,409 (trainer:737) INFO: 4epoch:train:12901-13000batch: iter_time=9.069e-05, forward_time=0.333, loss_ctc=80.907, loss_att=79.333, acc=0.563, loss=79.805, backward_time=0.473, grad_norm=39.611, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.100, optim0_lr0=4.693e-04, train_time=1.234
+[gpub052:0/16] 2024-01-23 02:24:04,408 (trainer:737) INFO: 4epoch:train:13001-13100batch: iter_time=8.777e-05, forward_time=0.337, loss_ctc=108.535, loss_att=115.590, acc=0.518, loss=113.473, backward_time=0.432, grad_norm=58.530, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.096, optim0_lr0=4.708e-04, train_time=1.600
+[gpub052:0/16] 2024-01-23 02:26:18,104 (trainer:737) INFO: 4epoch:train:13101-13200batch: iter_time=3.476e-04, forward_time=0.413, loss_ctc=90.315, loss_att=84.062, acc=0.549, loss=85.938, backward_time=0.461, grad_norm=41.655, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.100, optim0_lr0=4.723e-04, train_time=1.336
+[gpub052:0/16] 2024-01-23 02:28:40,391 (trainer:737) INFO: 4epoch:train:13201-13300batch: iter_time=8.341e-04, forward_time=0.388, loss_ctc=87.435, loss_att=97.041, acc=0.535, loss=94.159, backward_time=0.447, grad_norm=42.335, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.097, optim0_lr0=4.738e-04, train_time=1.424
+[gpub052:0/16] 2024-01-23 02:31:09,298 (trainer:737) INFO: 4epoch:train:13301-13400batch: iter_time=9.005e-05, forward_time=0.334, loss_ctc=103.023, loss_att=110.382, acc=0.547, loss=108.174, backward_time=0.406, grad_norm=46.410, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.094, optim0_lr0=4.753e-04, train_time=1.487
+[gpub052:0/16] 2024-01-23 02:33:26,737 (trainer:737) INFO: 4epoch:train:13401-13500batch: iter_time=3.277e-04, forward_time=0.396, loss_ctc=80.133, loss_att=87.581, acc=0.580, loss=85.347, backward_time=0.438, grad_norm=38.797, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.098, optim0_lr0=4.768e-04, train_time=1.375
+[gpub052:0/16] 2024-01-23 02:35:32,530 (trainer:737) INFO: 4epoch:train:13501-13600batch: iter_time=2.207e-04, forward_time=0.384, loss_ctc=83.451, loss_att=74.105, acc=0.577, loss=76.909, backward_time=0.434, grad_norm=42.986, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.101, optim0_lr0=4.783e-04, train_time=1.256
+[gpub052:0/16] 2024-01-23 02:38:15,564 (trainer:737) INFO: 4epoch:train:13601-13700batch: iter_time=4.028e-04, forward_time=0.364, loss_ctc=96.137, loss_att=86.721, acc=0.557, loss=89.546, backward_time=0.464, grad_norm=47.183, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.105, optim0_lr0=4.798e-04, train_time=1.632
+[gpub052:0/16] 2024-01-23 02:39:37,547 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub052:0/16] 2024-01-23 02:39:58,223 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 02:40:02,346 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58d273b5e0>)
+[gpub052:0/16] 2024-01-23 02:40:02,346 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub052:0/16] 2024-01-23 02:40:02,350 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 02:46:12,800 (trainer:737) INFO: 4epoch:train:13701-13800batch: iter_time=3.383, forward_time=0.450, loss_ctc=81.573, loss_att=91.915, acc=0.555, loss=88.812, backward_time=0.444, grad_norm=40.145, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.102, optim0_lr0=4.813e-04, train_time=4.771
+[gpub052:0/16] 2024-01-23 02:49:11,655 (trainer:737) INFO: 4epoch:train:13801-13900batch: iter_time=8.766e-05, forward_time=0.516, loss_ctc=91.227, loss_att=94.890, acc=0.566, loss=93.791, backward_time=0.478, grad_norm=46.014, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.113, optim0_lr0=4.828e-04, train_time=1.789
+[gpub052:0/16] 2024-01-23 02:51:59,203 (trainer:737) INFO: 4epoch:train:13901-14000batch: iter_time=3.368e-04, forward_time=0.452, loss_ctc=85.831, loss_att=101.515, acc=0.528, loss=96.809, backward_time=0.517, grad_norm=46.916, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.119, optim0_lr0=4.843e-04, train_time=1.675
+[gpub052:0/16] 2024-01-23 02:54:08,238 (trainer:737) INFO: 4epoch:train:14001-14100batch: iter_time=3.188e-04, forward_time=0.384, loss_ctc=79.802, loss_att=86.091, acc=0.539, loss=84.204, backward_time=0.443, grad_norm=40.048, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.106, optim0_lr0=4.858e-04, train_time=1.291
+[gpub052:0/16] 2024-01-23 02:56:52,007 (trainer:737) INFO: 4epoch:train:14101-14200batch: iter_time=8.703e-05, forward_time=0.352, loss_ctc=88.946, loss_att=82.017, acc=0.574, loss=84.096, backward_time=0.414, grad_norm=43.309, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.098, optim0_lr0=4.873e-04, train_time=1.637
+[gpub052:0/16] 2024-01-23 02:59:12,146 (trainer:737) INFO: 4epoch:train:14201-14300batch: iter_time=9.241e-04, forward_time=0.415, loss_ctc=85.470, loss_att=84.062, acc=0.535, loss=84.484, backward_time=0.487, grad_norm=44.762, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.114, optim0_lr0=4.888e-04, train_time=1.401
+[gpub052:0/16] 2024-01-23 03:01:17,668 (trainer:737) INFO: 4epoch:train:14301-14400batch: iter_time=8.424e-05, forward_time=0.366, loss_ctc=101.489, loss_att=105.491, acc=0.520, loss=104.290, backward_time=0.436, grad_norm=53.650, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.095, optim0_lr0=4.903e-04, train_time=1.255
+[gpub052:0/16] 2024-01-23 03:04:02,297 (trainer:737) INFO: 4epoch:train:14401-14500batch: iter_time=3.684e-04, forward_time=0.426, loss_ctc=93.228, loss_att=86.269, acc=0.549, loss=88.356, backward_time=0.445, grad_norm=44.661, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.099, optim0_lr0=4.918e-04, train_time=1.646
+[gpub052:0/16] 2024-01-23 03:06:27,507 (trainer:737) INFO: 4epoch:train:14501-14600batch: iter_time=6.367e-04, forward_time=0.426, loss_ctc=100.877, loss_att=113.739, acc=0.527, loss=109.880, backward_time=0.458, grad_norm=45.236, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.102, optim0_lr0=4.933e-04, train_time=1.452
+[gpub052:0/16] 2024-01-23 03:08:40,739 (trainer:737) INFO: 4epoch:train:14601-14700batch: iter_time=3.091e-04, forward_time=0.360, loss_ctc=78.817, loss_att=82.480, acc=0.551, loss=81.381, backward_time=0.465, grad_norm=38.789, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.103, optim0_lr0=4.948e-04, train_time=1.333
+[gpub052:0/16] 2024-01-23 03:11:01,395 (trainer:737) INFO: 4epoch:train:14701-14800batch: iter_time=8.456e-05, forward_time=0.305, loss_ctc=77.619, loss_att=84.948, acc=0.566, loss=82.749, backward_time=0.415, grad_norm=38.742, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.095, optim0_lr0=4.963e-04, train_time=1.405
+[gpub052:0/16] 2024-01-23 03:13:25,432 (trainer:737) INFO: 4epoch:train:14801-14900batch: iter_time=3.351e-04, forward_time=0.401, loss_ctc=93.715, loss_att=79.288, acc=0.571, loss=83.616, backward_time=0.437, grad_norm=49.759, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.104, optim0_lr0=4.978e-04, train_time=1.441
+[gpub052:0/16] 2024-01-23 03:15:44,765 (trainer:737) INFO: 4epoch:train:14901-15000batch: iter_time=2.645e-04, forward_time=0.380, loss_ctc=82.589, loss_att=79.064, acc=0.584, loss=80.121, backward_time=0.437, grad_norm=40.893, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.105, optim0_lr0=4.993e-04, train_time=1.392
+[gpub052:0/16] 2024-01-23 03:53:45,497 (trainer:343) INFO: 4epoch results: [train] iter_time=0.405, forward_time=0.374, loss_ctc=95.570, loss_att=102.883, acc=0.513, loss=100.689, backward_time=0.441, grad_norm=48.726, clip=100.000, loss_scale=9.328e+12, optim_step_time=0.102, optim0_lr0=3.875e-04, train_time=2.057, time=8 hours, 34 minutes and 45.4 seconds, total_count=60000, gpu_max_cached_mem_GB=40.771, [valid] loss_ctc=91.410, cer_ctc=0.434, loss_att=79.362, acc=0.457, cer=0.417, wer=1.000, loss=82.976, time=37 minutes and 36.86 seconds, total_count=18684, gpu_max_cached_mem_GB=40.771
+[gpub052:0/16] 2024-01-23 03:53:54,925 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub052:0/16] 2024-01-23 03:53:54,969 (trainer:272) INFO: 5/45epoch started. Estimated time to finish: 2 weeks, 4 days and 15 hours
+[gpub052:0/16] 2024-01-23 03:53:54,979 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub052:0/16] 2024-01-23 03:54:13,627 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 03:54:17,014 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f51656ccc70>)
+[gpub052:0/16] 2024-01-23 03:54:17,014 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub052:0/16] 2024-01-23 03:54:17,017 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 04:00:52,948 (trainer:737) INFO: 5epoch:train:1-100batch: iter_time=2.965, forward_time=0.355, loss_ctc=106.744, loss_att=106.730, acc=0.531, loss=106.734, backward_time=0.414, grad_norm=57.903, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.096, optim0_lr0=4.998e-04, train_time=4.179
+[gpub052:0/16] 2024-01-23 04:03:12,073 (trainer:737) INFO: 5epoch:train:101-200batch: iter_time=4.598e-04, forward_time=0.397, loss_ctc=99.898, loss_att=104.771, acc=0.534, loss=103.309, backward_time=0.462, grad_norm=51.391, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.098, optim0_lr0=4.994e-04, train_time=1.392
+[gpub052:0/16] 2024-01-23 04:05:32,306 (trainer:737) INFO: 5epoch:train:201-300batch: iter_time=8.064e-05, forward_time=0.292, loss_ctc=84.953, loss_att=79.915, acc=0.557, loss=81.426, backward_time=0.401, grad_norm=48.799, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.095, optim0_lr0=4.990e-04, train_time=1.402
+[gpub052:0/16] 2024-01-23 04:07:49,275 (trainer:737) INFO: 5epoch:train:301-400batch: iter_time=8.155e-05, forward_time=0.414, loss_ctc=96.488, loss_att=99.226, acc=0.558, loss=98.405, backward_time=0.429, grad_norm=45.699, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.101, optim0_lr0=4.985e-04, train_time=1.370
+[gpub052:0/16] 2024-01-23 04:10:15,663 (trainer:737) INFO: 5epoch:train:401-500batch: iter_time=6.097e-04, forward_time=0.299, loss_ctc=88.470, loss_att=87.246, acc=0.565, loss=87.613, backward_time=0.406, grad_norm=43.560, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.098, optim0_lr0=4.981e-04, train_time=1.463
+[gpub052:0/16] 2024-01-23 04:12:27,056 (trainer:737) INFO: 5epoch:train:501-600batch: iter_time=8.321e-05, forward_time=0.375, loss_ctc=76.126, loss_att=80.102, acc=0.560, loss=78.909, backward_time=0.439, grad_norm=40.891, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.101, optim0_lr0=4.977e-04, train_time=1.314
+[gpub052:0/16] 2024-01-23 04:14:42,208 (trainer:737) INFO: 5epoch:train:601-700batch: iter_time=4.384e-04, forward_time=0.322, loss_ctc=91.302, loss_att=102.695, acc=0.550, loss=99.277, backward_time=0.413, grad_norm=42.201, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.096, optim0_lr0=4.973e-04, train_time=1.350
+[gpub052:0/16] 2024-01-23 04:17:05,706 (trainer:737) INFO: 5epoch:train:701-800batch: iter_time=8.076e-05, forward_time=0.388, loss_ctc=100.989, loss_att=92.581, acc=0.541, loss=95.104, backward_time=0.419, grad_norm=46.487, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.097, optim0_lr0=4.969e-04, train_time=1.436
+[gpub052:0/16] 2024-01-23 04:19:18,262 (trainer:737) INFO: 5epoch:train:801-900batch: iter_time=4.124e-04, forward_time=0.325, loss_ctc=88.392, loss_att=87.127, acc=0.581, loss=87.507, backward_time=0.432, grad_norm=39.209, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.096, optim0_lr0=4.965e-04, train_time=1.325
+[gpub052:0/16] 2024-01-23 04:21:34,509 (trainer:737) INFO: 5epoch:train:901-1000batch: iter_time=8.386e-05, forward_time=0.365, loss_ctc=84.610, loss_att=90.723, acc=0.531, loss=88.889, backward_time=0.451, grad_norm=42.970, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.099, optim0_lr0=4.961e-04, train_time=1.362
+[gpub052:0/16] 2024-01-23 04:24:09,166 (trainer:737) INFO: 5epoch:train:1001-1100batch: iter_time=3.230e-04, forward_time=0.346, loss_ctc=88.873, loss_att=96.819, acc=0.543, loss=94.435, backward_time=0.474, grad_norm=43.160, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.098, optim0_lr0=4.957e-04, train_time=1.545
+[gpub052:0/16] 2024-01-23 04:26:12,153 (trainer:737) INFO: 5epoch:train:1101-1200batch: iter_time=7.936e-05, forward_time=0.294, loss_ctc=97.265, loss_att=103.452, acc=0.555, loss=101.596, backward_time=0.406, grad_norm=47.344, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.094, optim0_lr0=4.953e-04, train_time=1.231
+[gpub052:0/16] 2024-01-23 04:27:43,879 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub052:0/16] 2024-01-23 04:28:03,561 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 04:28:07,324 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58d5a2ff40>)
+[gpub052:0/16] 2024-01-23 04:28:07,324 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub052:0/16] 2024-01-23 04:28:07,328 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 04:34:08,545 (trainer:737) INFO: 5epoch:train:1201-1300batch: iter_time=3.323, forward_time=0.392, loss_ctc=104.861, loss_att=93.717, acc=0.560, loss=97.060, backward_time=0.428, grad_norm=51.442, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.098, optim0_lr0=4.949e-04, train_time=4.764
+[gpub052:0/16] 2024-01-23 04:36:14,142 (trainer:737) INFO: 5epoch:train:1301-1400batch: iter_time=7.910e-05, forward_time=0.294, loss_ctc=99.983, loss_att=100.005, acc=0.551, loss=99.998, backward_time=0.408, grad_norm=45.852, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.094, optim0_lr0=4.945e-04, train_time=1.256
+[gpub052:0/16] 2024-01-23 04:38:28,657 (trainer:737) INFO: 5epoch:train:1401-1500batch: iter_time=7.874e-05, forward_time=0.357, loss_ctc=73.937, loss_att=74.137, acc=0.571, loss=74.077, backward_time=0.456, grad_norm=37.153, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.103, optim0_lr0=4.941e-04, train_time=1.345
+[gpub052:0/16] 2024-01-23 04:40:39,221 (trainer:737) INFO: 5epoch:train:1501-1600batch: iter_time=7.831e-05, forward_time=0.295, loss_ctc=97.529, loss_att=93.171, acc=0.565, loss=94.478, backward_time=0.407, grad_norm=51.106, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.094, optim0_lr0=4.937e-04, train_time=1.306
+[gpub052:0/16] 2024-01-23 04:43:06,421 (trainer:737) INFO: 5epoch:train:1601-1700batch: iter_time=2.622e-04, forward_time=0.436, loss_ctc=87.552, loss_att=91.521, acc=0.584, loss=90.330, backward_time=0.435, grad_norm=40.140, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.107, optim0_lr0=4.933e-04, train_time=1.472
+[gpub052:0/16] 2024-01-23 04:45:09,170 (trainer:737) INFO: 5epoch:train:1701-1800batch: iter_time=7.955e-05, forward_time=0.289, loss_ctc=74.869, loss_att=78.914, acc=0.564, loss=77.701, backward_time=0.402, grad_norm=34.971, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.094, optim0_lr0=4.929e-04, train_time=1.227
+[gpub052:0/16] 2024-01-23 04:47:18,144 (trainer:737) INFO: 5epoch:train:1801-1900batch: iter_time=2.106e-04, forward_time=0.294, loss_ctc=80.179, loss_att=77.458, acc=0.584, loss=78.275, backward_time=0.406, grad_norm=39.389, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.095, optim0_lr0=4.925e-04, train_time=1.290
+[gpub052:0/16] 2024-01-23 04:49:27,130 (trainer:737) INFO: 5epoch:train:1901-2000batch: iter_time=8.808e-05, forward_time=0.395, loss_ctc=103.888, loss_att=107.047, acc=0.530, loss=106.100, backward_time=0.440, grad_norm=46.325, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.098, optim0_lr0=4.921e-04, train_time=1.289
+[gpub052:0/16] 2024-01-23 04:51:34,965 (trainer:737) INFO: 5epoch:train:2001-2100batch: iter_time=8.284e-05, forward_time=0.292, loss_ctc=83.518, loss_att=88.103, acc=0.580, loss=86.727, backward_time=0.405, grad_norm=37.456, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.917e-04, train_time=1.278
+[gpub052:0/16] 2024-01-23 04:53:55,439 (trainer:737) INFO: 5epoch:train:2101-2200batch: iter_time=8.278e-05, forward_time=0.347, loss_ctc=84.912, loss_att=79.054, acc=0.594, loss=80.812, backward_time=0.450, grad_norm=37.137, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.099, optim0_lr0=4.913e-04, train_time=1.405
+[gpub052:0/16] 2024-01-23 04:55:54,485 (trainer:737) INFO: 5epoch:train:2201-2300batch: iter_time=8.121e-05, forward_time=0.290, loss_ctc=77.555, loss_att=86.071, acc=0.528, loss=83.516, backward_time=0.402, grad_norm=39.969, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.909e-04, train_time=1.189
+[gpub052:0/16] 2024-01-23 04:58:02,441 (trainer:737) INFO: 5epoch:train:2301-2400batch: iter_time=8.540e-05, forward_time=0.299, loss_ctc=87.343, loss_att=103.261, acc=0.568, loss=98.485, backward_time=0.410, grad_norm=38.107, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.905e-04, train_time=1.279
+[gpub052:0/16] 2024-01-23 05:00:32,257 (trainer:737) INFO: 5epoch:train:2401-2500batch: iter_time=9.105e-05, forward_time=0.390, loss_ctc=101.952, loss_att=91.430, acc=0.567, loss=94.587, backward_time=0.440, grad_norm=48.432, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.099, optim0_lr0=4.901e-04, train_time=1.499
+[gpub052:0/16] 2024-01-23 05:00:52,408 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub052:0/16] 2024-01-23 05:01:12,159 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 05:01:15,990 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58cd6e7f40>)
+[gpub052:0/16] 2024-01-23 05:01:15,990 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub052:0/16] 2024-01-23 05:01:15,993 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 05:07:52,519 (trainer:737) INFO: 5epoch:train:2501-2600batch: iter_time=3.160, forward_time=0.298, loss_ctc=97.833, loss_att=94.359, acc=0.557, loss=95.401, backward_time=0.408, grad_norm=47.020, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.095, optim0_lr0=4.897e-04, train_time=4.402
+[gpub052:0/16] 2024-01-23 05:10:09,389 (trainer:737) INFO: 5epoch:train:2601-2700batch: iter_time=9.440e-05, forward_time=0.415, loss_ctc=92.719, loss_att=96.603, acc=0.552, loss=95.438, backward_time=0.432, grad_norm=43.740, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.097, optim0_lr0=4.893e-04, train_time=1.368
+[gpub052:0/16] 2024-01-23 05:12:35,116 (trainer:737) INFO: 5epoch:train:2701-2800batch: iter_time=9.032e-05, forward_time=0.327, loss_ctc=79.102, loss_att=72.986, acc=0.586, loss=74.821, backward_time=0.454, grad_norm=41.362, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.099, optim0_lr0=4.889e-04, train_time=1.457
+[gpub052:0/16] 2024-01-23 05:14:36,163 (trainer:737) INFO: 5epoch:train:2801-2900batch: iter_time=8.446e-05, forward_time=0.312, loss_ctc=89.596, loss_att=91.159, acc=0.578, loss=90.690, backward_time=0.416, grad_norm=37.582, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.095, optim0_lr0=4.885e-04, train_time=1.211
+[gpub052:0/16] 2024-01-23 05:17:01,843 (trainer:737) INFO: 5epoch:train:2901-3000batch: iter_time=8.938e-05, forward_time=0.382, loss_ctc=83.350, loss_att=80.628, acc=0.583, loss=81.445, backward_time=0.457, grad_norm=38.646, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.103, optim0_lr0=4.881e-04, train_time=1.456
+[gpub052:0/16] 2024-01-23 05:19:01,621 (trainer:737) INFO: 5epoch:train:3001-3100batch: iter_time=9.013e-05, forward_time=0.292, loss_ctc=71.722, loss_att=76.641, acc=0.568, loss=75.165, backward_time=0.403, grad_norm=35.349, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.878e-04, train_time=1.197
+[gpub052:0/16] 2024-01-23 05:21:34,187 (trainer:737) INFO: 5epoch:train:3101-3200batch: iter_time=9.474e-05, forward_time=0.377, loss_ctc=85.303, loss_att=95.628, acc=0.565, loss=92.530, backward_time=0.439, grad_norm=39.764, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.101, optim0_lr0=4.874e-04, train_time=1.525
+[gpub052:0/16] 2024-01-23 05:23:39,101 (trainer:737) INFO: 5epoch:train:3201-3300batch: iter_time=7.913e-05, forward_time=0.291, loss_ctc=92.585, loss_att=85.120, acc=0.550, loss=87.359, backward_time=0.404, grad_norm=43.958, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.870e-04, train_time=1.249
+[gpub052:0/16] 2024-01-23 05:26:06,440 (trainer:737) INFO: 5epoch:train:3301-3400batch: iter_time=2.604e-04, forward_time=0.417, loss_ctc=84.116, loss_att=82.639, acc=0.592, loss=83.082, backward_time=0.439, grad_norm=37.032, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.102, optim0_lr0=4.866e-04, train_time=1.473
+[gpub052:0/16] 2024-01-23 05:28:18,721 (trainer:737) INFO: 5epoch:train:3401-3500batch: iter_time=8.664e-05, forward_time=0.292, loss_ctc=78.270, loss_att=85.644, acc=0.542, loss=83.432, backward_time=0.404, grad_norm=37.920, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.862e-04, train_time=1.322
+[gpub052:0/16] 2024-01-23 05:30:44,919 (trainer:737) INFO: 5epoch:train:3501-3600batch: iter_time=3.651e-04, forward_time=0.351, loss_ctc=83.851, loss_att=92.462, acc=0.553, loss=89.879, backward_time=0.467, grad_norm=40.505, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.100, optim0_lr0=4.858e-04, train_time=1.462
+[gpub052:0/16] 2024-01-23 05:32:43,250 (trainer:737) INFO: 5epoch:train:3601-3700batch: iter_time=8.349e-05, forward_time=0.301, loss_ctc=90.125, loss_att=97.798, acc=0.562, loss=95.496, backward_time=0.409, grad_norm=41.843, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.094, optim0_lr0=4.854e-04, train_time=1.183
+[gpub052:0/16] 2024-01-23 05:34:17,154 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub052:0/16] 2024-01-23 05:34:36,941 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 05:34:40,922 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5a842637c0>)
+[gpub052:0/16] 2024-01-23 05:34:40,922 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub052:0/16] 2024-01-23 05:34:40,925 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 05:40:27,476 (trainer:737) INFO: 5epoch:train:3701-3800batch: iter_time=3.227, forward_time=0.385, loss_ctc=97.425, loss_att=84.934, acc=0.575, loss=88.681, backward_time=0.450, grad_norm=49.073, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.100, optim0_lr0=4.851e-04, train_time=4.642
+[gpub052:0/16] 2024-01-23 05:42:31,343 (trainer:737) INFO: 5epoch:train:3801-3900batch: iter_time=8.202e-05, forward_time=0.340, loss_ctc=96.973, loss_att=94.464, acc=0.561, loss=95.217, backward_time=0.440, grad_norm=46.986, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.102, optim0_lr0=4.847e-04, train_time=1.239
+[gpub052:0/16] 2024-01-23 05:44:46,560 (trainer:737) INFO: 5epoch:train:3901-4000batch: iter_time=7.774e-05, forward_time=0.312, loss_ctc=71.086, loss_att=70.956, acc=0.583, loss=70.995, backward_time=0.409, grad_norm=36.306, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.095, optim0_lr0=4.843e-04, train_time=1.351
+[gpub052:0/16] 2024-01-23 05:47:17,080 (trainer:737) INFO: 5epoch:train:4001-4100batch: iter_time=4.628e-04, forward_time=0.396, loss_ctc=93.520, loss_att=87.278, acc=0.583, loss=89.151, backward_time=0.439, grad_norm=42.964, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.099, optim0_lr0=4.839e-04, train_time=1.505
+[gpub052:0/16] 2024-01-23 05:49:29,142 (trainer:737) INFO: 5epoch:train:4101-4200batch: iter_time=7.986e-05, forward_time=0.293, loss_ctc=83.909, loss_att=87.440, acc=0.593, loss=86.380, backward_time=0.407, grad_norm=38.368, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.094, optim0_lr0=4.836e-04, train_time=1.320
+[gpub052:0/16] 2024-01-23 05:51:47,780 (trainer:737) INFO: 5epoch:train:4201-4300batch: iter_time=8.569e-05, forward_time=0.357, loss_ctc=71.685, loss_att=74.896, acc=0.579, loss=73.932, backward_time=0.474, grad_norm=33.118, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.100, optim0_lr0=4.832e-04, train_time=1.387
+[gpub052:0/16] 2024-01-23 05:54:07,249 (trainer:737) INFO: 5epoch:train:4301-4400batch: iter_time=8.543e-05, forward_time=0.290, loss_ctc=75.976, loss_att=74.111, acc=0.591, loss=74.670, backward_time=0.401, grad_norm=35.145, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.094, optim0_lr0=4.828e-04, train_time=1.394
+[gpub052:0/16] 2024-01-23 05:56:31,944 (trainer:737) INFO: 5epoch:train:4401-4500batch: iter_time=6.322e-04, forward_time=0.416, loss_ctc=99.352, loss_att=102.343, acc=0.532, loss=101.446, backward_time=0.454, grad_norm=44.548, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.112, optim0_lr0=4.824e-04, train_time=1.448
+[gpub052:0/16] 2024-01-23 05:58:46,661 (trainer:737) INFO: 5epoch:train:4501-4600batch: iter_time=8.911e-05, forward_time=0.301, loss_ctc=79.740, loss_att=83.103, acc=0.592, loss=82.094, backward_time=0.404, grad_norm=58.053, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.094, optim0_lr0=4.821e-04, train_time=1.347
+[gpub052:0/16] 2024-01-23 06:01:29,484 (trainer:737) INFO: 5epoch:train:4601-4700batch: iter_time=4.116e-04, forward_time=0.355, loss_ctc=81.712, loss_att=75.875, acc=0.599, loss=77.626, backward_time=0.471, grad_norm=36.768, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.098, optim0_lr0=4.817e-04, train_time=1.627
+[gpub052:0/16] 2024-01-23 06:03:36,283 (trainer:737) INFO: 5epoch:train:4701-4800batch: iter_time=3.136e-04, forward_time=0.358, loss_ctc=73.497, loss_att=83.665, acc=0.533, loss=80.615, backward_time=0.435, grad_norm=37.328, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.098, optim0_lr0=4.813e-04, train_time=1.268
+[gpub052:0/16] 2024-01-23 06:05:36,104 (trainer:737) INFO: 5epoch:train:4801-4900batch: iter_time=8.620e-05, forward_time=0.347, loss_ctc=83.966, loss_att=98.320, acc=0.580, loss=94.014, backward_time=0.422, grad_norm=36.589, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.096, optim0_lr0=4.809e-04, train_time=1.197
+[gpub052:0/16] 2024-01-23 06:08:12,992 (trainer:737) INFO: 5epoch:train:4901-5000batch: iter_time=3.281e-04, forward_time=0.400, loss_ctc=96.116, loss_att=86.782, acc=0.568, loss=89.582, backward_time=0.446, grad_norm=46.436, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.098, optim0_lr0=4.806e-04, train_time=1.569
+[gpub052:0/16] 2024-01-23 06:08:33,044 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub052:0/16] 2024-01-23 06:08:53,017 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 06:08:56,685 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5ba36ef640>)
+[gpub052:0/16] 2024-01-23 06:08:56,685 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub052:0/16] 2024-01-23 06:08:56,688 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 06:15:12,951 (trainer:737) INFO: 5epoch:train:5001-5100batch: iter_time=3.069, forward_time=0.306, loss_ctc=93.376, loss_att=92.165, acc=0.575, loss=92.528, backward_time=0.410, grad_norm=43.936, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.096, optim0_lr0=4.802e-04, train_time=4.200
+[gpub052:0/16] 2024-01-23 06:17:56,264 (trainer:737) INFO: 5epoch:train:5101-5200batch: iter_time=8.352e-05, forward_time=0.354, loss_ctc=87.314, loss_att=90.910, acc=0.582, loss=89.831, backward_time=0.466, grad_norm=40.795, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.101, optim0_lr0=4.798e-04, train_time=1.632
+[gpub052:0/16] 2024-01-23 06:19:48,441 (trainer:737) INFO: 5epoch:train:5201-5300batch: iter_time=7.975e-05, forward_time=0.289, loss_ctc=74.984, loss_att=69.837, acc=0.602, loss=71.381, backward_time=0.402, grad_norm=38.621, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.094, optim0_lr0=4.795e-04, train_time=1.120
+[gpub052:0/16] 2024-01-23 06:21:58,692 (trainer:737) INFO: 5epoch:train:5301-5400batch: iter_time=8.610e-05, forward_time=0.379, loss_ctc=86.173, loss_att=87.456, acc=0.597, loss=87.071, backward_time=0.436, grad_norm=35.809, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.114, optim0_lr0=4.791e-04, train_time=1.303
+[gpub052:0/16] 2024-01-23 06:24:30,470 (trainer:737) INFO: 5epoch:train:5401-5500batch: iter_time=8.153e-05, forward_time=0.292, loss_ctc=80.083, loss_att=78.165, acc=0.601, loss=78.741, backward_time=0.403, grad_norm=37.432, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.094, optim0_lr0=4.787e-04, train_time=1.516
+[gpub052:0/16] 2024-01-23 06:26:36,920 (trainer:737) INFO: 5epoch:train:5501-5600batch: iter_time=8.387e-05, forward_time=0.374, loss_ctc=68.924, loss_att=72.272, acc=0.592, loss=71.268, backward_time=0.458, grad_norm=34.084, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.101, optim0_lr0=4.784e-04, train_time=1.265
+[gpub052:0/16] 2024-01-23 06:28:36,582 (trainer:737) INFO: 5epoch:train:5601-5700batch: iter_time=8.490e-05, forward_time=0.301, loss_ctc=81.915, loss_att=93.046, acc=0.587, loss=89.707, backward_time=0.432, grad_norm=35.692, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.096, optim0_lr0=4.780e-04, train_time=1.197
+[gpub052:0/16] 2024-01-23 06:31:17,806 (trainer:737) INFO: 5epoch:train:5701-5800batch: iter_time=0.011, forward_time=0.372, loss_ctc=91.011, loss_att=82.664, acc=0.575, loss=85.168, backward_time=0.422, grad_norm=42.096, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.098, optim0_lr0=4.776e-04, train_time=1.612
+[gpub052:0/16] 2024-01-23 06:33:26,847 (trainer:737) INFO: 5epoch:train:5801-5900batch: iter_time=3.111e-04, forward_time=0.403, loss_ctc=79.631, loss_att=78.329, acc=0.615, loss=78.719, backward_time=0.444, grad_norm=32.524, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.100, optim0_lr0=4.773e-04, train_time=1.291
+[gpub052:0/16] 2024-01-23 06:36:00,234 (trainer:737) INFO: 5epoch:train:5901-6000batch: iter_time=0.024, forward_time=0.307, loss_ctc=75.729, loss_att=81.714, acc=0.568, loss=79.919, backward_time=0.409, grad_norm=35.740, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.095, optim0_lr0=4.769e-04, train_time=1.528
+[gpub052:0/16] 2024-01-23 06:39:09,455 (trainer:737) INFO: 5epoch:train:6001-6100batch: iter_time=3.455e-04, forward_time=0.398, loss_ctc=80.926, loss_att=89.046, acc=0.576, loss=86.610, backward_time=0.445, grad_norm=38.169, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.102, optim0_lr0=4.765e-04, train_time=1.897
+[gpub052:0/16] 2024-01-23 06:41:36,970 (trainer:737) INFO: 5epoch:train:6101-6200batch: iter_time=7.852e-05, forward_time=0.293, loss_ctc=86.412, loss_att=93.364, acc=0.585, loss=91.278, backward_time=0.407, grad_norm=39.896, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.094, optim0_lr0=4.762e-04, train_time=1.474
+[gpub052:0/16] 2024-01-23 06:42:56,960 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub052:0/16] 2024-01-23 06:43:26,057 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 06:43:30,951 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f588e253820>)
+[gpub052:0/16] 2024-01-23 06:43:30,951 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub052:0/16] 2024-01-23 06:43:31,073 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 06:49:19,343 (trainer:737) INFO: 5epoch:train:6201-6300batch: iter_time=3.357, forward_time=0.393, loss_ctc=93.571, loss_att=80.850, acc=0.596, loss=84.666, backward_time=0.418, grad_norm=45.823, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.099, optim0_lr0=4.758e-04, train_time=4.624
+[gpub052:0/16] 2024-01-23 06:52:08,493 (trainer:737) INFO: 5epoch:train:6301-6400batch: iter_time=0.088, forward_time=0.298, loss_ctc=92.719, loss_att=91.761, acc=0.575, loss=92.049, backward_time=0.424, grad_norm=39.921, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.094, optim0_lr0=4.755e-04, train_time=1.691
+[gpub052:0/16] 2024-01-23 06:54:32,569 (trainer:737) INFO: 5epoch:train:6401-6500batch: iter_time=8.059e-05, forward_time=0.352, loss_ctc=67.920, loss_att=67.494, acc=0.602, loss=67.621, backward_time=0.442, grad_norm=35.620, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.104, optim0_lr0=4.751e-04, train_time=1.441
+[gpub052:0/16] 2024-01-23 06:56:31,462 (trainer:737) INFO: 5epoch:train:6501-6600batch: iter_time=8.558e-05, forward_time=0.302, loss_ctc=89.740, loss_att=83.567, acc=0.597, loss=85.419, backward_time=0.419, grad_norm=39.745, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.095, optim0_lr0=4.748e-04, train_time=1.189
+[gpub052:0/16] 2024-01-23 06:58:55,010 (trainer:737) INFO: 5epoch:train:6601-6700batch: iter_time=8.289e-05, forward_time=0.420, loss_ctc=81.990, loss_att=85.452, acc=0.605, loss=84.413, backward_time=0.428, grad_norm=36.021, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.098, optim0_lr0=4.744e-04, train_time=1.435
+[gpub052:0/16] 2024-01-23 07:01:10,347 (trainer:737) INFO: 5epoch:train:6701-6800batch: iter_time=8.102e-05, forward_time=0.319, loss_ctc=68.825, loss_att=72.456, acc=0.593, loss=71.366, backward_time=0.417, grad_norm=31.143, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.099, optim0_lr0=4.740e-04, train_time=1.353
+[gpub052:0/16] 2024-01-23 07:03:20,013 (trainer:737) INFO: 5epoch:train:6801-6900batch: iter_time=8.461e-05, forward_time=0.337, loss_ctc=73.103, loss_att=71.436, acc=0.605, loss=71.936, backward_time=0.420, grad_norm=33.478, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.099, optim0_lr0=4.737e-04, train_time=1.297
+[gpub052:0/16] 2024-01-23 07:05:41,503 (trainer:737) INFO: 5epoch:train:6901-7000batch: iter_time=3.433e-04, forward_time=0.381, loss_ctc=96.311, loss_att=98.349, acc=0.546, loss=97.738, backward_time=0.466, grad_norm=45.446, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.104, optim0_lr0=4.733e-04, train_time=1.414
+[gpub052:0/16] 2024-01-23 07:07:34,886 (trainer:737) INFO: 5epoch:train:7001-7100batch: iter_time=7.818e-05, forward_time=0.292, loss_ctc=77.170, loss_att=80.023, acc=0.606, loss=79.167, backward_time=0.406, grad_norm=33.365, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.094, optim0_lr0=4.730e-04, train_time=1.134
+[gpub052:0/16] 2024-01-23 07:10:20,998 (trainer:737) INFO: 5epoch:train:7101-7200batch: iter_time=2.101e-04, forward_time=0.435, loss_ctc=76.807, loss_att=71.793, acc=0.614, loss=73.297, backward_time=0.435, grad_norm=32.774, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.099, optim0_lr0=4.726e-04, train_time=1.660
+[gpub052:0/16] 2024-01-23 07:12:18,247 (trainer:737) INFO: 5epoch:train:7201-7300batch: iter_time=8.269e-05, forward_time=0.295, loss_ctc=71.524, loss_att=81.163, acc=0.544, loss=78.271, backward_time=0.408, grad_norm=37.672, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.094, optim0_lr0=4.723e-04, train_time=1.173
+[gpub052:0/16] 2024-01-23 07:14:44,291 (trainer:737) INFO: 5epoch:train:7301-7400batch: iter_time=3.227e-04, forward_time=0.365, loss_ctc=80.686, loss_att=95.293, acc=0.593, loss=90.911, backward_time=0.463, grad_norm=34.596, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.100, optim0_lr0=4.719e-04, train_time=1.460
+[gpub052:0/16] 2024-01-23 07:17:19,232 (trainer:737) INFO: 5epoch:train:7401-7500batch: iter_time=6.368e-04, forward_time=0.320, loss_ctc=93.036, loss_att=83.548, acc=0.582, loss=86.395, backward_time=0.447, grad_norm=46.985, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.101, optim0_lr0=4.716e-04, train_time=1.549
+[gpub052:0/16] 2024-01-23 07:17:39,383 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub052:0/16] 2024-01-23 07:17:59,368 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 07:18:02,993 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58985ffc70>)
+[gpub052:0/16] 2024-01-23 07:18:02,993 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub052:0/16] 2024-01-23 07:18:02,996 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 07:24:26,636 (trainer:737) INFO: 5epoch:train:7501-7600batch: iter_time=3.046, forward_time=0.363, loss_ctc=90.899, loss_att=84.257, acc=0.593, loss=86.250, backward_time=0.416, grad_norm=42.488, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.095, optim0_lr0=4.712e-04, train_time=4.274
+[gpub052:0/16] 2024-01-23 07:27:03,349 (trainer:737) INFO: 5epoch:train:7601-7700batch: iter_time=7.722e-05, forward_time=0.364, loss_ctc=85.804, loss_att=85.981, acc=0.585, loss=85.928, backward_time=0.439, grad_norm=40.944, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.102, optim0_lr0=4.709e-04, train_time=1.567
+[gpub052:0/16] 2024-01-23 07:29:09,820 (trainer:737) INFO: 5epoch:train:7701-7800batch: iter_time=8.180e-05, forward_time=0.302, loss_ctc=72.096, loss_att=66.116, acc=0.617, loss=67.910, backward_time=0.405, grad_norm=36.144, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.096, optim0_lr0=4.705e-04, train_time=1.265
+[gpub052:0/16] 2024-01-23 07:31:20,450 (trainer:737) INFO: 5epoch:train:7801-7900batch: iter_time=8.227e-05, forward_time=0.361, loss_ctc=83.177, loss_att=83.413, acc=0.603, loss=83.342, backward_time=0.430, grad_norm=33.576, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.103, optim0_lr0=4.702e-04, train_time=1.306
+[gpub052:0/16] 2024-01-23 07:33:45,636 (trainer:737) INFO: 5epoch:train:7901-8000batch: iter_time=8.133e-05, forward_time=0.324, loss_ctc=77.196, loss_att=74.433, acc=0.612, loss=75.262, backward_time=0.433, grad_norm=37.338, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.105, optim0_lr0=4.698e-04, train_time=1.452
+[gpub052:0/16] 2024-01-23 07:35:49,354 (trainer:737) INFO: 5epoch:train:8001-8100batch: iter_time=1.809e-04, forward_time=0.372, loss_ctc=66.446, loss_att=69.413, acc=0.600, loss=68.522, backward_time=0.427, grad_norm=31.278, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.098, optim0_lr0=4.695e-04, train_time=1.237
+[gpub052:0/16] 2024-01-23 07:38:14,673 (trainer:737) INFO: 5epoch:train:8101-8200batch: iter_time=0.001, forward_time=0.370, loss_ctc=80.014, loss_att=89.537, acc=0.591, loss=86.680, backward_time=0.438, grad_norm=35.043, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.102, optim0_lr0=4.691e-04, train_time=1.452
+[gpub052:0/16] 2024-01-23 07:40:29,821 (trainer:737) INFO: 5epoch:train:8201-8300batch: iter_time=7.982e-05, forward_time=0.299, loss_ctc=86.869, loss_att=78.067, acc=0.577, loss=80.708, backward_time=0.412, grad_norm=40.715, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.095, optim0_lr0=4.688e-04, train_time=1.352
+[gpub052:0/16] 2024-01-23 07:42:46,350 (trainer:737) INFO: 5epoch:train:8301-8400batch: iter_time=9.146e-04, forward_time=0.343, loss_ctc=77.078, loss_att=75.045, acc=0.621, loss=75.655, backward_time=0.451, grad_norm=31.857, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.103, optim0_lr0=4.685e-04, train_time=1.363
+[gpub052:0/16] 2024-01-23 07:45:04,075 (trainer:737) INFO: 5epoch:train:8401-8500batch: iter_time=6.256e-04, forward_time=0.361, loss_ctc=73.274, loss_att=79.089, acc=0.567, loss=77.345, backward_time=0.444, grad_norm=34.757, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.101, optim0_lr0=4.681e-04, train_time=1.379
+[gpub052:0/16] 2024-01-23 07:47:19,964 (trainer:737) INFO: 5epoch:train:8501-8600batch: iter_time=8.712e-05, forward_time=0.328, loss_ctc=77.557, loss_att=85.212, acc=0.581, loss=82.916, backward_time=0.408, grad_norm=35.465, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.094, optim0_lr0=4.678e-04, train_time=1.359
+[gpub052:0/16] 2024-01-23 07:49:37,001 (trainer:737) INFO: 5epoch:train:8601-8700batch: iter_time=3.591e-04, forward_time=0.419, loss_ctc=83.521, loss_att=90.217, acc=0.592, loss=88.208, backward_time=0.430, grad_norm=38.781, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.098, optim0_lr0=4.674e-04, train_time=1.370
+[gpub052:0/16] 2024-01-23 07:50:53,987 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub052:0/16] 2024-01-23 07:51:13,822 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 07:51:17,440 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5b79356da0>)
+[gpub052:0/16] 2024-01-23 07:51:17,440 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub052:0/16] 2024-01-23 07:51:17,444 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 07:57:01,720 (trainer:737) INFO: 5epoch:train:8701-8800batch: iter_time=3.220, forward_time=0.379, loss_ctc=90.590, loss_att=81.689, acc=0.598, loss=84.359, backward_time=0.421, grad_norm=44.803, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.096, optim0_lr0=4.671e-04, train_time=4.447
+[gpub052:0/16] 2024-01-23 07:59:24,821 (trainer:737) INFO: 5epoch:train:8801-8900batch: iter_time=8.556e-05, forward_time=0.301, loss_ctc=88.075, loss_att=89.885, acc=0.593, loss=89.342, backward_time=0.408, grad_norm=38.798, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.094, optim0_lr0=4.668e-04, train_time=1.430
+[gpub052:0/16] 2024-01-23 08:01:31,865 (trainer:737) INFO: 5epoch:train:8901-9000batch: iter_time=2.146e-04, forward_time=0.404, loss_ctc=65.595, loss_att=66.405, acc=0.615, loss=66.162, backward_time=0.433, grad_norm=32.105, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.101, optim0_lr0=4.664e-04, train_time=1.271
+[gpub052:0/16] 2024-01-23 08:03:52,922 (trainer:737) INFO: 5epoch:train:9001-9100batch: iter_time=8.982e-05, forward_time=0.378, loss_ctc=86.706, loss_att=81.070, acc=0.606, loss=82.761, backward_time=0.434, grad_norm=36.607, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.096, optim0_lr0=4.661e-04, train_time=1.410
+[gpub052:0/16] 2024-01-23 08:06:04,060 (trainer:737) INFO: 5epoch:train:9101-9200batch: iter_time=8.821e-05, forward_time=0.314, loss_ctc=78.720, loss_att=83.516, acc=0.620, loss=82.077, backward_time=0.409, grad_norm=34.387, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.094, optim0_lr0=4.657e-04, train_time=1.311
+[gpub052:0/16] 2024-01-23 08:08:18,769 (trainer:737) INFO: 5epoch:train:9201-9300batch: iter_time=8.435e-05, forward_time=0.391, loss_ctc=67.189, loss_att=71.053, acc=0.602, loss=69.894, backward_time=0.447, grad_norm=29.687, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.100, optim0_lr0=4.654e-04, train_time=1.347
+[gpub052:0/16] 2024-01-23 08:10:27,570 (trainer:737) INFO: 5epoch:train:9301-9400batch: iter_time=8.222e-05, forward_time=0.291, loss_ctc=71.062, loss_att=69.789, acc=0.620, loss=70.171, backward_time=0.404, grad_norm=33.006, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.094, optim0_lr0=4.651e-04, train_time=1.288
+[gpub052:0/16] 2024-01-23 08:12:50,960 (trainer:737) INFO: 5epoch:train:9401-9500batch: iter_time=8.587e-05, forward_time=0.377, loss_ctc=92.788, loss_att=94.867, acc=0.571, loss=94.243, backward_time=0.453, grad_norm=40.694, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.101, optim0_lr0=4.647e-04, train_time=1.433
+[gpub052:0/16] 2024-01-23 08:15:11,441 (trainer:737) INFO: 5epoch:train:9501-9600batch: iter_time=2.943e-04, forward_time=0.358, loss_ctc=74.554, loss_att=78.805, acc=0.620, loss=77.530, backward_time=0.494, grad_norm=31.380, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.103, optim0_lr0=4.644e-04, train_time=1.405
+[gpub052:0/16] 2024-01-23 08:17:22,264 (trainer:737) INFO: 5epoch:train:9601-9700batch: iter_time=8.292e-05, forward_time=0.293, loss_ctc=76.318, loss_att=70.728, acc=0.633, loss=72.405, backward_time=0.405, grad_norm=33.211, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.094, optim0_lr0=4.641e-04, train_time=1.306
+[gpub052:0/16] 2024-01-23 08:19:51,161 (trainer:737) INFO: 5epoch:train:9701-9800batch: iter_time=5.867e-04, forward_time=0.452, loss_ctc=68.401, loss_att=77.856, acc=0.566, loss=75.020, backward_time=0.433, grad_norm=34.564, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.101, optim0_lr0=4.637e-04, train_time=1.488
+[gpub052:0/16] 2024-01-23 08:21:56,733 (trainer:737) INFO: 5epoch:train:9801-9900batch: iter_time=7.717e-05, forward_time=0.298, loss_ctc=78.483, loss_att=92.105, acc=0.609, loss=88.019, backward_time=0.411, grad_norm=34.239, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.095, optim0_lr0=4.634e-04, train_time=1.257
+[gpub052:0/16] 2024-01-23 08:24:22,866 (trainer:737) INFO: 5epoch:train:9901-10000batch: iter_time=4.393e-04, forward_time=0.369, loss_ctc=90.220, loss_att=79.646, acc=0.609, loss=82.818, backward_time=0.450, grad_norm=45.077, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.103, optim0_lr0=4.631e-04, train_time=1.461
+[gpub052:0/16] 2024-01-23 08:24:42,941 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub052:0/16] 2024-01-23 08:25:02,962 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 08:25:06,710 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5b793db790>)
+[gpub052:0/16] 2024-01-23 08:25:06,710 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub052:0/16] 2024-01-23 08:25:06,713 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 08:31:43,030 (trainer:737) INFO: 5epoch:train:10001-10100batch: iter_time=3.122, forward_time=0.303, loss_ctc=86.586, loss_att=84.892, acc=0.601, loss=85.400, backward_time=0.411, grad_norm=40.086, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.095, optim0_lr0=4.627e-04, train_time=4.402
+[gpub052:0/16] 2024-01-23 08:34:05,586 (trainer:737) INFO: 5epoch:train:10101-10200batch: iter_time=8.291e-05, forward_time=0.416, loss_ctc=82.084, loss_att=84.192, acc=0.605, loss=83.559, backward_time=0.446, grad_norm=35.926, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.100, optim0_lr0=4.624e-04, train_time=1.424
+[gpub052:0/16] 2024-01-23 08:36:37,308 (trainer:737) INFO: 5epoch:train:10201-10300batch: iter_time=7.804e-05, forward_time=0.377, loss_ctc=70.281, loss_att=64.853, acc=0.628, loss=66.481, backward_time=0.440, grad_norm=35.534, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.100, optim0_lr0=4.621e-04, train_time=1.518
+[gpub052:0/16] 2024-01-23 08:38:43,811 (trainer:737) INFO: 5epoch:train:10301-10400batch: iter_time=8.136e-05, forward_time=0.292, loss_ctc=80.526, loss_att=81.156, acc=0.618, loss=80.967, backward_time=0.405, grad_norm=33.394, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.095, optim0_lr0=4.618e-04, train_time=1.265
+[gpub052:0/16] 2024-01-23 08:40:57,219 (trainer:737) INFO: 5epoch:train:10401-10500batch: iter_time=8.217e-05, forward_time=0.426, loss_ctc=74.965, loss_att=72.634, acc=0.625, loss=73.333, backward_time=0.431, grad_norm=32.016, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.100, optim0_lr0=4.614e-04, train_time=1.335
+[gpub052:0/16] 2024-01-23 08:43:21,070 (trainer:737) INFO: 5epoch:train:10501-10600batch: iter_time=8.170e-05, forward_time=0.355, loss_ctc=64.080, loss_att=67.434, acc=0.614, loss=66.428, backward_time=0.436, grad_norm=30.736, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.102, optim0_lr0=4.611e-04, train_time=1.438
+[gpub052:0/16] 2024-01-23 08:45:23,497 (trainer:737) INFO: 5epoch:train:10601-10700batch: iter_time=8.260e-05, forward_time=0.293, loss_ctc=76.769, loss_att=86.366, acc=0.610, loss=83.487, backward_time=0.406, grad_norm=33.563, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.094, optim0_lr0=4.608e-04, train_time=1.223
+[gpub052:0/16] 2024-01-23 08:47:46,527 (trainer:737) INFO: 5epoch:train:10701-10800batch: iter_time=9.050e-05, forward_time=0.356, loss_ctc=84.132, loss_att=76.272, acc=0.596, loss=78.630, backward_time=0.478, grad_norm=38.252, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.101, optim0_lr0=4.604e-04, train_time=1.431
+[gpub052:0/16] 2024-01-23 08:50:02,124 (trainer:737) INFO: 5epoch:train:10801-10900batch: iter_time=4.308e-04, forward_time=0.376, loss_ctc=75.197, loss_att=73.617, acc=0.635, loss=74.091, backward_time=0.431, grad_norm=30.295, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.097, optim0_lr0=4.601e-04, train_time=1.356
+[gpub052:0/16] 2024-01-23 08:52:11,048 (trainer:737) INFO: 5epoch:train:10901-11000batch: iter_time=8.502e-05, forward_time=0.301, loss_ctc=70.912, loss_att=77.926, acc=0.584, loss=75.821, backward_time=0.406, grad_norm=31.793, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.094, optim0_lr0=4.598e-04, train_time=1.289
+[gpub052:0/16] 2024-01-23 08:54:31,383 (trainer:737) INFO: 5epoch:train:11001-11100batch: iter_time=4.389e-04, forward_time=0.360, loss_ctc=75.770, loss_att=82.262, acc=0.600, loss=80.314, backward_time=0.469, grad_norm=33.766, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.100, optim0_lr0=4.595e-04, train_time=1.402
+[gpub052:0/16] 2024-01-23 08:56:41,768 (trainer:737) INFO: 5epoch:train:11101-11200batch: iter_time=3.552e-04, forward_time=0.333, loss_ctc=80.473, loss_att=86.314, acc=0.613, loss=84.561, backward_time=0.443, grad_norm=35.410, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.101, optim0_lr0=4.591e-04, train_time=1.304
+[gpub052:0/16] 2024-01-23 08:58:10,809 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub052:0/16] 2024-01-23 08:58:31,395 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 08:58:35,020 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58db667c40>)
+[gpub052:0/16] 2024-01-23 08:58:35,020 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub052:0/16] 2024-01-23 08:58:35,024 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 09:04:16,121 (trainer:737) INFO: 5epoch:train:11201-11300batch: iter_time=3.193, forward_time=0.325, loss_ctc=86.457, loss_att=74.735, acc=0.619, loss=78.251, backward_time=0.409, grad_norm=42.844, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.095, optim0_lr0=4.588e-04, train_time=4.543
+[gpub052:0/16] 2024-01-23 09:06:25,051 (trainer:737) INFO: 5epoch:train:11301-11400batch: iter_time=8.010e-05, forward_time=0.409, loss_ctc=86.957, loss_att=85.719, acc=0.600, loss=86.090, backward_time=0.442, grad_norm=38.637, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.101, optim0_lr0=4.585e-04, train_time=1.289
+[gpub052:0/16] 2024-01-23 09:08:55,755 (trainer:737) INFO: 5epoch:train:11401-11500batch: iter_time=7.842e-05, forward_time=0.321, loss_ctc=63.465, loss_att=64.509, acc=0.621, loss=64.195, backward_time=0.401, grad_norm=31.306, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.095, optim0_lr0=4.582e-04, train_time=1.507
+[gpub052:0/16] 2024-01-23 09:11:19,340 (trainer:737) INFO: 5epoch:train:11501-11600batch: iter_time=8.704e-05, forward_time=0.359, loss_ctc=83.778, loss_att=79.049, acc=0.617, loss=80.467, backward_time=0.456, grad_norm=34.360, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.103, optim0_lr0=4.579e-04, train_time=1.435
+[gpub052:0/16] 2024-01-23 09:13:29,938 (trainer:737) INFO: 5epoch:train:11601-11700batch: iter_time=1.421e-04, forward_time=0.408, loss_ctc=75.846, loss_att=80.129, acc=0.625, loss=78.844, backward_time=0.452, grad_norm=31.589, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.100, optim0_lr0=4.575e-04, train_time=1.306
+[gpub052:0/16] 2024-01-23 09:16:04,936 (trainer:737) INFO: 5epoch:train:11701-11800batch: iter_time=8.383e-05, forward_time=0.289, loss_ctc=65.059, loss_att=68.748, acc=0.610, loss=67.641, backward_time=0.402, grad_norm=30.012, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.094, optim0_lr0=4.572e-04, train_time=1.550
+[gpub052:0/16] 2024-01-23 09:18:14,240 (trainer:737) INFO: 5epoch:train:11801-11900batch: iter_time=8.130e-05, forward_time=0.397, loss_ctc=68.709, loss_att=66.830, acc=0.626, loss=67.394, backward_time=0.451, grad_norm=30.021, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.101, optim0_lr0=4.569e-04, train_time=1.293
+[gpub052:0/16] 2024-01-23 09:20:37,527 (trainer:737) INFO: 5epoch:train:11901-12000batch: iter_time=1.632e-04, forward_time=0.340, loss_ctc=90.258, loss_att=91.701, acc=0.568, loss=91.268, backward_time=0.438, grad_norm=40.371, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.101, optim0_lr0=4.566e-04, train_time=1.433
+[gpub052:0/16] 2024-01-23 09:23:10,505 (trainer:737) INFO: 5epoch:train:12001-12100batch: iter_time=8.018e-05, forward_time=0.303, loss_ctc=72.228, loss_att=75.153, acc=0.625, loss=74.275, backward_time=0.407, grad_norm=30.182, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.095, optim0_lr0=4.563e-04, train_time=1.529
+[gpub052:0/16] 2024-01-23 09:25:37,639 (trainer:737) INFO: 5epoch:train:12101-12200batch: iter_time=2.064e-04, forward_time=0.408, loss_ctc=72.448, loss_att=68.266, acc=0.632, loss=69.521, backward_time=0.442, grad_norm=30.291, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.099, optim0_lr0=4.560e-04, train_time=1.471
+[gpub052:0/16] 2024-01-23 09:28:24,664 (trainer:737) INFO: 5epoch:train:12201-12300batch: iter_time=6.093e-04, forward_time=0.378, loss_ctc=66.426, loss_att=75.387, acc=0.567, loss=72.699, backward_time=0.439, grad_norm=33.160, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.107, optim0_lr0=4.556e-04, train_time=1.671
+[gpub052:0/16] 2024-01-23 09:30:47,608 (trainer:737) INFO: 5epoch:train:12301-12400batch: iter_time=8.328e-05, forward_time=0.298, loss_ctc=76.258, loss_att=90.395, acc=0.612, loss=86.154, backward_time=0.408, grad_norm=32.828, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.095, optim0_lr0=4.553e-04, train_time=1.428
+[gpub052:0/16] 2024-01-23 09:33:19,597 (trainer:737) INFO: 5epoch:train:12401-12500batch: iter_time=2.816e-04, forward_time=0.381, loss_ctc=87.831, loss_att=78.232, acc=0.603, loss=81.112, backward_time=0.464, grad_norm=42.840, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.104, optim0_lr0=4.550e-04, train_time=1.521
+[gpub052:0/16] 2024-01-23 09:33:39,808 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub052:0/16] 2024-01-23 09:33:59,915 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 09:34:03,564 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f583e173520>)
+[gpub052:0/16] 2024-01-23 09:34:03,564 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub052:0/16] 2024-01-23 09:34:03,567 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 09:41:07,504 (trainer:737) INFO: 5epoch:train:12501-12600batch: iter_time=3.348, forward_time=0.411, loss_ctc=84.717, loss_att=82.850, acc=0.611, loss=83.410, backward_time=0.432, grad_norm=38.247, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.097, optim0_lr0=4.547e-04, train_time=4.679
+[gpub052:0/16] 2024-01-23 09:43:16,095 (trainer:737) INFO: 5epoch:train:12601-12700batch: iter_time=7.946e-05, forward_time=0.421, loss_ctc=79.973, loss_att=82.361, acc=0.614, loss=81.645, backward_time=0.441, grad_norm=36.424, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.099, optim0_lr0=4.544e-04, train_time=1.285
+[gpub052:0/16] 2024-01-23 09:46:03,672 (trainer:737) INFO: 5epoch:train:12701-12800batch: iter_time=7.964e-05, forward_time=0.312, loss_ctc=67.957, loss_att=62.618, acc=0.639, loss=64.219, backward_time=0.466, grad_norm=32.667, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.106, optim0_lr0=4.541e-04, train_time=1.676
+[gpub052:0/16] 2024-01-23 09:48:05,914 (trainer:737) INFO: 5epoch:train:12801-12900batch: iter_time=8.120e-05, forward_time=0.349, loss_ctc=77.885, loss_att=78.926, acc=0.627, loss=78.614, backward_time=0.425, grad_norm=32.033, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.095, optim0_lr0=4.538e-04, train_time=1.222
+[gpub052:0/16] 2024-01-23 09:50:49,994 (trainer:737) INFO: 5epoch:train:12901-13000batch: iter_time=8.204e-05, forward_time=0.382, loss_ctc=73.003, loss_att=71.156, acc=0.634, loss=71.710, backward_time=0.431, grad_norm=30.776, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.103, optim0_lr0=4.534e-04, train_time=1.641
+[gpub052:0/16] 2024-01-23 09:53:14,001 (trainer:737) INFO: 5epoch:train:13001-13100batch: iter_time=3.605e-04, forward_time=0.365, loss_ctc=63.157, loss_att=66.094, acc=0.625, loss=65.213, backward_time=0.427, grad_norm=28.569, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.103, optim0_lr0=4.531e-04, train_time=1.439
+[gpub052:0/16] 2024-01-23 09:55:26,891 (trainer:737) INFO: 5epoch:train:13101-13200batch: iter_time=8.433e-05, forward_time=0.353, loss_ctc=75.055, loss_att=84.244, acc=0.618, loss=81.487, backward_time=0.453, grad_norm=33.090, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.099, optim0_lr0=4.528e-04, train_time=1.329
+[gpub052:0/16] 2024-01-23 09:57:58,946 (trainer:737) INFO: 5epoch:train:13201-13300batch: iter_time=2.369e-04, forward_time=0.338, loss_ctc=82.131, loss_att=74.007, acc=0.609, loss=76.444, backward_time=0.412, grad_norm=36.402, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.095, optim0_lr0=4.525e-04, train_time=1.519
+[gpub052:0/16] 2024-01-23 10:00:26,664 (trainer:737) INFO: 5epoch:train:13301-13400batch: iter_time=1.551e-04, forward_time=0.387, loss_ctc=73.574, loss_att=71.736, acc=0.644, loss=72.287, backward_time=0.448, grad_norm=29.910, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.104, optim0_lr0=4.522e-04, train_time=1.478
+[gpub052:0/16] 2024-01-23 10:02:47,518 (trainer:737) INFO: 5epoch:train:13401-13500batch: iter_time=8.094e-05, forward_time=0.327, loss_ctc=69.300, loss_att=74.646, acc=0.599, loss=73.042, backward_time=0.416, grad_norm=32.527, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.102, optim0_lr0=4.519e-04, train_time=1.408
+[gpub052:0/16] 2024-01-23 10:05:04,181 (trainer:737) INFO: 5epoch:train:13501-13600batch: iter_time=3.380e-04, forward_time=0.351, loss_ctc=73.119, loss_att=81.091, acc=0.606, loss=78.700, backward_time=0.413, grad_norm=31.889, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.096, optim0_lr0=4.516e-04, train_time=1.367
+[gpub052:0/16] 2024-01-23 10:07:32,888 (trainer:737) INFO: 5epoch:train:13601-13700batch: iter_time=3.370e-04, forward_time=0.363, loss_ctc=79.025, loss_att=85.857, acc=0.616, loss=83.807, backward_time=0.458, grad_norm=34.719, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.103, optim0_lr0=4.513e-04, train_time=1.486
+[gpub052:0/16] 2024-01-23 10:09:05,830 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub052:0/16] 2024-01-23 10:09:25,819 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 10:09:29,452 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58caf94940>)
+[gpub052:0/16] 2024-01-23 10:09:29,452 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub052:0/16] 2024-01-23 10:09:29,455 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 10:15:59,484 (trainer:737) INFO: 5epoch:train:13701-13800batch: iter_time=3.610, forward_time=0.425, loss_ctc=84.938, loss_att=73.829, acc=0.630, loss=77.162, backward_time=0.424, grad_norm=39.841, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.097, optim0_lr0=4.510e-04, train_time=5.067
+[gpub052:0/16] 2024-01-23 10:18:31,296 (trainer:737) INFO: 5epoch:train:13801-13900batch: iter_time=6.173e-04, forward_time=0.383, loss_ctc=83.328, loss_att=82.234, acc=0.612, loss=82.562, backward_time=0.433, grad_norm=37.008, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.120, optim0_lr0=4.507e-04, train_time=1.517
+[gpub052:0/16] 2024-01-23 10:21:00,612 (trainer:737) INFO: 5epoch:train:13901-14000batch: iter_time=8.988e-05, forward_time=0.288, loss_ctc=61.805, loss_att=61.709, acc=0.630, loss=61.737, backward_time=0.397, grad_norm=29.814, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.094, optim0_lr0=4.504e-04, train_time=1.493
+[gpub052:0/16] 2024-01-23 10:23:15,056 (trainer:737) INFO: 5epoch:train:14001-14100batch: iter_time=3.399e-04, forward_time=0.413, loss_ctc=81.824, loss_att=76.614, acc=0.627, loss=78.177, backward_time=0.464, grad_norm=37.390, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.104, optim0_lr0=4.501e-04, train_time=1.344
+[gpub052:0/16] 2024-01-23 10:25:28,665 (trainer:737) INFO: 5epoch:train:14101-14200batch: iter_time=8.990e-05, forward_time=0.292, loss_ctc=74.262, loss_att=77.859, acc=0.633, loss=76.780, backward_time=0.404, grad_norm=33.186, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.094, optim0_lr0=4.498e-04, train_time=1.336
+[gpub052:0/16] 2024-01-23 10:28:04,789 (trainer:737) INFO: 5epoch:train:14201-14300batch: iter_time=3.222e-04, forward_time=0.421, loss_ctc=63.732, loss_att=67.414, acc=0.618, loss=66.310, backward_time=0.434, grad_norm=30.245, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.101, optim0_lr0=4.495e-04, train_time=1.561
+[gpub052:0/16] 2024-01-23 10:29:58,870 (trainer:737) INFO: 5epoch:train:14301-14400batch: iter_time=8.785e-05, forward_time=0.289, loss_ctc=67.677, loss_att=65.865, acc=0.633, loss=66.409, backward_time=0.401, grad_norm=30.152, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.094, optim0_lr0=4.492e-04, train_time=1.141
+[gpub052:0/16] 2024-01-23 10:33:03,352 (trainer:737) INFO: 5epoch:train:14401-14500batch: iter_time=0.012, forward_time=0.416, loss_ctc=87.809, loss_att=89.655, acc=0.577, loss=89.101, backward_time=0.440, grad_norm=40.881, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.102, optim0_lr0=4.489e-04, train_time=1.843
+[gpub052:0/16] 2024-01-23 10:35:46,585 (trainer:737) INFO: 5epoch:train:14501-14600batch: iter_time=0.243, forward_time=0.291, loss_ctc=71.167, loss_att=73.524, acc=0.634, loss=72.817, backward_time=0.404, grad_norm=30.044, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.094, optim0_lr0=4.486e-04, train_time=1.633
+[gpub052:0/16] 2024-01-23 10:38:31,174 (trainer:737) INFO: 5epoch:train:14601-14700batch: iter_time=5.400e-04, forward_time=0.405, loss_ctc=71.314, loss_att=66.214, acc=0.640, loss=67.744, backward_time=0.434, grad_norm=30.709, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.111, optim0_lr0=4.483e-04, train_time=1.644
+[gpub052:0/16] 2024-01-23 10:40:27,309 (trainer:737) INFO: 5epoch:train:14701-14800batch: iter_time=8.545e-05, forward_time=0.289, loss_ctc=65.315, loss_att=73.893, acc=0.576, loss=71.320, backward_time=0.402, grad_norm=32.782, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.094, optim0_lr0=4.480e-04, train_time=1.162
+[gpub052:0/16] 2024-01-23 10:43:09,541 (trainer:737) INFO: 5epoch:train:14801-14900batch: iter_time=3.771e-04, forward_time=0.347, loss_ctc=74.192, loss_att=87.851, acc=0.621, loss=83.753, backward_time=0.455, grad_norm=31.965, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.106, optim0_lr0=4.477e-04, train_time=1.622
+[gpub052:0/16] 2024-01-23 10:45:16,704 (trainer:737) INFO: 5epoch:train:14901-15000batch: iter_time=1.709e-04, forward_time=0.294, loss_ctc=85.922, loss_att=76.012, acc=0.613, loss=78.985, backward_time=0.413, grad_norm=40.735, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.099, optim0_lr0=4.474e-04, train_time=1.271
+[gpub052:0/16] 2024-01-23 11:23:38,707 (trainer:343) INFO: 5epoch results: [train] iter_time=0.260, forward_time=0.348, loss_ctc=81.453, loss_att=82.310, acc=0.591, loss=82.053, backward_time=0.429, grad_norm=37.860, clip=100.000, loss_scale=1.792e+15, optim_step_time=0.099, optim0_lr0=4.721e-04, train_time=1.645, time=6 hours, 51 minutes and 47.95 seconds, total_count=75000, gpu_max_cached_mem_GB=41.549, [valid] loss_ctc=69.561, cer_ctc=0.353, loss_att=63.706, acc=0.509, cer=0.425, wer=1.000, loss=65.462, time=37 minutes and 55.51 seconds, total_count=23355, gpu_max_cached_mem_GB=41.549
+[gpub052:0/16] 2024-01-23 11:23:48,676 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub052:0/16] 2024-01-23 11:23:48,739 (trainer:272) INFO: 6/45epoch started. Estimated time to finish: 2 weeks, 2 days and 6 hours
+[gpub052:0/16] 2024-01-23 11:23:48,751 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub052:0/16] 2024-01-23 11:24:07,998 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 11:24:11,778 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5165462050>)
+[gpub052:0/16] 2024-01-23 11:24:11,778 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub052:0/16] 2024-01-23 11:24:11,781 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 11:31:07,961 (trainer:737) INFO: 6epoch:train:1-100batch: iter_time=3.166, forward_time=0.364, loss_ctc=89.308, loss_att=74.913, acc=0.593, loss=79.231, backward_time=0.414, grad_norm=38.709, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.097, optim0_lr0=4.471e-04, train_time=4.392
+[gpub052:0/16] 2024-01-23 11:33:08,290 (trainer:737) INFO: 6epoch:train:101-200batch: iter_time=8.120e-05, forward_time=0.347, loss_ctc=77.557, loss_att=77.988, acc=0.608, loss=77.859, backward_time=0.426, grad_norm=34.140, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.101, optim0_lr0=4.468e-04, train_time=1.203
+[gpub052:0/16] 2024-01-23 11:35:39,414 (trainer:737) INFO: 6epoch:train:201-300batch: iter_time=0.001, forward_time=0.343, loss_ctc=83.236, loss_att=84.132, acc=0.615, loss=83.863, backward_time=0.412, grad_norm=34.136, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.096, optim0_lr0=4.465e-04, train_time=1.511
+[gpub052:0/16] 2024-01-23 11:37:48,995 (trainer:737) INFO: 6epoch:train:301-400batch: iter_time=7.862e-05, forward_time=0.344, loss_ctc=74.854, loss_att=78.426, acc=0.622, loss=77.354, backward_time=0.434, grad_norm=33.212, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.097, optim0_lr0=4.462e-04, train_time=1.296
+[gpub052:0/16] 2024-01-23 11:40:14,668 (trainer:737) INFO: 6epoch:train:401-500batch: iter_time=3.923e-04, forward_time=0.318, loss_ctc=71.467, loss_att=80.238, acc=0.591, loss=77.607, backward_time=0.432, grad_norm=32.032, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.099, optim0_lr0=4.459e-04, train_time=1.457
+[gpub052:0/16] 2024-01-23 11:42:34,116 (trainer:737) INFO: 6epoch:train:501-600batch: iter_time=8.265e-05, forward_time=0.332, loss_ctc=73.054, loss_att=64.508, acc=0.636, loss=67.072, backward_time=0.418, grad_norm=31.622, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.096, optim0_lr0=4.456e-04, train_time=1.394
+[gpub052:0/16] 2024-01-23 11:45:00,314 (trainer:737) INFO: 6epoch:train:601-700batch: iter_time=3.430e-04, forward_time=0.353, loss_ctc=83.488, loss_att=74.309, acc=0.632, loss=77.062, backward_time=0.435, grad_norm=34.223, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.098, optim0_lr0=4.453e-04, train_time=1.461
+[gpub052:0/16] 2024-01-23 11:47:32,385 (trainer:737) INFO: 6epoch:train:701-800batch: iter_time=9.259e-05, forward_time=0.349, loss_ctc=85.708, loss_att=89.651, acc=0.586, loss=88.468, backward_time=0.458, grad_norm=40.589, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.102, optim0_lr0=4.450e-04, train_time=1.521
+[gpub052:0/16] 2024-01-23 11:49:56,503 (trainer:737) INFO: 6epoch:train:801-900batch: iter_time=1.985e-04, forward_time=0.319, loss_ctc=72.399, loss_att=72.762, acc=0.632, loss=72.653, backward_time=0.416, grad_norm=30.176, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.097, optim0_lr0=4.447e-04, train_time=1.440
+[gpub052:0/16] 2024-01-23 11:52:18,288 (trainer:737) INFO: 6epoch:train:901-1000batch: iter_time=8.643e-05, forward_time=0.379, loss_ctc=84.878, loss_att=93.890, acc=0.584, loss=91.186, backward_time=0.462, grad_norm=39.236, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.100, optim0_lr0=4.444e-04, train_time=1.418
+[gpub052:0/16] 2024-01-23 11:54:47,075 (trainer:737) INFO: 6epoch:train:1001-1100batch: iter_time=3.179e-04, forward_time=0.315, loss_ctc=69.183, loss_att=68.629, acc=0.622, loss=68.795, backward_time=0.402, grad_norm=32.877, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.094, optim0_lr0=4.441e-04, train_time=1.488
+[gpub052:0/16] 2024-01-23 11:57:21,742 (trainer:737) INFO: 6epoch:train:1101-1200batch: iter_time=8.261e-05, forward_time=0.327, loss_ctc=69.025, loss_att=70.537, acc=0.644, loss=70.083, backward_time=0.411, grad_norm=31.048, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.097, optim0_lr0=4.438e-04, train_time=1.546
+[gpub052:0/16] 2024-01-23 11:58:51,882 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub052:0/16] 2024-01-23 11:59:11,567 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 11:59:15,192 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5a4015bfd0>)
+[gpub052:0/16] 2024-01-23 11:59:15,192 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub052:0/16] 2024-01-23 11:59:15,196 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 12:04:57,337 (trainer:737) INFO: 6epoch:train:1201-1300batch: iter_time=3.294, forward_time=0.346, loss_ctc=90.490, loss_att=77.747, acc=0.606, loss=81.570, backward_time=0.448, grad_norm=38.273, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.100, optim0_lr0=4.435e-04, train_time=4.556
+[gpub052:0/16] 2024-01-23 12:07:18,768 (trainer:737) INFO: 6epoch:train:1301-1400batch: iter_time=8.306e-05, forward_time=0.363, loss_ctc=74.923, loss_att=74.172, acc=0.629, loss=74.397, backward_time=0.425, grad_norm=32.469, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.103, optim0_lr0=4.432e-04, train_time=1.414
+[gpub052:0/16] 2024-01-23 12:09:36,265 (trainer:737) INFO: 6epoch:train:1401-1500batch: iter_time=5.670e-04, forward_time=0.297, loss_ctc=74.784, loss_att=84.146, acc=0.622, loss=81.337, backward_time=0.412, grad_norm=31.725, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.095, optim0_lr0=4.429e-04, train_time=1.375
+[gpub052:0/16] 2024-01-23 12:11:44,939 (trainer:737) INFO: 6epoch:train:1501-1600batch: iter_time=8.861e-05, forward_time=0.407, loss_ctc=82.837, loss_att=83.562, acc=0.611, loss=83.345, backward_time=0.439, grad_norm=35.538, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.102, optim0_lr0=4.427e-04, train_time=1.287
+[gpub052:0/16] 2024-01-23 12:13:40,508 (trainer:737) INFO: 6epoch:train:1601-1700batch: iter_time=8.413e-05, forward_time=0.301, loss_ctc=75.647, loss_att=86.484, acc=0.612, loss=83.233, backward_time=0.418, grad_norm=30.989, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.096, optim0_lr0=4.424e-04, train_time=1.155
+[gpub052:0/16] 2024-01-23 12:16:22,383 (trainer:737) INFO: 6epoch:train:1701-1800batch: iter_time=2.805e-04, forward_time=0.371, loss_ctc=70.461, loss_att=64.177, acc=0.636, loss=66.062, backward_time=0.435, grad_norm=31.487, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.100, optim0_lr0=4.421e-04, train_time=1.618
+[gpub052:0/16] 2024-01-23 12:18:24,953 (trainer:737) INFO: 6epoch:train:1801-1900batch: iter_time=2.318e-04, forward_time=0.311, loss_ctc=66.065, loss_att=71.664, acc=0.638, loss=69.984, backward_time=0.416, grad_norm=28.716, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.104, optim0_lr0=4.418e-04, train_time=1.226
+[gpub052:0/16] 2024-01-23 12:20:30,554 (trainer:737) INFO: 6epoch:train:1901-2000batch: iter_time=4.996e-04, forward_time=0.393, loss_ctc=82.914, loss_att=86.459, acc=0.603, loss=85.396, backward_time=0.436, grad_norm=34.385, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.105, optim0_lr0=4.415e-04, train_time=1.255
+[gpub052:0/16] 2024-01-23 12:22:52,706 (trainer:737) INFO: 6epoch:train:2001-2100batch: iter_time=1.532e-04, forward_time=0.317, loss_ctc=78.308, loss_att=78.901, acc=0.627, loss=78.723, backward_time=0.420, grad_norm=32.478, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.099, optim0_lr0=4.412e-04, train_time=1.422
+[gpub052:0/16] 2024-01-23 12:25:12,812 (trainer:737) INFO: 6epoch:train:2101-2200batch: iter_time=6.757e-04, forward_time=0.361, loss_ctc=78.147, loss_att=72.958, acc=0.638, loss=74.514, backward_time=0.424, grad_norm=31.643, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.099, optim0_lr0=4.409e-04, train_time=1.401
+[gpub052:0/16] 2024-01-23 12:27:23,734 (trainer:737) INFO: 6epoch:train:2201-2300batch: iter_time=8.506e-05, forward_time=0.402, loss_ctc=72.921, loss_att=87.915, acc=0.623, loss=83.416, backward_time=0.442, grad_norm=31.595, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.102, optim0_lr0=4.406e-04, train_time=1.309
+[gpub052:0/16] 2024-01-23 12:29:21,245 (trainer:737) INFO: 6epoch:train:2301-2400batch: iter_time=8.279e-04, forward_time=0.307, loss_ctc=70.392, loss_att=73.128, acc=0.633, loss=72.307, backward_time=0.411, grad_norm=32.070, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.095, optim0_lr0=4.404e-04, train_time=1.174
+[gpub052:0/16] 2024-01-23 12:31:48,871 (trainer:737) INFO: 6epoch:train:2401-2500batch: iter_time=8.083e-05, forward_time=0.369, loss_ctc=72.358, loss_att=71.375, acc=0.639, loss=71.670, backward_time=0.437, grad_norm=33.512, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.107, optim0_lr0=4.401e-04, train_time=1.477
+[gpub052:0/16] 2024-01-23 12:32:09,006 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub052:0/16] 2024-01-23 12:32:28,670 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 12:32:32,256 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5163761330>)
+[gpub052:0/16] 2024-01-23 12:32:32,256 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub052:0/16] 2024-01-23 12:32:32,260 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 12:39:19,018 (trainer:737) INFO: 6epoch:train:2501-2600batch: iter_time=3.283, forward_time=0.309, loss_ctc=85.651, loss_att=73.972, acc=0.600, loss=77.476, backward_time=0.409, grad_norm=34.759, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.098, optim0_lr0=4.398e-04, train_time=4.501
+[gpub052:0/16] 2024-01-23 12:41:45,709 (trainer:737) INFO: 6epoch:train:2601-2700batch: iter_time=8.686e-05, forward_time=0.385, loss_ctc=74.407, loss_att=76.343, acc=0.618, loss=75.762, backward_time=0.448, grad_norm=34.737, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.105, optim0_lr0=4.395e-04, train_time=1.466
+[gpub052:0/16] 2024-01-23 12:44:04,866 (trainer:737) INFO: 6epoch:train:2701-2800batch: iter_time=8.324e-05, forward_time=0.301, loss_ctc=78.697, loss_att=80.045, acc=0.626, loss=79.641, backward_time=0.420, grad_norm=34.295, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.096, optim0_lr0=4.392e-04, train_time=1.391
+[gpub052:0/16] 2024-01-23 12:46:10,470 (trainer:737) INFO: 6epoch:train:2801-2900batch: iter_time=1.313e-04, forward_time=0.373, loss_ctc=71.904, loss_att=76.015, acc=0.633, loss=74.781, backward_time=0.435, grad_norm=34.199, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.102, optim0_lr0=4.389e-04, train_time=1.256
+[gpub052:0/16] 2024-01-23 12:48:12,745 (trainer:737) INFO: 6epoch:train:2901-3000batch: iter_time=8.459e-05, forward_time=0.329, loss_ctc=67.978, loss_att=76.260, acc=0.607, loss=73.775, backward_time=0.428, grad_norm=31.698, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.104, optim0_lr0=4.387e-04, train_time=1.222
+[gpub052:0/16] 2024-01-23 12:50:46,948 (trainer:737) INFO: 6epoch:train:3001-3100batch: iter_time=7.288e-04, forward_time=0.357, loss_ctc=69.380, loss_att=62.412, acc=0.646, loss=64.502, backward_time=0.421, grad_norm=29.108, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.097, optim0_lr0=4.384e-04, train_time=1.543
+[gpub052:0/16] 2024-01-23 12:53:10,435 (trainer:737) INFO: 6epoch:train:3101-3200batch: iter_time=7.843e-05, forward_time=0.394, loss_ctc=80.123, loss_att=71.462, acc=0.643, loss=74.060, backward_time=0.425, grad_norm=32.792, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.106, optim0_lr0=4.381e-04, train_time=1.434
+[gpub052:0/16] 2024-01-23 12:55:05,135 (trainer:737) INFO: 6epoch:train:3201-3300batch: iter_time=4.889e-04, forward_time=0.296, loss_ctc=82.135, loss_att=88.158, acc=0.594, loss=86.351, backward_time=0.422, grad_norm=37.601, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.095, optim0_lr0=4.378e-04, train_time=1.147
+[gpub052:0/16] 2024-01-23 12:57:45,279 (trainer:737) INFO: 6epoch:train:3301-3400batch: iter_time=3.515e-04, forward_time=0.382, loss_ctc=69.524, loss_att=70.414, acc=0.640, loss=70.147, backward_time=0.442, grad_norm=29.129, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.107, optim0_lr0=4.375e-04, train_time=1.601
+[gpub052:0/16] 2024-01-23 12:59:53,882 (trainer:737) INFO: 6epoch:train:3401-3500batch: iter_time=8.593e-05, forward_time=0.295, loss_ctc=78.800, loss_att=89.742, acc=0.598, loss=86.459, backward_time=0.408, grad_norm=34.583, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.095, optim0_lr0=4.373e-04, train_time=1.286
+[gpub052:0/16] 2024-01-23 13:02:03,515 (trainer:737) INFO: 6epoch:train:3501-3600batch: iter_time=2.752e-04, forward_time=0.419, loss_ctc=65.337, loss_att=66.429, acc=0.632, loss=66.101, backward_time=0.444, grad_norm=31.934, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.101, optim0_lr0=4.370e-04, train_time=1.296
+[gpub052:0/16] 2024-01-23 13:04:03,423 (trainer:737) INFO: 6epoch:train:3601-3700batch: iter_time=1.053e-04, forward_time=0.303, loss_ctc=66.423, loss_att=68.346, acc=0.654, loss=67.769, backward_time=0.414, grad_norm=30.214, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.096, optim0_lr0=4.367e-04, train_time=1.199
+[gpub052:0/16] 2024-01-23 13:05:57,530 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub052:0/16] 2024-01-23 13:06:17,404 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 13:06:21,012 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5887c93df0>)
+[gpub052:0/16] 2024-01-23 13:06:21,012 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub052:0/16] 2024-01-23 13:06:21,015 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 13:12:03,245 (trainer:737) INFO: 6epoch:train:3701-3800batch: iter_time=3.267, forward_time=0.370, loss_ctc=86.041, loss_att=72.762, acc=0.619, loss=76.746, backward_time=0.428, grad_norm=34.337, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.100, optim0_lr0=4.364e-04, train_time=4.798
+[gpub052:0/16] 2024-01-23 13:14:07,983 (trainer:737) INFO: 6epoch:train:3801-3900batch: iter_time=3.007e-04, forward_time=0.353, loss_ctc=73.429, loss_att=67.026, acc=0.633, loss=68.947, backward_time=0.421, grad_norm=34.839, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.104, optim0_lr0=4.362e-04, train_time=1.247
+[gpub052:0/16] 2024-01-23 13:16:48,700 (trainer:737) INFO: 6epoch:train:3901-4000batch: iter_time=1.916e-04, forward_time=0.345, loss_ctc=72.153, loss_att=76.387, acc=0.628, loss=75.117, backward_time=0.421, grad_norm=29.834, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.096, optim0_lr0=4.359e-04, train_time=1.607
+[gpub052:0/16] 2024-01-23 13:19:02,645 (trainer:737) INFO: 6epoch:train:4001-4100batch: iter_time=8.136e-05, forward_time=0.385, loss_ctc=80.059, loss_att=80.394, acc=0.608, loss=80.293, backward_time=0.440, grad_norm=36.584, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.109, optim0_lr0=4.356e-04, train_time=1.339
+[gpub052:0/16] 2024-01-23 13:21:23,965 (trainer:737) INFO: 6epoch:train:4101-4200batch: iter_time=8.866e-05, forward_time=0.322, loss_ctc=74.840, loss_att=82.781, acc=0.620, loss=80.399, backward_time=0.415, grad_norm=31.025, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.096, optim0_lr0=4.353e-04, train_time=1.413
+[gpub052:0/16] 2024-01-23 13:23:42,848 (trainer:737) INFO: 6epoch:train:4201-4300batch: iter_time=8.282e-05, forward_time=0.338, loss_ctc=68.977, loss_att=62.471, acc=0.639, loss=64.423, backward_time=0.423, grad_norm=30.380, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.095, optim0_lr0=4.351e-04, train_time=1.389
+[gpub052:0/16] 2024-01-23 13:25:57,816 (trainer:737) INFO: 6epoch:train:4301-4400batch: iter_time=3.827e-04, forward_time=0.395, loss_ctc=65.915, loss_att=64.105, acc=0.648, loss=64.648, backward_time=0.426, grad_norm=27.683, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.105, optim0_lr0=4.348e-04, train_time=1.350
+[gpub052:0/16] 2024-01-23 13:28:20,293 (trainer:737) INFO: 6epoch:train:4401-4500batch: iter_time=7.953e-05, forward_time=0.302, loss_ctc=80.371, loss_att=82.981, acc=0.605, loss=82.198, backward_time=0.410, grad_norm=33.372, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.094, optim0_lr0=4.345e-04, train_time=1.424
+[gpub052:0/16] 2024-01-23 13:30:43,499 (trainer:737) INFO: 6epoch:train:4501-4600batch: iter_time=3.965e-04, forward_time=0.378, loss_ctc=74.366, loss_att=72.344, acc=0.635, loss=72.951, backward_time=0.437, grad_norm=30.150, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.099, optim0_lr0=4.342e-04, train_time=1.430
+[gpub052:0/16] 2024-01-23 13:32:42,321 (trainer:737) INFO: 6epoch:train:4601-4700batch: iter_time=8.132e-05, forward_time=0.331, loss_ctc=75.614, loss_att=69.615, acc=0.640, loss=71.415, backward_time=0.449, grad_norm=31.087, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.098, optim0_lr0=4.340e-04, train_time=1.188
+[gpub052:0/16] 2024-01-23 13:35:16,304 (trainer:737) INFO: 6epoch:train:4701-4800batch: iter_time=1.998e-04, forward_time=0.340, loss_ctc=70.860, loss_att=84.396, acc=0.621, loss=80.335, backward_time=0.437, grad_norm=32.435, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.099, optim0_lr0=4.337e-04, train_time=1.539
+[gpub052:0/16] 2024-01-23 13:37:35,334 (trainer:737) INFO: 6epoch:train:4801-4900batch: iter_time=3.310e-04, forward_time=0.363, loss_ctc=67.819, loss_att=70.520, acc=0.630, loss=69.710, backward_time=0.444, grad_norm=31.823, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.102, optim0_lr0=4.334e-04, train_time=1.391
+[gpub052:0/16] 2024-01-23 13:39:50,078 (trainer:737) INFO: 6epoch:train:4901-5000batch: iter_time=8.121e-05, forward_time=0.292, loss_ctc=70.369, loss_att=67.566, acc=0.643, loss=68.407, backward_time=0.411, grad_norm=32.744, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.094, optim0_lr0=4.331e-04, train_time=1.346
+[gpub052:0/16] 2024-01-23 13:40:10,311 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub052:0/16] 2024-01-23 13:40:30,083 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 13:40:34,005 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58de97ff40>)
+[gpub052:0/16] 2024-01-23 13:40:34,005 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub052:0/16] 2024-01-23 13:40:34,008 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 13:46:23,764 (trainer:737) INFO: 6epoch:train:5001-5100batch: iter_time=2.645, forward_time=0.398, loss_ctc=82.972, loss_att=74.892, acc=0.610, loss=77.316, backward_time=0.441, grad_norm=34.043, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.100, optim0_lr0=4.329e-04, train_time=3.938
+[gpub052:0/16] 2024-01-23 13:48:51,209 (trainer:737) INFO: 6epoch:train:5101-5200batch: iter_time=3.416e-04, forward_time=0.383, loss_ctc=72.529, loss_att=78.861, acc=0.631, loss=76.961, backward_time=0.436, grad_norm=31.584, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.099, optim0_lr0=4.326e-04, train_time=1.475
+[gpub052:0/16] 2024-01-23 13:51:06,930 (trainer:737) INFO: 6epoch:train:5201-5300batch: iter_time=7.768e-05, forward_time=0.319, loss_ctc=75.796, loss_att=78.391, acc=0.642, loss=77.612, backward_time=0.417, grad_norm=33.179, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.095, optim0_lr0=4.323e-04, train_time=1.357
+[gpub052:0/16] 2024-01-23 13:53:41,121 (trainer:737) INFO: 6epoch:train:5301-5400batch: iter_time=3.681e-04, forward_time=0.388, loss_ctc=70.462, loss_att=75.914, acc=0.643, loss=74.278, backward_time=0.451, grad_norm=31.088, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.104, optim0_lr0=4.321e-04, train_time=1.542
+[gpub052:0/16] 2024-01-23 13:55:43,879 (trainer:737) INFO: 6epoch:train:5401-5500batch: iter_time=8.047e-05, forward_time=0.334, loss_ctc=66.354, loss_att=76.909, acc=0.609, loss=73.742, backward_time=0.424, grad_norm=32.603, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.096, optim0_lr0=4.318e-04, train_time=1.227
+[gpub052:0/16] 2024-01-23 13:58:23,582 (trainer:737) INFO: 6epoch:train:5501-5600batch: iter_time=1.492e-04, forward_time=0.347, loss_ctc=67.702, loss_att=63.012, acc=0.651, loss=64.419, backward_time=0.414, grad_norm=27.991, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.095, optim0_lr0=4.315e-04, train_time=1.597
+[gpub052:0/16] 2024-01-23 14:01:02,216 (trainer:737) INFO: 6epoch:train:5601-5700batch: iter_time=0.307, forward_time=0.398, loss_ctc=77.723, loss_att=74.103, acc=0.647, loss=75.189, backward_time=0.445, grad_norm=32.395, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.102, optim0_lr0=4.313e-04, train_time=1.586
+[gpub052:0/16] 2024-01-23 14:03:40,069 (trainer:737) INFO: 6epoch:train:5701-5800batch: iter_time=8.314e-05, forward_time=0.293, loss_ctc=79.175, loss_att=87.336, acc=0.609, loss=84.888, backward_time=0.405, grad_norm=34.306, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.094, optim0_lr0=4.310e-04, train_time=1.578
+[gpub052:0/16] 2024-01-23 14:05:49,894 (trainer:737) INFO: 6epoch:train:5801-5900batch: iter_time=2.953e-04, forward_time=0.433, loss_ctc=67.993, loss_att=70.604, acc=0.660, loss=69.821, backward_time=0.433, grad_norm=28.671, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.101, optim0_lr0=4.307e-04, train_time=1.298
+[gpub052:0/16] 2024-01-23 14:08:36,890 (trainer:737) INFO: 6epoch:train:5901-6000batch: iter_time=6.651e-04, forward_time=0.373, loss_ctc=77.626, loss_att=89.365, acc=0.614, loss=85.843, backward_time=0.438, grad_norm=34.335, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.114, optim0_lr0=4.305e-04, train_time=1.669
+[gpub052:0/16] 2024-01-23 14:11:00,048 (trainer:737) INFO: 6epoch:train:6001-6100batch: iter_time=7.728e-05, forward_time=0.291, loss_ctc=65.197, loss_att=66.571, acc=0.646, loss=66.159, backward_time=0.403, grad_norm=31.000, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.094, optim0_lr0=4.302e-04, train_time=1.432
+[gpub052:0/16] 2024-01-23 14:13:09,352 (trainer:737) INFO: 6epoch:train:6101-6200batch: iter_time=4.751e-04, forward_time=0.422, loss_ctc=64.019, loss_att=66.696, acc=0.667, loss=65.893, backward_time=0.429, grad_norm=27.994, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.104, optim0_lr0=4.299e-04, train_time=1.292
+[gpub052:0/16] 2024-01-23 14:14:58,466 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub052:0/16] 2024-01-23 14:15:18,114 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub052:0/16] 2024-01-23 14:15:21,791 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58c8cb7ac0>)
+[gpub052:0/16] 2024-01-23 14:15:21,791 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub052:0/16] 2024-01-23 14:15:21,794 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub052:0/16] 2024-01-23 14:21:16,621 (trainer:737) INFO: 6epoch:train:6201-6300batch: iter_time=3.309, forward_time=0.379, loss_ctc=83.622, loss_att=72.710, acc=0.626, loss=75.984, backward_time=0.425, grad_norm=34.758, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.098, optim0_lr0=4.297e-04, train_time=4.873
+[gpub052:0/16] 2024-01-23 14:23:42,764 (trainer:737) INFO: 6epoch:train:6301-6400batch: iter_time=7.469e-05, forward_time=0.424, loss_ctc=70.968, loss_att=67.491, acc=0.649, loss=68.534, backward_time=0.428, grad_norm=29.232, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.097, optim0_lr0=4.294e-04, train_time=1.461
+[gpub052:0/16] 2024-01-23 14:25:53,181 (trainer:737) INFO: 6epoch:train:6401-6500batch: iter_time=8.008e-05, forward_time=0.304, loss_ctc=70.543, loss_att=78.490, acc=0.642, loss=76.106, backward_time=0.413, grad_norm=29.811, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.094, optim0_lr0=4.291e-04, train_time=1.303
+[gpub052:0/16] 2024-01-23 14:28:09,859 (trainer:737) INFO: 6epoch:train:6501-6600batch: iter_time=8.593e-05, forward_time=0.354, loss_ctc=78.352, loss_att=78.608, acc=0.626, loss=78.531, backward_time=0.464, grad_norm=34.285, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.102, optim0_lr0=4.289e-04, train_time=1.368
+[gpub052:0/16] 2024-01-23 14:30:55,393 (trainer:737) INFO: 6epoch:train:6601-6700batch: iter_time=3.821e-04, forward_time=0.353, loss_ctc=71.847, loss_att=82.209, acc=0.628, loss=79.101, backward_time=0.446, grad_norm=30.166, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.104, optim0_lr0=4.286e-04, train_time=1.655
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2878933.0 ON gpub052 CANCELLED AT 2024-01-23T14:32:56 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.18.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.18.log
new file mode 100644
index 0000000000000000000000000000000000000000..f7cb3139711a195edc29081eda038455188bf749
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.18.log
@@ -0,0 +1,1991 @@
+# Running on gpub024.delta.ncsa.illinois.edu
+# Started at Sat Jan 20 18:27:57 CST 2024
+# SLURMD_NODENAME=gpub024
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2877332
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x8)'
+# SLURM_JOB_END_TIME=1705969637
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2877332
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[024,026-027,029,031,038,040,042]'
+# SLURM_JOB_NUM_NODES=8
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1705796837
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_NNODES=8
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[024,026-027,029,031,038,040,042]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login01.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x8)'
+# SLURM_TASK_PID=370659
+# SLURM_TOPOLOGY_ADDR=ss00.ss10.gpub024
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessi_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+ng_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_87808da5-0a77-452f-b2ef-79e88058766b
+[gpub024:0/32] 2024-01-20 18:37:38,742 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub024:0/32] 2024-01-20 18:37:53,683 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 32 nodes.
+[gpub024:0/32] 2024-01-20 18:37:53,755 (s2t:464) INFO: Vocabulary size: 50002
+[gpub024:0/32] 2024-01-20 18:38:03,970 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub024:0/32] 2024-01-20 18:38:03,976 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub024:0/32] 2024-01-20 18:38:03,976 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub024:0/32] 2024-01-20 18:38:03,976 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub024:0/32] 2024-01-20 18:38:03,978 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub024:0/32] 2024-01-20 18:38:09,356 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 18:38:10,279 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f63db07a380>)
+[gpub024:0/32] 2024-01-20 18:38:10,279 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub024:0/32] 2024-01-20 18:38:10,280 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+gpub024:370832:370832 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.124<0>
+gpub024:370832:370832 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub024:370832:370832 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub024:0/32] 2024-01-20 18:38:25,974 (trainer:284) INFO: 1/45epoch started
+[gpub024:0/32] 2024-01-20 18:38:26,023 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub024:0/32] 2024-01-20 18:38:44,427 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 18:38:48,114 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f61f29f09d0>)
+[gpub024:0/32] 2024-01-20 18:38:48,114 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub024:0/32] 2024-01-20 18:38:48,118 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub031:1194500:1194500 [3] NCCL INFO cudaDriverVersion 12020
+gpub031:1194500:1194500 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.131<0>
+gpub031:1194500:1194500 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub031:1194500:1194561 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.131<0>
+gpub031:1194500:1194561 [3] NCCL INFO Using network IB
+gpub031:1194500:1194561 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub031:1194500:1194561 [3] NCCL INFO Trees [0] -1/-1/-1->19->18 [1] -1/-1/-1->19->18
+gpub031:1194500:1194561 [3] NCCL INFO Channel 00/0 : 19[c7000] -> 20[7000] [send] via NET/IB/0
+gpub031:1194500:1194561 [3] NCCL INFO Channel 01/0 : 19[c7000] -> 20[7000] [send] via NET/IB/0
+gpub031:1194500:1194561 [3] NCCL INFO Connected all rings
+gpub031:1194500:1194561 [3] NCCL INFO Channel 00/0 : 19[c7000] -> 18[85000] via P2P/IPC
+gpub031:1194500:1194561 [3] NCCL INFO Channel 01/0 : 19[c7000] -> 18[85000] via P2P/IPC
+gpub031:1194500:1194561 [3] NCCL INFO Connected all trees
+gpub031:1194500:1194561 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub031:1194500:1194561 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub031:1194500:1194561 [3] NCCL INFO comm 0x144ee660 rank 19 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub031:1194498:1194498 [1] NCCL INFO cudaDriverVersion 12020
+gpub031:1194498:1194498 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.131<0>
+gpub031:1194498:1194498 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub031:1194498:1194562 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.131<0>
+gpub031:1194498:1194562 [1] NCCL INFO Using network IB
+gpub031:1194498:1194562 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub031:1194498:1194562 [1] NCCL INFO Trees [0] 18/8/-1->17->16 [1] 18/-1/-1->17->16
+gpub031:1194498:1194562 [1] NCCL INFO Channel 00/0 : 17[46000] -> 18[85000] via P2P/IPC
+gpub031:1194498:1194562 [1] NCCL INFO Channel 01/0 : 17[46000] -> 18[85000] via P2P/IPC
+gpub031:1194498:1194562 [1] NCCL INFO Connected all rings
+gpub031:1194498:1194562 [1] NCCL INFO Channel 00/0 : 8[7000] -> 17[46000] [receive] via NET/IB/0
+gpub031:1194498:1194562 [1] NCCL INFO Channel 00/0 : 17[46000] -> 8[7000] [send] via NET/IB/0
+gpub031:1194498:1194562 [1] NCCL INFO Channel 00/0 : 17[46000] -> 16[7000] via P2P/IPC
+gpub031:1194498:1194562 [1] NCCL INFO Channel 01/0 : 17[46000] -> 16[7000] via P2P/IPC
+gpub031:1194498:1194562 [1] NCCL INFO Connected all trees
+gpub031:1194498:1194562 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub031:1194498:1194562 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub031:1194498:1194562 [1] NCCL INFO comm 0x17154ae0 rank 17 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub031:1194497:1194497 [0] NCCL INFO cudaDriverVersion 12020
+gpub031:1194497:1194497 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.131<0>
+gpub031:1194497:1194497 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub031:1194497:1194564 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.131<0>
+gpub031:1194497:1194564 [0] NCCL INFO Using network IB
+gpub031:1194497:1194564 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub031:1194497:1194564 [0] NCCL INFO Trees [0] 17/24/-1->16->0 [1] 17/-1/-1->16->20
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 16[7000] [receive] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 16[7000] [receive] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 16[7000] -> 17[46000] via P2P/IPC
+gpub031:1194497:1194564 [0] NCCL INFO Channel 01/0 : 16[7000] -> 17[46000] via P2P/IPC
+gpub031:1194497:1194564 [0] NCCL INFO Connected all rings
+gpub031:1194497:1194564 [0] NCCL INFO Channel 01/0 : 16[7000] -> 20[7000] [send] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 16[7000] -> 24[7000] [send] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 0[7000] -> 16[7000] [receive] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 16[7000] -> 0[7000] [send] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 00/0 : 24[7000] -> 16[7000] [receive] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Channel 01/0 : 20[7000] -> 16[7000] [receive] via NET/IB/0
+gpub031:1194497:1194564 [0] NCCL INFO Connected all trees
+gpub031:1194497:1194564 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub031:1194497:1194564 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub031:1194497:1194564 [0] NCCL INFO comm 0x146eea40 rank 16 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub031:1194499:1194499 [2] NCCL INFO cudaDriverVersion 12020
+gpub031:1194499:1194499 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.131<0>
+gpub031:1194499:1194499 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub031:1194499:1194563 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.131<0>
+gpub031:1194499:1194563 [2] NCCL INFO Using network IB
+gpub031:1194499:1194563 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub031:1194499:1194563 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17
+gpub031:1194499:1194563 [2] NCCL INFO Channel 00/0 : 18[85000] -> 19[c7000] via P2P/IPC
+gpub031:1194499:1194563 [2] NCCL INFO Channel 01/0 : 18[85000] -> 19[c7000] via P2P/IPC
+gpub031:1194499:1194563 [2] NCCL INFO Connected all rings
+gpub031:1194499:1194563 [2] NCCL INFO Channel 00/0 : 18[85000] -> 17[46000] via P2P/IPC
+gpub031:1194499:1194563 [2] NCCL INFO Channel 01/0 : 18[85000] -> 17[46000] via P2P/IPC
+gpub031:1194499:1194563 [2] NCCL INFO Connected all trees
+gpub031:1194499:1194563 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub031:1194499:1194563 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub031:1194499:1194563 [2] NCCL INFO comm 0xe55c030 rank 18 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub026:3211453:3211453 [3] NCCL INFO cudaDriverVersion 12020
+gpub026:3211453:3211453 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.126<0>
+gpub026:3211453:3211453 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub026:3211453:3211510 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.126<0>
+gpub026:3211453:3211510 [3] NCCL INFO Using network IB
+gpub026:3211453:3211510 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub026:3211453:3211510 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub026:3211453:3211510 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/IB/0
+gpub026:3211453:3211510 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/IB/0
+gpub026:3211453:3211510 [3] NCCL INFO Connected all rings
+gpub026:3211453:3211510 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub026:3211453:3211510 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub026:3211453:3211510 [3] NCCL INFO Connected all trees
+gpub026:3211453:3211510 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub026:3211453:3211510 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub026:3211453:3211510 [3] NCCL INFO comm 0x1093d4c0 rank 7 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub026:3211452:3211452 [2] NCCL INFO cudaDriverVersion 12020
+gpub026:3211452:3211452 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.126<0>
+gpub026:3211452:3211452 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub026:3211452:3211513 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.126<0>
+gpub026:3211452:3211513 [2] NCCL INFO Using network IB
+gpub026:3211452:3211513 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub026:3211452:3211513 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub026:3211452:3211513 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub026:3211452:3211513 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub026:3211452:3211513 [2] NCCL INFO Connected all rings
+gpub026:3211452:3211513 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub026:3211452:3211513 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub026:3211452:3211513 [2] NCCL INFO Connected all trees
+gpub026:3211452:3211513 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub026:3211452:3211513 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub026:3211452:3211513 [2] NCCL INFO comm 0x14994570 rank 6 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub026:3211451:3211451 [1] NCCL INFO cudaDriverVersion 12020
+gpub026:3211451:3211451 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.126<0>
+gpub026:3211451:3211451 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub026:3211451:3211509 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.126<0>
+gpub026:3211451:3211509 [1] NCCL INFO Using network IB
+gpub026:3211451:3211509 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub026:3211451:3211509 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub026:3211451:3211509 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub026:3211451:3211509 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub026:3211451:3211509 [1] NCCL INFO Connected all rings
+gpub026:3211451:3211509 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/IB/0
+gpub026:3211451:3211509 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/IB/0
+gpub026:3211451:3211509 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub026:3211451:3211509 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub026:3211451:3211509 [1] NCCL INFO Connected all trees
+gpub026:3211451:3211509 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub026:3211451:3211509 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub026:3211451:3211509 [1] NCCL INFO comm 0x106512e0 rank 5 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub026:3211450:3211450 [0] NCCL INFO cudaDriverVersion 12020
+gpub026:3211450:3211450 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.126<0>
+gpub026:3211450:3211450 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub026:3211450:3211508 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.126<0>
+gpub026:3211450:3211508 [0] NCCL INFO Using network IB
+gpub026:3211450:3211508 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub026:3211450:3211508 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub026:3211450:3211508 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub026:3211450:3211508 [0] NCCL INFO Connected all rings
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/IB/0
+gpub026:3211450:3211508 [0] NCCL INFO Connected all trees
+gpub026:3211450:3211508 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub026:3211450:3211508 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub026:3211450:3211508 [0] NCCL INFO comm 0xff71720 rank 4 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub040:2133145:2133145 [2] NCCL INFO cudaDriverVersion 12020
+gpub040:2133145:2133145 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.140<0>
+gpub040:2133145:2133145 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub040:2133145:2133222 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.140<0>
+gpub040:2133145:2133222 [2] NCCL INFO Using network IB
+gpub040:2133145:2133222 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub040:2133145:2133222 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25
+gpub040:2133145:2133222 [2] NCCL INFO Channel 00/0 : 26[85000] -> 27[c7000] via P2P/IPC
+gpub040:2133145:2133222 [2] NCCL INFO Channel 01/0 : 26[85000] -> 27[c7000] via P2P/IPC
+gpub040:2133145:2133222 [2] NCCL INFO Connected all rings
+gpub040:2133145:2133222 [2] NCCL INFO Channel 00/0 : 26[85000] -> 25[46000] via P2P/IPC
+gpub040:2133145:2133222 [2] NCCL INFO Channel 01/0 : 26[85000] -> 25[46000] via P2P/IPC
+gpub040:2133145:2133222 [2] NCCL INFO Connected all trees
+gpub040:2133145:2133222 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub040:2133145:2133222 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub040:2133145:2133222 [2] NCCL INFO comm 0x104cb3f0 rank 26 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub029:521670:521670 [1] NCCL INFO cudaDriverVersion 12020
+gpub029:521670:521670 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.129<0>
+gpub029:521670:521670 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub029:521670:521730 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.129<0>
+gpub029:521670:521730 [1] NCCL INFO Using network IB
+gpub029:521670:521730 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub029:521670:521730 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/20/-1->13->12
+gpub029:521670:521730 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub029:521670:521730 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub029:521670:521730 [1] NCCL INFO Connected all rings
+gpub029:521670:521730 [1] NCCL INFO Channel 01/0 : 13[46000] -> 20[7000] [send] via NET/IB/0
+gpub029:521670:521730 [1] NCCL INFO Channel 01/0 : 20[7000] -> 13[46000] [receive] via NET/IB/0
+gpub029:521670:521730 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub029:521670:521730 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub029:521670:521730 [1] NCCL INFO Connected all trees
+gpub029:521670:521730 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub029:521670:521730 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub029:521670:521730 [1] NCCL INFO comm 0x1115f990 rank 13 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub029:521671:521671 [2] NCCL INFO cudaDriverVersion 12020
+gpub029:521671:521671 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.129<0>
+gpub029:521671:521671 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub029:521671:521732 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.129<0>
+gpub029:521671:521732 [2] NCCL INFO Using network IB
+gpub029:521671:521732 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub029:521671:521732 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub029:521671:521732 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub029:521671:521732 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub029:521671:521732 [2] NCCL INFO Connected all rings
+gpub029:521671:521732 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub029:521671:521732 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub029:521671:521732 [2] NCCL INFO Connected all trees
+gpub029:521671:521732 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub029:521671:521732 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub029:521671:521732 [2] NCCL INFO comm 0xf1ac6a0 rank 14 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub042:3340684:3340684 [1] NCCL INFO cudaDriverVersion 12020
+gpub042:3340684:3340684 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:3340684:3340684 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub042:3340684:3340737 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.142<0>
+gpub042:3340684:3340737 [1] NCCL INFO Using network IB
+gpub042:3340684:3340737 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub042:3340684:3340737 [1] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28
+gpub042:3340684:3340737 [1] NCCL INFO Channel 00/0 : 29[46000] -> 30[85000] via P2P/IPC
+gpub042:3340684:3340737 [1] NCCL INFO Channel 01/0 : 29[46000] -> 30[85000] via P2P/IPC
+gpub042:3340684:3340737 [1] NCCL INFO Connected all rings
+gpub042:3340684:3340737 [1] NCCL INFO Channel 00/0 : 29[46000] -> 28[7000] via P2P/IPC
+gpub042:3340684:3340737 [1] NCCL INFO Channel 01/0 : 29[46000] -> 28[7000] via P2P/IPC
+gpub042:3340684:3340737 [1] NCCL INFO Connected all trees
+gpub042:3340684:3340737 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub042:3340684:3340737 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:3340684:3340737 [1] NCCL INFO comm 0xf7b4e00 rank 29 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub029:521672:521672 [3] NCCL INFO cudaDriverVersion 12020
+gpub029:521672:521672 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.129<0>
+gpub029:521672:521672 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub029:521672:521731 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.129<0>
+gpub029:521672:521731 [3] NCCL INFO Using network IB
+gpub029:521672:521731 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub029:521672:521731 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub029:521672:521731 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 16[7000] [send] via NET/IB/0
+gpub029:521672:521731 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 16[7000] [send] via NET/IB/0
+gpub029:521672:521731 [3] NCCL INFO Connected all rings
+gpub029:521672:521731 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub029:521672:521731 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub029:521672:521731 [3] NCCL INFO Connected all trees
+gpub029:521672:521731 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub029:521672:521731 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub029:521672:521731 [3] NCCL INFO comm 0x1c3339f0 rank 15 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub038:1156124:1156124 [0] NCCL INFO cudaDriverVersion 12020
+gpub038:1156124:1156124 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.138<0>
+gpub038:1156124:1156124 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub038:1156124:1156219 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.138<0>
+gpub038:1156124:1156219 [0] NCCL INFO Using network IB
+gpub038:1156124:1156219 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub038:1156124:1156219 [0] NCCL INFO Trees [0] 21/-1/-1->20->25 [1] 21/16/-1->20->13
+gpub038:1156124:1156219 [0] NCCL INFO Channel 00/0 : 19[c7000] -> 20[7000] [receive] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 19[c7000] -> 20[7000] [receive] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 00/0 : 20[7000] -> 21[46000] via P2P/IPC
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 20[7000] -> 21[46000] via P2P/IPC
+gpub038:1156124:1156219 [0] NCCL INFO Connected all rings
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 16[7000] -> 20[7000] [receive] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 00/0 : 20[7000] -> 25[46000] [send] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 13[46000] -> 20[7000] [receive] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 20[7000] -> 13[46000] [send] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 00/0 : 25[46000] -> 20[7000] [receive] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Channel 01/0 : 20[7000] -> 16[7000] [send] via NET/IB/0
+gpub038:1156124:1156219 [0] NCCL INFO Connected all trees
+gpub038:1156124:1156219 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub038:1156124:1156219 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub038:1156124:1156219 [0] NCCL INFO comm 0x14d1c920 rank 20 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub038:1156126:1156126 [2] NCCL INFO cudaDriverVersion 12020
+gpub038:1156126:1156126 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.138<0>
+gpub038:1156126:1156126 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub038:1156126:1156220 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.138<0>
+gpub038:1156126:1156220 [2] NCCL INFO Using network IB
+gpub038:1156126:1156220 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub038:1156126:1156220 [2] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21
+gpub038:1156126:1156220 [2] NCCL INFO Channel 00/0 : 22[85000] -> 23[c7000] via P2P/IPC
+gpub038:1156126:1156220 [2] NCCL INFO Channel 01/0 : 22[85000] -> 23[c7000] via P2P/IPC
+gpub038:1156126:1156220 [2] NCCL INFO Connected all rings
+gpub038:1156126:1156220 [2] NCCL INFO Channel 00/0 : 22[85000] -> 21[46000] via P2P/IPC
+gpub038:1156126:1156220 [2] NCCL INFO Channel 01/0 : 22[85000] -> 21[46000] via P2P/IPC
+gpub038:1156126:1156220 [2] NCCL INFO Connected all trees
+gpub038:1156126:1156220 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub038:1156126:1156220 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub038:1156126:1156220 [2] NCCL INFO comm 0x17362910 rank 22 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub042:3340686:3340686 [3] NCCL INFO cudaDriverVersion 12020
+gpub042:3340686:3340686 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:3340686:3340686 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub042:3340686:3340739 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.142<0>
+gpub042:3340686:3340739 [3] NCCL INFO Using network IB
+gpub042:3340686:3340739 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub042:3340686:3340739 [3] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] -1/-1/-1->31->30
+gpub042:3340686:3340739 [3] NCCL INFO Channel 00/0 : 31[c7000] -> 0[7000] [send] via NET/IB/0
+gpub042:3340686:3340739 [3] NCCL INFO Channel 01/0 : 31[c7000] -> 0[7000] [send] via NET/IB/0
+gpub042:3340686:3340739 [3] NCCL INFO Connected all rings
+gpub042:3340686:3340739 [3] NCCL INFO Channel 00/0 : 31[c7000] -> 30[85000] via P2P/IPC
+gpub042:3340686:3340739 [3] NCCL INFO Channel 01/0 : 31[c7000] -> 30[85000] via P2P/IPC
+gpub042:3340686:3340739 [3] NCCL INFO Connected all trees
+gpub042:3340686:3340739 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub042:3340686:3340739 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:3340686:3340739 [3] NCCL INFO comm 0x163cf940 rank 31 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub040:2133144:2133144 [1] NCCL INFO cudaDriverVersion 12020
+gpub040:2133144:2133144 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.140<0>
+gpub040:2133144:2133144 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub040:2133144:2133218 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.140<0>
+gpub040:2133144:2133218 [1] NCCL INFO Using network IB
+gpub040:2133144:2133218 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub040:2133144:2133218 [1] NCCL INFO Trees [0] 26/20/-1->25->24 [1] 26/-1/-1->25->24
+gpub040:2133144:2133218 [1] NCCL INFO Channel 00/0 : 25[46000] -> 26[85000] via P2P/IPC
+gpub040:2133144:2133218 [1] NCCL INFO Channel 01/0 : 25[46000] -> 26[85000] via P2P/IPC
+gpub040:2133144:2133218 [1] NCCL INFO Connected all rings
+gpub040:2133144:2133218 [1] NCCL INFO Channel 00/0 : 20[7000] -> 25[46000] [receive] via NET/IB/0
+gpub040:2133144:2133218 [1] NCCL INFO Channel 00/0 : 25[46000] -> 20[7000] [send] via NET/IB/0
+gpub040:2133144:2133218 [1] NCCL INFO Channel 00/0 : 25[46000] -> 24[7000] via P2P/IPC
+gpub040:2133144:2133218 [1] NCCL INFO Channel 01/0 : 25[46000] -> 24[7000] via P2P/IPC
+gpub040:2133144:2133218 [1] NCCL INFO Connected all trees
+gpub040:2133144:2133218 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub040:2133144:2133218 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub040:2133144:2133218 [1] NCCL INFO comm 0x103eb6e0 rank 25 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub038:1156125:1156125 [1] NCCL INFO cudaDriverVersion 12020
+gpub038:1156125:1156125 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.138<0>
+gpub038:1156125:1156125 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub038:1156125:1156217 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.138<0>
+gpub038:1156125:1156217 [1] NCCL INFO Using network IB
+gpub038:1156125:1156217 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub038:1156125:1156217 [1] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/24/-1->21->20
+gpub038:1156125:1156217 [1] NCCL INFO Channel 00/0 : 21[46000] -> 22[85000] via P2P/IPC
+gpub038:1156125:1156217 [1] NCCL INFO Channel 01/0 : 21[46000] -> 22[85000] via P2P/IPC
+gpub038:1156125:1156217 [1] NCCL INFO Connected all rings
+gpub038:1156125:1156217 [1] NCCL INFO Channel 01/0 : 21[46000] -> 24[7000] [send] via NET/IB/0
+gpub038:1156125:1156217 [1] NCCL INFO Channel 01/0 : 24[7000] -> 21[46000] [receive] via NET/IB/0
+gpub038:1156125:1156217 [1] NCCL INFO Channel 00/0 : 21[46000] -> 20[7000] via P2P/IPC
+gpub038:1156125:1156217 [1] NCCL INFO Channel 01/0 : 21[46000] -> 20[7000] via P2P/IPC
+gpub038:1156125:1156217 [1] NCCL INFO Connected all trees
+gpub038:1156125:1156217 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub038:1156125:1156217 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub038:1156125:1156217 [1] NCCL INFO comm 0x15ca03c0 rank 21 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub027:248345:248345 [2] NCCL INFO cudaDriverVersion 12020
+gpub027:248345:248345 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.127<0>
+gpub027:248345:248345 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub027:248345:248512 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.127<0>
+gpub027:248345:248512 [2] NCCL INFO Using network IB
+gpub027:248345:248512 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub027:248345:248512 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub027:248345:248512 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub027:248345:248512 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub027:248345:248512 [2] NCCL INFO Connected all rings
+gpub027:248345:248512 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub027:248345:248512 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub027:248345:248512 [2] NCCL INFO Connected all trees
+gpub027:248345:248512 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub027:248345:248512 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub027:248345:248512 [2] NCCL INFO comm 0xef2d9a0 rank 10 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub040:2133146:2133146 [3] NCCL INFO cudaDriverVersion 12020
+gpub040:2133146:2133146 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.140<0>
+gpub040:2133146:2133146 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub040:2133146:2133219 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.140<0>
+gpub040:2133146:2133219 [3] NCCL INFO Using network IB
+gpub040:2133146:2133219 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub040:2133146:2133219 [3] NCCL INFO Trees [0] -1/-1/-1->27->26 [1] -1/-1/-1->27->26
+gpub040:2133146:2133219 [3] NCCL INFO Channel 00/0 : 27[c7000] -> 28[7000] [send] via NET/IB/0
+gpub040:2133146:2133219 [3] NCCL INFO Channel 01/0 : 27[c7000] -> 28[7000] [send] via NET/IB/0
+gpub040:2133146:2133219 [3] NCCL INFO Connected all rings
+gpub040:2133146:2133219 [3] NCCL INFO Channel 00/0 : 27[c7000] -> 26[85000] via P2P/IPC
+gpub040:2133146:2133219 [3] NCCL INFO Channel 01/0 : 27[c7000] -> 26[85000] via P2P/IPC
+gpub040:2133146:2133219 [3] NCCL INFO Connected all trees
+gpub040:2133146:2133219 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub040:2133146:2133219 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub040:2133146:2133219 [3] NCCL INFO comm 0x14c4db60 rank 27 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub027:248346:248346 [3] NCCL INFO cudaDriverVersion 12020
+gpub027:248346:248346 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.127<0>
+gpub027:248346:248346 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub027:248346:248511 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.127<0>
+gpub027:248346:248511 [3] NCCL INFO Using network IB
+gpub027:248346:248511 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub027:248346:248511 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub027:248346:248511 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/IB/0
+gpub027:248346:248511 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/IB/0
+gpub027:248346:248511 [3] NCCL INFO Connected all rings
+gpub027:248346:248511 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub027:248346:248511 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub027:248346:248511 [3] NCCL INFO Connected all trees
+gpub027:248346:248511 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub027:248346:248511 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub027:248346:248511 [3] NCCL INFO comm 0x1a3091a0 rank 11 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub029:521669:521669 [0] NCCL INFO cudaDriverVersion 12020
+gpub029:521669:521669 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.129<0>
+gpub029:521669:521669 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub029:521669:521734 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.129<0>
+gpub029:521669:521734 [0] NCCL INFO Using network IB
+gpub029:521669:521734 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub029:521669:521734 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->28
+gpub029:521669:521734 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub029:521669:521734 [0] NCCL INFO Connected all rings
+gpub029:521669:521734 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 28[7000] -> 12[7000] [receive] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 12[7000] -> 28[7000] [send] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/IB/0
+gpub029:521669:521734 [0] NCCL INFO Connected all trees
+gpub029:521669:521734 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub029:521669:521734 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub029:521669:521734 [0] NCCL INFO comm 0x11d80350 rank 12 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub027:248344:248344 [1] NCCL INFO cudaDriverVersion 12020
+gpub027:248344:248344 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.127<0>
+gpub027:248344:248344 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub027:248344:248509 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.127<0>
+gpub027:248344:248509 [1] NCCL INFO Using network IB
+gpub027:248344:248509 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub027:248344:248509 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub027:248344:248509 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub027:248344:248509 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub027:248344:248509 [1] NCCL INFO Connected all rings
+gpub027:248344:248509 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/IB/0
+gpub027:248344:248509 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/IB/0
+gpub027:248344:248509 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub027:248344:248509 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub027:248344:248509 [1] NCCL INFO Connected all trees
+gpub027:248344:248509 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub027:248344:248509 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub027:248344:248509 [1] NCCL INFO comm 0x12ad62e0 rank 9 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub040:2133143:2133143 [0] NCCL INFO cudaDriverVersion 12020
+gpub040:2133143:2133143 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.140<0>
+gpub040:2133143:2133143 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub040:2133143:2133224 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.140<0>
+gpub040:2133143:2133224 [0] NCCL INFO Using network IB
+gpub040:2133143:2133224 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub040:2133143:2133224 [0] NCCL INFO Trees [0] 25/28/-1->24->16 [1] 25/-1/-1->24->21
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 23[c7000] -> 24[7000] [receive] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 01/0 : 23[c7000] -> 24[7000] [receive] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 24[7000] -> 25[46000] via P2P/IPC
+gpub040:2133143:2133224 [0] NCCL INFO Channel 01/0 : 24[7000] -> 25[46000] via P2P/IPC
+gpub040:2133143:2133224 [0] NCCL INFO Connected all rings
+gpub040:2133143:2133224 [0] NCCL INFO Channel 01/0 : 21[46000] -> 24[7000] [receive] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 24[7000] -> 28[7000] [send] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 16[7000] -> 24[7000] [receive] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 24[7000] -> 16[7000] [send] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 00/0 : 28[7000] -> 24[7000] [receive] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Channel 01/0 : 24[7000] -> 21[46000] [send] via NET/IB/0
+gpub040:2133143:2133224 [0] NCCL INFO Connected all trees
+gpub040:2133143:2133224 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub040:2133143:2133224 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub040:2133143:2133224 [0] NCCL INFO comm 0x17e0fdc0 rank 24 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub027:248343:248343 [0] NCCL INFO cudaDriverVersion 12020
+gpub027:248343:248343 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.127<0>
+gpub027:248343:248343 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub027:248343:248510 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.127<0>
+gpub027:248343:248510 [0] NCCL INFO Using network IB
+gpub027:248343:248510 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub027:248343:248510 [0] NCCL INFO Trees [0] 9/12/-1->8->17 [1] 9/-1/-1->8->5
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub027:248343:248510 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub027:248343:248510 [0] NCCL INFO Connected all rings
+gpub027:248343:248510 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 8[7000] -> 17[46000] [send] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 17[46000] -> 8[7000] [receive] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/IB/0
+gpub027:248343:248510 [0] NCCL INFO Connected all trees
+gpub027:248343:248510 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub027:248343:248510 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub027:248343:248510 [0] NCCL INFO comm 0x1b7294c0 rank 8 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub038:1156127:1156127 [3] NCCL INFO cudaDriverVersion 12020
+gpub038:1156127:1156127 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.138<0>
+gpub038:1156127:1156127 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub038:1156127:1156218 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.138<0>
+gpub038:1156127:1156218 [3] NCCL INFO Using network IB
+gpub038:1156127:1156218 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub038:1156127:1156218 [3] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] -1/-1/-1->23->22
+gpub038:1156127:1156218 [3] NCCL INFO Channel 00/0 : 23[c7000] -> 24[7000] [send] via NET/IB/0
+gpub038:1156127:1156218 [3] NCCL INFO Channel 01/0 : 23[c7000] -> 24[7000] [send] via NET/IB/0
+gpub038:1156127:1156218 [3] NCCL INFO Connected all rings
+gpub038:1156127:1156218 [3] NCCL INFO Channel 00/0 : 23[c7000] -> 22[85000] via P2P/IPC
+gpub038:1156127:1156218 [3] NCCL INFO Channel 01/0 : 23[c7000] -> 22[85000] via P2P/IPC
+gpub038:1156127:1156218 [3] NCCL INFO Connected all trees
+gpub038:1156127:1156218 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub038:1156127:1156218 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub038:1156127:1156218 [3] NCCL INFO comm 0x1a2cf7e0 rank 23 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub042:3340683:3340683 [0] NCCL INFO cudaDriverVersion 12020
+gpub042:3340683:3340683 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:3340683:3340683 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub042:3340683:3340740 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.142<0>
+gpub042:3340683:3340740 [0] NCCL INFO Using network IB
+gpub042:3340683:3340740 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub042:3340683:3340740 [0] NCCL INFO Trees [0] 29/-1/-1->28->24 [1] 29/12/-1->28->-1
+gpub042:3340683:3340740 [0] NCCL INFO Channel 00/0 : 27[c7000] -> 28[7000] [receive] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Channel 01/0 : 27[c7000] -> 28[7000] [receive] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Channel 00/0 : 28[7000] -> 29[46000] via P2P/IPC
+gpub042:3340683:3340740 [0] NCCL INFO Channel 01/0 : 28[7000] -> 29[46000] via P2P/IPC
+gpub042:3340683:3340740 [0] NCCL INFO Connected all rings
+gpub042:3340683:3340740 [0] NCCL INFO Channel 00/0 : 24[7000] -> 28[7000] [receive] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Channel 01/0 : 12[7000] -> 28[7000] [receive] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Channel 01/0 : 28[7000] -> 12[7000] [send] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Channel 00/0 : 28[7000] -> 24[7000] [send] via NET/IB/0
+gpub042:3340683:3340740 [0] NCCL INFO Connected all trees
+gpub042:3340683:3340740 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub042:3340683:3340740 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:3340683:3340740 [0] NCCL INFO comm 0x1630c5f0 rank 28 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub042:3340685:3340685 [2] NCCL INFO cudaDriverVersion 12020
+gpub042:3340685:3340685 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:3340685:3340685 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub042:3340685:3340738 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.142<0>
+gpub042:3340685:3340738 [2] NCCL INFO Using network IB
+gpub042:3340685:3340738 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub042:3340685:3340738 [2] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29
+gpub042:3340685:3340738 [2] NCCL INFO Channel 00/0 : 30[85000] -> 31[c7000] via P2P/IPC
+gpub042:3340685:3340738 [2] NCCL INFO Channel 01/0 : 30[85000] -> 31[c7000] via P2P/IPC
+gpub042:3340685:3340738 [2] NCCL INFO Connected all rings
+gpub042:3340685:3340738 [2] NCCL INFO Channel 00/0 : 30[85000] -> 29[46000] via P2P/IPC
+gpub042:3340685:3340738 [2] NCCL INFO Channel 01/0 : 30[85000] -> 29[46000] via P2P/IPC
+gpub042:3340685:3340738 [2] NCCL INFO Connected all trees
+gpub042:3340685:3340738 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub042:3340685:3340738 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:3340685:3340738 [2] NCCL INFO comm 0x17ac17b0 rank 30 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+gpub024:370833:370833 [1] NCCL INFO cudaDriverVersion 12020
+gpub024:370833:370833 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.124<0>
+gpub024:370833:370833 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub024:370833:370886 [1] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.124<0>
+gpub024:370833:370886 [1] NCCL INFO Using network IB
+gpub024:370833:370886 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub024:370833:370886 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub024:370833:370886 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub024:370833:370886 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub024:370833:370886 [1] NCCL INFO Connected all rings
+gpub024:370833:370886 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub024:370833:370886 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub024:370833:370886 [1] NCCL INFO Connected all trees
+gpub024:370833:370886 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub024:370833:370886 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub024:370833:370886 [1] NCCL INFO comm 0x14718a70 rank 1 nranks 32 cudaDev 1 busId 46000 - Init COMPLETE
+gpub024:370832:370887 [0] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.124<0>
+gpub024:370832:370887 [0] NCCL INFO Using network IB
+gpub024:370832:370887 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub024:370832:370887 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+gpub024:370832:370887 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+gpub024:370832:370887 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] 1/-1/-1->0->4
+gpub024:370832:370887 [0] NCCL INFO Channel 00/0 : 31[c7000] -> 0[7000] [receive] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Channel 01/0 : 31[c7000] -> 0[7000] [receive] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub024:370832:370887 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub024:370832:370887 [0] NCCL INFO Connected all rings
+gpub024:370832:370887 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Channel 00/0 : 16[7000] -> 0[7000] [receive] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Channel 00/0 : 0[7000] -> 16[7000] [send] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/IB/0
+gpub024:370832:370887 [0] NCCL INFO Connected all trees
+gpub024:370832:370887 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub024:370832:370887 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub024:370832:370887 [0] NCCL INFO comm 0x1b1c4d20 rank 0 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+gpub024:370835:370835 [3] NCCL INFO cudaDriverVersion 12020
+gpub024:370835:370835 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.124<0>
+gpub024:370835:370835 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub024:370835:370888 [3] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.124<0>
+gpub024:370835:370888 [3] NCCL INFO Using network IB
+gpub024:370835:370888 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub024:370835:370888 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub024:370835:370888 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/IB/0
+gpub024:370835:370888 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/IB/0
+gpub024:370835:370888 [3] NCCL INFO Connected all rings
+gpub024:370835:370888 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub024:370835:370888 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub024:370835:370888 [3] NCCL INFO Connected all trees
+gpub024:370835:370888 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub024:370835:370888 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub024:370835:370888 [3] NCCL INFO comm 0x17ae9650 rank 3 nranks 32 cudaDev 3 busId c7000 - Init COMPLETE
+gpub024:370834:370834 [2] NCCL INFO cudaDriverVersion 12020
+gpub024:370834:370834 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.124<0>
+gpub024:370834:370834 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub024:370834:370885 [2] NCCL INFO NET/IB : Using [0]mlx5_0:1/RoCE [RO]; OOB eth1:172.28.23.124<0>
+gpub024:370834:370885 [2] NCCL INFO Using network IB
+gpub024:370834:370885 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub024:370834:370885 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub024:370834:370885 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub024:370834:370885 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub024:370834:370885 [2] NCCL INFO Connected all rings
+gpub024:370834:370885 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub024:370834:370885 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub024:370834:370885 [2] NCCL INFO Connected all trees
+gpub024:370834:370885 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+gpub024:370834:370885 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub024:370834:370885 [2] NCCL INFO comm 0x15a9a000 rank 2 nranks 32 cudaDev 2 busId 85000 - Init COMPLETE
+[gpub024:0/32] 2024-01-20 18:55:18,982 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub024:0/32] 2024-01-20 19:10:24,490 (trainer:737) INFO: 1epoch:train:1-100batch: iter_time=8.017, forward_time=5.067, loss_ctc=7.108e+03, loss_att=391.197, acc=1.527e-05, loss=2.406e+03, backward_time=1.579, grad_norm=1.413e+04, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.532, optim0_lr0=8.583e-08, train_time=19.173
+[gpub024:0/32] 2024-01-20 19:22:54,102 (trainer:737) INFO: 1epoch:train:101-200batch: iter_time=0.007, forward_time=3.200, loss_ctc=3.684e+03, loss_att=398.294, acc=1.718e-05, loss=1.384e+03, backward_time=1.665, grad_norm=1.747e+04, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.303, optim0_lr0=2.525e-07, train_time=7.505
+[gpub024:0/32] 2024-01-20 19:31:46,675 (trainer:737) INFO: 1epoch:train:201-300batch: iter_time=0.005, forward_time=1.924, loss_ctc=603.674, loss_att=431.352, acc=1.173e-05, loss=483.049, backward_time=0.938, grad_norm=2.342e+03, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.182, optim0_lr0=4.192e-07, train_time=5.328
+[gpub024:0/32] 2024-01-20 19:39:38,342 (trainer:737) INFO: 1epoch:train:301-400batch: iter_time=0.003, forward_time=1.344, loss_ctc=381.188, loss_att=373.060, acc=3.251e-05, loss=375.498, backward_time=0.993, grad_norm=252.988, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.179, optim0_lr0=5.858e-07, train_time=4.710
+[gpub024:0/32] 2024-01-20 19:45:26,480 (trainer:737) INFO: 1epoch:train:401-500batch: iter_time=0.001, forward_time=0.885, loss_ctc=392.954, loss_att=399.022, acc=2.600e-05, loss=397.201, backward_time=0.506, grad_norm=250.900, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.125, optim0_lr0=7.525e-07, train_time=3.485
+[gpub024:0/32] 2024-01-20 19:50:54,887 (trainer:737) INFO: 1epoch:train:501-600batch: iter_time=0.002, forward_time=0.709, loss_ctc=395.354, loss_att=398.357, acc=4.765e-05, loss=397.456, backward_time=0.481, grad_norm=228.791, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.097, optim0_lr0=9.192e-07, train_time=3.285
+[gpub024:0/32] 2024-01-20 19:55:50,223 (trainer:737) INFO: 1epoch:train:601-700batch: iter_time=6.128e-04, forward_time=0.391, loss_ctc=364.242, loss_att=393.167, acc=8.524e-05, loss=384.490, backward_time=0.439, grad_norm=201.715, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.075, optim0_lr0=1.086e-06, train_time=2.953
+[gpub024:0/32] 2024-01-20 20:01:13,591 (trainer:737) INFO: 1epoch:train:701-800batch: iter_time=8.934e-04, forward_time=0.984, loss_ctc=320.784, loss_att=343.284, acc=2.966e-04, loss=336.534, backward_time=0.588, grad_norm=159.260, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.134, optim0_lr0=1.252e-06, train_time=3.236
+[gpub024:0/32] 2024-01-20 20:07:24,330 (trainer:737) INFO: 1epoch:train:801-900batch: iter_time=0.002, forward_time=1.235, loss_ctc=378.305, loss_att=402.403, acc=8.029e-04, loss=395.173, backward_time=0.639, grad_norm=201.389, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.120, optim0_lr0=1.419e-06, train_time=3.707
+[gpub024:0/32] 2024-01-20 20:13:14,564 (trainer:737) INFO: 1epoch:train:901-1000batch: iter_time=0.002, forward_time=1.350, loss_ctc=349.828, loss_att=372.383, acc=0.004, loss=365.616, backward_time=0.619, grad_norm=185.687, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.112, optim0_lr0=1.586e-06, train_time=3.500
+[gpub024:0/32] 2024-01-20 20:19:08,728 (trainer:737) INFO: 1epoch:train:1001-1100batch: iter_time=0.004, forward_time=1.051, loss_ctc=359.903, loss_att=385.543, acc=0.018, loss=377.851, backward_time=0.544, grad_norm=160.993, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.140, optim0_lr0=1.752e-06, train_time=3.541
+[gpub024:0/32] 2024-01-20 20:25:02,408 (trainer:737) INFO: 1epoch:train:1101-1200batch: iter_time=0.005, forward_time=1.262, loss_ctc=375.377, loss_att=388.217, acc=0.035, loss=384.365, backward_time=0.672, grad_norm=191.985, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.166, optim0_lr0=1.919e-06, train_time=3.537
+[gpub024:0/32] 2024-01-20 20:28:26,831 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub024:0/32] 2024-01-20 20:28:45,839 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 20:28:49,381 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f61f29f1930>)
+[gpub024:0/32] 2024-01-20 20:28:49,382 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub024:0/32] 2024-01-20 20:28:49,385 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-20 20:41:25,726 (trainer:737) INFO: 1epoch:train:1201-1300batch: iter_time=6.545, forward_time=0.724, loss_ctc=348.215, loss_att=369.677, acc=0.050, loss=363.238, backward_time=0.501, grad_norm=161.461, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.108, optim0_lr0=2.086e-06, train_time=9.834
+[gpub024:0/32] 2024-01-20 20:45:38,679 (trainer:737) INFO: 1epoch:train:1301-1400batch: iter_time=9.093e-05, forward_time=0.325, loss_ctc=369.173, loss_att=417.840, acc=0.045, loss=403.240, backward_time=0.293, grad_norm=145.087, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.055, optim0_lr0=2.253e-06, train_time=2.528
+[gpub024:0/32] 2024-01-20 20:49:20,234 (trainer:737) INFO: 1epoch:train:1401-1500batch: iter_time=8.568e-05, forward_time=0.433, loss_ctc=356.292, loss_att=397.047, acc=0.043, loss=384.821, backward_time=0.360, grad_norm=154.715, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.077, optim0_lr0=2.419e-06, train_time=2.217
+[gpub024:0/32] 2024-01-20 20:52:29,993 (trainer:737) INFO: 1epoch:train:1501-1600batch: iter_time=9.064e-05, forward_time=0.467, loss_ctc=324.464, loss_att=363.911, acc=0.042, loss=352.077, backward_time=0.301, grad_norm=116.044, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.077, optim0_lr0=2.586e-06, train_time=1.897
+[gpub024:0/32] 2024-01-20 20:55:02,609 (trainer:737) INFO: 1epoch:train:1601-1700batch: iter_time=7.253e-04, forward_time=0.361, loss_ctc=349.119, loss_att=375.923, acc=0.044, loss=367.882, backward_time=0.351, grad_norm=135.732, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.080, optim0_lr0=2.752e-06, train_time=1.527
+[gpub024:0/32] 2024-01-20 20:57:19,221 (trainer:737) INFO: 1epoch:train:1701-1800batch: iter_time=8.372e-05, forward_time=0.415, loss_ctc=314.586, loss_att=345.320, acc=0.043, loss=336.100, backward_time=0.317, grad_norm=109.595, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.080, optim0_lr0=2.919e-06, train_time=1.363
+[gpub024:0/32] 2024-01-20 21:01:11,862 (trainer:737) INFO: 1epoch:train:1801-1900batch: iter_time=8.007e-04, forward_time=0.435, loss_ctc=324.254, loss_att=353.172, acc=0.055, loss=344.497, backward_time=0.265, grad_norm=138.683, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.079, optim0_lr0=3.086e-06, train_time=2.329
+[gpub024:0/32] 2024-01-20 21:03:41,235 (trainer:737) INFO: 1epoch:train:1901-2000batch: iter_time=4.039e-04, forward_time=0.258, loss_ctc=320.499, loss_att=352.307, acc=0.066, loss=342.764, backward_time=0.240, grad_norm=115.747, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.059, optim0_lr0=3.253e-06, train_time=1.491
+[gpub024:0/32] 2024-01-20 21:06:08,186 (trainer:737) INFO: 1epoch:train:2001-2100batch: iter_time=0.002, forward_time=0.366, loss_ctc=309.144, loss_att=334.432, acc=0.079, loss=326.846, backward_time=0.322, grad_norm=116.792, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.081, optim0_lr0=3.419e-06, train_time=1.471
+[gpub024:0/32] 2024-01-20 21:08:25,008 (trainer:737) INFO: 1epoch:train:2101-2200batch: iter_time=2.762e-04, forward_time=0.192, loss_ctc=308.376, loss_att=327.875, acc=0.101, loss=322.025, backward_time=0.241, grad_norm=104.830, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.055, optim0_lr0=3.586e-06, train_time=1.368
+[gpub024:0/32] 2024-01-20 21:11:23,364 (trainer:737) INFO: 1epoch:train:2201-2300batch: iter_time=0.001, forward_time=0.414, loss_ctc=299.648, loss_att=327.446, acc=0.095, loss=319.107, backward_time=0.305, grad_norm=103.609, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.065, optim0_lr0=3.753e-06, train_time=1.782
+[gpub024:0/32] 2024-01-20 21:14:03,227 (trainer:737) INFO: 1epoch:train:2301-2400batch: iter_time=3.871e-04, forward_time=0.439, loss_ctc=330.507, loss_att=357.165, acc=0.091, loss=349.168, backward_time=0.306, grad_norm=129.013, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.066, optim0_lr0=3.919e-06, train_time=1.599
+[gpub024:0/32] 2024-01-20 21:16:44,191 (trainer:737) INFO: 1epoch:train:2401-2500batch: iter_time=8.307e-05, forward_time=0.342, loss_ctc=320.025, loss_att=339.437, acc=0.097, loss=333.613, backward_time=0.317, grad_norm=103.409, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.065, optim0_lr0=4.086e-06, train_time=1.611
+[gpub024:0/32] 2024-01-20 21:17:04,238 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub024:0/32] 2024-01-20 21:17:23,142 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 21:17:26,703 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f61f29f3940>)
+[gpub024:0/32] 2024-01-20 21:17:26,703 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub024:0/32] 2024-01-20 21:17:26,706 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-20 21:36:04,615 (trainer:737) INFO: 1epoch:train:2501-2600batch: iter_time=6.291, forward_time=0.342, loss_ctc=301.069, loss_att=310.948, acc=0.109, loss=307.984, backward_time=0.275, grad_norm=108.040, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.066, optim0_lr0=4.253e-06, train_time=11.604
+[gpub024:0/32] 2024-01-20 21:38:59,998 (trainer:737) INFO: 1epoch:train:2601-2700batch: iter_time=8.703e-05, forward_time=0.304, loss_ctc=299.043, loss_att=314.306, acc=0.105, loss=309.727, backward_time=0.294, grad_norm=101.463, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.055, optim0_lr0=4.419e-06, train_time=1.754
+[gpub024:0/32] 2024-01-20 21:43:01,894 (trainer:737) INFO: 1epoch:train:2701-2800batch: iter_time=8.701e-05, forward_time=0.197, loss_ctc=316.568, loss_att=334.264, acc=0.107, loss=328.955, backward_time=0.248, grad_norm=96.280, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.055, optim0_lr0=4.586e-06, train_time=2.419
+[gpub024:0/32] 2024-01-20 21:46:01,444 (trainer:737) INFO: 1epoch:train:2801-2900batch: iter_time=8.414e-05, forward_time=0.441, loss_ctc=278.170, loss_att=285.617, acc=0.112, loss=283.383, backward_time=0.345, grad_norm=87.227, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.073, optim0_lr0=4.752e-06, train_time=1.795
+[gpub024:0/32] 2024-01-20 21:49:13,104 (trainer:737) INFO: 1epoch:train:2901-3000batch: iter_time=8.534e-05, forward_time=0.204, loss_ctc=299.573, loss_att=310.555, acc=0.106, loss=307.261, backward_time=0.226, grad_norm=100.655, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.050, optim0_lr0=4.919e-06, train_time=1.916
+[gpub024:0/32] 2024-01-20 21:52:25,305 (trainer:737) INFO: 1epoch:train:3001-3100batch: iter_time=8.473e-05, forward_time=0.347, loss_ctc=301.884, loss_att=306.786, acc=0.108, loss=305.315, backward_time=0.353, grad_norm=97.884, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.075, optim0_lr0=5.086e-06, train_time=1.922
+[gpub024:0/32] 2024-01-20 21:54:43,667 (trainer:737) INFO: 1epoch:train:3101-3200batch: iter_time=9.716e-05, forward_time=0.207, loss_ctc=290.382, loss_att=304.094, acc=0.116, loss=299.980, backward_time=0.260, grad_norm=105.457, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.050, optim0_lr0=5.252e-06, train_time=1.383
+[gpub024:0/32] 2024-01-20 21:57:53,427 (trainer:737) INFO: 1epoch:train:3201-3300batch: iter_time=9.168e-05, forward_time=0.229, loss_ctc=255.789, loss_att=266.257, acc=0.118, loss=263.117, backward_time=0.231, grad_norm=79.841, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.062, optim0_lr0=5.419e-06, train_time=1.898
+[gpub024:0/32] 2024-01-20 22:01:22,128 (trainer:737) INFO: 1epoch:train:3301-3400batch: iter_time=8.734e-05, forward_time=0.404, loss_ctc=302.288, loss_att=310.324, acc=0.111, loss=307.913, backward_time=0.344, grad_norm=106.725, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.073, optim0_lr0=5.586e-06, train_time=2.087
+[gpub024:0/32] 2024-01-20 22:03:43,082 (trainer:737) INFO: 1epoch:train:3401-3500batch: iter_time=8.881e-05, forward_time=0.183, loss_ctc=279.487, loss_att=286.844, acc=0.111, loss=284.637, backward_time=0.224, grad_norm=95.334, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.052, optim0_lr0=5.753e-06, train_time=1.408
+[gpub024:0/32] 2024-01-20 22:06:50,874 (trainer:737) INFO: 1epoch:train:3501-3600batch: iter_time=0.150, forward_time=0.184, loss_ctc=283.928, loss_att=296.965, acc=0.111, loss=293.054, backward_time=0.246, grad_norm=93.639, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.056, optim0_lr0=5.919e-06, train_time=1.878
+[gpub024:0/32] 2024-01-20 22:10:19,461 (trainer:737) INFO: 1epoch:train:3601-3700batch: iter_time=8.749e-05, forward_time=0.449, loss_ctc=300.761, loss_att=301.375, acc=0.111, loss=301.191, backward_time=0.293, grad_norm=99.802, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.065, optim0_lr0=6.086e-06, train_time=2.087
+[gpub024:0/32] 2024-01-20 22:12:08,052 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub024:0/32] 2024-01-20 22:12:26,700 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 22:12:30,288 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5894204100>)
+[gpub024:0/32] 2024-01-20 22:12:30,288 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub024:0/32] 2024-01-20 22:12:30,394 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-20 22:24:22,176 (trainer:737) INFO: 1epoch:train:3701-3800batch: iter_time=6.818, forward_time=0.171, loss_ctc=289.721, loss_att=289.438, acc=0.119, loss=289.523, backward_time=0.227, grad_norm=102.882, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.051, optim0_lr0=6.253e-06, train_time=8.427
+[gpub024:0/32] 2024-01-20 22:26:43,907 (trainer:737) INFO: 1epoch:train:3801-3900batch: iter_time=8.849e-05, forward_time=0.186, loss_ctc=300.682, loss_att=334.632, acc=0.105, loss=324.447, backward_time=0.230, grad_norm=100.017, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.051, optim0_lr0=6.419e-06, train_time=1.417
+[gpub024:0/32] 2024-01-20 22:29:28,883 (trainer:737) INFO: 1epoch:train:3901-4000batch: iter_time=9.533e-05, forward_time=0.389, loss_ctc=296.858, loss_att=317.228, acc=0.111, loss=311.117, backward_time=0.316, grad_norm=99.470, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.078, optim0_lr0=6.586e-06, train_time=1.650
+[gpub024:0/32] 2024-01-20 22:33:27,432 (trainer:737) INFO: 1epoch:train:4001-4100batch: iter_time=9.802e-05, forward_time=0.229, loss_ctc=266.485, loss_att=291.302, acc=0.116, loss=283.857, backward_time=0.275, grad_norm=84.792, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=6.752e-06, train_time=2.385
+[gpub024:0/32] 2024-01-20 22:35:25,892 (trainer:737) INFO: 1epoch:train:4101-4200batch: iter_time=1.029e-04, forward_time=0.164, loss_ctc=298.556, loss_att=312.756, acc=0.116, loss=308.496, backward_time=0.226, grad_norm=100.008, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=6.919e-06, train_time=1.185
+[gpub024:0/32] 2024-01-20 22:37:19,155 (trainer:737) INFO: 1epoch:train:4201-4300batch: iter_time=9.227e-05, forward_time=0.182, loss_ctc=265.312, loss_att=290.080, acc=0.115, loss=282.649, backward_time=0.226, grad_norm=107.829, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=7.086e-06, train_time=1.132
+[gpub024:0/32] 2024-01-20 22:39:32,458 (trainer:737) INFO: 1epoch:train:4301-4400batch: iter_time=8.702e-05, forward_time=0.369, loss_ctc=286.412, loss_att=305.438, acc=0.125, loss=299.730, backward_time=0.313, grad_norm=104.624, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.078, optim0_lr0=7.253e-06, train_time=1.332
+[gpub024:0/32] 2024-01-20 22:41:53,325 (trainer:737) INFO: 1epoch:train:4401-4500batch: iter_time=8.774e-05, forward_time=0.211, loss_ctc=279.545, loss_att=305.048, acc=0.118, loss=297.397, backward_time=0.273, grad_norm=98.376, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=7.419e-06, train_time=1.409
+[gpub024:0/32] 2024-01-20 22:43:49,902 (trainer:737) INFO: 1epoch:train:4501-4600batch: iter_time=8.875e-05, forward_time=0.161, loss_ctc=277.268, loss_att=294.900, acc=0.125, loss=289.610, backward_time=0.229, grad_norm=100.884, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=7.586e-06, train_time=1.165
+[gpub024:0/32] 2024-01-20 22:47:57,064 (trainer:737) INFO: 1epoch:train:4601-4700batch: iter_time=1.025e-04, forward_time=0.505, loss_ctc=281.201, loss_att=291.085, acc=0.126, loss=288.119, backward_time=0.315, grad_norm=103.691, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.078, optim0_lr0=7.753e-06, train_time=2.472
+[gpub024:0/32] 2024-01-20 22:50:01,404 (trainer:737) INFO: 1epoch:train:4701-4800batch: iter_time=9.262e-05, forward_time=0.181, loss_ctc=269.646, loss_att=291.312, acc=0.120, loss=284.812, backward_time=0.235, grad_norm=95.532, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=7.919e-06, train_time=1.243
+[gpub024:0/32] 2024-01-20 22:52:31,972 (trainer:737) INFO: 1epoch:train:4801-4900batch: iter_time=8.772e-05, forward_time=0.162, loss_ctc=300.142, loss_att=321.756, acc=0.117, loss=315.272, backward_time=0.225, grad_norm=116.778, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=8.086e-06, train_time=1.504
+[gpub024:0/32] 2024-01-20 22:54:31,936 (trainer:737) INFO: 1epoch:train:4901-5000batch: iter_time=2.798e-04, forward_time=0.246, loss_ctc=291.556, loss_att=305.632, acc=0.127, loss=301.410, backward_time=0.249, grad_norm=119.888, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.063, optim0_lr0=8.252e-06, train_time=1.201
+[gpub024:0/32] 2024-01-20 22:54:51,965 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub024:0/32] 2024-01-20 22:55:11,321 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 22:55:14,816 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f583646ff40>)
+[gpub024:0/32] 2024-01-20 22:55:14,816 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub024:0/32] 2024-01-20 22:55:14,819 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-20 23:09:21,101 (trainer:737) INFO: 1epoch:train:5001-5100batch: iter_time=6.748, forward_time=0.163, loss_ctc=280.352, loss_att=288.171, acc=0.135, loss=285.825, backward_time=0.225, grad_norm=109.360, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.051, optim0_lr0=8.419e-06, train_time=8.891
+[gpub024:0/32] 2024-01-20 23:10:46,449 (trainer:737) INFO: 1epoch:train:5101-5200batch: iter_time=1.039e-04, forward_time=0.161, loss_ctc=281.041, loss_att=293.839, acc=0.120, loss=289.999, backward_time=0.228, grad_norm=113.574, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.051, optim0_lr0=8.586e-06, train_time=0.853
+[gpub024:0/32] 2024-01-20 23:14:22,646 (trainer:737) INFO: 1epoch:train:5201-5300batch: iter_time=9.378e-05, forward_time=0.250, loss_ctc=295.798, loss_att=309.933, acc=0.126, loss=305.692, backward_time=0.283, grad_norm=116.875, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.062, optim0_lr0=8.753e-06, train_time=2.162
+[gpub024:0/32] 2024-01-20 23:16:53,960 (trainer:737) INFO: 1epoch:train:5301-5400batch: iter_time=4.429e-04, forward_time=0.304, loss_ctc=262.365, loss_att=268.039, acc=0.138, loss=266.337, backward_time=0.247, grad_norm=92.865, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.052, optim0_lr0=8.919e-06, train_time=1.513
+[gpub024:0/32] 2024-01-20 23:20:54,219 (trainer:737) INFO: 1epoch:train:5401-5500batch: iter_time=9.416e-05, forward_time=0.164, loss_ctc=285.595, loss_att=294.133, acc=0.130, loss=291.571, backward_time=0.224, grad_norm=115.563, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=9.086e-06, train_time=2.403
+[gpub024:0/32] 2024-01-20 23:22:37,100 (trainer:737) INFO: 1epoch:train:5501-5600batch: iter_time=9.800e-05, forward_time=0.161, loss_ctc=292.112, loss_att=293.890, acc=0.135, loss=293.357, backward_time=0.226, grad_norm=138.736, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.051, optim0_lr0=9.252e-06, train_time=1.029
+[gpub024:0/32] 2024-01-20 23:25:14,354 (trainer:737) INFO: 1epoch:train:5601-5700batch: iter_time=9.905e-05, forward_time=0.238, loss_ctc=281.139, loss_att=291.522, acc=0.134, loss=288.407, backward_time=0.231, grad_norm=120.229, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.056, optim0_lr0=9.419e-06, train_time=1.572
+[gpub024:0/32] 2024-01-20 23:27:32,104 (trainer:737) INFO: 1epoch:train:5701-5800batch: iter_time=9.042e-05, forward_time=0.160, loss_ctc=249.249, loss_att=256.116, acc=0.144, loss=254.056, backward_time=0.224, grad_norm=121.123, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.051, optim0_lr0=9.586e-06, train_time=1.377
+[gpub024:0/32] 2024-01-20 23:29:51,600 (trainer:737) INFO: 1epoch:train:5801-5900batch: iter_time=9.051e-05, forward_time=0.161, loss_ctc=293.006, loss_att=296.923, acc=0.136, loss=295.748, backward_time=0.225, grad_norm=128.273, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=9.753e-06, train_time=1.395
+[gpub024:0/32] 2024-01-20 23:32:09,015 (trainer:737) INFO: 1epoch:train:5901-6000batch: iter_time=9.189e-05, forward_time=0.160, loss_ctc=272.142, loss_att=275.457, acc=0.137, loss=274.462, backward_time=0.226, grad_norm=125.595, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.050, optim0_lr0=9.919e-06, train_time=1.374
+[gpub024:0/32] 2024-01-20 23:34:29,290 (trainer:737) INFO: 1epoch:train:6001-6100batch: iter_time=9.056e-05, forward_time=0.328, loss_ctc=276.483, loss_att=282.505, acc=0.137, loss=280.699, backward_time=0.313, grad_norm=184.316, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.073, optim0_lr0=1.009e-05, train_time=1.402
+[gpub024:0/32] 2024-01-20 23:38:53,601 (trainer:737) INFO: 1epoch:train:6101-6200batch: iter_time=9.595e-05, forward_time=0.232, loss_ctc=293.494, loss_att=286.853, acc=0.140, loss=288.845, backward_time=0.247, grad_norm=150.799, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.050, optim0_lr0=1.025e-05, train_time=2.643
+[gpub024:0/32] 2024-01-20 23:40:24,738 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub024:0/32] 2024-01-20 23:40:44,202 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-20 23:40:47,822 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f589424a8c0>)
+[gpub024:0/32] 2024-01-20 23:40:47,822 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub024:0/32] 2024-01-20 23:40:47,829 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 00:01:51,031 (trainer:737) INFO: 1epoch:train:6201-6300batch: iter_time=6.556, forward_time=0.231, loss_ctc=285.457, loss_att=276.496, acc=0.144, loss=279.184, backward_time=0.229, grad_norm=122.960, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.054, optim0_lr0=1.042e-05, train_time=13.775
+[gpub024:0/32] 2024-01-21 00:03:47,863 (trainer:737) INFO: 1epoch:train:6301-6400batch: iter_time=7.840e-05, forward_time=0.161, loss_ctc=294.854, loss_att=299.251, acc=0.133, loss=297.932, backward_time=0.226, grad_norm=145.793, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.050, optim0_lr0=1.059e-05, train_time=1.168
+[gpub024:0/32] 2024-01-21 00:06:11,304 (trainer:737) INFO: 1epoch:train:6401-6500batch: iter_time=8.345e-05, forward_time=0.161, loss_ctc=287.621, loss_att=297.049, acc=0.138, loss=294.221, backward_time=0.226, grad_norm=156.545, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.075e-05, train_time=1.434
+[gpub024:0/32] 2024-01-21 00:08:43,939 (trainer:737) INFO: 1epoch:train:6501-6600batch: iter_time=8.770e-05, forward_time=0.160, loss_ctc=259.478, loss_att=259.336, acc=0.146, loss=259.379, backward_time=0.226, grad_norm=104.308, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.092e-05, train_time=1.526
+[gpub024:0/32] 2024-01-21 00:10:41,126 (trainer:737) INFO: 1epoch:train:6601-6700batch: iter_time=8.612e-05, forward_time=0.161, loss_ctc=291.348, loss_att=285.394, acc=0.141, loss=287.180, backward_time=0.226, grad_norm=146.355, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.109e-05, train_time=1.172
+[gpub024:0/32] 2024-01-21 00:13:41,365 (trainer:737) INFO: 1epoch:train:6701-6800batch: iter_time=9.318e-05, forward_time=0.349, loss_ctc=259.395, loss_att=255.467, acc=0.144, loss=256.645, backward_time=0.295, grad_norm=132.878, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.067, optim0_lr0=1.125e-05, train_time=1.802
+[gpub024:0/32] 2024-01-21 00:15:23,595 (trainer:737) INFO: 1epoch:train:6801-6900batch: iter_time=9.421e-05, forward_time=0.165, loss_ctc=281.879, loss_att=279.304, acc=0.148, loss=280.077, backward_time=0.230, grad_norm=152.878, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.142e-05, train_time=1.022
+[gpub024:0/32] 2024-01-21 00:17:12,947 (trainer:737) INFO: 1epoch:train:6901-7000batch: iter_time=8.625e-05, forward_time=0.213, loss_ctc=275.708, loss_att=276.118, acc=0.142, loss=275.995, backward_time=0.230, grad_norm=149.103, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.058, optim0_lr0=1.159e-05, train_time=1.093
+[gpub024:0/32] 2024-01-21 00:19:05,440 (trainer:737) INFO: 1epoch:train:7001-7100batch: iter_time=8.654e-05, forward_time=0.161, loss_ctc=273.671, loss_att=268.499, acc=0.147, loss=270.050, backward_time=0.225, grad_norm=153.642, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.050, optim0_lr0=1.175e-05, train_time=1.125
+[gpub024:0/32] 2024-01-21 00:20:59,795 (trainer:737) INFO: 1epoch:train:7101-7200batch: iter_time=9.776e-05, forward_time=0.161, loss_ctc=277.449, loss_att=269.663, acc=0.151, loss=271.999, backward_time=0.223, grad_norm=129.304, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.050, optim0_lr0=1.192e-05, train_time=1.143
+[gpub024:0/32] 2024-01-21 00:23:01,780 (trainer:737) INFO: 1epoch:train:7201-7300batch: iter_time=9.702e-05, forward_time=0.161, loss_ctc=264.548, loss_att=264.024, acc=0.147, loss=264.181, backward_time=0.225, grad_norm=128.648, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.209e-05, train_time=1.220
+[gpub024:0/32] 2024-01-21 00:25:09,228 (trainer:737) INFO: 1epoch:train:7301-7400batch: iter_time=8.854e-05, forward_time=0.162, loss_ctc=294.659, loss_att=280.789, acc=0.144, loss=284.950, backward_time=0.226, grad_norm=124.475, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.225e-05, train_time=1.274
+[gpub024:0/32] 2024-01-21 00:28:16,872 (trainer:737) INFO: 1epoch:train:7401-7500batch: iter_time=9.139e-05, forward_time=0.428, loss_ctc=288.436, loss_att=273.891, acc=0.144, loss=278.254, backward_time=0.319, grad_norm=166.369, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.068, optim0_lr0=1.242e-05, train_time=1.876
+[gpub024:0/32] 2024-01-21 00:28:36,900 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub024:0/32] 2024-01-21 00:28:55,574 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 00:28:59,087 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58364a44c0>)
+[gpub024:0/32] 2024-01-21 00:28:59,087 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub024:0/32] 2024-01-21 00:28:59,248 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 00:40:21,220 (trainer:737) INFO: 1epoch:train:7501-7600batch: iter_time=6.127, forward_time=0.230, loss_ctc=277.883, loss_att=283.573, acc=0.154, loss=281.866, backward_time=0.230, grad_norm=121.577, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.053, optim0_lr0=1.259e-05, train_time=7.243
+[gpub024:0/32] 2024-01-21 00:42:24,844 (trainer:737) INFO: 1epoch:train:7601-7700batch: iter_time=8.879e-05, forward_time=0.162, loss_ctc=277.535, loss_att=286.191, acc=0.141, loss=283.594, backward_time=0.227, grad_norm=126.737, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.275e-05, train_time=1.236
+[gpub024:0/32] 2024-01-21 00:45:21,985 (trainer:737) INFO: 1epoch:train:7701-7800batch: iter_time=8.881e-05, forward_time=0.187, loss_ctc=289.913, loss_att=292.983, acc=0.148, loss=292.062, backward_time=0.225, grad_norm=121.086, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.051, optim0_lr0=1.292e-05, train_time=1.771
+[gpub024:0/32] 2024-01-21 00:47:55,822 (trainer:737) INFO: 1epoch:train:7801-7900batch: iter_time=9.194e-05, forward_time=0.419, loss_ctc=257.546, loss_att=263.915, acc=0.155, loss=262.005, backward_time=0.319, grad_norm=111.737, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.071, optim0_lr0=1.309e-05, train_time=1.538
+[gpub024:0/32] 2024-01-21 00:50:18,405 (trainer:737) INFO: 1epoch:train:7901-8000batch: iter_time=9.613e-05, forward_time=0.203, loss_ctc=282.081, loss_att=286.559, acc=0.145, loss=285.216, backward_time=0.235, grad_norm=131.767, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.058, optim0_lr0=1.325e-05, train_time=1.425
+[gpub024:0/32] 2024-01-21 00:51:46,881 (trainer:737) INFO: 1epoch:train:8001-8100batch: iter_time=9.102e-05, forward_time=0.163, loss_ctc=288.654, loss_att=290.892, acc=0.150, loss=290.221, backward_time=0.229, grad_norm=161.444, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.051, optim0_lr0=1.342e-05, train_time=0.885
+[gpub024:0/32] 2024-01-21 00:54:42,155 (trainer:737) INFO: 1epoch:train:8101-8200batch: iter_time=8.625e-05, forward_time=0.161, loss_ctc=279.314, loss_att=279.520, acc=0.151, loss=279.458, backward_time=0.224, grad_norm=216.407, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.050, optim0_lr0=1.359e-05, train_time=1.753
+[gpub024:0/32] 2024-01-21 00:58:11,092 (trainer:737) INFO: 1epoch:train:8201-8300batch: iter_time=8.959e-05, forward_time=0.300, loss_ctc=243.958, loss_att=251.134, acc=0.160, loss=248.981, backward_time=0.295, grad_norm=127.653, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.066, optim0_lr0=1.375e-05, train_time=2.089
+[gpub024:0/32] 2024-01-21 00:59:59,771 (trainer:737) INFO: 1epoch:train:8301-8400batch: iter_time=8.893e-05, forward_time=0.320, loss_ctc=288.230, loss_att=282.975, acc=0.155, loss=284.552, backward_time=0.267, grad_norm=155.130, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.066, optim0_lr0=1.392e-05, train_time=1.087
+[gpub024:0/32] 2024-01-21 01:02:11,491 (trainer:737) INFO: 1epoch:train:8401-8500batch: iter_time=8.713e-05, forward_time=0.161, loss_ctc=265.490, loss_att=264.901, acc=0.156, loss=265.077, backward_time=0.227, grad_norm=126.736, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.050, optim0_lr0=1.409e-05, train_time=1.315
+[gpub024:0/32] 2024-01-21 01:06:06,000 (trainer:737) INFO: 1epoch:train:8501-8600batch: iter_time=8.672e-05, forward_time=0.163, loss_ctc=273.922, loss_att=278.069, acc=0.152, loss=276.825, backward_time=0.225, grad_norm=141.164, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.050, optim0_lr0=1.425e-05, train_time=2.347
+[gpub024:0/32] 2024-01-21 01:07:53,740 (trainer:737) INFO: 1epoch:train:8601-8700batch: iter_time=9.219e-05, forward_time=0.185, loss_ctc=290.834, loss_att=281.552, acc=0.156, loss=284.337, backward_time=0.231, grad_norm=160.741, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.050, optim0_lr0=1.442e-05, train_time=1.078
+[gpub024:0/32] 2024-01-21 01:10:32,057 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub024:0/32] 2024-01-21 01:11:00,342 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 01:11:04,637 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f587d26a950>)
+[gpub024:0/32] 2024-01-21 01:11:04,637 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub024:0/32] 2024-01-21 01:11:04,641 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 01:34:03,599 (trainer:737) INFO: 1epoch:train:8701-8800batch: iter_time=7.727, forward_time=0.397, loss_ctc=278.410, loss_att=262.287, acc=0.162, loss=267.124, backward_time=0.290, grad_norm=128.774, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.068, optim0_lr0=1.459e-05, train_time=15.698
+[gpub024:0/32] 2024-01-21 01:35:54,043 (trainer:737) INFO: 1epoch:train:8801-8900batch: iter_time=8.431e-05, forward_time=0.318, loss_ctc=289.419, loss_att=290.680, acc=0.149, loss=290.302, backward_time=0.289, grad_norm=131.349, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.068, optim0_lr0=1.475e-05, train_time=1.105
+[gpub024:0/32] 2024-01-21 01:38:57,268 (trainer:737) INFO: 1epoch:train:8901-9000batch: iter_time=8.220e-05, forward_time=0.374, loss_ctc=280.590, loss_att=280.118, acc=0.157, loss=280.259, backward_time=0.260, grad_norm=144.201, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.068, optim0_lr0=1.492e-05, train_time=1.831
+[gpub024:0/32] 2024-01-21 01:42:42,364 (trainer:737) INFO: 1epoch:train:9001-9100batch: iter_time=8.654e-05, forward_time=0.426, loss_ctc=256.042, loss_att=253.245, acc=0.164, loss=254.084, backward_time=0.362, grad_norm=124.603, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.062, optim0_lr0=1.509e-05, train_time=2.252
+[gpub024:0/32] 2024-01-21 01:45:04,495 (trainer:737) INFO: 1epoch:train:9101-9200batch: iter_time=8.543e-05, forward_time=0.270, loss_ctc=286.072, loss_att=273.971, acc=0.160, loss=277.602, backward_time=0.264, grad_norm=141.570, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.064, optim0_lr0=1.525e-05, train_time=1.421
+[gpub024:0/32] 2024-01-21 01:47:44,070 (trainer:737) INFO: 1epoch:train:9201-9300batch: iter_time=8.352e-05, forward_time=0.337, loss_ctc=255.778, loss_att=249.880, acc=0.158, loss=251.649, backward_time=0.305, grad_norm=142.729, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.060, optim0_lr0=1.542e-05, train_time=1.596
+[gpub024:0/32] 2024-01-21 01:50:08,293 (trainer:737) INFO: 1epoch:train:9301-9400batch: iter_time=9.111e-05, forward_time=0.234, loss_ctc=277.610, loss_att=271.383, acc=0.167, loss=273.251, backward_time=0.255, grad_norm=175.939, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.064, optim0_lr0=1.559e-05, train_time=1.442
+[gpub024:0/32] 2024-01-21 01:54:10,177 (trainer:737) INFO: 1epoch:train:9401-9500batch: iter_time=9.213e-05, forward_time=0.397, loss_ctc=274.070, loss_att=270.063, acc=0.157, loss=271.265, backward_time=0.357, grad_norm=207.861, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.078, optim0_lr0=1.575e-05, train_time=2.417
+[gpub024:0/32] 2024-01-21 01:56:10,615 (trainer:737) INFO: 1epoch:train:9501-9600batch: iter_time=9.055e-05, forward_time=0.194, loss_ctc=268.234, loss_att=261.396, acc=0.165, loss=263.448, backward_time=0.263, grad_norm=135.097, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.061, optim0_lr0=1.592e-05, train_time=1.206
+[gpub024:0/32] 2024-01-21 01:59:07,664 (trainer:737) INFO: 1epoch:train:9601-9700batch: iter_time=7.795e-04, forward_time=0.536, loss_ctc=272.750, loss_att=258.889, acc=0.168, loss=263.048, backward_time=0.329, grad_norm=163.494, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.064, optim0_lr0=1.609e-05, train_time=1.768
+[gpub024:0/32] 2024-01-21 02:01:58,862 (trainer:737) INFO: 1epoch:train:9701-9800batch: iter_time=8.461e-05, forward_time=0.189, loss_ctc=258.195, loss_att=253.960, acc=0.163, loss=255.231, backward_time=0.232, grad_norm=128.764, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.050, optim0_lr0=1.625e-05, train_time=1.714
+[gpub024:0/32] 2024-01-21 02:06:03,753 (trainer:737) INFO: 1epoch:train:9801-9900batch: iter_time=0.002, forward_time=0.458, loss_ctc=290.756, loss_att=281.045, acc=0.159, loss=283.958, backward_time=0.318, grad_norm=130.748, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.081, optim0_lr0=1.642e-05, train_time=2.449
+[gpub024:0/32] 2024-01-21 02:07:33,302 (trainer:737) INFO: 1epoch:train:9901-10000batch: iter_time=8.674e-05, forward_time=0.196, loss_ctc=283.997, loss_att=267.150, acc=0.166, loss=272.204, backward_time=0.254, grad_norm=154.014, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.051, optim0_lr0=1.659e-05, train_time=0.894
+[gpub024:0/32] 2024-01-21 02:07:53,641 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub024:0/32] 2024-01-21 02:08:12,273 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 02:08:15,869 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5836af8670>)
+[gpub024:0/32] 2024-01-21 02:08:15,869 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub024:0/32] 2024-01-21 02:08:15,872 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 02:21:07,091 (trainer:737) INFO: 1epoch:train:10001-10100batch: iter_time=6.605, forward_time=0.161, loss_ctc=273.675, loss_att=256.946, acc=0.169, loss=261.965, backward_time=0.226, grad_norm=133.749, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.675e-05, train_time=8.139
+[gpub024:0/32] 2024-01-21 02:23:15,058 (trainer:737) INFO: 1epoch:train:10101-10200batch: iter_time=8.829e-05, forward_time=0.161, loss_ctc=272.759, loss_att=259.320, acc=0.156, loss=263.352, backward_time=0.226, grad_norm=150.880, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.692e-05, train_time=1.279
+[gpub024:0/32] 2024-01-21 02:26:38,602 (trainer:737) INFO: 1epoch:train:10201-10300batch: iter_time=8.969e-05, forward_time=0.164, loss_ctc=286.823, loss_att=272.190, acc=0.164, loss=276.580, backward_time=0.226, grad_norm=138.208, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.709e-05, train_time=2.035
+[gpub024:0/32] 2024-01-21 02:28:59,593 (trainer:737) INFO: 1epoch:train:10301-10400batch: iter_time=9.324e-05, forward_time=0.160, loss_ctc=251.876, loss_att=232.726, acc=0.172, loss=238.471, backward_time=0.224, grad_norm=129.368, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.725e-05, train_time=1.410
+[gpub024:0/32] 2024-01-21 02:31:53,662 (trainer:737) INFO: 1epoch:train:10401-10500batch: iter_time=8.964e-05, forward_time=0.312, loss_ctc=278.952, loss_att=258.765, acc=0.161, loss=264.821, backward_time=0.307, grad_norm=146.759, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.066, optim0_lr0=1.742e-05, train_time=1.740
+[gpub024:0/32] 2024-01-21 02:33:45,295 (trainer:737) INFO: 1epoch:train:10501-10600batch: iter_time=8.545e-05, forward_time=0.399, loss_ctc=282.027, loss_att=256.172, acc=0.168, loss=263.928, backward_time=0.249, grad_norm=166.966, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.077, optim0_lr0=1.759e-05, train_time=1.117
+[gpub024:0/32] 2024-01-21 02:37:17,762 (trainer:737) INFO: 1epoch:train:10601-10700batch: iter_time=8.880e-05, forward_time=0.161, loss_ctc=272.999, loss_att=256.802, acc=0.167, loss=261.661, backward_time=0.223, grad_norm=179.897, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.775e-05, train_time=2.122
+[gpub024:0/32] 2024-01-21 02:39:39,221 (trainer:737) INFO: 1epoch:train:10701-10800batch: iter_time=8.818e-05, forward_time=0.160, loss_ctc=239.982, loss_att=224.637, acc=0.178, loss=229.240, backward_time=0.225, grad_norm=126.363, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.792e-05, train_time=1.417
+[gpub024:0/32] 2024-01-21 02:41:42,389 (trainer:737) INFO: 1epoch:train:10801-10900batch: iter_time=9.326e-05, forward_time=0.161, loss_ctc=285.429, loss_att=261.823, acc=0.167, loss=268.905, backward_time=0.225, grad_norm=142.748, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.809e-05, train_time=1.231
+[gpub024:0/32] 2024-01-21 02:43:50,304 (trainer:737) INFO: 1epoch:train:10901-11000batch: iter_time=9.409e-05, forward_time=0.161, loss_ctc=260.227, loss_att=239.086, acc=0.174, loss=245.428, backward_time=0.225, grad_norm=115.044, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.825e-05, train_time=1.279
+[gpub024:0/32] 2024-01-21 02:45:31,853 (trainer:737) INFO: 1epoch:train:11001-11100batch: iter_time=8.988e-05, forward_time=0.161, loss_ctc=266.320, loss_att=247.667, acc=0.170, loss=253.263, backward_time=0.225, grad_norm=145.713, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.842e-05, train_time=1.015
+[gpub024:0/32] 2024-01-21 02:48:42,670 (trainer:737) INFO: 1epoch:train:11101-11200batch: iter_time=8.137e-05, forward_time=0.306, loss_ctc=285.860, loss_att=252.738, acc=0.176, loss=262.674, backward_time=0.288, grad_norm=167.609, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.073, optim0_lr0=1.859e-05, train_time=1.908
+[gpub024:0/32] 2024-01-21 02:49:52,391 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub024:0/32] 2024-01-21 02:50:11,504 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 02:50:15,128 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f587e1bfe50>)
+[gpub024:0/32] 2024-01-21 02:50:15,128 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub024:0/32] 2024-01-21 02:50:15,227 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 03:02:27,821 (trainer:737) INFO: 1epoch:train:11201-11300batch: iter_time=6.832, forward_time=0.363, loss_ctc=277.302, loss_att=241.976, acc=0.181, loss=252.574, backward_time=0.254, grad_norm=154.550, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.060, optim0_lr0=1.875e-05, train_time=8.251
+[gpub024:0/32] 2024-01-21 03:03:56,469 (trainer:737) INFO: 1epoch:train:11301-11400batch: iter_time=7.928e-05, forward_time=0.162, loss_ctc=285.277, loss_att=260.747, acc=0.165, loss=268.106, backward_time=0.227, grad_norm=133.876, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.892e-05, train_time=0.887
+[gpub024:0/32] 2024-01-21 03:06:59,988 (trainer:737) INFO: 1epoch:train:11401-11500batch: iter_time=8.308e-05, forward_time=0.162, loss_ctc=277.671, loss_att=262.178, acc=0.172, loss=266.826, backward_time=0.227, grad_norm=132.603, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.909e-05, train_time=1.835
+[gpub024:0/32] 2024-01-21 03:09:52,471 (trainer:737) INFO: 1epoch:train:11501-11600batch: iter_time=8.317e-05, forward_time=0.160, loss_ctc=250.428, loss_att=227.706, acc=0.182, loss=234.522, backward_time=0.223, grad_norm=105.524, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.050, optim0_lr0=1.925e-05, train_time=1.725
+[gpub024:0/32] 2024-01-21 03:13:05,110 (trainer:737) INFO: 1epoch:train:11601-11700batch: iter_time=9.427e-05, forward_time=0.161, loss_ctc=281.736, loss_att=252.370, acc=0.175, loss=261.179, backward_time=0.224, grad_norm=174.028, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.051, optim0_lr0=1.942e-05, train_time=1.926
+[gpub024:0/32] 2024-01-21 03:14:48,702 (trainer:737) INFO: 1epoch:train:11701-11800batch: iter_time=8.426e-05, forward_time=0.285, loss_ctc=252.192, loss_att=224.693, acc=0.176, loss=232.943, backward_time=0.298, grad_norm=137.661, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.059, optim0_lr0=1.959e-05, train_time=1.034
+[gpub024:0/32] 2024-01-21 03:18:18,883 (trainer:737) INFO: 1epoch:train:11801-11900batch: iter_time=8.260e-05, forward_time=0.213, loss_ctc=272.494, loss_att=247.291, acc=0.189, loss=254.852, backward_time=0.249, grad_norm=184.464, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.054, optim0_lr0=1.975e-05, train_time=2.103
+[gpub024:0/32] 2024-01-21 03:20:45,968 (trainer:737) INFO: 1epoch:train:11901-12000batch: iter_time=8.428e-05, forward_time=0.328, loss_ctc=266.780, loss_att=244.474, acc=0.174, loss=251.166, backward_time=0.235, grad_norm=170.144, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.064, optim0_lr0=1.992e-05, train_time=1.470
+[gpub024:0/32] 2024-01-21 03:24:31,243 (trainer:737) INFO: 1epoch:train:12001-12100batch: iter_time=9.120e-05, forward_time=0.163, loss_ctc=263.322, loss_att=238.195, acc=0.180, loss=245.733, backward_time=0.224, grad_norm=136.046, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.009e-05, train_time=2.253
+[gpub024:0/32] 2024-01-21 03:26:03,517 (trainer:737) INFO: 1epoch:train:12101-12200batch: iter_time=7.956e-05, forward_time=0.160, loss_ctc=268.671, loss_att=240.877, acc=0.184, loss=249.215, backward_time=0.225, grad_norm=132.638, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.025e-05, train_time=0.923
+[gpub024:0/32] 2024-01-21 03:27:58,533 (trainer:737) INFO: 1epoch:train:12201-12300batch: iter_time=7.989e-05, forward_time=0.160, loss_ctc=253.662, loss_att=234.867, acc=0.178, loss=240.505, backward_time=0.224, grad_norm=119.769, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.042e-05, train_time=1.150
+[gpub024:0/32] 2024-01-21 03:29:50,067 (trainer:737) INFO: 1epoch:train:12301-12400batch: iter_time=7.787e-05, forward_time=0.162, loss_ctc=286.632, loss_att=249.840, acc=0.176, loss=260.878, backward_time=0.228, grad_norm=127.209, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.059e-05, train_time=1.115
+[gpub024:0/32] 2024-01-21 03:31:24,231 (trainer:737) INFO: 1epoch:train:12401-12500batch: iter_time=8.204e-05, forward_time=0.216, loss_ctc=280.054, loss_att=244.005, acc=0.181, loss=254.820, backward_time=0.263, grad_norm=143.667, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.059, optim0_lr0=2.075e-05, train_time=0.941
+[gpub024:0/32] 2024-01-21 03:31:44,260 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub024:0/32] 2024-01-21 03:32:02,425 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 03:32:05,952 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f58a8fe7f70>)
+[gpub024:0/32] 2024-01-21 03:32:05,952 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub024:0/32] 2024-01-21 03:32:05,955 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 03:38:57,326 (trainer:737) INFO: 1epoch:train:12501-12600batch: iter_time=3.558, forward_time=0.209, loss_ctc=268.223, loss_att=251.450, acc=0.185, loss=256.482, backward_time=0.230, grad_norm=142.825, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.052, optim0_lr0=2.092e-05, train_time=4.529
+[gpub024:0/32] 2024-01-21 03:41:54,500 (trainer:737) INFO: 1epoch:train:12601-12700batch: iter_time=8.325e-05, forward_time=0.160, loss_ctc=265.980, loss_att=251.895, acc=0.169, loss=256.120, backward_time=0.224, grad_norm=125.970, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.109e-05, train_time=1.773
+[gpub024:0/32] 2024-01-21 03:44:31,962 (trainer:737) INFO: 1epoch:train:12701-12800batch: iter_time=7.995e-05, forward_time=0.162, loss_ctc=279.990, loss_att=261.107, acc=0.179, loss=266.772, backward_time=0.228, grad_norm=115.312, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.125e-05, train_time=1.574
+[gpub024:0/32] 2024-01-21 03:48:04,859 (trainer:737) INFO: 1epoch:train:12801-12900batch: iter_time=8.525e-05, forward_time=0.161, loss_ctc=249.224, loss_att=234.362, acc=0.188, loss=238.821, backward_time=0.224, grad_norm=144.959, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.142e-05, train_time=2.129
+[gpub024:0/32] 2024-01-21 03:49:49,050 (trainer:737) INFO: 1epoch:train:12901-13000batch: iter_time=8.209e-05, forward_time=0.161, loss_ctc=274.109, loss_att=254.440, acc=0.174, loss=260.341, backward_time=0.227, grad_norm=160.819, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.159e-05, train_time=1.042
+[gpub024:0/32] 2024-01-21 03:54:23,570 (trainer:737) INFO: 1epoch:train:13001-13100batch: iter_time=8.262e-05, forward_time=0.296, loss_ctc=276.021, loss_att=255.708, acc=0.185, loss=261.802, backward_time=0.292, grad_norm=139.119, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.073, optim0_lr0=2.175e-05, train_time=2.745
+[gpub024:0/32] 2024-01-21 03:57:42,111 (trainer:737) INFO: 1epoch:train:13101-13200batch: iter_time=8.509e-05, forward_time=0.321, loss_ctc=267.229, loss_att=248.614, acc=0.185, loss=254.198, backward_time=0.259, grad_norm=174.767, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.059, optim0_lr0=2.192e-05, train_time=1.984
+[gpub024:0/32] 2024-01-21 04:00:38,500 (trainer:737) INFO: 1epoch:train:13201-13300batch: iter_time=8.217e-05, forward_time=0.202, loss_ctc=235.487, loss_att=221.467, acc=0.189, loss=225.673, backward_time=0.226, grad_norm=145.404, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.209e-05, train_time=1.764
+[gpub024:0/32] 2024-01-21 04:02:00,119 (trainer:737) INFO: 1epoch:train:13301-13400batch: iter_time=9.035e-05, forward_time=0.161, loss_ctc=281.851, loss_att=254.881, acc=0.179, loss=262.972, backward_time=0.228, grad_norm=139.862, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.225e-05, train_time=0.817
+[gpub024:0/32] 2024-01-21 04:04:33,597 (trainer:737) INFO: 1epoch:train:13401-13500batch: iter_time=8.976e-05, forward_time=0.161, loss_ctc=255.035, loss_att=235.240, acc=0.187, loss=241.178, backward_time=0.226, grad_norm=103.883, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.242e-05, train_time=1.535
+[gpub024:0/32] 2024-01-21 04:08:30,555 (trainer:737) INFO: 1epoch:train:13501-13600batch: iter_time=8.333e-05, forward_time=0.172, loss_ctc=263.305, loss_att=247.813, acc=0.179, loss=252.460, backward_time=0.269, grad_norm=125.528, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.058, optim0_lr0=2.259e-05, train_time=2.369
+[gpub024:0/32] 2024-01-21 04:11:22,592 (trainer:737) INFO: 1epoch:train:13601-13700batch: iter_time=2.879e-04, forward_time=0.307, loss_ctc=280.670, loss_att=252.997, acc=0.185, loss=261.299, backward_time=0.355, grad_norm=149.947, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.068, optim0_lr0=2.275e-05, train_time=1.720
+[gpub024:0/32] 2024-01-21 04:12:51,360 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub024:0/32] 2024-01-21 04:13:09,996 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 04:13:13,618 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5837af9330>)
+[gpub024:0/32] 2024-01-21 04:13:13,618 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub024:0/32] 2024-01-21 04:13:13,675 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 04:24:34,471 (trainer:737) INFO: 1epoch:train:13701-13800batch: iter_time=6.534, forward_time=0.272, loss_ctc=272.963, loss_att=239.046, acc=0.189, loss=249.221, backward_time=0.237, grad_norm=138.956, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.053, optim0_lr0=2.292e-05, train_time=7.917
+[gpub024:0/32] 2024-01-21 04:25:56,710 (trainer:737) INFO: 1epoch:train:13801-13900batch: iter_time=7.908e-05, forward_time=0.162, loss_ctc=282.201, loss_att=263.690, acc=0.172, loss=269.243, backward_time=0.228, grad_norm=115.834, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.309e-05, train_time=0.824
+[gpub024:0/32] 2024-01-21 04:28:35,165 (trainer:737) INFO: 1epoch:train:13901-14000batch: iter_time=8.137e-05, forward_time=0.161, loss_ctc=271.403, loss_att=253.846, acc=0.180, loss=259.113, backward_time=0.226, grad_norm=119.145, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.050, optim0_lr0=2.325e-05, train_time=1.584
+[gpub024:0/32] 2024-01-21 04:31:01,455 (trainer:737) INFO: 1epoch:train:14001-14100batch: iter_time=8.272e-05, forward_time=0.161, loss_ctc=246.686, loss_att=228.488, acc=0.192, loss=233.947, backward_time=0.224, grad_norm=93.795, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.050, optim0_lr0=2.342e-05, train_time=1.463
+[gpub024:0/32] 2024-01-21 04:34:16,980 (trainer:737) INFO: 1epoch:train:14101-14200batch: iter_time=8.490e-05, forward_time=0.220, loss_ctc=277.616, loss_att=249.427, acc=0.186, loss=257.884, backward_time=0.237, grad_norm=149.300, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.058, optim0_lr0=2.359e-05, train_time=1.955
+[gpub024:0/32] 2024-01-21 04:37:14,736 (trainer:737) INFO: 1epoch:train:14201-14300batch: iter_time=0.007, forward_time=0.414, loss_ctc=246.626, loss_att=225.503, acc=0.184, loss=231.840, backward_time=0.315, grad_norm=136.346, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.074, optim0_lr0=2.375e-05, train_time=1.777
+[gpub024:0/32] 2024-01-21 04:39:59,694 (trainer:737) INFO: 1epoch:train:14301-14400batch: iter_time=8.136e-05, forward_time=0.164, loss_ctc=264.381, loss_att=244.665, acc=0.200, loss=250.579, backward_time=0.229, grad_norm=128.314, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.050, optim0_lr0=2.392e-05, train_time=1.649
+[gpub024:0/32] 2024-01-21 04:41:56,593 (trainer:737) INFO: 1epoch:train:14401-14500batch: iter_time=8.377e-05, forward_time=0.162, loss_ctc=262.655, loss_att=243.951, acc=0.182, loss=249.562, backward_time=0.227, grad_norm=150.819, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.050, optim0_lr0=2.409e-05, train_time=1.169
+[gpub024:0/32] 2024-01-21 04:44:41,922 (trainer:737) INFO: 1epoch:train:14501-14600batch: iter_time=8.189e-05, forward_time=0.161, loss_ctc=257.958, loss_att=235.954, acc=0.188, loss=242.555, backward_time=0.223, grad_norm=114.181, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.050, optim0_lr0=2.425e-05, train_time=1.654
+[gpub024:0/32] 2024-01-21 04:48:02,004 (trainer:737) INFO: 1epoch:train:14601-14700batch: iter_time=8.097e-05, forward_time=0.161, loss_ctc=265.878, loss_att=234.991, acc=0.192, loss=244.257, backward_time=0.224, grad_norm=125.507, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.051, optim0_lr0=2.442e-05, train_time=2.001
+[gpub024:0/32] 2024-01-21 04:49:32,776 (trainer:737) INFO: 1epoch:train:14701-14800batch: iter_time=8.562e-05, forward_time=0.161, loss_ctc=250.036, loss_att=232.270, acc=0.186, loss=237.600, backward_time=0.229, grad_norm=130.132, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.051, optim0_lr0=2.459e-05, train_time=0.907
+[gpub024:0/32] 2024-01-21 04:51:56,200 (trainer:737) INFO: 1epoch:train:14801-14900batch: iter_time=7.885e-05, forward_time=0.306, loss_ctc=282.079, loss_att=255.372, acc=0.180, loss=263.384, backward_time=0.303, grad_norm=121.399, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.066, optim0_lr0=2.475e-05, train_time=1.434
+[gpub024:0/32] 2024-01-21 04:55:21,942 (trainer:737) INFO: 1epoch:train:14901-15000batch: iter_time=8.552e-05, forward_time=0.308, loss_ctc=273.545, loss_att=244.581, acc=0.188, loss=253.270, backward_time=0.251, grad_norm=110.045, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.067, optim0_lr0=2.492e-05, train_time=2.057
+[gpub024:0/32] 2024-01-21 05:31:23,617 (trainer:343) INFO: 1epoch results: [train] iter_time=0.524, forward_time=0.359, loss_ctc=356.728, loss_att=287.104, acc=0.134, loss=307.991, backward_time=0.302, grad_norm=358.379, clip=100.000, loss_scale=1.669e+06, optim_step_time=0.068, optim0_lr0=1.250e-05, train_time=2.468, time=10 hours, 17 minutes and 19.89 seconds, total_count=15000, gpu_max_cached_mem_GB=24.412, [valid] loss_ctc=192.611, cer_ctc=0.981, loss_att=170.394, acc=0.159, cer=0.753, wer=1.000, loss=177.059, time=35 minutes and 37.51 seconds, total_count=4671, gpu_max_cached_mem_GB=24.412
+[gpub024:0/32] 2024-01-21 05:31:51,835 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub024:0/32] 2024-01-21 05:31:51,970 (trainer:272) INFO: 2/45epoch started. Estimated time to finish: 2 weeks, 5 days and 23 hours
+[gpub024:0/32] 2024-01-21 05:31:52,177 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub024:0/32] 2024-01-21 05:32:10,284 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 05:32:13,611 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f589aa0d210>)
+[gpub024:0/32] 2024-01-21 05:32:13,611 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub024:0/32] 2024-01-21 05:32:13,615 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 05:44:05,340 (trainer:737) INFO: 2epoch:train:1-100batch: iter_time=6.321, forward_time=0.277, loss_ctc=278.550, loss_att=243.434, acc=0.187, loss=253.969, backward_time=0.260, grad_norm=117.097, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.055, optim0_lr0=2.509e-05, train_time=7.332
+[gpub024:0/32] 2024-01-21 05:45:59,240 (trainer:737) INFO: 2epoch:train:101-200batch: iter_time=9.186e-05, forward_time=0.161, loss_ctc=255.487, loss_att=221.012, acc=0.197, loss=231.355, backward_time=0.229, grad_norm=116.468, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.051, optim0_lr0=2.525e-05, train_time=1.139
+[gpub024:0/32] 2024-01-21 05:48:09,893 (trainer:737) INFO: 2epoch:train:201-300batch: iter_time=9.197e-05, forward_time=0.216, loss_ctc=245.728, loss_att=216.154, acc=0.202, loss=225.026, backward_time=0.242, grad_norm=110.269, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.062, optim0_lr0=2.542e-05, train_time=1.306
+[gpub024:0/32] 2024-01-21 05:50:01,902 (trainer:737) INFO: 2epoch:train:301-400batch: iter_time=9.275e-05, forward_time=0.199, loss_ctc=260.214, loss_att=234.469, acc=0.185, loss=242.192, backward_time=0.261, grad_norm=94.800, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.061, optim0_lr0=2.559e-05, train_time=1.120
+[gpub024:0/32] 2024-01-21 05:52:40,201 (trainer:737) INFO: 2epoch:train:401-500batch: iter_time=8.907e-05, forward_time=0.224, loss_ctc=280.631, loss_att=236.690, acc=0.179, loss=249.872, backward_time=0.269, grad_norm=140.323, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.059, optim0_lr0=2.575e-05, train_time=1.583
+[gpub024:0/32] 2024-01-21 05:54:25,128 (trainer:737) INFO: 2epoch:train:501-600batch: iter_time=8.419e-05, forward_time=0.161, loss_ctc=244.359, loss_att=214.584, acc=0.194, loss=223.517, backward_time=0.224, grad_norm=113.378, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.051, optim0_lr0=2.592e-05, train_time=1.049
+[gpub024:0/32] 2024-01-21 05:57:29,351 (trainer:737) INFO: 2epoch:train:601-700batch: iter_time=9.334e-05, forward_time=0.196, loss_ctc=238.135, loss_att=206.846, acc=0.196, loss=216.233, backward_time=0.224, grad_norm=89.904, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.053, optim0_lr0=2.609e-05, train_time=1.842
+[gpub024:0/32] 2024-01-21 06:00:02,999 (trainer:737) INFO: 2epoch:train:701-800batch: iter_time=8.976e-05, forward_time=0.163, loss_ctc=229.877, loss_att=211.912, acc=0.199, loss=217.301, backward_time=0.230, grad_norm=96.500, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.051, optim0_lr0=2.625e-05, train_time=1.536
+[gpub024:0/32] 2024-01-21 06:01:57,144 (trainer:737) INFO: 2epoch:train:801-900batch: iter_time=9.453e-05, forward_time=0.226, loss_ctc=266.449, loss_att=226.756, acc=0.199, loss=238.664, backward_time=0.231, grad_norm=100.461, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.056, optim0_lr0=2.642e-05, train_time=1.141
+[gpub024:0/32] 2024-01-21 06:04:12,356 (trainer:737) INFO: 2epoch:train:901-1000batch: iter_time=9.299e-05, forward_time=0.305, loss_ctc=224.168, loss_att=200.005, acc=0.202, loss=207.254, backward_time=0.230, grad_norm=93.028, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.063, optim0_lr0=2.659e-05, train_time=1.352
+[gpub024:0/32] 2024-01-21 06:06:57,031 (trainer:737) INFO: 2epoch:train:1001-1100batch: iter_time=8.859e-05, forward_time=0.234, loss_ctc=258.210, loss_att=231.613, acc=0.185, loss=239.592, backward_time=0.302, grad_norm=111.040, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.061, optim0_lr0=2.675e-05, train_time=1.646
+[gpub024:0/32] 2024-01-21 06:08:46,365 (trainer:737) INFO: 2epoch:train:1101-1200batch: iter_time=9.132e-05, forward_time=0.170, loss_ctc=203.086, loss_att=180.971, acc=0.207, loss=187.605, backward_time=0.223, grad_norm=75.434, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.692e-05, train_time=1.093
+[gpub024:0/32] 2024-01-21 06:10:08,731 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub024:0/32] 2024-01-21 06:10:27,745 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 06:10:31,286 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f589aa9df30>)
+[gpub024:0/32] 2024-01-21 06:10:31,286 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub024:0/32] 2024-01-21 06:10:31,310 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 06:21:41,365 (trainer:737) INFO: 2epoch:train:1201-1300batch: iter_time=6.441, forward_time=0.202, loss_ctc=247.800, loss_att=211.359, acc=0.209, loss=222.291, backward_time=0.229, grad_norm=103.662, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.052, optim0_lr0=2.709e-05, train_time=7.750
+[gpub024:0/32] 2024-01-21 06:23:36,667 (trainer:737) INFO: 2epoch:train:1301-1400batch: iter_time=8.103e-05, forward_time=0.194, loss_ctc=294.381, loss_att=252.861, acc=0.184, loss=265.317, backward_time=0.226, grad_norm=115.322, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.063, optim0_lr0=2.725e-05, train_time=1.153
+[gpub024:0/32] 2024-01-21 06:25:32,387 (trainer:737) INFO: 2epoch:train:1401-1500batch: iter_time=8.503e-05, forward_time=0.213, loss_ctc=238.791, loss_att=207.607, acc=0.207, loss=216.962, backward_time=0.261, grad_norm=118.804, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.058, optim0_lr0=2.742e-05, train_time=1.157
+[gpub024:0/32] 2024-01-21 06:27:05,366 (trainer:737) INFO: 2epoch:train:1501-1600batch: iter_time=8.172e-05, forward_time=0.163, loss_ctc=231.221, loss_att=206.621, acc=0.203, loss=214.001, backward_time=0.227, grad_norm=85.674, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.759e-05, train_time=0.930
+[gpub024:0/32] 2024-01-21 06:29:10,253 (trainer:737) INFO: 2epoch:train:1601-1700batch: iter_time=9.047e-05, forward_time=0.181, loss_ctc=287.987, loss_att=246.891, acc=0.176, loss=259.220, backward_time=0.229, grad_norm=145.131, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.056, optim0_lr0=2.775e-05, train_time=1.249
+[gpub024:0/32] 2024-01-21 06:31:21,717 (trainer:737) INFO: 2epoch:train:1701-1800batch: iter_time=8.620e-05, forward_time=0.166, loss_ctc=270.589, loss_att=231.940, acc=0.186, loss=243.534, backward_time=0.226, grad_norm=120.163, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.792e-05, train_time=1.314
+[gpub024:0/32] 2024-01-21 06:33:51,629 (trainer:737) INFO: 2epoch:train:1801-1900batch: iter_time=9.134e-05, forward_time=0.189, loss_ctc=200.891, loss_att=178.210, acc=0.213, loss=185.015, backward_time=0.275, grad_norm=86.129, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.057, optim0_lr0=2.809e-05, train_time=1.499
+[gpub024:0/32] 2024-01-21 06:36:06,096 (trainer:737) INFO: 2epoch:train:1901-2000batch: iter_time=8.785e-05, forward_time=0.173, loss_ctc=261.200, loss_att=230.601, acc=0.193, loss=239.781, backward_time=0.232, grad_norm=98.514, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.072, optim0_lr0=2.825e-05, train_time=1.345
+[gpub024:0/32] 2024-01-21 06:37:56,697 (trainer:737) INFO: 2epoch:train:2001-2100batch: iter_time=9.503e-05, forward_time=0.338, loss_ctc=241.390, loss_att=213.947, acc=0.202, loss=222.180, backward_time=0.274, grad_norm=80.923, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.061, optim0_lr0=2.842e-05, train_time=1.106
+[gpub024:0/32] 2024-01-21 06:39:35,880 (trainer:737) INFO: 2epoch:train:2101-2200batch: iter_time=9.507e-05, forward_time=0.162, loss_ctc=245.696, loss_att=213.313, acc=0.204, loss=223.028, backward_time=0.226, grad_norm=104.165, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.859e-05, train_time=0.992
+[gpub024:0/32] 2024-01-21 06:41:22,762 (trainer:737) INFO: 2epoch:train:2201-2300batch: iter_time=1.042e-04, forward_time=0.161, loss_ctc=223.664, loss_att=195.760, acc=0.201, loss=204.131, backward_time=0.224, grad_norm=87.140, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.875e-05, train_time=1.069
+[gpub024:0/32] 2024-01-21 06:43:49,364 (trainer:737) INFO: 2epoch:train:2301-2400batch: iter_time=9.250e-05, forward_time=0.162, loss_ctc=234.589, loss_att=209.921, acc=0.201, loss=217.321, backward_time=0.236, grad_norm=96.139, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.892e-05, train_time=1.466
+[gpub024:0/32] 2024-01-21 06:45:47,764 (trainer:737) INFO: 2epoch:train:2401-2500batch: iter_time=8.881e-05, forward_time=0.161, loss_ctc=225.439, loss_att=196.216, acc=0.208, loss=204.983, backward_time=0.224, grad_norm=88.570, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.909e-05, train_time=1.184
+[gpub024:0/32] 2024-01-21 06:46:07,828 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub024:0/32] 2024-01-21 06:46:26,384 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 06:46:30,050 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f580c5aefb0>)
+[gpub024:0/32] 2024-01-21 06:46:30,050 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub024:0/32] 2024-01-21 06:46:30,053 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 06:56:40,674 (trainer:737) INFO: 2epoch:train:2501-2600batch: iter_time=5.503, forward_time=0.301, loss_ctc=268.540, loss_att=243.815, acc=0.194, loss=251.232, backward_time=0.261, grad_norm=103.385, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.055, optim0_lr0=2.925e-05, train_time=6.529
+[gpub024:0/32] 2024-01-21 06:58:44,102 (trainer:737) INFO: 2epoch:train:2601-2700batch: iter_time=8.341e-05, forward_time=0.161, loss_ctc=246.805, loss_att=217.914, acc=0.203, loss=226.582, backward_time=0.225, grad_norm=100.753, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.942e-05, train_time=1.234
+[gpub024:0/32] 2024-01-21 07:01:15,350 (trainer:737) INFO: 2epoch:train:2701-2800batch: iter_time=8.570e-05, forward_time=0.172, loss_ctc=240.234, loss_att=210.714, acc=0.209, loss=219.570, backward_time=0.238, grad_norm=101.011, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.052, optim0_lr0=2.959e-05, train_time=1.512
+[gpub024:0/32] 2024-01-21 07:04:01,118 (trainer:737) INFO: 2epoch:train:2801-2900batch: iter_time=9.133e-05, forward_time=0.178, loss_ctc=251.982, loss_att=234.885, acc=0.191, loss=240.015, backward_time=0.252, grad_norm=88.250, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.053, optim0_lr0=2.975e-05, train_time=1.657
+[gpub024:0/32] 2024-01-21 07:05:50,920 (trainer:737) INFO: 2epoch:train:2901-3000batch: iter_time=9.407e-05, forward_time=0.163, loss_ctc=275.032, loss_att=237.947, acc=0.184, loss=249.072, backward_time=0.226, grad_norm=147.593, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.051, optim0_lr0=2.992e-05, train_time=1.098
+[gpub024:0/32] 2024-01-21 07:07:29,752 (trainer:737) INFO: 2epoch:train:3001-3100batch: iter_time=8.584e-05, forward_time=0.248, loss_ctc=238.163, loss_att=212.849, acc=0.198, loss=220.443, backward_time=0.241, grad_norm=117.457, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.068, optim0_lr0=3.009e-05, train_time=0.988
+[gpub024:0/32] 2024-01-21 07:10:31,389 (trainer:737) INFO: 2epoch:train:3101-3200batch: iter_time=8.934e-05, forward_time=0.185, loss_ctc=234.672, loss_att=204.356, acc=0.202, loss=213.451, backward_time=0.255, grad_norm=92.657, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.053, optim0_lr0=3.025e-05, train_time=1.816
+[gpub024:0/32] 2024-01-21 07:13:15,922 (trainer:737) INFO: 2epoch:train:3201-3300batch: iter_time=9.161e-05, forward_time=0.232, loss_ctc=226.264, loss_att=210.248, acc=0.204, loss=215.053, backward_time=0.283, grad_norm=86.351, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.057, optim0_lr0=3.042e-05, train_time=1.645
+[gpub024:0/32] 2024-01-21 07:15:41,035 (trainer:737) INFO: 2epoch:train:3301-3400batch: iter_time=8.052e-05, forward_time=0.237, loss_ctc=259.525, loss_att=224.990, acc=0.204, loss=235.351, backward_time=0.226, grad_norm=82.395, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.056, optim0_lr0=3.059e-05, train_time=1.451
+[gpub024:0/32] 2024-01-21 07:18:07,983 (trainer:737) INFO: 2epoch:train:3401-3500batch: iter_time=8.787e-05, forward_time=0.163, loss_ctc=218.331, loss_att=197.243, acc=0.207, loss=203.569, backward_time=0.224, grad_norm=84.428, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.075e-05, train_time=1.469
+[gpub024:0/32] 2024-01-21 07:20:04,372 (trainer:737) INFO: 2epoch:train:3501-3600batch: iter_time=9.025e-05, forward_time=0.172, loss_ctc=252.357, loss_att=233.969, acc=0.190, loss=239.485, backward_time=0.249, grad_norm=95.587, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.092e-05, train_time=1.164
+[gpub024:0/32] 2024-01-21 07:21:44,682 (trainer:737) INFO: 2epoch:train:3601-3700batch: iter_time=9.147e-05, forward_time=0.168, loss_ctc=197.695, loss_att=176.588, acc=0.215, loss=182.920, backward_time=0.227, grad_norm=68.013, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.053, optim0_lr0=3.109e-05, train_time=1.003
+[gpub024:0/32] 2024-01-21 07:22:52,553 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub024:0/32] 2024-01-21 07:23:11,899 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 07:23:15,679 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f57ba132020>)
+[gpub024:0/32] 2024-01-21 07:23:15,679 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub024:0/32] 2024-01-21 07:23:15,685 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 07:34:19,606 (trainer:737) INFO: 2epoch:train:3701-3800batch: iter_time=6.472, forward_time=0.310, loss_ctc=241.358, loss_att=207.379, acc=0.215, loss=217.573, backward_time=0.251, grad_norm=91.132, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.060, optim0_lr0=3.125e-05, train_time=7.549
+[gpub024:0/32] 2024-01-21 07:35:43,509 (trainer:737) INFO: 2epoch:train:3801-3900batch: iter_time=8.507e-05, forward_time=0.163, loss_ctc=288.381, loss_att=249.592, acc=0.190, loss=261.229, backward_time=0.228, grad_norm=112.420, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.142e-05, train_time=0.839
+[gpub024:0/32] 2024-01-21 07:37:09,667 (trainer:737) INFO: 2epoch:train:3901-4000batch: iter_time=8.449e-05, forward_time=0.161, loss_ctc=230.682, loss_att=200.532, acc=0.214, loss=209.577, backward_time=0.227, grad_norm=87.220, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.159e-05, train_time=0.861
+[gpub024:0/32] 2024-01-21 07:39:41,474 (trainer:737) INFO: 2epoch:train:4001-4100batch: iter_time=8.844e-05, forward_time=0.215, loss_ctc=223.215, loss_att=199.348, acc=0.211, loss=206.508, backward_time=0.296, grad_norm=62.646, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.059, optim0_lr0=3.175e-05, train_time=1.518
+[gpub024:0/32] 2024-01-21 07:42:08,432 (trainer:737) INFO: 2epoch:train:4101-4200batch: iter_time=8.122e-05, forward_time=0.168, loss_ctc=277.420, loss_att=240.075, acc=0.184, loss=251.278, backward_time=0.226, grad_norm=116.529, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.192e-05, train_time=1.469
+[gpub024:0/32] 2024-01-21 07:44:11,007 (trainer:737) INFO: 2epoch:train:4201-4300batch: iter_time=8.398e-05, forward_time=0.199, loss_ctc=263.705, loss_att=225.890, acc=0.194, loss=237.235, backward_time=0.303, grad_norm=121.672, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.061, optim0_lr0=3.209e-05, train_time=1.225
+[gpub024:0/32] 2024-01-21 07:46:15,888 (trainer:737) INFO: 2epoch:train:4301-4400batch: iter_time=8.033e-05, forward_time=0.160, loss_ctc=196.307, loss_att=174.026, acc=0.218, loss=180.710, backward_time=0.223, grad_norm=82.089, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.225e-05, train_time=1.249
+[gpub024:0/32] 2024-01-21 07:48:21,008 (trainer:737) INFO: 2epoch:train:4401-4500batch: iter_time=8.508e-05, forward_time=0.161, loss_ctc=255.716, loss_att=224.966, acc=0.197, loss=234.191, backward_time=0.226, grad_norm=105.065, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.242e-05, train_time=1.251
+[gpub024:0/32] 2024-01-21 07:50:51,488 (trainer:737) INFO: 2epoch:train:4501-4600batch: iter_time=8.319e-05, forward_time=0.196, loss_ctc=235.227, loss_att=208.637, acc=0.207, loss=216.614, backward_time=0.224, grad_norm=64.723, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.051, optim0_lr0=3.259e-05, train_time=1.505
+[gpub024:0/32] 2024-01-21 07:52:27,390 (trainer:737) INFO: 2epoch:train:4601-4700batch: iter_time=8.582e-05, forward_time=0.220, loss_ctc=239.126, loss_att=207.929, acc=0.210, loss=217.288, backward_time=0.236, grad_norm=84.245, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.055, optim0_lr0=3.275e-05, train_time=0.959
+[gpub024:0/32] 2024-01-21 07:54:24,285 (trainer:737) INFO: 2epoch:train:4701-4800batch: iter_time=8.323e-05, forward_time=0.248, loss_ctc=216.670, loss_att=190.129, acc=0.208, loss=198.091, backward_time=0.236, grad_norm=71.871, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.055, optim0_lr0=3.292e-05, train_time=1.169
+[gpub024:0/32] 2024-01-21 07:57:10,992 (trainer:737) INFO: 2epoch:train:4801-4900batch: iter_time=8.758e-05, forward_time=0.259, loss_ctc=229.003, loss_att=204.945, acc=0.205, loss=212.163, backward_time=0.243, grad_norm=79.685, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.069, optim0_lr0=3.309e-05, train_time=1.667
+[gpub024:0/32] 2024-01-21 07:59:45,528 (trainer:737) INFO: 2epoch:train:4901-5000batch: iter_time=8.339e-05, forward_time=0.185, loss_ctc=219.063, loss_att=190.733, acc=0.215, loss=199.232, backward_time=0.263, grad_norm=84.716, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.052, optim0_lr0=3.325e-05, train_time=1.545
+[gpub024:0/32] 2024-01-21 08:00:05,807 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub024:0/32] 2024-01-21 08:00:24,647 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 08:00:28,295 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f580c569cf0>)
+[gpub024:0/32] 2024-01-21 08:00:28,295 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub024:0/32] 2024-01-21 08:00:28,451 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 08:12:21,330 (trainer:737) INFO: 2epoch:train:5001-5100batch: iter_time=6.579, forward_time=0.199, loss_ctc=262.118, loss_att=238.490, acc=0.199, loss=245.578, backward_time=0.231, grad_norm=91.844, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.052, optim0_lr0=3.342e-05, train_time=7.558
+[gpub024:0/32] 2024-01-21 08:13:41,639 (trainer:737) INFO: 2epoch:train:5101-5200batch: iter_time=8.684e-05, forward_time=0.161, loss_ctc=237.366, loss_att=211.685, acc=0.210, loss=219.389, backward_time=0.228, grad_norm=79.965, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.359e-05, train_time=0.803
+[gpub024:0/32] 2024-01-21 08:15:02,405 (trainer:737) INFO: 2epoch:train:5201-5300batch: iter_time=8.781e-05, forward_time=0.162, loss_ctc=232.605, loss_att=204.885, acc=0.216, loss=213.201, backward_time=0.227, grad_norm=77.109, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.375e-05, train_time=0.807
+[gpub024:0/32] 2024-01-21 08:17:33,556 (trainer:737) INFO: 2epoch:train:5301-5400batch: iter_time=9.406e-05, forward_time=0.261, loss_ctc=244.863, loss_att=230.471, acc=0.196, loss=234.789, backward_time=0.245, grad_norm=69.615, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.060, optim0_lr0=3.392e-05, train_time=1.511
+[gpub024:0/32] 2024-01-21 08:20:05,462 (trainer:737) INFO: 2epoch:train:5401-5500batch: iter_time=8.910e-05, forward_time=0.162, loss_ctc=265.929, loss_att=232.565, acc=0.192, loss=242.574, backward_time=0.229, grad_norm=109.310, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.409e-05, train_time=1.519
+[gpub024:0/32] 2024-01-21 08:23:04,770 (trainer:737) INFO: 2epoch:train:5501-5600batch: iter_time=8.741e-05, forward_time=0.375, loss_ctc=228.963, loss_att=206.505, acc=0.206, loss=213.242, backward_time=0.266, grad_norm=119.623, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.057, optim0_lr0=3.425e-05, train_time=1.793
+[gpub024:0/32] 2024-01-21 08:25:16,783 (trainer:737) INFO: 2epoch:train:5601-5700batch: iter_time=9.252e-05, forward_time=0.208, loss_ctc=223.296, loss_att=198.104, acc=0.210, loss=205.662, backward_time=0.239, grad_norm=79.582, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.053, optim0_lr0=3.442e-05, train_time=1.320
+[gpub024:0/32] 2024-01-21 08:27:52,758 (trainer:737) INFO: 2epoch:train:5701-5800batch: iter_time=9.247e-05, forward_time=0.161, loss_ctc=217.604, loss_att=204.479, acc=0.211, loss=208.417, backward_time=0.225, grad_norm=69.104, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.459e-05, train_time=1.560
+[gpub024:0/32] 2024-01-21 08:29:21,767 (trainer:737) INFO: 2epoch:train:5801-5900batch: iter_time=8.738e-05, forward_time=0.198, loss_ctc=247.443, loss_att=218.163, acc=0.211, loss=226.947, backward_time=0.229, grad_norm=70.090, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.475e-05, train_time=0.890
+[gpub024:0/32] 2024-01-21 08:31:23,160 (trainer:737) INFO: 2epoch:train:5901-6000batch: iter_time=9.375e-05, forward_time=0.174, loss_ctc=208.974, loss_att=192.924, acc=0.213, loss=197.739, backward_time=0.231, grad_norm=70.760, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.052, optim0_lr0=3.492e-05, train_time=1.214
+[gpub024:0/32] 2024-01-21 08:34:14,397 (trainer:737) INFO: 2epoch:train:6001-6100batch: iter_time=8.782e-05, forward_time=0.260, loss_ctc=242.242, loss_att=228.397, acc=0.197, loss=232.550, backward_time=0.242, grad_norm=80.398, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.056, optim0_lr0=3.509e-05, train_time=1.712
+[gpub024:0/32] 2024-01-21 08:36:47,875 (trainer:737) INFO: 2epoch:train:6101-6200batch: iter_time=2.561e-04, forward_time=0.323, loss_ctc=188.975, loss_att=172.481, acc=0.223, loss=177.429, backward_time=0.256, grad_norm=59.407, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.058, optim0_lr0=3.525e-05, train_time=1.534
+[gpub024:0/32] 2024-01-21 08:38:51,607 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub024:0/32] 2024-01-21 08:39:10,542 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 08:39:14,257 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f57807fec80>)
+[gpub024:0/32] 2024-01-21 08:39:14,257 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub024:0/32] 2024-01-21 08:39:14,262 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 08:49:44,113 (trainer:737) INFO: 2epoch:train:6201-6300batch: iter_time=5.987, forward_time=0.230, loss_ctc=228.740, loss_att=204.508, acc=0.221, loss=211.778, backward_time=0.231, grad_norm=83.068, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.053, optim0_lr0=3.542e-05, train_time=7.762
+[gpub024:0/32] 2024-01-21 08:51:05,776 (trainer:737) INFO: 2epoch:train:6301-6400batch: iter_time=8.207e-05, forward_time=0.162, loss_ctc=271.393, loss_att=248.959, acc=0.196, loss=255.689, backward_time=0.229, grad_norm=90.159, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.559e-05, train_time=0.817
+[gpub024:0/32] 2024-01-21 08:52:30,426 (trainer:737) INFO: 2epoch:train:6401-6500batch: iter_time=8.196e-05, forward_time=0.161, loss_ctc=218.958, loss_att=196.417, acc=0.220, loss=203.179, backward_time=0.228, grad_norm=76.215, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.575e-05, train_time=0.846
+[gpub024:0/32] 2024-01-21 08:54:51,499 (trainer:737) INFO: 2epoch:train:6501-6600batch: iter_time=9.141e-05, forward_time=0.272, loss_ctc=210.582, loss_att=197.155, acc=0.219, loss=201.183, backward_time=0.250, grad_norm=59.152, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.066, optim0_lr0=3.592e-05, train_time=1.411
+[gpub024:0/32] 2024-01-21 08:57:42,416 (trainer:737) INFO: 2epoch:train:6601-6700batch: iter_time=9.791e-05, forward_time=0.254, loss_ctc=260.506, loss_att=242.017, acc=0.190, loss=247.564, backward_time=0.292, grad_norm=90.511, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.061, optim0_lr0=3.609e-05, train_time=1.709
+[gpub024:0/32] 2024-01-21 09:00:00,617 (trainer:737) INFO: 2epoch:train:6701-6800batch: iter_time=9.461e-05, forward_time=0.216, loss_ctc=248.125, loss_att=226.360, acc=0.200, loss=232.889, backward_time=0.236, grad_norm=119.333, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.625e-05, train_time=1.381
+[gpub024:0/32] 2024-01-21 09:03:00,907 (trainer:737) INFO: 2epoch:train:6801-6900batch: iter_time=9.318e-05, forward_time=0.198, loss_ctc=184.510, loss_att=169.757, acc=0.227, loss=174.183, backward_time=0.243, grad_norm=79.053, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.056, optim0_lr0=3.642e-05, train_time=1.803
+[gpub024:0/32] 2024-01-21 09:05:27,883 (trainer:737) INFO: 2epoch:train:6901-7000batch: iter_time=9.256e-05, forward_time=0.163, loss_ctc=238.134, loss_att=223.327, acc=0.204, loss=227.769, backward_time=0.226, grad_norm=83.362, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.051, optim0_lr0=3.659e-05, train_time=1.470
+[gpub024:0/32] 2024-01-21 09:07:07,969 (trainer:737) INFO: 2epoch:train:7001-7100batch: iter_time=9.336e-05, forward_time=0.307, loss_ctc=216.896, loss_att=207.462, acc=0.214, loss=210.292, backward_time=0.239, grad_norm=60.200, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.068, optim0_lr0=3.675e-05, train_time=1.001
+[gpub024:0/32] 2024-01-21 09:09:40,290 (trainer:737) INFO: 2epoch:train:7101-7200batch: iter_time=9.464e-05, forward_time=0.284, loss_ctc=220.562, loss_att=205.600, acc=0.214, loss=210.089, backward_time=0.279, grad_norm=74.969, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.072, optim0_lr0=3.692e-05, train_time=1.523
+[gpub024:0/32] 2024-01-21 09:11:57,556 (trainer:737) INFO: 2epoch:train:7201-7300batch: iter_time=9.826e-05, forward_time=0.210, loss_ctc=202.133, loss_att=189.144, acc=0.214, loss=193.040, backward_time=0.225, grad_norm=69.761, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.709e-05, train_time=1.373
+[gpub024:0/32] 2024-01-21 09:14:55,274 (trainer:737) INFO: 2epoch:train:7301-7400batch: iter_time=1.023e-04, forward_time=0.160, loss_ctc=211.126, loss_att=202.855, acc=0.213, loss=205.337, backward_time=0.224, grad_norm=66.765, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.725e-05, train_time=1.777
+[gpub024:0/32] 2024-01-21 09:17:02,278 (trainer:737) INFO: 2epoch:train:7401-7500batch: iter_time=9.630e-05, forward_time=0.160, loss_ctc=199.691, loss_att=188.392, acc=0.222, loss=191.782, backward_time=0.225, grad_norm=86.864, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.742e-05, train_time=1.270
+[gpub024:0/32] 2024-01-21 09:17:22,307 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub024:0/32] 2024-01-21 09:17:41,572 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 09:17:45,223 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f57c7bb2770>)
+[gpub024:0/32] 2024-01-21 09:17:45,223 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub024:0/32] 2024-01-21 09:17:45,249 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 09:28:51,163 (trainer:737) INFO: 2epoch:train:7501-7600batch: iter_time=5.842, forward_time=0.196, loss_ctc=239.677, loss_att=224.630, acc=0.206, loss=229.144, backward_time=0.230, grad_norm=82.967, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.052, optim0_lr0=3.759e-05, train_time=7.089
+[gpub024:0/32] 2024-01-21 09:31:17,265 (trainer:737) INFO: 2epoch:train:7601-7700batch: iter_time=8.296e-05, forward_time=0.196, loss_ctc=219.073, loss_att=203.958, acc=0.218, loss=208.492, backward_time=0.236, grad_norm=71.932, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.055, optim0_lr0=3.775e-05, train_time=1.461
+[gpub024:0/32] 2024-01-21 09:33:05,428 (trainer:737) INFO: 2epoch:train:7701-7800batch: iter_time=8.374e-05, forward_time=0.304, loss_ctc=213.855, loss_att=198.193, acc=0.225, loss=202.891, backward_time=0.259, grad_norm=68.242, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.079, optim0_lr0=3.792e-05, train_time=1.081
+[gpub024:0/32] 2024-01-21 09:34:41,034 (trainer:737) INFO: 2epoch:train:7801-7900batch: iter_time=8.207e-05, forward_time=0.162, loss_ctc=221.706, loss_att=216.057, acc=0.206, loss=217.751, backward_time=0.230, grad_norm=65.031, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.809e-05, train_time=0.956
+[gpub024:0/32] 2024-01-21 09:36:53,141 (trainer:737) INFO: 2epoch:train:7901-8000batch: iter_time=8.311e-05, forward_time=0.161, loss_ctc=239.975, loss_att=220.096, acc=0.200, loss=226.060, backward_time=0.227, grad_norm=85.575, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.825e-05, train_time=1.321
+[gpub024:0/32] 2024-01-21 09:39:51,136 (trainer:737) INFO: 2epoch:train:8001-8100batch: iter_time=8.436e-05, forward_time=0.161, loss_ctc=208.967, loss_att=198.476, acc=0.214, loss=201.623, backward_time=0.223, grad_norm=84.455, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.842e-05, train_time=1.780
+[gpub024:0/32] 2024-01-21 09:43:18,224 (trainer:737) INFO: 2epoch:train:8101-8200batch: iter_time=8.673e-05, forward_time=0.190, loss_ctc=200.780, loss_att=191.624, acc=0.217, loss=194.371, backward_time=0.252, grad_norm=67.507, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.055, optim0_lr0=3.859e-05, train_time=2.070
+[gpub024:0/32] 2024-01-21 09:45:27,691 (trainer:737) INFO: 2epoch:train:8201-8300batch: iter_time=8.910e-05, forward_time=0.211, loss_ctc=198.574, loss_att=195.617, acc=0.220, loss=196.504, backward_time=0.240, grad_norm=61.646, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.055, optim0_lr0=3.875e-05, train_time=1.295
+[gpub024:0/32] 2024-01-21 09:48:10,554 (trainer:737) INFO: 2epoch:train:8301-8400batch: iter_time=9.070e-05, forward_time=0.453, loss_ctc=221.790, loss_att=209.065, acc=0.218, loss=212.882, backward_time=0.274, grad_norm=68.268, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.064, optim0_lr0=3.892e-05, train_time=1.629
+[gpub024:0/32] 2024-01-21 09:49:41,631 (trainer:737) INFO: 2epoch:train:8401-8500batch: iter_time=8.523e-05, forward_time=0.160, loss_ctc=187.897, loss_att=184.793, acc=0.223, loss=185.724, backward_time=0.227, grad_norm=66.366, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.909e-05, train_time=0.911
+[gpub024:0/32] 2024-01-21 09:52:20,281 (trainer:737) INFO: 2epoch:train:8501-8600batch: iter_time=8.635e-05, forward_time=0.162, loss_ctc=218.076, loss_att=215.271, acc=0.205, loss=216.113, backward_time=0.225, grad_norm=68.686, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.925e-05, train_time=1.586
+[gpub024:0/32] 2024-01-21 09:54:56,536 (trainer:737) INFO: 2epoch:train:8601-8700batch: iter_time=8.389e-05, forward_time=0.159, loss_ctc=169.898, loss_att=167.304, acc=0.227, loss=168.082, backward_time=0.223, grad_norm=51.693, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.942e-05, train_time=1.563
+[gpub024:0/32] 2024-01-21 09:56:31,531 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub024:0/32] 2024-01-21 09:56:50,296 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 09:56:54,253 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f57848978e0>)
+[gpub024:0/32] 2024-01-21 09:56:54,253 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub024:0/32] 2024-01-21 09:56:54,256 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 10:06:24,976 (trainer:737) INFO: 2epoch:train:8701-8800batch: iter_time=5.535, forward_time=0.265, loss_ctc=204.846, loss_att=201.447, acc=0.227, loss=202.467, backward_time=0.239, grad_norm=77.903, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.055, optim0_lr0=3.959e-05, train_time=6.884
+[gpub024:0/32] 2024-01-21 10:08:03,109 (trainer:737) INFO: 2epoch:train:8801-8900batch: iter_time=8.002e-05, forward_time=0.163, loss_ctc=242.449, loss_att=246.009, acc=0.201, loss=244.941, backward_time=0.228, grad_norm=89.561, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.975e-05, train_time=0.982
+[gpub024:0/32] 2024-01-21 10:10:38,522 (trainer:737) INFO: 2epoch:train:8901-9000batch: iter_time=8.131e-05, forward_time=0.161, loss_ctc=201.129, loss_att=192.800, acc=0.227, loss=195.299, backward_time=0.226, grad_norm=77.445, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.051, optim0_lr0=3.992e-05, train_time=1.554
+[gpub024:0/32] 2024-01-21 10:13:00,254 (trainer:737) INFO: 2epoch:train:9001-9100batch: iter_time=8.223e-05, forward_time=0.161, loss_ctc=187.176, loss_att=192.050, acc=0.226, loss=190.588, backward_time=0.226, grad_norm=58.931, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.009e-05, train_time=1.417
+[gpub024:0/32] 2024-01-21 10:15:33,656 (trainer:737) INFO: 2epoch:train:9101-9200batch: iter_time=8.065e-05, forward_time=0.163, loss_ctc=232.664, loss_att=237.841, acc=0.196, loss=236.288, backward_time=0.227, grad_norm=75.046, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.025e-05, train_time=1.534
+[gpub024:0/32] 2024-01-21 10:18:06,971 (trainer:737) INFO: 2epoch:train:9201-9300batch: iter_time=8.247e-05, forward_time=0.162, loss_ctc=220.542, loss_att=221.388, acc=0.206, loss=221.135, backward_time=0.226, grad_norm=85.394, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.042e-05, train_time=1.533
+[gpub024:0/32] 2024-01-21 10:21:37,086 (trainer:737) INFO: 2epoch:train:9301-9400batch: iter_time=8.920e-05, forward_time=0.345, loss_ctc=163.426, loss_att=164.934, acc=0.235, loss=164.482, backward_time=0.412, grad_norm=58.719, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.086, optim0_lr0=4.059e-05, train_time=2.101
+[gpub024:0/32] 2024-01-21 10:24:48,473 (trainer:737) INFO: 2epoch:train:9401-9500batch: iter_time=9.323e-05, forward_time=0.162, loss_ctc=211.375, loss_att=217.735, acc=0.212, loss=215.827, backward_time=0.226, grad_norm=64.868, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.075e-05, train_time=1.914
+[gpub024:0/32] 2024-01-21 10:27:07,156 (trainer:737) INFO: 2epoch:train:9501-9600batch: iter_time=9.232e-05, forward_time=0.162, loss_ctc=191.529, loss_att=203.279, acc=0.220, loss=199.754, backward_time=0.226, grad_norm=55.924, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.092e-05, train_time=1.387
+[gpub024:0/32] 2024-01-21 10:28:36,263 (trainer:737) INFO: 2epoch:train:9601-9700batch: iter_time=8.804e-05, forward_time=0.161, loss_ctc=196.695, loss_att=200.832, acc=0.221, loss=199.591, backward_time=0.227, grad_norm=68.243, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.109e-05, train_time=0.891
+[gpub024:0/32] 2024-01-21 10:31:30,236 (trainer:737) INFO: 2epoch:train:9701-9800batch: iter_time=8.059e-05, forward_time=0.160, loss_ctc=180.509, loss_att=184.686, acc=0.221, loss=183.433, backward_time=0.224, grad_norm=62.165, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.125e-05, train_time=1.740
+[gpub024:0/32] 2024-01-21 10:33:54,061 (trainer:737) INFO: 2epoch:train:9801-9900batch: iter_time=8.463e-05, forward_time=0.161, loss_ctc=189.776, loss_att=198.824, acc=0.220, loss=196.110, backward_time=0.224, grad_norm=66.974, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.142e-05, train_time=1.438
+[gpub024:0/32] 2024-01-21 10:35:41,480 (trainer:737) INFO: 2epoch:train:9901-10000batch: iter_time=8.250e-05, forward_time=0.208, loss_ctc=177.553, loss_att=185.585, acc=0.229, loss=183.175, backward_time=0.270, grad_norm=66.750, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.069, optim0_lr0=4.159e-05, train_time=1.074
+[gpub024:0/32] 2024-01-21 10:36:01,508 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub024:0/32] 2024-01-21 10:36:20,612 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 10:36:24,315 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f581cfcf490>)
+[gpub024:0/32] 2024-01-21 10:36:24,315 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub024:0/32] 2024-01-21 10:36:24,318 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 10:46:19,711 (trainer:737) INFO: 2epoch:train:10001-10100batch: iter_time=5.291, forward_time=0.212, loss_ctc=211.537, loss_att=225.597, acc=0.213, loss=221.379, backward_time=0.238, grad_norm=76.161, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.053, optim0_lr0=4.175e-05, train_time=6.382
+[gpub024:0/32] 2024-01-21 10:48:01,575 (trainer:737) INFO: 2epoch:train:10101-10200batch: iter_time=8.986e-05, forward_time=0.161, loss_ctc=197.284, loss_att=203.088, acc=0.223, loss=201.347, backward_time=0.227, grad_norm=71.489, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.192e-05, train_time=1.019
+[gpub024:0/32] 2024-01-21 10:51:31,483 (trainer:737) INFO: 2epoch:train:10201-10300batch: iter_time=9.015e-05, forward_time=0.180, loss_ctc=191.913, loss_att=194.213, acc=0.233, loss=193.523, backward_time=0.275, grad_norm=62.049, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.059, optim0_lr0=4.209e-05, train_time=2.099
+[gpub024:0/32] 2024-01-21 10:54:13,703 (trainer:737) INFO: 2epoch:train:10301-10400batch: iter_time=8.781e-05, forward_time=0.163, loss_ctc=195.989, loss_att=217.526, acc=0.211, loss=211.065, backward_time=0.227, grad_norm=59.191, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.225e-05, train_time=1.622
+[gpub024:0/32] 2024-01-21 10:56:03,623 (trainer:737) INFO: 2epoch:train:10401-10500batch: iter_time=9.124e-05, forward_time=0.162, loss_ctc=213.058, loss_att=221.674, acc=0.203, loss=219.089, backward_time=0.228, grad_norm=75.719, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.242e-05, train_time=1.099
+[gpub024:0/32] 2024-01-21 10:58:44,097 (trainer:737) INFO: 2epoch:train:10501-10600batch: iter_time=2.653e-04, forward_time=0.360, loss_ctc=186.351, loss_att=197.085, acc=0.219, loss=193.865, backward_time=0.295, grad_norm=69.272, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.064, optim0_lr0=4.259e-05, train_time=1.605
+[gpub024:0/32] 2024-01-21 11:00:14,028 (trainer:737) INFO: 2epoch:train:10601-10700batch: iter_time=1.024e-04, forward_time=0.162, loss_ctc=179.379, loss_att=190.717, acc=0.223, loss=187.316, backward_time=0.228, grad_norm=62.268, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.275e-05, train_time=0.899
+[gpub024:0/32] 2024-01-21 11:02:04,285 (trainer:737) INFO: 2epoch:train:10701-10800batch: iter_time=8.979e-05, forward_time=0.237, loss_ctc=177.860, loss_att=194.007, acc=0.226, loss=189.163, backward_time=0.232, grad_norm=54.215, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.052, optim0_lr0=4.292e-05, train_time=1.102
+[gpub024:0/32] 2024-01-21 11:05:05,767 (trainer:737) INFO: 2epoch:train:10801-10900batch: iter_time=9.648e-05, forward_time=0.162, loss_ctc=196.944, loss_att=209.120, acc=0.222, loss=205.467, backward_time=0.226, grad_norm=60.768, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.309e-05, train_time=1.815
+[gpub024:0/32] 2024-01-21 11:07:20,467 (trainer:737) INFO: 2epoch:train:10901-11000batch: iter_time=9.434e-05, forward_time=0.161, loss_ctc=167.046, loss_att=183.033, acc=0.227, loss=178.237, backward_time=0.225, grad_norm=58.283, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.051, optim0_lr0=4.325e-05, train_time=1.347
+[gpub024:0/32] 2024-01-21 11:09:34,137 (trainer:737) INFO: 2epoch:train:11001-11100batch: iter_time=8.874e-05, forward_time=0.162, loss_ctc=194.706, loss_att=217.799, acc=0.211, loss=210.871, backward_time=0.228, grad_norm=62.817, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.342e-05, train_time=1.336
+[gpub024:0/32] 2024-01-21 11:11:29,609 (trainer:737) INFO: 2epoch:train:11101-11200batch: iter_time=8.806e-05, forward_time=0.192, loss_ctc=152.200, loss_att=164.902, acc=0.235, loss=161.092, backward_time=0.297, grad_norm=51.139, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.058, optim0_lr0=4.359e-05, train_time=1.155
+[gpub024:0/32] 2024-01-21 11:12:52,230 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub024:0/32] 2024-01-21 11:13:11,431 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 11:13:15,040 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f57bc7d4cd0>)
+[gpub024:0/32] 2024-01-21 11:13:15,040 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub024:0/32] 2024-01-21 11:13:15,044 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 11:22:24,187 (trainer:737) INFO: 2epoch:train:11201-11300batch: iter_time=5.274, forward_time=0.316, loss_ctc=182.443, loss_att=195.045, acc=0.234, loss=191.265, backward_time=0.260, grad_norm=67.292, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.060, optim0_lr0=4.375e-05, train_time=6.546
+[gpub024:0/32] 2024-01-21 11:23:48,392 (trainer:737) INFO: 2epoch:train:11301-11400batch: iter_time=8.878e-05, forward_time=0.163, loss_ctc=216.806, loss_att=235.576, acc=0.208, loss=229.945, backward_time=0.228, grad_norm=82.055, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.392e-05, train_time=0.842
+[gpub024:0/32] 2024-01-21 11:26:12,208 (trainer:737) INFO: 2epoch:train:11401-11500batch: iter_time=8.791e-05, forward_time=0.161, loss_ctc=183.590, loss_att=187.158, acc=0.234, loss=186.088, backward_time=0.226, grad_norm=69.442, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.409e-05, train_time=1.438
+[gpub024:0/32] 2024-01-21 11:29:12,492 (trainer:737) INFO: 2epoch:train:11501-11600batch: iter_time=8.466e-05, forward_time=0.164, loss_ctc=166.442, loss_att=186.899, acc=0.233, loss=180.762, backward_time=0.225, grad_norm=51.578, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.425e-05, train_time=1.803
+[gpub024:0/32] 2024-01-21 11:31:25,695 (trainer:737) INFO: 2epoch:train:11601-11700batch: iter_time=8.335e-05, forward_time=0.162, loss_ctc=208.882, loss_att=228.034, acc=0.200, loss=222.289, backward_time=0.228, grad_norm=65.881, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.442e-05, train_time=1.332
+[gpub024:0/32] 2024-01-21 11:33:42,231 (trainer:737) INFO: 2epoch:train:11701-11800batch: iter_time=8.475e-05, forward_time=0.372, loss_ctc=195.765, loss_att=213.355, acc=0.212, loss=208.078, backward_time=0.265, grad_norm=73.868, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.078, optim0_lr0=4.459e-05, train_time=1.365
+[gpub024:0/32] 2024-01-21 11:36:20,515 (trainer:737) INFO: 2epoch:train:11801-11900batch: iter_time=8.881e-05, forward_time=0.176, loss_ctc=147.160, loss_att=160.891, acc=0.240, loss=156.772, backward_time=0.281, grad_norm=58.215, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.071, optim0_lr0=4.475e-05, train_time=1.583
+[gpub024:0/32] 2024-01-21 11:39:02,234 (trainer:737) INFO: 2epoch:train:11901-12000batch: iter_time=9.437e-05, forward_time=0.162, loss_ctc=190.632, loss_att=210.824, acc=0.219, loss=204.766, backward_time=0.226, grad_norm=64.620, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.492e-05, train_time=1.617
+[gpub024:0/32] 2024-01-21 11:42:01,528 (trainer:737) INFO: 2epoch:train:12001-12100batch: iter_time=9.269e-05, forward_time=0.161, loss_ctc=171.226, loss_att=195.649, acc=0.227, loss=188.322, backward_time=0.226, grad_norm=50.466, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.509e-05, train_time=1.792
+[gpub024:0/32] 2024-01-21 11:43:57,004 (trainer:737) INFO: 2epoch:train:12101-12200batch: iter_time=8.575e-05, forward_time=0.161, loss_ctc=175.791, loss_att=194.657, acc=0.229, loss=188.997, backward_time=0.225, grad_norm=58.546, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.525e-05, train_time=1.155
+[gpub024:0/32] 2024-01-21 11:45:41,952 (trainer:737) INFO: 2epoch:train:12201-12300batch: iter_time=8.361e-05, forward_time=0.160, loss_ctc=162.117, loss_att=177.448, acc=0.228, loss=172.849, backward_time=0.224, grad_norm=56.202, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.542e-05, train_time=1.049
+[gpub024:0/32] 2024-01-21 11:47:53,370 (trainer:737) INFO: 2epoch:train:12301-12400batch: iter_time=8.567e-05, forward_time=0.315, loss_ctc=170.480, loss_att=191.648, acc=0.224, loss=185.297, backward_time=0.298, grad_norm=56.002, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.078, optim0_lr0=4.559e-05, train_time=1.314
+[gpub024:0/32] 2024-01-21 11:50:21,709 (trainer:737) INFO: 2epoch:train:12401-12500batch: iter_time=8.174e-05, forward_time=0.227, loss_ctc=158.129, loss_att=179.541, acc=0.236, loss=173.117, backward_time=0.250, grad_norm=56.338, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.058, optim0_lr0=4.575e-05, train_time=1.483
+[gpub024:0/32] 2024-01-21 11:50:41,737 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub024:0/32] 2024-01-21 11:51:01,089 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 11:51:04,991 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5816bf75b0>)
+[gpub024:0/32] 2024-01-21 11:51:04,991 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub024:0/32] 2024-01-21 11:51:04,994 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 12:03:54,844 (trainer:737) INFO: 2epoch:train:12501-12600batch: iter_time=6.743, forward_time=0.166, loss_ctc=189.494, loss_att=214.330, acc=0.221, loss=206.879, backward_time=0.228, grad_norm=67.290, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.592e-05, train_time=8.131
+[gpub024:0/32] 2024-01-21 12:05:15,765 (trainer:737) INFO: 2epoch:train:12601-12700batch: iter_time=8.742e-05, forward_time=0.161, loss_ctc=176.820, loss_att=195.331, acc=0.231, loss=189.777, backward_time=0.227, grad_norm=62.074, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.609e-05, train_time=0.809
+[gpub024:0/32] 2024-01-21 12:06:59,941 (trainer:737) INFO: 2epoch:train:12701-12800batch: iter_time=8.314e-05, forward_time=0.162, loss_ctc=173.997, loss_att=190.113, acc=0.237, loss=185.278, backward_time=0.228, grad_norm=57.607, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.051, optim0_lr0=4.625e-05, train_time=1.042
+[gpub024:0/32] 2024-01-21 12:10:07,465 (trainer:737) INFO: 2epoch:train:12801-12900batch: iter_time=8.637e-05, forward_time=0.323, loss_ctc=175.678, loss_att=206.467, acc=0.219, loss=197.230, backward_time=0.377, grad_norm=55.567, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.078, optim0_lr0=4.642e-05, train_time=1.875
+[gpub024:0/32] 2024-01-21 12:12:26,216 (trainer:737) INFO: 2epoch:train:12901-13000batch: iter_time=8.736e-05, forward_time=0.225, loss_ctc=193.775, loss_att=214.246, acc=0.212, loss=208.105, backward_time=0.236, grad_norm=72.197, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.075, optim0_lr0=4.659e-05, train_time=1.387
+[gpub024:0/32] 2024-01-21 12:15:47,450 (trainer:737) INFO: 2epoch:train:13001-13100batch: iter_time=9.301e-05, forward_time=0.161, loss_ctc=167.376, loss_att=189.572, acc=0.227, loss=182.913, backward_time=0.224, grad_norm=67.216, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.675e-05, train_time=2.012
+[gpub024:0/32] 2024-01-21 12:19:23,314 (trainer:737) INFO: 2epoch:train:13101-13200batch: iter_time=8.833e-05, forward_time=0.161, loss_ctc=160.247, loss_att=184.753, acc=0.231, loss=177.401, backward_time=0.223, grad_norm=57.673, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.692e-05, train_time=2.158
+[gpub024:0/32] 2024-01-21 12:21:00,301 (trainer:737) INFO: 2epoch:train:13201-13300batch: iter_time=8.285e-05, forward_time=0.164, loss_ctc=160.907, loss_att=189.457, acc=0.232, loss=180.892, backward_time=0.227, grad_norm=49.905, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.709e-05, train_time=0.970
+[gpub024:0/32] 2024-01-21 12:22:49,558 (trainer:737) INFO: 2epoch:train:13301-13400batch: iter_time=8.492e-05, forward_time=0.161, loss_ctc=176.327, loss_att=201.275, acc=0.229, loss=193.790, backward_time=0.227, grad_norm=55.806, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.725e-05, train_time=1.092
+[gpub024:0/32] 2024-01-21 12:25:07,496 (trainer:737) INFO: 2epoch:train:13401-13500batch: iter_time=8.740e-05, forward_time=0.421, loss_ctc=150.789, loss_att=178.223, acc=0.233, loss=169.993, backward_time=0.312, grad_norm=52.181, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.064, optim0_lr0=4.742e-05, train_time=1.379
+[gpub024:0/32] 2024-01-21 12:26:49,183 (trainer:737) INFO: 2epoch:train:13501-13600batch: iter_time=8.187e-05, forward_time=0.228, loss_ctc=175.408, loss_att=207.893, acc=0.216, loss=198.147, backward_time=0.232, grad_norm=59.868, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.058, optim0_lr0=4.759e-05, train_time=1.017
+[gpub024:0/32] 2024-01-21 12:29:05,911 (trainer:737) INFO: 2epoch:train:13601-13700batch: iter_time=8.967e-05, forward_time=0.159, loss_ctc=137.150, loss_att=160.655, acc=0.239, loss=153.604, backward_time=0.224, grad_norm=48.053, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.775e-05, train_time=1.367
+[gpub024:0/32] 2024-01-21 12:30:37,269 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub024:0/32] 2024-01-21 12:30:56,249 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 12:30:59,817 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f581612fd30>)
+[gpub024:0/32] 2024-01-21 12:30:59,817 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub024:0/32] 2024-01-21 12:30:59,821 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 12:42:01,547 (trainer:737) INFO: 2epoch:train:13701-13800batch: iter_time=6.356, forward_time=0.164, loss_ctc=163.751, loss_att=193.859, acc=0.241, loss=184.827, backward_time=0.225, grad_norm=63.970, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.792e-05, train_time=7.755
+[gpub024:0/32] 2024-01-21 12:43:22,498 (trainer:737) INFO: 2epoch:train:13801-13900batch: iter_time=8.267e-05, forward_time=0.162, loss_ctc=192.258, loss_att=237.489, acc=0.213, loss=223.920, backward_time=0.229, grad_norm=64.545, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.809e-05, train_time=0.810
+[gpub024:0/32] 2024-01-21 12:44:55,692 (trainer:737) INFO: 2epoch:train:13901-14000batch: iter_time=8.501e-05, forward_time=0.209, loss_ctc=166.963, loss_att=185.217, acc=0.240, loss=179.741, backward_time=0.257, grad_norm=65.861, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.066, optim0_lr0=4.825e-05, train_time=0.932
+[gpub024:0/32] 2024-01-21 12:47:12,062 (trainer:737) INFO: 2epoch:train:14001-14100batch: iter_time=8.791e-05, forward_time=0.430, loss_ctc=147.813, loss_att=184.558, acc=0.239, loss=173.535, backward_time=0.277, grad_norm=47.799, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.074, optim0_lr0=4.842e-05, train_time=1.363
+[gpub024:0/32] 2024-01-21 12:51:09,162 (trainer:737) INFO: 2epoch:train:14101-14200batch: iter_time=8.721e-05, forward_time=0.172, loss_ctc=188.929, loss_att=230.608, acc=0.206, loss=218.104, backward_time=0.238, grad_norm=61.021, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.060, optim0_lr0=4.859e-05, train_time=2.372
+[gpub024:0/32] 2024-01-21 12:53:27,429 (trainer:737) INFO: 2epoch:train:14201-14300batch: iter_time=9.045e-05, forward_time=0.162, loss_ctc=176.423, loss_att=214.171, acc=0.219, loss=202.846, backward_time=0.226, grad_norm=62.499, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.875e-05, train_time=1.382
+[gpub024:0/32] 2024-01-21 12:56:43,908 (trainer:737) INFO: 2epoch:train:14301-14400batch: iter_time=8.738e-05, forward_time=0.161, loss_ctc=131.590, loss_att=157.790, acc=0.248, loss=149.930, backward_time=0.221, grad_norm=49.061, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.892e-05, train_time=1.964
+[gpub024:0/32] 2024-01-21 12:59:03,330 (trainer:737) INFO: 2epoch:train:14401-14500batch: iter_time=8.467e-05, forward_time=0.162, loss_ctc=172.757, loss_att=211.088, acc=0.225, loss=199.588, backward_time=0.228, grad_norm=60.452, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.909e-05, train_time=1.395
+[gpub024:0/32] 2024-01-21 13:00:57,080 (trainer:737) INFO: 2epoch:train:14501-14600batch: iter_time=8.548e-05, forward_time=0.348, loss_ctc=153.311, loss_att=196.703, acc=0.232, loss=183.685, backward_time=0.280, grad_norm=48.713, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.080, optim0_lr0=4.925e-05, train_time=1.137
+[gpub024:0/32] 2024-01-21 13:03:52,273 (trainer:737) INFO: 2epoch:train:14601-14700batch: iter_time=8.819e-05, forward_time=0.232, loss_ctc=158.150, loss_att=192.988, acc=0.234, loss=182.537, backward_time=0.235, grad_norm=56.852, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.070, optim0_lr0=4.942e-05, train_time=1.751
+[gpub024:0/32] 2024-01-21 13:06:57,468 (trainer:737) INFO: 2epoch:train:14701-14800batch: iter_time=8.096e-05, forward_time=0.162, loss_ctc=145.524, loss_att=177.890, acc=0.234, loss=168.180, backward_time=0.223, grad_norm=50.138, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.959e-05, train_time=1.852
+[gpub024:0/32] 2024-01-21 13:08:39,278 (trainer:737) INFO: 2epoch:train:14801-14900batch: iter_time=8.262e-05, forward_time=0.161, loss_ctc=153.451, loss_att=190.574, acc=0.232, loss=179.437, backward_time=0.227, grad_norm=52.955, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.975e-05, train_time=1.018
+[gpub024:0/32] 2024-01-21 13:11:53,994 (trainer:737) INFO: 2epoch:train:14901-15000batch: iter_time=8.009e-05, forward_time=0.160, loss_ctc=142.911, loss_att=178.184, acc=0.242, loss=167.602, backward_time=0.223, grad_norm=54.443, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.051, optim0_lr0=4.992e-05, train_time=1.947
+[gpub024:0/32] 2024-01-21 13:48:22,540 (trainer:343) INFO: 2epoch results: [train] iter_time=0.482, forward_time=0.204, loss_ctc=211.941, loss_att=206.146, acc=0.213, loss=207.885, backward_time=0.242, grad_norm=78.034, clip=100.000, loss_scale=2.847e+08, optim_step_time=0.056, optim0_lr0=3.750e-05, train_time=1.840, time=7 hours, 40 minutes and 26.73 seconds, total_count=30000, gpu_max_cached_mem_GB=25.521, [valid] loss_ctc=147.384, cer_ctc=0.733, loss_att=149.420, acc=0.185, cer=0.708, wer=1.000, loss=148.809, time=36 minutes and 3.37 seconds, total_count=9342, gpu_max_cached_mem_GB=25.521
+[gpub024:0/32] 2024-01-21 13:48:35,183 (trainer:391) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub024:0/32] 2024-01-21 13:48:35,184 (trainer:272) INFO: 3/45epoch started. Estimated time to finish: 2 weeks, 3 days and 4 hours
+[gpub024:0/32] 2024-01-21 13:48:35,195 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub024:0/32] 2024-01-21 13:48:53,028 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 13:48:56,417 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f589aa86230>)
+[gpub024:0/32] 2024-01-21 13:48:56,417 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub024:0/32] 2024-01-21 13:48:56,420 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 14:00:42,551 (trainer:737) INFO: 3epoch:train:1-100batch: iter_time=6.334, forward_time=0.236, loss_ctc=164.879, loss_att=197.839, acc=0.227, loss=187.951, backward_time=0.261, grad_norm=67.852, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.059, optim0_lr0=5.077e-05, train_time=7.273
+[gpub024:0/32] 2024-01-21 14:02:03,118 (trainer:737) INFO: 3epoch:train:101-200batch: iter_time=7.964e-05, forward_time=0.161, loss_ctc=150.342, loss_att=196.520, acc=0.231, loss=182.667, backward_time=0.235, grad_norm=56.883, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=5.227e-05, train_time=0.805
+[gpub024:0/32] 2024-01-21 14:03:39,420 (trainer:737) INFO: 3epoch:train:201-300batch: iter_time=8.633e-05, forward_time=0.162, loss_ctc=156.844, loss_att=200.873, acc=0.225, loss=187.664, backward_time=0.228, grad_norm=61.095, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=5.377e-05, train_time=0.963
+[gpub024:0/32] 2024-01-21 14:05:44,638 (trainer:737) INFO: 3epoch:train:301-400batch: iter_time=8.713e-05, forward_time=0.161, loss_ctc=160.456, loss_att=204.208, acc=0.226, loss=191.083, backward_time=0.228, grad_norm=53.251, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=5.527e-05, train_time=1.252
+[gpub024:0/32] 2024-01-21 14:08:09,494 (trainer:737) INFO: 3epoch:train:401-500batch: iter_time=8.428e-05, forward_time=0.161, loss_ctc=167.574, loss_att=203.804, acc=0.228, loss=192.935, backward_time=0.226, grad_norm=70.335, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=5.677e-05, train_time=1.447
+[gpub024:0/32] 2024-01-21 14:10:54,410 (trainer:737) INFO: 3epoch:train:501-600batch: iter_time=8.870e-05, forward_time=0.161, loss_ctc=151.030, loss_att=193.038, acc=0.226, loss=180.436, backward_time=0.225, grad_norm=62.493, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=5.827e-05, train_time=1.650
+[gpub024:0/32] 2024-01-21 14:13:24,214 (trainer:737) INFO: 3epoch:train:601-700batch: iter_time=9.219e-05, forward_time=0.269, loss_ctc=151.594, loss_att=194.530, acc=0.232, loss=181.649, backward_time=0.261, grad_norm=52.749, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.066, optim0_lr0=5.977e-05, train_time=1.498
+[gpub024:0/32] 2024-01-21 14:16:36,449 (trainer:737) INFO: 3epoch:train:701-800batch: iter_time=4.938e-04, forward_time=0.416, loss_ctc=186.657, loss_att=215.152, acc=0.221, loss=206.604, backward_time=0.306, grad_norm=80.094, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.097, optim0_lr0=6.127e-05, train_time=1.922
+[gpub024:0/32] 2024-01-21 14:18:59,718 (trainer:737) INFO: 3epoch:train:801-900batch: iter_time=9.184e-05, forward_time=0.161, loss_ctc=156.436, loss_att=210.908, acc=0.224, loss=194.567, backward_time=0.224, grad_norm=56.058, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=6.277e-05, train_time=1.432
+[gpub024:0/32] 2024-01-21 14:21:10,574 (trainer:737) INFO: 3epoch:train:901-1000batch: iter_time=8.679e-05, forward_time=0.162, loss_ctc=153.718, loss_att=209.820, acc=0.226, loss=192.989, backward_time=0.226, grad_norm=59.704, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=6.427e-05, train_time=1.308
+[gpub024:0/32] 2024-01-21 14:23:17,185 (trainer:737) INFO: 3epoch:train:1001-1100batch: iter_time=8.682e-05, forward_time=0.161, loss_ctc=149.274, loss_att=177.560, acc=0.239, loss=169.074, backward_time=0.225, grad_norm=66.689, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=6.577e-05, train_time=1.266
+[gpub024:0/32] 2024-01-21 14:25:56,105 (trainer:737) INFO: 3epoch:train:1101-1200batch: iter_time=8.427e-05, forward_time=0.161, loss_ctc=161.349, loss_att=207.731, acc=0.224, loss=193.816, backward_time=0.226, grad_norm=57.282, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=6.727e-05, train_time=1.589
+[gpub024:0/32] 2024-01-21 14:27:55,600 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub024:0/32] 2024-01-21 14:28:14,801 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub024:0/32] 2024-01-21 14:28:18,408 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f5c0e88b1c0>)
+[gpub024:0/32] 2024-01-21 14:28:18,408 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub024:0/32] 2024-01-21 14:28:18,411 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub024:0/32] 2024-01-21 14:39:54,078 (trainer:737) INFO: 3epoch:train:1201-1300batch: iter_time=6.851, forward_time=0.230, loss_ctc=143.889, loss_att=177.295, acc=0.245, loss=167.273, backward_time=0.242, grad_norm=61.605, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.054, optim0_lr0=6.877e-05, train_time=8.380
+[gpub024:0/32] 2024-01-21 14:41:35,222 (trainer:737) INFO: 3epoch:train:1301-1400batch: iter_time=8.133e-05, forward_time=0.163, loss_ctc=168.320, loss_att=209.472, acc=0.220, loss=197.126, backward_time=0.228, grad_norm=69.991, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=7.027e-05, train_time=1.011
+[gpub024:0/32] 2024-01-21 14:43:38,843 (trainer:737) INFO: 3epoch:train:1401-1500batch: iter_time=8.230e-05, forward_time=0.161, loss_ctc=152.600, loss_att=207.012, acc=0.231, loss=190.689, backward_time=0.224, grad_norm=59.556, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=7.177e-05, train_time=1.236
+[gpub024:0/32] 2024-01-21 14:45:16,055 (trainer:737) INFO: 3epoch:train:1501-1600batch: iter_time=8.089e-05, forward_time=0.160, loss_ctc=139.486, loss_att=175.690, acc=0.230, loss=164.829, backward_time=0.226, grad_norm=57.351, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=7.327e-05, train_time=0.972
+[gpub024:0/32] 2024-01-21 14:47:49,377 (trainer:737) INFO: 3epoch:train:1601-1700batch: iter_time=7.928e-05, forward_time=0.164, loss_ctc=172.898, loss_att=231.329, acc=0.226, loss=213.800, backward_time=0.236, grad_norm=68.368, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=7.477e-05, train_time=1.533
+[gpub024:0/32] 2024-01-21 14:50:05,665 (trainer:737) INFO: 3epoch:train:1701-1800batch: iter_time=8.141e-05, forward_time=0.161, loss_ctc=145.924, loss_att=187.528, acc=0.227, loss=175.047, backward_time=0.225, grad_norm=58.942, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.052, optim0_lr0=7.627e-05, train_time=1.363
+[gpub024:0/32] 2024-01-21 14:53:09,718 (trainer:737) INFO: 3epoch:train:1801-1900batch: iter_time=8.646e-05, forward_time=0.160, loss_ctc=142.683, loss_att=184.343, acc=0.235, loss=171.845, backward_time=0.224, grad_norm=58.001, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.051, optim0_lr0=7.777e-05, train_time=1.840
+[gpub024:0/32] 2024-01-21 14:56:10,488 (trainer:737) INFO: 3epoch:train:1901-2000batch: iter_time=8.438e-05, forward_time=0.342, loss_ctc=160.193, loss_att=203.916, acc=0.224, loss=190.799, backward_time=0.312, grad_norm=57.292, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.070, optim0_lr0=7.927e-05, train_time=1.807
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2877332.0 ON gpub024 CANCELLED AT 2024-01-21T14:57:39 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.2.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.2.log
new file mode 100644
index 0000000000000000000000000000000000000000..afe900b3880166436bbb55ee6874198d86be7688
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.2.log
@@ -0,0 +1,2649 @@
+# Running on gpub011.delta.ncsa.illinois.edu
+# Started at Mon Feb 5 19:20:14 CST 2024
+# SLURMD_NODENAME=gpub011
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2944938
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1707355192
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2944938
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[011,042,044,098]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1707182392
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[011,042,044,098]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2351959
+# SLURM_TOPOLOGY_ADDR=ss00.ss09.gpub011
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_eb219d8a-88e5-437d-9b13-11b6fb1761dd 
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_eb219d8a-88e5-437d-9b13-11b6fb1761dd
+ape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_eb219d8a-88e5-437d-9b13-11b6fb1761dd
+ape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_eb219d8a-88e5-437d-9b13-11b6fb1761dd
+ape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_eb219d8a-88e5-437d-9b13-11b6fb1761dd
+[gpub011:0/16] 2024-02-05 19:25:40,541 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub011:0/16] 2024-02-05 19:25:40,561 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub011:0/16] 2024-02-05 19:25:40,728 (s2t:464) INFO: Vocabulary size: 50002
+[gpub011:0/16] 2024-02-05 19:25:53,456 (abs_task:1271) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub011:0/16] 2024-02-05 19:25:53,462 (abs_task:1272) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub011:0/16] 2024-02-05 19:25:53,462 (abs_task:1275) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub011:0/16] 2024-02-05 19:25:53,462 (abs_task:1276) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub011:0/16] 2024-02-05 19:25:53,464 (abs_task:1285) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub011:0/16] 2024-02-05 19:25:59,255 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 19:26:00,202 (abs_task:1663) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8165b951e0>)
+[gpub011:0/16] 2024-02-05 19:26:00,202 (abs_task:1664) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub011:0/16] 2024-02-05 19:26:00,256 (abs_task:1665) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 19:26:19,189 (trainer:168) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub011:2352043:2352043 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2352043:2352043 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2352043:2352043 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2352043:2352043 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub011:0/16] 2024-02-05 19:26:27,335 (trainer:302) INFO: 29/45epoch started
+[gpub011:0/16] 2024-02-05 19:26:27,376 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-05 19:26:45,423 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 19:26:48,936 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f803257bf40>)
+[gpub011:0/16] 2024-02-05 19:26:48,936 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-05 19:26:48,939 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub044:436186:436186 [1] NCCL INFO cudaDriverVersion 12020
+gpub044:436186:436186 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.144<0>
+gpub044:436186:436186 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub044:436186:436186 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub044:436186:436249 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub044:436186:436249 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub044:436186:436249 [1] NCCL INFO Using network AWS Libfabric
+gpub044:436186:436249 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub044:436186:436249 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub044:436186:436249 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub044:436186:436249 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub044:436186:436249 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub044:436186:436249 [1] NCCL INFO Connected all rings
+gpub044:436186:436249 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/AWS Libfabric/1
+gpub044:436186:436249 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub044:436186:436249 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub044:436186:436249 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub044:436186:436249 [1] NCCL INFO Connected all trees
+gpub044:436186:436249 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub044:436186:436249 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub044:436186:436249 [1] NCCL INFO comm 0x55d99450d260 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub044:436187:436187 [2] NCCL INFO cudaDriverVersion 12020
+gpub044:436187:436187 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.144<0>
+gpub044:436187:436187 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub044:436187:436187 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub044:436187:436251 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub044:436187:436251 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub044:436187:436251 [2] NCCL INFO Using network AWS Libfabric
+gpub044:436187:436251 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub044:436187:436251 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub044:436187:436251 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub044:436187:436251 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub044:436187:436251 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub044:436187:436251 [2] NCCL INFO Connected all rings
+gpub044:436187:436251 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub044:436187:436251 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub044:436187:436251 [2] NCCL INFO Connected all trees
+gpub044:436187:436251 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub044:436187:436251 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub044:436187:436251 [2] NCCL INFO comm 0x559867b7eef0 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub044:436185:436185 [0] NCCL INFO cudaDriverVersion 12020
+gpub044:436185:436185 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.144<0>
+gpub044:436185:436185 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub044:436185:436185 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub044:436185:436250 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub044:436185:436250 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub044:436185:436250 [0] NCCL INFO Using network AWS Libfabric
+gpub044:436185:436250 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub044:436185:436250 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub044:436185:436250 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub044:436185:436250 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub044:436185:436250 [0] NCCL INFO Connected all rings
+gpub044:436185:436250 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/AWS Libfabric/1
+gpub044:436185:436250 [0] NCCL INFO Connected all trees
+gpub044:436185:436250 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub044:436185:436250 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub044:436185:436250 [0] NCCL INFO comm 0x55837b1d3b60 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub044:436188:436188 [3] NCCL INFO cudaDriverVersion 12020
+gpub044:436188:436188 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.144<0>
+gpub044:436188:436188 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub044:436188:436188 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub044:436188:436252 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub044:436188:436252 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub044:436188:436252 [3] NCCL INFO Using network AWS Libfabric
+gpub044:436188:436252 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub044:436188:436252 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub044:436188:436252 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub044:436188:436252 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub044:436188:436252 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub044:436188:436252 [3] NCCL INFO Connected all rings
+gpub044:436188:436252 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub044:436188:436252 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub044:436188:436252 [3] NCCL INFO Connected all trees
+gpub044:436188:436252 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub044:436188:436252 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub044:436188:436252 [3] NCCL INFO comm 0x55afd4f9c990 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub011:2352046:2352046 [3] NCCL INFO cudaDriverVersion 12020
+gpub011:2352046:2352046 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2352046:2352046 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2352046:2352046 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2352046:2352106 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2352046:2352106 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2352046:2352106 [3] NCCL INFO Using network AWS Libfabric
+gpub011:2352046:2352106 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub011:2352046:2352106 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2352046:2352106 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub011:2352046:2352106 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub011:2352046:2352106 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub011:2352046:2352106 [3] NCCL INFO Connected all rings
+gpub011:2352046:2352106 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub011:2352046:2352106 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub011:2352046:2352106 [3] NCCL INFO Connected all trees
+gpub011:2352046:2352106 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2352046:2352106 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2352046:2352106 [3] NCCL INFO comm 0x56380d66ed60 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub011:2352045:2352045 [2] NCCL INFO cudaDriverVersion 12020
+gpub011:2352045:2352045 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2352045:2352045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2352045:2352045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2352045:2352105 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2352045:2352105 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2352045:2352105 [2] NCCL INFO Using network AWS Libfabric
+gpub011:2352045:2352105 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub011:2352045:2352105 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2352045:2352105 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub011:2352045:2352105 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub011:2352045:2352105 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub011:2352045:2352105 [2] NCCL INFO Connected all rings
+gpub011:2352045:2352105 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub011:2352045:2352105 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub011:2352045:2352105 [2] NCCL INFO Connected all trees
+gpub011:2352045:2352105 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2352045:2352105 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2352045:2352105 [2] NCCL INFO comm 0x5611d88aa8d0 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub011:2352044:2352044 [1] NCCL INFO cudaDriverVersion 12020
+gpub011:2352044:2352044 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.111<0>
+gpub011:2352044:2352044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub011:2352044:2352044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub011:2352044:2352104 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2352044:2352104 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2352044:2352104 [1] NCCL INFO Using network AWS Libfabric
+gpub011:2352044:2352104 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub011:2352044:2352104 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2352044:2352104 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub011:2352044:2352104 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub011:2352044:2352104 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub011:2352044:2352104 [1] NCCL INFO Connected all rings
+gpub011:2352044:2352104 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub011:2352044:2352104 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub011:2352044:2352104 [1] NCCL INFO Connected all trees
+gpub011:2352044:2352104 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2352044:2352104 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2352044:2352104 [1] NCCL INFO comm 0x55d410f2c6a0 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub011:2352043:2352107 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub011:2352043:2352107 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub011:2352043:2352107 [0] NCCL INFO Using network AWS Libfabric
+gpub011:2352043:2352107 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub011:2352043:2352107 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub011:2352043:2352107 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub011:2352043:2352107 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub011:2352043:2352107 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub011:2352043:2352107 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub011:2352043:2352107 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub011:2352043:2352107 [0] NCCL INFO Connected all rings
+gpub011:2352043:2352107 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub011:2352043:2352107 [0] NCCL INFO Connected all trees
+gpub011:2352043:2352107 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub011:2352043:2352107 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub011:2352043:2352107 [0] NCCL INFO comm 0x556276893b30 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub098:1865146:1865146 [2] NCCL INFO cudaDriverVersion 12020
+gpub098:1865146:1865146 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.198<0>
+gpub098:1865146:1865146 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub098:1865146:1865146 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub098:1865146:1865207 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub098:1865146:1865207 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub098:1865146:1865207 [2] NCCL INFO Using network AWS Libfabric
+gpub098:1865146:1865207 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub098:1865146:1865207 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub098:1865146:1865207 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub098:1865146:1865207 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub098:1865146:1865207 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub098:1865146:1865207 [2] NCCL INFO Connected all rings
+gpub098:1865146:1865207 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub098:1865146:1865207 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub098:1865146:1865207 [2] NCCL INFO Connected all trees
+gpub098:1865146:1865207 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub098:1865146:1865207 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub098:1865146:1865207 [2] NCCL INFO comm 0x5603c9eca3f0 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub098:1865145:1865145 [1] NCCL INFO cudaDriverVersion 12020
+gpub098:1865145:1865145 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.198<0>
+gpub098:1865145:1865145 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub098:1865145:1865145 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub098:1865145:1865206 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub098:1865145:1865206 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub098:1865145:1865206 [1] NCCL INFO Using network AWS Libfabric
+gpub098:1865145:1865206 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub098:1865145:1865206 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub098:1865145:1865206 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub098:1865145:1865206 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub098:1865145:1865206 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub098:1865145:1865206 [1] NCCL INFO Connected all rings
+gpub098:1865145:1865206 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub098:1865145:1865206 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub098:1865145:1865206 [1] NCCL INFO Connected all trees
+gpub098:1865145:1865206 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub098:1865145:1865206 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub098:1865145:1865206 [1] NCCL INFO comm 0x55ee8b529330 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub098:1865144:1865144 [0] NCCL INFO cudaDriverVersion 12020
+gpub098:1865144:1865144 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.198<0>
+gpub098:1865144:1865144 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub098:1865144:1865144 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub098:1865144:1865204 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub098:1865144:1865204 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub098:1865144:1865204 [0] NCCL INFO Using network AWS Libfabric
+gpub098:1865144:1865204 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub098:1865144:1865204 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub098:1865144:1865204 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub098:1865144:1865204 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub098:1865144:1865204 [0] NCCL INFO Connected all rings
+gpub098:1865144:1865204 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub098:1865144:1865204 [0] NCCL INFO Connected all trees
+gpub098:1865144:1865204 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub098:1865144:1865204 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub098:1865144:1865204 [0] NCCL INFO comm 0x558641031470 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub042:2862845:2862845 [3] NCCL INFO cudaDriverVersion 12020
+gpub042:2862845:2862845 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:2862845:2862845 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub042:2862845:2862845 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub042:2862845:2862902 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub042:2862845:2862902 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub042:2862845:2862902 [3] NCCL INFO Using network AWS Libfabric
+gpub042:2862845:2862902 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub042:2862845:2862902 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub042:2862845:2862902 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub042:2862845:2862902 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub042:2862845:2862902 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub042:2862845:2862902 [3] NCCL INFO Connected all rings
+gpub042:2862845:2862902 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub042:2862845:2862902 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub042:2862845:2862902 [3] NCCL INFO Connected all trees
+gpub042:2862845:2862902 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub042:2862845:2862902 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:2862845:2862902 [3] NCCL INFO comm 0x561137684d90 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub042:2862843:2862843 [1] NCCL INFO cudaDriverVersion 12020
+gpub042:2862843:2862843 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:2862843:2862843 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub042:2862843:2862843 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub042:2862843:2862900 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub042:2862843:2862900 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub042:2862843:2862900 [1] NCCL INFO Using network AWS Libfabric
+gpub042:2862843:2862900 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub042:2862843:2862900 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub042:2862843:2862900 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub042:2862843:2862900 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub042:2862843:2862900 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub042:2862843:2862900 [1] NCCL INFO Connected all rings
+gpub042:2862843:2862900 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub042:2862843:2862900 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/AWS Libfabric/1
+gpub042:2862843:2862900 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub042:2862843:2862900 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub042:2862843:2862900 [1] NCCL INFO Connected all trees
+gpub042:2862843:2862900 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub042:2862843:2862900 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:2862843:2862900 [1] NCCL INFO comm 0x560712643a50 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub042:2862842:2862842 [0] NCCL INFO cudaDriverVersion 12020
+gpub042:2862842:2862842 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:2862842:2862842 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub042:2862842:2862842 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub042:2862842:2862903 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub042:2862842:2862903 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub042:2862842:2862903 [0] NCCL INFO Using network AWS Libfabric
+gpub042:2862842:2862903 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub042:2862842:2862903 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub042:2862842:2862903 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub042:2862842:2862903 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub042:2862842:2862903 [0] NCCL INFO Connected all rings
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub042:2862842:2862903 [0] NCCL INFO Connected all trees
+gpub042:2862842:2862903 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub042:2862842:2862903 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:2862842:2862903 [0] NCCL INFO comm 0x55d6392f0d80 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub098:1865147:1865147 [3] NCCL INFO cudaDriverVersion 12020
+gpub098:1865147:1865147 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.198<0>
+gpub098:1865147:1865147 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub098:1865147:1865147 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub098:1865147:1865205 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub098:1865147:1865205 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub098:1865147:1865205 [3] NCCL INFO Using network AWS Libfabric
+gpub098:1865147:1865205 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub098:1865147:1865205 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub098:1865147:1865205 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub098:1865147:1865205 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub098:1865147:1865205 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub098:1865147:1865205 [3] NCCL INFO Connected all rings
+gpub098:1865147:1865205 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub098:1865147:1865205 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub098:1865147:1865205 [3] NCCL INFO Connected all trees
+gpub098:1865147:1865205 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub098:1865147:1865205 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub098:1865147:1865205 [3] NCCL INFO comm 0x560ec625b210 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub042:2862844:2862844 [2] NCCL INFO cudaDriverVersion 12020
+gpub042:2862844:2862844 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.142<0>
+gpub042:2862844:2862844 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub042:2862844:2862844 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub042:2862844:2862901 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub042:2862844:2862901 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub042:2862844:2862901 [2] NCCL INFO Using network AWS Libfabric
+gpub042:2862844:2862901 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub042:2862844:2862901 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub042:2862844:2862901 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub042:2862844:2862901 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub042:2862844:2862901 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub042:2862844:2862901 [2] NCCL INFO Connected all rings
+gpub042:2862844:2862901 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub042:2862844:2862901 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub042:2862844:2862901 [2] NCCL INFO Connected all trees
+gpub042:2862844:2862901 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub042:2862844:2862901 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub042:2862844:2862901 [2] NCCL INFO comm 0x55b48fc49950 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+[gpub011:0/16] 2024-02-05 19:31:47,732 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub011:0/16] 2024-02-05 19:33:46,028 (trainer:762) INFO: 29epoch:train:1-100batch: iter_time=1.159, forward_time=0.535, loss_ctc=45.751, loss_att=37.549, acc=0.779, loss=40.010, backward_time=0.330, grad_norm=39.679, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.890e-04, train_time=4.364
+[gpub011:0/16] 2024-02-05 19:36:13,837 (trainer:762) INFO: 29epoch:train:101-200batch: iter_time=9.318e-05, forward_time=0.435, loss_ctc=53.238, loss_att=50.932, acc=0.740, loss=51.624, backward_time=0.361, grad_norm=44.831, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.890e-04, train_time=1.500
+[gpub011:0/16] 2024-02-05 19:38:43,513 (trainer:762) INFO: 29epoch:train:201-300batch: iter_time=8.148e-05, forward_time=0.386, loss_ctc=51.649, loss_att=49.608, acc=0.751, loss=50.220, backward_time=0.324, grad_norm=44.247, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.889e-04, train_time=1.497
+[gpub011:0/16] 2024-02-05 19:40:57,066 (trainer:762) INFO: 29epoch:train:301-400batch: iter_time=8.546e-05, forward_time=0.315, loss_ctc=61.299, loss_att=52.789, acc=0.731, loss=55.342, backward_time=0.314, grad_norm=53.932, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.889e-04, train_time=1.335
+[gpub011:0/16] 2024-02-05 19:43:54,291 (trainer:762) INFO: 29epoch:train:401-500batch: iter_time=8.279e-05, forward_time=0.415, loss_ctc=45.254, loss_att=47.239, acc=0.734, loss=46.644, backward_time=0.339, grad_norm=40.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.889e-04, train_time=1.772
+[gpub011:0/16] 2024-02-05 19:46:32,037 (trainer:762) INFO: 29epoch:train:501-600batch: iter_time=8.024e-05, forward_time=0.323, loss_ctc=48.515, loss_att=43.992, acc=0.745, loss=45.349, backward_time=0.298, grad_norm=44.145, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.889e-04, train_time=1.578
+[gpub011:0/16] 2024-02-05 19:49:28,702 (trainer:762) INFO: 29epoch:train:601-700batch: iter_time=0.001, forward_time=0.453, loss_ctc=54.414, loss_att=48.819, acc=0.751, loss=50.498, backward_time=0.374, grad_norm=39.723, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.889e-04, train_time=1.766
+[gpub011:0/16] 2024-02-05 19:51:52,050 (trainer:762) INFO: 29epoch:train:701-800batch: iter_time=8.334e-05, forward_time=0.298, loss_ctc=53.375, loss_att=51.747, acc=0.733, loss=52.235, backward_time=0.298, grad_norm=39.904, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.888e-04, train_time=1.434
+[gpub011:0/16] 2024-02-05 19:53:42,743 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-05 19:55:04,605 (trainer:762) INFO: 29epoch:train:801-900batch: iter_time=3.502e-04, forward_time=0.408, loss_ctc=61.251, loss_att=53.019, acc=0.747, loss=55.488, backward_time=0.407, grad_norm=48.037, clip=100.000, loss_scale=3.986e+33, optim_step_time=0.107, optim0_lr0=1.888e-04, train_time=1.925
+[gpub011:0/16] 2024-02-05 19:57:13,705 (trainer:762) INFO: 29epoch:train:901-1000batch: iter_time=8.402e-05, forward_time=0.296, loss_ctc=46.817, loss_att=42.474, acc=0.749, loss=43.777, backward_time=0.298, grad_norm=35.655, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.888e-04, train_time=1.291
+[gpub011:0/16] 2024-02-05 20:00:35,310 (trainer:762) INFO: 29epoch:train:1001-1100batch: iter_time=8.275e-05, forward_time=0.450, loss_ctc=47.829, loss_att=46.221, acc=0.755, loss=46.703, backward_time=0.367, grad_norm=36.867, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.105, optim0_lr0=1.888e-04, train_time=2.015
+[gpub011:0/16] 2024-02-05 20:02:32,507 (trainer:762) INFO: 29epoch:train:1101-1200batch: iter_time=7.738e-05, forward_time=0.302, loss_ctc=52.013, loss_att=50.143, acc=0.741, loss=50.704, backward_time=0.301, grad_norm=53.205, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.887e-04, train_time=1.172
+[gpub011:0/16] 2024-02-05 20:04:39,060 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-05 20:04:57,952 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 20:05:01,473 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8162683e50>)
+[gpub011:0/16] 2024-02-05 20:05:01,473 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-05 20:05:01,478 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 20:12:36,766 (trainer:762) INFO: 29epoch:train:1201-1300batch: iter_time=4.113, forward_time=0.374, loss_ctc=42.747, loss_att=44.605, acc=0.751, loss=44.048, backward_time=0.335, grad_norm=38.387, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.887e-04, train_time=6.043
+[gpub011:0/16] 2024-02-05 20:14:55,316 (trainer:762) INFO: 29epoch:train:1301-1400batch: iter_time=8.422e-05, forward_time=0.401, loss_ctc=46.665, loss_att=43.039, acc=0.757, loss=44.127, backward_time=0.314, grad_norm=41.448, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.887e-04, train_time=1.385
+[gpub011:0/16] 2024-02-05 20:17:18,508 (trainer:762) INFO: 29epoch:train:1401-1500batch: iter_time=8.363e-05, forward_time=0.313, loss_ctc=57.086, loss_att=52.263, acc=0.748, loss=53.710, backward_time=0.307, grad_norm=45.118, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.887e-04, train_time=1.432
+[gpub011:0/16] 2024-02-05 20:20:05,391 (trainer:762) INFO: 29epoch:train:1501-1600batch: iter_time=8.166e-05, forward_time=0.442, loss_ctc=48.394, loss_att=44.548, acc=0.754, loss=45.702, backward_time=0.341, grad_norm=42.812, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.109, optim0_lr0=1.887e-04, train_time=1.669
+[gpub011:0/16] 2024-02-05 20:22:08,647 (trainer:762) INFO: 29epoch:train:1601-1700batch: iter_time=7.728e-05, forward_time=0.302, loss_ctc=57.273, loss_att=53.901, acc=0.725, loss=54.913, backward_time=0.316, grad_norm=48.533, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.886e-04, train_time=1.232
+[gpub011:0/16] 2024-02-05 20:24:23,813 (trainer:762) INFO: 29epoch:train:1701-1800batch: iter_time=7.791e-05, forward_time=0.346, loss_ctc=39.617, loss_att=33.785, acc=0.768, loss=35.534, backward_time=0.331, grad_norm=40.884, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.886e-04, train_time=1.352
+[gpub011:0/16] 2024-02-05 20:26:44,269 (trainer:762) INFO: 29epoch:train:1801-1900batch: iter_time=4.164e-04, forward_time=0.395, loss_ctc=55.014, loss_att=49.717, acc=0.750, loss=51.306, backward_time=0.324, grad_norm=40.466, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.886e-04, train_time=1.403
+[gpub011:0/16] 2024-02-05 20:29:24,608 (trainer:762) INFO: 29epoch:train:1901-2000batch: iter_time=7.862e-05, forward_time=0.392, loss_ctc=48.087, loss_att=48.554, acc=0.730, loss=48.414, backward_time=0.348, grad_norm=37.874, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.886e-04, train_time=1.604
+[gpub011:0/16] 2024-02-05 20:31:23,820 (trainer:762) INFO: 29epoch:train:2001-2100batch: iter_time=8.088e-05, forward_time=0.296, loss_ctc=57.560, loss_att=47.611, acc=0.754, loss=50.596, backward_time=0.305, grad_norm=42.271, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.885e-04, train_time=1.192
+[gpub011:0/16] 2024-02-05 20:33:26,543 (trainer:762) INFO: 29epoch:train:2101-2200batch: iter_time=8.385e-05, forward_time=0.308, loss_ctc=55.734, loss_att=53.950, acc=0.736, loss=54.485, backward_time=0.305, grad_norm=45.241, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.885e-04, train_time=1.226
+[gpub011:0/16] 2024-02-05 20:35:59,940 (trainer:762) INFO: 29epoch:train:2201-2300batch: iter_time=4.490e-04, forward_time=0.411, loss_ctc=45.072, loss_att=43.733, acc=0.754, loss=44.135, backward_time=0.378, grad_norm=36.919, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=1.885e-04, train_time=1.535
+[gpub011:0/16] 2024-02-05 20:38:38,757 (trainer:762) INFO: 29epoch:train:2301-2400batch: iter_time=8.054e-05, forward_time=0.355, loss_ctc=52.275, loss_att=49.663, acc=0.749, loss=50.447, backward_time=0.308, grad_norm=47.989, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.885e-04, train_time=1.587
+[gpub011:0/16] 2024-02-05 20:40:53,532 (trainer:762) INFO: 29epoch:train:2401-2500batch: iter_time=7.819e-05, forward_time=0.301, loss_ctc=42.864, loss_att=46.370, acc=0.728, loss=45.318, backward_time=0.315, grad_norm=36.562, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.885e-04, train_time=1.347
+[gpub011:0/16] 2024-02-05 20:41:13,649 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-05 20:41:32,234 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 20:41:35,639 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8162438ac0>)
+[gpub011:0/16] 2024-02-05 20:41:35,639 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-05 20:41:35,646 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 20:49:45,657 (trainer:762) INFO: 29epoch:train:2501-2600batch: iter_time=3.594, forward_time=0.410, loss_ctc=44.299, loss_att=37.456, acc=0.786, loss=39.509, backward_time=0.315, grad_norm=37.796, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.884e-04, train_time=5.323
+[gpub011:0/16] 2024-02-05 20:52:01,858 (trainer:762) INFO: 29epoch:train:2601-2700batch: iter_time=3.865e-04, forward_time=0.316, loss_ctc=50.795, loss_att=52.446, acc=0.744, loss=51.950, backward_time=0.309, grad_norm=47.588, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.884e-04, train_time=1.362
+[gpub011:0/16] 2024-02-05 20:54:17,712 (trainer:762) INFO: 29epoch:train:2701-2800batch: iter_time=7.781e-05, forward_time=0.310, loss_ctc=49.339, loss_att=47.872, acc=0.765, loss=48.312, backward_time=0.317, grad_norm=42.217, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.884e-04, train_time=1.358
+[gpub011:0/16] 2024-02-05 20:56:50,039 (trainer:762) INFO: 29epoch:train:2801-2900batch: iter_time=8.228e-05, forward_time=0.390, loss_ctc=56.850, loss_att=50.902, acc=0.741, loss=52.686, backward_time=0.315, grad_norm=51.021, clip=100.000, loss_scale=3.790e+33, optim_step_time=0.111, optim0_lr0=1.884e-04, train_time=1.524
+[gpub011:0/16] 2024-02-05 20:59:16,469 (trainer:762) INFO: 29epoch:train:2901-3000batch: iter_time=8.299e-05, forward_time=0.341, loss_ctc=43.959, loss_att=45.750, acc=0.755, loss=45.213, backward_time=0.298, grad_norm=38.645, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.883e-04, train_time=1.464
+[gpub011:0/16] 2024-02-05 21:01:38,693 (trainer:762) INFO: 29epoch:train:3001-3100batch: iter_time=8.135e-05, forward_time=0.302, loss_ctc=46.555, loss_att=42.559, acc=0.761, loss=43.758, backward_time=0.316, grad_norm=40.314, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.883e-04, train_time=1.422
+[gpub011:0/16] 2024-02-05 21:03:49,714 (trainer:762) INFO: 29epoch:train:3101-3200batch: iter_time=8.230e-05, forward_time=0.373, loss_ctc=52.518, loss_att=48.425, acc=0.759, loss=49.653, backward_time=0.323, grad_norm=39.125, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.883e-04, train_time=1.309
+[gpub011:0/16] 2024-02-05 21:06:17,563 (trainer:762) INFO: 29epoch:train:3201-3300batch: iter_time=8.138e-05, forward_time=0.303, loss_ctc=53.233, loss_att=53.598, acc=0.737, loss=53.489, backward_time=0.302, grad_norm=41.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.883e-04, train_time=1.479
+[gpub011:0/16] 2024-02-05 21:08:50,659 (trainer:762) INFO: 29epoch:train:3301-3400batch: iter_time=8.354e-05, forward_time=0.309, loss_ctc=57.701, loss_att=51.359, acc=0.760, loss=53.261, backward_time=0.313, grad_norm=46.479, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.883e-04, train_time=1.531
+[gpub011:0/16] 2024-02-05 21:10:46,374 (trainer:762) INFO: 29epoch:train:3401-3500batch: iter_time=7.752e-05, forward_time=0.306, loss_ctc=46.233, loss_att=41.963, acc=0.757, loss=43.244, backward_time=0.310, grad_norm=37.833, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.882e-04, train_time=1.157
+[gpub011:0/16] 2024-02-05 21:13:40,479 (trainer:762) INFO: 29epoch:train:3501-3600batch: iter_time=1.954e-04, forward_time=0.397, loss_ctc=47.493, loss_att=45.000, acc=0.767, loss=45.748, backward_time=0.311, grad_norm=36.052, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.882e-04, train_time=1.740
+[gpub011:0/16] 2024-02-05 21:16:06,146 (trainer:762) INFO: 29epoch:train:3601-3700batch: iter_time=8.678e-05, forward_time=0.313, loss_ctc=49.532, loss_att=49.807, acc=0.749, loss=49.724, backward_time=0.302, grad_norm=46.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.882e-04, train_time=1.458
+[gpub011:0/16] 2024-02-05 21:17:32,317 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-05 21:17:51,006 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 21:17:54,490 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7af5251e40>)
+[gpub011:0/16] 2024-02-05 21:17:54,491 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-05 21:17:54,496 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 21:24:29,205 (trainer:762) INFO: 29epoch:train:3701-3800batch: iter_time=3.694, forward_time=0.312, loss_ctc=42.326, loss_att=43.918, acc=0.763, loss=43.440, backward_time=0.305, grad_norm=35.945, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.882e-04, train_time=5.030
+[gpub011:0/16] 2024-02-05 21:26:42,865 (trainer:762) INFO: 29epoch:train:3801-3900batch: iter_time=7.249e-05, forward_time=0.378, loss_ctc=44.978, loss_att=45.082, acc=0.756, loss=45.051, backward_time=0.322, grad_norm=40.714, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.881e-04, train_time=1.336
+[gpub011:0/16] 2024-02-05 21:29:13,899 (trainer:762) INFO: 29epoch:train:3901-4000batch: iter_time=7.898e-05, forward_time=0.315, loss_ctc=54.695, loss_att=51.889, acc=0.761, loss=52.731, backward_time=0.319, grad_norm=50.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.881e-04, train_time=1.510
+[gpub011:0/16] 2024-02-05 21:31:20,611 (trainer:762) INFO: 29epoch:train:4001-4100batch: iter_time=8.106e-05, forward_time=0.323, loss_ctc=47.141, loss_att=43.964, acc=0.761, loss=44.917, backward_time=0.307, grad_norm=41.889, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.881e-04, train_time=1.267
+[gpub011:0/16] 2024-02-05 21:33:22,276 (trainer:762) INFO: 29epoch:train:4101-4200batch: iter_time=8.360e-05, forward_time=0.307, loss_ctc=56.067, loss_att=53.654, acc=0.738, loss=54.378, backward_time=0.307, grad_norm=46.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.881e-04, train_time=1.217
+[gpub011:0/16] 2024-02-05 21:36:01,446 (trainer:762) INFO: 29epoch:train:4201-4300batch: iter_time=8.149e-05, forward_time=0.348, loss_ctc=38.681, loss_att=34.063, acc=0.775, loss=35.448, backward_time=0.312, grad_norm=40.355, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.881e-04, train_time=1.591
+[gpub011:0/16] 2024-02-05 21:38:08,288 (trainer:762) INFO: 29epoch:train:4301-4400batch: iter_time=7.794e-05, forward_time=0.320, loss_ctc=54.505, loss_att=49.096, acc=0.763, loss=50.718, backward_time=0.312, grad_norm=39.779, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.880e-04, train_time=1.267
+[gpub011:0/16] 2024-02-05 21:40:32,875 (trainer:762) INFO: 29epoch:train:4401-4500batch: iter_time=8.090e-05, forward_time=0.306, loss_ctc=47.115, loss_att=50.424, acc=0.735, loss=49.431, backward_time=0.309, grad_norm=37.841, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.880e-04, train_time=1.446
+[gpub011:0/16] 2024-02-05 21:42:35,366 (trainer:762) INFO: 29epoch:train:4501-4600batch: iter_time=7.978e-05, forward_time=0.309, loss_ctc=55.361, loss_att=47.285, acc=0.762, loss=49.708, backward_time=0.311, grad_norm=41.280, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.880e-04, train_time=1.225
+[gpub011:0/16] 2024-02-05 21:44:51,511 (trainer:762) INFO: 29epoch:train:4601-4700batch: iter_time=7.930e-05, forward_time=0.397, loss_ctc=54.412, loss_att=53.393, acc=0.747, loss=53.698, backward_time=0.321, grad_norm=46.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.880e-04, train_time=1.361
+[gpub011:0/16] 2024-02-05 21:47:24,194 (trainer:762) INFO: 29epoch:train:4701-4800batch: iter_time=8.062e-05, forward_time=0.298, loss_ctc=44.240, loss_att=44.010, acc=0.758, loss=44.079, backward_time=0.301, grad_norm=36.087, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.879e-04, train_time=1.526
+[gpub011:0/16] 2024-02-05 21:49:27,081 (trainer:762) INFO: 29epoch:train:4801-4900batch: iter_time=8.036e-05, forward_time=0.321, loss_ctc=51.272, loss_att=48.691, acc=0.763, loss=49.465, backward_time=0.307, grad_norm=43.480, clip=100.000, loss_scale=7.581e+33, optim_step_time=0.098, optim0_lr0=1.879e-04, train_time=1.229
+[gpub011:0/16] 2024-02-05 21:51:33,484 (trainer:762) INFO: 29epoch:train:4901-5000batch: iter_time=8.072e-05, forward_time=0.330, loss_ctc=42.188, loss_att=45.504, acc=0.746, loss=44.509, backward_time=0.307, grad_norm=35.619, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.879e-04, train_time=1.264
+[gpub011:0/16] 2024-02-05 21:51:53,513 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-05 21:52:12,309 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 21:52:15,760 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7fd88e35e0>)
+[gpub011:0/16] 2024-02-05 21:52:15,760 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-05 21:52:15,803 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 22:00:33,559 (trainer:762) INFO: 29epoch:train:5001-5100batch: iter_time=3.791, forward_time=0.381, loss_ctc=43.081, loss_att=36.918, acc=0.788, loss=38.767, backward_time=0.306, grad_norm=37.478, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.879e-04, train_time=5.400
+[gpub011:0/16] 2024-02-05 22:02:37,359 (trainer:762) INFO: 29epoch:train:5101-5200batch: iter_time=8.259e-05, forward_time=0.325, loss_ctc=50.757, loss_att=50.522, acc=0.748, loss=50.592, backward_time=0.303, grad_norm=44.214, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.879e-04, train_time=1.238
+[gpub011:0/16] 2024-02-05 22:04:57,660 (trainer:762) INFO: 29epoch:train:5201-5300batch: iter_time=8.389e-05, forward_time=0.316, loss_ctc=48.861, loss_att=47.035, acc=0.767, loss=47.583, backward_time=0.310, grad_norm=41.259, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.878e-04, train_time=1.403
+[gpub011:0/16] 2024-02-05 22:07:40,583 (trainer:762) INFO: 29epoch:train:5301-5400batch: iter_time=8.887e-05, forward_time=0.389, loss_ctc=56.147, loss_att=50.636, acc=0.744, loss=52.289, backward_time=0.318, grad_norm=52.913, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=1.878e-04, train_time=1.629
+[gpub011:0/16] 2024-02-05 22:09:42,878 (trainer:762) INFO: 29epoch:train:5401-5500batch: iter_time=7.953e-05, forward_time=0.316, loss_ctc=43.839, loss_att=45.067, acc=0.758, loss=44.698, backward_time=0.307, grad_norm=37.677, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.878e-04, train_time=1.222
+[gpub011:0/16] 2024-02-05 22:11:40,118 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-05 22:11:56,172 (trainer:762) INFO: 29epoch:train:5501-5600batch: iter_time=8.661e-05, forward_time=0.314, loss_ctc=45.882, loss_att=42.393, acc=0.763, loss=43.440, backward_time=0.309, grad_norm=40.214, clip=100.000, loss_scale=9.598e+33, optim_step_time=0.097, optim0_lr0=1.878e-04, train_time=1.333
+[gpub011:0/16] 2024-02-05 22:14:43,908 (trainer:762) INFO: 29epoch:train:5601-5700batch: iter_time=8.142e-05, forward_time=0.422, loss_ctc=52.150, loss_att=47.617, acc=0.761, loss=48.977, backward_time=0.337, grad_norm=41.111, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.877e-04, train_time=1.677
+[gpub011:0/16] 2024-02-05 22:16:42,092 (trainer:762) INFO: 29epoch:train:5701-5800batch: iter_time=8.127e-05, forward_time=0.322, loss_ctc=52.581, loss_att=53.197, acc=0.737, loss=53.012, backward_time=0.308, grad_norm=38.104, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.877e-04, train_time=1.182
+[gpub011:0/16] 2024-02-05 22:18:58,057 (trainer:762) INFO: 29epoch:train:5801-5900batch: iter_time=8.265e-05, forward_time=0.323, loss_ctc=55.838, loss_att=50.927, acc=0.761, loss=52.400, backward_time=0.311, grad_norm=43.877, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.877e-04, train_time=1.359
+[gpub011:0/16] 2024-02-05 22:21:23,985 (trainer:762) INFO: 29epoch:train:5901-6000batch: iter_time=8.585e-05, forward_time=0.383, loss_ctc=46.025, loss_att=41.441, acc=0.759, loss=42.816, backward_time=0.326, grad_norm=37.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.877e-04, train_time=1.460
+[gpub011:0/16] 2024-02-05 22:23:58,099 (trainer:762) INFO: 29epoch:train:6001-6100batch: iter_time=8.938e-05, forward_time=0.324, loss_ctc=46.535, loss_att=44.275, acc=0.771, loss=44.953, backward_time=0.305, grad_norm=35.440, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.877e-04, train_time=1.541
+[gpub011:0/16] 2024-02-05 22:26:17,218 (trainer:762) INFO: 29epoch:train:6101-6200batch: iter_time=9.558e-05, forward_time=0.300, loss_ctc=49.477, loss_att=49.699, acc=0.750, loss=49.633, backward_time=0.327, grad_norm=45.999, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.876e-04, train_time=1.391
+[gpub011:0/16] 2024-02-05 22:27:33,748 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-05 22:27:52,483 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 22:27:55,954 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71fcacb670>)
+[gpub011:0/16] 2024-02-05 22:27:55,954 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-05 22:27:55,959 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 22:35:14,067 (trainer:762) INFO: 29epoch:train:6201-6300batch: iter_time=3.843, forward_time=0.365, loss_ctc=42.054, loss_att=43.445, acc=0.765, loss=43.028, backward_time=0.309, grad_norm=36.742, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.876e-04, train_time=5.369
+[gpub011:0/16] 2024-02-05 22:37:11,090 (trainer:762) INFO: 29epoch:train:6301-6400batch: iter_time=6.171e-04, forward_time=0.349, loss_ctc=44.908, loss_att=44.767, acc=0.760, loss=44.810, backward_time=0.308, grad_norm=42.784, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.876e-04, train_time=1.170
+[gpub011:0/16] 2024-02-05 22:39:38,848 (trainer:762) INFO: 29epoch:train:6401-6500batch: iter_time=7.985e-05, forward_time=0.308, loss_ctc=53.465, loss_att=51.253, acc=0.763, loss=51.916, backward_time=0.310, grad_norm=44.629, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.876e-04, train_time=1.477
+[gpub011:0/16] 2024-02-05 22:42:05,166 (trainer:762) INFO: 29epoch:train:6501-6600batch: iter_time=7.910e-05, forward_time=0.382, loss_ctc=47.495, loss_att=44.082, acc=0.762, loss=45.106, backward_time=0.310, grad_norm=41.913, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.120, optim0_lr0=1.875e-04, train_time=1.463
+[gpub011:0/16] 2024-02-05 22:44:11,666 (trainer:762) INFO: 29epoch:train:6601-6700batch: iter_time=7.900e-05, forward_time=0.306, loss_ctc=55.346, loss_att=52.601, acc=0.740, loss=53.424, backward_time=0.308, grad_norm=48.015, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.875e-04, train_time=1.265
+[gpub011:0/16] 2024-02-05 22:46:56,780 (trainer:762) INFO: 29epoch:train:6701-6800batch: iter_time=7.975e-05, forward_time=0.318, loss_ctc=37.901, loss_att=33.819, acc=0.778, loss=35.044, backward_time=0.309, grad_norm=39.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.875e-04, train_time=1.651
+[gpub011:0/16] 2024-02-05 22:48:59,910 (trainer:762) INFO: 29epoch:train:6801-6900batch: iter_time=7.903e-05, forward_time=0.293, loss_ctc=54.344, loss_att=48.696, acc=0.763, loss=50.390, backward_time=0.300, grad_norm=41.396, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.875e-04, train_time=1.231
+[gpub011:0/16] 2024-02-05 22:51:15,562 (trainer:762) INFO: 29epoch:train:6901-7000batch: iter_time=2.974e-04, forward_time=0.379, loss_ctc=47.196, loss_att=50.571, acc=0.737, loss=49.558, backward_time=0.316, grad_norm=36.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.875e-04, train_time=1.356
+[gpub011:0/16] 2024-02-05 22:53:56,834 (trainer:762) INFO: 29epoch:train:7001-7100batch: iter_time=7.964e-05, forward_time=0.321, loss_ctc=55.191, loss_att=47.023, acc=0.764, loss=49.473, backward_time=0.320, grad_norm=40.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.874e-04, train_time=1.613
+[gpub011:0/16] 2024-02-05 22:55:59,702 (trainer:762) INFO: 29epoch:train:7101-7200batch: iter_time=7.890e-05, forward_time=0.307, loss_ctc=53.785, loss_att=52.542, acc=0.750, loss=52.915, backward_time=0.310, grad_norm=41.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.874e-04, train_time=1.228
+[gpub011:0/16] 2024-02-05 22:58:32,712 (trainer:762) INFO: 29epoch:train:7201-7300batch: iter_time=7.733e-05, forward_time=0.369, loss_ctc=44.287, loss_att=44.065, acc=0.759, loss=44.132, backward_time=0.306, grad_norm=36.139, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.874e-04, train_time=1.530
+[gpub011:0/16] 2024-02-05 23:00:47,191 (trainer:762) INFO: 29epoch:train:7301-7400batch: iter_time=4.982e-04, forward_time=0.313, loss_ctc=51.196, loss_att=48.816, acc=0.764, loss=49.530, backward_time=0.302, grad_norm=44.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.874e-04, train_time=1.345
+[gpub011:0/16] 2024-02-05 23:02:47,274 (trainer:762) INFO: 29epoch:train:7401-7500batch: iter_time=7.825e-05, forward_time=0.317, loss_ctc=41.967, loss_att=45.259, acc=0.747, loss=44.271, backward_time=0.310, grad_norm=34.889, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.873e-04, train_time=1.201
+[gpub011:0/16] 2024-02-05 23:03:07,303 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-05 23:03:26,467 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 23:03:30,178 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71fcaab7c0>)
+[gpub011:0/16] 2024-02-05 23:03:30,178 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-05 23:03:30,182 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 23:11:10,270 (trainer:762) INFO: 29epoch:train:7501-7600batch: iter_time=3.702, forward_time=0.355, loss_ctc=43.004, loss_att=37.907, acc=0.786, loss=39.436, backward_time=0.314, grad_norm=37.104, clip=100.000, loss_scale=5.971e+33, optim_step_time=0.099, optim0_lr0=1.873e-04, train_time=5.030
+[gpub011:0/16] 2024-02-05 23:13:27,523 (trainer:762) INFO: 29epoch:train:7601-7700batch: iter_time=8.325e-05, forward_time=0.317, loss_ctc=49.778, loss_att=50.457, acc=0.743, loss=50.253, backward_time=0.317, grad_norm=43.960, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.873e-04, train_time=1.372
+[gpub011:0/16] 2024-02-05 23:15:58,902 (trainer:762) INFO: 29epoch:train:7701-7800batch: iter_time=7.850e-05, forward_time=0.389, loss_ctc=47.109, loss_att=48.355, acc=0.756, loss=47.981, backward_time=0.324, grad_norm=42.352, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.873e-04, train_time=1.513
+[gpub011:0/16] 2024-02-05 23:18:13,826 (trainer:762) INFO: 29epoch:train:7801-7900batch: iter_time=3.912e-04, forward_time=0.346, loss_ctc=55.129, loss_att=49.921, acc=0.744, loss=51.483, backward_time=0.307, grad_norm=47.168, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.873e-04, train_time=1.349
+[gpub011:0/16] 2024-02-05 23:20:54,570 (trainer:762) INFO: 29epoch:train:7901-8000batch: iter_time=7.849e-05, forward_time=0.314, loss_ctc=43.506, loss_att=46.221, acc=0.742, loss=45.407, backward_time=0.304, grad_norm=40.563, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.872e-04, train_time=1.607
+[gpub011:0/16] 2024-02-05 23:23:16,745 (trainer:762) INFO: 29epoch:train:8001-8100batch: iter_time=1.157e-04, forward_time=0.376, loss_ctc=45.575, loss_att=43.107, acc=0.751, loss=43.848, backward_time=0.326, grad_norm=43.303, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=1.872e-04, train_time=1.422
+[gpub011:0/16] 2024-02-05 23:25:21,767 (trainer:762) INFO: 29epoch:train:8101-8200batch: iter_time=8.110e-05, forward_time=0.320, loss_ctc=52.001, loss_att=47.944, acc=0.754, loss=49.161, backward_time=0.314, grad_norm=39.658, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.872e-04, train_time=1.250
+[gpub011:0/16] 2024-02-05 23:27:17,043 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-05 23:28:11,099 (trainer:762) INFO: 29epoch:train:8201-8300batch: iter_time=8.805e-05, forward_time=0.361, loss_ctc=51.963, loss_att=51.134, acc=0.739, loss=51.383, backward_time=0.312, grad_norm=38.608, clip=100.000, loss_scale=8.339e+33, optim_step_time=0.112, optim0_lr0=1.872e-04, train_time=1.693
+[gpub011:0/16] 2024-02-05 23:30:12,866 (trainer:762) INFO: 29epoch:train:8301-8400batch: iter_time=8.946e-04, forward_time=0.337, loss_ctc=55.970, loss_att=51.663, acc=0.752, loss=52.955, backward_time=0.314, grad_norm=47.188, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.872e-04, train_time=1.218
+[gpub011:0/16] 2024-02-05 23:32:38,195 (trainer:762) INFO: 29epoch:train:8401-8500batch: iter_time=7.758e-05, forward_time=0.354, loss_ctc=45.842, loss_att=42.193, acc=0.752, loss=43.288, backward_time=0.308, grad_norm=37.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.871e-04, train_time=1.453
+[gpub011:0/16] 2024-02-05 23:35:14,606 (trainer:762) INFO: 29epoch:train:8501-8600batch: iter_time=7.953e-05, forward_time=0.392, loss_ctc=46.118, loss_att=45.249, acc=0.759, loss=45.510, backward_time=0.319, grad_norm=37.361, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.871e-04, train_time=1.564
+[gpub011:0/16] 2024-02-05 23:37:39,313 (trainer:762) INFO: 29epoch:train:8601-8700batch: iter_time=8.562e-05, forward_time=0.334, loss_ctc=48.053, loss_att=49.141, acc=0.742, loss=48.815, backward_time=0.326, grad_norm=47.951, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.871e-04, train_time=1.448
+[gpub011:0/16] 2024-02-05 23:39:03,278 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-05 23:39:22,165 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-05 23:39:25,698 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b7aec7ee0>)
+[gpub011:0/16] 2024-02-05 23:39:25,698 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-05 23:39:25,749 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-05 23:46:50,827 (trainer:762) INFO: 29epoch:train:8701-8800batch: iter_time=4.071, forward_time=0.392, loss_ctc=41.394, loss_att=43.817, acc=0.760, loss=43.090, backward_time=0.307, grad_norm=34.378, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.871e-04, train_time=5.515
+[gpub011:0/16] 2024-02-05 23:48:54,503 (trainer:762) INFO: 29epoch:train:8801-8900batch: iter_time=7.855e-05, forward_time=0.293, loss_ctc=44.527, loss_att=46.084, acc=0.757, loss=45.617, backward_time=0.308, grad_norm=41.463, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.870e-04, train_time=1.236
+[gpub011:0/16] 2024-02-05 23:51:38,880 (trainer:762) INFO: 29epoch:train:8901-9000batch: iter_time=8.104e-05, forward_time=0.317, loss_ctc=53.163, loss_att=51.876, acc=0.763, loss=52.262, backward_time=0.316, grad_norm=44.824, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.870e-04, train_time=1.644
+[gpub011:0/16] 2024-02-05 23:53:49,580 (trainer:762) INFO: 29epoch:train:9001-9100batch: iter_time=8.036e-05, forward_time=0.393, loss_ctc=46.171, loss_att=43.211, acc=0.764, loss=44.099, backward_time=0.309, grad_norm=41.211, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.870e-04, train_time=1.307
+[gpub011:0/16] 2024-02-05 23:56:18,902 (trainer:762) INFO: 29epoch:train:9101-9200batch: iter_time=8.215e-05, forward_time=0.326, loss_ctc=54.588, loss_att=52.269, acc=0.743, loss=52.964, backward_time=0.302, grad_norm=45.987, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.870e-04, train_time=1.492
+[gpub011:0/16] 2024-02-05 23:58:38,685 (trainer:762) INFO: 29epoch:train:9201-9300batch: iter_time=7.927e-05, forward_time=0.339, loss_ctc=37.674, loss_att=33.793, acc=0.780, loss=34.958, backward_time=0.307, grad_norm=36.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.870e-04, train_time=1.398
+[gpub011:0/16] 2024-02-06 00:01:04,541 (trainer:762) INFO: 29epoch:train:9301-9400batch: iter_time=8.076e-05, forward_time=0.353, loss_ctc=53.533, loss_att=47.893, acc=0.766, loss=49.585, backward_time=0.330, grad_norm=41.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.869e-04, train_time=1.458
+[gpub011:0/16] 2024-02-06 00:03:20,441 (trainer:762) INFO: 29epoch:train:9401-9500batch: iter_time=4.183e-04, forward_time=0.326, loss_ctc=46.323, loss_att=50.111, acc=0.737, loss=48.974, backward_time=0.300, grad_norm=43.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.869e-04, train_time=1.358
+[gpub011:0/16] 2024-02-06 00:06:04,457 (trainer:762) INFO: 29epoch:train:9501-9600batch: iter_time=8.122e-05, forward_time=0.356, loss_ctc=54.178, loss_att=46.818, acc=0.764, loss=49.026, backward_time=0.349, grad_norm=40.403, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.869e-04, train_time=1.641
+[gpub011:0/16] 2024-02-06 00:08:08,169 (trainer:762) INFO: 29epoch:train:9601-9700batch: iter_time=8.295e-05, forward_time=0.310, loss_ctc=54.230, loss_att=51.610, acc=0.754, loss=52.396, backward_time=0.312, grad_norm=45.093, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.869e-04, train_time=1.237
+[gpub011:0/16] 2024-02-06 00:10:36,949 (trainer:762) INFO: 29epoch:train:9701-9800batch: iter_time=8.025e-05, forward_time=0.309, loss_ctc=43.982, loss_att=43.824, acc=0.761, loss=43.871, backward_time=0.318, grad_norm=36.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.868e-04, train_time=1.487
+[gpub011:0/16] 2024-02-06 00:12:59,171 (trainer:762) INFO: 29epoch:train:9801-9900batch: iter_time=8.366e-05, forward_time=0.352, loss_ctc=50.916, loss_att=48.540, acc=0.764, loss=49.253, backward_time=0.323, grad_norm=46.552, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.868e-04, train_time=1.422
+[gpub011:0/16] 2024-02-06 00:13:58,719 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 00:15:13,753 (trainer:762) INFO: 29epoch:train:9901-10000batch: iter_time=7.952e-05, forward_time=0.311, loss_ctc=41.817, loss_att=46.371, acc=0.743, loss=45.005, backward_time=0.304, grad_norm=36.762, clip=100.000, loss_scale=3.566e+33, optim_step_time=0.094, optim0_lr0=1.868e-04, train_time=1.346
+[gpub011:0/16] 2024-02-06 00:15:33,841 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-06 00:15:53,021 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 00:15:56,489 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f72319ef7f0>)
+[gpub011:0/16] 2024-02-06 00:15:56,489 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-06 00:15:56,614 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 00:23:41,947 (trainer:762) INFO: 29epoch:train:10001-10100batch: iter_time=3.824, forward_time=0.392, loss_ctc=42.940, loss_att=37.264, acc=0.787, loss=38.967, backward_time=0.318, grad_norm=38.044, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.868e-04, train_time=5.082
+[gpub011:0/16] 2024-02-06 00:26:04,158 (trainer:762) INFO: 29epoch:train:10101-10200batch: iter_time=8.671e-05, forward_time=0.291, loss_ctc=49.539, loss_att=49.680, acc=0.746, loss=49.638, backward_time=0.295, grad_norm=43.794, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.868e-04, train_time=1.422
+[gpub011:0/16] 2024-02-06 00:28:44,124 (trainer:762) INFO: 29epoch:train:10201-10300batch: iter_time=8.485e-05, forward_time=0.408, loss_ctc=46.790, loss_att=48.178, acc=0.757, loss=47.762, backward_time=0.328, grad_norm=41.689, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=1.867e-04, train_time=1.599
+[gpub011:0/16] 2024-02-06 00:30:47,026 (trainer:762) INFO: 29epoch:train:10301-10400batch: iter_time=7.751e-05, forward_time=0.294, loss_ctc=54.799, loss_att=49.031, acc=0.745, loss=50.761, backward_time=0.298, grad_norm=46.957, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.867e-04, train_time=1.229
+[gpub011:0/16] 2024-02-06 00:33:24,827 (trainer:762) INFO: 29epoch:train:10401-10500batch: iter_time=7.707e-05, forward_time=0.288, loss_ctc=43.163, loss_att=45.332, acc=0.744, loss=44.682, backward_time=0.292, grad_norm=38.590, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.867e-04, train_time=1.577
+[gpub011:0/16] 2024-02-06 00:35:39,918 (trainer:762) INFO: 29epoch:train:10501-10600batch: iter_time=1.286e-04, forward_time=0.379, loss_ctc=45.211, loss_att=42.591, acc=0.754, loss=43.377, backward_time=0.341, grad_norm=41.561, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.867e-04, train_time=1.351
+[gpub011:0/16] 2024-02-06 00:38:16,578 (trainer:762) INFO: 29epoch:train:10601-10700batch: iter_time=7.619e-05, forward_time=0.325, loss_ctc=52.154, loss_att=47.566, acc=0.754, loss=48.942, backward_time=0.313, grad_norm=39.208, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.867e-04, train_time=1.566
+[gpub011:0/16] 2024-02-06 00:40:55,135 (trainer:762) INFO: 29epoch:train:10701-10800batch: iter_time=7.756e-05, forward_time=0.291, loss_ctc=52.056, loss_att=50.634, acc=0.742, loss=51.061, backward_time=0.295, grad_norm=38.100, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.866e-04, train_time=1.586
+[gpub011:0/16] 2024-02-06 00:43:13,336 (trainer:762) INFO: 29epoch:train:10801-10900batch: iter_time=1.821e-04, forward_time=0.363, loss_ctc=55.232, loss_att=51.038, acc=0.754, loss=52.296, backward_time=0.302, grad_norm=46.326, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.866e-04, train_time=1.380
+[gpub011:0/16] 2024-02-06 00:45:53,350 (trainer:762) INFO: 29epoch:train:10901-11000batch: iter_time=7.856e-05, forward_time=0.388, loss_ctc=46.214, loss_att=42.149, acc=0.753, loss=43.369, backward_time=0.331, grad_norm=37.833, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=1.866e-04, train_time=1.602
+[gpub011:0/16] 2024-02-06 00:48:10,540 (trainer:762) INFO: 29epoch:train:11001-11100batch: iter_time=7.872e-05, forward_time=0.291, loss_ctc=46.445, loss_att=44.794, acc=0.761, loss=45.289, backward_time=0.297, grad_norm=36.886, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.866e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 00:50:40,220 (trainer:762) INFO: 29epoch:train:11101-11200batch: iter_time=8.025e-05, forward_time=0.300, loss_ctc=47.796, loss_att=49.106, acc=0.743, loss=48.713, backward_time=0.300, grad_norm=45.076, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.865e-04, train_time=1.497
+[gpub011:0/16] 2024-02-06 00:52:12,344 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-06 00:52:31,673 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 00:52:35,255 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71c751bb80>)
+[gpub011:0/16] 2024-02-06 00:52:35,255 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-06 00:52:35,262 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 00:59:24,973 (trainer:762) INFO: 29epoch:train:11201-11300batch: iter_time=3.770, forward_time=0.408, loss_ctc=41.499, loss_att=43.718, acc=0.760, loss=43.052, backward_time=0.334, grad_norm=35.480, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.865e-04, train_time=5.247
+[gpub011:0/16] 2024-02-06 01:01:50,433 (trainer:762) INFO: 29epoch:train:11301-11400batch: iter_time=8.236e-05, forward_time=0.333, loss_ctc=44.034, loss_att=45.356, acc=0.761, loss=44.959, backward_time=0.300, grad_norm=41.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.865e-04, train_time=1.454
+[gpub011:0/16] 2024-02-06 01:04:07,740 (trainer:762) INFO: 29epoch:train:11401-11500batch: iter_time=8.165e-05, forward_time=0.423, loss_ctc=52.435, loss_att=51.778, acc=0.763, loss=51.976, backward_time=0.328, grad_norm=46.307, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.865e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 01:06:43,013 (trainer:762) INFO: 29epoch:train:11501-11600batch: iter_time=8.172e-05, forward_time=0.292, loss_ctc=46.261, loss_att=43.696, acc=0.764, loss=44.466, backward_time=0.298, grad_norm=40.296, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.865e-04, train_time=1.553
+[gpub011:0/16] 2024-02-06 01:09:09,543 (trainer:762) INFO: 29epoch:train:11601-11700batch: iter_time=8.088e-05, forward_time=0.441, loss_ctc=54.335, loss_att=52.947, acc=0.741, loss=53.363, backward_time=0.344, grad_norm=45.863, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=1.864e-04, train_time=1.465
+[gpub011:0/16] 2024-02-06 01:11:32,856 (trainer:762) INFO: 29epoch:train:11701-11800batch: iter_time=8.433e-05, forward_time=0.287, loss_ctc=37.064, loss_att=33.578, acc=0.779, loss=34.624, backward_time=0.292, grad_norm=37.433, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.864e-04, train_time=1.431
+[gpub011:0/16] 2024-02-06 01:13:30,230 (trainer:762) INFO: 29epoch:train:11801-11900batch: iter_time=2.106e-04, forward_time=0.300, loss_ctc=54.024, loss_att=48.609, acc=0.765, loss=50.234, backward_time=0.310, grad_norm=39.844, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.864e-04, train_time=1.175
+[gpub011:0/16] 2024-02-06 01:15:55,542 (trainer:762) INFO: 29epoch:train:11901-12000batch: iter_time=7.994e-05, forward_time=0.397, loss_ctc=46.318, loss_att=50.137, acc=0.737, loss=48.991, backward_time=0.341, grad_norm=37.311, clip=100.000, loss_scale=4.206e+33, optim_step_time=0.098, optim0_lr0=1.864e-04, train_time=1.452
+[gpub011:0/16] 2024-02-06 01:18:13,582 (trainer:762) INFO: 29epoch:train:12001-12100batch: iter_time=8.039e-05, forward_time=0.291, loss_ctc=54.197, loss_att=47.036, acc=0.764, loss=49.184, backward_time=0.296, grad_norm=40.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.863e-04, train_time=1.381
+[gpub011:0/16] 2024-02-06 01:20:39,036 (trainer:762) INFO: 29epoch:train:12101-12200batch: iter_time=9.028e-05, forward_time=0.390, loss_ctc=53.139, loss_att=52.619, acc=0.751, loss=52.775, backward_time=0.356, grad_norm=43.974, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.863e-04, train_time=1.454
+[gpub011:0/16] 2024-02-06 01:22:57,867 (trainer:762) INFO: 29epoch:train:12201-12300batch: iter_time=8.529e-05, forward_time=0.289, loss_ctc=43.670, loss_att=43.686, acc=0.760, loss=43.681, backward_time=0.294, grad_norm=36.119, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.863e-04, train_time=1.388
+[gpub011:0/16] 2024-02-06 01:25:05,238 (trainer:762) INFO: 29epoch:train:12301-12400batch: iter_time=2.072e-04, forward_time=0.301, loss_ctc=49.643, loss_att=47.576, acc=0.766, loss=48.196, backward_time=0.302, grad_norm=41.180, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.863e-04, train_time=1.274
+[gpub011:0/16] 2024-02-06 01:27:44,969 (trainer:762) INFO: 29epoch:train:12401-12500batch: iter_time=8.539e-05, forward_time=0.367, loss_ctc=41.485, loss_att=45.463, acc=0.748, loss=44.270, backward_time=0.380, grad_norm=36.565, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.863e-04, train_time=1.597
+[gpub011:0/16] 2024-02-06 01:28:04,997 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-06 01:28:24,349 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 01:28:28,187 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71bfb58df0>)
+[gpub011:0/16] 2024-02-06 01:28:28,187 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-06 01:28:28,190 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 01:36:08,979 (trainer:762) INFO: 29epoch:train:12501-12600batch: iter_time=3.566, forward_time=0.287, loss_ctc=42.791, loss_att=36.737, acc=0.788, loss=38.553, backward_time=0.292, grad_norm=36.920, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.862e-04, train_time=5.040
+[gpub011:0/16] 2024-02-06 01:38:49,419 (trainer:762) INFO: 29epoch:train:12601-12700batch: iter_time=8.649e-05, forward_time=0.401, loss_ctc=49.693, loss_att=48.866, acc=0.747, loss=49.114, backward_time=0.348, grad_norm=42.743, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.862e-04, train_time=1.604
+[gpub011:0/16] 2024-02-06 01:41:11,441 (trainer:762) INFO: 29epoch:train:12701-12800batch: iter_time=8.065e-05, forward_time=0.291, loss_ctc=47.113, loss_att=47.814, acc=0.760, loss=47.604, backward_time=0.296, grad_norm=42.981, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.862e-04, train_time=1.419
+[gpub011:0/16] 2024-02-06 01:43:32,612 (trainer:762) INFO: 29epoch:train:12801-12900batch: iter_time=1.814e-04, forward_time=0.311, loss_ctc=54.538, loss_att=49.952, acc=0.742, loss=51.328, backward_time=0.301, grad_norm=50.652, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.862e-04, train_time=1.412
+[gpub011:0/16] 2024-02-06 01:46:11,851 (trainer:762) INFO: 29epoch:train:12901-13000batch: iter_time=8.486e-05, forward_time=0.357, loss_ctc=43.274, loss_att=45.724, acc=0.742, loss=44.989, backward_time=0.358, grad_norm=37.844, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.862e-04, train_time=1.592
+[gpub011:0/16] 2024-02-06 01:48:33,270 (trainer:762) INFO: 29epoch:train:13001-13100batch: iter_time=8.281e-05, forward_time=0.306, loss_ctc=44.567, loss_att=42.460, acc=0.754, loss=43.092, backward_time=0.293, grad_norm=40.679, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.861e-04, train_time=1.414
+[gpub011:0/16] 2024-02-06 01:50:43,900 (trainer:762) INFO: 29epoch:train:13101-13200batch: iter_time=8.608e-05, forward_time=0.293, loss_ctc=51.508, loss_att=46.503, acc=0.759, loss=48.005, backward_time=0.298, grad_norm=38.386, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.861e-04, train_time=1.306
+[gpub011:0/16] 2024-02-06 01:53:56,378 (trainer:762) INFO: 29epoch:train:13201-13300batch: iter_time=8.636e-05, forward_time=0.428, loss_ctc=51.439, loss_att=50.582, acc=0.740, loss=50.839, backward_time=0.345, grad_norm=36.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.861e-04, train_time=1.925
+[gpub011:0/16] 2024-02-06 01:56:06,360 (trainer:762) INFO: 29epoch:train:13301-13400batch: iter_time=8.324e-05, forward_time=0.291, loss_ctc=54.660, loss_att=50.719, acc=0.754, loss=51.901, backward_time=0.298, grad_norm=44.924, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.861e-04, train_time=1.300
+[gpub011:0/16] 2024-02-06 01:58:31,537 (trainer:762) INFO: 29epoch:train:13401-13500batch: iter_time=7.866e-05, forward_time=0.288, loss_ctc=45.494, loss_att=41.469, acc=0.755, loss=42.676, backward_time=0.294, grad_norm=39.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.860e-04, train_time=1.452
+[gpub011:0/16] 2024-02-06 02:01:01,297 (trainer:762) INFO: 29epoch:train:13501-13600batch: iter_time=8.278e-05, forward_time=0.326, loss_ctc=46.540, loss_att=44.618, acc=0.763, loss=45.194, backward_time=0.307, grad_norm=37.374, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.860e-04, train_time=1.497
+[gpub011:0/16] 2024-02-06 02:03:26,659 (trainer:762) INFO: 29epoch:train:13601-13700batch: iter_time=7.638e-05, forward_time=0.390, loss_ctc=47.801, loss_att=48.068, acc=0.750, loss=47.988, backward_time=0.348, grad_norm=44.984, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.860e-04, train_time=1.453
+[gpub011:0/16] 2024-02-06 02:05:13,868 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-06 02:05:33,282 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 02:05:36,885 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b464c16f0>)
+[gpub011:0/16] 2024-02-06 02:05:36,885 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-06 02:05:36,929 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 02:12:43,607 (trainer:762) INFO: 29epoch:train:13701-13800batch: iter_time=4.043, forward_time=0.297, loss_ctc=41.142, loss_att=43.036, acc=0.758, loss=42.468, backward_time=0.292, grad_norm=35.492, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.860e-04, train_time=5.569
+[gpub011:0/16] 2024-02-06 02:15:10,284 (trainer:762) INFO: 29epoch:train:13801-13900batch: iter_time=8.731e-05, forward_time=0.427, loss_ctc=43.957, loss_att=41.892, acc=0.765, loss=42.511, backward_time=0.313, grad_norm=41.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.860e-04, train_time=1.467
+[gpub011:0/16] 2024-02-06 02:17:44,990 (trainer:762) INFO: 29epoch:train:13901-14000batch: iter_time=8.623e-05, forward_time=0.293, loss_ctc=52.353, loss_att=51.111, acc=0.756, loss=51.484, backward_time=0.298, grad_norm=44.408, clip=100.000, loss_scale=8.412e+33, optim_step_time=0.094, optim0_lr0=1.859e-04, train_time=1.546
+[gpub011:0/16] 2024-02-06 02:19:55,297 (trainer:762) INFO: 29epoch:train:14001-14100batch: iter_time=8.838e-05, forward_time=0.307, loss_ctc=45.706, loss_att=42.734, acc=0.763, loss=43.626, backward_time=0.310, grad_norm=39.000, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.859e-04, train_time=1.304
+[gpub011:0/16] 2024-02-06 02:22:31,093 (trainer:762) INFO: 29epoch:train:14101-14200batch: iter_time=8.907e-05, forward_time=0.381, loss_ctc=54.352, loss_att=53.144, acc=0.728, loss=53.507, backward_time=0.315, grad_norm=47.041, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.859e-04, train_time=1.558
+[gpub011:0/16] 2024-02-06 02:24:44,767 (trainer:762) INFO: 29epoch:train:14201-14300batch: iter_time=9.095e-05, forward_time=0.295, loss_ctc=37.283, loss_att=32.653, acc=0.775, loss=34.042, backward_time=0.293, grad_norm=36.833, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.859e-04, train_time=1.337
+[gpub011:0/16] 2024-02-06 02:27:16,986 (trainer:762) INFO: 29epoch:train:14301-14400batch: iter_time=9.149e-05, forward_time=0.293, loss_ctc=53.340, loss_att=47.932, acc=0.758, loss=49.555, backward_time=0.297, grad_norm=39.560, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.859e-04, train_time=1.521
+[gpub011:0/16] 2024-02-06 02:29:27,981 (trainer:762) INFO: 29epoch:train:14401-14500batch: iter_time=0.001, forward_time=0.392, loss_ctc=46.538, loss_att=47.833, acc=0.736, loss=47.444, backward_time=0.350, grad_norm=35.876, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.114, optim0_lr0=1.858e-04, train_time=1.309
+[gpub011:0/16] 2024-02-06 02:32:10,615 (trainer:762) INFO: 29epoch:train:14501-14600batch: iter_time=9.340e-05, forward_time=0.296, loss_ctc=54.131, loss_att=46.935, acc=0.758, loss=49.094, backward_time=0.294, grad_norm=40.751, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.858e-04, train_time=1.628
+[gpub011:0/16] 2024-02-06 02:34:26,694 (trainer:762) INFO: 29epoch:train:14601-14700batch: iter_time=5.988e-04, forward_time=0.318, loss_ctc=52.778, loss_att=52.475, acc=0.743, loss=52.566, backward_time=0.296, grad_norm=43.272, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.858e-04, train_time=1.361
+[gpub011:0/16] 2024-02-06 02:36:58,572 (trainer:762) INFO: 29epoch:train:14701-14800batch: iter_time=3.714e-04, forward_time=0.375, loss_ctc=43.724, loss_att=42.991, acc=0.761, loss=43.211, backward_time=0.369, grad_norm=36.927, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.858e-04, train_time=1.517
+[gpub011:0/16] 2024-02-06 02:39:30,466 (trainer:762) INFO: 29epoch:train:14801-14900batch: iter_time=8.441e-05, forward_time=0.298, loss_ctc=49.671, loss_att=48.055, acc=0.758, loss=48.540, backward_time=0.302, grad_norm=44.908, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.857e-04, train_time=1.520
+[gpub011:0/16] 2024-02-06 02:42:01,265 (trainer:762) INFO: 29epoch:train:14901-15000batch: iter_time=9.147e-05, forward_time=0.311, loss_ctc=41.421, loss_att=45.267, acc=0.736, loss=44.113, backward_time=0.295, grad_norm=35.526, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.857e-04, train_time=1.508
+[gpub011:0/16] 2024-02-06 03:21:57,339 (trainer:361) INFO: 29epoch results: [train] iter_time=0.288, forward_time=0.343, loss_ctc=48.829, loss_att=46.758, acc=0.754, loss=47.379, backward_time=0.316, grad_norm=41.335, clip=100.000, loss_scale=5.354e+33, optim_step_time=0.098, optim0_lr0=1.873e-04, train_time=1.742, time=7 hours, 15 minutes and 57.84 seconds, total_count=465000, gpu_max_cached_mem_GB=40.000, [valid] loss_ctc=37.023, cer_ctc=0.190, loss_att=39.435, acc=0.674, cer=0.340, wer=0.994, loss=38.711, time=39 minutes and 31.62 seconds, total_count=144801, gpu_max_cached_mem_GB=40.000
+[gpub011:0/16] 2024-02-06 03:22:19,309 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-06 03:22:19,420 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/23epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/24epoch.pth
+[gpub011:0/16] 2024-02-06 03:22:19,420 (trainer:290) INFO: 30/45epoch started. Estimated time to finish: 5 days, 6 hours and 53 minutes
+[gpub011:0/16] 2024-02-06 03:22:19,430 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-06 03:22:37,796 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 03:22:41,174 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b3ed545e0>)
+[gpub011:0/16] 2024-02-06 03:22:41,174 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-06 03:22:41,177 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 03:30:29,892 (trainer:762) INFO: 30epoch:train:1-100batch: iter_time=3.552, forward_time=0.304, loss_ctc=47.900, loss_att=44.820, acc=0.762, loss=45.744, backward_time=0.307, grad_norm=36.143, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.857e-04, train_time=4.904
+[gpub011:0/16] 2024-02-06 03:32:33,977 (trainer:762) INFO: 30epoch:train:101-200batch: iter_time=6.909e-04, forward_time=0.379, loss_ctc=50.111, loss_att=50.083, acc=0.762, loss=50.091, backward_time=0.336, grad_norm=45.272, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.857e-04, train_time=1.241
+[gpub011:0/16] 2024-02-06 03:33:50,973 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 03:34:40,753 (trainer:762) INFO: 30epoch:train:201-300batch: iter_time=8.330e-05, forward_time=0.289, loss_ctc=48.536, loss_att=42.408, acc=0.745, loss=44.247, backward_time=0.301, grad_norm=41.241, clip=100.000, loss_scale=8.549e+33, optim_step_time=0.095, optim0_lr0=1.857e-04, train_time=1.268
+[gpub011:0/16] 2024-02-06 03:37:17,846 (trainer:762) INFO: 30epoch:train:301-400batch: iter_time=8.105e-05, forward_time=0.314, loss_ctc=43.371, loss_att=41.672, acc=0.754, loss=42.182, backward_time=0.312, grad_norm=40.092, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.856e-04, train_time=1.570
+[gpub011:0/16] 2024-02-06 03:39:15,995 (trainer:762) INFO: 30epoch:train:401-500batch: iter_time=8.173e-05, forward_time=0.322, loss_ctc=46.125, loss_att=46.222, acc=0.760, loss=46.193, backward_time=0.310, grad_norm=37.703, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.856e-04, train_time=1.182
+[gpub011:0/16] 2024-02-06 03:41:43,945 (trainer:762) INFO: 30epoch:train:501-600batch: iter_time=8.725e-05, forward_time=0.351, loss_ctc=52.627, loss_att=44.209, acc=0.757, loss=46.735, backward_time=0.322, grad_norm=39.650, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.856e-04, train_time=1.479
+[gpub011:0/16] 2024-02-06 03:44:10,651 (trainer:762) INFO: 30epoch:train:601-700batch: iter_time=7.803e-05, forward_time=0.303, loss_ctc=54.151, loss_att=45.805, acc=0.740, loss=48.309, backward_time=0.305, grad_norm=43.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.856e-04, train_time=1.467
+[gpub011:0/16] 2024-02-06 03:46:15,078 (trainer:762) INFO: 30epoch:train:701-800batch: iter_time=7.970e-05, forward_time=0.304, loss_ctc=50.272, loss_att=47.992, acc=0.747, loss=48.676, backward_time=0.330, grad_norm=43.690, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.856e-04, train_time=1.243
+[gpub011:0/16] 2024-02-06 03:48:53,195 (trainer:762) INFO: 30epoch:train:801-900batch: iter_time=7.938e-05, forward_time=0.335, loss_ctc=46.237, loss_att=44.018, acc=0.767, loss=44.684, backward_time=0.304, grad_norm=36.753, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.855e-04, train_time=1.582
+[gpub011:0/16] 2024-02-06 03:51:05,833 (trainer:762) INFO: 30epoch:train:901-1000batch: iter_time=6.965e-04, forward_time=0.354, loss_ctc=55.861, loss_att=53.451, acc=0.740, loss=54.174, backward_time=0.298, grad_norm=46.198, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.855e-04, train_time=1.326
+[gpub011:0/16] 2024-02-06 03:53:16,994 (trainer:762) INFO: 30epoch:train:1001-1100batch: iter_time=2.655e-04, forward_time=0.312, loss_ctc=48.242, loss_att=46.487, acc=0.732, loss=47.014, backward_time=0.327, grad_norm=40.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.855e-04, train_time=1.311
+[gpub011:0/16] 2024-02-06 03:55:49,639 (trainer:762) INFO: 30epoch:train:1101-1200batch: iter_time=8.136e-05, forward_time=0.302, loss_ctc=47.946, loss_att=41.903, acc=0.759, loss=43.716, backward_time=0.305, grad_norm=39.704, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.855e-04, train_time=1.527
+[gpub011:0/16] 2024-02-06 03:57:14,485 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-06 03:57:33,523 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 03:57:37,052 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f8162594e50>)
+[gpub011:0/16] 2024-02-06 03:57:37,052 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-06 03:57:37,056 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 04:04:02,157 (trainer:762) INFO: 30epoch:train:1201-1300batch: iter_time=3.621, forward_time=0.369, loss_ctc=48.698, loss_att=46.466, acc=0.772, loss=47.136, backward_time=0.311, grad_norm=39.795, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.854e-04, train_time=4.925
+[gpub011:0/16] 2024-02-06 04:06:40,060 (trainer:762) INFO: 30epoch:train:1301-1400batch: iter_time=8.304e-05, forward_time=0.307, loss_ctc=46.731, loss_att=48.962, acc=0.763, loss=48.293, backward_time=0.313, grad_norm=42.061, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.854e-04, train_time=1.578
+[gpub011:0/16] 2024-02-06 04:09:03,002 (trainer:762) INFO: 30epoch:train:1401-1500batch: iter_time=8.244e-05, forward_time=0.300, loss_ctc=49.943, loss_att=47.846, acc=0.757, loss=48.475, backward_time=0.297, grad_norm=38.569, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.854e-04, train_time=1.430
+[gpub011:0/16] 2024-02-06 04:11:33,647 (trainer:762) INFO: 30epoch:train:1501-1600batch: iter_time=8.457e-05, forward_time=0.380, loss_ctc=43.511, loss_att=40.217, acc=0.766, loss=41.205, backward_time=0.326, grad_norm=38.262, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.854e-04, train_time=1.506
+[gpub011:0/16] 2024-02-06 04:13:59,277 (trainer:762) INFO: 30epoch:train:1601-1700batch: iter_time=8.785e-05, forward_time=0.293, loss_ctc=45.448, loss_att=46.030, acc=0.746, loss=45.856, backward_time=0.298, grad_norm=39.171, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.854e-04, train_time=1.456
+[gpub011:0/16] 2024-02-06 04:16:29,032 (trainer:762) INFO: 30epoch:train:1701-1800batch: iter_time=8.051e-05, forward_time=0.317, loss_ctc=48.988, loss_att=46.460, acc=0.771, loss=47.218, backward_time=0.297, grad_norm=38.233, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.853e-04, train_time=1.497
+[gpub011:0/16] 2024-02-06 04:18:56,612 (trainer:762) INFO: 30epoch:train:1801-1900batch: iter_time=8.374e-05, forward_time=0.356, loss_ctc=50.387, loss_att=43.007, acc=0.760, loss=45.221, backward_time=0.326, grad_norm=38.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.853e-04, train_time=1.476
+[gpub011:0/16] 2024-02-06 04:21:13,300 (trainer:762) INFO: 30epoch:train:1901-2000batch: iter_time=8.099e-05, forward_time=0.297, loss_ctc=51.354, loss_att=48.053, acc=0.746, loss=49.043, backward_time=0.298, grad_norm=42.991, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.853e-04, train_time=1.367
+[gpub011:0/16] 2024-02-06 04:23:41,318 (trainer:762) INFO: 30epoch:train:2001-2100batch: iter_time=8.556e-05, forward_time=0.316, loss_ctc=46.251, loss_att=44.548, acc=0.768, loss=45.059, backward_time=0.302, grad_norm=40.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.853e-04, train_time=1.480
+[gpub011:0/16] 2024-02-06 04:26:04,663 (trainer:762) INFO: 30epoch:train:2101-2200batch: iter_time=8.121e-05, forward_time=0.307, loss_ctc=55.454, loss_att=53.898, acc=0.750, loss=54.365, backward_time=0.301, grad_norm=43.921, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.853e-04, train_time=1.433
+[gpub011:0/16] 2024-02-06 04:28:41,612 (trainer:762) INFO: 30epoch:train:2201-2300batch: iter_time=8.324e-05, forward_time=0.354, loss_ctc=49.425, loss_att=44.621, acc=0.753, loss=46.063, backward_time=0.316, grad_norm=36.699, clip=100.000, loss_scale=7.010e+33, optim_step_time=0.097, optim0_lr0=1.852e-04, train_time=1.570
+[gpub011:0/16] 2024-02-06 04:30:57,848 (trainer:762) INFO: 30epoch:train:2301-2400batch: iter_time=8.291e-05, forward_time=0.293, loss_ctc=48.326, loss_att=45.804, acc=0.769, loss=46.561, backward_time=0.301, grad_norm=38.951, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.852e-04, train_time=1.362
+[gpub011:0/16] 2024-02-06 04:33:25,356 (trainer:762) INFO: 30epoch:train:2401-2500batch: iter_time=8.178e-05, forward_time=0.301, loss_ctc=47.887, loss_att=43.046, acc=0.767, loss=44.498, backward_time=0.304, grad_norm=39.469, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.852e-04, train_time=1.475
+[gpub011:0/16] 2024-02-06 04:33:45,407 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-06 04:34:04,506 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 04:34:08,388 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f72045e01f0>)
+[gpub011:0/16] 2024-02-06 04:34:08,388 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-06 04:34:08,392 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 04:41:42,097 (trainer:762) INFO: 30epoch:train:2501-2600batch: iter_time=3.611, forward_time=0.379, loss_ctc=47.022, loss_att=43.748, acc=0.770, loss=44.730, backward_time=0.319, grad_norm=34.220, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.852e-04, train_time=4.967
+[gpub011:0/16] 2024-02-06 04:43:55,885 (trainer:762) INFO: 30epoch:train:2601-2700batch: iter_time=8.816e-05, forward_time=0.312, loss_ctc=49.728, loss_att=51.461, acc=0.769, loss=50.941, backward_time=0.299, grad_norm=42.490, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.852e-04, train_time=1.338
+[gpub011:0/16] 2024-02-06 04:46:13,231 (trainer:762) INFO: 30epoch:train:2701-2800batch: iter_time=7.945e-05, forward_time=0.346, loss_ctc=47.270, loss_att=41.458, acc=0.760, loss=43.202, backward_time=0.352, grad_norm=39.023, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.851e-04, train_time=1.373
+[gpub011:0/16] 2024-02-06 04:48:12,143 (trainer:762) INFO: 30epoch:train:2801-2900batch: iter_time=8.003e-05, forward_time=0.308, loss_ctc=42.548, loss_att=41.529, acc=0.760, loss=41.835, backward_time=0.298, grad_norm=37.299, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.851e-04, train_time=1.189
+[gpub011:0/16] 2024-02-06 04:50:37,538 (trainer:762) INFO: 30epoch:train:2901-3000batch: iter_time=8.775e-05, forward_time=0.346, loss_ctc=45.308, loss_att=46.581, acc=0.765, loss=46.199, backward_time=0.340, grad_norm=35.820, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=1.851e-04, train_time=1.454
+[gpub011:0/16] 2024-02-06 04:53:00,435 (trainer:762) INFO: 30epoch:train:3001-3100batch: iter_time=8.616e-05, forward_time=0.313, loss_ctc=51.213, loss_att=44.623, acc=0.765, loss=46.600, backward_time=0.300, grad_norm=40.445, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.851e-04, train_time=1.429
+[gpub011:0/16] 2024-02-06 04:55:05,328 (trainer:762) INFO: 30epoch:train:3101-3200batch: iter_time=1.326e-04, forward_time=0.371, loss_ctc=51.895, loss_att=44.723, acc=0.747, loss=46.875, backward_time=0.340, grad_norm=40.147, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.850e-04, train_time=1.248
+[gpub011:0/16] 2024-02-06 04:57:25,396 (trainer:762) INFO: 30epoch:train:3201-3300batch: iter_time=9.318e-05, forward_time=0.307, loss_ctc=49.028, loss_att=47.253, acc=0.761, loss=47.786, backward_time=0.313, grad_norm=43.026, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.850e-04, train_time=1.401
+[gpub011:0/16] 2024-02-06 04:59:53,715 (trainer:762) INFO: 30epoch:train:3301-3400batch: iter_time=1.002e-04, forward_time=0.334, loss_ctc=45.740, loss_att=43.747, acc=0.777, loss=44.345, backward_time=0.371, grad_norm=35.324, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.850e-04, train_time=1.483
+[gpub011:0/16] 2024-02-06 05:01:52,134 (trainer:762) INFO: 30epoch:train:3401-3500batch: iter_time=8.765e-05, forward_time=0.316, loss_ctc=54.579, loss_att=53.163, acc=0.746, loss=53.588, backward_time=0.306, grad_norm=43.214, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.850e-04, train_time=1.183
+[gpub011:0/16] 2024-02-06 05:04:24,793 (trainer:762) INFO: 30epoch:train:3501-3600batch: iter_time=4.233e-04, forward_time=0.361, loss_ctc=47.217, loss_att=45.167, acc=0.748, loss=45.782, backward_time=0.313, grad_norm=38.763, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.850e-04, train_time=1.527
+[gpub011:0/16] 2024-02-06 05:06:29,728 (trainer:762) INFO: 30epoch:train:3601-3700batch: iter_time=8.908e-05, forward_time=0.301, loss_ctc=47.286, loss_att=41.934, acc=0.770, loss=43.540, backward_time=0.319, grad_norm=38.782, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.849e-04, train_time=1.249
+[gpub011:0/16] 2024-02-06 05:07:58,977 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-06 05:08:18,089 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 05:08:21,655 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b70d175e0>)
+[gpub011:0/16] 2024-02-06 05:08:21,655 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-06 05:08:21,659 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 05:14:24,749 (trainer:762) INFO: 30epoch:train:3701-3800batch: iter_time=3.475, forward_time=0.304, loss_ctc=47.287, loss_att=44.810, acc=0.777, loss=45.553, backward_time=0.309, grad_norm=38.203, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.849e-04, train_time=4.750
+[gpub011:0/16] 2024-02-06 05:16:53,834 (trainer:762) INFO: 30epoch:train:3801-3900batch: iter_time=0.003, forward_time=0.392, loss_ctc=45.582, loss_att=48.074, acc=0.766, loss=47.326, backward_time=0.313, grad_norm=39.479, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.849e-04, train_time=1.490
+[gpub011:0/16] 2024-02-06 05:18:57,084 (trainer:762) INFO: 30epoch:train:3901-4000batch: iter_time=8.386e-05, forward_time=0.291, loss_ctc=49.291, loss_att=45.983, acc=0.763, loss=46.975, backward_time=0.297, grad_norm=38.866, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.849e-04, train_time=1.233
+[gpub011:0/16] 2024-02-06 05:20:54,435 (trainer:762) INFO: 30epoch:train:4001-4100batch: iter_time=8.271e-05, forward_time=0.314, loss_ctc=43.158, loss_att=39.066, acc=0.771, loss=40.294, backward_time=0.298, grad_norm=37.386, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.849e-04, train_time=1.172
+[gpub011:0/16] 2024-02-06 05:23:02,497 (trainer:762) INFO: 30epoch:train:4101-4200batch: iter_time=8.424e-05, forward_time=0.290, loss_ctc=44.562, loss_att=44.496, acc=0.752, loss=44.516, backward_time=0.296, grad_norm=38.078, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.848e-04, train_time=1.282
+[gpub011:0/16] 2024-02-06 05:25:29,303 (trainer:762) INFO: 30epoch:train:4201-4300batch: iter_time=8.887e-05, forward_time=0.353, loss_ctc=47.972, loss_att=45.593, acc=0.774, loss=46.307, backward_time=0.330, grad_norm=37.077, clip=100.000, loss_scale=1.402e+34, optim_step_time=0.099, optim0_lr0=1.848e-04, train_time=1.467
+[gpub011:0/16] 2024-02-06 05:26:53,809 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 05:27:36,016 (trainer:762) INFO: 30epoch:train:4301-4400batch: iter_time=8.560e-05, forward_time=0.320, loss_ctc=49.496, loss_att=42.256, acc=0.764, loss=44.428, backward_time=0.297, grad_norm=38.372, clip=100.000, loss_scale=1.752e+34, optim_step_time=0.094, optim0_lr0=1.848e-04, train_time=1.267
+[gpub011:0/16] 2024-02-06 05:29:39,644 (trainer:762) INFO: 30epoch:train:4401-4500batch: iter_time=8.084e-05, forward_time=0.314, loss_ctc=50.723, loss_att=47.036, acc=0.749, loss=48.142, backward_time=0.299, grad_norm=42.097, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.848e-04, train_time=1.236
+[gpub011:0/16] 2024-02-06 05:31:53,775 (trainer:762) INFO: 30epoch:train:4501-4600batch: iter_time=9.170e-05, forward_time=0.289, loss_ctc=45.255, loss_att=43.655, acc=0.770, loss=44.135, backward_time=0.296, grad_norm=39.824, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.848e-04, train_time=1.342
+[gpub011:0/16] 2024-02-06 05:34:26,832 (trainer:762) INFO: 30epoch:train:4601-4700batch: iter_time=5.360e-04, forward_time=0.367, loss_ctc=54.500, loss_att=52.683, acc=0.755, loss=53.228, backward_time=0.400, grad_norm=42.483, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=1.847e-04, train_time=1.530
+[gpub011:0/16] 2024-02-06 05:36:26,659 (trainer:762) INFO: 30epoch:train:4701-4800batch: iter_time=8.455e-05, forward_time=0.288, loss_ctc=48.396, loss_att=44.058, acc=0.756, loss=45.360, backward_time=0.296, grad_norm=37.509, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.847e-04, train_time=1.198
+[gpub011:0/16] 2024-02-06 05:38:42,450 (trainer:762) INFO: 30epoch:train:4801-4900batch: iter_time=8.610e-05, forward_time=0.291, loss_ctc=47.938, loss_att=45.760, acc=0.770, loss=46.413, backward_time=0.296, grad_norm=40.408, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.847e-04, train_time=1.358
+[gpub011:0/16] 2024-02-06 05:41:09,267 (trainer:762) INFO: 30epoch:train:4901-5000batch: iter_time=8.393e-05, forward_time=0.409, loss_ctc=47.015, loss_att=42.618, acc=0.767, loss=43.937, backward_time=0.317, grad_norm=38.375, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.847e-04, train_time=1.468
+[gpub011:0/16] 2024-02-06 05:41:29,317 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-06 05:41:48,317 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 05:41:51,876 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f722efd0af0>)
+[gpub011:0/16] 2024-02-06 05:41:51,876 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-06 05:41:51,880 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 05:49:03,404 (trainer:762) INFO: 30epoch:train:5001-5100batch: iter_time=3.518, forward_time=0.311, loss_ctc=46.815, loss_att=44.464, acc=0.769, loss=45.169, backward_time=0.299, grad_norm=35.391, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.846e-04, train_time=4.741
+[gpub011:0/16] 2024-02-06 05:51:09,850 (trainer:762) INFO: 30epoch:train:5101-5200batch: iter_time=7.940e-05, forward_time=0.292, loss_ctc=49.609, loss_att=50.790, acc=0.773, loss=50.436, backward_time=0.299, grad_norm=42.424, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.846e-04, train_time=1.264
+[gpub011:0/16] 2024-02-06 05:53:42,488 (trainer:762) INFO: 30epoch:train:5201-5300batch: iter_time=8.082e-05, forward_time=0.337, loss_ctc=46.726, loss_att=40.873, acc=0.764, loss=42.629, backward_time=0.386, grad_norm=37.523, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.846e-04, train_time=1.525
+[gpub011:0/16] 2024-02-06 05:55:47,527 (trainer:762) INFO: 30epoch:train:5301-5400batch: iter_time=7.890e-05, forward_time=0.331, loss_ctc=42.231, loss_att=41.341, acc=0.762, loss=41.608, backward_time=0.304, grad_norm=37.323, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.846e-04, train_time=1.250
+[gpub011:0/16] 2024-02-06 05:58:06,900 (trainer:762) INFO: 30epoch:train:5401-5500batch: iter_time=8.995e-05, forward_time=0.292, loss_ctc=45.082, loss_att=45.623, acc=0.768, loss=45.461, backward_time=0.296, grad_norm=36.926, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.846e-04, train_time=1.394
+[gpub011:0/16] 2024-02-06 06:00:39,373 (trainer:762) INFO: 30epoch:train:5501-5600batch: iter_time=0.002, forward_time=0.392, loss_ctc=51.128, loss_att=43.972, acc=0.768, loss=46.119, backward_time=0.319, grad_norm=38.158, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.845e-04, train_time=1.524
+[gpub011:0/16] 2024-02-06 06:02:45,603 (trainer:762) INFO: 30epoch:train:5601-5700batch: iter_time=7.760e-05, forward_time=0.319, loss_ctc=51.363, loss_att=44.446, acc=0.749, loss=46.521, backward_time=0.304, grad_norm=41.394, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.845e-04, train_time=1.262
+[gpub011:0/16] 2024-02-06 06:04:56,777 (trainer:762) INFO: 30epoch:train:5701-5800batch: iter_time=7.758e-05, forward_time=0.291, loss_ctc=48.743, loss_att=47.198, acc=0.762, loss=47.662, backward_time=0.297, grad_norm=42.256, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.845e-04, train_time=1.311
+[gpub011:0/16] 2024-02-06 06:07:18,962 (trainer:762) INFO: 30epoch:train:5801-5900batch: iter_time=8.059e-05, forward_time=0.292, loss_ctc=45.828, loss_att=43.395, acc=0.779, loss=44.125, backward_time=0.297, grad_norm=35.088, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.845e-04, train_time=1.422
+[gpub011:0/16] 2024-02-06 06:09:25,837 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 06:09:38,130 (trainer:762) INFO: 30epoch:train:5901-6000batch: iter_time=7.894e-05, forward_time=0.361, loss_ctc=53.560, loss_att=52.367, acc=0.749, loss=52.725, backward_time=0.389, grad_norm=42.811, clip=100.000, loss_scale=9.808e+33, optim_step_time=0.099, optim0_lr0=1.845e-04, train_time=1.391
+[gpub011:0/16] 2024-02-06 06:11:49,602 (trainer:762) INFO: 30epoch:train:6001-6100batch: iter_time=8.022e-05, forward_time=0.291, loss_ctc=46.312, loss_att=45.501, acc=0.750, loss=45.744, backward_time=0.296, grad_norm=38.017, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.844e-04, train_time=1.314
+[gpub011:0/16] 2024-02-06 06:14:08,493 (trainer:762) INFO: 30epoch:train:6101-6200batch: iter_time=7.944e-05, forward_time=0.290, loss_ctc=47.036, loss_att=42.083, acc=0.770, loss=43.569, backward_time=0.295, grad_norm=37.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.844e-04, train_time=1.388
+[gpub011:0/16] 2024-02-06 06:15:31,270 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-06 06:15:50,605 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 06:15:54,238 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f792aea84c0>)
+[gpub011:0/16] 2024-02-06 06:15:54,238 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-06 06:15:54,242 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 06:22:01,342 (trainer:762) INFO: 30epoch:train:6201-6300batch: iter_time=3.463, forward_time=0.362, loss_ctc=47.976, loss_att=46.325, acc=0.771, loss=46.820, backward_time=0.307, grad_norm=37.466, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.844e-04, train_time=4.728
+[gpub011:0/16] 2024-02-06 06:24:08,220 (trainer:762) INFO: 30epoch:train:6301-6400batch: iter_time=7.374e-05, forward_time=0.289, loss_ctc=45.372, loss_att=48.312, acc=0.766, loss=47.430, backward_time=0.294, grad_norm=41.245, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.844e-04, train_time=1.269
+[gpub011:0/16] 2024-02-06 06:26:32,495 (trainer:762) INFO: 30epoch:train:6401-6500batch: iter_time=8.057e-05, forward_time=0.414, loss_ctc=49.318, loss_att=45.850, acc=0.751, loss=46.891, backward_time=0.320, grad_norm=39.212, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.844e-04, train_time=1.443
+[gpub011:0/16] 2024-02-06 06:27:19,765 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 06:28:48,401 (trainer:762) INFO: 30epoch:train:6501-6600batch: iter_time=8.401e-05, forward_time=0.287, loss_ctc=42.842, loss_att=39.528, acc=0.769, loss=40.522, backward_time=0.292, grad_norm=36.313, clip=100.000, loss_scale=3.252e+33, optim_step_time=0.094, optim0_lr0=1.843e-04, train_time=1.359
+[gpub011:0/16] 2024-02-06 06:31:23,289 (trainer:762) INFO: 30epoch:train:6601-6700batch: iter_time=8.663e-05, forward_time=0.385, loss_ctc=44.058, loss_att=44.862, acc=0.747, loss=44.621, backward_time=0.361, grad_norm=37.296, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.843e-04, train_time=1.548
+[gpub011:0/16] 2024-02-06 06:33:37,740 (trainer:762) INFO: 30epoch:train:6701-6800batch: iter_time=8.313e-05, forward_time=0.290, loss_ctc=47.450, loss_att=45.622, acc=0.772, loss=46.170, backward_time=0.295, grad_norm=36.553, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.843e-04, train_time=1.345
+[gpub011:0/16] 2024-02-06 06:36:00,995 (trainer:762) INFO: 30epoch:train:6801-6900batch: iter_time=8.067e-05, forward_time=0.358, loss_ctc=49.096, loss_att=41.643, acc=0.762, loss=43.879, backward_time=0.364, grad_norm=40.065, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.843e-04, train_time=1.432
+[gpub011:0/16] 2024-02-06 06:38:32,198 (trainer:762) INFO: 30epoch:train:6901-7000batch: iter_time=8.110e-05, forward_time=0.292, loss_ctc=50.163, loss_att=46.821, acc=0.748, loss=47.824, backward_time=0.296, grad_norm=42.244, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.842e-04, train_time=1.513
+[gpub011:0/16] 2024-02-06 06:40:41,672 (trainer:762) INFO: 30epoch:train:7001-7100batch: iter_time=7.927e-05, forward_time=0.290, loss_ctc=45.019, loss_att=44.279, acc=0.761, loss=44.501, backward_time=0.296, grad_norm=41.066, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.842e-04, train_time=1.295
+[gpub011:0/16] 2024-02-06 06:43:11,960 (trainer:762) INFO: 30epoch:train:7101-7200batch: iter_time=7.954e-05, forward_time=0.456, loss_ctc=53.687, loss_att=53.109, acc=0.749, loss=53.282, backward_time=0.338, grad_norm=43.231, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.842e-04, train_time=1.502
+[gpub011:0/16] 2024-02-06 06:45:29,770 (trainer:762) INFO: 30epoch:train:7201-7300batch: iter_time=8.008e-05, forward_time=0.290, loss_ctc=47.654, loss_att=43.630, acc=0.754, loss=44.838, backward_time=0.295, grad_norm=37.946, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.842e-04, train_time=1.379
+[gpub011:0/16] 2024-02-06 06:47:34,352 (trainer:762) INFO: 30epoch:train:7301-7400batch: iter_time=7.999e-05, forward_time=0.290, loss_ctc=47.484, loss_att=44.755, acc=0.765, loss=45.574, backward_time=0.296, grad_norm=40.424, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.842e-04, train_time=1.246
+[gpub011:0/16] 2024-02-06 06:50:31,946 (trainer:762) INFO: 30epoch:train:7401-7500batch: iter_time=3.282e-04, forward_time=0.384, loss_ctc=46.568, loss_att=42.465, acc=0.764, loss=43.696, backward_time=0.363, grad_norm=40.481, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.841e-04, train_time=1.775
+[gpub011:0/16] 2024-02-06 06:50:52,106 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-06 06:51:11,179 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 06:51:14,743 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f792aeab7c0>)
+[gpub011:0/16] 2024-02-06 06:51:14,744 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-06 06:51:14,748 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 06:58:39,759 (trainer:762) INFO: 30epoch:train:7501-7600batch: iter_time=3.476, forward_time=0.370, loss_ctc=46.589, loss_att=44.498, acc=0.772, loss=45.125, backward_time=0.395, grad_norm=35.382, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.841e-04, train_time=4.879
+[gpub011:0/16] 2024-02-06 07:01:02,572 (trainer:762) INFO: 30epoch:train:7601-7700batch: iter_time=8.284e-05, forward_time=0.290, loss_ctc=49.158, loss_att=51.585, acc=0.771, loss=50.857, backward_time=0.298, grad_norm=43.439, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.841e-04, train_time=1.426
+[gpub011:0/16] 2024-02-06 07:03:35,679 (trainer:762) INFO: 30epoch:train:7701-7800batch: iter_time=3.135e-04, forward_time=0.444, loss_ctc=46.215, loss_att=40.907, acc=0.765, loss=42.499, backward_time=0.334, grad_norm=38.873, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.841e-04, train_time=1.532
+[gpub011:0/16] 2024-02-06 07:06:12,227 (trainer:762) INFO: 30epoch:train:7801-7900batch: iter_time=4.035e-04, forward_time=0.389, loss_ctc=41.602, loss_att=41.173, acc=0.765, loss=41.302, backward_time=0.350, grad_norm=38.249, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.841e-04, train_time=1.565
+[gpub011:0/16] 2024-02-06 07:08:38,277 (trainer:762) INFO: 30epoch:train:7901-8000batch: iter_time=8.993e-05, forward_time=0.288, loss_ctc=44.474, loss_att=45.906, acc=0.768, loss=45.477, backward_time=0.295, grad_norm=36.100, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.840e-04, train_time=1.460
+[gpub011:0/16] 2024-02-06 07:11:10,973 (trainer:762) INFO: 30epoch:train:8001-8100batch: iter_time=5.236e-04, forward_time=0.396, loss_ctc=50.307, loss_att=44.133, acc=0.769, loss=45.985, backward_time=0.324, grad_norm=38.655, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.112, optim0_lr0=1.840e-04, train_time=1.526
+[gpub011:0/16] 2024-02-06 07:13:14,346 (trainer:762) INFO: 30epoch:train:8101-8200batch: iter_time=9.059e-05, forward_time=0.290, loss_ctc=50.529, loss_att=43.751, acc=0.751, loss=45.784, backward_time=0.296, grad_norm=41.772, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.840e-04, train_time=1.234
+[gpub011:0/16] 2024-02-06 07:15:56,149 (trainer:762) INFO: 30epoch:train:8201-8300batch: iter_time=8.766e-05, forward_time=0.356, loss_ctc=48.242, loss_att=47.217, acc=0.762, loss=47.525, backward_time=0.322, grad_norm=43.699, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.840e-04, train_time=1.618
+[gpub011:0/16] 2024-02-06 07:18:10,042 (trainer:762) INFO: 30epoch:train:8301-8400batch: iter_time=8.457e-05, forward_time=0.290, loss_ctc=45.136, loss_att=43.052, acc=0.779, loss=43.677, backward_time=0.298, grad_norm=37.163, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.840e-04, train_time=1.338
+[gpub011:0/16] 2024-02-06 07:20:18,289 (trainer:762) INFO: 30epoch:train:8401-8500batch: iter_time=8.746e-05, forward_time=0.363, loss_ctc=53.086, loss_att=52.120, acc=0.750, loss=52.410, backward_time=0.313, grad_norm=44.372, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.839e-04, train_time=1.283
+[gpub011:0/16] 2024-02-06 07:23:06,113 (trainer:762) INFO: 30epoch:train:8501-8600batch: iter_time=8.292e-05, forward_time=0.318, loss_ctc=46.840, loss_att=44.976, acc=0.752, loss=45.535, backward_time=0.337, grad_norm=39.109, clip=100.000, loss_scale=4.517e+33, optim_step_time=0.096, optim0_lr0=1.839e-04, train_time=1.678
+[gpub011:0/16] 2024-02-06 07:25:10,686 (trainer:762) INFO: 30epoch:train:8601-8700batch: iter_time=8.455e-05, forward_time=0.289, loss_ctc=47.223, loss_att=41.759, acc=0.771, loss=43.398, backward_time=0.297, grad_norm=38.150, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.839e-04, train_time=1.245
+[gpub011:0/16] 2024-02-06 07:26:28,271 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-06 07:26:47,650 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 07:26:51,496 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c0fb9fe20>)
+[gpub011:0/16] 2024-02-06 07:26:51,496 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-06 07:26:51,499 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 07:32:39,413 (trainer:762) INFO: 30epoch:train:8701-8800batch: iter_time=3.275, forward_time=0.353, loss_ctc=47.302, loss_att=45.571, acc=0.773, loss=46.090, backward_time=0.308, grad_norm=38.102, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.839e-04, train_time=4.487
+[gpub011:0/16] 2024-02-06 07:34:44,041 (trainer:762) INFO: 30epoch:train:8801-8900batch: iter_time=8.063e-05, forward_time=0.289, loss_ctc=45.041, loss_att=47.627, acc=0.765, loss=46.851, backward_time=0.295, grad_norm=39.607, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.839e-04, train_time=1.245
+[gpub011:0/16] 2024-02-06 07:37:19,216 (trainer:762) INFO: 30epoch:train:8901-9000batch: iter_time=7.799e-05, forward_time=0.453, loss_ctc=49.031, loss_att=45.463, acc=0.754, loss=46.533, backward_time=0.346, grad_norm=38.242, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.838e-04, train_time=1.552
+[gpub011:0/16] 2024-02-06 07:39:28,993 (trainer:762) INFO: 30epoch:train:9001-9100batch: iter_time=7.932e-05, forward_time=0.286, loss_ctc=42.405, loss_att=39.147, acc=0.772, loss=40.125, backward_time=0.294, grad_norm=39.415, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.838e-04, train_time=1.297
+[gpub011:0/16] 2024-02-06 07:42:11,795 (trainer:762) INFO: 30epoch:train:9101-9200batch: iter_time=8.100e-05, forward_time=0.361, loss_ctc=43.254, loss_att=44.172, acc=0.748, loss=43.896, backward_time=0.406, grad_norm=38.219, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.838e-04, train_time=1.628
+[gpub011:0/16] 2024-02-06 07:44:27,011 (trainer:762) INFO: 30epoch:train:9201-9300batch: iter_time=7.645e-05, forward_time=0.290, loss_ctc=47.666, loss_att=45.386, acc=0.773, loss=46.070, backward_time=0.297, grad_norm=37.564, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.838e-04, train_time=1.352
+[gpub011:0/16] 2024-02-06 07:46:32,609 (trainer:762) INFO: 30epoch:train:9301-9400batch: iter_time=7.747e-05, forward_time=0.290, loss_ctc=49.058, loss_att=41.628, acc=0.762, loss=43.857, backward_time=0.296, grad_norm=38.055, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.838e-04, train_time=1.256
+[gpub011:0/16] 2024-02-06 07:49:17,417 (trainer:762) INFO: 30epoch:train:9401-9500batch: iter_time=7.874e-05, forward_time=0.354, loss_ctc=49.742, loss_att=46.587, acc=0.747, loss=47.533, backward_time=0.390, grad_norm=41.639, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.837e-04, train_time=1.648
+[gpub011:0/16] 2024-02-06 07:51:37,825 (trainer:762) INFO: 30epoch:train:9501-9600batch: iter_time=7.885e-05, forward_time=0.421, loss_ctc=44.914, loss_att=44.061, acc=0.763, loss=44.317, backward_time=0.343, grad_norm=42.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.837e-04, train_time=1.404
+[gpub011:0/16] 2024-02-06 07:53:52,553 (trainer:762) INFO: 30epoch:train:9601-9700batch: iter_time=7.728e-05, forward_time=0.292, loss_ctc=53.648, loss_att=52.689, acc=0.750, loss=52.977, backward_time=0.300, grad_norm=44.773, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.837e-04, train_time=1.347
+[gpub011:0/16] 2024-02-06 07:56:28,219 (trainer:762) INFO: 30epoch:train:9701-9800batch: iter_time=8.140e-05, forward_time=0.444, loss_ctc=47.828, loss_att=44.095, acc=0.753, loss=45.215, backward_time=0.321, grad_norm=37.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.837e-04, train_time=1.556
+[gpub011:0/16] 2024-02-06 07:58:46,830 (trainer:762) INFO: 30epoch:train:9801-9900batch: iter_time=7.915e-05, forward_time=0.311, loss_ctc=47.387, loss_att=44.478, acc=0.766, loss=45.351, backward_time=0.296, grad_norm=40.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.836e-04, train_time=1.386
+[gpub011:0/16] 2024-02-06 08:00:58,372 (trainer:762) INFO: 30epoch:train:9901-10000batch: iter_time=7.119e-05, forward_time=0.290, loss_ctc=46.607, loss_att=42.350, acc=0.766, loss=43.627, backward_time=0.296, grad_norm=40.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.836e-04, train_time=1.315
+[gpub011:0/16] 2024-02-06 08:01:18,401 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-06 08:01:38,003 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 08:01:41,587 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78b79e5360>)
+[gpub011:0/16] 2024-02-06 08:01:41,587 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-06 08:01:41,660 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 08:08:52,290 (trainer:762) INFO: 30epoch:train:10001-10100batch: iter_time=3.566, forward_time=0.346, loss_ctc=46.261, loss_att=43.444, acc=0.770, loss=44.289, backward_time=0.310, grad_norm=35.098, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.836e-04, train_time=4.739
+[gpub011:0/16] 2024-02-06 08:10:47,814 (trainer:762) INFO: 30epoch:train:10101-10200batch: iter_time=8.408e-05, forward_time=0.291, loss_ctc=48.657, loss_att=49.102, acc=0.771, loss=48.968, backward_time=0.299, grad_norm=41.939, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.836e-04, train_time=1.155
+[gpub011:0/16] 2024-02-06 08:13:49,267 (trainer:762) INFO: 30epoch:train:10201-10300batch: iter_time=1.938e-04, forward_time=0.373, loss_ctc=45.200, loss_att=40.540, acc=0.756, loss=41.938, backward_time=0.378, grad_norm=36.038, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.836e-04, train_time=1.814
+[gpub011:0/16] 2024-02-06 08:15:41,175 (trainer:762) INFO: 30epoch:train:10301-10400batch: iter_time=8.442e-05, forward_time=0.287, loss_ctc=41.593, loss_att=40.448, acc=0.762, loss=40.792, backward_time=0.295, grad_norm=37.222, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.835e-04, train_time=1.119
+[gpub011:0/16] 2024-02-06 08:18:05,936 (trainer:762) INFO: 30epoch:train:10401-10500batch: iter_time=8.111e-05, forward_time=0.369, loss_ctc=44.443, loss_att=45.209, acc=0.767, loss=44.979, backward_time=0.355, grad_norm=35.103, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.835e-04, train_time=1.447
+[gpub011:0/16] 2024-02-06 08:20:35,909 (trainer:762) INFO: 30epoch:train:10501-10600batch: iter_time=8.540e-05, forward_time=0.291, loss_ctc=50.884, loss_att=43.510, acc=0.765, loss=45.722, backward_time=0.298, grad_norm=37.051, clip=100.000, loss_scale=9.035e+33, optim_step_time=0.094, optim0_lr0=1.835e-04, train_time=1.500
+[gpub011:0/16] 2024-02-06 08:23:05,643 (trainer:762) INFO: 30epoch:train:10601-10700batch: iter_time=8.760e-05, forward_time=0.427, loss_ctc=50.061, loss_att=43.408, acc=0.749, loss=45.404, backward_time=0.344, grad_norm=40.025, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.835e-04, train_time=1.497
+[gpub011:0/16] 2024-02-06 08:25:00,148 (trainer:762) INFO: 30epoch:train:10701-10800batch: iter_time=8.627e-05, forward_time=0.290, loss_ctc=48.279, loss_att=47.027, acc=0.753, loss=47.402, backward_time=0.299, grad_norm=45.478, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.835e-04, train_time=1.145
+[gpub011:0/16] 2024-02-06 08:27:45,981 (trainer:762) INFO: 30epoch:train:10801-10900batch: iter_time=8.102e-05, forward_time=0.428, loss_ctc=44.787, loss_att=43.048, acc=0.774, loss=43.570, backward_time=0.334, grad_norm=35.129, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.834e-04, train_time=1.658
+[gpub011:0/16] 2024-02-06 08:29:44,315 (trainer:762) INFO: 30epoch:train:10901-11000batch: iter_time=8.418e-05, forward_time=0.291, loss_ctc=52.867, loss_att=51.868, acc=0.747, loss=52.168, backward_time=0.298, grad_norm=44.692, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.834e-04, train_time=1.183
+[gpub011:0/16] 2024-02-06 08:31:56,119 (trainer:762) INFO: 30epoch:train:11001-11100batch: iter_time=8.510e-05, forward_time=0.463, loss_ctc=46.733, loss_att=45.717, acc=0.738, loss=46.022, backward_time=0.322, grad_norm=41.271, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.834e-04, train_time=1.318
+[gpub011:0/16] 2024-02-06 08:32:12,745 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 08:34:34,474 (trainer:762) INFO: 30epoch:train:11101-11200batch: iter_time=8.498e-05, forward_time=0.291, loss_ctc=46.477, loss_att=41.295, acc=0.765, loss=42.850, backward_time=0.295, grad_norm=39.574, clip=100.000, loss_scale=5.612e+33, optim_step_time=0.094, optim0_lr0=1.834e-04, train_time=1.583
+[gpub011:0/16] 2024-02-06 08:36:13,016 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-06 08:36:32,621 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 08:36:36,239 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c0fb48610>)
+[gpub011:0/16] 2024-02-06 08:36:36,239 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-06 08:36:36,243 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 08:42:50,077 (trainer:762) INFO: 30epoch:train:11201-11300batch: iter_time=3.572, forward_time=0.456, loss_ctc=46.910, loss_att=44.825, acc=0.774, loss=45.450, backward_time=0.334, grad_norm=38.567, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.834e-04, train_time=4.956
+[gpub011:0/16] 2024-02-06 08:45:08,696 (trainer:762) INFO: 30epoch:train:11301-11400batch: iter_time=8.303e-05, forward_time=0.289, loss_ctc=44.633, loss_att=46.409, acc=0.769, loss=45.876, backward_time=0.296, grad_norm=38.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.833e-04, train_time=1.386
+[gpub011:0/16] 2024-02-06 08:47:09,230 (trainer:762) INFO: 30epoch:train:11401-11500batch: iter_time=8.644e-05, forward_time=0.291, loss_ctc=49.229, loss_att=44.971, acc=0.753, loss=46.248, backward_time=0.298, grad_norm=38.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.833e-04, train_time=1.205
+[gpub011:0/16] 2024-02-06 08:49:21,952 (trainer:762) INFO: 30epoch:train:11501-11600batch: iter_time=9.937e-05, forward_time=0.438, loss_ctc=42.190, loss_att=38.898, acc=0.772, loss=39.885, backward_time=0.342, grad_norm=36.044, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.833e-04, train_time=1.327
+[gpub011:0/16] 2024-02-06 08:51:52,658 (trainer:762) INFO: 30epoch:train:11601-11700batch: iter_time=8.134e-05, forward_time=0.290, loss_ctc=43.587, loss_att=43.181, acc=0.753, loss=43.303, backward_time=0.295, grad_norm=37.535, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.833e-04, train_time=1.507
+[gpub011:0/16] 2024-02-06 08:53:45,924 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 08:54:14,601 (trainer:762) INFO: 30epoch:train:11701-11800batch: iter_time=8.462e-05, forward_time=0.356, loss_ctc=47.291, loss_att=44.845, acc=0.772, loss=45.579, backward_time=0.405, grad_norm=36.746, clip=100.000, loss_scale=4.642e+33, optim_step_time=0.099, optim0_lr0=1.833e-04, train_time=1.419
+[gpub011:0/16] 2024-02-06 08:56:02,953 (trainer:762) INFO: 30epoch:train:11801-11900batch: iter_time=9.416e-05, forward_time=0.288, loss_ctc=48.590, loss_att=41.078, acc=0.762, loss=43.331, backward_time=0.297, grad_norm=39.971, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.832e-04, train_time=1.084
+[gpub011:0/16] 2024-02-06 08:58:47,707 (trainer:762) INFO: 30epoch:train:11901-12000batch: iter_time=8.421e-05, forward_time=0.318, loss_ctc=50.007, loss_att=46.376, acc=0.749, loss=47.465, backward_time=0.296, grad_norm=42.574, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.832e-04, train_time=1.647
+[gpub011:0/16] 2024-02-06 09:01:04,936 (trainer:762) INFO: 30epoch:train:12001-12100batch: iter_time=8.420e-05, forward_time=0.384, loss_ctc=44.494, loss_att=43.731, acc=0.762, loss=43.960, backward_time=0.344, grad_norm=41.412, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=1.832e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 09:03:09,155 (trainer:762) INFO: 30epoch:train:12101-12200batch: iter_time=9.161e-05, forward_time=0.292, loss_ctc=53.311, loss_att=52.262, acc=0.750, loss=52.577, backward_time=0.301, grad_norm=42.621, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.832e-04, train_time=1.243
+[gpub011:0/16] 2024-02-06 09:05:58,150 (trainer:762) INFO: 30epoch:train:12201-12300batch: iter_time=9.327e-05, forward_time=0.349, loss_ctc=47.220, loss_att=43.577, acc=0.755, loss=44.670, backward_time=0.425, grad_norm=36.672, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.832e-04, train_time=1.690
+[gpub011:0/16] 2024-02-06 09:08:02,278 (trainer:762) INFO: 30epoch:train:12301-12400batch: iter_time=9.010e-05, forward_time=0.289, loss_ctc=47.205, loss_att=44.854, acc=0.764, loss=45.559, backward_time=0.298, grad_norm=40.342, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.831e-04, train_time=1.241
+[gpub011:0/16] 2024-02-06 09:10:32,450 (trainer:762) INFO: 30epoch:train:12401-12500batch: iter_time=9.451e-05, forward_time=0.291, loss_ctc=45.825, loss_att=42.281, acc=0.765, loss=43.344, backward_time=0.296, grad_norm=39.588, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.831e-04, train_time=1.502
+[gpub011:0/16] 2024-02-06 09:10:52,478 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-06 09:11:12,188 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 09:11:15,766 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f790771c2b0>)
+[gpub011:0/16] 2024-02-06 09:11:15,766 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-06 09:11:15,794 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 09:18:50,744 (trainer:762) INFO: 30epoch:train:12501-12600batch: iter_time=3.720, forward_time=0.407, loss_ctc=46.288, loss_att=43.452, acc=0.770, loss=44.303, backward_time=0.319, grad_norm=34.888, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.831e-04, train_time=4.983
+[gpub011:0/16] 2024-02-06 09:21:16,704 (trainer:762) INFO: 30epoch:train:12601-12700batch: iter_time=8.097e-05, forward_time=0.290, loss_ctc=48.191, loss_att=47.907, acc=0.773, loss=47.992, backward_time=0.296, grad_norm=38.458, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.831e-04, train_time=1.459
+[gpub011:0/16] 2024-02-06 09:23:52,399 (trainer:762) INFO: 30epoch:train:12701-12800batch: iter_time=0.001, forward_time=0.449, loss_ctc=45.300, loss_att=40.562, acc=0.757, loss=41.984, backward_time=0.322, grad_norm=36.762, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.831e-04, train_time=1.557
+[gpub011:0/16] 2024-02-06 09:26:03,544 (trainer:762) INFO: 30epoch:train:12801-12900batch: iter_time=8.403e-05, forward_time=0.289, loss_ctc=41.518, loss_att=41.148, acc=0.760, loss=41.259, backward_time=0.294, grad_norm=37.903, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.830e-04, train_time=1.311
+[gpub011:0/16] 2024-02-06 09:28:25,209 (trainer:762) INFO: 30epoch:train:12901-13000batch: iter_time=8.690e-05, forward_time=0.290, loss_ctc=43.781, loss_att=45.172, acc=0.767, loss=44.755, backward_time=0.295, grad_norm=34.972, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.830e-04, train_time=1.416
+[gpub011:0/16] 2024-02-06 09:31:14,194 (trainer:762) INFO: 30epoch:train:13001-13100batch: iter_time=8.599e-05, forward_time=0.515, loss_ctc=49.911, loss_att=42.775, acc=0.766, loss=44.916, backward_time=0.353, grad_norm=38.295, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.830e-04, train_time=1.690
+[gpub011:0/16] 2024-02-06 09:33:25,267 (trainer:762) INFO: 30epoch:train:13101-13200batch: iter_time=9.040e-05, forward_time=0.289, loss_ctc=50.433, loss_att=43.396, acc=0.752, loss=45.507, backward_time=0.295, grad_norm=40.951, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.830e-04, train_time=1.311
+[gpub011:0/16] 2024-02-06 09:35:55,716 (trainer:762) INFO: 30epoch:train:13201-13300batch: iter_time=8.762e-05, forward_time=0.290, loss_ctc=48.077, loss_att=46.210, acc=0.755, loss=46.770, backward_time=0.295, grad_norm=42.455, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.829e-04, train_time=1.504
+[gpub011:0/16] 2024-02-06 09:38:23,692 (trainer:762) INFO: 30epoch:train:13301-13400batch: iter_time=8.582e-05, forward_time=0.291, loss_ctc=44.852, loss_att=42.920, acc=0.776, loss=43.500, backward_time=0.298, grad_norm=36.613, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.829e-04, train_time=1.480
+[gpub011:0/16] 2024-02-06 09:40:53,463 (trainer:762) INFO: 30epoch:train:13401-13500batch: iter_time=8.004e-05, forward_time=0.333, loss_ctc=52.710, loss_att=51.665, acc=0.750, loss=51.978, backward_time=0.372, grad_norm=44.767, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.829e-04, train_time=1.497
+[gpub011:0/16] 2024-02-06 09:43:06,272 (trainer:762) INFO: 30epoch:train:13501-13600batch: iter_time=8.352e-05, forward_time=0.290, loss_ctc=46.157, loss_att=44.471, acc=0.744, loss=44.977, backward_time=0.295, grad_norm=37.442, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.829e-04, train_time=1.328
+[gpub011:0/16] 2024-02-06 09:45:13,388 (trainer:762) INFO: 30epoch:train:13601-13700batch: iter_time=8.544e-05, forward_time=0.289, loss_ctc=46.340, loss_att=40.871, acc=0.768, loss=42.511, backward_time=0.296, grad_norm=38.366, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.829e-04, train_time=1.271
+[gpub011:0/16] 2024-02-06 09:47:01,124 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-06 09:47:20,385 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 09:47:23,942 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7bbcf34880>)
+[gpub011:0/16] 2024-02-06 09:47:23,942 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-06 09:47:23,946 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 09:54:09,592 (trainer:762) INFO: 30epoch:train:13701-13800batch: iter_time=3.719, forward_time=0.444, loss_ctc=46.813, loss_att=45.907, acc=0.776, loss=46.179, backward_time=0.330, grad_norm=37.749, clip=100.000, loss_scale=3.141e+33, optim_step_time=0.097, optim0_lr0=1.828e-04, train_time=5.361
+[gpub011:0/16] 2024-02-06 09:56:32,396 (trainer:762) INFO: 30epoch:train:13801-13900batch: iter_time=7.751e-05, forward_time=0.289, loss_ctc=44.759, loss_att=48.390, acc=0.767, loss=47.301, backward_time=0.295, grad_norm=44.080, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.828e-04, train_time=1.429
+[gpub011:0/16] 2024-02-06 09:58:52,470 (trainer:762) INFO: 30epoch:train:13901-14000batch: iter_time=8.165e-05, forward_time=0.313, loss_ctc=48.657, loss_att=47.070, acc=0.764, loss=47.546, backward_time=0.302, grad_norm=37.648, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.828e-04, train_time=1.401
+[gpub011:0/16] 2024-02-06 10:01:44,119 (trainer:762) INFO: 30epoch:train:14001-14100batch: iter_time=8.176e-05, forward_time=0.363, loss_ctc=42.165, loss_att=38.718, acc=0.774, loss=39.752, backward_time=0.341, grad_norm=36.704, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.828e-04, train_time=1.716
+[gpub011:0/16] 2024-02-06 10:03:48,929 (trainer:762) INFO: 30epoch:train:14101-14200batch: iter_time=7.932e-05, forward_time=0.295, loss_ctc=43.125, loss_att=44.285, acc=0.756, loss=43.937, backward_time=0.296, grad_norm=36.831, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.828e-04, train_time=1.248
+[gpub011:0/16] 2024-02-06 10:06:35,762 (trainer:762) INFO: 30epoch:train:14201-14300batch: iter_time=8.344e-05, forward_time=0.414, loss_ctc=47.178, loss_att=45.795, acc=0.777, loss=46.210, backward_time=0.327, grad_norm=36.777, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.827e-04, train_time=1.668
+[gpub011:0/16] 2024-02-06 10:08:52,006 (trainer:762) INFO: 30epoch:train:14301-14400batch: iter_time=8.587e-05, forward_time=0.290, loss_ctc=48.274, loss_att=41.661, acc=0.768, loss=43.645, backward_time=0.295, grad_norm=37.458, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.827e-04, train_time=1.362
+[gpub011:0/16] 2024-02-06 10:11:20,294 (trainer:762) INFO: 30epoch:train:14401-14500batch: iter_time=8.385e-05, forward_time=0.293, loss_ctc=49.750, loss_att=47.063, acc=0.753, loss=47.869, backward_time=0.299, grad_norm=41.863, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.827e-04, train_time=1.483
+[gpub011:0/16] 2024-02-06 10:14:02,851 (trainer:762) INFO: 30epoch:train:14501-14600batch: iter_time=8.403e-05, forward_time=0.417, loss_ctc=44.785, loss_att=44.067, acc=0.771, loss=44.282, backward_time=0.369, grad_norm=38.945, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.827e-04, train_time=1.625
+[gpub011:0/16] 2024-02-06 10:16:26,634 (trainer:762) INFO: 30epoch:train:14601-14700batch: iter_time=8.350e-05, forward_time=0.293, loss_ctc=53.148, loss_att=52.790, acc=0.757, loss=52.897, backward_time=0.300, grad_norm=40.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.827e-04, train_time=1.437
+[gpub011:0/16] 2024-02-06 10:18:59,773 (trainer:762) INFO: 30epoch:train:14701-14800batch: iter_time=0.003, forward_time=0.405, loss_ctc=46.840, loss_att=43.049, acc=0.760, loss=44.186, backward_time=0.349, grad_norm=36.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.826e-04, train_time=1.532
+[gpub011:0/16] 2024-02-06 10:21:32,812 (trainer:762) INFO: 30epoch:train:14801-14900batch: iter_time=8.412e-05, forward_time=0.290, loss_ctc=47.045, loss_att=45.376, acc=0.773, loss=45.877, backward_time=0.296, grad_norm=37.944, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.826e-04, train_time=1.530
+[gpub011:0/16] 2024-02-06 10:23:47,137 (trainer:762) INFO: 30epoch:train:14901-15000batch: iter_time=8.294e-05, forward_time=0.290, loss_ctc=45.733, loss_att=42.670, acc=0.770, loss=43.589, backward_time=0.296, grad_norm=38.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.826e-04, train_time=1.342
+[gpub011:0/16] 2024-02-06 11:03:10,452 (trainer:361) INFO: 30epoch results: [train] iter_time=0.284, forward_time=0.333, loss_ctc=47.645, loss_att=45.083, acc=0.761, loss=45.851, backward_time=0.317, grad_norm=39.317, clip=100.000, loss_scale=6.153e+33, optim_step_time=0.097, optim0_lr0=1.841e-04, train_time=1.686, time=7 hours, 1 minute and 52.35 seconds, total_count=480000, gpu_max_cached_mem_GB=40.000, [valid] loss_ctc=36.219, cer_ctc=0.191, loss_att=37.548, acc=0.675, cer=0.339, wer=0.999, loss=37.149, time=38 minutes and 58.39 seconds, total_count=149472, gpu_max_cached_mem_GB=40.000
+[gpub011:0/16] 2024-02-06 11:03:20,048 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub011:0/16] 2024-02-06 11:03:20,134 (trainer:290) INFO: 31/45epoch started. Estimated time to finish: 4 days, 21 hours and 6 minutes
+[gpub011:0/16] 2024-02-06 11:03:20,144 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-06 11:03:39,110 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 11:03:42,661 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f723101cd90>)
+[gpub011:0/16] 2024-02-06 11:03:42,661 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-06 11:03:42,664 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 11:11:08,489 (trainer:762) INFO: 31epoch:train:1-100batch: iter_time=3.486, forward_time=0.319, loss_ctc=39.512, loss_att=37.717, acc=0.738, loss=38.255, backward_time=0.304, grad_norm=39.708, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.826e-04, train_time=4.683
+[gpub011:0/16] 2024-02-06 11:13:22,318 (trainer:762) INFO: 31epoch:train:101-200batch: iter_time=6.913e-04, forward_time=0.340, loss_ctc=52.914, loss_att=44.821, acc=0.747, loss=47.249, backward_time=0.315, grad_norm=41.858, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.826e-04, train_time=1.338
+[gpub011:0/16] 2024-02-06 11:16:07,776 (trainer:762) INFO: 31epoch:train:201-300batch: iter_time=7.880e-05, forward_time=0.355, loss_ctc=48.386, loss_att=49.342, acc=0.739, loss=49.055, backward_time=0.331, grad_norm=40.974, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.825e-04, train_time=1.654
+[gpub011:0/16] 2024-02-06 11:18:36,956 (trainer:762) INFO: 31epoch:train:301-400batch: iter_time=7.726e-05, forward_time=0.333, loss_ctc=51.856, loss_att=48.773, acc=0.739, loss=49.698, backward_time=0.305, grad_norm=44.899, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.825e-04, train_time=1.492
+[gpub011:0/16] 2024-02-06 11:21:29,206 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 11:22:42,316 (trainer:762) INFO: 31epoch:train:401-500batch: iter_time=8.026e-05, forward_time=0.460, loss_ctc=50.585, loss_att=53.152, acc=0.744, loss=52.382, backward_time=0.585, grad_norm=41.523, clip=100.000, loss_scale=4.353e+33, optim_step_time=0.118, optim0_lr0=1.825e-04, train_time=2.453
+[gpub011:0/16] 2024-02-06 11:25:01,997 (trainer:762) INFO: 31epoch:train:501-600batch: iter_time=8.161e-05, forward_time=0.305, loss_ctc=47.830, loss_att=43.851, acc=0.759, loss=45.044, backward_time=0.320, grad_norm=42.656, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.825e-04, train_time=1.396
+[gpub011:0/16] 2024-02-06 11:27:22,659 (trainer:762) INFO: 31epoch:train:601-700batch: iter_time=7.906e-05, forward_time=0.398, loss_ctc=56.433, loss_att=45.007, acc=0.754, loss=48.435, backward_time=0.303, grad_norm=53.122, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.825e-04, train_time=1.407
+[gpub011:0/16] 2024-02-06 11:30:49,375 (trainer:762) INFO: 31epoch:train:701-800batch: iter_time=0.090, forward_time=0.497, loss_ctc=44.310, loss_att=36.781, acc=0.759, loss=39.040, backward_time=0.449, grad_norm=38.782, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.824e-04, train_time=2.062
+[gpub011:0/16] 2024-02-06 11:33:10,673 (trainer:762) INFO: 31epoch:train:801-900batch: iter_time=8.977e-05, forward_time=0.287, loss_ctc=46.573, loss_att=37.135, acc=0.776, loss=39.966, backward_time=0.295, grad_norm=39.919, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.824e-04, train_time=1.417
+[gpub011:0/16] 2024-02-06 11:35:39,746 (trainer:762) INFO: 31epoch:train:901-1000batch: iter_time=8.609e-05, forward_time=0.432, loss_ctc=43.272, loss_att=40.438, acc=0.748, loss=41.288, backward_time=0.339, grad_norm=37.657, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.128, optim0_lr0=1.824e-04, train_time=1.491
+[gpub011:0/16] 2024-02-06 11:38:28,274 (trainer:762) INFO: 31epoch:train:1001-1100batch: iter_time=8.430e-05, forward_time=0.464, loss_ctc=52.390, loss_att=58.160, acc=0.737, loss=56.429, backward_time=0.324, grad_norm=43.495, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=1.824e-04, train_time=1.685
+[gpub011:0/16] 2024-02-06 11:40:38,631 (trainer:762) INFO: 31epoch:train:1101-1200batch: iter_time=7.925e-05, forward_time=0.289, loss_ctc=57.569, loss_att=54.934, acc=0.729, loss=55.724, backward_time=0.297, grad_norm=49.960, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.824e-04, train_time=1.303
+[gpub011:0/16] 2024-02-06 11:41:54,607 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-06 11:42:13,955 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 11:42:17,502 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b75fa7d60>)
+[gpub011:0/16] 2024-02-06 11:42:17,502 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-06 11:42:17,505 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 11:50:27,503 (trainer:762) INFO: 31epoch:train:1201-1300batch: iter_time=3.924, forward_time=0.302, loss_ctc=42.638, loss_att=39.397, acc=0.746, loss=40.369, backward_time=0.353, grad_norm=37.864, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.130, optim0_lr0=1.823e-04, train_time=5.888
+[gpub011:0/16] 2024-02-06 11:52:52,842 (trainer:762) INFO: 31epoch:train:1301-1400batch: iter_time=8.303e-05, forward_time=0.407, loss_ctc=51.527, loss_att=45.502, acc=0.737, loss=47.309, backward_time=0.334, grad_norm=44.855, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.823e-04, train_time=1.453
+[gpub011:0/16] 2024-02-06 11:55:23,226 (trainer:762) INFO: 31epoch:train:1401-1500batch: iter_time=7.968e-05, forward_time=0.287, loss_ctc=47.448, loss_att=44.191, acc=0.745, loss=45.168, backward_time=0.295, grad_norm=39.249, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.823e-04, train_time=1.504
+[gpub011:0/16] 2024-02-06 11:57:20,977 (trainer:762) INFO: 31epoch:train:1501-1600batch: iter_time=8.550e-05, forward_time=0.302, loss_ctc=40.200, loss_att=38.728, acc=0.757, loss=39.169, backward_time=0.302, grad_norm=36.080, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.823e-04, train_time=1.177
+[gpub011:0/16] 2024-02-06 12:00:06,559 (trainer:762) INFO: 31epoch:train:1601-1700batch: iter_time=8.749e-05, forward_time=0.423, loss_ctc=55.831, loss_att=52.818, acc=0.751, loss=53.722, backward_time=0.340, grad_norm=44.642, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.109, optim0_lr0=1.823e-04, train_time=1.655
+[gpub011:0/16] 2024-02-06 12:02:13,840 (trainer:762) INFO: 31epoch:train:1701-1800batch: iter_time=8.688e-05, forward_time=0.290, loss_ctc=52.581, loss_att=50.601, acc=0.748, loss=51.195, backward_time=0.296, grad_norm=43.687, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.822e-04, train_time=1.273
+[gpub011:0/16] 2024-02-06 12:04:13,080 (trainer:762) INFO: 31epoch:train:1801-1900batch: iter_time=9.184e-05, forward_time=0.299, loss_ctc=50.098, loss_att=47.387, acc=0.747, loss=48.200, backward_time=0.317, grad_norm=45.493, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.822e-04, train_time=1.192
+[gpub011:0/16] 2024-02-06 12:07:00,135 (trainer:762) INFO: 31epoch:train:1901-2000batch: iter_time=2.059e-04, forward_time=0.392, loss_ctc=50.099, loss_att=40.247, acc=0.763, loss=43.202, backward_time=0.311, grad_norm=38.524, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.116, optim0_lr0=1.822e-04, train_time=1.670
+[gpub011:0/16] 2024-02-06 12:09:07,998 (trainer:762) INFO: 31epoch:train:2001-2100batch: iter_time=7.955e-05, forward_time=0.289, loss_ctc=46.396, loss_att=35.528, acc=0.776, loss=38.788, backward_time=0.294, grad_norm=42.479, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.822e-04, train_time=1.278
+[gpub011:0/16] 2024-02-06 12:11:15,023 (trainer:762) INFO: 31epoch:train:2101-2200batch: iter_time=8.555e-05, forward_time=0.291, loss_ctc=42.409, loss_att=40.230, acc=0.754, loss=40.884, backward_time=0.298, grad_norm=37.564, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.822e-04, train_time=1.270
+[gpub011:0/16] 2024-02-06 12:13:25,355 (trainer:762) INFO: 31epoch:train:2201-2300batch: iter_time=8.318e-05, forward_time=0.319, loss_ctc=44.402, loss_att=45.404, acc=0.757, loss=45.103, backward_time=0.308, grad_norm=36.231, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.821e-04, train_time=1.303
+[gpub011:0/16] 2024-02-06 12:16:13,400 (trainer:762) INFO: 31epoch:train:2301-2400batch: iter_time=8.104e-05, forward_time=0.388, loss_ctc=49.009, loss_att=53.915, acc=0.734, loss=52.443, backward_time=0.307, grad_norm=43.598, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=1.821e-04, train_time=1.680
+[gpub011:0/16] 2024-02-06 12:18:07,121 (trainer:762) INFO: 31epoch:train:2401-2500batch: iter_time=8.248e-05, forward_time=0.290, loss_ctc=58.451, loss_att=53.397, acc=0.732, loss=54.913, backward_time=0.297, grad_norm=53.613, clip=100.000, loss_scale=3.427e+33, optim_step_time=0.094, optim0_lr0=1.821e-04, train_time=1.137
+[gpub011:0/16] 2024-02-06 12:18:27,173 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-06 12:18:46,057 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 12:18:49,574 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71d6ba75e0>)
+[gpub011:0/16] 2024-02-06 12:18:49,574 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-06 12:18:49,615 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 12:26:45,113 (trainer:762) INFO: 31epoch:train:2501-2600batch: iter_time=3.843, forward_time=0.304, loss_ctc=37.998, loss_att=36.672, acc=0.755, loss=37.069, backward_time=0.299, grad_norm=39.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.821e-04, train_time=5.180
+[gpub011:0/16] 2024-02-06 12:29:18,941 (trainer:762) INFO: 31epoch:train:2601-2700batch: iter_time=8.197e-05, forward_time=0.367, loss_ctc=50.420, loss_att=44.466, acc=0.762, loss=46.253, backward_time=0.326, grad_norm=40.981, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.821e-04, train_time=1.538
+[gpub011:0/16] 2024-02-06 12:31:31,333 (trainer:762) INFO: 31epoch:train:2701-2800batch: iter_time=8.282e-05, forward_time=0.287, loss_ctc=46.886, loss_att=49.319, acc=0.748, loss=48.589, backward_time=0.296, grad_norm=40.425, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.820e-04, train_time=1.324
+[gpub011:0/16] 2024-02-06 12:33:40,674 (trainer:762) INFO: 31epoch:train:2801-2900batch: iter_time=8.199e-05, forward_time=0.305, loss_ctc=50.692, loss_att=48.629, acc=0.749, loss=49.248, backward_time=0.304, grad_norm=43.297, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.820e-04, train_time=1.293
+[gpub011:0/16] 2024-02-06 12:36:09,694 (trainer:762) INFO: 31epoch:train:2901-3000batch: iter_time=8.074e-05, forward_time=0.406, loss_ctc=49.621, loss_att=55.127, acc=0.751, loss=53.475, backward_time=0.321, grad_norm=41.805, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.820e-04, train_time=1.490
+[gpub011:0/16] 2024-02-06 12:38:36,647 (trainer:762) INFO: 31epoch:train:3001-3100batch: iter_time=8.013e-05, forward_time=0.320, loss_ctc=46.983, loss_att=44.247, acc=0.769, loss=45.067, backward_time=0.294, grad_norm=40.887, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.820e-04, train_time=1.469
+[gpub011:0/16] 2024-02-06 12:40:39,739 (trainer:762) INFO: 31epoch:train:3101-3200batch: iter_time=8.205e-05, forward_time=0.305, loss_ctc=54.334, loss_att=46.808, acc=0.760, loss=49.066, backward_time=0.310, grad_norm=51.067, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.820e-04, train_time=1.231
+[gpub011:0/16] 2024-02-06 12:43:12,847 (trainer:762) INFO: 31epoch:train:3201-3300batch: iter_time=8.805e-05, forward_time=0.286, loss_ctc=43.182, loss_att=35.815, acc=0.767, loss=38.025, backward_time=0.292, grad_norm=36.832, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.819e-04, train_time=1.531
+[gpub011:0/16] 2024-02-06 12:45:31,275 (trainer:762) INFO: 31epoch:train:3301-3400batch: iter_time=8.395e-05, forward_time=0.454, loss_ctc=45.130, loss_att=36.438, acc=0.783, loss=39.046, backward_time=0.314, grad_norm=38.785, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.819e-04, train_time=1.384
+[gpub011:0/16] 2024-02-06 12:48:00,199 (trainer:762) INFO: 31epoch:train:3401-3500batch: iter_time=8.848e-05, forward_time=0.298, loss_ctc=42.263, loss_att=39.202, acc=0.758, loss=40.120, backward_time=0.305, grad_norm=35.833, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.819e-04, train_time=1.489
+[gpub011:0/16] 2024-02-06 12:50:19,337 (trainer:762) INFO: 31epoch:train:3501-3600batch: iter_time=8.466e-05, forward_time=0.294, loss_ctc=51.069, loss_att=61.326, acc=0.741, loss=58.249, backward_time=0.302, grad_norm=43.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.819e-04, train_time=1.391
+[gpub011:0/16] 2024-02-06 12:52:57,368 (trainer:762) INFO: 31epoch:train:3601-3700batch: iter_time=1.951e-04, forward_time=0.411, loss_ctc=53.764, loss_att=53.576, acc=0.741, loss=53.632, backward_time=0.323, grad_norm=46.902, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.819e-04, train_time=1.580
+[gpub011:0/16] 2024-02-06 12:54:15,778 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-06 12:54:35,317 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 12:54:38,886 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78c78c6320>)
+[gpub011:0/16] 2024-02-06 12:54:38,886 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-06 12:54:38,889 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 13:01:03,376 (trainer:762) INFO: 31epoch:train:3701-3800batch: iter_time=3.581, forward_time=0.289, loss_ctc=41.835, loss_att=39.278, acc=0.752, loss=40.045, backward_time=0.295, grad_norm=36.674, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.818e-04, train_time=4.860
+[gpub011:0/16] 2024-02-06 13:03:17,756 (trainer:762) INFO: 31epoch:train:3801-3900batch: iter_time=7.806e-05, forward_time=0.299, loss_ctc=50.179, loss_att=46.308, acc=0.737, loss=47.469, backward_time=0.304, grad_norm=42.915, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.818e-04, train_time=1.343
+[gpub011:0/16] 2024-02-06 13:05:39,033 (trainer:762) INFO: 31epoch:train:3901-4000batch: iter_time=7.585e-05, forward_time=0.467, loss_ctc=46.804, loss_att=44.258, acc=0.749, loss=45.022, backward_time=0.313, grad_norm=39.153, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.818e-04, train_time=1.413
+[gpub011:0/16] 2024-02-06 13:07:48,061 (trainer:762) INFO: 31epoch:train:4001-4100batch: iter_time=7.780e-05, forward_time=0.288, loss_ctc=39.902, loss_att=38.365, acc=0.759, loss=38.826, backward_time=0.294, grad_norm=35.260, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.818e-04, train_time=1.290
+[gpub011:0/16] 2024-02-06 13:10:05,192 (trainer:762) INFO: 31epoch:train:4101-4200batch: iter_time=7.816e-05, forward_time=0.305, loss_ctc=54.710, loss_att=52.727, acc=0.754, loss=53.322, backward_time=0.309, grad_norm=44.827, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.818e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 13:12:34,991 (trainer:762) INFO: 31epoch:train:4201-4300batch: iter_time=7.588e-05, forward_time=0.342, loss_ctc=51.905, loss_att=50.059, acc=0.754, loss=50.613, backward_time=0.371, grad_norm=43.494, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.817e-04, train_time=1.498
+[gpub011:0/16] 2024-02-06 13:15:08,531 (trainer:762) INFO: 31epoch:train:4301-4400batch: iter_time=7.575e-05, forward_time=0.287, loss_ctc=49.338, loss_att=47.444, acc=0.747, loss=48.012, backward_time=0.293, grad_norm=51.503, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.817e-04, train_time=1.535
+[gpub011:0/16] 2024-02-06 13:17:03,543 (trainer:762) INFO: 31epoch:train:4401-4500batch: iter_time=7.534e-05, forward_time=0.300, loss_ctc=49.239, loss_att=39.774, acc=0.765, loss=42.614, backward_time=0.316, grad_norm=38.903, clip=100.000, loss_scale=6.854e+33, optim_step_time=0.097, optim0_lr0=1.817e-04, train_time=1.150
+[gpub011:0/16] 2024-02-06 13:19:23,720 (trainer:762) INFO: 31epoch:train:4501-4600batch: iter_time=7.534e-05, forward_time=0.290, loss_ctc=45.852, loss_att=35.606, acc=0.777, loss=38.680, backward_time=0.294, grad_norm=40.232, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.817e-04, train_time=1.402
+[gpub011:0/16] 2024-02-06 13:22:01,339 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 13:22:01,347 (trainer:762) INFO: 31epoch:train:4601-4700batch: iter_time=1.726e-04, forward_time=0.397, loss_ctc=42.571, loss_att=40.453, acc=0.756, loss=41.089, backward_time=0.310, grad_norm=38.017, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=1.817e-04, train_time=1.576
+[gpub011:0/16] 2024-02-06 13:24:14,429 (trainer:762) INFO: 31epoch:train:4701-4800batch: iter_time=7.624e-05, forward_time=0.290, loss_ctc=44.398, loss_att=45.377, acc=0.760, loss=45.083, backward_time=0.294, grad_norm=36.673, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.816e-04, train_time=1.331
+[gpub011:0/16] 2024-02-06 13:26:08,216 (trainer:762) INFO: 31epoch:train:4801-4900batch: iter_time=7.709e-05, forward_time=0.300, loss_ctc=49.156, loss_att=53.761, acc=0.734, loss=52.380, backward_time=0.310, grad_norm=44.168, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.816e-04, train_time=1.138
+[gpub011:0/16] 2024-02-06 13:29:05,826 (trainer:762) INFO: 31epoch:train:4901-5000batch: iter_time=4.804e-04, forward_time=0.376, loss_ctc=57.858, loss_att=53.990, acc=0.734, loss=55.150, backward_time=0.330, grad_norm=53.003, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.816e-04, train_time=1.776
+[gpub011:0/16] 2024-02-06 13:29:25,855 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-06 13:29:45,272 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 13:29:49,117 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f79270673a0>)
+[gpub011:0/16] 2024-02-06 13:29:49,117 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-06 13:29:49,121 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 13:36:49,805 (trainer:762) INFO: 31epoch:train:5001-5100batch: iter_time=3.452, forward_time=0.299, loss_ctc=37.812, loss_att=35.812, acc=0.748, loss=36.412, backward_time=0.304, grad_norm=38.909, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.816e-04, train_time=4.640
+[gpub011:0/16] 2024-02-06 13:39:16,768 (trainer:762) INFO: 31epoch:train:5101-5200batch: iter_time=8.644e-05, forward_time=0.339, loss_ctc=50.369, loss_att=43.061, acc=0.757, loss=45.253, backward_time=0.350, grad_norm=38.523, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.816e-04, train_time=1.469
+[gpub011:0/16] 2024-02-06 13:41:37,305 (trainer:762) INFO: 31epoch:train:5201-5300batch: iter_time=7.890e-05, forward_time=0.297, loss_ctc=46.328, loss_att=47.805, acc=0.745, loss=47.362, backward_time=0.308, grad_norm=38.363, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.815e-04, train_time=1.405
+[gpub011:0/16] 2024-02-06 13:43:41,067 (trainer:762) INFO: 31epoch:train:5301-5400batch: iter_time=7.778e-05, forward_time=0.288, loss_ctc=49.678, loss_att=47.161, acc=0.747, loss=47.916, backward_time=0.295, grad_norm=40.806, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.815e-04, train_time=1.237
+[gpub011:0/16] 2024-02-06 13:46:09,213 (trainer:762) INFO: 31epoch:train:5401-5500batch: iter_time=4.080e-04, forward_time=0.412, loss_ctc=49.407, loss_att=52.205, acc=0.750, loss=51.366, backward_time=0.345, grad_norm=42.139, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.815e-04, train_time=1.481
+[gpub011:0/16] 2024-02-06 13:48:44,799 (trainer:762) INFO: 31epoch:train:5501-5600batch: iter_time=8.000e-05, forward_time=0.289, loss_ctc=46.311, loss_att=42.807, acc=0.766, loss=43.858, backward_time=0.294, grad_norm=40.917, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.815e-04, train_time=1.556
+[gpub011:0/16] 2024-02-06 13:51:01,571 (trainer:762) INFO: 31epoch:train:5601-5700batch: iter_time=8.645e-05, forward_time=0.415, loss_ctc=53.854, loss_att=44.223, acc=0.759, loss=47.112, backward_time=0.320, grad_norm=49.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.127, optim0_lr0=1.815e-04, train_time=1.367
+[gpub011:0/16] 2024-02-06 13:53:11,285 (trainer:762) INFO: 31epoch:train:5701-5800batch: iter_time=8.216e-05, forward_time=0.291, loss_ctc=43.178, loss_att=35.799, acc=0.765, loss=38.013, backward_time=0.296, grad_norm=38.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.814e-04, train_time=1.297
+[gpub011:0/16] 2024-02-06 13:55:44,522 (trainer:762) INFO: 31epoch:train:5801-5900batch: iter_time=8.869e-05, forward_time=0.377, loss_ctc=44.646, loss_att=35.554, acc=0.782, loss=38.282, backward_time=0.314, grad_norm=37.969, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.814e-04, train_time=1.532
+[gpub011:0/16] 2024-02-06 13:57:46,310 (trainer:762) INFO: 31epoch:train:5901-6000batch: iter_time=7.895e-05, forward_time=0.296, loss_ctc=42.304, loss_att=39.392, acc=0.755, loss=40.265, backward_time=0.303, grad_norm=35.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.814e-04, train_time=1.218
+[gpub011:0/16] 2024-02-06 14:00:16,395 (trainer:762) INFO: 31epoch:train:6001-6100batch: iter_time=7.632e-05, forward_time=0.420, loss_ctc=51.161, loss_att=57.067, acc=0.743, loss=55.295, backward_time=0.350, grad_norm=41.251, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.814e-04, train_time=1.500
+[gpub011:0/16] 2024-02-06 14:02:24,940 (trainer:762) INFO: 31epoch:train:6101-6200batch: iter_time=7.840e-05, forward_time=0.308, loss_ctc=53.164, loss_att=53.192, acc=0.736, loss=53.183, backward_time=0.316, grad_norm=49.807, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.814e-04, train_time=1.286
+[gpub011:0/16] 2024-02-06 14:04:09,928 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-06 14:04:29,182 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 14:04:32,782 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f790ba8bf40>)
+[gpub011:0/16] 2024-02-06 14:04:32,782 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-06 14:04:32,786 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 14:11:37,472 (trainer:762) INFO: 31epoch:train:6201-6300batch: iter_time=3.935, forward_time=0.287, loss_ctc=41.848, loss_att=39.060, acc=0.756, loss=39.896, backward_time=0.292, grad_norm=38.221, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.813e-04, train_time=5.525
+[gpub011:0/16] 2024-02-06 14:14:16,550 (trainer:762) INFO: 31epoch:train:6301-6400batch: iter_time=7.526e-05, forward_time=0.382, loss_ctc=48.998, loss_att=45.633, acc=0.752, loss=46.643, backward_time=0.342, grad_norm=42.493, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.813e-04, train_time=1.590
+[gpub011:0/16] 2024-02-06 14:16:17,035 (trainer:762) INFO: 31epoch:train:6401-6500batch: iter_time=8.238e-05, forward_time=0.304, loss_ctc=46.549, loss_att=45.764, acc=0.755, loss=45.999, backward_time=0.324, grad_norm=38.777, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.813e-04, train_time=1.205
+[gpub011:0/16] 2024-02-06 14:19:05,940 (trainer:762) INFO: 31epoch:train:6501-6600batch: iter_time=8.175e-05, forward_time=0.285, loss_ctc=39.560, loss_att=39.535, acc=0.763, loss=39.542, backward_time=0.292, grad_norm=37.390, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.813e-04, train_time=1.689
+[gpub011:0/16] 2024-02-06 14:21:24,463 (trainer:762) INFO: 31epoch:train:6601-6700batch: iter_time=7.994e-05, forward_time=0.291, loss_ctc=54.845, loss_att=52.747, acc=0.763, loss=53.376, backward_time=0.298, grad_norm=45.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.813e-04, train_time=1.385
+[gpub011:0/16] 2024-02-06 14:21:30,867 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 14:24:13,727 (trainer:762) INFO: 31epoch:train:6701-6800batch: iter_time=8.736e-05, forward_time=0.388, loss_ctc=51.856, loss_att=53.328, acc=0.753, loss=52.886, backward_time=0.354, grad_norm=44.615, clip=100.000, loss_scale=5.455e+33, optim_step_time=0.105, optim0_lr0=1.812e-04, train_time=1.692
+[gpub011:0/16] 2024-02-06 14:26:37,595 (trainer:762) INFO: 31epoch:train:6801-6900batch: iter_time=8.336e-05, forward_time=0.290, loss_ctc=48.622, loss_att=48.943, acc=0.758, loss=48.847, backward_time=0.295, grad_norm=48.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.812e-04, train_time=1.439
+[gpub011:0/16] 2024-02-06 14:27:16,624 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 14:29:12,966 (trainer:762) INFO: 31epoch:train:6901-7000batch: iter_time=8.132e-05, forward_time=0.288, loss_ctc=48.737, loss_att=40.050, acc=0.770, loss=42.656, backward_time=0.294, grad_norm=37.346, clip=100.000, loss_scale=3.147e+33, optim_step_time=0.094, optim0_lr0=1.812e-04, train_time=1.554
+[gpub011:0/16] 2024-02-06 14:31:44,490 (trainer:762) INFO: 31epoch:train:7001-7100batch: iter_time=8.351e-05, forward_time=0.380, loss_ctc=44.861, loss_att=35.246, acc=0.785, loss=38.131, backward_time=0.365, grad_norm=39.662, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.812e-04, train_time=1.515
+[gpub011:0/16] 2024-02-06 14:34:14,405 (trainer:762) INFO: 31epoch:train:7101-7200batch: iter_time=8.261e-05, forward_time=0.305, loss_ctc=41.566, loss_att=39.824, acc=0.762, loss=40.347, backward_time=0.305, grad_norm=39.005, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.812e-04, train_time=1.499
+[gpub011:0/16] 2024-02-06 14:36:29,736 (trainer:762) INFO: 31epoch:train:7201-7300batch: iter_time=8.633e-05, forward_time=0.290, loss_ctc=44.066, loss_att=47.696, acc=0.761, loss=46.607, backward_time=0.299, grad_norm=35.406, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.811e-04, train_time=1.353
+[gpub011:0/16] 2024-02-06 14:39:09,834 (trainer:762) INFO: 31epoch:train:7301-7400batch: iter_time=8.354e-05, forward_time=0.291, loss_ctc=48.911, loss_att=55.709, acc=0.739, loss=53.670, backward_time=0.296, grad_norm=44.185, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.811e-04, train_time=1.601
+[gpub011:0/16] 2024-02-06 14:41:38,892 (trainer:762) INFO: 31epoch:train:7401-7500batch: iter_time=5.695e-04, forward_time=0.405, loss_ctc=56.566, loss_att=55.057, acc=0.736, loss=55.510, backward_time=0.318, grad_norm=51.636, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.127, optim0_lr0=1.811e-04, train_time=1.490
+[gpub011:0/16] 2024-02-06 14:41:58,921 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-06 14:42:18,320 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 14:42:21,882 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78f8babf40>)
+[gpub011:0/16] 2024-02-06 14:42:21,882 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-06 14:42:21,962 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 14:50:48,250 (trainer:762) INFO: 31epoch:train:7501-7600batch: iter_time=3.987, forward_time=0.302, loss_ctc=37.296, loss_att=36.226, acc=0.749, loss=36.547, backward_time=0.299, grad_norm=36.292, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.811e-04, train_time=5.493
+[gpub011:0/16] 2024-02-06 14:53:30,296 (trainer:762) INFO: 31epoch:train:7601-7700batch: iter_time=7.900e-05, forward_time=0.437, loss_ctc=49.284, loss_att=43.471, acc=0.755, loss=45.215, backward_time=0.311, grad_norm=39.221, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=1.811e-04, train_time=1.620
+[gpub011:0/16] 2024-02-06 14:55:50,047 (trainer:762) INFO: 31epoch:train:7701-7800batch: iter_time=7.903e-05, forward_time=0.306, loss_ctc=46.910, loss_att=48.234, acc=0.747, loss=47.836, backward_time=0.315, grad_norm=40.855, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.810e-04, train_time=1.398
+[gpub011:0/16] 2024-02-06 14:58:49,409 (trainer:762) INFO: 31epoch:train:7801-7900batch: iter_time=8.766e-05, forward_time=0.395, loss_ctc=48.999, loss_att=46.910, acc=0.748, loss=47.537, backward_time=0.302, grad_norm=39.963, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.105, optim0_lr0=1.810e-04, train_time=1.794
+[gpub011:0/16] 2024-02-06 15:00:49,932 (trainer:762) INFO: 31epoch:train:7901-8000batch: iter_time=7.806e-05, forward_time=0.340, loss_ctc=49.221, loss_att=51.850, acc=0.752, loss=51.061, backward_time=0.324, grad_norm=40.872, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.810e-04, train_time=1.205
+[gpub011:0/16] 2024-02-06 15:03:31,592 (trainer:762) INFO: 31epoch:train:8001-8100batch: iter_time=8.189e-05, forward_time=0.288, loss_ctc=45.662, loss_att=43.124, acc=0.765, loss=43.886, backward_time=0.293, grad_norm=39.911, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.810e-04, train_time=1.617
+[gpub011:0/16] 2024-02-06 15:06:11,427 (trainer:762) INFO: 31epoch:train:8101-8200batch: iter_time=9.314e-05, forward_time=0.428, loss_ctc=53.845, loss_att=43.961, acc=0.762, loss=46.926, backward_time=0.331, grad_norm=49.461, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=1.810e-04, train_time=1.598
+[gpub011:0/16] 2024-02-06 15:08:25,098 (trainer:762) INFO: 31epoch:train:8201-8300batch: iter_time=9.601e-05, forward_time=0.288, loss_ctc=42.493, loss_att=36.073, acc=0.765, loss=37.999, backward_time=0.296, grad_norm=37.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.809e-04, train_time=1.337
+[gpub011:0/16] 2024-02-06 15:11:15,950 (trainer:762) INFO: 31epoch:train:8301-8400batch: iter_time=8.995e-05, forward_time=0.379, loss_ctc=44.589, loss_att=35.748, acc=0.783, loss=38.400, backward_time=0.360, grad_norm=37.522, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.809e-04, train_time=1.709
+[gpub011:0/16] 2024-02-06 15:13:40,202 (trainer:762) INFO: 31epoch:train:8401-8500batch: iter_time=9.662e-05, forward_time=0.303, loss_ctc=42.205, loss_att=39.786, acc=0.752, loss=40.512, backward_time=0.311, grad_norm=38.023, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.809e-04, train_time=1.442
+[gpub011:0/16] 2024-02-06 15:15:30,478 (trainer:762) INFO: 31epoch:train:8501-8600batch: iter_time=8.336e-05, forward_time=0.294, loss_ctc=50.606, loss_att=57.536, acc=0.741, loss=55.457, backward_time=0.300, grad_norm=44.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.809e-04, train_time=1.103
+[gpub011:0/16] 2024-02-06 15:18:15,411 (trainer:762) INFO: 31epoch:train:8601-8700batch: iter_time=9.547e-05, forward_time=0.390, loss_ctc=52.891, loss_att=54.050, acc=0.734, loss=53.702, backward_time=0.320, grad_norm=46.962, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.112, optim0_lr0=1.809e-04, train_time=1.648
+[gpub011:0/16] 2024-02-06 15:19:45,787 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-06 15:20:05,345 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 15:20:08,889 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f790bac8a90>)
+[gpub011:0/16] 2024-02-06 15:20:08,889 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-06 15:20:08,892 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 15:27:07,751 (trainer:762) INFO: 31epoch:train:8701-8800batch: iter_time=3.914, forward_time=0.308, loss_ctc=41.543, loss_att=38.213, acc=0.752, loss=39.212, backward_time=0.309, grad_norm=40.001, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.808e-04, train_time=5.324
+[gpub011:0/16] 2024-02-06 15:29:33,560 (trainer:762) INFO: 31epoch:train:8801-8900batch: iter_time=7.808e-05, forward_time=0.373, loss_ctc=48.594, loss_att=44.519, acc=0.742, loss=45.741, backward_time=0.320, grad_norm=44.603, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.808e-04, train_time=1.458
+[gpub011:0/16] 2024-02-06 15:31:53,091 (trainer:762) INFO: 31epoch:train:8901-9000batch: iter_time=7.423e-05, forward_time=0.297, loss_ctc=46.095, loss_att=43.671, acc=0.751, loss=44.398, backward_time=0.302, grad_norm=40.648, clip=100.000, loss_scale=4.621e+33, optim_step_time=0.094, optim0_lr0=1.808e-04, train_time=1.395
+[gpub011:0/16] 2024-02-06 15:34:02,671 (trainer:762) INFO: 31epoch:train:9001-9100batch: iter_time=7.512e-05, forward_time=0.296, loss_ctc=39.881, loss_att=38.515, acc=0.762, loss=38.925, backward_time=0.312, grad_norm=34.566, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.808e-04, train_time=1.296
+[gpub011:0/16] 2024-02-06 15:37:01,380 (trainer:762) INFO: 31epoch:train:9101-9200batch: iter_time=7.622e-05, forward_time=0.429, loss_ctc=54.340, loss_att=52.175, acc=0.755, loss=52.825, backward_time=0.306, grad_norm=44.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.808e-04, train_time=1.787
+[gpub011:0/16] 2024-02-06 15:39:14,897 (trainer:762) INFO: 31epoch:train:9201-9300batch: iter_time=7.750e-05, forward_time=0.302, loss_ctc=51.852, loss_att=50.024, acc=0.753, loss=50.573, backward_time=0.305, grad_norm=43.442, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.807e-04, train_time=1.335
+[gpub011:0/16] 2024-02-06 15:41:23,935 (trainer:762) INFO: 31epoch:train:9301-9400batch: iter_time=8.044e-05, forward_time=0.307, loss_ctc=47.996, loss_att=46.607, acc=0.752, loss=47.024, backward_time=0.312, grad_norm=44.689, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.807e-04, train_time=1.291
+[gpub011:0/16] 2024-02-06 15:44:28,358 (trainer:762) INFO: 31epoch:train:9401-9500batch: iter_time=8.033e-05, forward_time=0.405, loss_ctc=48.776, loss_att=39.537, acc=0.766, loss=42.309, backward_time=0.327, grad_norm=40.773, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.807e-04, train_time=1.844
+[gpub011:0/16] 2024-02-06 15:46:16,613 (trainer:762) INFO: 31epoch:train:9501-9600batch: iter_time=7.722e-05, forward_time=0.289, loss_ctc=44.769, loss_att=34.822, acc=0.781, loss=37.806, backward_time=0.298, grad_norm=40.220, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.807e-04, train_time=1.082
+[gpub011:0/16] 2024-02-06 15:48:54,750 (trainer:762) INFO: 31epoch:train:9601-9700batch: iter_time=8.306e-05, forward_time=0.299, loss_ctc=42.014, loss_att=39.680, acc=0.759, loss=40.380, backward_time=0.302, grad_norm=38.994, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.807e-04, train_time=1.582
+[gpub011:0/16] 2024-02-06 15:51:23,478 (trainer:762) INFO: 31epoch:train:9701-9800batch: iter_time=7.727e-05, forward_time=0.450, loss_ctc=44.030, loss_att=44.890, acc=0.761, loss=44.632, backward_time=0.321, grad_norm=60.957, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.806e-04, train_time=1.487
+[gpub011:0/16] 2024-02-06 15:53:49,120 (trainer:762) INFO: 31epoch:train:9801-9900batch: iter_time=7.866e-05, forward_time=0.303, loss_ctc=48.261, loss_att=52.665, acc=0.739, loss=51.344, backward_time=0.318, grad_norm=42.724, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.806e-04, train_time=1.456
+[gpub011:0/16] 2024-02-06 15:56:10,553 (trainer:762) INFO: 31epoch:train:9901-10000batch: iter_time=7.787e-05, forward_time=0.383, loss_ctc=56.070, loss_att=53.009, acc=0.738, loss=53.927, backward_time=0.311, grad_norm=50.004, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.806e-04, train_time=1.415
+[gpub011:0/16] 2024-02-06 15:56:30,619 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-06 15:56:50,505 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 15:56:54,418 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78f8b90130>)
+[gpub011:0/16] 2024-02-06 15:56:54,418 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-06 15:56:54,421 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 16:05:05,592 (trainer:762) INFO: 31epoch:train:10001-10100batch: iter_time=3.941, forward_time=0.318, loss_ctc=37.474, loss_att=36.318, acc=0.759, loss=36.665, backward_time=0.298, grad_norm=39.228, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.806e-04, train_time=5.350
+[gpub011:0/16] 2024-02-06 16:07:15,249 (trainer:762) INFO: 31epoch:train:10101-10200batch: iter_time=7.837e-05, forward_time=0.315, loss_ctc=49.156, loss_att=44.238, acc=0.765, loss=45.714, backward_time=0.310, grad_norm=38.964, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.806e-04, train_time=1.296
+[gpub011:0/16] 2024-02-06 16:09:45,899 (trainer:762) INFO: 31epoch:train:10201-10300batch: iter_time=7.945e-05, forward_time=0.338, loss_ctc=46.314, loss_att=49.313, acc=0.750, loss=48.413, backward_time=0.332, grad_norm=42.181, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.805e-04, train_time=1.507
+[gpub011:0/16] 2024-02-06 16:12:18,336 (trainer:762) INFO: 31epoch:train:10301-10400batch: iter_time=7.918e-05, forward_time=0.322, loss_ctc=48.740, loss_att=47.681, acc=0.752, loss=47.999, backward_time=0.301, grad_norm=42.646, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.805e-04, train_time=1.524
+[gpub011:0/16] 2024-02-06 16:15:15,046 (trainer:762) INFO: 31epoch:train:10401-10500batch: iter_time=8.072e-05, forward_time=0.292, loss_ctc=48.685, loss_att=54.432, acc=0.754, loss=52.708, backward_time=0.298, grad_norm=41.036, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.805e-04, train_time=1.766
+[gpub011:0/16] 2024-02-06 16:17:19,892 (trainer:762) INFO: 31epoch:train:10501-10600batch: iter_time=7.822e-05, forward_time=0.317, loss_ctc=46.032, loss_att=43.645, acc=0.772, loss=44.361, backward_time=0.311, grad_norm=41.413, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.805e-04, train_time=1.248
+[gpub011:0/16] 2024-02-06 16:20:27,999 (trainer:762) INFO: 31epoch:train:10601-10700batch: iter_time=8.115e-05, forward_time=0.406, loss_ctc=53.545, loss_att=46.093, acc=0.763, loss=48.328, backward_time=0.320, grad_norm=47.932, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.805e-04, train_time=1.882
+[gpub011:0/16] 2024-02-06 16:22:21,916 (trainer:762) INFO: 31epoch:train:10701-10800batch: iter_time=7.915e-05, forward_time=0.290, loss_ctc=42.567, loss_att=35.785, acc=0.772, loss=37.820, backward_time=0.295, grad_norm=38.163, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.805e-04, train_time=1.139
+[gpub011:0/16] 2024-02-06 16:25:21,759 (trainer:762) INFO: 31epoch:train:10801-10900batch: iter_time=8.213e-05, forward_time=0.287, loss_ctc=44.388, loss_att=35.776, acc=0.787, loss=38.360, backward_time=0.295, grad_norm=36.196, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.804e-04, train_time=1.798
+[gpub011:0/16] 2024-02-06 16:27:24,164 (trainer:762) INFO: 31epoch:train:10901-11000batch: iter_time=7.907e-05, forward_time=0.304, loss_ctc=41.680, loss_att=39.018, acc=0.762, loss=39.816, backward_time=0.304, grad_norm=34.690, clip=100.000, loss_scale=9.242e+33, optim_step_time=0.107, optim0_lr0=1.804e-04, train_time=1.224
+[gpub011:0/16] 2024-02-06 16:29:48,484 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 16:30:27,366 (trainer:762) INFO: 31epoch:train:11001-11100batch: iter_time=8.107e-05, forward_time=0.426, loss_ctc=50.937, loss_att=61.007, acc=0.743, loss=57.986, backward_time=0.315, grad_norm=43.610, clip=100.000, loss_scale=8.811e+33, optim_step_time=0.099, optim0_lr0=1.804e-04, train_time=1.833
+[gpub011:0/16] 2024-02-06 16:32:26,913 (trainer:762) INFO: 31epoch:train:11101-11200batch: iter_time=8.035e-05, forward_time=0.292, loss_ctc=51.643, loss_att=52.659, acc=0.744, loss=52.354, backward_time=0.298, grad_norm=47.045, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.804e-04, train_time=1.195
+[gpub011:0/16] 2024-02-06 16:34:37,474 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-06 16:34:56,638 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 16:35:00,206 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7924883fd0>)
+[gpub011:0/16] 2024-02-06 16:35:00,206 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-06 16:35:00,209 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 16:41:40,281 (trainer:762) INFO: 31epoch:train:11201-11300batch: iter_time=3.621, forward_time=0.443, loss_ctc=41.059, loss_att=38.098, acc=0.762, loss=38.986, backward_time=0.323, grad_norm=35.831, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.804e-04, train_time=5.533
+[gpub011:0/16] 2024-02-06 16:43:52,095 (trainer:762) INFO: 31epoch:train:11301-11400batch: iter_time=7.922e-05, forward_time=0.291, loss_ctc=48.862, loss_att=44.171, acc=0.757, loss=45.579, backward_time=0.297, grad_norm=42.544, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.803e-04, train_time=1.318
+[gpub011:0/16] 2024-02-06 16:46:27,476 (trainer:762) INFO: 31epoch:train:11401-11500batch: iter_time=8.118e-05, forward_time=0.290, loss_ctc=45.959, loss_att=44.366, acc=0.756, loss=44.844, backward_time=0.297, grad_norm=38.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.803e-04, train_time=1.554
+[gpub011:0/16] 2024-02-06 16:48:42,867 (trainer:762) INFO: 31epoch:train:11501-11600batch: iter_time=8.776e-05, forward_time=0.400, loss_ctc=39.464, loss_att=38.811, acc=0.765, loss=39.007, backward_time=0.346, grad_norm=34.912, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.803e-04, train_time=1.354
+[gpub011:0/16] 2024-02-06 16:51:10,130 (trainer:762) INFO: 31epoch:train:11601-11700batch: iter_time=8.501e-05, forward_time=0.294, loss_ctc=53.457, loss_att=52.506, acc=0.763, loss=52.791, backward_time=0.303, grad_norm=42.866, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.803e-04, train_time=1.472
+[gpub011:0/16] 2024-02-06 16:52:32,690 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 16:53:28,830 (trainer:762) INFO: 31epoch:train:11701-11800batch: iter_time=8.747e-05, forward_time=0.289, loss_ctc=51.444, loss_att=52.578, acc=0.757, loss=52.238, backward_time=0.298, grad_norm=44.896, clip=100.000, loss_scale=3.934e+33, optim_step_time=0.094, optim0_lr0=1.803e-04, train_time=1.387
+[gpub011:0/16] 2024-02-06 16:55:46,535 (trainer:762) INFO: 31epoch:train:11801-11900batch: iter_time=3.422e-04, forward_time=0.290, loss_ctc=48.625, loss_att=49.225, acc=0.757, loss=49.045, backward_time=0.307, grad_norm=45.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.802e-04, train_time=1.377
+[gpub011:0/16] 2024-02-06 16:58:39,548 (trainer:762) INFO: 31epoch:train:11901-12000batch: iter_time=8.225e-05, forward_time=0.347, loss_ctc=48.233, loss_att=39.692, acc=0.771, loss=42.254, backward_time=0.390, grad_norm=35.732, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.802e-04, train_time=1.730
+[gpub011:0/16] 2024-02-06 17:01:20,902 (trainer:762) INFO: 31epoch:train:12001-12100batch: iter_time=8.478e-05, forward_time=0.289, loss_ctc=43.866, loss_att=34.143, acc=0.789, loss=37.060, backward_time=0.294, grad_norm=37.938, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.802e-04, train_time=1.613
+[gpub011:0/16] 2024-02-06 17:03:21,914 (trainer:762) INFO: 31epoch:train:12101-12200batch: iter_time=8.656e-05, forward_time=0.288, loss_ctc=41.873, loss_att=39.654, acc=0.764, loss=40.320, backward_time=0.295, grad_norm=36.446, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.802e-04, train_time=1.210
+[gpub011:0/16] 2024-02-06 17:06:24,573 (trainer:762) INFO: 31epoch:train:12201-12300batch: iter_time=0.001, forward_time=0.477, loss_ctc=43.613, loss_att=46.247, acc=0.764, loss=45.457, backward_time=0.334, grad_norm=36.697, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.802e-04, train_time=1.826
+[gpub011:0/16] 2024-02-06 17:08:33,055 (trainer:762) INFO: 31epoch:train:12301-12400batch: iter_time=7.972e-05, forward_time=0.290, loss_ctc=48.066, loss_att=54.544, acc=0.744, loss=52.601, backward_time=0.298, grad_norm=43.881, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.801e-04, train_time=1.285
+[gpub011:0/16] 2024-02-06 17:11:12,135 (trainer:762) INFO: 31epoch:train:12401-12500batch: iter_time=7.497e-05, forward_time=0.290, loss_ctc=55.323, loss_att=53.435, acc=0.743, loss=54.001, backward_time=0.296, grad_norm=50.556, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.801e-04, train_time=1.591
+[gpub011:0/16] 2024-02-06 17:11:32,188 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-06 17:11:51,634 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 17:11:55,212 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78b35c76d0>)
+[gpub011:0/16] 2024-02-06 17:11:55,212 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-06 17:11:55,215 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 17:20:52,304 (trainer:762) INFO: 31epoch:train:12501-12600batch: iter_time=4.174, forward_time=0.394, loss_ctc=37.331, loss_att=35.063, acc=0.763, loss=35.744, backward_time=0.307, grad_norm=37.789, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.801e-04, train_time=5.801
+[gpub011:0/16] 2024-02-06 17:23:25,242 (trainer:762) INFO: 31epoch:train:12601-12700batch: iter_time=7.864e-05, forward_time=0.290, loss_ctc=48.813, loss_att=42.415, acc=0.770, loss=44.335, backward_time=0.296, grad_norm=38.881, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.801e-04, train_time=1.529
+[gpub011:0/16] 2024-02-06 17:25:43,312 (trainer:762) INFO: 31epoch:train:12701-12800batch: iter_time=8.519e-05, forward_time=0.291, loss_ctc=46.329, loss_att=48.176, acc=0.754, loss=47.622, backward_time=0.295, grad_norm=44.921, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.801e-04, train_time=1.381
+[gpub011:0/16] 2024-02-06 17:28:54,776 (trainer:762) INFO: 31epoch:train:12801-12900batch: iter_time=8.261e-05, forward_time=0.391, loss_ctc=48.403, loss_att=46.964, acc=0.754, loss=47.396, backward_time=0.375, grad_norm=44.107, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.800e-04, train_time=1.914
+[gpub011:0/16] 2024-02-06 17:30:57,483 (trainer:762) INFO: 31epoch:train:12901-13000batch: iter_time=7.811e-05, forward_time=0.292, loss_ctc=49.575, loss_att=54.532, acc=0.754, loss=53.045, backward_time=0.299, grad_norm=41.761, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.800e-04, train_time=1.227
+[gpub011:0/16] 2024-02-06 17:33:52,466 (trainer:762) INFO: 31epoch:train:13001-13100batch: iter_time=8.452e-05, forward_time=0.289, loss_ctc=45.592, loss_att=42.731, acc=0.775, loss=43.589, backward_time=0.293, grad_norm=39.218, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.800e-04, train_time=1.749
+[gpub011:0/16] 2024-02-06 17:36:26,185 (trainer:762) INFO: 31epoch:train:13101-13200batch: iter_time=8.194e-05, forward_time=0.375, loss_ctc=52.527, loss_att=45.585, acc=0.766, loss=47.667, backward_time=0.408, grad_norm=47.782, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.800e-04, train_time=1.537
+[gpub011:0/16] 2024-02-06 17:38:42,627 (trainer:762) INFO: 31epoch:train:13201-13300batch: iter_time=8.081e-05, forward_time=0.290, loss_ctc=42.124, loss_att=35.855, acc=0.772, loss=37.736, backward_time=0.292, grad_norm=36.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.800e-04, train_time=1.364
+[gpub011:0/16] 2024-02-06 17:41:29,144 (trainer:762) INFO: 31epoch:train:13301-13400batch: iter_time=8.164e-05, forward_time=0.288, loss_ctc=44.160, loss_att=35.353, acc=0.789, loss=37.995, backward_time=0.294, grad_norm=37.146, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.799e-04, train_time=1.665
+[gpub011:0/16] 2024-02-06 17:44:13,894 (trainer:762) INFO: 31epoch:train:13401-13500batch: iter_time=8.361e-05, forward_time=0.314, loss_ctc=41.819, loss_att=39.629, acc=0.759, loss=40.286, backward_time=0.295, grad_norm=36.210, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=1.799e-04, train_time=1.647
+[gpub011:0/16] 2024-02-06 17:47:05,859 (trainer:762) INFO: 31epoch:train:13501-13600batch: iter_time=8.422e-05, forward_time=0.393, loss_ctc=50.150, loss_att=60.519, acc=0.747, loss=57.408, backward_time=0.321, grad_norm=42.918, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.108, optim0_lr0=1.799e-04, train_time=1.720
+[gpub011:0/16] 2024-02-06 17:49:18,939 (trainer:762) INFO: 31epoch:train:13601-13700batch: iter_time=8.395e-05, forward_time=0.288, loss_ctc=53.127, loss_att=54.149, acc=0.740, loss=53.842, backward_time=0.297, grad_norm=48.860, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.799e-04, train_time=1.331
+[gpub011:0/16] 2024-02-06 17:51:13,580 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-06 17:51:32,847 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 17:51:36,411 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f79248441c0>)
+[gpub011:0/16] 2024-02-06 17:51:36,411 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-06 17:51:36,414 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 17:59:14,476 (trainer:762) INFO: 31epoch:train:13701-13800batch: iter_time=4.268, forward_time=0.369, loss_ctc=40.957, loss_att=38.576, acc=0.760, loss=39.290, backward_time=0.307, grad_norm=38.654, clip=100.000, loss_scale=3.842e+33, optim_step_time=0.097, optim0_lr0=1.799e-04, train_time=5.955
+[gpub011:0/16] 2024-02-06 18:01:14,336 (trainer:762) INFO: 31epoch:train:13801-13900batch: iter_time=7.756e-05, forward_time=0.290, loss_ctc=48.145, loss_att=43.950, acc=0.756, loss=45.208, backward_time=0.296, grad_norm=40.097, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.798e-04, train_time=1.199
+[gpub011:0/16] 2024-02-06 18:04:03,616 (trainer:762) INFO: 31epoch:train:13901-14000batch: iter_time=8.131e-05, forward_time=0.420, loss_ctc=46.080, loss_att=43.832, acc=0.759, loss=44.506, backward_time=0.344, grad_norm=38.946, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.798e-04, train_time=1.692
+[gpub011:0/16] 2024-02-06 18:06:31,331 (trainer:762) INFO: 31epoch:train:14001-14100batch: iter_time=8.231e-05, forward_time=0.286, loss_ctc=38.914, loss_att=38.715, acc=0.765, loss=38.774, backward_time=0.293, grad_norm=34.935, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.798e-04, train_time=1.477
+[gpub011:0/16] 2024-02-06 18:09:15,427 (trainer:762) INFO: 31epoch:train:14101-14200batch: iter_time=8.510e-05, forward_time=0.303, loss_ctc=54.533, loss_att=51.539, acc=0.765, loss=52.437, backward_time=0.308, grad_norm=44.515, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.798e-04, train_time=1.641
+[gpub011:0/16] 2024-02-06 18:11:56,233 (trainer:762) INFO: 31epoch:train:14201-14300batch: iter_time=8.564e-05, forward_time=0.386, loss_ctc=51.433, loss_att=52.546, acc=0.756, loss=52.212, backward_time=0.328, grad_norm=44.077, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.798e-04, train_time=1.608
+[gpub011:0/16] 2024-02-06 18:14:03,744 (trainer:762) INFO: 31epoch:train:14301-14400batch: iter_time=8.274e-05, forward_time=0.327, loss_ctc=47.353, loss_att=48.365, acc=0.760, loss=48.062, backward_time=0.296, grad_norm=46.993, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.798e-04, train_time=1.275
+[gpub011:0/16] 2024-02-06 18:16:52,190 (trainer:762) INFO: 31epoch:train:14401-14500batch: iter_time=6.913e-04, forward_time=0.382, loss_ctc=48.119, loss_att=39.459, acc=0.773, loss=42.057, backward_time=0.375, grad_norm=36.978, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.797e-04, train_time=1.683
+[gpub011:0/16] 2024-02-06 18:19:00,837 (trainer:762) INFO: 31epoch:train:14501-14600batch: iter_time=8.356e-05, forward_time=0.287, loss_ctc=44.410, loss_att=34.952, acc=0.786, loss=37.790, backward_time=0.296, grad_norm=37.982, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.797e-04, train_time=1.287
+[gpub011:0/16] 2024-02-06 18:22:10,119 (trainer:762) INFO: 31epoch:train:14601-14700batch: iter_time=8.434e-05, forward_time=0.348, loss_ctc=41.539, loss_att=39.605, acc=0.765, loss=40.185, backward_time=0.307, grad_norm=38.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.797e-04, train_time=1.892
+[gpub011:0/16] 2024-02-06 18:24:15,539 (trainer:762) INFO: 31epoch:train:14701-14800batch: iter_time=8.403e-05, forward_time=0.368, loss_ctc=44.159, loss_att=46.400, acc=0.765, loss=45.728, backward_time=0.315, grad_norm=35.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.797e-04, train_time=1.253
+[gpub011:0/16] 2024-02-06 18:26:54,677 (trainer:762) INFO: 31epoch:train:14801-14900batch: iter_time=8.462e-05, forward_time=0.297, loss_ctc=48.223, loss_att=55.037, acc=0.742, loss=52.993, backward_time=0.298, grad_norm=42.199, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.797e-04, train_time=1.592
+[gpub011:0/16] 2024-02-06 18:29:55,561 (trainer:762) INFO: 31epoch:train:14901-15000batch: iter_time=2.028e-04, forward_time=0.387, loss_ctc=54.818, loss_att=53.669, acc=0.738, loss=54.014, backward_time=0.326, grad_norm=52.056, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.115, optim0_lr0=1.796e-04, train_time=1.809
+[gpub011:0/16] 2024-02-06 19:09:35,929 (trainer:361) INFO: 31epoch results: [train] iter_time=0.308, forward_time=0.336, loss_ctc=47.479, loss_att=45.237, acc=0.756, loss=45.909, backward_time=0.315, grad_norm=41.613, clip=100.000, loss_scale=4.286e+33, optim_step_time=0.099, optim0_lr0=1.811e-04, train_time=1.786, time=7 hours, 26 minutes and 59.31 seconds, total_count=495000, gpu_max_cached_mem_GB=41.025, [valid] loss_ctc=35.901, cer_ctc=0.186, loss_att=38.143, acc=0.685, cer=0.310, wer=0.994, loss=37.471, time=39 minutes and 16.21 seconds, total_count=154143, gpu_max_cached_mem_GB=41.025
+[gpub011:0/16] 2024-02-06 19:09:46,610 (trainer:416) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub011:0/16] 2024-02-06 19:09:46,668 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/26epoch.pth
+[gpub011:0/16] 2024-02-06 19:09:46,669 (trainer:290) INFO: 32/45epoch started. Estimated time to finish: 4 days, 14 hours and 42 minutes
+[gpub011:0/16] 2024-02-06 19:09:47,029 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-06 19:10:05,676 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 19:10:09,368 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7fd93c5ea0>)
+[gpub011:0/16] 2024-02-06 19:10:09,368 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-06 19:10:09,371 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 19:18:21,946 (trainer:762) INFO: 32epoch:train:1-100batch: iter_time=3.796, forward_time=0.376, loss_ctc=49.884, loss_att=50.368, acc=0.749, loss=50.223, backward_time=0.313, grad_norm=44.868, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.796e-04, train_time=5.152
+[gpub011:0/16] 2024-02-06 19:20:38,214 (trainer:762) INFO: 32epoch:train:101-200batch: iter_time=8.590e-05, forward_time=0.297, loss_ctc=49.821, loss_att=47.669, acc=0.754, loss=48.314, backward_time=0.296, grad_norm=40.496, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.796e-04, train_time=1.362
+[gpub011:0/16] 2024-02-06 19:23:19,071 (trainer:762) INFO: 32epoch:train:201-300batch: iter_time=9.055e-05, forward_time=0.313, loss_ctc=47.292, loss_att=46.896, acc=0.761, loss=47.015, backward_time=0.315, grad_norm=38.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.796e-04, train_time=1.608
+[gpub011:0/16] 2024-02-06 19:26:20,097 (trainer:762) INFO: 32epoch:train:301-400batch: iter_time=8.651e-04, forward_time=0.342, loss_ctc=53.747, loss_att=51.500, acc=0.747, loss=52.174, backward_time=0.325, grad_norm=41.829, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.796e-04, train_time=1.811
+[gpub011:0/16] 2024-02-06 19:28:59,114 (trainer:762) INFO: 32epoch:train:401-500batch: iter_time=8.694e-05, forward_time=0.302, loss_ctc=44.197, loss_att=37.928, acc=0.773, loss=39.809, backward_time=0.295, grad_norm=38.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.795e-04, train_time=1.590
+[gpub011:0/16] 2024-02-06 19:31:24,598 (trainer:762) INFO: 32epoch:train:501-600batch: iter_time=9.343e-05, forward_time=0.320, loss_ctc=41.490, loss_att=42.534, acc=0.754, loss=42.220, backward_time=0.308, grad_norm=36.079, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.795e-04, train_time=1.455
+[gpub011:0/16] 2024-02-06 19:34:08,852 (trainer:762) INFO: 32epoch:train:601-700batch: iter_time=8.838e-05, forward_time=0.364, loss_ctc=45.483, loss_att=40.809, acc=0.762, loss=42.211, backward_time=0.302, grad_norm=38.422, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.114, optim0_lr0=1.795e-04, train_time=1.642
+[gpub011:0/16] 2024-02-06 19:36:47,869 (trainer:762) INFO: 32epoch:train:701-800batch: iter_time=8.196e-05, forward_time=0.292, loss_ctc=49.949, loss_att=49.371, acc=0.738, loss=49.544, backward_time=0.297, grad_norm=40.883, clip=100.000, loss_scale=7.685e+33, optim_step_time=0.096, optim0_lr0=1.795e-04, train_time=1.591
+[gpub011:0/16] 2024-02-06 19:39:15,899 (trainer:762) INFO: 32epoch:train:801-900batch: iter_time=8.714e-05, forward_time=0.299, loss_ctc=51.910, loss_att=55.740, acc=0.743, loss=54.591, backward_time=0.305, grad_norm=43.679, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.795e-04, train_time=1.480
+[gpub011:0/16] 2024-02-06 19:41:38,397 (trainer:762) INFO: 32epoch:train:901-1000batch: iter_time=2.462e-04, forward_time=0.381, loss_ctc=45.897, loss_att=41.778, acc=0.769, loss=43.014, backward_time=0.317, grad_norm=39.065, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.794e-04, train_time=1.425
+[gpub011:0/16] 2024-02-06 19:42:58,349 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 19:44:03,157 (trainer:762) INFO: 32epoch:train:1001-1100batch: iter_time=8.070e-05, forward_time=0.304, loss_ctc=63.488, loss_att=58.213, acc=0.749, loss=59.796, backward_time=0.302, grad_norm=51.783, clip=100.000, loss_scale=8.654e+33, optim_step_time=0.094, optim0_lr0=1.794e-04, train_time=1.447
+[gpub011:0/16] 2024-02-06 19:46:03,678 (trainer:762) INFO: 32epoch:train:1101-1200batch: iter_time=8.311e-05, forward_time=0.295, loss_ctc=51.985, loss_att=50.339, acc=0.754, loss=50.833, backward_time=0.309, grad_norm=42.794, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.794e-04, train_time=1.205
+[gpub011:0/16] 2024-02-06 19:47:31,962 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-06 19:47:51,325 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 19:47:54,867 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b7e7d9630>)
+[gpub011:0/16] 2024-02-06 19:47:54,867 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-06 19:47:54,870 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 19:55:33,140 (trainer:762) INFO: 32epoch:train:1201-1300batch: iter_time=4.326, forward_time=0.412, loss_ctc=46.589, loss_att=48.471, acc=0.752, loss=47.906, backward_time=0.331, grad_norm=44.689, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.794e-04, train_time=5.694
+[gpub011:0/16] 2024-02-06 19:58:15,092 (trainer:762) INFO: 32epoch:train:1301-1400batch: iter_time=7.867e-05, forward_time=0.310, loss_ctc=46.543, loss_att=42.785, acc=0.747, loss=43.913, backward_time=0.292, grad_norm=43.301, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.794e-04, train_time=1.620
+[gpub011:0/16] 2024-02-06 20:00:17,543 (trainer:762) INFO: 32epoch:train:1401-1500batch: iter_time=7.933e-05, forward_time=0.307, loss_ctc=49.399, loss_att=47.288, acc=0.752, loss=47.921, backward_time=0.305, grad_norm=37.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.793e-04, train_time=1.224
+[gpub011:0/16] 2024-02-06 20:02:38,941 (trainer:762) INFO: 32epoch:train:1501-1600batch: iter_time=3.920e-04, forward_time=0.419, loss_ctc=49.877, loss_att=45.972, acc=0.764, loss=47.144, backward_time=0.329, grad_norm=38.448, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.119, optim0_lr0=1.793e-04, train_time=1.414
+[gpub011:0/16] 2024-02-06 20:05:22,313 (trainer:762) INFO: 32epoch:train:1601-1700batch: iter_time=8.411e-05, forward_time=0.290, loss_ctc=50.215, loss_att=47.601, acc=0.749, loss=48.385, backward_time=0.294, grad_norm=43.040, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.793e-04, train_time=1.633
+[gpub011:0/16] 2024-02-06 20:07:23,090 (trainer:762) INFO: 32epoch:train:1701-1800batch: iter_time=8.679e-05, forward_time=0.311, loss_ctc=45.350, loss_att=43.639, acc=0.753, loss=44.152, backward_time=0.314, grad_norm=37.680, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.793e-04, train_time=1.208
+[gpub011:0/16] 2024-02-06 20:10:24,471 (trainer:762) INFO: 32epoch:train:1801-1900batch: iter_time=6.792e-04, forward_time=0.344, loss_ctc=40.134, loss_att=37.349, acc=0.762, loss=38.185, backward_time=0.361, grad_norm=33.421, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.793e-04, train_time=1.813
+[gpub011:0/16] 2024-02-06 20:12:27,717 (trainer:762) INFO: 32epoch:train:1901-2000batch: iter_time=8.365e-05, forward_time=0.290, loss_ctc=42.591, loss_att=40.754, acc=0.758, loss=41.305, backward_time=0.295, grad_norm=36.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.792e-04, train_time=1.232
+[gpub011:0/16] 2024-02-06 20:14:37,486 (trainer:762) INFO: 32epoch:train:2001-2100batch: iter_time=8.204e-05, forward_time=0.325, loss_ctc=51.255, loss_att=53.734, acc=0.715, loss=52.991, backward_time=0.306, grad_norm=43.533, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.792e-04, train_time=1.297
+[gpub011:0/16] 2024-02-06 20:17:30,620 (trainer:762) INFO: 32epoch:train:2101-2200batch: iter_time=7.892e-04, forward_time=0.364, loss_ctc=48.989, loss_att=49.637, acc=0.745, loss=49.443, backward_time=0.340, grad_norm=43.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.792e-04, train_time=1.732
+[gpub011:0/16] 2024-02-06 20:19:37,125 (trainer:762) INFO: 32epoch:train:2201-2300batch: iter_time=8.516e-05, forward_time=0.290, loss_ctc=52.785, loss_att=49.601, acc=0.754, loss=50.556, backward_time=0.298, grad_norm=45.902, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.792e-04, train_time=1.265
+[gpub011:0/16] 2024-02-06 20:21:41,237 (trainer:762) INFO: 32epoch:train:2301-2400batch: iter_time=8.510e-05, forward_time=0.301, loss_ctc=54.484, loss_att=53.593, acc=0.740, loss=53.860, backward_time=0.303, grad_norm=48.293, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.792e-04, train_time=1.241
+[gpub011:0/16] 2024-02-06 20:24:22,759 (trainer:762) INFO: 32epoch:train:2401-2500batch: iter_time=8.674e-05, forward_time=0.335, loss_ctc=50.887, loss_att=46.640, acc=0.758, loss=47.914, backward_time=0.302, grad_norm=45.872, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.792e-04, train_time=1.615
+[gpub011:0/16] 2024-02-06 20:24:42,912 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-06 20:25:02,205 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 20:25:05,770 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b46825ff0>)
+[gpub011:0/16] 2024-02-06 20:25:05,770 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-06 20:25:05,773 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 20:33:24,423 (trainer:762) INFO: 32epoch:train:2501-2600batch: iter_time=4.150, forward_time=0.413, loss_ctc=48.413, loss_att=48.367, acc=0.746, loss=48.381, backward_time=0.319, grad_norm=42.568, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.791e-04, train_time=5.416
+[gpub011:0/16] 2024-02-06 20:36:00,894 (trainer:762) INFO: 32epoch:train:2601-2700batch: iter_time=7.957e-05, forward_time=0.294, loss_ctc=47.905, loss_att=44.892, acc=0.755, loss=45.796, backward_time=0.302, grad_norm=39.509, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.791e-04, train_time=1.565
+[gpub011:0/16] 2024-02-06 20:38:11,202 (trainer:762) INFO: 32epoch:train:2701-2800batch: iter_time=8.305e-05, forward_time=0.311, loss_ctc=46.743, loss_att=45.988, acc=0.764, loss=46.215, backward_time=0.304, grad_norm=39.005, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.791e-04, train_time=1.302
+[gpub011:0/16] 2024-02-06 20:40:50,448 (trainer:762) INFO: 32epoch:train:2801-2900batch: iter_time=3.501e-04, forward_time=0.414, loss_ctc=52.134, loss_att=48.296, acc=0.749, loss=49.448, backward_time=0.320, grad_norm=39.805, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.131, optim0_lr0=1.791e-04, train_time=1.593
+[gpub011:0/16] 2024-02-06 20:43:15,485 (trainer:762) INFO: 32epoch:train:2901-3000batch: iter_time=7.935e-05, forward_time=0.288, loss_ctc=43.883, loss_att=37.822, acc=0.775, loss=39.640, backward_time=0.292, grad_norm=39.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.791e-04, train_time=1.450
+[gpub011:0/16] 2024-02-06 20:44:40,683 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 20:45:08,497 (trainer:762) INFO: 32epoch:train:3001-3100batch: iter_time=7.956e-05, forward_time=0.330, loss_ctc=40.764, loss_att=38.256, acc=0.759, loss=39.008, backward_time=0.306, grad_norm=33.469, clip=100.000, loss_scale=5.507e+33, optim_step_time=0.096, optim0_lr0=1.790e-04, train_time=1.130
+[gpub011:0/16] 2024-02-06 20:47:47,651 (trainer:762) INFO: 32epoch:train:3101-3200batch: iter_time=0.001, forward_time=0.408, loss_ctc=44.041, loss_att=39.919, acc=0.761, loss=41.156, backward_time=0.319, grad_norm=37.244, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.114, optim0_lr0=1.790e-04, train_time=1.590
+[gpub011:0/16] 2024-02-06 20:50:26,445 (trainer:762) INFO: 32epoch:train:3201-3300batch: iter_time=8.802e-05, forward_time=0.289, loss_ctc=48.064, loss_att=48.566, acc=0.739, loss=48.415, backward_time=0.293, grad_norm=40.738, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.790e-04, train_time=1.589
+[gpub011:0/16] 2024-02-06 20:52:19,756 (trainer:762) INFO: 32epoch:train:3301-3400batch: iter_time=8.152e-05, forward_time=0.314, loss_ctc=51.455, loss_att=55.149, acc=0.731, loss=54.041, backward_time=0.315, grad_norm=46.545, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.790e-04, train_time=1.133
+[gpub011:0/16] 2024-02-06 20:55:05,692 (trainer:762) INFO: 32epoch:train:3401-3500batch: iter_time=8.661e-05, forward_time=0.287, loss_ctc=44.812, loss_att=42.127, acc=0.758, loss=42.933, backward_time=0.293, grad_norm=38.426, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.790e-04, train_time=1.659
+[gpub011:0/16] 2024-02-06 20:57:36,019 (trainer:762) INFO: 32epoch:train:3501-3600batch: iter_time=8.157e-05, forward_time=0.394, loss_ctc=60.644, loss_att=56.643, acc=0.745, loss=57.843, backward_time=0.325, grad_norm=54.947, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.128, optim0_lr0=1.789e-04, train_time=1.501
+[gpub011:0/16] 2024-02-06 20:59:37,730 (trainer:762) INFO: 32epoch:train:3601-3700batch: iter_time=8.246e-05, forward_time=0.318, loss_ctc=50.945, loss_att=50.024, acc=0.748, loss=50.300, backward_time=0.312, grad_norm=44.651, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.789e-04, train_time=1.219
+[gpub011:0/16] 2024-02-06 21:01:13,057 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-06 21:01:32,217 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 21:01:35,817 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78ec3a6bc0>)
+[gpub011:0/16] 2024-02-06 21:01:35,817 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-06 21:01:35,821 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 21:09:15,397 (trainer:762) INFO: 32epoch:train:3701-3800batch: iter_time=4.342, forward_time=0.396, loss_ctc=44.854, loss_att=47.334, acc=0.757, loss=46.590, backward_time=0.314, grad_norm=43.870, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.789e-04, train_time=5.776
+[gpub011:0/16] 2024-02-06 21:11:20,690 (trainer:762) INFO: 32epoch:train:3801-3900batch: iter_time=8.188e-05, forward_time=0.304, loss_ctc=45.461, loss_att=44.996, acc=0.757, loss=45.136, backward_time=0.297, grad_norm=42.437, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.789e-04, train_time=1.253
+[gpub011:0/16] 2024-02-06 21:13:13,344 (trainer:762) INFO: 32epoch:train:3901-4000batch: iter_time=8.659e-05, forward_time=0.311, loss_ctc=49.167, loss_att=47.128, acc=0.756, loss=47.740, backward_time=0.306, grad_norm=38.264, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.789e-04, train_time=1.126
+[gpub011:0/16] 2024-02-06 21:16:23,270 (trainer:762) INFO: 32epoch:train:4001-4100batch: iter_time=9.391e-05, forward_time=0.383, loss_ctc=48.375, loss_att=45.745, acc=0.776, loss=46.534, backward_time=0.351, grad_norm=35.324, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.788e-04, train_time=1.899
+[gpub011:0/16] 2024-02-06 21:18:29,461 (trainer:762) INFO: 32epoch:train:4101-4200batch: iter_time=8.724e-05, forward_time=0.298, loss_ctc=49.804, loss_att=47.038, acc=0.755, loss=47.868, backward_time=0.297, grad_norm=40.536, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.788e-04, train_time=1.262
+[gpub011:0/16] 2024-02-06 21:20:30,749 (trainer:762) INFO: 32epoch:train:4201-4300batch: iter_time=9.302e-05, forward_time=0.309, loss_ctc=44.503, loss_att=46.447, acc=0.759, loss=45.864, backward_time=0.306, grad_norm=37.668, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.788e-04, train_time=1.213
+[gpub011:0/16] 2024-02-06 21:23:20,263 (trainer:762) INFO: 32epoch:train:4301-4400batch: iter_time=8.885e-05, forward_time=0.397, loss_ctc=40.782, loss_att=37.228, acc=0.766, loss=38.294, backward_time=0.343, grad_norm=33.961, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.788e-04, train_time=1.695
+[gpub011:0/16] 2024-02-06 21:25:21,689 (trainer:762) INFO: 32epoch:train:4401-4500batch: iter_time=8.893e-05, forward_time=0.289, loss_ctc=42.036, loss_att=39.865, acc=0.766, loss=40.517, backward_time=0.296, grad_norm=38.447, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.788e-04, train_time=1.214
+[gpub011:0/16] 2024-02-06 21:27:21,692 (trainer:762) INFO: 32epoch:train:4501-4600batch: iter_time=7.935e-05, forward_time=0.323, loss_ctc=50.364, loss_att=53.029, acc=0.731, loss=52.230, backward_time=0.309, grad_norm=43.313, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.788e-04, train_time=1.200
+[gpub011:0/16] 2024-02-06 21:30:27,721 (trainer:762) INFO: 32epoch:train:4601-4700batch: iter_time=9.050e-05, forward_time=0.398, loss_ctc=48.307, loss_att=48.560, acc=0.764, loss=48.484, backward_time=0.359, grad_norm=41.622, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.787e-04, train_time=1.860
+[gpub011:0/16] 2024-02-06 21:32:34,971 (trainer:762) INFO: 32epoch:train:4701-4800batch: iter_time=8.003e-05, forward_time=0.290, loss_ctc=51.707, loss_att=48.127, acc=0.769, loss=49.201, backward_time=0.298, grad_norm=42.790, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.787e-04, train_time=1.272
+[gpub011:0/16] 2024-02-06 21:34:36,173 (trainer:762) INFO: 32epoch:train:4801-4900batch: iter_time=8.802e-05, forward_time=0.327, loss_ctc=52.630, loss_att=52.780, acc=0.752, loss=52.735, backward_time=0.312, grad_norm=48.452, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.787e-04, train_time=1.212
+[gpub011:0/16] 2024-02-06 21:37:44,641 (trainer:762) INFO: 32epoch:train:4901-5000batch: iter_time=0.001, forward_time=0.363, loss_ctc=49.733, loss_att=47.832, acc=0.764, loss=48.403, backward_time=0.322, grad_norm=46.340, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.787e-04, train_time=1.883
+[gpub011:0/16] 2024-02-06 21:38:04,960 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-06 21:38:24,484 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 21:38:28,362 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f792743f520>)
+[gpub011:0/16] 2024-02-06 21:38:28,362 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-06 21:38:28,366 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 21:46:38,121 (trainer:762) INFO: 32epoch:train:5001-5100batch: iter_time=3.974, forward_time=0.365, loss_ctc=46.896, loss_att=48.617, acc=0.747, loss=48.100, backward_time=0.305, grad_norm=41.715, clip=100.000, loss_scale=6.542e+33, optim_step_time=0.095, optim0_lr0=1.787e-04, train_time=5.336
+[gpub011:0/16] 2024-02-06 21:48:30,443 (trainer:762) INFO: 32epoch:train:5101-5200batch: iter_time=8.489e-05, forward_time=0.311, loss_ctc=47.870, loss_att=45.003, acc=0.755, loss=45.863, backward_time=0.308, grad_norm=37.684, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.786e-04, train_time=1.123
+[gpub011:0/16] 2024-02-06 21:51:06,233 (trainer:762) INFO: 32epoch:train:5201-5300batch: iter_time=8.465e-05, forward_time=0.296, loss_ctc=46.676, loss_att=46.249, acc=0.765, loss=46.377, backward_time=0.300, grad_norm=38.189, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.786e-04, train_time=1.558
+[gpub011:0/16] 2024-02-06 21:53:21,515 (trainer:762) INFO: 32epoch:train:5301-5400batch: iter_time=2.050e-04, forward_time=0.374, loss_ctc=51.817, loss_att=48.300, acc=0.751, loss=49.355, backward_time=0.327, grad_norm=41.035, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.109, optim0_lr0=1.786e-04, train_time=1.351
+[gpub011:0/16] 2024-02-06 21:55:24,357 (trainer:762) INFO: 32epoch:train:5401-5500batch: iter_time=8.978e-05, forward_time=0.295, loss_ctc=43.380, loss_att=37.448, acc=0.777, loss=39.227, backward_time=0.303, grad_norm=37.706, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.786e-04, train_time=1.229
+[gpub011:0/16] 2024-02-06 21:58:06,197 (trainer:762) INFO: 32epoch:train:5501-5600batch: iter_time=9.109e-05, forward_time=0.300, loss_ctc=40.708, loss_att=38.555, acc=0.758, loss=39.200, backward_time=0.302, grad_norm=34.854, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.786e-04, train_time=1.618
+[gpub011:0/16] 2024-02-06 21:59:15,377 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 22:00:20,942 (trainer:762) INFO: 32epoch:train:5601-5700batch: iter_time=4.495e-04, forward_time=0.386, loss_ctc=43.983, loss_att=40.187, acc=0.761, loss=41.326, backward_time=0.336, grad_norm=36.627, clip=100.000, loss_scale=8.129e+33, optim_step_time=0.110, optim0_lr0=1.785e-04, train_time=1.347
+[gpub011:0/16] 2024-02-06 22:02:25,557 (trainer:762) INFO: 32epoch:train:5701-5800batch: iter_time=8.275e-05, forward_time=0.288, loss_ctc=47.836, loss_att=48.612, acc=0.741, loss=48.380, backward_time=0.295, grad_norm=41.263, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.785e-04, train_time=1.244
+[gpub011:0/16] 2024-02-06 22:05:18,564 (trainer:762) INFO: 32epoch:train:5801-5900batch: iter_time=9.428e-05, forward_time=0.313, loss_ctc=51.063, loss_att=55.278, acc=0.732, loss=54.014, backward_time=0.308, grad_norm=46.409, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.785e-04, train_time=1.732
+[gpub011:0/16] 2024-02-06 22:07:19,048 (trainer:762) INFO: 32epoch:train:5901-6000batch: iter_time=9.203e-05, forward_time=0.352, loss_ctc=44.892, loss_att=42.073, acc=0.760, loss=42.919, backward_time=0.319, grad_norm=38.393, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.785e-04, train_time=1.204
+[gpub011:0/16] 2024-02-06 22:09:52,194 (trainer:762) INFO: 32epoch:train:6001-6100batch: iter_time=9.652e-05, forward_time=0.366, loss_ctc=59.138, loss_att=56.629, acc=0.747, loss=57.382, backward_time=0.313, grad_norm=54.568, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.785e-04, train_time=1.531
+[gpub011:0/16] 2024-02-06 22:11:57,316 (trainer:762) INFO: 32epoch:train:6101-6200batch: iter_time=9.279e-05, forward_time=0.315, loss_ctc=50.497, loss_att=49.318, acc=0.750, loss=49.672, backward_time=0.302, grad_norm=43.969, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.784e-04, train_time=1.250
+[gpub011:0/16] 2024-02-06 22:13:32,965 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-06 22:13:52,297 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 22:13:55,891 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f790fa93d60>)
+[gpub011:0/16] 2024-02-06 22:13:55,891 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-06 22:13:55,895 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 22:20:46,057 (trainer:762) INFO: 32epoch:train:6201-6300batch: iter_time=3.805, forward_time=0.393, loss_ctc=44.242, loss_att=46.853, acc=0.759, loss=46.070, backward_time=0.320, grad_norm=42.930, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.784e-04, train_time=5.288
+[gpub011:0/16] 2024-02-06 22:22:45,031 (trainer:762) INFO: 32epoch:train:6301-6400batch: iter_time=7.836e-05, forward_time=0.291, loss_ctc=44.522, loss_att=43.914, acc=0.758, loss=44.096, backward_time=0.296, grad_norm=39.614, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.784e-04, train_time=1.190
+[gpub011:0/16] 2024-02-06 22:25:02,337 (trainer:762) INFO: 32epoch:train:6401-6500batch: iter_time=8.271e-05, forward_time=0.310, loss_ctc=49.345, loss_att=47.801, acc=0.756, loss=48.264, backward_time=0.315, grad_norm=40.131, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.784e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 22:27:48,695 (trainer:762) INFO: 32epoch:train:6501-6600batch: iter_time=8.308e-05, forward_time=0.407, loss_ctc=48.797, loss_att=46.661, acc=0.775, loss=47.301, backward_time=0.348, grad_norm=37.848, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.784e-04, train_time=1.664
+[gpub011:0/16] 2024-02-06 22:29:51,877 (trainer:762) INFO: 32epoch:train:6601-6700batch: iter_time=7.562e-05, forward_time=0.291, loss_ctc=49.280, loss_att=46.598, acc=0.757, loss=47.403, backward_time=0.298, grad_norm=42.543, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.784e-04, train_time=1.231
+[gpub011:0/16] 2024-02-06 22:32:27,477 (trainer:762) INFO: 32epoch:train:6701-6800batch: iter_time=7.626e-05, forward_time=0.316, loss_ctc=44.446, loss_att=46.100, acc=0.761, loss=45.604, backward_time=0.304, grad_norm=35.950, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.783e-04, train_time=1.555
+[gpub011:0/16] 2024-02-06 22:34:54,663 (trainer:762) INFO: 32epoch:train:6801-6900batch: iter_time=4.672e-04, forward_time=0.423, loss_ctc=39.945, loss_att=36.884, acc=0.768, loss=37.803, backward_time=0.333, grad_norm=33.523, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.783e-04, train_time=1.472
+[gpub011:0/16] 2024-02-06 22:36:43,199 (trainer:762) INFO: 32epoch:train:6901-7000batch: iter_time=8.291e-05, forward_time=0.290, loss_ctc=41.702, loss_att=39.654, acc=0.768, loss=40.268, backward_time=0.297, grad_norm=37.658, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.783e-04, train_time=1.086
+[gpub011:0/16] 2024-02-06 22:39:16,743 (trainer:762) INFO: 32epoch:train:7001-7100batch: iter_time=8.225e-05, forward_time=0.318, loss_ctc=50.046, loss_att=53.355, acc=0.732, loss=52.362, backward_time=0.306, grad_norm=42.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.783e-04, train_time=1.535
+[gpub011:0/16] 2024-02-06 22:41:44,728 (trainer:762) INFO: 32epoch:train:7101-7200batch: iter_time=2.829e-04, forward_time=0.353, loss_ctc=48.326, loss_att=48.577, acc=0.764, loss=48.502, backward_time=0.304, grad_norm=39.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.783e-04, train_time=1.479
+[gpub011:0/16] 2024-02-06 22:43:43,538 (trainer:762) INFO: 32epoch:train:7201-7300batch: iter_time=8.573e-05, forward_time=0.348, loss_ctc=51.697, loss_att=47.829, acc=0.770, loss=48.990, backward_time=0.323, grad_norm=47.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.782e-04, train_time=1.188
+[gpub011:0/16] 2024-02-06 22:46:17,419 (trainer:762) INFO: 32epoch:train:7301-7400batch: iter_time=8.873e-05, forward_time=0.310, loss_ctc=52.474, loss_att=53.287, acc=0.750, loss=53.043, backward_time=0.311, grad_norm=51.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.782e-04, train_time=1.539
+[gpub011:0/16] 2024-02-06 22:48:15,661 (trainer:762) INFO: 32epoch:train:7401-7500batch: iter_time=8.260e-05, forward_time=0.292, loss_ctc=49.393, loss_att=47.719, acc=0.763, loss=48.222, backward_time=0.298, grad_norm=45.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.782e-04, train_time=1.182
+[gpub011:0/16] 2024-02-06 22:48:35,763 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-06 22:48:54,944 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 22:48:58,501 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7910a063b0>)
+[gpub011:0/16] 2024-02-06 22:48:58,501 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-06 22:48:58,505 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 22:57:27,501 (trainer:762) INFO: 32epoch:train:7501-7600batch: iter_time=4.174, forward_time=0.416, loss_ctc=47.123, loss_att=48.158, acc=0.757, loss=47.847, backward_time=0.319, grad_norm=43.930, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.782e-04, train_time=5.519
+[gpub011:0/16] 2024-02-06 22:59:51,722 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-06 22:59:56,509 (trainer:762) INFO: 32epoch:train:7601-7700batch: iter_time=8.489e-05, forward_time=0.328, loss_ctc=47.372, loss_att=46.244, acc=0.761, loss=46.582, backward_time=0.299, grad_norm=37.364, clip=100.000, loss_scale=7.185e+33, optim_step_time=0.095, optim0_lr0=1.782e-04, train_time=1.490
+[gpub011:0/16] 2024-02-06 23:02:33,837 (trainer:762) INFO: 32epoch:train:7701-7800batch: iter_time=8.315e-05, forward_time=0.426, loss_ctc=46.100, loss_att=45.955, acc=0.769, loss=45.998, backward_time=0.328, grad_norm=36.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.781e-04, train_time=1.573
+[gpub011:0/16] 2024-02-06 23:05:21,064 (trainer:762) INFO: 32epoch:train:7801-7900batch: iter_time=9.875e-05, forward_time=0.290, loss_ctc=51.453, loss_att=50.392, acc=0.753, loss=50.710, backward_time=0.297, grad_norm=40.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.781e-04, train_time=1.672
+[gpub011:0/16] 2024-02-06 23:07:37,242 (trainer:762) INFO: 32epoch:train:7901-8000batch: iter_time=1.008e-04, forward_time=0.361, loss_ctc=43.288, loss_att=37.282, acc=0.780, loss=39.084, backward_time=0.334, grad_norm=38.024, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.781e-04, train_time=1.362
+[gpub011:0/16] 2024-02-06 23:10:00,440 (trainer:762) INFO: 32epoch:train:8001-8100batch: iter_time=8.520e-05, forward_time=0.343, loss_ctc=40.620, loss_att=41.499, acc=0.760, loss=41.235, backward_time=0.313, grad_norm=34.539, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.781e-04, train_time=1.432
+[gpub011:0/16] 2024-02-06 23:12:24,690 (trainer:762) INFO: 32epoch:train:8101-8200batch: iter_time=8.369e-05, forward_time=0.288, loss_ctc=43.455, loss_att=39.728, acc=0.767, loss=40.846, backward_time=0.295, grad_norm=36.253, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.781e-04, train_time=1.442
+[gpub011:0/16] 2024-02-06 23:15:19,177 (trainer:762) INFO: 32epoch:train:8201-8300batch: iter_time=8.771e-05, forward_time=0.480, loss_ctc=47.638, loss_att=48.818, acc=0.744, loss=48.464, backward_time=0.324, grad_norm=39.813, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.781e-04, train_time=1.745
+[gpub011:0/16] 2024-02-06 23:17:33,995 (trainer:762) INFO: 32epoch:train:8301-8400batch: iter_time=8.210e-05, forward_time=0.293, loss_ctc=50.958, loss_att=55.356, acc=0.748, loss=54.037, backward_time=0.299, grad_norm=44.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.780e-04, train_time=1.348
+[gpub011:0/16] 2024-02-06 23:20:10,119 (trainer:762) INFO: 32epoch:train:8401-8500batch: iter_time=8.892e-05, forward_time=0.288, loss_ctc=44.355, loss_att=41.185, acc=0.775, loss=42.136, backward_time=0.295, grad_norm=38.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.780e-04, train_time=1.561
+[gpub011:0/16] 2024-02-06 23:22:49,917 (trainer:762) INFO: 32epoch:train:8501-8600batch: iter_time=0.007, forward_time=0.438, loss_ctc=57.611, loss_att=56.446, acc=0.755, loss=56.795, backward_time=0.333, grad_norm=53.687, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.780e-04, train_time=1.598
+[gpub011:0/16] 2024-02-06 23:25:09,228 (trainer:762) INFO: 32epoch:train:8601-8700batch: iter_time=8.330e-05, forward_time=0.288, loss_ctc=50.042, loss_att=49.681, acc=0.759, loss=49.790, backward_time=0.297, grad_norm=42.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.780e-04, train_time=1.393
+[gpub011:0/16] 2024-02-06 23:26:46,882 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-06 23:27:06,434 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-06 23:27:10,032 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7910a537c0>)
+[gpub011:0/16] 2024-02-06 23:27:10,032 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-06 23:27:10,092 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-06 23:34:31,336 (trainer:762) INFO: 32epoch:train:8701-8800batch: iter_time=4.058, forward_time=0.400, loss_ctc=43.871, loss_att=47.380, acc=0.757, loss=46.327, backward_time=0.317, grad_norm=41.694, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.780e-04, train_time=5.621
+[gpub011:0/16] 2024-02-06 23:36:50,544 (trainer:762) INFO: 32epoch:train:8801-8900batch: iter_time=7.128e-05, forward_time=0.288, loss_ctc=44.617, loss_att=41.116, acc=0.754, loss=42.166, backward_time=0.294, grad_norm=39.132, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.779e-04, train_time=1.392
+[gpub011:0/16] 2024-02-06 23:39:20,716 (trainer:762) INFO: 32epoch:train:8901-9000batch: iter_time=7.888e-05, forward_time=0.290, loss_ctc=48.679, loss_att=46.691, acc=0.756, loss=47.287, backward_time=0.295, grad_norm=38.250, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.779e-04, train_time=1.501
+[gpub011:0/16] 2024-02-06 23:42:06,189 (trainer:762) INFO: 32epoch:train:9001-9100batch: iter_time=7.792e-05, forward_time=0.395, loss_ctc=48.533, loss_att=44.724, acc=0.773, loss=45.866, backward_time=0.353, grad_norm=37.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.779e-04, train_time=1.655
+[gpub011:0/16] 2024-02-06 23:44:18,634 (trainer:762) INFO: 32epoch:train:9101-9200batch: iter_time=8.019e-05, forward_time=0.291, loss_ctc=48.901, loss_att=46.467, acc=0.755, loss=47.197, backward_time=0.298, grad_norm=44.229, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.779e-04, train_time=1.324
+[gpub011:0/16] 2024-02-06 23:46:30,935 (trainer:762) INFO: 32epoch:train:9201-9300batch: iter_time=8.896e-05, forward_time=0.291, loss_ctc=44.851, loss_att=42.776, acc=0.759, loss=43.398, backward_time=0.296, grad_norm=36.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.779e-04, train_time=1.323
+[gpub011:0/16] 2024-02-06 23:49:30,644 (trainer:762) INFO: 32epoch:train:9301-9400batch: iter_time=9.353e-05, forward_time=0.387, loss_ctc=39.453, loss_att=36.672, acc=0.767, loss=37.506, backward_time=0.330, grad_norm=34.426, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.124, optim0_lr0=1.778e-04, train_time=1.797
+[gpub011:0/16] 2024-02-06 23:51:47,883 (trainer:762) INFO: 32epoch:train:9401-9500batch: iter_time=8.513e-05, forward_time=0.289, loss_ctc=41.518, loss_att=40.198, acc=0.763, loss=40.594, backward_time=0.296, grad_norm=49.644, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.778e-04, train_time=1.372
+[gpub011:0/16] 2024-02-06 23:54:01,449 (trainer:762) INFO: 32epoch:train:9501-9600batch: iter_time=8.983e-05, forward_time=0.289, loss_ctc=50.200, loss_att=53.217, acc=0.720, loss=52.312, backward_time=0.296, grad_norm=42.879, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.778e-04, train_time=1.335
+[gpub011:0/16] 2024-02-06 23:56:44,642 (trainer:762) INFO: 32epoch:train:9601-9700batch: iter_time=9.458e-05, forward_time=0.353, loss_ctc=48.082, loss_att=48.523, acc=0.750, loss=48.390, backward_time=0.331, grad_norm=40.858, clip=100.000, loss_scale=5.400e+33, optim_step_time=0.106, optim0_lr0=1.778e-04, train_time=1.632
+[gpub011:0/16] 2024-02-06 23:59:05,873 (trainer:762) INFO: 32epoch:train:9701-9800batch: iter_time=8.444e-05, forward_time=0.316, loss_ctc=52.478, loss_att=48.840, acc=0.757, loss=49.932, backward_time=0.326, grad_norm=44.351, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.778e-04, train_time=1.412
+[gpub011:0/16] 2024-02-07 00:01:13,189 (trainer:762) INFO: 32epoch:train:9801-9900batch: iter_time=8.076e-05, forward_time=0.291, loss_ctc=51.652, loss_att=53.245, acc=0.745, loss=52.767, backward_time=0.296, grad_norm=49.721, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.778e-04, train_time=1.273
+[gpub011:0/16] 2024-02-07 00:03:42,639 (trainer:762) INFO: 32epoch:train:9901-10000batch: iter_time=8.277e-05, forward_time=0.304, loss_ctc=49.049, loss_att=45.406, acc=0.763, loss=46.499, backward_time=0.312, grad_norm=45.332, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.777e-04, train_time=1.494
+[gpub011:0/16] 2024-02-07 00:04:02,772 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-07 00:04:21,924 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 00:04:25,496 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78c8ac2e00>)
+[gpub011:0/16] 2024-02-07 00:04:25,496 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-07 00:04:25,499 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 00:13:53,571 (trainer:762) INFO: 32epoch:train:10001-10100batch: iter_time=4.185, forward_time=0.455, loss_ctc=46.464, loss_att=49.103, acc=0.757, loss=48.311, backward_time=0.328, grad_norm=42.274, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.777e-04, train_time=6.110
+[gpub011:0/16] 2024-02-07 00:16:44,246 (trainer:762) INFO: 32epoch:train:10101-10200batch: iter_time=8.569e-05, forward_time=0.290, loss_ctc=46.880, loss_att=46.030, acc=0.762, loss=46.285, backward_time=0.295, grad_norm=40.640, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.777e-04, train_time=1.706
+[gpub011:0/16] 2024-02-07 00:19:35,570 (trainer:762) INFO: 32epoch:train:10201-10300batch: iter_time=8.956e-05, forward_time=0.292, loss_ctc=46.086, loss_att=46.269, acc=0.769, loss=46.214, backward_time=0.298, grad_norm=37.001, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.777e-04, train_time=1.713
+[gpub011:0/16] 2024-02-07 00:22:33,468 (trainer:762) INFO: 32epoch:train:10301-10400batch: iter_time=9.497e-05, forward_time=0.504, loss_ctc=51.425, loss_att=50.263, acc=0.754, loss=50.612, backward_time=0.339, grad_norm=41.020, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.777e-04, train_time=1.778
+[gpub011:0/16] 2024-02-07 00:25:41,551 (trainer:762) INFO: 32epoch:train:10401-10500batch: iter_time=9.862e-05, forward_time=0.287, loss_ctc=43.053, loss_att=37.257, acc=0.779, loss=38.996, backward_time=0.292, grad_norm=37.019, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.776e-04, train_time=1.881
+[gpub011:0/16] 2024-02-07 00:27:58,448 (trainer:762) INFO: 32epoch:train:10501-10600batch: iter_time=9.735e-05, forward_time=0.288, loss_ctc=40.102, loss_att=41.379, acc=0.761, loss=40.996, backward_time=0.304, grad_norm=34.512, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.776e-04, train_time=1.368
+[gpub011:0/16] 2024-02-07 00:30:55,426 (trainer:762) INFO: 32epoch:train:10601-10700batch: iter_time=9.176e-05, forward_time=0.362, loss_ctc=43.687, loss_att=39.894, acc=0.768, loss=41.032, backward_time=0.389, grad_norm=34.826, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.776e-04, train_time=1.770
+[gpub011:0/16] 2024-02-07 00:33:40,019 (trainer:762) INFO: 32epoch:train:10701-10800batch: iter_time=9.083e-05, forward_time=0.289, loss_ctc=47.178, loss_att=48.321, acc=0.746, loss=47.978, backward_time=0.295, grad_norm=39.689, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.776e-04, train_time=1.646
+[gpub011:0/16] 2024-02-07 00:36:46,658 (trainer:762) INFO: 32epoch:train:10801-10900batch: iter_time=9.418e-05, forward_time=0.373, loss_ctc=50.616, loss_att=55.304, acc=0.748, loss=53.898, backward_time=0.480, grad_norm=44.333, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.776e-04, train_time=1.866
+[gpub011:0/16] 2024-02-07 00:39:48,438 (trainer:762) INFO: 32epoch:train:10901-11000batch: iter_time=8.062e-05, forward_time=0.290, loss_ctc=44.315, loss_att=40.722, acc=0.778, loss=41.800, backward_time=0.292, grad_norm=37.601, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.775e-04, train_time=1.818
+[gpub011:0/16] 2024-02-07 00:42:28,885 (trainer:762) INFO: 32epoch:train:11001-11100batch: iter_time=9.044e-05, forward_time=0.293, loss_ctc=57.783, loss_att=56.202, acc=0.756, loss=56.676, backward_time=0.299, grad_norm=55.729, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.775e-04, train_time=1.604
+[gpub011:0/16] 2024-02-07 00:45:16,046 (trainer:762) INFO: 32epoch:train:11101-11200batch: iter_time=4.002e-04, forward_time=0.422, loss_ctc=50.049, loss_att=50.002, acc=0.757, loss=50.016, backward_time=0.320, grad_norm=42.951, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.121, optim0_lr0=1.775e-04, train_time=1.671
+[gpub011:0/16] 2024-02-07 00:47:18,784 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-07 00:47:38,168 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 00:47:41,798 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c0d733fa0>)
+[gpub011:0/16] 2024-02-07 00:47:41,799 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-07 00:47:41,802 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 00:55:24,482 (trainer:762) INFO: 32epoch:train:11201-11300batch: iter_time=4.302, forward_time=0.435, loss_ctc=44.036, loss_att=46.879, acc=0.762, loss=46.026, backward_time=0.324, grad_norm=41.802, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.775e-04, train_time=6.085
+[gpub011:0/16] 2024-02-07 00:57:50,762 (trainer:762) INFO: 32epoch:train:11301-11400batch: iter_time=8.486e-05, forward_time=0.288, loss_ctc=44.898, loss_att=43.150, acc=0.762, loss=43.675, backward_time=0.294, grad_norm=40.471, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.775e-04, train_time=1.463
+[gpub011:0/16] 2024-02-07 01:00:04,740 (trainer:762) INFO: 32epoch:train:11401-11500batch: iter_time=8.010e-05, forward_time=0.290, loss_ctc=48.358, loss_att=46.149, acc=0.759, loss=46.812, backward_time=0.297, grad_norm=37.311, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.775e-04, train_time=1.340
+[gpub011:0/16] 2024-02-07 01:03:06,191 (trainer:762) INFO: 32epoch:train:11501-11600batch: iter_time=8.636e-05, forward_time=0.455, loss_ctc=48.483, loss_att=45.709, acc=0.778, loss=46.541, backward_time=0.336, grad_norm=37.217, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=1.774e-04, train_time=1.814
+[gpub011:0/16] 2024-02-07 01:05:45,613 (trainer:762) INFO: 32epoch:train:11601-11700batch: iter_time=8.442e-05, forward_time=0.291, loss_ctc=48.644, loss_att=46.043, acc=0.758, loss=46.823, backward_time=0.296, grad_norm=40.931, clip=100.000, loss_scale=1.080e+34, optim_step_time=0.094, optim0_lr0=1.774e-04, train_time=1.594
+[gpub011:0/16] 2024-02-07 01:07:42,764 (trainer:762) INFO: 32epoch:train:11701-11800batch: iter_time=8.233e-05, forward_time=0.289, loss_ctc=44.553, loss_att=45.574, acc=0.762, loss=45.268, backward_time=0.298, grad_norm=37.144, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.094, optim0_lr0=1.774e-04, train_time=1.171
+[gpub011:0/16] 2024-02-07 01:10:28,344 (trainer:762) INFO: 32epoch:train:11801-11900batch: iter_time=8.857e-05, forward_time=0.374, loss_ctc=39.446, loss_att=36.510, acc=0.770, loss=37.391, backward_time=0.389, grad_norm=34.917, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.105, optim0_lr0=1.774e-04, train_time=1.655
+[gpub011:0/16] 2024-02-07 01:12:41,638 (trainer:762) INFO: 32epoch:train:11901-12000batch: iter_time=8.186e-05, forward_time=0.293, loss_ctc=41.093, loss_att=39.459, acc=0.769, loss=39.949, backward_time=0.295, grad_norm=36.770, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.095, optim0_lr0=1.774e-04, train_time=1.332
+[gpub011:0/16] 2024-02-07 01:15:20,936 (trainer:762) INFO: 32epoch:train:12001-12100batch: iter_time=8.386e-05, forward_time=0.289, loss_ctc=49.486, loss_att=52.730, acc=0.734, loss=51.757, backward_time=0.294, grad_norm=42.297, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.095, optim0_lr0=1.773e-04, train_time=1.594
+[gpub011:0/16] 2024-02-07 01:18:17,324 (trainer:762) INFO: 32epoch:train:12101-12200batch: iter_time=8.502e-05, forward_time=0.350, loss_ctc=47.297, loss_att=47.919, acc=0.767, loss=47.733, backward_time=0.366, grad_norm=38.471, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.105, optim0_lr0=1.773e-04, train_time=1.764
+[gpub011:0/16] 2024-02-07 01:20:53,874 (trainer:762) INFO: 32epoch:train:12201-12300batch: iter_time=8.402e-05, forward_time=0.291, loss_ctc=51.491, loss_att=47.765, acc=0.772, loss=48.883, backward_time=0.299, grad_norm=40.947, clip=100.000, loss_scale=2.077e+34, optim_step_time=0.094, optim0_lr0=1.773e-04, train_time=1.564
+[gpub011:0/16] 2024-02-07 01:21:10,120 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 01:23:14,252 (trainer:762) INFO: 32epoch:train:12301-12400batch: iter_time=8.707e-05, forward_time=0.288, loss_ctc=51.392, loss_att=52.493, acc=0.753, loss=52.163, backward_time=0.297, grad_norm=46.862, clip=100.000, loss_scale=1.154e+34, optim_step_time=0.095, optim0_lr0=1.773e-04, train_time=1.404
+[gpub011:0/16] 2024-02-07 01:25:29,481 (trainer:762) INFO: 32epoch:train:12401-12500batch: iter_time=0.001, forward_time=0.414, loss_ctc=49.091, loss_att=47.047, acc=0.765, loss=47.660, backward_time=0.343, grad_norm=43.552, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=1.773e-04, train_time=1.352
+[gpub011:0/16] 2024-02-07 01:25:49,556 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-07 01:26:09,421 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 01:26:13,011 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c0d7f0af0>)
+[gpub011:0/16] 2024-02-07 01:26:13,011 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-07 01:26:13,014 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 01:34:56,610 (trainer:762) INFO: 32epoch:train:12501-12600batch: iter_time=4.193, forward_time=0.310, loss_ctc=46.743, loss_att=48.778, acc=0.750, loss=48.167, backward_time=0.296, grad_norm=41.492, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.772e-04, train_time=5.671
+[gpub011:0/16] 2024-02-07 01:36:46,034 (trainer:762) INFO: 32epoch:train:12601-12700batch: iter_time=7.629e-05, forward_time=0.292, loss_ctc=46.676, loss_att=44.967, acc=0.757, loss=45.479, backward_time=0.299, grad_norm=38.495, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.772e-04, train_time=1.094
+[gpub011:0/16] 2024-02-07 01:40:11,883 (trainer:762) INFO: 32epoch:train:12701-12800batch: iter_time=8.396e-05, forward_time=0.469, loss_ctc=45.843, loss_att=46.282, acc=0.764, loss=46.150, backward_time=0.342, grad_norm=37.737, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.123, optim0_lr0=1.772e-04, train_time=2.058
+[gpub011:0/16] 2024-02-07 01:41:13,373 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 01:42:02,555 (trainer:762) INFO: 32epoch:train:12801-12900batch: iter_time=8.481e-05, forward_time=0.290, loss_ctc=51.188, loss_att=48.653, acc=0.750, loss=49.414, backward_time=0.298, grad_norm=42.773, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.094, optim0_lr0=1.772e-04, train_time=1.107
+[gpub011:0/16] 2024-02-07 01:44:41,685 (trainer:762) INFO: 32epoch:train:12901-13000batch: iter_time=8.516e-05, forward_time=0.317, loss_ctc=42.631, loss_att=37.040, acc=0.780, loss=38.717, backward_time=0.292, grad_norm=36.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.772e-04, train_time=1.591
+[gpub011:0/16] 2024-02-07 01:47:00,836 (trainer:762) INFO: 32epoch:train:13001-13100batch: iter_time=8.451e-05, forward_time=0.317, loss_ctc=40.289, loss_att=38.652, acc=0.758, loss=39.143, backward_time=0.295, grad_norm=34.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.772e-04, train_time=1.392
+[gpub011:0/16] 2024-02-07 01:49:23,009 (trainer:762) INFO: 32epoch:train:13101-13200batch: iter_time=8.416e-05, forward_time=0.477, loss_ctc=43.557, loss_att=40.463, acc=0.762, loss=41.391, backward_time=0.337, grad_norm=37.424, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.771e-04, train_time=1.422
+[gpub011:0/16] 2024-02-07 01:52:15,222 (trainer:762) INFO: 32epoch:train:13201-13300batch: iter_time=8.139e-05, forward_time=0.291, loss_ctc=47.044, loss_att=48.783, acc=0.739, loss=48.261, backward_time=0.293, grad_norm=40.207, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.771e-04, train_time=1.722
+[gpub011:0/16] 2024-02-07 01:54:07,468 (trainer:762) INFO: 32epoch:train:13301-13400batch: iter_time=0.003, forward_time=0.317, loss_ctc=49.787, loss_att=54.632, acc=0.736, loss=53.178, backward_time=0.298, grad_norm=45.113, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.771e-04, train_time=1.122
+[gpub011:0/16] 2024-02-07 01:57:11,251 (trainer:762) INFO: 32epoch:train:13401-13500batch: iter_time=1.908e-04, forward_time=0.487, loss_ctc=43.478, loss_att=41.437, acc=0.763, loss=42.050, backward_time=0.335, grad_norm=37.793, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.771e-04, train_time=1.837
+[gpub011:0/16] 2024-02-07 01:59:37,049 (trainer:762) INFO: 32epoch:train:13501-13600batch: iter_time=8.343e-05, forward_time=0.295, loss_ctc=57.288, loss_att=56.055, acc=0.747, loss=56.425, backward_time=0.300, grad_norm=53.120, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.771e-04, train_time=1.458
+[gpub011:0/16] 2024-02-07 02:02:46,084 (trainer:762) INFO: 32epoch:train:13601-13700batch: iter_time=8.382e-05, forward_time=0.289, loss_ctc=49.822, loss_att=48.553, acc=0.756, loss=48.934, backward_time=0.293, grad_norm=41.790, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.770e-04, train_time=1.890
+[gpub011:0/16] 2024-02-07 02:04:26,381 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-07 02:04:46,102 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 02:04:49,712 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78a64022c0>)
+[gpub011:0/16] 2024-02-07 02:04:49,713 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-07 02:04:49,716 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 02:12:08,558 (trainer:762) INFO: 32epoch:train:13701-13800batch: iter_time=4.017, forward_time=0.454, loss_ctc=43.502, loss_att=46.051, acc=0.759, loss=45.286, backward_time=0.365, grad_norm=41.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.770e-04, train_time=5.624
+[gpub011:0/16] 2024-02-07 02:14:29,269 (trainer:762) INFO: 32epoch:train:13801-13900batch: iter_time=7.444e-05, forward_time=0.286, loss_ctc=44.554, loss_att=40.434, acc=0.754, loss=41.670, backward_time=0.294, grad_norm=39.844, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.770e-04, train_time=1.407
+[gpub011:0/16] 2024-02-07 02:16:50,238 (trainer:762) INFO: 32epoch:train:13901-14000batch: iter_time=8.325e-05, forward_time=0.290, loss_ctc=48.991, loss_att=46.120, acc=0.759, loss=46.982, backward_time=0.297, grad_norm=39.046, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.770e-04, train_time=1.409
+[gpub011:0/16] 2024-02-07 02:19:53,929 (trainer:762) INFO: 32epoch:train:14001-14100batch: iter_time=7.065e-04, forward_time=0.492, loss_ctc=48.504, loss_att=44.033, acc=0.777, loss=45.374, backward_time=0.357, grad_norm=37.615, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.770e-04, train_time=1.834
+[gpub011:0/16] 2024-02-07 02:23:12,079 (trainer:762) INFO: 32epoch:train:14101-14200batch: iter_time=8.101e-05, forward_time=0.290, loss_ctc=48.670, loss_att=45.842, acc=0.756, loss=46.690, backward_time=0.295, grad_norm=40.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.770e-04, train_time=1.984
+[gpub011:0/16] 2024-02-07 02:25:30,752 (trainer:762) INFO: 32epoch:train:14201-14300batch: iter_time=8.172e-05, forward_time=0.290, loss_ctc=43.942, loss_att=42.335, acc=0.760, loss=42.817, backward_time=0.296, grad_norm=36.376, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.769e-04, train_time=1.387
+[gpub011:0/16] 2024-02-07 02:28:56,239 (trainer:762) INFO: 32epoch:train:14301-14400batch: iter_time=8.816e-05, forward_time=0.476, loss_ctc=39.539, loss_att=36.432, acc=0.767, loss=37.364, backward_time=0.335, grad_norm=34.444, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.769e-04, train_time=2.054
+[gpub011:0/16] 2024-02-07 02:31:26,790 (trainer:762) INFO: 32epoch:train:14401-14500batch: iter_time=8.171e-05, forward_time=0.287, loss_ctc=40.768, loss_att=39.218, acc=0.766, loss=39.683, backward_time=0.294, grad_norm=36.546, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.769e-04, train_time=1.504
+[gpub011:0/16] 2024-02-07 02:33:47,155 (trainer:762) INFO: 32epoch:train:14501-14600batch: iter_time=8.129e-05, forward_time=0.288, loss_ctc=49.482, loss_att=52.007, acc=0.724, loss=51.250, backward_time=0.295, grad_norm=44.006, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.769e-04, train_time=1.405
+[gpub011:0/16] 2024-02-07 02:37:17,914 (trainer:762) INFO: 32epoch:train:14601-14700batch: iter_time=2.222e-04, forward_time=0.431, loss_ctc=47.930, loss_att=48.852, acc=0.749, loss=48.576, backward_time=0.325, grad_norm=41.734, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.769e-04, train_time=2.107
+[gpub011:0/16] 2024-02-07 02:39:32,130 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 02:39:33,289 (trainer:762) INFO: 32epoch:train:14701-14800batch: iter_time=8.094e-05, forward_time=0.290, loss_ctc=50.739, loss_att=48.046, acc=0.759, loss=48.854, backward_time=0.298, grad_norm=42.282, clip=100.000, loss_scale=5.166e+33, optim_step_time=0.095, optim0_lr0=1.768e-04, train_time=1.354
+[gpub011:0/16] 2024-02-07 02:41:49,380 (trainer:762) INFO: 32epoch:train:14801-14900batch: iter_time=7.921e-05, forward_time=0.291, loss_ctc=51.085, loss_att=52.604, acc=0.747, loss=52.148, backward_time=0.297, grad_norm=47.546, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.768e-04, train_time=1.359
+[gpub011:0/16] 2024-02-07 02:45:24,409 (trainer:762) INFO: 32epoch:train:14901-15000batch: iter_time=1.417e-04, forward_time=0.375, loss_ctc=48.661, loss_att=46.015, acc=0.761, loss=46.809, backward_time=0.462, grad_norm=42.854, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.768e-04, train_time=2.151
+[gpub011:0/16] 2024-02-07 03:24:11,724 (trainer:361) INFO: 32epoch results: [train] iter_time=0.329, forward_time=0.337, loss_ctc=47.595, loss_att=46.402, acc=0.757, loss=46.760, backward_time=0.314, grad_norm=40.971, clip=100.000, loss_scale=7.002e+33, optim_step_time=0.099, optim0_lr0=1.782e-04, train_time=1.822, time=7 hours, 36 minutes and 2.07 seconds, total_count=510000, gpu_max_cached_mem_GB=41.025, [valid] loss_ctc=35.374, cer_ctc=0.183, loss_att=36.633, acc=0.691, cer=0.285, wer=0.991, loss=36.256, time=38 minutes and 22.69 seconds, total_count=158814, gpu_max_cached_mem_GB=41.025
+[gpub011:0/16] 2024-02-07 03:24:22,066 (trainer:416) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub011:0/16] 2024-02-07 03:24:22,115 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/22epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/27epoch.pth
+[gpub011:0/16] 2024-02-07 03:24:22,115 (trainer:290) INFO: 33/45epoch started. Estimated time to finish: 4 days, 7 hours and 53 minutes
+[gpub011:0/16] 2024-02-07 03:24:22,126 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-07 03:24:41,154 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 03:24:44,753 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7231454a00>)
+[gpub011:0/16] 2024-02-07 03:24:44,753 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-07 03:24:44,757 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 03:32:08,086 (trainer:762) INFO: 33epoch:train:1-100batch: iter_time=3.384, forward_time=0.410, loss_ctc=54.758, loss_att=52.377, acc=0.749, loss=53.091, backward_time=0.314, grad_norm=57.261, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.768e-04, train_time=4.659
+[gpub011:0/16] 2024-02-07 03:34:37,824 (trainer:762) INFO: 33epoch:train:101-200batch: iter_time=9.064e-05, forward_time=0.292, loss_ctc=43.453, loss_att=45.231, acc=0.772, loss=44.697, backward_time=0.296, grad_norm=36.954, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.768e-04, train_time=1.497
+[gpub011:0/16] 2024-02-07 03:37:24,443 (trainer:762) INFO: 33epoch:train:201-300batch: iter_time=8.309e-05, forward_time=0.323, loss_ctc=38.588, loss_att=38.635, acc=0.758, loss=38.621, backward_time=0.303, grad_norm=34.128, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.768e-04, train_time=1.666
+[gpub011:0/16] 2024-02-07 03:39:44,321 (trainer:762) INFO: 33epoch:train:301-400batch: iter_time=8.845e-05, forward_time=0.389, loss_ctc=46.180, loss_att=44.825, acc=0.763, loss=45.231, backward_time=0.307, grad_norm=40.561, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.767e-04, train_time=1.399
+[gpub011:0/16] 2024-02-07 03:42:17,520 (trainer:762) INFO: 33epoch:train:401-500batch: iter_time=9.048e-05, forward_time=0.290, loss_ctc=44.692, loss_att=39.778, acc=0.761, loss=41.252, backward_time=0.293, grad_norm=43.749, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.767e-04, train_time=1.532
+[gpub011:0/16] 2024-02-07 03:44:24,488 (trainer:762) INFO: 33epoch:train:501-600batch: iter_time=8.376e-05, forward_time=0.303, loss_ctc=51.935, loss_att=51.103, acc=0.738, loss=51.353, backward_time=0.305, grad_norm=46.306, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.767e-04, train_time=1.269
+[gpub011:0/16] 2024-02-07 03:47:20,141 (trainer:762) INFO: 33epoch:train:601-700batch: iter_time=8.229e-05, forward_time=0.388, loss_ctc=43.189, loss_att=41.058, acc=0.768, loss=41.698, backward_time=0.341, grad_norm=35.520, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=1.767e-04, train_time=1.756
+[gpub011:0/16] 2024-02-07 03:49:22,329 (trainer:762) INFO: 33epoch:train:701-800batch: iter_time=8.435e-05, forward_time=0.298, loss_ctc=51.420, loss_att=50.426, acc=0.738, loss=50.724, backward_time=0.304, grad_norm=43.462, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.767e-04, train_time=1.222
+[gpub011:0/16] 2024-02-07 03:51:38,302 (trainer:762) INFO: 33epoch:train:801-900batch: iter_time=7.986e-05, forward_time=0.293, loss_ctc=43.078, loss_att=39.074, acc=0.757, loss=40.275, backward_time=0.296, grad_norm=39.667, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.766e-04, train_time=1.360
+[gpub011:0/16] 2024-02-07 03:53:48,283 (trainer:762) INFO: 33epoch:train:901-1000batch: iter_time=8.112e-05, forward_time=0.319, loss_ctc=49.383, loss_att=45.825, acc=0.759, loss=46.893, backward_time=0.305, grad_norm=47.822, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.766e-04, train_time=1.300
+[gpub011:0/16] 2024-02-07 03:56:20,920 (trainer:762) INFO: 33epoch:train:1001-1100batch: iter_time=8.014e-05, forward_time=0.331, loss_ctc=47.513, loss_att=53.905, acc=0.733, loss=51.987, backward_time=0.357, grad_norm=42.537, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.766e-04, train_time=1.525
+[gpub011:0/16] 2024-02-07 03:58:36,769 (trainer:762) INFO: 33epoch:train:1101-1200batch: iter_time=7.826e-05, forward_time=0.291, loss_ctc=43.589, loss_att=44.628, acc=0.754, loss=44.316, backward_time=0.295, grad_norm=38.426, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.766e-04, train_time=1.359
+[gpub011:0/16] 2024-02-07 03:59:59,429 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-07 04:00:18,934 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 04:00:22,788 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f71fc933e80>)
+[gpub011:0/16] 2024-02-07 04:00:22,788 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub011:0/16] 2024-02-07 04:00:22,791 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 04:06:40,462 (trainer:762) INFO: 33epoch:train:1201-1300batch: iter_time=3.467, forward_time=0.335, loss_ctc=47.919, loss_att=46.715, acc=0.767, loss=47.076, backward_time=0.309, grad_norm=38.638, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.766e-04, train_time=4.837
+[gpub011:0/16] 2024-02-07 04:09:16,610 (trainer:762) INFO: 33epoch:train:1301-1400batch: iter_time=8.603e-05, forward_time=0.338, loss_ctc=42.785, loss_att=41.247, acc=0.769, loss=41.709, backward_time=0.308, grad_norm=48.851, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.765e-04, train_time=1.561
+[gpub011:0/16] 2024-02-07 04:11:20,842 (trainer:762) INFO: 33epoch:train:1401-1500batch: iter_time=7.885e-05, forward_time=0.290, loss_ctc=44.192, loss_att=45.119, acc=0.767, loss=44.841, backward_time=0.296, grad_norm=35.573, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.765e-04, train_time=1.242
+[gpub011:0/16] 2024-02-07 04:13:45,320 (trainer:762) INFO: 33epoch:train:1501-1600batch: iter_time=7.771e-05, forward_time=0.292, loss_ctc=42.895, loss_att=40.070, acc=0.755, loss=40.918, backward_time=0.297, grad_norm=37.156, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.765e-04, train_time=1.445
+[gpub011:0/16] 2024-02-07 04:16:02,725 (trainer:762) INFO: 33epoch:train:1601-1700batch: iter_time=8.815e-05, forward_time=0.312, loss_ctc=44.009, loss_att=43.664, acc=0.746, loss=43.767, backward_time=0.324, grad_norm=41.142, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.765e-04, train_time=1.374
+[gpub011:0/16] 2024-02-07 04:18:53,818 (trainer:762) INFO: 33epoch:train:1701-1800batch: iter_time=8.873e-05, forward_time=0.382, loss_ctc=51.410, loss_att=45.763, acc=0.755, loss=47.457, backward_time=0.303, grad_norm=42.954, clip=100.000, loss_scale=2.622e+33, optim_step_time=0.131, optim0_lr0=1.765e-04, train_time=1.711
+[gpub011:0/16] 2024-02-07 04:21:06,022 (trainer:762) INFO: 33epoch:train:1801-1900batch: iter_time=9.146e-05, forward_time=0.287, loss_ctc=45.743, loss_att=42.914, acc=0.743, loss=43.763, backward_time=0.293, grad_norm=40.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.765e-04, train_time=1.322
+[gpub011:0/16] 2024-02-07 04:23:26,942 (trainer:762) INFO: 33epoch:train:1901-2000batch: iter_time=8.782e-05, forward_time=0.332, loss_ctc=45.452, loss_att=47.758, acc=0.739, loss=47.066, backward_time=0.311, grad_norm=39.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.764e-04, train_time=1.409
+[gpub011:0/16] 2024-02-07 04:26:16,414 (trainer:762) INFO: 33epoch:train:2001-2100batch: iter_time=8.359e-05, forward_time=0.425, loss_ctc=44.169, loss_att=39.793, acc=0.757, loss=41.106, backward_time=0.302, grad_norm=38.235, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.764e-04, train_time=1.695
+[gpub011:0/16] 2024-02-07 04:28:13,563 (trainer:762) INFO: 33epoch:train:2101-2200batch: iter_time=8.344e-05, forward_time=0.290, loss_ctc=51.202, loss_att=50.205, acc=0.726, loss=50.504, backward_time=0.295, grad_norm=50.712, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.764e-04, train_time=1.171
+[gpub011:0/16] 2024-02-07 04:30:41,776 (trainer:762) INFO: 33epoch:train:2201-2300batch: iter_time=8.247e-05, forward_time=0.288, loss_ctc=44.153, loss_att=42.818, acc=0.761, loss=43.219, backward_time=0.295, grad_norm=37.223, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.764e-04, train_time=1.482
+[gpub011:0/16] 2024-02-07 04:33:08,235 (trainer:762) INFO: 33epoch:train:2301-2400batch: iter_time=7.642e-05, forward_time=0.325, loss_ctc=47.628, loss_att=53.193, acc=0.734, loss=51.523, backward_time=0.310, grad_norm=39.733, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.764e-04, train_time=1.464
+[gpub011:0/16] 2024-02-07 04:35:27,559 (trainer:762) INFO: 33epoch:train:2401-2500batch: iter_time=9.193e-05, forward_time=0.334, loss_ctc=42.135, loss_att=39.916, acc=0.761, loss=40.581, backward_time=0.341, grad_norm=36.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.763e-04, train_time=1.393
+[gpub011:0/16] 2024-02-07 04:35:47,609 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-07 04:36:06,563 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 04:36:10,142 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f794aad7670>)
+[gpub011:0/16] 2024-02-07 04:36:10,142 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub011:0/16] 2024-02-07 04:36:10,145 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 04:44:12,790 (trainer:762) INFO: 33epoch:train:2501-2600batch: iter_time=4.039, forward_time=0.291, loss_ctc=49.418, loss_att=50.917, acc=0.754, loss=50.467, backward_time=0.296, grad_norm=52.029, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.763e-04, train_time=5.252
+[gpub011:0/16] 2024-02-07 04:47:02,741 (trainer:762) INFO: 33epoch:train:2601-2700batch: iter_time=8.202e-05, forward_time=0.293, loss_ctc=42.309, loss_att=44.118, acc=0.777, loss=43.575, backward_time=0.298, grad_norm=34.433, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.763e-04, train_time=1.699
+[gpub011:0/16] 2024-02-07 04:49:11,319 (trainer:762) INFO: 33epoch:train:2701-2800batch: iter_time=3.836e-04, forward_time=0.389, loss_ctc=38.022, loss_att=38.128, acc=0.763, loss=38.096, backward_time=0.346, grad_norm=33.455, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.763e-04, train_time=1.286
+[gpub011:0/16] 2024-02-07 04:51:49,682 (trainer:762) INFO: 33epoch:train:2801-2900batch: iter_time=8.651e-05, forward_time=0.292, loss_ctc=44.923, loss_att=43.542, acc=0.769, loss=43.956, backward_time=0.296, grad_norm=39.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.763e-04, train_time=1.581
+[gpub011:0/16] 2024-02-07 04:54:07,202 (trainer:762) INFO: 33epoch:train:2901-3000batch: iter_time=8.259e-05, forward_time=0.297, loss_ctc=43.359, loss_att=38.484, acc=0.768, loss=39.946, backward_time=0.307, grad_norm=42.588, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.763e-04, train_time=1.377
+[gpub011:0/16] 2024-02-07 04:56:30,652 (trainer:762) INFO: 33epoch:train:3001-3100batch: iter_time=8.582e-05, forward_time=0.341, loss_ctc=50.629, loss_att=49.280, acc=0.742, loss=49.685, backward_time=0.307, grad_norm=44.269, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.762e-04, train_time=1.435
+[gpub011:0/16] 2024-02-07 04:58:48,398 (trainer:762) INFO: 33epoch:train:3101-3200batch: iter_time=1.209e-04, forward_time=0.376, loss_ctc=42.131, loss_att=39.985, acc=0.772, loss=40.629, backward_time=0.325, grad_norm=33.971, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.762e-04, train_time=1.377
+[gpub011:0/16] 2024-02-07 05:01:32,048 (trainer:762) INFO: 33epoch:train:3201-3300batch: iter_time=8.644e-05, forward_time=0.295, loss_ctc=49.351, loss_att=49.026, acc=0.742, loss=49.124, backward_time=0.307, grad_norm=38.322, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.762e-04, train_time=1.636
+[gpub011:0/16] 2024-02-07 05:03:51,131 (trainer:762) INFO: 33epoch:train:3301-3400batch: iter_time=8.306e-05, forward_time=0.331, loss_ctc=41.613, loss_att=39.344, acc=0.756, loss=40.025, backward_time=0.308, grad_norm=38.963, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.762e-04, train_time=1.390
+[gpub011:0/16] 2024-02-07 05:06:34,915 (trainer:762) INFO: 33epoch:train:3401-3500batch: iter_time=8.610e-05, forward_time=0.382, loss_ctc=48.968, loss_att=45.804, acc=0.761, loss=46.753, backward_time=0.312, grad_norm=46.152, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.762e-04, train_time=1.638
+[gpub011:0/16] 2024-02-07 05:08:29,022 (trainer:762) INFO: 33epoch:train:3501-3600batch: iter_time=8.751e-05, forward_time=0.293, loss_ctc=46.660, loss_att=53.535, acc=0.736, loss=51.472, backward_time=0.298, grad_norm=41.426, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.761e-04, train_time=1.141
+[gpub011:0/16] 2024-02-07 05:11:22,413 (trainer:762) INFO: 33epoch:train:3601-3700batch: iter_time=8.540e-05, forward_time=0.308, loss_ctc=42.695, loss_att=43.540, acc=0.759, loss=43.287, backward_time=0.317, grad_norm=37.000, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.761e-04, train_time=1.733
+[gpub011:0/16] 2024-02-07 05:12:45,341 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub011:0/16] 2024-02-07 05:13:05,075 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 05:13:08,950 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7b7f9a1e40>)
+[gpub011:0/16] 2024-02-07 05:13:08,951 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub011:0/16] 2024-02-07 05:13:08,954 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 05:20:26,742 (trainer:762) INFO: 33epoch:train:3701-3800batch: iter_time=3.971, forward_time=0.314, loss_ctc=47.139, loss_att=46.255, acc=0.770, loss=46.520, backward_time=0.298, grad_norm=37.580, clip=100.000, loss_scale=5.244e+33, optim_step_time=0.095, optim0_lr0=1.761e-04, train_time=5.444
+[gpub011:0/16] 2024-02-07 05:22:51,295 (trainer:762) INFO: 33epoch:train:3801-3900batch: iter_time=7.417e-05, forward_time=0.355, loss_ctc=41.709, loss_att=41.232, acc=0.772, loss=41.375, backward_time=0.298, grad_norm=47.352, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=1.761e-04, train_time=1.445
+[gpub011:0/16] 2024-02-07 05:25:25,637 (trainer:762) INFO: 33epoch:train:3901-4000batch: iter_time=8.013e-05, forward_time=0.299, loss_ctc=43.609, loss_att=45.107, acc=0.771, loss=44.658, backward_time=0.315, grad_norm=34.993, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.761e-04, train_time=1.543
+[gpub011:0/16] 2024-02-07 05:27:34,987 (trainer:762) INFO: 33epoch:train:4001-4100batch: iter_time=2.128e-04, forward_time=0.352, loss_ctc=42.344, loss_att=40.789, acc=0.756, loss=41.256, backward_time=0.395, grad_norm=36.383, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.761e-04, train_time=1.293
+[gpub011:0/16] 2024-02-07 05:30:08,133 (trainer:762) INFO: 33epoch:train:4101-4200batch: iter_time=8.198e-05, forward_time=0.286, loss_ctc=43.301, loss_att=44.130, acc=0.748, loss=43.881, backward_time=0.295, grad_norm=42.833, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.760e-04, train_time=1.531
+[gpub011:0/16] 2024-02-07 05:32:24,538 (trainer:762) INFO: 33epoch:train:4201-4300batch: iter_time=8.341e-05, forward_time=0.288, loss_ctc=50.856, loss_att=45.235, acc=0.757, loss=46.921, backward_time=0.296, grad_norm=43.357, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.760e-04, train_time=1.364
+[gpub011:0/16] 2024-02-07 05:34:46,485 (trainer:762) INFO: 33epoch:train:4301-4400batch: iter_time=8.563e-05, forward_time=0.329, loss_ctc=45.482, loss_att=43.088, acc=0.746, loss=43.806, backward_time=0.322, grad_norm=39.211, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.760e-04, train_time=1.419
+[gpub011:0/16] 2024-02-07 05:37:11,323 (trainer:762) INFO: 33epoch:train:4401-4500batch: iter_time=7.969e-05, forward_time=0.351, loss_ctc=45.219, loss_att=47.647, acc=0.741, loss=46.919, backward_time=0.346, grad_norm=38.857, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.760e-04, train_time=1.448
+[gpub011:0/16] 2024-02-07 05:40:00,285 (trainer:762) INFO: 33epoch:train:4501-4600batch: iter_time=8.123e-05, forward_time=0.287, loss_ctc=43.883, loss_att=39.633, acc=0.761, loss=40.908, backward_time=0.293, grad_norm=37.347, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.760e-04, train_time=1.690
+[gpub011:0/16] 2024-02-07 05:41:50,537 (trainer:762) INFO: 33epoch:train:4601-4700batch: iter_time=8.075e-05, forward_time=0.292, loss_ctc=49.736, loss_att=48.962, acc=0.730, loss=49.194, backward_time=0.297, grad_norm=50.559, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.759e-04, train_time=1.102
+[gpub011:0/16] 2024-02-07 05:44:24,377 (trainer:762) INFO: 33epoch:train:4701-4800batch: iter_time=7.816e-05, forward_time=0.322, loss_ctc=43.458, loss_att=42.098, acc=0.765, loss=42.506, backward_time=0.310, grad_norm=38.647, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.759e-04, train_time=1.538
+[gpub011:0/16] 2024-02-07 05:47:02,959 (trainer:762) INFO: 33epoch:train:4801-4900batch: iter_time=8.242e-05, forward_time=0.422, loss_ctc=47.841, loss_att=52.995, acc=0.736, loss=51.449, backward_time=0.318, grad_norm=38.490, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.759e-04, train_time=1.585
+[gpub011:0/16] 2024-02-07 05:49:18,338 (trainer:762) INFO: 33epoch:train:4901-5000batch: iter_time=7.864e-05, forward_time=0.286, loss_ctc=41.431, loss_att=39.158, acc=0.766, loss=39.840, backward_time=0.294, grad_norm=54.414, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.759e-04, train_time=1.354
+[gpub011:0/16] 2024-02-07 05:49:38,366 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub011:0/16] 2024-02-07 05:49:57,944 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 05:50:01,509 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f794dab75e0>)
+[gpub011:0/16] 2024-02-07 05:50:01,509 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub011:0/16] 2024-02-07 05:50:01,513 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 05:58:11,932 (trainer:762) INFO: 33epoch:train:5001-5100batch: iter_time=3.852, forward_time=0.438, loss_ctc=49.209, loss_att=48.970, acc=0.751, loss=49.041, backward_time=0.328, grad_norm=50.787, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.147, optim0_lr0=1.759e-04, train_time=5.336
+[gpub011:0/16] 2024-02-07 06:00:31,970 (trainer:762) INFO: 33epoch:train:5101-5200batch: iter_time=8.127e-05, forward_time=0.290, loss_ctc=42.094, loss_att=43.216, acc=0.779, loss=42.880, backward_time=0.296, grad_norm=34.174, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.759e-04, train_time=1.400
+[gpub011:0/16] 2024-02-07 06:02:44,479 (trainer:762) INFO: 33epoch:train:5201-5300batch: iter_time=8.026e-05, forward_time=0.287, loss_ctc=37.523, loss_att=35.637, acc=0.768, loss=36.203, backward_time=0.294, grad_norm=33.518, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.758e-04, train_time=1.325
+[gpub011:0/16] 2024-02-07 06:03:33,289 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 06:05:38,877 (trainer:762) INFO: 33epoch:train:5301-5400batch: iter_time=8.468e-05, forward_time=0.429, loss_ctc=44.499, loss_att=41.894, acc=0.767, loss=42.675, backward_time=0.337, grad_norm=38.620, clip=100.000, loss_scale=6.766e+33, optim_step_time=0.101, optim0_lr0=1.758e-04, train_time=1.744
+[gpub011:0/16] 2024-02-07 06:07:32,029 (trainer:762) INFO: 33epoch:train:5401-5500batch: iter_time=8.465e-05, forward_time=0.315, loss_ctc=43.125, loss_att=38.644, acc=0.757, loss=39.988, backward_time=0.296, grad_norm=41.488, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.758e-04, train_time=1.131
+[gpub011:0/16] 2024-02-07 06:10:30,034 (trainer:762) INFO: 33epoch:train:5501-5600batch: iter_time=8.444e-05, forward_time=0.290, loss_ctc=50.085, loss_att=48.620, acc=0.742, loss=49.059, backward_time=0.295, grad_norm=45.031, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.758e-04, train_time=1.780
+[gpub011:0/16] 2024-02-07 06:12:26,402 (trainer:762) INFO: 33epoch:train:5601-5700batch: iter_time=8.689e-05, forward_time=0.289, loss_ctc=42.259, loss_att=40.003, acc=0.771, loss=40.680, backward_time=0.294, grad_norm=35.655, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.758e-04, train_time=1.163
+[gpub011:0/16] 2024-02-07 06:15:15,244 (trainer:762) INFO: 33epoch:train:5701-5800batch: iter_time=2.398e-04, forward_time=0.397, loss_ctc=49.971, loss_att=49.803, acc=0.733, loss=49.854, backward_time=0.334, grad_norm=40.083, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.757e-04, train_time=1.689
+[gpub011:0/16] 2024-02-07 06:17:11,319 (trainer:762) INFO: 33epoch:train:5801-5900batch: iter_time=8.305e-05, forward_time=0.290, loss_ctc=41.071, loss_att=39.221, acc=0.749, loss=39.776, backward_time=0.294, grad_norm=38.920, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.757e-04, train_time=1.161
+[gpub011:0/16] 2024-02-07 06:19:51,284 (trainer:762) INFO: 33epoch:train:5901-6000batch: iter_time=8.961e-05, forward_time=0.289, loss_ctc=48.410, loss_att=45.240, acc=0.756, loss=46.191, backward_time=0.294, grad_norm=46.846, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.757e-04, train_time=1.599
+[gpub011:0/16] 2024-02-07 06:22:01,050 (trainer:762) INFO: 33epoch:train:6001-6100batch: iter_time=2.836e-04, forward_time=0.367, loss_ctc=46.167, loss_att=51.562, acc=0.729, loss=49.943, backward_time=0.342, grad_norm=39.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.757e-04, train_time=1.297
+[gpub011:0/16] 2024-02-07 06:24:31,874 (trainer:762) INFO: 33epoch:train:6101-6200batch: iter_time=8.718e-05, forward_time=0.322, loss_ctc=42.730, loss_att=42.694, acc=0.757, loss=42.704, backward_time=0.312, grad_norm=36.165, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.757e-04, train_time=1.508
+[gpub011:0/16] 2024-02-07 06:25:46,174 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub011:0/16] 2024-02-07 06:26:05,492 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 06:26:09,089 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78cbc59240>)
+[gpub011:0/16] 2024-02-07 06:26:09,089 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub011:0/16] 2024-02-07 06:26:09,093 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 06:32:37,049 (trainer:762) INFO: 33epoch:train:6201-6300batch: iter_time=3.602, forward_time=0.290, loss_ctc=46.828, loss_att=47.840, acc=0.767, loss=47.537, backward_time=0.296, grad_norm=37.915, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.757e-04, train_time=4.852
+[gpub011:0/16] 2024-02-07 06:34:56,859 (trainer:762) INFO: 33epoch:train:6301-6400batch: iter_time=3.283e-04, forward_time=0.403, loss_ctc=41.762, loss_att=41.600, acc=0.779, loss=41.649, backward_time=0.333, grad_norm=45.692, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.118, optim0_lr0=1.756e-04, train_time=1.398
+[gpub011:0/16] 2024-02-07 06:37:09,728 (trainer:762) INFO: 33epoch:train:6401-6500batch: iter_time=8.071e-05, forward_time=0.290, loss_ctc=43.078, loss_att=45.004, acc=0.776, loss=44.426, backward_time=0.296, grad_norm=35.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.756e-04, train_time=1.329
+[gpub011:0/16] 2024-02-07 06:39:26,232 (trainer:762) INFO: 33epoch:train:6501-6600batch: iter_time=8.680e-05, forward_time=0.288, loss_ctc=42.213, loss_att=43.687, acc=0.754, loss=43.245, backward_time=0.298, grad_norm=37.346, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.756e-04, train_time=1.365
+[gpub011:0/16] 2024-02-07 06:42:18,813 (trainer:762) INFO: 33epoch:train:6601-6700batch: iter_time=8.490e-05, forward_time=0.364, loss_ctc=43.060, loss_att=43.107, acc=0.767, loss=43.093, backward_time=0.374, grad_norm=42.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.756e-04, train_time=1.725
+[gpub011:0/16] 2024-02-07 06:44:44,452 (trainer:762) INFO: 33epoch:train:6701-6800batch: iter_time=9.209e-05, forward_time=0.291, loss_ctc=50.609, loss_att=44.783, acc=0.764, loss=46.531, backward_time=0.300, grad_norm=43.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.756e-04, train_time=1.457
+[gpub011:0/16] 2024-02-07 06:46:57,949 (trainer:762) INFO: 33epoch:train:6801-6900batch: iter_time=8.248e-05, forward_time=0.291, loss_ctc=45.177, loss_att=43.203, acc=0.754, loss=43.795, backward_time=0.298, grad_norm=41.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.755e-04, train_time=1.335
+[gpub011:0/16] 2024-02-07 06:49:18,933 (trainer:762) INFO: 33epoch:train:6901-7000batch: iter_time=8.098e-05, forward_time=0.377, loss_ctc=44.956, loss_att=47.214, acc=0.749, loss=46.537, backward_time=0.354, grad_norm=38.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.755e-04, train_time=1.410
+[gpub011:0/16] 2024-02-07 06:51:54,387 (trainer:762) INFO: 33epoch:train:7001-7100batch: iter_time=8.769e-05, forward_time=0.289, loss_ctc=43.684, loss_att=39.718, acc=0.766, loss=40.908, backward_time=0.297, grad_norm=37.827, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.755e-04, train_time=1.554
+[gpub011:0/16] 2024-02-07 06:54:48,961 (trainer:762) INFO: 33epoch:train:7101-7200batch: iter_time=9.671e-04, forward_time=0.323, loss_ctc=49.118, loss_att=48.495, acc=0.741, loss=48.682, backward_time=0.328, grad_norm=51.388, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.755e-04, train_time=1.746
+[gpub011:0/16] 2024-02-07 06:56:49,452 (trainer:762) INFO: 33epoch:train:7201-7300batch: iter_time=8.372e-05, forward_time=0.372, loss_ctc=43.119, loss_att=43.940, acc=0.769, loss=43.694, backward_time=0.322, grad_norm=36.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.755e-04, train_time=1.205
+[gpub011:0/16] 2024-02-07 06:57:41,496 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 06:59:24,151 (trainer:762) INFO: 33epoch:train:7301-7400batch: iter_time=8.996e-05, forward_time=0.291, loss_ctc=46.732, loss_att=53.129, acc=0.748, loss=51.210, backward_time=0.301, grad_norm=37.671, clip=100.000, loss_scale=3.330e+33, optim_step_time=0.094, optim0_lr0=1.755e-04, train_time=1.547
+[gpub011:0/16] 2024-02-07 07:02:01,151 (trainer:762) INFO: 33epoch:train:7401-7500batch: iter_time=9.134e-05, forward_time=0.321, loss_ctc=41.397, loss_att=39.478, acc=0.766, loss=40.054, backward_time=0.317, grad_norm=35.571, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.754e-04, train_time=1.570
+[gpub011:0/16] 2024-02-07 07:02:21,180 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub011:0/16] 2024-02-07 07:02:40,743 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 07:02:44,619 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78c1b33f40>)
+[gpub011:0/16] 2024-02-07 07:02:44,619 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub011:0/16] 2024-02-07 07:02:44,623 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 07:10:58,131 (trainer:762) INFO: 33epoch:train:7501-7600batch: iter_time=3.900, forward_time=0.393, loss_ctc=48.330, loss_att=50.087, acc=0.751, loss=49.560, backward_time=0.306, grad_norm=51.360, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.754e-04, train_time=5.370
+[gpub011:0/16] 2024-02-07 07:13:11,750 (trainer:762) INFO: 33epoch:train:7601-7700batch: iter_time=8.196e-05, forward_time=0.289, loss_ctc=41.887, loss_att=43.789, acc=0.778, loss=43.219, backward_time=0.299, grad_norm=32.933, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.754e-04, train_time=1.336
+[gpub011:0/16] 2024-02-07 07:15:55,937 (trainer:762) INFO: 33epoch:train:7701-7800batch: iter_time=8.626e-05, forward_time=0.335, loss_ctc=37.443, loss_att=35.629, acc=0.768, loss=36.173, backward_time=0.302, grad_norm=33.942, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=1.754e-04, train_time=1.642
+[gpub011:0/16] 2024-02-07 07:18:31,965 (trainer:762) INFO: 33epoch:train:7801-7900batch: iter_time=3.527e-04, forward_time=0.336, loss_ctc=44.291, loss_att=42.476, acc=0.766, loss=43.020, backward_time=0.379, grad_norm=39.947, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.754e-04, train_time=1.560
+[gpub011:0/16] 2024-02-07 07:20:34,674 (trainer:762) INFO: 33epoch:train:7901-8000batch: iter_time=8.207e-05, forward_time=0.288, loss_ctc=43.192, loss_att=39.293, acc=0.755, loss=40.463, backward_time=0.295, grad_norm=41.128, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.754e-04, train_time=1.227
+[gpub011:0/16] 2024-02-07 07:23:28,648 (trainer:762) INFO: 33epoch:train:8001-8100batch: iter_time=8.390e-05, forward_time=0.325, loss_ctc=49.714, loss_att=47.940, acc=0.745, loss=48.472, backward_time=0.322, grad_norm=43.118, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.753e-04, train_time=1.739
+[gpub011:0/16] 2024-02-07 07:26:03,828 (trainer:762) INFO: 33epoch:train:8101-8200batch: iter_time=8.547e-05, forward_time=0.385, loss_ctc=41.546, loss_att=39.438, acc=0.772, loss=40.071, backward_time=0.330, grad_norm=38.274, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.753e-04, train_time=1.552
+[gpub011:0/16] 2024-02-07 07:28:18,224 (trainer:762) INFO: 33epoch:train:8201-8300batch: iter_time=8.306e-05, forward_time=0.291, loss_ctc=48.806, loss_att=48.857, acc=0.735, loss=48.842, backward_time=0.298, grad_norm=40.487, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.753e-04, train_time=1.344
+[gpub011:0/16] 2024-02-07 07:30:49,607 (trainer:762) INFO: 33epoch:train:8301-8400batch: iter_time=8.643e-05, forward_time=0.286, loss_ctc=40.815, loss_att=39.223, acc=0.750, loss=39.701, backward_time=0.292, grad_norm=39.423, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.753e-04, train_time=1.514
+[gpub011:0/16] 2024-02-07 07:33:32,173 (trainer:762) INFO: 33epoch:train:8401-8500batch: iter_time=8.310e-04, forward_time=0.421, loss_ctc=48.942, loss_att=45.378, acc=0.756, loss=46.447, backward_time=0.324, grad_norm=48.492, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.112, optim0_lr0=1.753e-04, train_time=1.625
+[gpub011:0/16] 2024-02-07 07:35:44,059 (trainer:762) INFO: 33epoch:train:8501-8600batch: iter_time=7.970e-05, forward_time=0.289, loss_ctc=45.549, loss_att=51.392, acc=0.732, loss=49.639, backward_time=0.295, grad_norm=39.873, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.752e-04, train_time=1.319
+[gpub011:0/16] 2024-02-07 07:38:19,054 (trainer:762) INFO: 33epoch:train:8601-8700batch: iter_time=7.842e-05, forward_time=0.293, loss_ctc=42.124, loss_att=41.721, acc=0.760, loss=41.842, backward_time=0.297, grad_norm=37.067, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.752e-04, train_time=1.550
+[gpub011:0/16] 2024-02-07 07:39:37,235 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub011:0/16] 2024-02-07 07:39:56,504 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 07:40:00,116 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f790a8cf610>)
+[gpub011:0/16] 2024-02-07 07:40:00,116 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub011:0/16] 2024-02-07 07:40:00,119 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 07:48:30,754 (trainer:762) INFO: 33epoch:train:8701-8800batch: iter_time=4.685, forward_time=0.530, loss_ctc=46.953, loss_att=47.738, acc=0.768, loss=47.502, backward_time=0.327, grad_norm=38.035, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.752e-04, train_time=6.117
+[gpub011:0/16] 2024-02-07 07:50:49,779 (trainer:762) INFO: 33epoch:train:8801-8900batch: iter_time=8.345e-05, forward_time=0.288, loss_ctc=39.753, loss_att=40.715, acc=0.779, loss=40.426, backward_time=0.294, grad_norm=43.601, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.752e-04, train_time=1.390
+[gpub011:0/16] 2024-02-07 07:53:10,028 (trainer:762) INFO: 33epoch:train:8901-9000batch: iter_time=8.159e-05, forward_time=0.292, loss_ctc=43.538, loss_att=44.906, acc=0.776, loss=44.496, backward_time=0.302, grad_norm=34.509, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.752e-04, train_time=1.403
+[gpub011:0/16] 2024-02-07 07:55:50,246 (trainer:762) INFO: 33epoch:train:9001-9100batch: iter_time=8.429e-05, forward_time=0.363, loss_ctc=41.879, loss_att=43.353, acc=0.753, loss=42.911, backward_time=0.429, grad_norm=37.862, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.109, optim0_lr0=1.752e-04, train_time=1.602
+[gpub011:0/16] 2024-02-07 07:58:31,429 (trainer:762) INFO: 33epoch:train:9101-9200batch: iter_time=8.211e-05, forward_time=0.289, loss_ctc=42.991, loss_att=43.051, acc=0.767, loss=43.033, backward_time=0.295, grad_norm=40.455, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.751e-04, train_time=1.612
+[gpub011:0/16] 2024-02-07 08:00:48,528 (trainer:762) INFO: 33epoch:train:9201-9300batch: iter_time=8.090e-05, forward_time=0.292, loss_ctc=50.204, loss_att=44.498, acc=0.764, loss=46.210, backward_time=0.300, grad_norm=41.608, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.751e-04, train_time=1.371
+[gpub011:0/16] 2024-02-07 08:03:30,149 (trainer:762) INFO: 33epoch:train:9301-9400batch: iter_time=8.456e-05, forward_time=0.289, loss_ctc=44.574, loss_att=42.724, acc=0.755, loss=43.279, backward_time=0.294, grad_norm=40.990, clip=100.000, loss_scale=4.439e+33, optim_step_time=0.094, optim0_lr0=1.751e-04, train_time=1.616
+[gpub011:0/16] 2024-02-07 08:06:21,284 (trainer:762) INFO: 33epoch:train:9401-9500batch: iter_time=8.335e-05, forward_time=0.485, loss_ctc=44.331, loss_att=46.982, acc=0.751, loss=46.187, backward_time=0.339, grad_norm=37.142, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.751e-04, train_time=1.711
+[gpub011:0/16] 2024-02-07 08:09:04,580 (trainer:762) INFO: 33epoch:train:9501-9600batch: iter_time=8.143e-05, forward_time=0.291, loss_ctc=43.619, loss_att=39.471, acc=0.766, loss=40.715, backward_time=0.294, grad_norm=37.294, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.751e-04, train_time=1.632
+[gpub011:0/16] 2024-02-07 08:11:06,317 (trainer:762) INFO: 33epoch:train:9601-9700batch: iter_time=7.939e-05, forward_time=0.289, loss_ctc=48.512, loss_att=48.235, acc=0.742, loss=48.318, backward_time=0.299, grad_norm=47.941, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.750e-04, train_time=1.217
+[gpub011:0/16] 2024-02-07 08:13:22,881 (trainer:762) INFO: 33epoch:train:9701-9800batch: iter_time=8.674e-05, forward_time=0.290, loss_ctc=43.085, loss_att=43.779, acc=0.769, loss=43.571, backward_time=0.295, grad_norm=37.131, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.750e-04, train_time=1.366
+[gpub011:0/16] 2024-02-07 08:16:43,072 (trainer:762) INFO: 33epoch:train:9801-9900batch: iter_time=8.200e-05, forward_time=0.472, loss_ctc=46.926, loss_att=53.512, acc=0.747, loss=51.536, backward_time=0.328, grad_norm=38.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.750e-04, train_time=2.001
+[gpub011:0/16] 2024-02-07 08:19:03,655 (trainer:762) INFO: 33epoch:train:9901-10000batch: iter_time=7.804e-05, forward_time=0.288, loss_ctc=40.950, loss_att=39.538, acc=0.768, loss=39.962, backward_time=0.295, grad_norm=36.370, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.750e-04, train_time=1.406
+[gpub011:0/16] 2024-02-07 08:19:23,683 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub011:0/16] 2024-02-07 08:19:43,465 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 08:19:47,035 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7953a048e0>)
+[gpub011:0/16] 2024-02-07 08:19:47,036 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub011:0/16] 2024-02-07 08:19:47,039 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 08:27:47,697 (trainer:762) INFO: 33epoch:train:10001-10100batch: iter_time=3.758, forward_time=0.289, loss_ctc=48.504, loss_att=49.867, acc=0.751, loss=49.458, backward_time=0.296, grad_norm=53.130, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.750e-04, train_time=5.240
+[gpub011:0/16] 2024-02-07 08:30:37,350 (trainer:762) INFO: 33epoch:train:10101-10200batch: iter_time=8.003e-05, forward_time=0.431, loss_ctc=42.034, loss_att=43.325, acc=0.781, loss=42.938, backward_time=0.330, grad_norm=34.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.750e-04, train_time=1.696
+[gpub011:0/16] 2024-02-07 08:32:54,052 (trainer:762) INFO: 33epoch:train:10201-10300batch: iter_time=8.091e-05, forward_time=0.288, loss_ctc=37.415, loss_att=35.957, acc=0.767, loss=36.395, backward_time=0.293, grad_norm=31.521, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.749e-04, train_time=1.367
+[gpub011:0/16] 2024-02-07 08:35:00,497 (trainer:762) INFO: 33epoch:train:10301-10400batch: iter_time=8.636e-05, forward_time=0.289, loss_ctc=44.085, loss_att=41.805, acc=0.769, loss=42.489, backward_time=0.295, grad_norm=37.628, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.749e-04, train_time=1.264
+[gpub011:0/16] 2024-02-07 08:37:57,088 (trainer:762) INFO: 33epoch:train:10401-10500batch: iter_time=8.272e-05, forward_time=0.458, loss_ctc=42.990, loss_att=38.873, acc=0.758, loss=40.108, backward_time=0.327, grad_norm=41.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.749e-04, train_time=1.765
+[gpub011:0/16] 2024-02-07 08:40:04,200 (trainer:762) INFO: 33epoch:train:10501-10600batch: iter_time=7.862e-05, forward_time=0.298, loss_ctc=49.698, loss_att=48.423, acc=0.743, loss=48.806, backward_time=0.300, grad_norm=43.109, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.749e-04, train_time=1.271
+[gpub011:0/16] 2024-02-07 08:42:37,302 (trainer:762) INFO: 33epoch:train:10601-10700batch: iter_time=8.075e-05, forward_time=0.289, loss_ctc=41.410, loss_att=39.544, acc=0.772, loss=40.104, backward_time=0.293, grad_norm=34.624, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.749e-04, train_time=1.531
+[gpub011:0/16] 2024-02-07 08:44:45,153 (trainer:762) INFO: 33epoch:train:10701-10800batch: iter_time=8.718e-05, forward_time=0.291, loss_ctc=48.792, loss_att=49.385, acc=0.736, loss=49.207, backward_time=0.306, grad_norm=40.669, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.748e-04, train_time=1.278
+[gpub011:0/16] 2024-02-07 08:47:26,813 (trainer:762) INFO: 33epoch:train:10801-10900batch: iter_time=8.857e-05, forward_time=0.396, loss_ctc=40.595, loss_att=39.125, acc=0.751, loss=39.566, backward_time=0.309, grad_norm=40.555, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.748e-04, train_time=1.616
+[gpub011:0/16] 2024-02-07 08:49:31,261 (trainer:762) INFO: 33epoch:train:10901-11000batch: iter_time=8.270e-05, forward_time=0.290, loss_ctc=47.264, loss_att=44.616, acc=0.760, loss=45.410, backward_time=0.295, grad_norm=45.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.748e-04, train_time=1.244
+[gpub011:0/16] 2024-02-07 08:51:45,406 (trainer:762) INFO: 33epoch:train:11001-11100batch: iter_time=7.848e-05, forward_time=0.291, loss_ctc=46.199, loss_att=51.759, acc=0.730, loss=50.091, backward_time=0.296, grad_norm=41.155, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.748e-04, train_time=1.342
+[gpub011:0/16] 2024-02-07 08:54:29,764 (trainer:762) INFO: 33epoch:train:11101-11200batch: iter_time=2.147e-04, forward_time=0.430, loss_ctc=41.785, loss_att=42.580, acc=0.757, loss=42.341, backward_time=0.335, grad_norm=36.322, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.748e-04, train_time=1.642
+[gpub011:0/16] 2024-02-07 08:55:44,536 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub011:0/16] 2024-02-07 08:56:04,322 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 08:56:07,923 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f795c05f820>)
+[gpub011:0/16] 2024-02-07 08:56:07,923 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-07 08:56:07,927 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 09:05:43,551 (trainer:762) INFO: 33epoch:train:11201-11300batch: iter_time=5.443, forward_time=0.291, loss_ctc=47.064, loss_att=46.170, acc=0.768, loss=46.438, backward_time=0.298, grad_norm=37.224, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.748e-04, train_time=6.738
+[gpub011:0/16] 2024-02-07 09:10:44,775 (trainer:762) INFO: 33epoch:train:11301-11400batch: iter_time=8.226e-05, forward_time=0.298, loss_ctc=39.728, loss_att=39.949, acc=0.775, loss=39.883, backward_time=0.296, grad_norm=42.542, clip=100.000, loss_scale=8.879e+33, optim_step_time=0.097, optim0_lr0=1.747e-04, train_time=3.012
+[gpub011:0/16] 2024-02-07 09:13:39,536 (trainer:762) INFO: 33epoch:train:11401-11500batch: iter_time=8.233e-05, forward_time=0.425, loss_ctc=42.951, loss_att=44.258, acc=0.774, loss=43.866, backward_time=0.315, grad_norm=34.318, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.747e-04, train_time=1.747
+[gpub011:0/16] 2024-02-07 09:17:31,970 (trainer:762) INFO: 33epoch:train:11501-11600batch: iter_time=7.811e-05, forward_time=0.292, loss_ctc=41.725, loss_att=40.260, acc=0.759, loss=40.699, backward_time=0.296, grad_norm=36.064, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.747e-04, train_time=2.323
+[gpub011:0/16] 2024-02-07 09:19:52,762 (trainer:762) INFO: 33epoch:train:11601-11700batch: iter_time=8.512e-05, forward_time=0.287, loss_ctc=43.116, loss_att=43.050, acc=0.753, loss=43.070, backward_time=0.294, grad_norm=40.152, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.747e-04, train_time=1.409
+[gpub011:0/16] 2024-02-07 09:22:51,761 (trainer:762) INFO: 33epoch:train:11701-11800batch: iter_time=8.350e-05, forward_time=0.391, loss_ctc=50.514, loss_att=45.222, acc=0.760, loss=46.809, backward_time=0.363, grad_norm=45.404, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.747e-04, train_time=1.790
+[gpub011:0/16] 2024-02-07 09:24:57,633 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 09:25:03,052 (trainer:762) INFO: 33epoch:train:11801-11900batch: iter_time=8.318e-05, forward_time=0.288, loss_ctc=44.713, loss_att=42.415, acc=0.750, loss=43.104, backward_time=0.294, grad_norm=40.692, clip=100.000, loss_scale=1.012e+34, optim_step_time=0.094, optim0_lr0=1.747e-04, train_time=1.313
+[gpub011:0/16] 2024-02-07 09:28:15,288 (trainer:762) INFO: 33epoch:train:11901-12000batch: iter_time=8.803e-05, forward_time=0.292, loss_ctc=44.331, loss_att=47.317, acc=0.743, loss=46.422, backward_time=0.293, grad_norm=38.311, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.746e-04, train_time=1.921
+[gpub011:0/16] 2024-02-07 09:30:38,878 (trainer:762) INFO: 33epoch:train:12001-12100batch: iter_time=8.892e-05, forward_time=0.288, loss_ctc=43.310, loss_att=39.290, acc=0.763, loss=40.496, backward_time=0.298, grad_norm=39.135, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.746e-04, train_time=1.437
+[gpub011:0/16] 2024-02-07 09:33:29,218 (trainer:762) INFO: 33epoch:train:12101-12200batch: iter_time=9.052e-05, forward_time=0.427, loss_ctc=48.273, loss_att=48.180, acc=0.733, loss=48.208, backward_time=0.352, grad_norm=49.055, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.746e-04, train_time=1.703
+[gpub011:0/16] 2024-02-07 09:36:22,020 (trainer:762) INFO: 33epoch:train:12201-12300batch: iter_time=8.708e-05, forward_time=0.289, loss_ctc=42.643, loss_att=41.525, acc=0.768, loss=41.860, backward_time=0.294, grad_norm=37.967, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.746e-04, train_time=1.728
+[gpub011:0/16] 2024-02-07 09:38:38,632 (trainer:762) INFO: 33epoch:train:12301-12400batch: iter_time=2.029e-04, forward_time=0.301, loss_ctc=46.555, loss_att=51.619, acc=0.741, loss=50.100, backward_time=0.309, grad_norm=38.633, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.746e-04, train_time=1.365
+[gpub011:0/16] 2024-02-07 09:41:44,685 (trainer:762) INFO: 33epoch:train:12401-12500batch: iter_time=9.317e-05, forward_time=0.374, loss_ctc=41.259, loss_att=38.547, acc=0.769, loss=39.361, backward_time=0.378, grad_norm=35.519, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.745e-04, train_time=1.861
+[gpub011:0/16] 2024-02-07 09:42:04,713 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub011:0/16] 2024-02-07 09:42:24,413 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 09:42:27,977 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c0e41ec50>)
+[gpub011:0/16] 2024-02-07 09:42:27,977 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub011:0/16] 2024-02-07 09:42:27,981 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 09:50:29,714 (trainer:762) INFO: 33epoch:train:12501-12600batch: iter_time=3.866, forward_time=0.293, loss_ctc=49.518, loss_att=51.049, acc=0.758, loss=50.590, backward_time=0.296, grad_norm=52.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.745e-04, train_time=5.250
+[gpub011:0/16] 2024-02-07 09:52:50,389 (trainer:762) INFO: 33epoch:train:12601-12700batch: iter_time=8.983e-05, forward_time=0.289, loss_ctc=41.633, loss_att=43.650, acc=0.781, loss=43.045, backward_time=0.295, grad_norm=36.159, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.745e-04, train_time=1.406
+[gpub011:0/16] 2024-02-07 09:55:41,668 (trainer:762) INFO: 33epoch:train:12701-12800batch: iter_time=7.836e-05, forward_time=0.431, loss_ctc=37.203, loss_att=38.013, acc=0.766, loss=37.770, backward_time=0.321, grad_norm=33.306, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.745e-04, train_time=1.713
+[gpub011:0/16] 2024-02-07 09:58:21,394 (trainer:762) INFO: 33epoch:train:12801-12900batch: iter_time=8.087e-05, forward_time=0.293, loss_ctc=44.368, loss_att=44.152, acc=0.769, loss=44.217, backward_time=0.296, grad_norm=38.324, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.745e-04, train_time=1.597
+[gpub011:0/16] 2024-02-07 10:00:34,379 (trainer:762) INFO: 33epoch:train:12901-13000batch: iter_time=8.749e-05, forward_time=0.289, loss_ctc=42.944, loss_att=38.859, acc=0.769, loss=40.084, backward_time=0.296, grad_norm=39.803, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.745e-04, train_time=1.330
+[gpub011:0/16] 2024-02-07 10:03:27,146 (trainer:762) INFO: 33epoch:train:13001-13100batch: iter_time=8.305e-05, forward_time=0.376, loss_ctc=49.182, loss_att=48.845, acc=0.746, loss=48.946, backward_time=0.397, grad_norm=43.816, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.744e-04, train_time=1.727
+[gpub011:0/16] 2024-02-07 10:05:16,349 (trainer:762) INFO: 33epoch:train:13101-13200batch: iter_time=8.129e-05, forward_time=0.290, loss_ctc=41.822, loss_att=39.600, acc=0.775, loss=40.267, backward_time=0.296, grad_norm=37.990, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.744e-04, train_time=1.092
+[gpub011:0/16] 2024-02-07 10:08:22,060 (trainer:762) INFO: 33epoch:train:13201-13300batch: iter_time=7.988e-05, forward_time=0.290, loss_ctc=48.773, loss_att=48.814, acc=0.747, loss=48.802, backward_time=0.296, grad_norm=40.377, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.744e-04, train_time=1.857
+[gpub011:0/16] 2024-02-07 10:10:58,758 (trainer:762) INFO: 33epoch:train:13301-13400batch: iter_time=2.660e-04, forward_time=0.387, loss_ctc=40.498, loss_att=37.991, acc=0.765, loss=38.743, backward_time=0.362, grad_norm=38.417, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.744e-04, train_time=1.567
+[gpub011:0/16] 2024-02-07 10:13:10,604 (trainer:762) INFO: 33epoch:train:13401-13500batch: iter_time=8.092e-05, forward_time=0.290, loss_ctc=47.492, loss_att=44.658, acc=0.767, loss=45.508, backward_time=0.296, grad_norm=45.047, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.744e-04, train_time=1.318
+[gpub011:0/16] 2024-02-07 10:15:48,384 (trainer:762) INFO: 33epoch:train:13501-13600batch: iter_time=8.257e-05, forward_time=0.290, loss_ctc=45.449, loss_att=52.856, acc=0.740, loss=50.634, backward_time=0.295, grad_norm=39.335, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.744e-04, train_time=1.578
+[gpub011:0/16] 2024-02-07 10:18:41,089 (trainer:762) INFO: 33epoch:train:13601-13700batch: iter_time=1.740e-04, forward_time=0.399, loss_ctc=42.001, loss_att=43.294, acc=0.763, loss=42.906, backward_time=0.339, grad_norm=36.350, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.743e-04, train_time=1.727
+[gpub011:0/16] 2024-02-07 10:19:56,732 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub011:0/16] 2024-02-07 10:20:16,178 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 10:20:20,031 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f794ddc3b20>)
+[gpub011:0/16] 2024-02-07 10:20:20,031 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-07 10:20:20,034 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 10:27:13,479 (trainer:762) INFO: 33epoch:train:13701-13800batch: iter_time=3.814, forward_time=0.294, loss_ctc=46.746, loss_att=47.063, acc=0.772, loss=46.968, backward_time=0.298, grad_norm=37.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.743e-04, train_time=5.124
+[gpub011:0/16] 2024-02-07 10:29:50,101 (trainer:762) INFO: 33epoch:train:13801-13900batch: iter_time=7.953e-05, forward_time=0.296, loss_ctc=39.826, loss_att=39.435, acc=0.783, loss=39.552, backward_time=0.316, grad_norm=43.817, clip=100.000, loss_scale=5.452e+33, optim_step_time=0.096, optim0_lr0=1.743e-04, train_time=1.566
+[gpub011:0/16] 2024-02-07 10:32:40,039 (trainer:762) INFO: 33epoch:train:13901-14000batch: iter_time=3.159e-04, forward_time=0.406, loss_ctc=42.826, loss_att=44.540, acc=0.778, loss=44.026, backward_time=0.315, grad_norm=34.650, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.111, optim0_lr0=1.743e-04, train_time=1.699
+[gpub011:0/16] 2024-02-07 10:34:58,953 (trainer:762) INFO: 33epoch:train:14001-14100batch: iter_time=7.692e-05, forward_time=0.292, loss_ctc=41.726, loss_att=43.652, acc=0.753, loss=43.074, backward_time=0.296, grad_norm=36.791, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.743e-04, train_time=1.389
+[gpub011:0/16] 2024-02-07 10:38:03,323 (trainer:762) INFO: 33epoch:train:14101-14200batch: iter_time=1.910e-04, forward_time=0.371, loss_ctc=42.556, loss_att=42.281, acc=0.769, loss=42.363, backward_time=0.381, grad_norm=37.517, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.742e-04, train_time=1.844
+[gpub011:0/16] 2024-02-07 10:40:43,876 (trainer:762) INFO: 33epoch:train:14201-14300batch: iter_time=8.023e-05, forward_time=0.288, loss_ctc=49.766, loss_att=44.434, acc=0.765, loss=46.034, backward_time=0.297, grad_norm=41.561, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.742e-04, train_time=1.605
+[gpub011:0/16] 2024-02-07 10:43:52,350 (trainer:762) INFO: 33epoch:train:14301-14400batch: iter_time=8.259e-05, forward_time=0.287, loss_ctc=44.837, loss_att=42.828, acc=0.755, loss=43.431, backward_time=0.293, grad_norm=40.900, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.742e-04, train_time=1.884
+[gpub011:0/16] 2024-02-07 10:46:23,168 (trainer:762) INFO: 33epoch:train:14401-14500batch: iter_time=1.869e-04, forward_time=0.388, loss_ctc=44.548, loss_att=47.518, acc=0.750, loss=46.627, backward_time=0.355, grad_norm=37.903, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.101, optim0_lr0=1.742e-04, train_time=1.508
+[gpub011:0/16] 2024-02-07 10:49:01,505 (trainer:762) INFO: 33epoch:train:14501-14600batch: iter_time=8.323e-05, forward_time=0.288, loss_ctc=42.598, loss_att=38.393, acc=0.772, loss=39.655, backward_time=0.293, grad_norm=35.734, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.742e-04, train_time=1.583
+[gpub011:0/16] 2024-02-07 10:51:39,905 (trainer:762) INFO: 33epoch:train:14601-14700batch: iter_time=8.161e-05, forward_time=0.289, loss_ctc=48.074, loss_att=47.355, acc=0.745, loss=47.571, backward_time=0.295, grad_norm=48.705, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.742e-04, train_time=1.584
+[gpub011:0/16] 2024-02-07 10:54:43,867 (trainer:762) INFO: 33epoch:train:14701-14800batch: iter_time=3.993e-04, forward_time=0.466, loss_ctc=42.380, loss_att=43.290, acc=0.773, loss=43.017, backward_time=0.320, grad_norm=35.832, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.102, optim0_lr0=1.741e-04, train_time=1.839
+[gpub011:0/16] 2024-02-07 10:57:29,434 (trainer:762) INFO: 33epoch:train:14801-14900batch: iter_time=9.116e-05, forward_time=0.293, loss_ctc=46.716, loss_att=52.743, acc=0.752, loss=50.935, backward_time=0.295, grad_norm=37.630, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.741e-04, train_time=1.655
+[gpub011:0/16] 2024-02-07 10:59:49,311 (trainer:762) INFO: 33epoch:train:14901-15000batch: iter_time=9.363e-05, forward_time=0.289, loss_ctc=41.152, loss_att=39.614, acc=0.768, loss=40.075, backward_time=0.294, grad_norm=35.578, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.741e-04, train_time=1.398
+[gpub011:0/16] 2024-02-07 11:40:37,449 (trainer:361) INFO: 33epoch results: [train] iter_time=0.319, forward_time=0.328, loss_ctc=44.786, loss_att=44.310, acc=0.758, loss=44.453, backward_time=0.312, grad_norm=40.231, clip=100.000, loss_scale=5.643e+33, optim_step_time=0.098, optim0_lr0=1.754e-04, train_time=1.822, time=7 hours, 35 minutes and 51.39 seconds, total_count=525000, gpu_max_cached_mem_GB=41.025, [valid] loss_ctc=35.538, cer_ctc=0.182, loss_att=36.677, acc=0.696, cer=0.301, wer=0.991, loss=36.335, time=40 minutes and 23.58 seconds, total_count=163485, gpu_max_cached_mem_GB=41.025
+[gpub011:0/16] 2024-02-07 11:40:47,889 (trainer:416) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub011:0/16] 2024-02-07 11:40:47,932 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/28epoch.pth
+[gpub011:0/16] 2024-02-07 11:40:47,932 (trainer:290) INFO: 34/45epoch started. Estimated time to finish: 4 days, 34 minutes and 25.43 seconds
+[gpub011:0/16] 2024-02-07 11:40:47,941 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub011:0/16] 2024-02-07 11:41:06,904 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 11:41:10,406 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f72311df730>)
+[gpub011:0/16] 2024-02-07 11:41:10,406 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub011:0/16] 2024-02-07 11:41:10,410 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 11:49:33,549 (trainer:762) INFO: 34epoch:train:1-100batch: iter_time=4.083, forward_time=0.324, loss_ctc=57.803, loss_att=56.908, acc=0.726, loss=57.177, backward_time=0.308, grad_norm=51.142, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=1.741e-04, train_time=5.256
+[gpub011:0/16] 2024-02-07 11:51:44,168 (trainer:762) INFO: 34epoch:train:101-200batch: iter_time=2.070e-04, forward_time=0.375, loss_ctc=51.589, loss_att=45.672, acc=0.748, loss=47.447, backward_time=0.298, grad_norm=41.423, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.107, optim0_lr0=1.741e-04, train_time=1.306
+[gpub011:0/16] 2024-02-07 11:53:50,337 (trainer:762) INFO: 34epoch:train:201-300batch: iter_time=8.738e-05, forward_time=0.297, loss_ctc=39.117, loss_att=35.778, acc=0.763, loss=36.780, backward_time=0.306, grad_norm=35.847, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.741e-04, train_time=1.261
+[gpub011:0/16] 2024-02-07 11:56:14,867 (trainer:762) INFO: 34epoch:train:301-400batch: iter_time=8.591e-05, forward_time=0.325, loss_ctc=49.054, loss_att=51.009, acc=0.742, loss=50.422, backward_time=0.318, grad_norm=44.612, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.740e-04, train_time=1.444
+[gpub011:0/16] 2024-02-07 11:58:17,990 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub011:0/16] 2024-02-07 11:58:21,232 (trainer:762) INFO: 34epoch:train:401-500batch: iter_time=8.245e-05, forward_time=0.306, loss_ctc=46.921, loss_att=43.434, acc=0.746, loss=44.480, backward_time=0.294, grad_norm=41.854, clip=100.000, loss_scale=1.023e+34, optim_step_time=0.095, optim0_lr0=1.740e-04, train_time=1.264
+[gpub011:0/16] 2024-02-07 12:00:31,160 (trainer:762) INFO: 34epoch:train:501-600batch: iter_time=2.320e-04, forward_time=0.340, loss_ctc=46.608, loss_att=47.161, acc=0.753, loss=46.995, backward_time=0.326, grad_norm=44.347, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.740e-04, train_time=1.299
+[gpub011:0/16] 2024-02-07 12:02:46,530 (trainer:762) INFO: 34epoch:train:601-700batch: iter_time=9.218e-05, forward_time=0.335, loss_ctc=45.118, loss_att=46.794, acc=0.763, loss=46.291, backward_time=0.326, grad_norm=37.205, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.740e-04, train_time=1.353
+[gpub011:0/16] 2024-02-07 12:04:58,805 (trainer:762) INFO: 34epoch:train:701-800batch: iter_time=9.103e-05, forward_time=0.307, loss_ctc=37.397, loss_att=35.741, acc=0.783, loss=36.238, backward_time=0.297, grad_norm=33.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.740e-04, train_time=1.323
+[gpub011:0/16] 2024-02-07 12:07:19,098 (trainer:762) INFO: 34epoch:train:801-900batch: iter_time=5.311e-04, forward_time=0.404, loss_ctc=47.204, loss_att=42.097, acc=0.756, loss=43.629, backward_time=0.335, grad_norm=44.033, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.739e-04, train_time=1.403
+[gpub011:0/16] 2024-02-07 12:09:27,701 (trainer:762) INFO: 34epoch:train:901-1000batch: iter_time=8.877e-05, forward_time=0.320, loss_ctc=52.580, loss_att=51.226, acc=0.747, loss=51.632, backward_time=0.306, grad_norm=45.479, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.739e-04, train_time=1.285
+[gpub011:0/16] 2024-02-07 12:11:54,628 (trainer:762) INFO: 34epoch:train:1001-1100batch: iter_time=4.150e-04, forward_time=0.384, loss_ctc=47.013, loss_att=40.146, acc=0.764, loss=42.206, backward_time=0.298, grad_norm=44.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.739e-04, train_time=1.469
+[gpub011:0/16] 2024-02-07 12:14:04,646 (trainer:762) INFO: 34epoch:train:1101-1200batch: iter_time=8.979e-05, forward_time=0.295, loss_ctc=46.837, loss_att=40.938, acc=0.749, loss=42.708, backward_time=0.304, grad_norm=41.385, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.739e-04, train_time=1.301
+[gpub011:0/16] 2024-02-07 12:15:32,357 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub011:0/16] 2024-02-07 12:15:51,505 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 12:15:55,104 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f7c123bc490>)
+[gpub011:0/16] 2024-02-07 12:15:55,104 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub011:0/16] 2024-02-07 12:15:55,133 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub011:0/16] 2024-02-07 12:22:58,478 (trainer:762) INFO: 34epoch:train:1201-1300batch: iter_time=3.944, forward_time=0.336, loss_ctc=48.432, loss_att=49.120, acc=0.748, loss=48.913, backward_time=0.324, grad_norm=42.455, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.739e-04, train_time=5.338
+[gpub011:0/16] 2024-02-07 12:25:11,841 (trainer:762) INFO: 34epoch:train:1301-1400batch: iter_time=2.427e-04, forward_time=0.349, loss_ctc=52.650, loss_att=50.247, acc=0.758, loss=50.968, backward_time=0.333, grad_norm=45.551, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.739e-04, train_time=1.334
+[gpub011:0/16] 2024-02-07 12:27:37,843 (trainer:762) INFO: 34epoch:train:1401-1500batch: iter_time=7.695e-05, forward_time=0.295, loss_ctc=43.650, loss_att=39.755, acc=0.758, loss=40.924, backward_time=0.309, grad_norm=35.772, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.738e-04, train_time=1.459
+[gpub011:0/16] 2024-02-07 12:30:07,240 (trainer:762) INFO: 34epoch:train:1501-1600batch: iter_time=2.417e-04, forward_time=0.330, loss_ctc=42.673, loss_att=38.455, acc=0.768, loss=39.721, backward_time=0.314, grad_norm=38.815, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.738e-04, train_time=1.495
+[gpub011:0/16] 2024-02-07 12:32:27,151 (trainer:762) INFO: 34epoch:train:1601-1700batch: iter_time=7.926e-05, forward_time=0.338, loss_ctc=49.872, loss_att=51.821, acc=0.753, loss=51.236, backward_time=0.304, grad_norm=40.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.738e-04, train_time=1.399
+[gpub011:0/16] 2024-02-07 12:34:52,240 (trainer:762) INFO: 34epoch:train:1701-1800batch: iter_time=8.687e-04, forward_time=0.357, loss_ctc=46.736, loss_att=49.068, acc=0.738, loss=48.368, backward_time=0.304, grad_norm=41.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.738e-04, train_time=1.451
+[gpub011:0/16] 2024-02-07 12:37:32,852 (trainer:762) INFO: 34epoch:train:1801-1900batch: iter_time=7.864e-05, forward_time=0.380, loss_ctc=43.029, loss_att=43.211, acc=0.788, loss=43.156, backward_time=0.338, grad_norm=36.760, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.738e-04, train_time=1.604
+[gpub011:0/16] 2024-02-07 12:39:39,293 (trainer:762) INFO: 34epoch:train:1901-2000batch: iter_time=8.192e-05, forward_time=0.303, loss_ctc=41.553, loss_att=38.434, acc=0.790, loss=39.370, backward_time=0.295, grad_norm=32.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.738e-04, train_time=1.266
+[gpub011:0/16] 2024-02-07 12:42:21,982 (trainer:762) INFO: 34epoch:train:2001-2100batch: iter_time=1.493e-04, forward_time=0.346, loss_ctc=42.009, loss_att=40.585, acc=0.767, loss=41.013, backward_time=0.320, grad_norm=39.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.737e-04, train_time=1.626
+[gpub011:0/16] 2024-02-07 12:44:39,212 (trainer:762) INFO: 34epoch:train:2101-2200batch: iter_time=8.302e-05, forward_time=0.330, loss_ctc=48.605, loss_att=49.716, acc=0.752, loss=49.383, backward_time=0.351, grad_norm=39.157, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.737e-04, train_time=1.372
+[gpub011:0/16] 2024-02-07 12:47:08,228 (trainer:762) INFO: 34epoch:train:2201-2300batch: iter_time=7.907e-05, forward_time=0.302, loss_ctc=49.412, loss_att=46.259, acc=0.759, loss=47.205, backward_time=0.301, grad_norm=45.881, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.737e-04, train_time=1.487
+[gpub011:0/16] 2024-02-07 12:49:51,300 (trainer:762) INFO: 34epoch:train:2301-2400batch: iter_time=2.586e-04, forward_time=0.349, loss_ctc=43.064, loss_att=37.859, acc=0.775, loss=39.420, backward_time=0.317, grad_norm=36.658, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.737e-04, train_time=1.633
+[gpub011:0/16] 2024-02-07 12:52:00,280 (trainer:762) INFO: 34epoch:train:2401-2500batch: iter_time=7.787e-05, forward_time=0.323, loss_ctc=44.749, loss_att=40.300, acc=0.767, loss=41.635, backward_time=0.314, grad_norm=37.986, clip=100.000, loss_scale=5.348e+33, optim_step_time=0.095, optim0_lr0=1.737e-04, train_time=1.289
+[gpub011:0/16] 2024-02-07 12:52:20,337 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub011:0/16] 2024-02-07 12:52:39,597 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub011:0/16] 2024-02-07 12:52:43,217 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7f78a0b8f490>)
+[gpub011:0/16] 2024-02-07 12:52:43,217 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub011:0/16] 2024-02-07 12:52:43,220 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.7.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.7.log
new file mode 100644
index 0000000000000000000000000000000000000000..45027d5beec42fd68ea96d47b14ab744be740466
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.7.log
@@ -0,0 +1,1729 @@
+# Running on gpub058.delta.ncsa.illinois.edu
+# Started at Sun Feb 4 17:38:07 CST 2024
+# SLURMD_NODENAME=gpub058
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2938841
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1707262664
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2938841
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[058-061]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1707089864
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[058-061]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2127211
+# SLURM_TOPOLOGY_ADDR=ss00.ss11.gpub058
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_f3531bf8-892e-4ab6-b64a-bfb8246edf7b 
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_f3531bf8-892e-4ab6-b64a-bfb8246edf7b
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_f3531bf8-892e-4ab6-b64a-bfb8246edf7b
+/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold/scratch/bbjs/peng6/espnet-whisper-public/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-whisper-public/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_f3531bf8-892e-4ab6-b64a-bfb8246edf7b
+_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-whisper-public/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_f3531bf8-892e-4ab6-b64a-bfb8246edf7b
+[gpub058:0/16] 2024-02-04 17:40:58,608 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub058:0/16] 2024-02-04 17:41:00,622 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub058:0/16] 2024-02-04 17:41:00,693 (s2t:464) INFO: Vocabulary size: 50002
+[gpub058:0/16] 2024-02-04 17:41:09,108 (abs_task:1231) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub058:0/16] 2024-02-04 17:41:09,114 (abs_task:1232) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub058:0/16] 2024-02-04 17:41:09,114 (abs_task:1235) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub058:0/16] 2024-02-04 17:41:09,114 (abs_task:1236) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub058:0/16] 2024-02-04 17:41:09,117 (abs_task:1245) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub058:0/16] 2024-02-04 17:41:14,453 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 17:41:15,372 (abs_task:1616) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fea8721e410>)
+[gpub058:0/16] 2024-02-04 17:41:15,372 (abs_task:1617) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub058:0/16] 2024-02-04 17:41:15,373 (abs_task:1618) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 17:41:29,051 (trainer:159) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub058:2127280:2127280 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2127280:2127280 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2127280:2127280 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub058:0/16] 2024-02-04 17:41:34,492 (trainer:287) INFO: 27/45epoch started
+[gpub058:0/16] 2024-02-04 17:41:34,551 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-04 17:41:52,523 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 17:41:55,961 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe89e908220>)
+[gpub058:0/16] 2024-02-04 17:41:55,961 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-04 17:41:55,964 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub059:2626468:2626468 [2] NCCL INFO cudaDriverVersion 12020
+gpub059:2626468:2626468 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.159<0>
+gpub059:2626468:2626468 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub059:2626468:2626532 [2] NCCL INFO NET/IB : No device found.
+gpub059:2626468:2626532 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.159<0> [1]hsn0:141.142.145.159<0>
+gpub059:2626468:2626532 [2] NCCL INFO Using network Socket
+gpub059:2626468:2626532 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub059:2626468:2626532 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub059:2626468:2626532 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub059:2626468:2626532 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub059:2626468:2626532 [2] NCCL INFO Connected all rings
+gpub059:2626468:2626532 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub059:2626468:2626532 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub059:2626468:2626532 [2] NCCL INFO Connected all trees
+gpub059:2626468:2626532 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub059:2626468:2626532 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub059:2626468:2626532 [2] NCCL INFO comm 0x1609ff70 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub059:2626467:2626467 [1] NCCL INFO cudaDriverVersion 12020
+gpub059:2626467:2626467 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.159<0>
+gpub059:2626467:2626467 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub059:2626467:2626533 [1] NCCL INFO NET/IB : No device found.
+gpub059:2626467:2626533 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.159<0> [1]hsn0:141.142.145.159<0>
+gpub059:2626467:2626533 [1] NCCL INFO Using network Socket
+gpub059:2626467:2626533 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub059:2626467:2626533 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub059:2626467:2626533 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub059:2626467:2626533 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub059:2626467:2626533 [1] NCCL INFO Connected all rings
+gpub059:2626467:2626533 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/Socket/1
+gpub059:2626467:2626533 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/Socket/1
+gpub059:2626467:2626533 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub059:2626467:2626533 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub059:2626467:2626533 [1] NCCL INFO Connected all trees
+gpub059:2626467:2626533 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub059:2626467:2626533 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub059:2626467:2626533 [1] NCCL INFO comm 0x17c88d70 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub059:2626466:2626466 [0] NCCL INFO cudaDriverVersion 12020
+gpub059:2626466:2626466 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.159<0>
+gpub059:2626466:2626466 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub059:2626466:2626534 [0] NCCL INFO NET/IB : No device found.
+gpub059:2626466:2626534 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.159<0> [1]hsn0:141.142.145.159<0>
+gpub059:2626466:2626534 [0] NCCL INFO Using network Socket
+gpub059:2626466:2626534 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub059:2626466:2626534 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub059:2626466:2626534 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub059:2626466:2626534 [0] NCCL INFO Connected all rings
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/Socket/1
+gpub059:2626466:2626534 [0] NCCL INFO Connected all trees
+gpub059:2626466:2626534 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub059:2626466:2626534 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub059:2626466:2626534 [0] NCCL INFO comm 0x16200a70 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub059:2626469:2626469 [3] NCCL INFO cudaDriverVersion 12020
+gpub059:2626469:2626469 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.159<0>
+gpub059:2626469:2626469 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub059:2626469:2626535 [3] NCCL INFO NET/IB : No device found.
+gpub059:2626469:2626535 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.159<0> [1]hsn0:141.142.145.159<0>
+gpub059:2626469:2626535 [3] NCCL INFO Using network Socket
+gpub059:2626469:2626535 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub059:2626469:2626535 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub059:2626469:2626535 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub059:2626469:2626535 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/Socket/1
+gpub059:2626469:2626535 [3] NCCL INFO Connected all rings
+gpub059:2626469:2626535 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub059:2626469:2626535 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub059:2626469:2626535 [3] NCCL INFO Connected all trees
+gpub059:2626469:2626535 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub059:2626469:2626535 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub059:2626469:2626535 [3] NCCL INFO comm 0xfed8830 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub058:2127281:2127281 [1] NCCL INFO cudaDriverVersion 12020
+gpub058:2127281:2127281 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2127281:2127281 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2127281:2127336 [1] NCCL INFO NET/IB : No device found.
+gpub058:2127281:2127336 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2127281:2127336 [1] NCCL INFO Using network Socket
+gpub058:2127281:2127336 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub058:2127281:2127336 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub058:2127281:2127336 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub058:2127281:2127336 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub058:2127281:2127336 [1] NCCL INFO Connected all rings
+gpub058:2127281:2127336 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub058:2127281:2127336 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub058:2127281:2127336 [1] NCCL INFO Connected all trees
+gpub058:2127281:2127336 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2127281:2127336 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2127281:2127336 [1] NCCL INFO comm 0xc9e7a80 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub058:2127280:2127337 [0] NCCL INFO NET/IB : No device found.
+gpub058:2127280:2127337 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2127280:2127337 [0] NCCL INFO Using network Socket
+gpub058:2127280:2127337 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub058:2127280:2127337 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub058:2127280:2127337 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub058:2127280:2127337 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub058:2127280:2127337 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub058:2127280:2127337 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub058:2127280:2127337 [0] NCCL INFO Connected all rings
+gpub058:2127280:2127337 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/Socket/1
+gpub058:2127280:2127337 [0] NCCL INFO Connected all trees
+gpub058:2127280:2127337 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2127280:2127337 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2127280:2127337 [0] NCCL INFO comm 0x75b2a1a0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub058:2127282:2127282 [2] NCCL INFO cudaDriverVersion 12020
+gpub058:2127282:2127282 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2127282:2127282 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2127282:2127338 [2] NCCL INFO NET/IB : No device found.
+gpub058:2127282:2127338 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2127282:2127338 [2] NCCL INFO Using network Socket
+gpub058:2127282:2127338 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub058:2127282:2127338 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub058:2127282:2127338 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub058:2127282:2127338 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub058:2127282:2127338 [2] NCCL INFO Connected all rings
+gpub058:2127282:2127338 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub058:2127282:2127338 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub058:2127282:2127338 [2] NCCL INFO Connected all trees
+gpub058:2127282:2127338 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2127282:2127338 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2127282:2127338 [2] NCCL INFO comm 0xe0e31d0 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub058:2127283:2127283 [3] NCCL INFO cudaDriverVersion 12020
+gpub058:2127283:2127283 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.158<0>
+gpub058:2127283:2127283 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub058:2127283:2127339 [3] NCCL INFO NET/IB : No device found.
+gpub058:2127283:2127339 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.158<0> [1]hsn0:141.142.145.158<0> [2]eth0:fe80::5e76:5fb4:a207:b9dd%eth0<0>
+gpub058:2127283:2127339 [3] NCCL INFO Using network Socket
+gpub058:2127283:2127339 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub058:2127283:2127339 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub058:2127283:2127339 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2127283:2127339 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/Socket/1
+gpub058:2127283:2127339 [3] NCCL INFO Connected all rings
+gpub058:2127283:2127339 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub058:2127283:2127339 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub058:2127283:2127339 [3] NCCL INFO Connected all trees
+gpub058:2127283:2127339 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub058:2127283:2127339 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub058:2127283:2127339 [3] NCCL INFO comm 0x1cb25c80 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub061:2529191:2529191 [3] NCCL INFO cudaDriverVersion 12020
+gpub061:2529191:2529191 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2529191:2529191 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2529191:2529247 [3] NCCL INFO NET/IB : No device found.
+gpub061:2529191:2529247 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2529191:2529247 [3] NCCL INFO Using network Socket
+gpub061:2529191:2529247 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub061:2529191:2529247 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub061:2529191:2529247 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub061:2529191:2529247 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/Socket/1
+gpub061:2529191:2529247 [3] NCCL INFO Connected all rings
+gpub061:2529191:2529247 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub061:2529191:2529247 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub061:2529191:2529247 [3] NCCL INFO Connected all trees
+gpub061:2529191:2529247 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2529191:2529247 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2529191:2529247 [3] NCCL INFO comm 0x1456cd30 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub061:2529190:2529190 [2] NCCL INFO cudaDriverVersion 12020
+gpub061:2529190:2529190 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2529190:2529190 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2529190:2529249 [2] NCCL INFO NET/IB : No device found.
+gpub061:2529190:2529249 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2529190:2529249 [2] NCCL INFO Using network Socket
+gpub061:2529190:2529249 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub061:2529190:2529249 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub061:2529190:2529249 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub061:2529190:2529249 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub061:2529190:2529249 [2] NCCL INFO Connected all rings
+gpub061:2529190:2529249 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub061:2529190:2529249 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub061:2529190:2529249 [2] NCCL INFO Connected all trees
+gpub061:2529190:2529249 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2529190:2529249 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2529190:2529249 [2] NCCL INFO comm 0x1b848230 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub061:2529189:2529189 [1] NCCL INFO cudaDriverVersion 12020
+gpub061:2529189:2529189 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2529189:2529189 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2529189:2529250 [1] NCCL INFO NET/IB : No device found.
+gpub061:2529189:2529250 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2529189:2529250 [1] NCCL INFO Using network Socket
+gpub061:2529189:2529250 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub061:2529189:2529250 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub061:2529189:2529250 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub061:2529189:2529250 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub061:2529189:2529250 [1] NCCL INFO Connected all rings
+gpub061:2529189:2529250 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub061:2529189:2529250 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub061:2529189:2529250 [1] NCCL INFO Connected all trees
+gpub061:2529189:2529250 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2529189:2529250 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2529189:2529250 [1] NCCL INFO comm 0xcf04490 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub060:3090775:3090775 [2] NCCL INFO cudaDriverVersion 12020
+gpub060:3090775:3090775 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.160<0>
+gpub060:3090775:3090775 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub060:3090775:3090841 [2] NCCL INFO NET/IB : No device found.
+gpub060:3090775:3090841 [2] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.160<0> [1]hsn0:141.142.145.160<0> [2]eth0:fe80::60cc:de69:5e72:f184%eth0<0>
+gpub060:3090775:3090841 [2] NCCL INFO Using network Socket
+gpub060:3090775:3090841 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub060:3090775:3090841 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub060:3090775:3090841 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub060:3090775:3090841 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub060:3090775:3090841 [2] NCCL INFO Connected all rings
+gpub060:3090775:3090841 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub060:3090775:3090841 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub060:3090775:3090841 [2] NCCL INFO Connected all trees
+gpub060:3090775:3090841 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub060:3090775:3090841 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub060:3090775:3090841 [2] NCCL INFO comm 0x19fb48b0 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub060:3090776:3090776 [3] NCCL INFO cudaDriverVersion 12020
+gpub060:3090776:3090776 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.160<0>
+gpub060:3090776:3090776 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub060:3090776:3090840 [3] NCCL INFO NET/IB : No device found.
+gpub060:3090776:3090840 [3] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.160<0> [1]hsn0:141.142.145.160<0> [2]eth0:fe80::60cc:de69:5e72:f184%eth0<0>
+gpub060:3090776:3090840 [3] NCCL INFO Using network Socket
+gpub060:3090776:3090840 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub060:3090776:3090840 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub060:3090776:3090840 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub060:3090776:3090840 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/Socket/1
+gpub060:3090776:3090840 [3] NCCL INFO Connected all rings
+gpub060:3090776:3090840 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub060:3090776:3090840 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub060:3090776:3090840 [3] NCCL INFO Connected all trees
+gpub060:3090776:3090840 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub060:3090776:3090840 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub060:3090776:3090840 [3] NCCL INFO comm 0x76cb3130 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub060:3090773:3090773 [0] NCCL INFO cudaDriverVersion 12020
+gpub060:3090773:3090773 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.160<0>
+gpub060:3090773:3090773 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub060:3090773:3090839 [0] NCCL INFO NET/IB : No device found.
+gpub060:3090773:3090839 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.160<0> [1]hsn0:141.142.145.160<0> [2]eth0:fe80::60cc:de69:5e72:f184%eth0<0>
+gpub060:3090773:3090839 [0] NCCL INFO Using network Socket
+gpub060:3090773:3090839 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub060:3090773:3090839 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub060:3090773:3090839 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub060:3090773:3090839 [0] NCCL INFO Connected all rings
+gpub060:3090773:3090839 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/Socket/1
+gpub060:3090773:3090839 [0] NCCL INFO Connected all trees
+gpub060:3090773:3090839 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub060:3090773:3090839 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub060:3090773:3090839 [0] NCCL INFO comm 0x17eefe80 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub060:3090774:3090774 [1] NCCL INFO cudaDriverVersion 12020
+gpub060:3090774:3090774 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.160<0>
+gpub060:3090774:3090774 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub060:3090774:3090838 [1] NCCL INFO NET/IB : No device found.
+gpub060:3090774:3090838 [1] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.160<0> [1]hsn0:141.142.145.160<0> [2]eth0:fe80::60cc:de69:5e72:f184%eth0<0>
+gpub060:3090774:3090838 [1] NCCL INFO Using network Socket
+gpub060:3090774:3090838 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub060:3090774:3090838 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub060:3090774:3090838 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub060:3090774:3090838 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub060:3090774:3090838 [1] NCCL INFO Connected all rings
+gpub060:3090774:3090838 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/Socket/1
+gpub060:3090774:3090838 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/Socket/1
+gpub060:3090774:3090838 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub060:3090774:3090838 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub060:3090774:3090838 [1] NCCL INFO Connected all trees
+gpub060:3090774:3090838 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub060:3090774:3090838 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub060:3090774:3090838 [1] NCCL INFO comm 0x10c3c050 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub061:2529188:2529188 [0] NCCL INFO cudaDriverVersion 12020
+gpub061:2529188:2529188 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.161<0>
+gpub061:2529188:2529188 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+gpub061:2529188:2529248 [0] NCCL INFO NET/IB : No device found.
+gpub061:2529188:2529248 [0] NCCL INFO NET/Socket : Using [0]eth1:172.28.23.161<0> [1]hsn0:141.142.145.161<0> [2]eth0:fe80::29e2:373:e40d:cf83%eth0<0>
+gpub061:2529188:2529248 [0] NCCL INFO Using network Socket
+gpub061:2529188:2529248 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub061:2529188:2529248 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub061:2529188:2529248 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub061:2529188:2529248 [0] NCCL INFO Connected all rings
+gpub061:2529188:2529248 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/Socket/1
+gpub061:2529188:2529248 [0] NCCL INFO Connected all trees
+gpub061:2529188:2529248 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub061:2529188:2529248 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub061:2529188:2529248 [0] NCCL INFO comm 0x78945ad0 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+[gpub058:0/16] 2024-02-04 17:46:29,106 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub058:0/16] 2024-02-04 17:49:01,736 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 17:49:18,251 (trainer:740) INFO: 27epoch:train:1-100batch: iter_time=1.259, forward_time=0.499, loss_ctc=54.760, loss_att=57.381, acc=0.711, loss=56.595, backward_time=0.312, grad_norm=43.713, clip=100.000, loss_scale=4.851e+33, optim_step_time=0.096, optim0_lr0=1.961e-04, train_time=4.622
+[gpub058:0/16] 2024-02-04 17:52:06,083 (trainer:740) INFO: 27epoch:train:101-200batch: iter_time=9.253e-05, forward_time=0.382, loss_ctc=52.047, loss_att=60.543, acc=0.727, loss=57.994, backward_time=0.328, grad_norm=41.866, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.961e-04, train_time=1.693
+[gpub058:0/16] 2024-02-04 17:54:35,625 (trainer:740) INFO: 27epoch:train:201-300batch: iter_time=9.590e-05, forward_time=0.393, loss_ctc=49.940, loss_att=44.887, acc=0.762, loss=46.403, backward_time=0.380, grad_norm=37.811, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.961e-04, train_time=1.494
+[gpub058:0/16] 2024-02-04 17:56:52,271 (trainer:740) INFO: 27epoch:train:301-400batch: iter_time=9.816e-05, forward_time=0.308, loss_ctc=47.556, loss_att=43.425, acc=0.775, loss=44.665, backward_time=0.296, grad_norm=36.856, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.960e-04, train_time=1.367
+[gpub058:0/16] 2024-02-04 17:59:33,295 (trainer:740) INFO: 27epoch:train:401-500batch: iter_time=9.924e-05, forward_time=0.411, loss_ctc=46.270, loss_att=43.206, acc=0.761, loss=44.125, backward_time=0.314, grad_norm=35.962, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.960e-04, train_time=1.610
+[gpub058:0/16] 2024-02-04 18:02:32,042 (trainer:740) INFO: 27epoch:train:501-600batch: iter_time=8.737e-05, forward_time=0.372, loss_ctc=42.226, loss_att=36.180, acc=0.761, loss=37.994, backward_time=0.328, grad_norm=35.579, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.960e-04, train_time=1.787
+[gpub058:0/16] 2024-02-04 18:05:14,571 (trainer:740) INFO: 27epoch:train:601-700batch: iter_time=6.522e-04, forward_time=0.415, loss_ctc=48.940, loss_att=40.745, acc=0.751, loss=43.203, backward_time=0.335, grad_norm=39.062, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.102, optim0_lr0=1.960e-04, train_time=1.626
+[gpub058:0/16] 2024-02-04 18:08:00,034 (trainer:740) INFO: 27epoch:train:701-800batch: iter_time=8.653e-05, forward_time=0.331, loss_ctc=50.212, loss_att=49.336, acc=0.744, loss=49.599, backward_time=0.315, grad_norm=38.879, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.959e-04, train_time=1.653
+[gpub058:0/16] 2024-02-04 18:11:22,267 (trainer:740) INFO: 27epoch:train:801-900batch: iter_time=2.474e-04, forward_time=0.395, loss_ctc=40.518, loss_att=42.937, acc=0.748, loss=42.211, backward_time=0.313, grad_norm=36.438, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.959e-04, train_time=2.024
+[gpub058:0/16] 2024-02-04 18:14:18,745 (trainer:740) INFO: 27epoch:train:901-1000batch: iter_time=8.188e-05, forward_time=0.348, loss_ctc=51.677, loss_att=50.205, acc=0.721, loss=50.646, backward_time=0.321, grad_norm=42.830, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.959e-04, train_time=1.764
+[gpub058:0/16] 2024-02-04 18:17:31,879 (trainer:740) INFO: 27epoch:train:1001-1100batch: iter_time=2.029e-04, forward_time=0.325, loss_ctc=51.675, loss_att=43.353, acc=0.754, loss=45.850, backward_time=0.305, grad_norm=42.824, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.959e-04, train_time=1.932
+[gpub058:0/16] 2024-02-04 18:20:47,875 (trainer:740) INFO: 27epoch:train:1101-1200batch: iter_time=3.151e-04, forward_time=0.413, loss_ctc=49.734, loss_att=50.180, acc=0.740, loss=50.046, backward_time=0.349, grad_norm=37.476, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.958e-04, train_time=1.959
+[gpub058:0/16] 2024-02-04 18:22:32,279 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-04 18:22:51,051 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 18:22:54,492 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe363034b50>)
+[gpub058:0/16] 2024-02-04 18:22:54,492 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-04 18:22:54,747 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 18:29:07,987 (trainer:740) INFO: 27epoch:train:1201-1300batch: iter_time=3.313, forward_time=0.291, loss_ctc=51.249, loss_att=51.175, acc=0.723, loss=51.197, backward_time=0.283, grad_norm=45.196, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=1.958e-04, train_time=5.001
+[gpub058:0/16] 2024-02-04 18:31:55,425 (trainer:740) INFO: 27epoch:train:1301-1400batch: iter_time=9.312e-05, forward_time=0.386, loss_ctc=55.860, loss_att=61.776, acc=0.713, loss=60.001, backward_time=0.340, grad_norm=44.908, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.958e-04, train_time=1.675
+[gpub058:0/16] 2024-02-04 18:34:04,963 (trainer:740) INFO: 27epoch:train:1401-1500batch: iter_time=9.067e-05, forward_time=0.291, loss_ctc=50.349, loss_att=52.601, acc=0.763, loss=51.925, backward_time=0.287, grad_norm=37.103, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=1.958e-04, train_time=1.294
+[gpub058:0/16] 2024-02-04 18:36:32,926 (trainer:740) INFO: 27epoch:train:1501-1600batch: iter_time=9.723e-05, forward_time=0.286, loss_ctc=43.794, loss_att=42.500, acc=0.766, loss=42.888, backward_time=0.282, grad_norm=35.945, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.957e-04, train_time=1.481
+[gpub058:0/16] 2024-02-04 18:39:25,279 (trainer:740) INFO: 27epoch:train:1601-1700batch: iter_time=5.263e-04, forward_time=0.599, loss_ctc=48.221, loss_att=42.708, acc=0.769, loss=44.362, backward_time=0.384, grad_norm=38.115, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=1.957e-04, train_time=1.723
+[gpub058:0/16] 2024-02-04 18:42:10,855 (trainer:740) INFO: 27epoch:train:1701-1800batch: iter_time=9.247e-05, forward_time=0.288, loss_ctc=42.209, loss_att=38.665, acc=0.769, loss=39.728, backward_time=0.281, grad_norm=32.795, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=1.957e-04, train_time=1.656
+[gpub058:0/16] 2024-02-04 18:44:08,114 (trainer:740) INFO: 27epoch:train:1801-1900batch: iter_time=8.968e-05, forward_time=0.289, loss_ctc=44.425, loss_att=36.016, acc=0.779, loss=38.539, backward_time=0.285, grad_norm=31.109, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.091, optim0_lr0=1.957e-04, train_time=1.170
+[gpub058:0/16] 2024-02-04 18:46:36,850 (trainer:740) INFO: 27epoch:train:1901-2000batch: iter_time=9.426e-05, forward_time=0.311, loss_ctc=50.135, loss_att=49.121, acc=0.740, loss=49.425, backward_time=0.304, grad_norm=37.788, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.956e-04, train_time=1.489
+[gpub058:0/16] 2024-02-04 18:49:11,140 (trainer:740) INFO: 27epoch:train:2001-2100batch: iter_time=4.084e-04, forward_time=0.369, loss_ctc=47.112, loss_att=46.943, acc=0.756, loss=46.994, backward_time=0.333, grad_norm=40.023, clip=100.000, loss_scale=2.934e+33, optim_step_time=0.100, optim0_lr0=1.956e-04, train_time=1.543
+[gpub058:0/16] 2024-02-04 18:51:23,729 (trainer:740) INFO: 27epoch:train:2101-2200batch: iter_time=8.695e-05, forward_time=0.291, loss_ctc=45.555, loss_att=48.771, acc=0.750, loss=47.806, backward_time=0.286, grad_norm=37.583, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.956e-04, train_time=1.326
+[gpub058:0/16] 2024-02-04 18:53:24,377 (trainer:740) INFO: 27epoch:train:2201-2300batch: iter_time=9.214e-05, forward_time=0.289, loss_ctc=48.350, loss_att=44.144, acc=0.739, loss=45.405, backward_time=0.285, grad_norm=40.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.956e-04, train_time=1.206
+[gpub058:0/16] 2024-02-04 18:56:03,532 (trainer:740) INFO: 27epoch:train:2301-2400batch: iter_time=1.731e-04, forward_time=0.416, loss_ctc=48.962, loss_att=44.420, acc=0.752, loss=45.783, backward_time=0.324, grad_norm=37.732, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.955e-04, train_time=1.592
+[gpub058:0/16] 2024-02-04 18:58:10,822 (trainer:740) INFO: 27epoch:train:2401-2500batch: iter_time=9.135e-05, forward_time=0.318, loss_ctc=51.647, loss_att=52.544, acc=0.732, loss=52.275, backward_time=0.286, grad_norm=40.293, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.955e-04, train_time=1.273
+[gpub058:0/16] 2024-02-04 18:58:31,324 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-04 18:58:49,382 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 18:58:52,798 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe89f4b37c0>)
+[gpub058:0/16] 2024-02-04 18:58:52,799 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-04 18:58:52,802 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 19:05:54,022 (trainer:740) INFO: 27epoch:train:2501-2600batch: iter_time=3.347, forward_time=0.289, loss_ctc=53.346, loss_att=55.702, acc=0.723, loss=54.995, backward_time=0.286, grad_norm=42.803, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.955e-04, train_time=4.632
+[gpub058:0/16] 2024-02-04 19:08:30,635 (trainer:740) INFO: 27epoch:train:2601-2700batch: iter_time=9.266e-05, forward_time=0.315, loss_ctc=50.976, loss_att=59.311, acc=0.737, loss=56.810, backward_time=0.300, grad_norm=40.005, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.955e-04, train_time=1.565
+[gpub058:0/16] 2024-02-04 19:10:44,426 (trainer:740) INFO: 27epoch:train:2701-2800batch: iter_time=9.445e-05, forward_time=0.385, loss_ctc=48.436, loss_att=44.084, acc=0.772, loss=45.390, backward_time=0.319, grad_norm=41.159, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.954e-04, train_time=1.339
+[gpub058:0/16] 2024-02-04 19:12:42,198 (trainer:740) INFO: 27epoch:train:2801-2900batch: iter_time=8.516e-05, forward_time=0.292, loss_ctc=46.797, loss_att=43.060, acc=0.782, loss=44.181, backward_time=0.288, grad_norm=36.376, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.954e-04, train_time=1.178
+[gpub058:0/16] 2024-02-04 19:15:00,039 (trainer:740) INFO: 27epoch:train:2901-3000batch: iter_time=9.560e-05, forward_time=0.292, loss_ctc=45.244, loss_att=42.673, acc=0.768, loss=43.444, backward_time=0.286, grad_norm=35.347, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.954e-04, train_time=1.378
+[gpub058:0/16] 2024-02-04 19:17:05,493 (trainer:740) INFO: 27epoch:train:3001-3100batch: iter_time=9.676e-05, forward_time=0.292, loss_ctc=41.817, loss_att=36.441, acc=0.765, loss=38.053, backward_time=0.287, grad_norm=33.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.954e-04, train_time=1.254
+[gpub058:0/16] 2024-02-04 19:19:19,073 (trainer:740) INFO: 27epoch:train:3101-3200batch: iter_time=9.224e-05, forward_time=0.371, loss_ctc=47.801, loss_att=39.572, acc=0.765, loss=42.041, backward_time=0.337, grad_norm=34.775, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.953e-04, train_time=1.335
+[gpub058:0/16] 2024-02-04 19:21:26,874 (trainer:740) INFO: 27epoch:train:3201-3300batch: iter_time=9.980e-05, forward_time=0.310, loss_ctc=49.109, loss_att=48.131, acc=0.756, loss=48.424, backward_time=0.286, grad_norm=37.791, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.953e-04, train_time=1.278
+[gpub058:0/16] 2024-02-04 19:23:37,270 (trainer:740) INFO: 27epoch:train:3301-3400batch: iter_time=9.845e-05, forward_time=0.288, loss_ctc=39.531, loss_att=42.247, acc=0.762, loss=41.432, backward_time=0.282, grad_norm=35.559, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.953e-04, train_time=1.304
+[gpub058:0/16] 2024-02-04 19:25:41,983 (trainer:740) INFO: 27epoch:train:3401-3500batch: iter_time=9.335e-05, forward_time=0.291, loss_ctc=49.778, loss_att=50.103, acc=0.729, loss=50.005, backward_time=0.291, grad_norm=43.365, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.953e-04, train_time=1.247
+[gpub058:0/16] 2024-02-04 19:28:09,153 (trainer:740) INFO: 27epoch:train:3501-3600batch: iter_time=8.601e-05, forward_time=0.401, loss_ctc=50.371, loss_att=44.242, acc=0.760, loss=46.080, backward_time=0.365, grad_norm=39.364, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.952e-04, train_time=1.471
+[gpub058:0/16] 2024-02-04 19:30:18,548 (trainer:740) INFO: 27epoch:train:3601-3700batch: iter_time=8.764e-05, forward_time=0.300, loss_ctc=48.495, loss_att=49.907, acc=0.747, loss=49.483, backward_time=0.289, grad_norm=35.990, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.952e-04, train_time=1.294
+[gpub058:0/16] 2024-02-04 19:31:43,939 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-04 19:32:02,455 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 19:32:06,125 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3ab0cfca0>)
+[gpub058:0/16] 2024-02-04 19:32:06,126 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-04 19:32:06,151 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 19:38:17,946 (trainer:740) INFO: 27epoch:train:3701-3800batch: iter_time=3.257, forward_time=0.292, loss_ctc=49.495, loss_att=50.450, acc=0.725, loss=50.164, backward_time=0.283, grad_norm=41.445, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.952e-04, train_time=4.794
+[gpub058:0/16] 2024-02-04 19:41:07,603 (trainer:740) INFO: 27epoch:train:3801-3900batch: iter_time=7.794e-04, forward_time=0.415, loss_ctc=54.969, loss_att=61.843, acc=0.710, loss=59.781, backward_time=0.356, grad_norm=44.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.952e-04, train_time=1.696
+[gpub058:0/16] 2024-02-04 19:43:18,499 (trainer:740) INFO: 27epoch:train:3901-4000batch: iter_time=7.772e-05, forward_time=0.300, loss_ctc=49.968, loss_att=53.136, acc=0.758, loss=52.186, backward_time=0.289, grad_norm=38.625, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.951e-04, train_time=1.309
+[gpub058:0/16] 2024-02-04 19:45:28,498 (trainer:740) INFO: 27epoch:train:4001-4100batch: iter_time=7.967e-05, forward_time=0.288, loss_ctc=43.648, loss_att=41.721, acc=0.770, loss=42.299, backward_time=0.283, grad_norm=37.502, clip=100.000, loss_scale=5.867e+33, optim_step_time=0.091, optim0_lr0=1.951e-04, train_time=1.300
+[gpub058:0/16] 2024-02-04 19:47:56,579 (trainer:740) INFO: 27epoch:train:4101-4200batch: iter_time=8.188e-05, forward_time=0.319, loss_ctc=48.695, loss_att=42.322, acc=0.764, loss=44.234, backward_time=0.288, grad_norm=39.268, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.951e-04, train_time=1.481
+[gpub058:0/16] 2024-02-04 19:50:30,564 (trainer:740) INFO: 27epoch:train:4201-4300batch: iter_time=8.729e-05, forward_time=0.394, loss_ctc=41.588, loss_att=37.690, acc=0.777, loss=38.859, backward_time=0.365, grad_norm=32.277, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.951e-04, train_time=1.540
+[gpub058:0/16] 2024-02-04 19:53:02,997 (trainer:740) INFO: 27epoch:train:4301-4400batch: iter_time=8.443e-05, forward_time=0.299, loss_ctc=44.604, loss_att=36.181, acc=0.773, loss=38.708, backward_time=0.284, grad_norm=32.926, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.950e-04, train_time=1.524
+[gpub058:0/16] 2024-02-04 19:55:10,590 (trainer:740) INFO: 27epoch:train:4401-4500batch: iter_time=8.660e-05, forward_time=0.292, loss_ctc=49.839, loss_att=47.597, acc=0.737, loss=48.270, backward_time=0.285, grad_norm=39.092, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.950e-04, train_time=1.276
+[gpub058:0/16] 2024-02-04 19:57:35,074 (trainer:740) INFO: 27epoch:train:4501-4600batch: iter_time=8.498e-05, forward_time=0.291, loss_ctc=46.598, loss_att=46.465, acc=0.749, loss=46.505, backward_time=0.285, grad_norm=39.171, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.950e-04, train_time=1.444
+[gpub058:0/16] 2024-02-04 20:00:07,152 (trainer:740) INFO: 27epoch:train:4601-4700batch: iter_time=8.447e-05, forward_time=0.381, loss_ctc=45.107, loss_att=49.453, acc=0.738, loss=48.149, backward_time=0.314, grad_norm=37.534, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.109, optim0_lr0=1.950e-04, train_time=1.521
+[gpub058:0/16] 2024-02-04 20:02:40,088 (trainer:740) INFO: 27epoch:train:4701-4800batch: iter_time=8.701e-05, forward_time=0.325, loss_ctc=47.091, loss_att=41.041, acc=0.743, loss=42.856, backward_time=0.292, grad_norm=40.113, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.950e-04, train_time=1.529
+[gpub058:0/16] 2024-02-04 20:04:57,056 (trainer:740) INFO: 27epoch:train:4801-4900batch: iter_time=8.643e-05, forward_time=0.297, loss_ctc=48.147, loss_att=43.772, acc=0.748, loss=45.085, backward_time=0.284, grad_norm=37.569, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.949e-04, train_time=1.369
+[gpub058:0/16] 2024-02-04 20:07:36,729 (trainer:740) INFO: 27epoch:train:4901-5000batch: iter_time=8.753e-05, forward_time=0.291, loss_ctc=51.270, loss_att=53.018, acc=0.724, loss=52.494, backward_time=0.286, grad_norm=41.416, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=1.949e-04, train_time=1.596
+[gpub058:0/16] 2024-02-04 20:07:56,772 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-04 20:08:15,182 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 20:08:18,652 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3b5aa2f80>)
+[gpub058:0/16] 2024-02-04 20:08:18,653 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-04 20:08:18,770 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 20:24:17,509 (trainer:740) INFO: 27epoch:train:5001-5100batch: iter_time=3.606, forward_time=0.453, loss_ctc=52.840, loss_att=56.628, acc=0.723, loss=55.492, backward_time=0.307, grad_norm=42.503, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.949e-04, train_time=10.009
+[gpub058:0/16] 2024-02-04 20:29:29,875 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 20:33:15,845 (trainer:740) INFO: 27epoch:train:5101-5200batch: iter_time=9.567e-05, forward_time=0.289, loss_ctc=50.422, loss_att=59.089, acc=0.742, loss=56.489, backward_time=0.285, grad_norm=38.492, clip=100.000, loss_scale=8.339e+33, optim_step_time=0.092, optim0_lr0=1.949e-04, train_time=5.383
+[gpub058:0/16] 2024-02-04 20:45:43,751 (trainer:740) INFO: 27epoch:train:5201-5300batch: iter_time=9.848e-05, forward_time=0.289, loss_ctc=48.239, loss_att=44.160, acc=0.772, loss=45.384, backward_time=0.286, grad_norm=37.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.948e-04, train_time=7.479
+[gpub058:0/16] 2024-02-04 21:00:33,989 (trainer:740) INFO: 27epoch:train:5301-5400batch: iter_time=9.130e-05, forward_time=0.370, loss_ctc=46.139, loss_att=42.616, acc=0.785, loss=43.673, backward_time=0.421, grad_norm=35.079, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.948e-04, train_time=8.901
+[gpub058:0/16] 2024-02-04 21:14:18,330 (trainer:740) INFO: 27epoch:train:5401-5500batch: iter_time=9.172e-05, forward_time=0.289, loss_ctc=44.952, loss_att=43.130, acc=0.768, loss=43.677, backward_time=0.280, grad_norm=36.253, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.948e-04, train_time=8.245
+[gpub058:0/16] 2024-02-04 21:20:01,103 (trainer:740) INFO: 27epoch:train:5501-5600batch: iter_time=9.072e-05, forward_time=0.298, loss_ctc=41.714, loss_att=36.119, acc=0.767, loss=37.798, backward_time=0.288, grad_norm=32.405, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.948e-04, train_time=3.427
+[gpub058:0/16] 2024-02-04 21:30:52,773 (trainer:740) INFO: 27epoch:train:5601-5700batch: iter_time=8.951e-05, forward_time=0.287, loss_ctc=47.793, loss_att=39.825, acc=0.764, loss=42.216, backward_time=0.280, grad_norm=35.367, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.947e-04, train_time=6.516
+[gpub058:0/16] 2024-02-04 21:44:00,225 (trainer:740) INFO: 27epoch:train:5701-5800batch: iter_time=9.916e-05, forward_time=0.407, loss_ctc=49.260, loss_att=48.744, acc=0.754, loss=48.899, backward_time=0.315, grad_norm=37.406, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.947e-04, train_time=7.874
+[gpub058:0/16] 2024-02-04 21:56:05,443 (trainer:740) INFO: 27epoch:train:5801-5900batch: iter_time=9.465e-05, forward_time=0.285, loss_ctc=39.565, loss_att=42.007, acc=0.765, loss=41.275, backward_time=0.278, grad_norm=35.548, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.947e-04, train_time=7.253
+[gpub058:0/16] 2024-02-04 22:04:40,096 (trainer:740) INFO: 27epoch:train:5901-6000batch: iter_time=1.109e-04, forward_time=0.297, loss_ctc=49.674, loss_att=49.113, acc=0.732, loss=49.281, backward_time=0.284, grad_norm=43.267, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.947e-04, train_time=5.146
+[gpub058:0/16] 2024-02-04 22:20:41,611 (trainer:740) INFO: 27epoch:train:6001-6100batch: iter_time=1.088e-04, forward_time=0.288, loss_ctc=49.440, loss_att=43.808, acc=0.761, loss=45.498, backward_time=0.278, grad_norm=39.326, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.946e-04, train_time=9.615
+[gpub058:0/16] 2024-02-04 22:27:58,534 (trainer:740) INFO: 27epoch:train:6101-6200batch: iter_time=1.177e-04, forward_time=0.429, loss_ctc=48.177, loss_att=49.691, acc=0.751, loss=49.237, backward_time=0.310, grad_norm=36.868, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.946e-04, train_time=4.369
+[gpub058:0/16] 2024-02-04 22:37:04,354 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-04 22:37:23,182 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 22:37:26,731 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd96ee23dc0>)
+[gpub058:0/16] 2024-02-04 22:37:26,731 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-04 22:37:26,863 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 22:45:02,961 (trainer:740) INFO: 27epoch:train:6201-6300batch: iter_time=4.118, forward_time=0.472, loss_ctc=49.468, loss_att=49.098, acc=0.732, loss=49.209, backward_time=0.308, grad_norm=44.876, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.946e-04, train_time=10.244
+[gpub058:0/16] 2024-02-04 22:48:01,297 (trainer:740) INFO: 27epoch:train:6301-6400batch: iter_time=8.757e-05, forward_time=0.291, loss_ctc=55.243, loss_att=60.623, acc=0.718, loss=59.009, backward_time=0.285, grad_norm=43.489, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.946e-04, train_time=1.784
+[gpub058:0/16] 2024-02-04 22:50:57,413 (trainer:740) INFO: 27epoch:train:6401-6500batch: iter_time=8.880e-05, forward_time=0.309, loss_ctc=49.391, loss_att=51.735, acc=0.766, loss=51.032, backward_time=0.291, grad_norm=38.171, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.945e-04, train_time=1.761
+[gpub058:0/16] 2024-02-04 22:53:11,905 (trainer:740) INFO: 27epoch:train:6501-6600batch: iter_time=9.475e-05, forward_time=0.433, loss_ctc=43.298, loss_att=41.709, acc=0.773, loss=42.186, backward_time=0.329, grad_norm=35.465, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.945e-04, train_time=1.345
+[gpub058:0/16] 2024-02-04 22:55:33,018 (trainer:740) INFO: 27epoch:train:6601-6700batch: iter_time=1.006e-04, forward_time=0.291, loss_ctc=47.683, loss_att=41.519, acc=0.775, loss=43.369, backward_time=0.288, grad_norm=38.923, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.945e-04, train_time=1.411
+[gpub058:0/16] 2024-02-04 22:58:16,613 (trainer:740) INFO: 27epoch:train:6701-6800batch: iter_time=9.617e-05, forward_time=0.287, loss_ctc=41.155, loss_att=38.434, acc=0.773, loss=39.251, backward_time=0.281, grad_norm=31.720, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.945e-04, train_time=1.636
+[gpub058:0/16] 2024-02-04 23:01:23,056 (trainer:740) INFO: 27epoch:train:6801-6900batch: iter_time=3.646e-04, forward_time=0.430, loss_ctc=44.262, loss_att=35.702, acc=0.781, loss=38.270, backward_time=0.354, grad_norm=32.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.944e-04, train_time=1.864
+[gpub058:0/16] 2024-02-04 23:03:21,115 (trainer:740) INFO: 27epoch:train:6901-7000batch: iter_time=9.047e-05, forward_time=0.294, loss_ctc=49.561, loss_att=47.533, acc=0.747, loss=48.142, backward_time=0.286, grad_norm=38.564, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.944e-04, train_time=1.180
+[gpub058:0/16] 2024-02-04 23:05:30,859 (trainer:740) INFO: 27epoch:train:7001-7100batch: iter_time=9.077e-05, forward_time=0.292, loss_ctc=46.351, loss_att=45.805, acc=0.760, loss=45.969, backward_time=0.289, grad_norm=38.791, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.944e-04, train_time=1.297
+[gpub058:0/16] 2024-02-04 23:08:19,747 (trainer:740) INFO: 27epoch:train:7101-7200batch: iter_time=9.027e-05, forward_time=0.341, loss_ctc=45.270, loss_att=49.709, acc=0.747, loss=48.378, backward_time=0.284, grad_norm=37.582, clip=100.000, loss_scale=7.217e+33, optim_step_time=0.092, optim0_lr0=1.944e-04, train_time=1.689
+[gpub058:0/16] 2024-02-04 23:10:54,330 (trainer:740) INFO: 27epoch:train:7201-7300batch: iter_time=9.100e-05, forward_time=0.345, loss_ctc=46.818, loss_att=42.804, acc=0.743, loss=44.008, backward_time=0.387, grad_norm=39.769, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.943e-04, train_time=1.546
+[gpub058:0/16] 2024-02-04 23:13:15,132 (trainer:740) INFO: 27epoch:train:7301-7400batch: iter_time=9.934e-05, forward_time=0.290, loss_ctc=47.672, loss_att=43.402, acc=0.756, loss=44.683, backward_time=0.285, grad_norm=37.917, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.943e-04, train_time=1.408
+[gpub058:0/16] 2024-02-04 23:15:30,658 (trainer:740) INFO: 27epoch:train:7401-7500batch: iter_time=8.038e-05, forward_time=0.290, loss_ctc=51.333, loss_att=52.747, acc=0.732, loss=52.323, backward_time=0.284, grad_norm=42.071, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.943e-04, train_time=1.355
+[gpub058:0/16] 2024-02-04 23:15:50,698 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-04 23:16:09,666 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 23:16:13,223 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3d70caec0>)
+[gpub058:0/16] 2024-02-04 23:16:13,223 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-04 23:16:13,272 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 23:24:11,535 (trainer:740) INFO: 27epoch:train:7501-7600batch: iter_time=3.754, forward_time=0.461, loss_ctc=52.606, loss_att=56.716, acc=0.717, loss=55.483, backward_time=0.319, grad_norm=43.217, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.943e-04, train_time=5.209
+[gpub058:0/16] 2024-02-04 23:26:27,385 (trainer:740) INFO: 27epoch:train:7601-7700batch: iter_time=8.868e-05, forward_time=0.292, loss_ctc=50.597, loss_att=59.666, acc=0.731, loss=56.945, backward_time=0.287, grad_norm=41.802, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.942e-04, train_time=1.358
+[gpub058:0/16] 2024-02-04 23:27:23,880 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-04 23:29:02,701 (trainer:740) INFO: 27epoch:train:7701-7800batch: iter_time=8.741e-05, forward_time=0.289, loss_ctc=47.787, loss_att=43.932, acc=0.770, loss=45.088, backward_time=0.283, grad_norm=39.838, clip=100.000, loss_scale=6.923e+33, optim_step_time=0.092, optim0_lr0=1.942e-04, train_time=1.553
+[gpub058:0/16] 2024-02-04 23:31:20,686 (trainer:740) INFO: 27epoch:train:7801-7900batch: iter_time=6.298e-04, forward_time=0.444, loss_ctc=46.266, loss_att=43.203, acc=0.780, loss=44.122, backward_time=0.349, grad_norm=35.582, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.942e-04, train_time=1.380
+[gpub058:0/16] 2024-02-04 23:33:59,203 (trainer:740) INFO: 27epoch:train:7901-8000batch: iter_time=8.727e-05, forward_time=0.316, loss_ctc=44.586, loss_att=42.530, acc=0.766, loss=43.147, backward_time=0.283, grad_norm=35.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.942e-04, train_time=1.585
+[gpub058:0/16] 2024-02-04 23:36:01,264 (trainer:740) INFO: 27epoch:train:8001-8100batch: iter_time=9.342e-05, forward_time=0.287, loss_ctc=41.219, loss_att=35.291, acc=0.768, loss=37.069, backward_time=0.281, grad_norm=32.225, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.941e-04, train_time=1.220
+[gpub058:0/16] 2024-02-04 23:38:39,300 (trainer:740) INFO: 27epoch:train:8101-8200batch: iter_time=9.679e-05, forward_time=0.413, loss_ctc=47.047, loss_att=39.747, acc=0.757, loss=41.937, backward_time=0.336, grad_norm=35.298, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.941e-04, train_time=1.580
+[gpub058:0/16] 2024-02-04 23:41:12,926 (trainer:740) INFO: 27epoch:train:8201-8300batch: iter_time=9.884e-05, forward_time=0.289, loss_ctc=48.676, loss_att=48.025, acc=0.750, loss=48.220, backward_time=0.284, grad_norm=37.629, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.941e-04, train_time=1.536
+[gpub058:0/16] 2024-02-04 23:43:10,309 (trainer:740) INFO: 27epoch:train:8301-8400batch: iter_time=8.890e-05, forward_time=0.287, loss_ctc=39.026, loss_att=42.075, acc=0.755, loss=41.160, backward_time=0.282, grad_norm=34.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.941e-04, train_time=1.174
+[gpub058:0/16] 2024-02-04 23:45:31,055 (trainer:740) INFO: 27epoch:train:8401-8500batch: iter_time=9.294e-05, forward_time=0.367, loss_ctc=48.934, loss_att=48.889, acc=0.727, loss=48.902, backward_time=0.325, grad_norm=41.105, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.940e-04, train_time=1.407
+[gpub058:0/16] 2024-02-04 23:48:35,548 (trainer:740) INFO: 27epoch:train:8501-8600batch: iter_time=9.397e-05, forward_time=0.332, loss_ctc=49.032, loss_att=42.099, acc=0.759, loss=44.179, backward_time=0.344, grad_norm=39.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.940e-04, train_time=1.845
+[gpub058:0/16] 2024-02-04 23:50:32,259 (trainer:740) INFO: 27epoch:train:8601-8700batch: iter_time=8.963e-05, forward_time=0.291, loss_ctc=48.097, loss_att=49.590, acc=0.744, loss=49.142, backward_time=0.286, grad_norm=37.073, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.940e-04, train_time=1.167
+[gpub058:0/16] 2024-02-04 23:51:46,550 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-04 23:52:05,397 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-04 23:52:08,896 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd9b13ebb50>)
+[gpub058:0/16] 2024-02-04 23:52:08,896 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-04 23:52:08,927 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-04 23:58:50,856 (trainer:740) INFO: 27epoch:train:8701-8800batch: iter_time=3.557, forward_time=0.427, loss_ctc=48.828, loss_att=48.665, acc=0.725, loss=48.714, backward_time=0.305, grad_norm=41.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.940e-04, train_time=4.986
+[gpub058:0/16] 2024-02-05 00:01:36,620 (trainer:740) INFO: 27epoch:train:8801-8900batch: iter_time=0.002, forward_time=0.395, loss_ctc=54.646, loss_att=59.902, acc=0.715, loss=58.325, backward_time=0.360, grad_norm=44.047, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.939e-04, train_time=1.657
+[gpub058:0/16] 2024-02-05 00:03:48,726 (trainer:740) INFO: 27epoch:train:8901-9000batch: iter_time=7.408e-05, forward_time=0.432, loss_ctc=48.968, loss_att=52.023, acc=0.761, loss=51.107, backward_time=0.334, grad_norm=39.053, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.939e-04, train_time=1.321
+[gpub058:0/16] 2024-02-05 00:06:03,731 (trainer:740) INFO: 27epoch:train:9001-9100batch: iter_time=9.361e-05, forward_time=0.301, loss_ctc=43.436, loss_att=41.445, acc=0.773, loss=42.042, backward_time=0.288, grad_norm=34.271, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.939e-04, train_time=1.350
+[gpub058:0/16] 2024-02-05 00:08:40,216 (trainer:740) INFO: 27epoch:train:9101-9200batch: iter_time=8.740e-04, forward_time=0.400, loss_ctc=46.463, loss_att=41.119, acc=0.769, loss=42.722, backward_time=0.326, grad_norm=37.979, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.939e-04, train_time=1.565
+[gpub058:0/16] 2024-02-05 00:10:53,592 (trainer:740) INFO: 27epoch:train:9201-9300batch: iter_time=0.001, forward_time=0.386, loss_ctc=40.863, loss_att=37.422, acc=0.779, loss=38.455, backward_time=0.380, grad_norm=31.772, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.112, optim0_lr0=1.938e-04, train_time=1.334
+[gpub058:0/16] 2024-02-05 00:13:14,082 (trainer:740) INFO: 27epoch:train:9301-9400batch: iter_time=9.001e-05, forward_time=0.313, loss_ctc=44.349, loss_att=35.843, acc=0.775, loss=38.395, backward_time=0.295, grad_norm=31.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.938e-04, train_time=1.405
+[gpub058:0/16] 2024-02-05 00:16:02,101 (trainer:740) INFO: 27epoch:train:9401-9500batch: iter_time=9.264e-05, forward_time=0.402, loss_ctc=48.737, loss_att=46.751, acc=0.741, loss=47.347, backward_time=0.302, grad_norm=38.001, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.938e-04, train_time=1.680
+[gpub058:0/16] 2024-02-05 00:18:20,563 (trainer:740) INFO: 27epoch:train:9501-9600batch: iter_time=1.547e-04, forward_time=0.423, loss_ctc=45.993, loss_att=45.411, acc=0.751, loss=45.586, backward_time=0.330, grad_norm=39.209, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.938e-04, train_time=1.380
+[gpub058:0/16] 2024-02-05 00:20:41,977 (trainer:740) INFO: 27epoch:train:9601-9700batch: iter_time=2.452e-04, forward_time=0.433, loss_ctc=44.416, loss_att=48.835, acc=0.740, loss=47.509, backward_time=0.309, grad_norm=37.864, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.938e-04, train_time=1.418
+[gpub058:0/16] 2024-02-05 00:23:15,926 (trainer:740) INFO: 27epoch:train:9701-9800batch: iter_time=8.279e-05, forward_time=0.316, loss_ctc=46.660, loss_att=40.736, acc=0.745, loss=42.513, backward_time=0.310, grad_norm=41.136, clip=100.000, loss_scale=8.619e+33, optim_step_time=0.093, optim0_lr0=1.937e-04, train_time=1.537
+[gpub058:0/16] 2024-02-05 00:25:57,910 (trainer:740) INFO: 27epoch:train:9801-9900batch: iter_time=7.196e-04, forward_time=0.436, loss_ctc=47.420, loss_att=43.323, acc=0.750, loss=44.552, backward_time=0.334, grad_norm=38.265, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.937e-04, train_time=1.622
+[gpub058:0/16] 2024-02-05 00:28:13,710 (trainer:740) INFO: 27epoch:train:9901-10000batch: iter_time=0.001, forward_time=0.438, loss_ctc=49.975, loss_att=52.687, acc=0.726, loss=51.873, backward_time=0.356, grad_norm=41.545, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=1.937e-04, train_time=1.358
+[gpub058:0/16] 2024-02-05 00:28:34,013 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-05 00:28:52,654 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 00:28:56,175 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd9e0d7f790>)
+[gpub058:0/16] 2024-02-05 00:28:56,175 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-05 00:28:56,209 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 00:37:29,309 (trainer:740) INFO: 27epoch:train:10001-10100batch: iter_time=4.108, forward_time=0.290, loss_ctc=52.018, loss_att=56.784, acc=0.724, loss=55.355, backward_time=0.286, grad_norm=42.709, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.937e-04, train_time=5.556
+[gpub058:0/16] 2024-02-05 00:39:44,464 (trainer:740) INFO: 27epoch:train:10101-10200batch: iter_time=8.630e-05, forward_time=0.293, loss_ctc=50.317, loss_att=59.473, acc=0.742, loss=56.726, backward_time=0.288, grad_norm=38.406, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=1.936e-04, train_time=1.351
+[gpub058:0/16] 2024-02-05 00:42:25,287 (trainer:740) INFO: 27epoch:train:10201-10300batch: iter_time=8.770e-05, forward_time=0.418, loss_ctc=47.414, loss_att=43.966, acc=0.772, loss=45.000, backward_time=0.345, grad_norm=39.107, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=1.936e-04, train_time=1.608
+[gpub058:0/16] 2024-02-05 00:44:33,394 (trainer:740) INFO: 27epoch:train:10301-10400batch: iter_time=8.378e-05, forward_time=0.290, loss_ctc=46.111, loss_att=42.057, acc=0.787, loss=43.273, backward_time=0.286, grad_norm=34.729, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=1.936e-04, train_time=1.281
+[gpub058:0/16] 2024-02-05 00:45:20,906 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 00:46:44,510 (trainer:740) INFO: 27epoch:train:10401-10500batch: iter_time=8.677e-05, forward_time=0.291, loss_ctc=44.292, loss_att=42.939, acc=0.769, loss=43.345, backward_time=0.286, grad_norm=39.870, clip=100.000, loss_scale=7.448e+33, optim_step_time=0.091, optim0_lr0=1.936e-04, train_time=1.311
+[gpub058:0/16] 2024-02-05 00:49:26,846 (trainer:740) INFO: 27epoch:train:10501-10600batch: iter_time=8.411e-05, forward_time=0.397, loss_ctc=41.178, loss_att=35.908, acc=0.768, loss=37.489, backward_time=0.360, grad_norm=32.385, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.935e-04, train_time=1.623
+[gpub058:0/16] 2024-02-05 00:51:46,027 (trainer:740) INFO: 27epoch:train:10601-10700batch: iter_time=8.952e-05, forward_time=0.305, loss_ctc=47.141, loss_att=39.839, acc=0.766, loss=42.030, backward_time=0.284, grad_norm=35.490, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.935e-04, train_time=1.391
+[gpub058:0/16] 2024-02-05 00:53:47,332 (trainer:740) INFO: 27epoch:train:10701-10800batch: iter_time=8.853e-05, forward_time=0.292, loss_ctc=48.819, loss_att=48.561, acc=0.755, loss=48.638, backward_time=0.288, grad_norm=38.158, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.935e-04, train_time=1.213
+[gpub058:0/16] 2024-02-05 00:56:29,992 (trainer:740) INFO: 27epoch:train:10801-10900batch: iter_time=7.331e-04, forward_time=0.364, loss_ctc=38.913, loss_att=41.935, acc=0.767, loss=41.028, backward_time=0.347, grad_norm=35.164, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.935e-04, train_time=1.626
+[gpub058:0/16] 2024-02-05 00:58:56,988 (trainer:740) INFO: 27epoch:train:10901-11000batch: iter_time=8.458e-05, forward_time=0.336, loss_ctc=48.673, loss_att=49.359, acc=0.734, loss=49.153, backward_time=0.306, grad_norm=41.505, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.934e-04, train_time=1.469
+[gpub058:0/16] 2024-02-05 01:00:53,915 (trainer:740) INFO: 27epoch:train:11001-11100batch: iter_time=8.804e-05, forward_time=0.290, loss_ctc=48.846, loss_att=43.889, acc=0.762, loss=45.376, backward_time=0.285, grad_norm=38.768, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.934e-04, train_time=1.169
+[gpub058:0/16] 2024-02-05 01:03:09,677 (trainer:740) INFO: 27epoch:train:11101-11200batch: iter_time=9.224e-05, forward_time=0.291, loss_ctc=47.734, loss_att=49.106, acc=0.750, loss=48.695, backward_time=0.285, grad_norm=35.353, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.934e-04, train_time=1.358
+[gpub058:0/16] 2024-02-05 01:05:03,074 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-05 01:05:21,784 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 01:05:25,587 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd9910c3d90>)
+[gpub058:0/16] 2024-02-05 01:05:25,587 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-05 01:05:25,590 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 01:12:14,706 (trainer:740) INFO: 27epoch:train:11201-11300batch: iter_time=3.762, forward_time=0.462, loss_ctc=48.872, loss_att=49.664, acc=0.728, loss=49.426, backward_time=0.346, grad_norm=41.745, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.934e-04, train_time=5.449
+[gpub058:0/16] 2024-02-05 01:14:29,160 (trainer:740) INFO: 27epoch:train:11301-11400batch: iter_time=7.760e-05, forward_time=0.290, loss_ctc=54.402, loss_att=60.486, acc=0.712, loss=58.661, backward_time=0.286, grad_norm=43.289, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.933e-04, train_time=1.345
+[gpub058:0/16] 2024-02-05 01:17:32,782 (trainer:740) INFO: 27epoch:train:11401-11500batch: iter_time=8.385e-05, forward_time=0.477, loss_ctc=48.669, loss_att=52.007, acc=0.763, loss=51.006, backward_time=0.329, grad_norm=37.424, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.933e-04, train_time=1.836
+[gpub058:0/16] 2024-02-05 01:20:21,038 (trainer:740) INFO: 27epoch:train:11501-11600batch: iter_time=8.640e-05, forward_time=0.287, loss_ctc=42.848, loss_att=41.313, acc=0.773, loss=41.773, backward_time=0.279, grad_norm=35.417, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.933e-04, train_time=1.682
+[gpub058:0/16] 2024-02-05 01:22:38,859 (trainer:740) INFO: 27epoch:train:11601-11700batch: iter_time=8.432e-05, forward_time=0.290, loss_ctc=47.473, loss_att=41.604, acc=0.769, loss=43.365, backward_time=0.284, grad_norm=37.456, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.933e-04, train_time=1.379
+[gpub058:0/16] 2024-02-05 01:25:30,504 (trainer:740) INFO: 27epoch:train:11701-11800batch: iter_time=3.973e-04, forward_time=0.416, loss_ctc=40.973, loss_att=37.969, acc=0.776, loss=38.870, backward_time=0.352, grad_norm=31.846, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.932e-04, train_time=1.716
+[gpub058:0/16] 2024-02-05 01:28:14,657 (trainer:740) INFO: 27epoch:train:11801-11900batch: iter_time=9.010e-05, forward_time=0.287, loss_ctc=43.989, loss_att=35.533, acc=0.778, loss=38.070, backward_time=0.281, grad_norm=30.872, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.932e-04, train_time=1.641
+[gpub058:0/16] 2024-02-05 01:30:27,167 (trainer:740) INFO: 27epoch:train:11901-12000batch: iter_time=9.269e-05, forward_time=0.288, loss_ctc=49.238, loss_att=47.884, acc=0.737, loss=48.290, backward_time=0.284, grad_norm=38.729, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.932e-04, train_time=1.325
+[gpub058:0/16] 2024-02-05 01:33:07,961 (trainer:740) INFO: 27epoch:train:12001-12100batch: iter_time=8.508e-05, forward_time=0.409, loss_ctc=45.750, loss_att=44.646, acc=0.756, loss=44.977, backward_time=0.353, grad_norm=38.152, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.932e-04, train_time=1.608
+[gpub058:0/16] 2024-02-05 01:35:58,089 (trainer:740) INFO: 27epoch:train:12101-12200batch: iter_time=8.376e-05, forward_time=0.318, loss_ctc=44.445, loss_att=48.118, acc=0.744, loss=47.016, backward_time=0.287, grad_norm=36.351, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.931e-04, train_time=1.701
+[gpub058:0/16] 2024-02-05 01:38:10,422 (trainer:740) INFO: 27epoch:train:12201-12300batch: iter_time=8.794e-05, forward_time=0.288, loss_ctc=45.717, loss_att=40.647, acc=0.747, loss=42.168, backward_time=0.281, grad_norm=40.244, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.091, optim0_lr0=1.931e-04, train_time=1.323
+[gpub058:0/16] 2024-02-05 01:40:48,744 (trainer:740) INFO: 27epoch:train:12301-12400batch: iter_time=8.311e-05, forward_time=0.338, loss_ctc=47.730, loss_att=43.471, acc=0.748, loss=44.749, backward_time=0.332, grad_norm=39.359, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.931e-04, train_time=1.583
+[gpub058:0/16] 2024-02-05 01:43:39,997 (trainer:740) INFO: 27epoch:train:12401-12500batch: iter_time=0.002, forward_time=0.357, loss_ctc=49.958, loss_att=51.923, acc=0.730, loss=51.334, backward_time=0.308, grad_norm=39.487, clip=100.000, loss_scale=8.100e+33, optim_step_time=0.096, optim0_lr0=1.931e-04, train_time=1.712
+[gpub058:0/16] 2024-02-05 01:44:00,171 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-05 01:44:18,986 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 01:44:22,781 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd9a6c1f7c0>)
+[gpub058:0/16] 2024-02-05 01:44:22,781 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-05 01:44:22,784 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 01:55:12,445 (trainer:740) INFO: 27epoch:train:12501-12600batch: iter_time=3.510, forward_time=0.307, loss_ctc=51.768, loss_att=53.649, acc=0.726, loss=53.085, backward_time=0.292, grad_norm=41.941, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.931e-04, train_time=6.925
+[gpub058:0/16] 2024-02-05 01:57:31,655 (trainer:740) INFO: 27epoch:train:12601-12700batch: iter_time=8.295e-05, forward_time=0.397, loss_ctc=50.334, loss_att=59.071, acc=0.733, loss=56.450, backward_time=0.341, grad_norm=39.903, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.930e-04, train_time=1.392
+[gpub058:0/16] 2024-02-05 02:00:15,262 (trainer:740) INFO: 27epoch:train:12701-12800batch: iter_time=8.369e-05, forward_time=0.323, loss_ctc=47.159, loss_att=43.460, acc=0.771, loss=44.570, backward_time=0.291, grad_norm=38.349, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.930e-04, train_time=1.635
+[gpub058:0/16] 2024-02-05 02:02:21,611 (trainer:740) INFO: 27epoch:train:12801-12900batch: iter_time=8.538e-05, forward_time=0.302, loss_ctc=45.907, loss_att=42.467, acc=0.782, loss=43.499, backward_time=0.286, grad_norm=34.341, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.930e-04, train_time=1.264
+[gpub058:0/16] 2024-02-05 02:04:25,653 (trainer:740) INFO: 27epoch:train:12901-13000batch: iter_time=7.895e-05, forward_time=0.392, loss_ctc=44.215, loss_att=41.657, acc=0.769, loss=42.424, backward_time=0.328, grad_norm=37.803, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.930e-04, train_time=1.240
+[gpub058:0/16] 2024-02-05 02:06:46,888 (trainer:740) INFO: 27epoch:train:13001-13100batch: iter_time=7.969e-05, forward_time=0.296, loss_ctc=41.140, loss_att=34.876, acc=0.772, loss=36.756, backward_time=0.302, grad_norm=32.083, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.929e-04, train_time=1.411
+[gpub058:0/16] 2024-02-05 02:09:17,045 (trainer:740) INFO: 27epoch:train:13101-13200batch: iter_time=8.035e-05, forward_time=0.289, loss_ctc=46.601, loss_att=38.729, acc=0.763, loss=41.091, backward_time=0.281, grad_norm=36.308, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=1.929e-04, train_time=1.499
+[gpub058:0/16] 2024-02-05 02:11:32,412 (trainer:740) INFO: 27epoch:train:13201-13300batch: iter_time=8.362e-05, forward_time=0.469, loss_ctc=48.557, loss_att=47.289, acc=0.753, loss=47.669, backward_time=0.344, grad_norm=37.195, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.929e-04, train_time=1.356
+[gpub058:0/16] 2024-02-05 02:13:25,219 (trainer:740) INFO: 27epoch:train:13301-13400batch: iter_time=9.006e-05, forward_time=0.299, loss_ctc=38.823, loss_att=41.773, acc=0.754, loss=40.888, backward_time=0.300, grad_norm=35.790, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.929e-04, train_time=1.128
+[gpub058:0/16] 2024-02-05 02:16:17,894 (trainer:740) INFO: 27epoch:train:13401-13500batch: iter_time=7.983e-05, forward_time=0.289, loss_ctc=48.330, loss_att=48.982, acc=0.728, loss=48.786, backward_time=0.281, grad_norm=41.900, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.091, optim0_lr0=1.928e-04, train_time=1.726
+[gpub058:0/16] 2024-02-05 02:18:33,270 (trainer:740) INFO: 27epoch:train:13501-13600batch: iter_time=8.065e-05, forward_time=0.460, loss_ctc=48.611, loss_att=41.946, acc=0.760, loss=43.946, backward_time=0.350, grad_norm=36.325, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.928e-04, train_time=1.354
+[gpub058:0/16] 2024-02-05 02:20:24,458 (trainer:740) INFO: 27epoch:train:13601-13700batch: iter_time=8.822e-05, forward_time=0.302, loss_ctc=47.689, loss_att=49.728, acc=0.745, loss=49.117, backward_time=0.302, grad_norm=36.197, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.928e-04, train_time=1.112
+[gpub058:0/16] 2024-02-05 02:22:00,897 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-05 02:22:19,764 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 02:22:23,591 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd98e5ae890>)
+[gpub058:0/16] 2024-02-05 02:22:23,591 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-05 02:22:23,594 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 02:29:09,451 (trainer:740) INFO: 27epoch:train:13701-13800batch: iter_time=3.680, forward_time=0.389, loss_ctc=48.832, loss_att=51.047, acc=0.725, loss=50.382, backward_time=0.305, grad_norm=42.732, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.928e-04, train_time=5.249
+[gpub058:0/16] 2024-02-05 02:31:02,978 (trainer:740) INFO: 27epoch:train:13801-13900batch: iter_time=8.030e-05, forward_time=0.302, loss_ctc=54.604, loss_att=62.464, acc=0.716, loss=60.106, backward_time=0.293, grad_norm=47.160, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.927e-04, train_time=1.135
+[gpub058:0/16] 2024-02-05 02:34:21,989 (trainer:740) INFO: 27epoch:train:13901-14000batch: iter_time=7.990e-05, forward_time=0.422, loss_ctc=48.734, loss_att=52.008, acc=0.767, loss=51.025, backward_time=0.307, grad_norm=37.923, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.119, optim0_lr0=1.927e-04, train_time=1.990
+[gpub058:0/16] 2024-02-05 02:36:22,907 (trainer:740) INFO: 27epoch:train:14001-14100batch: iter_time=8.502e-05, forward_time=0.318, loss_ctc=43.020, loss_att=41.881, acc=0.773, loss=42.223, backward_time=0.295, grad_norm=34.679, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.927e-04, train_time=1.209
+[gpub058:0/16] 2024-02-05 02:38:58,638 (trainer:740) INFO: 27epoch:train:14101-14200batch: iter_time=8.526e-05, forward_time=0.291, loss_ctc=47.171, loss_att=41.659, acc=0.775, loss=43.312, backward_time=0.285, grad_norm=40.059, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.927e-04, train_time=1.558
+[gpub058:0/16] 2024-02-05 02:41:38,607 (trainer:740) INFO: 27epoch:train:14201-14300batch: iter_time=8.758e-05, forward_time=0.350, loss_ctc=40.630, loss_att=38.276, acc=0.775, loss=38.982, backward_time=0.357, grad_norm=32.236, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.926e-04, train_time=1.599
+[gpub058:0/16] 2024-02-05 02:43:56,568 (trainer:740) INFO: 27epoch:train:14301-14400batch: iter_time=8.391e-05, forward_time=0.289, loss_ctc=43.977, loss_att=35.825, acc=0.782, loss=38.270, backward_time=0.298, grad_norm=31.336, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.926e-04, train_time=1.379
+[gpub058:0/16] 2024-02-05 02:45:42,971 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 02:46:22,556 (trainer:740) INFO: 27epoch:train:14401-14500batch: iter_time=9.055e-05, forward_time=0.294, loss_ctc=48.676, loss_att=47.958, acc=0.748, loss=48.174, backward_time=0.286, grad_norm=38.818, clip=100.000, loss_scale=1.385e+34, optim_step_time=0.091, optim0_lr0=1.926e-04, train_time=1.459
+[gpub058:0/16] 2024-02-05 02:46:41,578 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 02:48:59,488 (trainer:740) INFO: 27epoch:train:14501-14600batch: iter_time=8.555e-05, forward_time=0.369, loss_ctc=45.756, loss_att=45.696, acc=0.762, loss=45.714, backward_time=0.369, grad_norm=39.801, clip=100.000, loss_scale=5.927e+33, optim_step_time=0.097, optim0_lr0=1.926e-04, train_time=1.569
+[gpub058:0/16] 2024-02-05 02:50:57,545 (trainer:740) INFO: 27epoch:train:14601-14700batch: iter_time=8.239e-05, forward_time=0.296, loss_ctc=44.109, loss_att=49.235, acc=0.750, loss=47.697, backward_time=0.287, grad_norm=35.820, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.926e-04, train_time=1.180
+[gpub058:0/16] 2024-02-05 02:53:38,038 (trainer:740) INFO: 27epoch:train:14701-14800batch: iter_time=8.603e-05, forward_time=0.290, loss_ctc=46.175, loss_att=43.075, acc=0.744, loss=44.005, backward_time=0.290, grad_norm=43.400, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.925e-04, train_time=1.605
+[gpub058:0/16] 2024-02-05 02:55:50,663 (trainer:740) INFO: 27epoch:train:14801-14900batch: iter_time=8.611e-05, forward_time=0.387, loss_ctc=47.216, loss_att=43.422, acc=0.758, loss=44.561, backward_time=0.301, grad_norm=38.770, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.925e-04, train_time=1.326
+[gpub058:0/16] 2024-02-05 02:58:05,930 (trainer:740) INFO: 27epoch:train:14901-15000batch: iter_time=5.440e-04, forward_time=0.310, loss_ctc=50.156, loss_att=52.655, acc=0.734, loss=51.905, backward_time=0.302, grad_norm=43.164, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.925e-04, train_time=1.350
+[gpub058:0/16] 2024-02-05 03:36:54,782 (trainer:346) INFO: 27epoch results: [train] iter_time=0.275, forward_time=0.342, loss_ctc=47.375, loss_att=46.224, acc=0.752, loss=46.569, backward_time=0.308, grad_norm=38.151, clip=100.000, loss_scale=6.401e+33, optim_step_time=0.096, optim0_lr0=1.943e-04, train_time=2.226, time=9 hours, 16 minutes and 55.03 seconds, total_count=435000, gpu_max_cached_mem_GB=41.105, [valid] loss_ctc=38.776, cer_ctc=0.195, loss_att=42.847, acc=0.662, cer=0.333, wer=0.997, loss=41.625, time=38 minutes and 24.8 seconds, total_count=135459, gpu_max_cached_mem_GB=41.105
+[gpub058:0/16] 2024-02-05 03:37:16,882 (trainer:394) INFO: The best model has been updated: valid.total_count
+[gpub058:0/16] 2024-02-05 03:37:16,986 (trainer:275) INFO: 28/45epoch started. Estimated time to finish: 1 week, 10 hours and 42 minutes
+[gpub058:0/16] 2024-02-05 03:37:16,998 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-05 03:37:34,627 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 03:37:37,965 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3b4acd780>)
+[gpub058:0/16] 2024-02-05 03:37:37,965 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub058:0/16] 2024-02-05 03:37:37,969 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 03:44:54,039 (trainer:740) INFO: 28epoch:train:1-100batch: iter_time=3.252, forward_time=0.391, loss_ctc=47.164, loss_att=53.798, acc=0.732, loss=51.808, backward_time=0.297, grad_norm=43.256, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.925e-04, train_time=4.570
+[gpub058:0/16] 2024-02-05 03:47:07,409 (trainer:740) INFO: 28epoch:train:101-200batch: iter_time=8.986e-05, forward_time=0.293, loss_ctc=51.828, loss_att=53.255, acc=0.729, loss=52.827, backward_time=0.288, grad_norm=44.816, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.924e-04, train_time=1.333
+[gpub058:0/16] 2024-02-05 03:49:51,260 (trainer:740) INFO: 28epoch:train:201-300batch: iter_time=8.328e-05, forward_time=0.384, loss_ctc=60.384, loss_att=60.863, acc=0.732, loss=60.719, backward_time=0.328, grad_norm=56.044, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.924e-04, train_time=1.639
+[gpub058:0/16] 2024-02-05 03:52:16,205 (trainer:740) INFO: 28epoch:train:301-400batch: iter_time=8.430e-05, forward_time=0.291, loss_ctc=56.676, loss_att=51.297, acc=0.737, loss=52.911, backward_time=0.285, grad_norm=51.799, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.924e-04, train_time=1.449
+[gpub058:0/16] 2024-02-05 03:54:27,121 (trainer:740) INFO: 28epoch:train:401-500batch: iter_time=9.103e-05, forward_time=0.352, loss_ctc=52.934, loss_att=50.603, acc=0.737, loss=51.303, backward_time=0.312, grad_norm=39.038, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.924e-04, train_time=1.309
+[gpub058:0/16] 2024-02-05 03:56:56,645 (trainer:740) INFO: 28epoch:train:501-600batch: iter_time=9.576e-05, forward_time=0.322, loss_ctc=40.904, loss_att=40.145, acc=0.752, loss=40.373, backward_time=0.307, grad_norm=37.841, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.923e-04, train_time=1.495
+[gpub058:0/16] 2024-02-05 03:59:00,386 (trainer:740) INFO: 28epoch:train:601-700batch: iter_time=9.076e-05, forward_time=0.305, loss_ctc=48.447, loss_att=46.923, acc=0.725, loss=47.380, backward_time=0.288, grad_norm=47.289, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.923e-04, train_time=1.237
+[gpub058:0/16] 2024-02-05 04:01:16,159 (trainer:740) INFO: 28epoch:train:701-800batch: iter_time=9.111e-05, forward_time=0.382, loss_ctc=49.188, loss_att=42.753, acc=0.751, loss=44.683, backward_time=0.340, grad_norm=38.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.923e-04, train_time=1.357
+[gpub058:0/16] 2024-02-05 04:03:53,455 (trainer:740) INFO: 28epoch:train:801-900batch: iter_time=8.597e-05, forward_time=0.290, loss_ctc=54.711, loss_att=43.822, acc=0.746, loss=47.089, backward_time=0.284, grad_norm=43.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.923e-04, train_time=1.573
+[gpub058:0/16] 2024-02-05 04:06:02,855 (trainer:740) INFO: 28epoch:train:901-1000batch: iter_time=8.997e-05, forward_time=0.312, loss_ctc=48.201, loss_att=50.356, acc=0.726, loss=49.709, backward_time=0.307, grad_norm=40.562, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.922e-04, train_time=1.294
+[gpub058:0/16] 2024-02-05 04:08:32,222 (trainer:740) INFO: 28epoch:train:1001-1100batch: iter_time=8.517e-05, forward_time=0.389, loss_ctc=48.925, loss_att=50.682, acc=0.727, loss=50.155, backward_time=0.343, grad_norm=38.186, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.922e-04, train_time=1.493
+[gpub058:0/16] 2024-02-05 04:10:36,069 (trainer:740) INFO: 28epoch:train:1101-1200batch: iter_time=8.893e-05, forward_time=0.292, loss_ctc=54.764, loss_att=51.018, acc=0.731, loss=52.142, backward_time=0.287, grad_norm=65.466, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.922e-04, train_time=1.238
+[gpub058:0/16] 2024-02-05 04:12:08,475 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub058:0/16] 2024-02-05 04:12:27,189 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 04:12:30,724 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fd9e074dd50>)
+[gpub058:0/16] 2024-02-05 04:12:30,724 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub058:0/16] 2024-02-05 04:12:30,729 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 04:19:07,063 (trainer:740) INFO: 28epoch:train:1201-1300batch: iter_time=3.656, forward_time=0.392, loss_ctc=46.142, loss_att=49.986, acc=0.743, loss=48.832, backward_time=0.314, grad_norm=36.486, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.922e-04, train_time=5.110
+[gpub058:0/16] 2024-02-05 04:20:56,746 (trainer:740) INFO: 28epoch:train:1301-1400batch: iter_time=8.567e-05, forward_time=0.293, loss_ctc=49.340, loss_att=55.265, acc=0.738, loss=53.488, backward_time=0.289, grad_norm=43.032, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.921e-04, train_time=1.097
+[gpub058:0/16] 2024-02-05 04:23:14,699 (trainer:740) INFO: 28epoch:train:1401-1500batch: iter_time=8.779e-05, forward_time=0.297, loss_ctc=59.807, loss_att=59.774, acc=0.738, loss=59.784, backward_time=0.300, grad_norm=57.817, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.921e-04, train_time=1.379
+[gpub058:0/16] 2024-02-05 04:25:51,093 (trainer:740) INFO: 28epoch:train:1501-1600batch: iter_time=9.659e-05, forward_time=0.346, loss_ctc=51.418, loss_att=48.096, acc=0.757, loss=49.093, backward_time=0.347, grad_norm=45.055, clip=100.000, loss_scale=9.606e+33, optim_step_time=0.097, optim0_lr0=1.921e-04, train_time=1.563
+[gpub058:0/16] 2024-02-05 04:28:07,752 (trainer:740) INFO: 28epoch:train:1601-1700batch: iter_time=9.652e-05, forward_time=0.338, loss_ctc=52.249, loss_att=50.549, acc=0.754, loss=51.059, backward_time=0.290, grad_norm=41.293, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.921e-04, train_time=1.367
+[gpub058:0/16] 2024-02-05 04:30:03,870 (trainer:740) INFO: 28epoch:train:1701-1800batch: iter_time=9.362e-05, forward_time=0.311, loss_ctc=48.808, loss_att=52.450, acc=0.736, loss=51.357, backward_time=0.304, grad_norm=39.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.921e-04, train_time=1.161
+[gpub058:0/16] 2024-02-05 04:32:48,535 (trainer:740) INFO: 28epoch:train:1801-1900batch: iter_time=9.390e-05, forward_time=0.334, loss_ctc=46.185, loss_att=42.859, acc=0.756, loss=43.857, backward_time=0.338, grad_norm=45.651, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.920e-04, train_time=1.646
+[gpub058:0/16] 2024-02-05 04:34:45,257 (trainer:740) INFO: 28epoch:train:1901-2000batch: iter_time=9.427e-05, forward_time=0.288, loss_ctc=43.458, loss_att=37.455, acc=0.757, loss=39.256, backward_time=0.284, grad_norm=36.243, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.920e-04, train_time=1.167
+[gpub058:0/16] 2024-02-05 04:37:14,859 (trainer:740) INFO: 28epoch:train:2001-2100batch: iter_time=9.007e-05, forward_time=0.307, loss_ctc=46.147, loss_att=41.202, acc=0.756, loss=42.685, backward_time=0.288, grad_norm=37.495, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.920e-04, train_time=1.496
+[gpub058:0/16] 2024-02-05 04:39:50,161 (trainer:740) INFO: 28epoch:train:2101-2200batch: iter_time=8.856e-05, forward_time=0.406, loss_ctc=52.034, loss_att=46.519, acc=0.745, loss=48.174, backward_time=0.307, grad_norm=43.221, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.920e-04, train_time=1.553
+[gpub058:0/16] 2024-02-05 04:42:06,305 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 04:42:15,221 (trainer:740) INFO: 28epoch:train:2201-2300batch: iter_time=8.974e-05, forward_time=0.291, loss_ctc=48.591, loss_att=51.806, acc=0.745, loss=50.841, backward_time=0.286, grad_norm=38.122, clip=100.000, loss_scale=9.965e+33, optim_step_time=0.092, optim0_lr0=1.919e-04, train_time=1.450
+[gpub058:0/16] 2024-02-05 04:44:14,914 (trainer:740) INFO: 28epoch:train:2301-2400batch: iter_time=8.515e-05, forward_time=0.305, loss_ctc=53.417, loss_att=54.251, acc=0.725, loss=54.001, backward_time=0.304, grad_norm=40.608, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.919e-04, train_time=1.197
+[gpub058:0/16] 2024-02-05 04:46:29,200 (trainer:740) INFO: 28epoch:train:2401-2500batch: iter_time=8.144e-05, forward_time=0.297, loss_ctc=49.999, loss_att=50.562, acc=0.754, loss=50.393, backward_time=0.286, grad_norm=38.747, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.919e-04, train_time=1.343
+[gpub058:0/16] 2024-02-05 04:46:49,228 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub058:0/16] 2024-02-05 04:47:08,047 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 04:47:11,526 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3ab03dd50>)
+[gpub058:0/16] 2024-02-05 04:47:11,526 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub058:0/16] 2024-02-05 04:47:11,549 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 04:54:20,670 (trainer:740) INFO: 28epoch:train:2501-2600batch: iter_time=3.490, forward_time=0.388, loss_ctc=45.988, loss_att=52.563, acc=0.739, loss=50.591, backward_time=0.296, grad_norm=41.545, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.919e-04, train_time=4.714
+[gpub058:0/16] 2024-02-05 04:56:40,198 (trainer:740) INFO: 28epoch:train:2601-2700batch: iter_time=8.131e-05, forward_time=0.310, loss_ctc=49.830, loss_att=50.833, acc=0.736, loss=50.532, backward_time=0.286, grad_norm=44.111, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.918e-04, train_time=1.395
+[gpub058:0/16] 2024-02-05 04:59:15,855 (trainer:740) INFO: 28epoch:train:2701-2800batch: iter_time=8.532e-05, forward_time=0.369, loss_ctc=56.456, loss_att=58.841, acc=0.736, loss=58.126, backward_time=0.353, grad_norm=52.460, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.918e-04, train_time=1.557
+[gpub058:0/16] 2024-02-05 05:01:40,556 (trainer:740) INFO: 28epoch:train:2801-2900batch: iter_time=8.155e-05, forward_time=0.293, loss_ctc=53.205, loss_att=49.202, acc=0.743, loss=50.403, backward_time=0.293, grad_norm=48.258, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.918e-04, train_time=1.447
+[gpub058:0/16] 2024-02-05 05:03:42,893 (trainer:740) INFO: 28epoch:train:2901-3000batch: iter_time=8.315e-05, forward_time=0.292, loss_ctc=51.925, loss_att=48.940, acc=0.743, loss=49.835, backward_time=0.289, grad_norm=38.997, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.918e-04, train_time=1.223
+[gpub058:0/16] 2024-02-05 05:06:02,847 (trainer:740) INFO: 28epoch:train:3001-3100batch: iter_time=8.096e-05, forward_time=0.315, loss_ctc=40.035, loss_att=39.314, acc=0.757, loss=39.530, backward_time=0.283, grad_norm=37.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.917e-04, train_time=1.399
+[gpub058:0/16] 2024-02-05 05:08:48,366 (trainer:740) INFO: 28epoch:train:3101-3200batch: iter_time=9.025e-05, forward_time=0.400, loss_ctc=47.482, loss_att=46.486, acc=0.728, loss=46.785, backward_time=0.311, grad_norm=43.571, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.917e-04, train_time=1.655
+[gpub058:0/16] 2024-02-05 05:10:38,288 (trainer:740) INFO: 28epoch:train:3201-3300batch: iter_time=9.027e-05, forward_time=0.291, loss_ctc=47.972, loss_att=42.118, acc=0.753, loss=43.874, backward_time=0.284, grad_norm=37.896, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.917e-04, train_time=1.099
+[gpub058:0/16] 2024-02-05 05:12:59,374 (trainer:740) INFO: 28epoch:train:3301-3400batch: iter_time=8.831e-05, forward_time=0.290, loss_ctc=52.295, loss_att=43.193, acc=0.748, loss=45.923, backward_time=0.286, grad_norm=40.527, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.917e-04, train_time=1.411
+[gpub058:0/16] 2024-02-05 05:15:13,805 (trainer:740) INFO: 28epoch:train:3401-3500batch: iter_time=8.399e-05, forward_time=0.295, loss_ctc=46.888, loss_att=50.126, acc=0.731, loss=49.155, backward_time=0.294, grad_norm=38.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.917e-04, train_time=1.343
+[gpub058:0/16] 2024-02-05 05:17:30,461 (trainer:740) INFO: 28epoch:train:3501-3600batch: iter_time=8.445e-05, forward_time=0.357, loss_ctc=47.716, loss_att=49.715, acc=0.731, loss=49.115, backward_time=0.314, grad_norm=37.524, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.916e-04, train_time=1.367
+[gpub058:0/16] 2024-02-05 05:20:01,200 (trainer:740) INFO: 28epoch:train:3601-3700batch: iter_time=8.379e-05, forward_time=0.292, loss_ctc=54.829, loss_att=50.961, acc=0.732, loss=52.121, backward_time=0.286, grad_norm=39.795, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.916e-04, train_time=1.507
+[gpub058:0/16] 2024-02-05 05:21:15,623 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub058:0/16] 2024-02-05 05:21:34,904 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 05:21:38,818 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe15f92fdf0>)
+[gpub058:0/16] 2024-02-05 05:21:38,819 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-05 05:21:38,822 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 05:27:38,488 (trainer:740) INFO: 28epoch:train:3701-3800batch: iter_time=3.447, forward_time=0.298, loss_ctc=45.511, loss_att=49.346, acc=0.743, loss=48.195, backward_time=0.286, grad_norm=36.823, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.916e-04, train_time=4.572
+[gpub058:0/16] 2024-02-05 05:30:13,138 (trainer:740) INFO: 28epoch:train:3801-3900batch: iter_time=8.454e-05, forward_time=0.341, loss_ctc=48.676, loss_att=53.234, acc=0.731, loss=51.867, backward_time=0.313, grad_norm=43.321, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.916e-04, train_time=1.547
+[gpub058:0/16] 2024-02-05 05:32:18,857 (trainer:740) INFO: 28epoch:train:3901-4000batch: iter_time=8.267e-05, forward_time=0.290, loss_ctc=58.231, loss_att=58.463, acc=0.738, loss=58.394, backward_time=0.286, grad_norm=58.213, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.915e-04, train_time=1.257
+[gpub058:0/16] 2024-02-05 05:34:21,425 (trainer:740) INFO: 28epoch:train:4001-4100batch: iter_time=8.473e-05, forward_time=0.301, loss_ctc=49.926, loss_att=46.608, acc=0.751, loss=47.603, backward_time=0.292, grad_norm=44.349, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.915e-04, train_time=1.225
+[gpub058:0/16] 2024-02-05 05:37:23,687 (trainer:740) INFO: 28epoch:train:4101-4200batch: iter_time=8.532e-05, forward_time=0.343, loss_ctc=51.708, loss_att=50.265, acc=0.746, loss=50.698, backward_time=0.368, grad_norm=42.448, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.915e-04, train_time=1.823
+[gpub058:0/16] 2024-02-05 05:39:16,734 (trainer:740) INFO: 28epoch:train:4201-4300batch: iter_time=8.519e-05, forward_time=0.290, loss_ctc=48.300, loss_att=50.500, acc=0.732, loss=49.840, backward_time=0.284, grad_norm=40.086, clip=100.000, loss_scale=5.608e+33, optim_step_time=0.093, optim0_lr0=1.915e-04, train_time=1.130
+[gpub058:0/16] 2024-02-05 05:41:00,973 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 05:41:25,238 (trainer:740) INFO: 28epoch:train:4301-4400batch: iter_time=8.409e-05, forward_time=0.307, loss_ctc=45.472, loss_att=43.025, acc=0.744, loss=43.760, backward_time=0.288, grad_norm=43.977, clip=100.000, loss_scale=9.283e+33, optim_step_time=0.094, optim0_lr0=1.914e-04, train_time=1.285
+[gpub058:0/16] 2024-02-05 05:43:56,411 (trainer:740) INFO: 28epoch:train:4401-4500batch: iter_time=8.743e-05, forward_time=0.375, loss_ctc=42.943, loss_att=36.808, acc=0.760, loss=38.648, backward_time=0.305, grad_norm=37.688, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.914e-04, train_time=1.512
+[gpub058:0/16] 2024-02-05 05:46:06,975 (trainer:740) INFO: 28epoch:train:4501-4600batch: iter_time=8.468e-05, forward_time=0.289, loss_ctc=45.627, loss_att=40.642, acc=0.752, loss=42.138, backward_time=0.287, grad_norm=36.878, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.914e-04, train_time=1.305
+[gpub058:0/16] 2024-02-05 05:48:26,014 (trainer:740) INFO: 28epoch:train:4601-4700batch: iter_time=6.109e-04, forward_time=0.302, loss_ctc=51.648, loss_att=45.380, acc=0.740, loss=47.260, backward_time=0.308, grad_norm=44.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.914e-04, train_time=1.390
+[gpub058:0/16] 2024-02-05 05:50:49,638 (trainer:740) INFO: 28epoch:train:4701-4800batch: iter_time=8.949e-05, forward_time=0.340, loss_ctc=47.812, loss_att=49.786, acc=0.744, loss=49.194, backward_time=0.338, grad_norm=36.191, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.914e-04, train_time=1.436
+[gpub058:0/16] 2024-02-05 05:53:10,555 (trainer:740) INFO: 28epoch:train:4801-4900batch: iter_time=8.406e-05, forward_time=0.330, loss_ctc=53.120, loss_att=52.803, acc=0.717, loss=52.898, backward_time=0.289, grad_norm=42.554, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.913e-04, train_time=1.409
+[gpub058:0/16] 2024-02-05 05:55:26,668 (trainer:740) INFO: 28epoch:train:4901-5000batch: iter_time=8.571e-05, forward_time=0.290, loss_ctc=49.974, loss_att=50.219, acc=0.750, loss=50.145, backward_time=0.286, grad_norm=39.000, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.913e-04, train_time=1.361
+[gpub058:0/16] 2024-02-05 05:55:46,697 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub058:0/16] 2024-02-05 05:56:05,609 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 05:56:09,161 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe160534340>)
+[gpub058:0/16] 2024-02-05 05:56:09,161 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub058:0/16] 2024-02-05 05:56:09,166 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 06:03:16,211 (trainer:740) INFO: 28epoch:train:5001-5100batch: iter_time=3.258, forward_time=0.390, loss_ctc=45.446, loss_att=51.677, acc=0.743, loss=49.808, backward_time=0.310, grad_norm=37.981, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.913e-04, train_time=4.695
+[gpub058:0/16] 2024-02-05 06:03:32,786 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 06:05:45,543 (trainer:740) INFO: 28epoch:train:5101-5200batch: iter_time=8.543e-05, forward_time=0.307, loss_ctc=49.566, loss_att=51.050, acc=0.737, loss=50.604, backward_time=0.286, grad_norm=42.905, clip=100.000, loss_scale=2.885e+33, optim_step_time=0.093, optim0_lr0=1.913e-04, train_time=1.494
+[gpub058:0/16] 2024-02-05 06:07:55,639 (trainer:740) INFO: 28epoch:train:5201-5300batch: iter_time=9.053e-05, forward_time=0.292, loss_ctc=54.753, loss_att=57.315, acc=0.739, loss=56.546, backward_time=0.289, grad_norm=52.802, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.912e-04, train_time=1.301
+[gpub058:0/16] 2024-02-05 06:10:21,435 (trainer:740) INFO: 28epoch:train:5301-5400batch: iter_time=8.988e-05, forward_time=0.388, loss_ctc=52.811, loss_att=48.964, acc=0.747, loss=50.118, backward_time=0.323, grad_norm=50.837, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.912e-04, train_time=1.458
+[gpub058:0/16] 2024-02-05 06:12:28,216 (trainer:740) INFO: 28epoch:train:5401-5500batch: iter_time=8.541e-05, forward_time=0.300, loss_ctc=50.971, loss_att=48.727, acc=0.745, loss=49.400, backward_time=0.291, grad_norm=39.234, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.912e-04, train_time=1.267
+[gpub058:0/16] 2024-02-05 06:14:35,221 (trainer:740) INFO: 28epoch:train:5501-5600batch: iter_time=9.116e-05, forward_time=0.288, loss_ctc=39.890, loss_att=38.373, acc=0.761, loss=38.828, backward_time=0.282, grad_norm=36.788, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.912e-04, train_time=1.271
+[gpub058:0/16] 2024-02-05 06:17:17,801 (trainer:740) INFO: 28epoch:train:5601-5700batch: iter_time=9.191e-05, forward_time=0.356, loss_ctc=46.714, loss_att=45.533, acc=0.731, loss=45.887, backward_time=0.340, grad_norm=43.297, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.911e-04, train_time=1.626
+[gpub058:0/16] 2024-02-05 06:19:33,297 (trainer:740) INFO: 28epoch:train:5701-5800batch: iter_time=9.497e-05, forward_time=0.299, loss_ctc=47.625, loss_att=41.302, acc=0.756, loss=43.199, backward_time=0.285, grad_norm=37.278, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.911e-04, train_time=1.355
+[gpub058:0/16] 2024-02-05 06:21:31,488 (trainer:740) INFO: 28epoch:train:5801-5900batch: iter_time=9.603e-05, forward_time=0.300, loss_ctc=52.013, loss_att=42.563, acc=0.754, loss=45.398, backward_time=0.292, grad_norm=41.874, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.911e-04, train_time=1.181
+[gpub058:0/16] 2024-02-05 06:24:11,817 (trainer:740) INFO: 28epoch:train:5901-6000batch: iter_time=8.729e-05, forward_time=0.292, loss_ctc=46.652, loss_att=49.125, acc=0.732, loss=48.383, backward_time=0.284, grad_norm=37.831, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.911e-04, train_time=1.604
+[gpub058:0/16] 2024-02-05 06:26:33,912 (trainer:740) INFO: 28epoch:train:6001-6100batch: iter_time=9.435e-05, forward_time=0.430, loss_ctc=47.466, loss_att=49.477, acc=0.734, loss=48.874, backward_time=0.342, grad_norm=39.430, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.910e-04, train_time=1.421
+[gpub058:0/16] 2024-02-05 06:29:11,659 (trainer:740) INFO: 28epoch:train:6101-6200batch: iter_time=9.351e-05, forward_time=0.300, loss_ctc=53.904, loss_att=50.497, acc=0.735, loss=51.519, backward_time=0.294, grad_norm=39.475, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.910e-04, train_time=1.577
+[gpub058:0/16] 2024-02-05 06:30:26,469 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub058:0/16] 2024-02-05 06:30:45,572 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 06:30:49,145 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1b20de6e0>)
+[gpub058:0/16] 2024-02-05 06:30:49,145 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub058:0/16] 2024-02-05 06:30:49,148 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 06:37:07,188 (trainer:740) INFO: 28epoch:train:6201-6300batch: iter_time=3.537, forward_time=0.382, loss_ctc=44.676, loss_att=48.449, acc=0.746, loss=47.317, backward_time=0.303, grad_norm=35.784, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.910e-04, train_time=4.755
+[gpub058:0/16] 2024-02-05 06:39:48,581 (trainer:740) INFO: 28epoch:train:6301-6400batch: iter_time=8.426e-05, forward_time=0.290, loss_ctc=48.180, loss_att=52.861, acc=0.732, loss=51.457, backward_time=0.285, grad_norm=42.074, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.910e-04, train_time=1.614
+[gpub058:0/16] 2024-02-05 06:41:56,163 (trainer:740) INFO: 28epoch:train:6401-6500batch: iter_time=8.094e-05, forward_time=0.309, loss_ctc=56.192, loss_att=58.194, acc=0.739, loss=57.594, backward_time=0.297, grad_norm=59.012, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.910e-04, train_time=1.275
+[gpub058:0/16] 2024-02-05 06:44:43,999 (trainer:740) INFO: 28epoch:train:6501-6600batch: iter_time=9.073e-05, forward_time=0.348, loss_ctc=49.900, loss_att=45.917, acc=0.754, loss=47.112, backward_time=0.415, grad_norm=42.993, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.106, optim0_lr0=1.909e-04, train_time=1.678
+[gpub058:0/16] 2024-02-05 06:46:45,237 (trainer:740) INFO: 28epoch:train:6601-6700batch: iter_time=8.862e-05, forward_time=0.292, loss_ctc=51.305, loss_att=50.046, acc=0.750, loss=50.424, backward_time=0.287, grad_norm=40.067, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.909e-04, train_time=1.212
+[gpub058:0/16] 2024-02-05 06:49:03,678 (trainer:740) INFO: 28epoch:train:6701-6800batch: iter_time=8.921e-05, forward_time=0.310, loss_ctc=47.321, loss_att=49.619, acc=0.736, loss=48.929, backward_time=0.286, grad_norm=39.235, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.092, optim0_lr0=1.909e-04, train_time=1.384
+[gpub058:0/16] 2024-02-05 06:51:14,645 (trainer:740) INFO: 28epoch:train:6801-6900batch: iter_time=8.957e-05, forward_time=0.318, loss_ctc=45.069, loss_att=42.913, acc=0.745, loss=43.560, backward_time=0.346, grad_norm=40.608, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.909e-04, train_time=1.309
+[gpub058:0/16] 2024-02-05 06:53:42,705 (trainer:740) INFO: 28epoch:train:6901-7000batch: iter_time=8.541e-05, forward_time=0.306, loss_ctc=42.641, loss_att=36.931, acc=0.759, loss=38.644, backward_time=0.292, grad_norm=37.864, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.908e-04, train_time=1.480
+[gpub058:0/16] 2024-02-05 06:55:44,531 (trainer:740) INFO: 28epoch:train:7001-7100batch: iter_time=8.100e-05, forward_time=0.289, loss_ctc=45.355, loss_att=40.469, acc=0.754, loss=41.934, backward_time=0.284, grad_norm=36.625, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.908e-04, train_time=1.218
+[gpub058:0/16] 2024-02-05 06:57:47,597 (trainer:740) INFO: 28epoch:train:7101-7200batch: iter_time=8.675e-05, forward_time=0.299, loss_ctc=50.998, loss_att=45.037, acc=0.742, loss=46.825, backward_time=0.292, grad_norm=46.151, clip=100.000, loss_scale=4.881e+33, optim_step_time=0.093, optim0_lr0=1.908e-04, train_time=1.230
+[gpub058:0/16] 2024-02-05 07:00:43,539 (trainer:740) INFO: 28epoch:train:7201-7300batch: iter_time=8.145e-05, forward_time=0.419, loss_ctc=47.913, loss_att=50.057, acc=0.744, loss=49.414, backward_time=0.304, grad_norm=36.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.908e-04, train_time=1.759
+[gpub058:0/16] 2024-02-05 07:02:40,670 (trainer:740) INFO: 28epoch:train:7301-7400batch: iter_time=8.249e-05, forward_time=0.291, loss_ctc=52.539, loss_att=51.887, acc=0.719, loss=52.083, backward_time=0.285, grad_norm=41.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.907e-04, train_time=1.171
+[gpub058:0/16] 2024-02-05 07:04:57,416 (trainer:740) INFO: 28epoch:train:7401-7500batch: iter_time=8.193e-05, forward_time=0.304, loss_ctc=49.739, loss_att=49.983, acc=0.751, loss=49.910, backward_time=0.288, grad_norm=37.622, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.907e-04, train_time=1.367
+[gpub058:0/16] 2024-02-05 07:05:17,445 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub058:0/16] 2024-02-05 07:05:36,577 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 07:05:40,209 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe18af87ca0>)
+[gpub058:0/16] 2024-02-05 07:05:40,210 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-05 07:05:40,213 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 07:13:11,531 (trainer:740) INFO: 28epoch:train:7501-7600batch: iter_time=3.467, forward_time=0.361, loss_ctc=44.744, loss_att=50.614, acc=0.745, loss=48.853, backward_time=0.297, grad_norm=37.346, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.907e-04, train_time=4.941
+[gpub058:0/16] 2024-02-05 07:15:44,758 (trainer:740) INFO: 28epoch:train:7601-7700batch: iter_time=8.157e-05, forward_time=0.289, loss_ctc=49.116, loss_att=50.184, acc=0.739, loss=49.863, backward_time=0.284, grad_norm=41.809, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.907e-04, train_time=1.532
+[gpub058:0/16] 2024-02-05 07:19:25,512 (trainer:740) INFO: 28epoch:train:7701-7800batch: iter_time=2.591e-04, forward_time=0.366, loss_ctc=54.096, loss_att=57.609, acc=0.741, loss=56.555, backward_time=0.407, grad_norm=52.450, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.907e-04, train_time=2.207
+[gpub058:0/16] 2024-02-05 07:22:11,246 (trainer:740) INFO: 28epoch:train:7801-7900batch: iter_time=7.880e-05, forward_time=0.291, loss_ctc=52.212, loss_att=48.234, acc=0.747, loss=49.427, backward_time=0.283, grad_norm=45.086, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.906e-04, train_time=1.657
+[gpub058:0/16] 2024-02-05 07:24:50,809 (trainer:740) INFO: 28epoch:train:7901-8000batch: iter_time=8.014e-05, forward_time=0.290, loss_ctc=50.654, loss_att=48.203, acc=0.747, loss=48.939, backward_time=0.285, grad_norm=41.554, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.906e-04, train_time=1.595
+[gpub058:0/16] 2024-02-05 07:27:35,334 (trainer:740) INFO: 28epoch:train:8001-8100batch: iter_time=8.073e-05, forward_time=0.286, loss_ctc=39.493, loss_att=38.597, acc=0.761, loss=38.866, backward_time=0.279, grad_norm=35.655, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.906e-04, train_time=1.645
+[gpub058:0/16] 2024-02-05 07:30:43,260 (trainer:740) INFO: 28epoch:train:8101-8200batch: iter_time=8.696e-05, forward_time=0.337, loss_ctc=46.283, loss_att=45.115, acc=0.732, loss=45.466, backward_time=0.372, grad_norm=44.044, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.906e-04, train_time=1.879
+[gpub058:0/16] 2024-02-05 07:33:20,122 (trainer:740) INFO: 28epoch:train:8201-8300batch: iter_time=9.278e-05, forward_time=0.291, loss_ctc=47.412, loss_att=41.420, acc=0.756, loss=43.217, backward_time=0.284, grad_norm=40.078, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.905e-04, train_time=1.569
+[gpub058:0/16] 2024-02-05 07:36:10,284 (trainer:740) INFO: 28epoch:train:8301-8400batch: iter_time=9.274e-05, forward_time=0.288, loss_ctc=51.432, loss_att=42.466, acc=0.753, loss=45.156, backward_time=0.283, grad_norm=40.208, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.905e-04, train_time=1.701
+[gpub058:0/16] 2024-02-05 07:38:43,511 (trainer:740) INFO: 28epoch:train:8401-8500batch: iter_time=8.881e-05, forward_time=0.348, loss_ctc=46.819, loss_att=49.252, acc=0.734, loss=48.522, backward_time=0.357, grad_norm=40.789, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.905e-04, train_time=1.532
+[gpub058:0/16] 2024-02-05 07:41:22,711 (trainer:740) INFO: 28epoch:train:8501-8600batch: iter_time=9.447e-05, forward_time=0.289, loss_ctc=47.061, loss_att=49.249, acc=0.736, loss=48.592, backward_time=0.284, grad_norm=37.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.905e-04, train_time=1.591
+[gpub058:0/16] 2024-02-05 07:43:56,527 (trainer:740) INFO: 28epoch:train:8601-8700batch: iter_time=8.552e-05, forward_time=0.292, loss_ctc=53.502, loss_att=50.019, acc=0.737, loss=51.064, backward_time=0.286, grad_norm=39.808, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.904e-04, train_time=1.539
+[gpub058:0/16] 2024-02-05 07:45:19,008 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub058:0/16] 2024-02-05 07:45:37,986 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 07:45:41,578 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe5b6b778b0>)
+[gpub058:0/16] 2024-02-05 07:45:41,578 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub058:0/16] 2024-02-05 07:45:41,581 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 07:52:16,166 (trainer:740) INFO: 28epoch:train:8701-8800batch: iter_time=3.492, forward_time=0.445, loss_ctc=44.445, loss_att=49.254, acc=0.749, loss=47.811, backward_time=0.318, grad_norm=36.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.904e-04, train_time=4.996
+[gpub058:0/16] 2024-02-05 07:54:27,442 (trainer:740) INFO: 28epoch:train:8801-8900batch: iter_time=8.767e-05, forward_time=0.293, loss_ctc=47.960, loss_att=55.367, acc=0.741, loss=53.145, backward_time=0.286, grad_norm=42.213, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.904e-04, train_time=1.312
+[gpub058:0/16] 2024-02-05 07:56:34,899 (trainer:740) INFO: 28epoch:train:8901-9000batch: iter_time=8.490e-05, forward_time=0.294, loss_ctc=55.230, loss_att=58.787, acc=0.745, loss=57.720, backward_time=0.286, grad_norm=56.142, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.904e-04, train_time=1.275
+[gpub058:0/16] 2024-02-05 07:59:47,429 (trainer:740) INFO: 28epoch:train:9001-9100batch: iter_time=9.611e-05, forward_time=0.360, loss_ctc=49.405, loss_att=47.431, acc=0.762, loss=48.023, backward_time=0.424, grad_norm=42.186, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.904e-04, train_time=1.925
+[gpub058:0/16] 2024-02-05 08:02:02,668 (trainer:740) INFO: 28epoch:train:9101-9200batch: iter_time=9.346e-05, forward_time=0.292, loss_ctc=50.639, loss_att=50.580, acc=0.758, loss=50.597, backward_time=0.288, grad_norm=41.308, clip=100.000, loss_scale=9.762e+33, optim_step_time=0.093, optim0_lr0=1.903e-04, train_time=1.352
+[gpub058:0/16] 2024-02-05 08:04:20,205 (trainer:740) INFO: 28epoch:train:9201-9300batch: iter_time=8.899e-05, forward_time=0.290, loss_ctc=47.575, loss_att=51.500, acc=0.745, loss=50.323, backward_time=0.286, grad_norm=40.521, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.903e-04, train_time=1.375
+[gpub058:0/16] 2024-02-05 08:07:10,631 (trainer:740) INFO: 28epoch:train:9301-9400batch: iter_time=8.870e-05, forward_time=0.368, loss_ctc=45.053, loss_att=43.052, acc=0.758, loss=43.652, backward_time=0.389, grad_norm=42.331, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.903e-04, train_time=1.704
+[gpub058:0/16] 2024-02-05 08:09:27,104 (trainer:740) INFO: 28epoch:train:9401-9500batch: iter_time=8.732e-05, forward_time=0.288, loss_ctc=42.349, loss_att=37.160, acc=0.763, loss=38.717, backward_time=0.282, grad_norm=35.678, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.903e-04, train_time=1.365
+[gpub058:0/16] 2024-02-05 08:11:20,055 (trainer:740) INFO: 28epoch:train:9501-9600batch: iter_time=8.747e-05, forward_time=0.290, loss_ctc=45.117, loss_att=40.870, acc=0.759, loss=42.144, backward_time=0.285, grad_norm=36.530, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.902e-04, train_time=1.129
+[gpub058:0/16] 2024-02-05 08:13:42,429 (trainer:740) INFO: 28epoch:train:9601-9700batch: iter_time=8.279e-05, forward_time=0.344, loss_ctc=50.782, loss_att=46.573, acc=0.749, loss=47.836, backward_time=0.306, grad_norm=46.250, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.902e-04, train_time=1.423
+[gpub058:0/16] 2024-02-05 08:16:32,331 (trainer:740) INFO: 28epoch:train:9701-9800batch: iter_time=8.805e-05, forward_time=0.352, loss_ctc=47.468, loss_att=51.631, acc=0.749, loss=50.382, backward_time=0.323, grad_norm=36.442, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.902e-04, train_time=1.699
+[gpub058:0/16] 2024-02-05 08:18:50,802 (trainer:740) INFO: 28epoch:train:9801-9900batch: iter_time=8.796e-05, forward_time=0.292, loss_ctc=52.748, loss_att=55.006, acc=0.723, loss=54.329, backward_time=0.288, grad_norm=41.233, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.092, optim0_lr0=1.902e-04, train_time=1.384
+[gpub058:0/16] 2024-02-05 08:21:05,935 (trainer:740) INFO: 28epoch:train:9901-10000batch: iter_time=8.446e-05, forward_time=0.291, loss_ctc=49.165, loss_att=49.622, acc=0.759, loss=49.485, backward_time=0.286, grad_norm=35.866, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.901e-04, train_time=1.351
+[gpub058:0/16] 2024-02-05 08:21:25,959 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub058:0/16] 2024-02-05 08:21:45,221 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 08:21:48,947 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe5b8f6fe20>)
+[gpub058:0/16] 2024-02-05 08:21:48,947 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub058:0/16] 2024-02-05 08:21:48,951 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 08:27:38,359 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 08:29:21,898 (trainer:740) INFO: 28epoch:train:10001-10100batch: iter_time=3.558, forward_time=0.360, loss_ctc=44.849, loss_att=50.778, acc=0.753, loss=49.000, backward_time=0.298, grad_norm=37.827, clip=100.000, loss_scale=6.451e+33, optim_step_time=0.095, optim0_lr0=1.901e-04, train_time=4.956
+[gpub058:0/16] 2024-02-05 08:31:42,820 (trainer:740) INFO: 28epoch:train:10101-10200batch: iter_time=0.091, forward_time=0.291, loss_ctc=48.872, loss_att=51.216, acc=0.745, loss=50.512, backward_time=0.286, grad_norm=42.840, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.901e-04, train_time=1.413
+[gpub058:0/16] 2024-02-05 08:36:05,294 (trainer:740) INFO: 28epoch:train:10201-10300batch: iter_time=8.443e-05, forward_time=0.353, loss_ctc=53.102, loss_att=57.181, acc=0.755, loss=55.957, backward_time=0.314, grad_norm=53.233, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.901e-04, train_time=2.625
+[gpub058:0/16] 2024-02-05 08:39:53,439 (trainer:740) INFO: 28epoch:train:10301-10400batch: iter_time=8.631e-05, forward_time=0.369, loss_ctc=52.351, loss_att=48.703, acc=0.757, loss=49.798, backward_time=0.300, grad_norm=47.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.901e-04, train_time=2.278
+[gpub058:0/16] 2024-02-05 08:42:41,683 (trainer:740) INFO: 28epoch:train:10401-10500batch: iter_time=8.608e-05, forward_time=0.293, loss_ctc=50.715, loss_att=50.361, acc=0.752, loss=50.467, backward_time=0.286, grad_norm=38.194, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.900e-04, train_time=1.686
+[gpub058:0/16] 2024-02-05 08:45:00,715 (trainer:740) INFO: 28epoch:train:10501-10600batch: iter_time=9.029e-05, forward_time=0.301, loss_ctc=39.531, loss_att=37.727, acc=0.774, loss=38.268, backward_time=0.290, grad_norm=35.272, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.900e-04, train_time=1.390
+[gpub058:0/16] 2024-02-05 08:48:45,112 (trainer:740) INFO: 28epoch:train:10601-10700batch: iter_time=0.300, forward_time=0.488, loss_ctc=46.156, loss_att=45.468, acc=0.743, loss=45.674, backward_time=0.333, grad_norm=42.037, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.118, optim0_lr0=1.900e-04, train_time=2.244
+[gpub058:0/16] 2024-02-05 08:51:12,615 (trainer:740) INFO: 28epoch:train:10701-10800batch: iter_time=8.762e-05, forward_time=0.293, loss_ctc=46.831, loss_att=41.301, acc=0.762, loss=42.960, backward_time=0.283, grad_norm=35.930, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.900e-04, train_time=1.475
+[gpub058:0/16] 2024-02-05 08:53:36,559 (trainer:740) INFO: 28epoch:train:10801-10900batch: iter_time=9.473e-05, forward_time=0.302, loss_ctc=50.998, loss_att=42.770, acc=0.758, loss=45.238, backward_time=0.304, grad_norm=42.087, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.899e-04, train_time=1.439
+[gpub058:0/16] 2024-02-05 08:56:11,695 (trainer:740) INFO: 28epoch:train:10901-11000batch: iter_time=8.786e-05, forward_time=0.370, loss_ctc=46.500, loss_att=50.408, acc=0.741, loss=49.236, backward_time=0.347, grad_norm=38.234, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.899e-04, train_time=1.551
+[gpub058:0/16] 2024-02-05 08:58:37,011 (trainer:740) INFO: 28epoch:train:11001-11100batch: iter_time=9.047e-05, forward_time=0.292, loss_ctc=46.479, loss_att=50.388, acc=0.741, loss=49.215, backward_time=0.287, grad_norm=37.639, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.092, optim0_lr0=1.899e-04, train_time=1.453
+[gpub058:0/16] 2024-02-05 09:01:05,036 (trainer:740) INFO: 28epoch:train:11101-11200batch: iter_time=2.680e-04, forward_time=0.424, loss_ctc=53.257, loss_att=52.092, acc=0.740, loss=52.441, backward_time=0.320, grad_norm=40.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.118, optim0_lr0=1.899e-04, train_time=1.479
+[gpub058:0/16] 2024-02-05 09:02:43,006 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub058:0/16] 2024-02-05 09:03:01,989 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 09:03:05,635 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1c9ab7c10>)
+[gpub058:0/16] 2024-02-05 09:03:05,635 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub058:0/16] 2024-02-05 09:03:05,683 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 09:10:08,961 (trainer:740) INFO: 28epoch:train:11201-11300batch: iter_time=3.899, forward_time=0.444, loss_ctc=44.754, loss_att=47.299, acc=0.757, loss=46.536, backward_time=0.310, grad_norm=36.952, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.899e-04, train_time=5.439
+[gpub058:0/16] 2024-02-05 09:12:47,675 (trainer:740) INFO: 28epoch:train:11301-11400batch: iter_time=8.967e-05, forward_time=0.467, loss_ctc=47.760, loss_att=52.960, acc=0.745, loss=51.400, backward_time=0.319, grad_norm=42.152, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.898e-04, train_time=1.588
+[gpub058:0/16] 2024-02-05 09:14:53,047 (trainer:740) INFO: 28epoch:train:11401-11500batch: iter_time=9.265e-05, forward_time=0.291, loss_ctc=55.532, loss_att=58.064, acc=0.747, loss=57.305, backward_time=0.286, grad_norm=54.059, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.898e-04, train_time=1.253
+[gpub058:0/16] 2024-02-05 09:17:58,032 (trainer:740) INFO: 28epoch:train:11501-11600batch: iter_time=9.426e-05, forward_time=0.417, loss_ctc=49.647, loss_att=47.151, acc=0.761, loss=47.900, backward_time=0.418, grad_norm=43.821, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.898e-04, train_time=1.850
+[gpub058:0/16] 2024-02-05 09:20:24,804 (trainer:740) INFO: 28epoch:train:11601-11700batch: iter_time=0.001, forward_time=0.384, loss_ctc=50.656, loss_att=49.321, acc=0.761, loss=49.722, backward_time=0.408, grad_norm=40.757, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.126, optim0_lr0=1.898e-04, train_time=1.466
+[gpub058:0/16] 2024-02-05 09:23:37,611 (trainer:740) INFO: 28epoch:train:11701-11800batch: iter_time=9.082e-05, forward_time=0.386, loss_ctc=47.453, loss_att=50.427, acc=0.748, loss=49.534, backward_time=0.386, grad_norm=37.307, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.897e-04, train_time=1.929
+[gpub058:0/16] 2024-02-05 09:25:51,542 (trainer:740) INFO: 28epoch:train:11801-11900batch: iter_time=9.365e-05, forward_time=0.289, loss_ctc=44.825, loss_att=41.995, acc=0.761, loss=42.844, backward_time=0.284, grad_norm=42.431, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.897e-04, train_time=1.338
+[gpub058:0/16] 2024-02-05 09:28:04,099 (trainer:740) INFO: 28epoch:train:11901-12000batch: iter_time=8.632e-05, forward_time=0.288, loss_ctc=42.338, loss_att=37.128, acc=0.762, loss=38.691, backward_time=0.282, grad_norm=36.235, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.897e-04, train_time=1.327
+[gpub058:0/16] 2024-02-05 09:30:50,038 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 09:31:09,649 (trainer:740) INFO: 28epoch:train:12001-12100batch: iter_time=9.358e-05, forward_time=0.456, loss_ctc=44.752, loss_att=40.386, acc=0.762, loss=41.696, backward_time=0.350, grad_norm=36.871, clip=100.000, loss_scale=8.234e+33, optim_step_time=0.100, optim0_lr0=1.897e-04, train_time=1.854
+[gpub058:0/16] 2024-02-05 09:33:14,011 (trainer:740) INFO: 28epoch:train:12101-12200batch: iter_time=9.148e-05, forward_time=0.290, loss_ctc=50.223, loss_att=45.599, acc=0.750, loss=46.986, backward_time=0.286, grad_norm=43.121, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.896e-04, train_time=1.245
+[gpub058:0/16] 2024-02-05 09:36:11,510 (trainer:740) INFO: 28epoch:train:12201-12300batch: iter_time=8.023e-04, forward_time=0.446, loss_ctc=47.184, loss_att=51.018, acc=0.749, loss=49.868, backward_time=0.378, grad_norm=36.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.896e-04, train_time=1.773
+[gpub058:0/16] 2024-02-05 09:39:10,509 (trainer:740) INFO: 28epoch:train:12301-12400batch: iter_time=9.810e-05, forward_time=0.400, loss_ctc=52.417, loss_att=54.077, acc=0.725, loss=53.579, backward_time=0.350, grad_norm=40.881, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.896e-04, train_time=1.792
+[gpub058:0/16] 2024-02-05 09:41:14,230 (trainer:740) INFO: 28epoch:train:12401-12500batch: iter_time=8.423e-05, forward_time=0.292, loss_ctc=49.344, loss_att=49.569, acc=0.759, loss=49.502, backward_time=0.287, grad_norm=38.016, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.896e-04, train_time=1.233
+[gpub058:0/16] 2024-02-05 09:41:34,520 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub058:0/16] 2024-02-05 09:41:53,253 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 09:41:56,800 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe138e23580>)
+[gpub058:0/16] 2024-02-05 09:41:56,800 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub058:0/16] 2024-02-05 09:41:56,803 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 09:50:20,423 (trainer:740) INFO: 28epoch:train:12501-12600batch: iter_time=3.973, forward_time=0.376, loss_ctc=44.605, loss_att=50.092, acc=0.758, loss=48.446, backward_time=0.300, grad_norm=37.484, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.896e-04, train_time=5.465
+[gpub058:0/16] 2024-02-05 09:52:57,900 (trainer:740) INFO: 28epoch:train:12601-12700batch: iter_time=9.166e-05, forward_time=0.448, loss_ctc=48.549, loss_att=51.036, acc=0.746, loss=50.290, backward_time=0.355, grad_norm=40.535, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.895e-04, train_time=1.574
+[gpub058:0/16] 2024-02-05 09:55:29,942 (trainer:740) INFO: 28epoch:train:12701-12800batch: iter_time=8.938e-05, forward_time=0.293, loss_ctc=52.042, loss_att=56.945, acc=0.755, loss=55.474, backward_time=0.288, grad_norm=50.594, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.895e-04, train_time=1.521
+[gpub058:0/16] 2024-02-05 09:58:00,629 (trainer:740) INFO: 28epoch:train:12801-12900batch: iter_time=9.634e-05, forward_time=0.386, loss_ctc=51.752, loss_att=48.332, acc=0.759, loss=49.358, backward_time=0.362, grad_norm=45.688, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.895e-04, train_time=1.506
+[gpub058:0/16] 2024-02-05 10:00:25,535 (trainer:740) INFO: 28epoch:train:12901-13000batch: iter_time=9.595e-05, forward_time=0.292, loss_ctc=50.467, loss_att=49.587, acc=0.755, loss=49.851, backward_time=0.286, grad_norm=37.418, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.895e-04, train_time=1.449
+[gpub058:0/16] 2024-02-05 10:02:40,291 (trainer:740) INFO: 28epoch:train:13001-13100batch: iter_time=0.009, forward_time=0.461, loss_ctc=38.950, loss_att=37.502, acc=0.773, loss=37.936, backward_time=0.318, grad_norm=37.777, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.894e-04, train_time=1.347
+[gpub058:0/16] 2024-02-05 10:05:15,080 (trainer:740) INFO: 28epoch:train:13101-13200batch: iter_time=9.298e-05, forward_time=0.290, loss_ctc=46.132, loss_att=44.625, acc=0.746, loss=45.077, backward_time=0.289, grad_norm=42.104, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.894e-04, train_time=1.548
+[gpub058:0/16] 2024-02-05 10:08:18,481 (trainer:740) INFO: 28epoch:train:13201-13300batch: iter_time=6.383e-04, forward_time=0.448, loss_ctc=47.011, loss_att=41.156, acc=0.764, loss=42.913, backward_time=0.329, grad_norm=36.233, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.894e-04, train_time=1.833
+[gpub058:0/16] 2024-02-05 10:10:21,634 (trainer:740) INFO: 28epoch:train:13301-13400batch: iter_time=9.415e-05, forward_time=0.289, loss_ctc=50.811, loss_att=42.447, acc=0.760, loss=44.956, backward_time=0.284, grad_norm=37.929, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.894e-04, train_time=1.232
+[gpub058:0/16] 2024-02-05 10:13:15,675 (trainer:740) INFO: 28epoch:train:13401-13500batch: iter_time=9.068e-05, forward_time=0.396, loss_ctc=46.159, loss_att=49.945, acc=0.744, loss=48.809, backward_time=0.358, grad_norm=36.826, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.894e-04, train_time=1.740
+[gpub058:0/16] 2024-02-05 10:15:15,353 (trainer:740) INFO: 28epoch:train:13501-13600batch: iter_time=9.482e-05, forward_time=0.292, loss_ctc=46.474, loss_att=49.744, acc=0.745, loss=48.763, backward_time=0.287, grad_norm=37.187, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.893e-04, train_time=1.194
+[gpub058:0/16] 2024-02-05 10:18:04,025 (trainer:740) INFO: 28epoch:train:13601-13700batch: iter_time=9.932e-05, forward_time=0.433, loss_ctc=52.672, loss_att=51.756, acc=0.742, loss=52.031, backward_time=0.310, grad_norm=39.878, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.893e-04, train_time=1.689
+[gpub058:0/16] 2024-02-05 10:19:36,253 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub058:0/16] 2024-02-05 10:19:55,687 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 10:19:59,306 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe1b72a76a0>)
+[gpub058:0/16] 2024-02-05 10:19:59,306 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub058:0/16] 2024-02-05 10:19:59,328 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 10:27:20,718 (trainer:740) INFO: 28epoch:train:13701-13800batch: iter_time=4.104, forward_time=0.288, loss_ctc=44.016, loss_att=47.434, acc=0.759, loss=46.408, backward_time=0.283, grad_norm=36.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.893e-04, train_time=5.567
+[gpub058:0/16] 2024-02-05 10:30:13,961 (trainer:740) INFO: 28epoch:train:13801-13900batch: iter_time=7.582e-05, forward_time=0.292, loss_ctc=47.584, loss_att=52.231, acc=0.750, loss=50.837, backward_time=0.285, grad_norm=39.878, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.893e-04, train_time=1.732
+[gpub058:0/16] 2024-02-05 10:32:55,028 (trainer:740) INFO: 28epoch:train:13901-14000batch: iter_time=8.566e-05, forward_time=0.459, loss_ctc=54.711, loss_att=57.915, acc=0.747, loss=56.954, backward_time=0.352, grad_norm=53.239, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.892e-04, train_time=1.610
+[gpub058:0/16] 2024-02-05 10:35:24,905 (trainer:740) INFO: 28epoch:train:14001-14100batch: iter_time=8.682e-05, forward_time=0.291, loss_ctc=49.084, loss_att=45.796, acc=0.765, loss=46.782, backward_time=0.286, grad_norm=44.815, clip=100.000, loss_scale=6.023e+33, optim_step_time=0.093, optim0_lr0=1.892e-04, train_time=1.499
+[gpub058:0/16] 2024-02-05 10:37:42,647 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub058:0/16] 2024-02-05 10:38:21,768 (trainer:740) INFO: 28epoch:train:14101-14200batch: iter_time=0.001, forward_time=0.469, loss_ctc=50.074, loss_att=49.123, acc=0.762, loss=49.408, backward_time=0.317, grad_norm=39.529, clip=100.000, loss_scale=1.012e+34, optim_step_time=0.110, optim0_lr0=1.892e-04, train_time=1.768
+[gpub058:0/16] 2024-02-05 10:40:45,617 (trainer:740) INFO: 28epoch:train:14201-14300batch: iter_time=8.466e-05, forward_time=0.292, loss_ctc=46.960, loss_att=50.880, acc=0.744, loss=49.704, backward_time=0.285, grad_norm=39.564, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.892e-04, train_time=1.439
+[gpub058:0/16] 2024-02-05 10:43:15,471 (trainer:740) INFO: 28epoch:train:14301-14400batch: iter_time=8.298e-05, forward_time=0.289, loss_ctc=45.064, loss_att=41.440, acc=0.762, loss=42.527, backward_time=0.282, grad_norm=40.669, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.891e-04, train_time=1.498
+[gpub058:0/16] 2024-02-05 10:46:11,931 (trainer:740) INFO: 28epoch:train:14401-14500batch: iter_time=9.149e-05, forward_time=0.521, loss_ctc=42.157, loss_att=36.650, acc=0.763, loss=38.302, backward_time=0.320, grad_norm=38.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.891e-04, train_time=1.764
+[gpub058:0/16] 2024-02-05 10:49:00,798 (trainer:740) INFO: 28epoch:train:14501-14600batch: iter_time=8.861e-05, forward_time=0.289, loss_ctc=44.508, loss_att=40.134, acc=0.763, loss=41.446, backward_time=0.282, grad_norm=34.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.891e-04, train_time=1.688
+[gpub058:0/16] 2024-02-05 10:51:19,202 (trainer:740) INFO: 28epoch:train:14601-14700batch: iter_time=2.418e-04, forward_time=0.408, loss_ctc=50.315, loss_att=45.149, acc=0.752, loss=46.699, backward_time=0.314, grad_norm=41.102, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.891e-04, train_time=1.384
+[gpub058:0/16] 2024-02-05 10:54:06,461 (trainer:740) INFO: 28epoch:train:14701-14800batch: iter_time=8.563e-05, forward_time=0.290, loss_ctc=47.082, loss_att=50.079, acc=0.752, loss=49.180, backward_time=0.286, grad_norm=36.997, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.891e-04, train_time=1.672
+[gpub058:0/16] 2024-02-05 10:56:46,141 (trainer:740) INFO: 28epoch:train:14801-14900batch: iter_time=8.588e-05, forward_time=0.292, loss_ctc=51.789, loss_att=52.983, acc=0.731, loss=52.625, backward_time=0.286, grad_norm=40.012, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.890e-04, train_time=1.597
+[gpub058:0/16] 2024-02-05 11:00:07,614 (trainer:740) INFO: 28epoch:train:14901-15000batch: iter_time=3.673e-04, forward_time=0.365, loss_ctc=48.712, loss_att=49.442, acc=0.759, loss=49.223, backward_time=0.440, grad_norm=36.158, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.890e-04, train_time=2.013
+[gpub058:0/16] 2024-02-05 11:40:06,981 (trainer:346) INFO: 28epoch results: [train] iter_time=0.290, forward_time=0.334, loss_ctc=48.866, loss_att=48.308, acc=0.747, loss=48.476, backward_time=0.308, grad_norm=41.482, clip=100.000, loss_scale=5.519e+33, optim_step_time=0.096, optim0_lr0=1.907e-04, train_time=1.771, time=7 hours, 23 minutes and 14.91 seconds, total_count=450000, gpu_max_cached_mem_GB=41.105, [valid] loss_ctc=38.769, cer_ctc=0.193, loss_att=40.854, acc=0.668, cer=0.355, wer=1.000, loss=40.229, time=39 minutes and 34.66 seconds, total_count=140130, gpu_max_cached_mem_GB=41.105
+[gpub058:0/16] 2024-02-05 11:40:17,869 (trainer:394) INFO: The best model has been updated: valid.total_count
+[gpub058:0/16] 2024-02-05 11:40:17,875 (trainer:275) INFO: 29/45epoch started. Estimated time to finish: 6 days, 8 hours and 49 minutes
+[gpub058:0/16] 2024-02-05 11:40:17,886 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub058:0/16] 2024-02-05 11:40:35,703 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 11:40:39,131 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe3b4acd780>)
+[gpub058:0/16] 2024-02-05 11:40:39,132 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub058:0/16] 2024-02-05 11:40:39,135 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 11:48:27,862 (trainer:740) INFO: 29epoch:train:1-100batch: iter_time=3.704, forward_time=0.360, loss_ctc=45.736, loss_att=37.560, acc=0.779, loss=40.013, backward_time=0.294, grad_norm=40.355, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.890e-04, train_time=4.900
+[gpub058:0/16] 2024-02-05 11:50:29,479 (trainer:740) INFO: 29epoch:train:101-200batch: iter_time=8.141e-05, forward_time=0.290, loss_ctc=53.227, loss_att=50.924, acc=0.740, loss=51.615, backward_time=0.284, grad_norm=44.349, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.890e-04, train_time=1.216
+[gpub058:0/16] 2024-02-05 11:52:55,039 (trainer:740) INFO: 29epoch:train:201-300batch: iter_time=8.428e-05, forward_time=0.293, loss_ctc=51.624, loss_att=49.611, acc=0.751, loss=50.215, backward_time=0.299, grad_norm=43.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.889e-04, train_time=1.453
+[gpub058:0/16] 2024-02-05 11:55:36,002 (trainer:740) INFO: 29epoch:train:301-400batch: iter_time=9.736e-05, forward_time=0.426, loss_ctc=61.337, loss_att=52.807, acc=0.731, loss=55.366, backward_time=0.347, grad_norm=54.220, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.889e-04, train_time=1.612
+[gpub058:0/16] 2024-02-05 11:57:56,423 (trainer:740) INFO: 29epoch:train:401-500batch: iter_time=9.799e-05, forward_time=0.289, loss_ctc=45.291, loss_att=47.243, acc=0.734, loss=46.657, backward_time=0.285, grad_norm=40.422, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.889e-04, train_time=1.404
+[gpub058:0/16] 2024-02-05 11:59:58,209 (trainer:740) INFO: 29epoch:train:501-600batch: iter_time=9.651e-05, forward_time=0.320, loss_ctc=48.554, loss_att=44.044, acc=0.744, loss=45.397, backward_time=0.292, grad_norm=44.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.889e-04, train_time=1.217
+[gpub058:0/16] 2024-02-05 12:02:18,327 (trainer:740) INFO: 29epoch:train:601-700batch: iter_time=9.003e-05, forward_time=0.406, loss_ctc=54.479, loss_att=48.799, acc=0.750, loss=50.503, backward_time=0.369, grad_norm=39.726, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.889e-04, train_time=1.402
+[gpub058:0/16] 2024-02-05 12:04:46,638 (trainer:740) INFO: 29epoch:train:701-800batch: iter_time=9.552e-05, forward_time=0.290, loss_ctc=53.371, loss_att=51.759, acc=0.733, loss=52.243, backward_time=0.284, grad_norm=40.872, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.888e-04, train_time=1.483
+[gpub058:0/16] 2024-02-05 12:06:56,305 (trainer:740) INFO: 29epoch:train:801-900batch: iter_time=9.018e-05, forward_time=0.332, loss_ctc=61.209, loss_att=53.019, acc=0.747, loss=55.476, backward_time=0.310, grad_norm=48.408, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.888e-04, train_time=1.296
+[gpub058:0/16] 2024-02-05 12:09:44,872 (trainer:740) INFO: 29epoch:train:901-1000batch: iter_time=9.521e-05, forward_time=0.338, loss_ctc=46.789, loss_att=42.506, acc=0.749, loss=43.791, backward_time=0.333, grad_norm=35.252, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.888e-04, train_time=1.686
+[gpub058:0/16] 2024-02-05 12:11:59,651 (trainer:740) INFO: 29epoch:train:1001-1100batch: iter_time=9.519e-05, forward_time=0.290, loss_ctc=47.823, loss_att=46.240, acc=0.755, loss=46.715, backward_time=0.284, grad_norm=38.391, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.888e-04, train_time=1.348
+[gpub058:0/16] 2024-02-05 12:14:31,644 (trainer:740) INFO: 29epoch:train:1101-1200batch: iter_time=9.070e-05, forward_time=0.435, loss_ctc=51.511, loss_att=50.112, acc=0.741, loss=50.532, backward_time=0.346, grad_norm=55.198, clip=100.000, loss_scale=5.452e+33, optim_step_time=0.102, optim0_lr0=1.887e-04, train_time=1.519
+[gpub058:0/16] 2024-02-05 12:16:07,416 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+
+gpub059:2626467:2626539 [1] proxy.cc:1059 NCCL WARN [Proxy Service] Poll failed with error 1
+[gpub058:0/16] 2024-02-05 12:16:26,842 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub058:0/16] 2024-02-05 12:16:30,431 (abs_task:1616) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fe14f5b3550>)
+[gpub058:0/16] 2024-02-05 12:16:30,431 (abs_task:1617) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub058:0/16] 2024-02-05 12:16:30,505 (abs_task:1618) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub058:0/16] 2024-02-05 12:23:23,544 (trainer:740) INFO: 29epoch:train:1201-1300batch: iter_time=3.955, forward_time=0.301, loss_ctc=42.756, loss_att=44.642, acc=0.751, loss=44.076, backward_time=0.285, grad_norm=36.088, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.887e-04, train_time=5.319
+[gpub058:0/16] 2024-02-05 12:25:57,116 (trainer:740) INFO: 29epoch:train:1301-1400batch: iter_time=8.834e-05, forward_time=0.429, loss_ctc=46.583, loss_att=43.040, acc=0.757, loss=44.103, backward_time=0.358, grad_norm=40.776, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=1.887e-04, train_time=1.535
+[gpub058:0/16] 2024-02-05 12:28:02,781 (trainer:740) INFO: 29epoch:train:1401-1500batch: iter_time=8.676e-05, forward_time=0.293, loss_ctc=57.213, loss_att=52.280, acc=0.748, loss=53.760, backward_time=0.286, grad_norm=47.402, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.887e-04, train_time=1.256
+[gpub058:0/16] 2024-02-05 12:30:16,651 (trainer:740) INFO: 29epoch:train:1501-1600batch: iter_time=9.515e-05, forward_time=0.303, loss_ctc=48.562, loss_att=44.553, acc=0.754, loss=45.756, backward_time=0.290, grad_norm=42.866, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.887e-04, train_time=1.339
+[gpub058:0/16] 2024-02-05 12:32:55,225 (trainer:740) INFO: 29epoch:train:1601-1700batch: iter_time=9.610e-05, forward_time=0.352, loss_ctc=57.465, loss_att=53.993, acc=0.724, loss=55.034, backward_time=0.302, grad_norm=49.336, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.886e-04, train_time=1.586
+[gpub058:0/16] 2024-02-05 12:34:10,026 (trainer:671) WARNING: The grad norm is nan. Skipping updating the model.
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 2938841.0 ON gpub058 CANCELLED AT 2024-02-05T12:34:53 ***
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..2ea517431109d1d3a24f257b92d73ebf06484d7a
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
@@ -0,0 +1,2927 @@
+# Running on gpub010.delta.ncsa.illinois.edu
+# Started at Sat Feb 10 11:47:58 CST 2024
+# SLURMD_NODENAME=gpub010
+# SLURM_CLUSTER_NAME=delta
+# SLURM_CONF=/var/spool/slurmd/conf-cache/slurm.conf
+# SLURM_CPUS_ON_NODE=64
+# SLURM_CPUS_PER_TASK=64
+# SLURM_EXPORT_ENV=PATH
+# SLURM_GET_USER_ENV=1
+# SLURM_GPUS_ON_NODE=4
+# SLURM_GTIDS=0
+# SLURM_JOBID=2968474
+# SLURM_JOB_ACCOUNT=bbjs-delta-gpu
+# SLURM_JOB_CPUS_PER_NODE='64(x4)'
+# SLURM_JOB_END_TIME=1707760063
+# SLURM_JOB_GID=202
+# SLURM_JOB_GPUS=0,1,2,3
+# SLURM_JOB_ID=2968474
+# SLURM_JOB_NAME=exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/train.log
+# SLURM_JOB_NODELIST='gpub[010,037,076,082]'
+# SLURM_JOB_NUM_NODES=4
+# SLURM_JOB_PARTITION=gpuA40x4
+# SLURM_JOB_QOS=bbjs-delta-gpu
+# SLURM_JOB_START_TIME=1707587263
+# SLURM_JOB_UID=68077
+# SLURM_JOB_USER=peng6
+# SLURM_LOCALID=0
+# SLURM_MEM_PER_NODE=240000
+# SLURM_MPI_TYPE=pmi2
+# SLURM_NNODES=4
+# SLURM_NODEID=0
+# SLURM_NODELIST='gpub[010,037,076,082]'
+# SLURM_NODE_ALIASES='(null)'
+# SLURM_OPEN_MODE=a
+# SLURM_PRIO_PROCESS=0
+# SLURM_PROCID=0
+# SLURM_SUBMIT_DIR=/scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1
+# SLURM_SUBMIT_HOST=dt-login03.delta.ncsa.illinois.edu
+# SLURM_TASKS_PER_NODE='1(x4)'
+# SLURM_TASK_PID=2415375
+# SLURM_TOPOLOGY_ADDR=ss00.ss09.gpub010
+# SLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.node
+# SLURM_WORKING_CLUSTER=delta:dt-sched:6817:9984:109
+# srun --export=ALL python3 -m espnet2.bin.s2t_train --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6 
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+GpuFreq=control_disabled
+/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_sh/scratch/bbjs/peng6/espnet-owsm-train-20240205/tools/miniconda/envs/espnet/bin/python3 /scratch/bbjs/peng6/espnet-owsm-train-20240205/espnet2/bin/s2t_train.py --use_preprocessor true --bpemodel data/token_list/bpe_unigram50000/bpe.model --token_type bpe --token_list data/token_list/bpe_unigram50000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev_v3/wav.scp,speech,kaldi_ark --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/speech_shape --resume true --fold_length 80000 --output_dir exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000 --config conf/train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/wav.scp,speech,kaldi_ark --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/speech_shape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6
+ape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --mulape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6
+ape --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.prev,text_prev,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_prev_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text.ctc,text_ctc,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_ctc_shape.bpe --fold_length 150 --train_data_path_and_name_and_type exp/s2t_stats_raw_bpe50000/splits12/text,text,text --train_shape_file exp/s2t_stats_raw_bpe50000/splits12/text_shape.bpe --multiple_iterator true --valid_data_path_and_name_and_type dump/raw/dev_v3/text.prev,text_prev,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_prev_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text.ctc,text_ctc,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_ctc_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev_v3/text,text,text --valid_shape_file exp/s2t_stats_raw_bpe50000/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6
+tiprocessing_distributed true --dist_launcher slurm --dist_init_method file:///scratch/bbjs/peng6/espnet-owsm-train-20240205/egs2/owsm_v3.1/s2t1/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/.dist_init_01501f2e-ee88-4157-a837-1cdd946f45c6
+[gpub010:0/16] 2024-02-10 11:52:06,129 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[gpub010:0/16] 2024-02-10 11:52:06,297 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 16 nodes.
+[gpub010:0/16] 2024-02-10 11:52:06,382 (s2t:464) INFO: Vocabulary size: 50002
+[gpub010:0/16] 2024-02-10 11:52:15,378 (abs_task:1271) INFO: pytorch.version=1.13.1, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[gpub010:0/16] 2024-02-10 11:52:15,384 (abs_task:1272) INFO: Model structure:
+ESPnetS2TModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=400, hop_length=160, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 27], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxisVariableMaxWidth(mask_width_ratio_range=[0.0, 0.05], num_mask=4, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/s2t_stats_raw_bpe50000/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): EBranchformerEncoder(
+    (embed): Conv2dSubsampling(
+      (conv): Sequential(
+        (0): Conv2d(1, 768, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(768, 768, kernel_size=(3, 3), stride=(2, 2))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=14592, out_features=768, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (1): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (2): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (3): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (4): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (5): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (6): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (7): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+      (8): EBranchformerEncoderLayer(
+        (attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (cgmlp): ConvolutionalGatingMLP(
+          (channel_proj1): Sequential(
+            (0): Linear(in_features=768, out_features=3072, bias=True)
+            (1): GELU(approximate='none')
+          )
+          (csgu): ConvolutionalSpatialGatingUnit(
+            (norm): LayerNorm((1536,), eps=1e-12, elementwise_affine=True)
+            (conv): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+            (act): Identity()
+            (dropout): Dropout(p=0.1, inplace=False)
+          )
+          (channel_proj2): Linear(in_features=1536, out_features=768, bias=True)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (feed_forward_macaron): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): Swish()
+        )
+        (norm_ff): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_ff_macaron): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mha): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_mlp): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm_final): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (depthwise_conv_fusion): Conv1d(1536, 1536, kernel_size=(31,), stride=(1,), padding=(15,), groups=1536)
+        (merge_proj): Linear(in_features=1536, out_features=768, bias=True)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(50002, 768)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=768, out_features=50002, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (6): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (7): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (8): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=768, out_features=768, bias=True)
+          (linear_k): Linear(in_features=768, out_features=768, bias=True)
+          (linear_v): Linear(in_features=768, out_features=768, bias=True)
+          (linear_out): Linear(in_features=768, out_features=768, bias=True)
+          (dropout): Identity()
+          (q_norm): Identity()
+          (k_norm): Identity()
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=768, out_features=3072, bias=True)
+          (w_2): Linear(in_features=3072, out_features=768, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=768, out_features=50002, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetS2TModel
+    Total Number of model parameters: 366.65 M
+    Number of trainable parameters: 366.65 M (100.0%)
+    Size: 1.47 GB
+    Type: torch.float32
+[gpub010:0/16] 2024-02-10 11:52:15,385 (abs_task:1275) INFO: Optimizer:
+AdamW (
+Parameter Group 0
+    amsgrad: False
+    betas: [0.9, 0.98]
+    capturable: False
+    eps: 1e-06
+    foreach: None
+    initial_lr: 0.0005
+    lr: 1.6666666666666667e-09
+    maximize: False
+    weight_decay: 0.0
+)
+[gpub010:0/16] 2024-02-10 11:52:15,385 (abs_task:1276) INFO: Scheduler: PiecewiseLinearWarmupLR(warmup_steps_list=[0, 30000, 60000], warmup_lr_list=[0.0, 5e-05, 0.0005])
+[gpub010:0/16] 2024-02-10 11:52:15,393 (abs_task:1285) INFO: Saving the configuration in exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml
+[gpub010:0/16] 2024-02-10 11:52:20,873 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 11:52:21,776 (abs_task:1663) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev_v3/wav.scp", "type": "kaldi_ark"}
+  text_prev: {"path": "dump/raw/dev_v3/text.prev", "type": "text"}
+  text_ctc: {"path": "dump/raw/dev_v3/text.ctc", "type": "text"}
+  text: {"path": "dump/raw/dev_v3/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc2f9cc7e80>)
+[gpub010:0/16] 2024-02-10 11:52:21,776 (abs_task:1664) INFO: [valid] Batch sampler: UnsortedBatchSampler(N-batch=4671, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/valid/speech_shape, 
+[gpub010:0/16] 2024-02-10 11:52:21,777 (abs_task:1665) INFO: [valid] mini-batch sizes summary: N-batch=4671, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 11:52:32,362 (trainer:168) INFO: The training was resumed using exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/checkpoint.pth
+gpub010:2415458:2415458 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:2415458:2415458 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub010:2415458:2415458 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub010:2415458:2415458 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.14.3+cuda11.7
+[gpub010:0/16] 2024-02-10 11:52:37,461 (trainer:302) INFO: 40/45epoch started
+[gpub010:0/16] 2024-02-10 11:52:37,515 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-10 11:52:55,705 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 11:52:59,362 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc0fb41be80>)
+[gpub010:0/16] 2024-02-10 11:52:59,363 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-10 11:52:59,366 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+gpub037:2479731:2479731 [0] NCCL INFO cudaDriverVersion 12020
+gpub037:2479731:2479731 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.137<0>
+gpub037:2479731:2479731 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub037:2479731:2479731 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub037:2479731:2479783 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub037:2479731:2479783 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub037:2479731:2479783 [0] NCCL INFO Using network AWS Libfabric
+gpub037:2479731:2479783 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub037:2479731:2479783 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub037:2479731:2479783 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/0/-1->4->12
+gpub037:2479731:2479783 [0] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 00/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 4[7000] -> 5[46000] via P2P/IPC
+gpub037:2479731:2479783 [0] NCCL INFO Connected all rings
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [send] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [receive] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub037:2479731:2479783 [0] NCCL INFO Connected all trees
+gpub037:2479731:2479783 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub037:2479731:2479783 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub037:2479731:2479783 [0] NCCL INFO comm 0x556289adbe40 rank 4 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub010:2415461:2415461 [3] NCCL INFO cudaDriverVersion 12020
+gpub010:2415461:2415461 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:2415461:2415461 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub010:2415461:2415461 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub010:2415461:2415512 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub010:2415461:2415512 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub010:2415461:2415512 [3] NCCL INFO Using network AWS Libfabric
+gpub010:2415461:2415512 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub010:2415461:2415512 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub010:2415461:2415512 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2
+gpub010:2415461:2415512 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub010:2415461:2415512 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub010:2415461:2415512 [3] NCCL INFO Connected all rings
+gpub010:2415461:2415512 [3] NCCL INFO Channel 00/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub010:2415461:2415512 [3] NCCL INFO Channel 01/0 : 3[c7000] -> 2[85000] via P2P/IPC
+gpub010:2415461:2415512 [3] NCCL INFO Connected all trees
+gpub010:2415461:2415512 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:2415461:2415512 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:2415461:2415512 [3] NCCL INFO comm 0x56066bfecd40 rank 3 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub010:2415459:2415459 [1] NCCL INFO cudaDriverVersion 12020
+gpub010:2415459:2415459 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:2415459:2415459 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub010:2415459:2415459 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub010:2415459:2415511 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub010:2415459:2415511 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub010:2415459:2415511 [1] NCCL INFO Using network AWS Libfabric
+gpub010:2415459:2415511 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub010:2415459:2415511 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub010:2415459:2415511 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+gpub010:2415459:2415511 [1] NCCL INFO Channel 00/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub010:2415459:2415511 [1] NCCL INFO Channel 01/0 : 1[46000] -> 2[85000] via P2P/IPC
+gpub010:2415459:2415511 [1] NCCL INFO Connected all rings
+gpub010:2415459:2415511 [1] NCCL INFO Channel 00/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub010:2415459:2415511 [1] NCCL INFO Channel 01/0 : 1[46000] -> 0[7000] via P2P/IPC
+gpub010:2415459:2415511 [1] NCCL INFO Connected all trees
+gpub010:2415459:2415511 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:2415459:2415511 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:2415459:2415511 [1] NCCL INFO comm 0x5576d44db4b0 rank 1 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub037:2479733:2479733 [2] NCCL INFO cudaDriverVersion 12020
+gpub037:2479733:2479733 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.137<0>
+gpub037:2479733:2479733 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub037:2479733:2479733 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub037:2479733:2479782 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub037:2479733:2479782 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub037:2479733:2479782 [2] NCCL INFO Using network AWS Libfabric
+gpub037:2479733:2479782 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub037:2479733:2479782 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub037:2479733:2479782 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+gpub037:2479733:2479782 [2] NCCL INFO Channel 00/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub037:2479733:2479782 [2] NCCL INFO Channel 01/0 : 6[85000] -> 7[c7000] via P2P/IPC
+gpub037:2479733:2479782 [2] NCCL INFO Connected all rings
+gpub037:2479733:2479782 [2] NCCL INFO Channel 00/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub037:2479733:2479782 [2] NCCL INFO Channel 01/0 : 6[85000] -> 5[46000] via P2P/IPC
+gpub037:2479733:2479782 [2] NCCL INFO Connected all trees
+gpub037:2479733:2479782 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub037:2479733:2479782 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub037:2479733:2479782 [2] NCCL INFO comm 0x55c8aa09ba70 rank 6 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub037:2479732:2479732 [1] NCCL INFO cudaDriverVersion 12020
+gpub037:2479732:2479732 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.137<0>
+gpub037:2479732:2479732 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub037:2479732:2479732 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub037:2479732:2479784 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub037:2479732:2479784 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub037:2479732:2479784 [1] NCCL INFO Using network AWS Libfabric
+gpub037:2479732:2479784 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub037:2479732:2479784 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub037:2479732:2479784 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/8/-1->5->4
+gpub037:2479732:2479784 [1] NCCL INFO Channel 00/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub037:2479732:2479784 [1] NCCL INFO Channel 01/0 : 5[46000] -> 6[85000] via P2P/IPC
+gpub010:2415460:2415460 [2] NCCL INFO cudaDriverVersion 12020
+gpub010:2415460:2415460 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.110<0>
+gpub010:2415460:2415460 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub010:2415460:2415460 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub010:2415460:2415513 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub010:2415460:2415513 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub010:2415460:2415513 [2] NCCL INFO Using network AWS Libfabric
+gpub010:2415460:2415513 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub010:2415460:2415513 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub010:2415460:2415513 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+gpub010:2415460:2415513 [2] NCCL INFO Channel 00/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub010:2415460:2415513 [2] NCCL INFO Channel 01/0 : 2[85000] -> 3[c7000] via P2P/IPC
+gpub037:2479732:2479784 [1] NCCL INFO Connected all rings
+gpub037:2479732:2479784 [1] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub037:2479732:2479784 [1] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [receive] via NET/AWS Libfabric/1
+gpub037:2479732:2479784 [1] NCCL INFO Channel 00/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub037:2479732:2479784 [1] NCCL INFO Channel 01/0 : 5[46000] -> 4[7000] via P2P/IPC
+gpub037:2479732:2479784 [1] NCCL INFO Connected all trees
+gpub037:2479732:2479784 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub037:2479732:2479784 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub037:2479732:2479784 [1] NCCL INFO comm 0x5641d0bbff50 rank 5 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub010:2415460:2415513 [2] NCCL INFO Connected all rings
+gpub010:2415460:2415513 [2] NCCL INFO Channel 00/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub010:2415460:2415513 [2] NCCL INFO Channel 01/0 : 2[85000] -> 1[46000] via P2P/IPC
+gpub010:2415460:2415513 [2] NCCL INFO Connected all trees
+gpub010:2415460:2415513 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:2415460:2415513 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:2415460:2415513 [2] NCCL INFO comm 0x560007013060 rank 2 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub082:3674493:3674493 [0] NCCL INFO cudaDriverVersion 12020
+gpub082:3674493:3674493 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.182<0>
+gpub082:3674493:3674493 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub082:3674493:3674493 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub082:3674493:3674547 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub082:3674493:3674547 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub082:3674493:3674547 [0] NCCL INFO Using network AWS Libfabric
+gpub082:3674493:3674547 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub082:3674493:3674547 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub082:3674493:3674547 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/4/-1->12->-1
+gpub082:3674493:3674547 [0] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub010:2415458:2415510 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub010:2415458:2415510 [0] NCCL INFO Using network AWS Libfabric
+gpub010:2415458:2415510 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub010:2415458:2415510 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub010:2415458:2415510 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub010:2415458:2415510 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+gpub010:2415458:2415510 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->4
+gpub010:2415458:2415510 [0] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Channel 00/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub082:3674493:3674547 [0] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub082:3674493:3674547 [0] NCCL INFO Channel 00/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub082:3674493:3674547 [0] NCCL INFO Channel 01/0 : 12[7000] -> 13[46000] via P2P/IPC
+gpub082:3674493:3674547 [0] NCCL INFO Connected all rings
+gpub082:3674493:3674547 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub082:3674493:3674547 [0] NCCL INFO Channel 01/0 : 4[7000] -> 12[7000] [receive] via NET/AWS Libfabric/1
+gpub082:3674493:3674547 [0] NCCL INFO Channel 01/0 : 12[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub082:3674493:3674547 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub082:3674493:3674547 [0] NCCL INFO Connected all trees
+gpub082:3674493:3674547 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub082:3674493:3674547 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:2415458:2415510 [0] NCCL INFO Channel 01/0 : 0[7000] -> 1[46000] via P2P/IPC
+gpub010:2415458:2415510 [0] NCCL INFO Connected all rings
+gpub010:2415458:2415510 [0] NCCL INFO Channel 01/0 : 0[7000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Channel 01/0 : 4[7000] -> 0[7000] [receive] via NET/AWS Libfabric/1
+gpub010:2415458:2415510 [0] NCCL INFO Connected all trees
+gpub010:2415458:2415510 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub010:2415458:2415510 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub010:2415458:2415510 [0] NCCL INFO comm 0x5627bbb736f0 rank 0 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub082:3674493:3674547 [0] NCCL INFO comm 0x5595c37939f0 rank 12 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+gpub082:3674496:3674496 [3] NCCL INFO cudaDriverVersion 12020
+gpub082:3674496:3674496 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.182<0>
+gpub082:3674496:3674496 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub082:3674496:3674496 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub082:3674496:3674548 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub082:3674496:3674548 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub082:3674496:3674548 [3] NCCL INFO Using network AWS Libfabric
+gpub082:3674496:3674548 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub082:3674496:3674548 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub082:3674496:3674548 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+gpub082:3674496:3674548 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub082:3674496:3674548 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub082:3674496:3674548 [3] NCCL INFO Connected all rings
+gpub082:3674496:3674548 [3] NCCL INFO Channel 00/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub082:3674496:3674548 [3] NCCL INFO Channel 01/0 : 15[c7000] -> 14[85000] via P2P/IPC
+gpub082:3674496:3674548 [3] NCCL INFO Connected all trees
+gpub082:3674496:3674548 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub082:3674496:3674548 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub082:3674496:3674548 [3] NCCL INFO comm 0x55cb4fbfa630 rank 15 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub076:3268963:3268963 [3] NCCL INFO cudaDriverVersion 12020
+gpub076:3268963:3268963 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3268963:3268963 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3268963:3268963 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3268963:3269014 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3268963:3269014 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3268963:3269014 [3] NCCL INFO Using network AWS Libfabric
+gpub076:3268963:3269014 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub076:3268963:3269014 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3268963:3269014 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10
+gpub076:3268963:3269014 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub076:3268963:3269014 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub076:3268963:3269014 [3] NCCL INFO Connected all rings
+gpub076:3268963:3269014 [3] NCCL INFO Channel 00/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub076:3268963:3269014 [3] NCCL INFO Channel 01/0 : 11[c7000] -> 10[85000] via P2P/IPC
+gpub076:3268963:3269014 [3] NCCL INFO Connected all trees
+gpub076:3268963:3269014 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3268963:3269014 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3268963:3269014 [3] NCCL INFO comm 0x56218608af50 rank 11 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub082:3674494:3674494 [1] NCCL INFO cudaDriverVersion 12020
+gpub082:3674494:3674494 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.182<0>
+gpub082:3674494:3674494 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub082:3674494:3674494 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub082:3674494:3674545 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub082:3674494:3674545 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub082:3674494:3674545 [1] NCCL INFO Using network AWS Libfabric
+gpub082:3674494:3674545 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub082:3674494:3674545 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub082:3674494:3674545 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+gpub082:3674494:3674545 [1] NCCL INFO Channel 00/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub082:3674494:3674545 [1] NCCL INFO Channel 01/0 : 13[46000] -> 14[85000] via P2P/IPC
+gpub082:3674494:3674545 [1] NCCL INFO Connected all rings
+gpub082:3674494:3674545 [1] NCCL INFO Channel 00/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub082:3674494:3674545 [1] NCCL INFO Channel 01/0 : 13[46000] -> 12[7000] via P2P/IPC
+gpub082:3674494:3674545 [1] NCCL INFO Connected all trees
+gpub082:3674494:3674545 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub082:3674494:3674545 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub082:3674494:3674545 [1] NCCL INFO comm 0x5605a17f6130 rank 13 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub082:3674495:3674495 [2] NCCL INFO cudaDriverVersion 12020
+gpub082:3674495:3674495 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.182<0>
+gpub082:3674495:3674495 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub082:3674495:3674495 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub082:3674495:3674546 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub082:3674495:3674546 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub082:3674495:3674546 [2] NCCL INFO Using network AWS Libfabric
+gpub082:3674495:3674546 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub082:3674495:3674546 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub082:3674495:3674546 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+gpub082:3674495:3674546 [2] NCCL INFO Channel 00/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub082:3674495:3674546 [2] NCCL INFO Channel 01/0 : 14[85000] -> 15[c7000] via P2P/IPC
+gpub082:3674495:3674546 [2] NCCL INFO Connected all rings
+gpub082:3674495:3674546 [2] NCCL INFO Channel 00/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub082:3674495:3674546 [2] NCCL INFO Channel 01/0 : 14[85000] -> 13[46000] via P2P/IPC
+gpub082:3674495:3674546 [2] NCCL INFO Connected all trees
+gpub082:3674495:3674546 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub082:3674495:3674546 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub082:3674495:3674546 [2] NCCL INFO comm 0x556db2db58a0 rank 14 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub037:2479734:2479734 [3] NCCL INFO cudaDriverVersion 12020
+gpub037:2479734:2479734 [3] NCCL INFO Bootstrap : Using eth1:172.28.23.137<0>
+gpub037:2479734:2479734 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub037:2479734:2479734 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub037:2479734:2479781 [3] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub037:2479734:2479781 [3] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub037:2479734:2479781 [3] NCCL INFO Using network AWS Libfabric
+gpub037:2479734:2479781 [3] NCCL INFO Setting affinity for GPU 3 to ffff
+gpub037:2479734:2479781 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub037:2479734:2479781 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+gpub037:2479734:2479781 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub037:2479734:2479781 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [send] via NET/AWS Libfabric/1
+gpub037:2479734:2479781 [3] NCCL INFO Connected all rings
+gpub037:2479734:2479781 [3] NCCL INFO Channel 00/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub037:2479734:2479781 [3] NCCL INFO Channel 01/0 : 7[c7000] -> 6[85000] via P2P/IPC
+gpub037:2479734:2479781 [3] NCCL INFO Connected all trees
+gpub037:2479734:2479781 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub037:2479734:2479781 [3] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub037:2479734:2479781 [3] NCCL INFO comm 0x560fe900a3e0 rank 7 nranks 16 cudaDev 3 busId c7000 - Init COMPLETE
+gpub076:3268961:3268961 [1] NCCL INFO cudaDriverVersion 12020
+gpub076:3268961:3268961 [1] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3268961:3268961 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3268961:3268961 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3268961:3269015 [1] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3268961:3269015 [1] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3268961:3269015 [1] NCCL INFO Using network AWS Libfabric
+gpub076:3268961:3269015 [1] NCCL INFO Setting affinity for GPU 1 to ffff,00000000
+gpub076:3268961:3269015 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3268961:3269015 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] 10/-1/-1->9->8
+gpub076:3268961:3269015 [1] NCCL INFO Channel 00/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub076:3268961:3269015 [1] NCCL INFO Channel 01/0 : 9[46000] -> 10[85000] via P2P/IPC
+gpub076:3268961:3269015 [1] NCCL INFO Connected all rings
+gpub076:3268961:3269015 [1] NCCL INFO Channel 00/0 : 4[7000] -> 9[46000] [receive] via NET/AWS Libfabric/1
+gpub076:3268961:3269015 [1] NCCL INFO Channel 00/0 : 9[46000] -> 4[7000] [send] via NET/AWS Libfabric/1
+gpub076:3268961:3269015 [1] NCCL INFO Channel 00/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub076:3268961:3269015 [1] NCCL INFO Channel 01/0 : 9[46000] -> 8[7000] via P2P/IPC
+gpub076:3268961:3269015 [1] NCCL INFO Connected all trees
+gpub076:3268961:3269015 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3268961:3269015 [1] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3268961:3269015 [1] NCCL INFO comm 0x558968564390 rank 9 nranks 16 cudaDev 1 busId 46000 - Init COMPLETE
+gpub076:3268962:3268962 [2] NCCL INFO cudaDriverVersion 12020
+gpub076:3268962:3268962 [2] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3268962:3268962 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3268962:3268962 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3268962:3269013 [2] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3268962:3269013 [2] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3268962:3269013 [2] NCCL INFO Using network AWS Libfabric
+gpub076:3268962:3269013 [2] NCCL INFO Setting affinity for GPU 2 to ffff0000
+gpub076:3268962:3269013 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3268962:3269013 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+gpub076:3268962:3269013 [2] NCCL INFO Channel 00/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub076:3268962:3269013 [2] NCCL INFO Channel 01/0 : 10[85000] -> 11[c7000] via P2P/IPC
+gpub076:3268962:3269013 [2] NCCL INFO Connected all rings
+gpub076:3268962:3269013 [2] NCCL INFO Channel 00/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub076:3268962:3269013 [2] NCCL INFO Channel 01/0 : 10[85000] -> 9[46000] via P2P/IPC
+gpub076:3268962:3269013 [2] NCCL INFO Connected all trees
+gpub076:3268962:3269013 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3268962:3269013 [2] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3268962:3269013 [2] NCCL INFO comm 0x561f7f19bb70 rank 10 nranks 16 cudaDev 2 busId 85000 - Init COMPLETE
+gpub076:3268960:3268960 [0] NCCL INFO cudaDriverVersion 12020
+gpub076:3268960:3268960 [0] NCCL INFO Bootstrap : Using eth1:172.28.23.176<0>
+gpub076:3268960:3268960 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+gpub076:3268960:3268960 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+gpub076:3268960:3269012 [0] NCCL INFO NET/OFI Using aws-ofi-nccl 1.6.0
+gpub076:3268960:3269012 [0] NCCL INFO NET/OFI Selected Provider is cxi (found 2 nics)
+gpub076:3268960:3269012 [0] NCCL INFO Using network AWS Libfabric
+gpub076:3268960:3269012 [0] NCCL INFO Setting affinity for GPU 0 to ffff0000,00000000
+gpub076:3268960:3269012 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 1.
+gpub076:3268960:3269012 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->5
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 01/0 : 7[c7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub076:3268960:3269012 [0] NCCL INFO Channel 01/0 : 8[7000] -> 9[46000] via P2P/IPC
+gpub076:3268960:3269012 [0] NCCL INFO Connected all rings
+gpub076:3268960:3269012 [0] NCCL INFO Channel 01/0 : 5[46000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 8[7000] -> 12[7000] [send] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 0[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 8[7000] -> 0[7000] [send] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 00/0 : 12[7000] -> 8[7000] [receive] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Channel 01/0 : 8[7000] -> 5[46000] [send] via NET/AWS Libfabric/1
+gpub076:3268960:3269012 [0] NCCL INFO Connected all trees
+gpub076:3268960:3269012 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+gpub076:3268960:3269012 [0] NCCL INFO 2 coll channels, 2 p2p channels, 2 p2p channels per peer
+gpub076:3268960:3269012 [0] NCCL INFO comm 0x55f0b7620d20 rank 8 nranks 16 cudaDev 0 busId 7000 - Init COMPLETE
+[gpub010:0/16] 2024-02-10 11:58:55,244 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[gpub010:0/16] 2024-02-10 12:00:43,387 (trainer:762) INFO: 40epoch:train:1-100batch: iter_time=1.260, forward_time=0.398, loss_ctc=47.901, loss_att=39.871, acc=0.778, loss=42.280, backward_time=0.310, grad_norm=43.063, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.601e-04, train_time=4.855
+[gpub010:0/16] 2024-02-10 12:02:43,424 (trainer:762) INFO: 40epoch:train:101-200batch: iter_time=8.535e-05, forward_time=0.312, loss_ctc=44.696, loss_att=49.824, acc=0.743, loss=48.286, backward_time=0.318, grad_norm=52.937, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.601e-04, train_time=1.204
+[gpub010:0/16] 2024-02-10 12:04:53,197 (trainer:762) INFO: 40epoch:train:201-300batch: iter_time=8.491e-05, forward_time=0.330, loss_ctc=39.988, loss_att=35.494, acc=0.781, loss=36.842, backward_time=0.314, grad_norm=38.837, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.601e-04, train_time=1.298
+[gpub010:0/16] 2024-02-10 12:07:03,306 (trainer:762) INFO: 40epoch:train:301-400batch: iter_time=8.791e-05, forward_time=0.334, loss_ctc=57.195, loss_att=55.241, acc=0.736, loss=55.828, backward_time=0.309, grad_norm=56.273, clip=100.000, loss_scale=6.023e+33, optim_step_time=0.095, optim0_lr0=1.601e-04, train_time=1.301
+[gpub010:0/16] 2024-02-10 12:09:16,703 (trainer:762) INFO: 40epoch:train:401-500batch: iter_time=9.050e-05, forward_time=0.293, loss_ctc=51.568, loss_att=50.811, acc=0.741, loss=51.038, backward_time=0.297, grad_norm=82.978, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.601e-04, train_time=1.334
+[gpub010:0/16] 2024-02-10 12:11:21,822 (trainer:762) INFO: 40epoch:train:501-600batch: iter_time=9.397e-05, forward_time=0.352, loss_ctc=46.301, loss_att=44.629, acc=0.786, loss=45.131, backward_time=0.319, grad_norm=38.650, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.601e-04, train_time=1.251
+[gpub010:0/16] 2024-02-10 12:13:26,807 (trainer:762) INFO: 40epoch:train:601-700batch: iter_time=2.602e-04, forward_time=0.346, loss_ctc=50.860, loss_att=49.494, acc=0.761, loss=49.904, backward_time=0.318, grad_norm=49.514, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.601e-04, train_time=1.250
+[gpub010:0/16] 2024-02-10 12:15:24,808 (trainer:762) INFO: 40epoch:train:701-800batch: iter_time=9.262e-05, forward_time=0.321, loss_ctc=45.788, loss_att=45.309, acc=0.763, loss=45.453, backward_time=0.305, grad_norm=43.529, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.600e-04, train_time=1.180
+[gpub010:0/16] 2024-02-10 12:17:56,775 (trainer:762) INFO: 40epoch:train:801-900batch: iter_time=9.242e-05, forward_time=0.318, loss_ctc=45.886, loss_att=41.052, acc=0.769, loss=42.502, backward_time=0.309, grad_norm=40.589, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.600e-04, train_time=1.520
+[gpub010:0/16] 2024-02-10 12:20:02,489 (trainer:762) INFO: 40epoch:train:901-1000batch: iter_time=2.359e-04, forward_time=0.330, loss_ctc=54.152, loss_att=54.472, acc=0.743, loss=54.376, backward_time=0.314, grad_norm=56.519, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.109, optim0_lr0=1.600e-04, train_time=1.257
+[gpub010:0/16] 2024-02-10 12:22:25,406 (trainer:762) INFO: 40epoch:train:1001-1100batch: iter_time=8.610e-05, forward_time=0.296, loss_ctc=51.097, loss_att=48.635, acc=0.758, loss=49.374, backward_time=0.300, grad_norm=49.692, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.600e-04, train_time=1.429
+[gpub010:0/16] 2024-02-10 12:24:37,323 (trainer:762) INFO: 40epoch:train:1101-1200batch: iter_time=8.466e-05, forward_time=0.347, loss_ctc=49.612, loss_att=47.865, acc=0.759, loss=48.389, backward_time=0.311, grad_norm=43.913, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.600e-04, train_time=1.318
+[gpub010:0/16] 2024-02-10 12:26:02,634 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-10 12:26:21,405 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 12:26:24,841 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc216ac6950>)
+[gpub010:0/16] 2024-02-10 12:26:24,841 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-10 12:26:24,889 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 12:32:54,774 (trainer:762) INFO: 40epoch:train:1201-1300batch: iter_time=3.673, forward_time=0.365, loss_ctc=43.073, loss_att=40.587, acc=0.772, loss=41.333, backward_time=0.307, grad_norm=40.794, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.600e-04, train_time=4.974
+[gpub010:0/16] 2024-02-10 12:34:55,715 (trainer:762) INFO: 40epoch:train:1301-1400batch: iter_time=2.625e-04, forward_time=0.309, loss_ctc=45.773, loss_att=47.790, acc=0.754, loss=47.185, backward_time=0.303, grad_norm=45.002, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.600e-04, train_time=1.210
+[gpub010:0/16] 2024-02-10 12:37:25,511 (trainer:762) INFO: 40epoch:train:1401-1500batch: iter_time=8.048e-05, forward_time=0.340, loss_ctc=41.423, loss_att=40.243, acc=0.755, loss=40.597, backward_time=0.348, grad_norm=48.436, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.599e-04, train_time=1.498
+[gpub010:0/16] 2024-02-10 12:39:20,998 (trainer:762) INFO: 40epoch:train:1501-1600batch: iter_time=8.306e-05, forward_time=0.288, loss_ctc=42.795, loss_att=40.155, acc=0.761, loss=40.947, backward_time=0.295, grad_norm=41.239, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.599e-04, train_time=1.155
+[gpub010:0/16] 2024-02-10 12:41:41,703 (trainer:762) INFO: 40epoch:train:1601-1700batch: iter_time=2.731e-04, forward_time=0.379, loss_ctc=48.970, loss_att=51.471, acc=0.746, loss=50.721, backward_time=0.321, grad_norm=51.183, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.599e-04, train_time=1.406
+[gpub010:0/16] 2024-02-10 12:43:47,392 (trainer:762) INFO: 40epoch:train:1701-1800batch: iter_time=8.992e-05, forward_time=0.290, loss_ctc=53.638, loss_att=47.475, acc=0.748, loss=49.324, backward_time=0.297, grad_norm=48.194, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.599e-04, train_time=1.258
+[gpub010:0/16] 2024-02-10 12:46:14,131 (trainer:762) INFO: 40epoch:train:1801-1900batch: iter_time=8.856e-05, forward_time=0.361, loss_ctc=49.602, loss_att=48.341, acc=0.766, loss=48.719, backward_time=0.349, grad_norm=43.906, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.599e-04, train_time=1.468
+[gpub010:0/16] 2024-02-10 12:47:15,171 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 12:48:12,268 (trainer:762) INFO: 40epoch:train:1901-2000batch: iter_time=8.882e-05, forward_time=0.290, loss_ctc=43.860, loss_att=42.668, acc=0.768, loss=43.025, backward_time=0.296, grad_norm=37.385, clip=100.000, loss_scale=7.762e+33, optim_step_time=0.094, optim0_lr0=1.599e-04, train_time=1.181
+[gpub010:0/16] 2024-02-10 12:50:18,148 (trainer:762) INFO: 40epoch:train:2001-2100batch: iter_time=8.596e-05, forward_time=0.290, loss_ctc=44.531, loss_att=39.965, acc=0.770, loss=41.334, backward_time=0.296, grad_norm=42.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.599e-04, train_time=1.259
+[gpub010:0/16] 2024-02-10 12:52:32,393 (trainer:762) INFO: 40epoch:train:2101-2200batch: iter_time=7.192e-04, forward_time=0.391, loss_ctc=47.893, loss_att=45.931, acc=0.758, loss=46.519, backward_time=0.337, grad_norm=44.277, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.599e-04, train_time=1.343
+[gpub010:0/16] 2024-02-10 12:54:43,724 (trainer:762) INFO: 40epoch:train:2201-2300batch: iter_time=8.314e-05, forward_time=0.290, loss_ctc=51.415, loss_att=49.611, acc=0.739, loss=50.152, backward_time=0.294, grad_norm=56.291, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.598e-04, train_time=1.312
+[gpub010:0/16] 2024-02-10 12:56:55,187 (trainer:762) INFO: 40epoch:train:2301-2400batch: iter_time=1.920e-04, forward_time=0.401, loss_ctc=51.454, loss_att=51.156, acc=0.753, loss=51.246, backward_time=0.337, grad_norm=44.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.598e-04, train_time=1.315
+[gpub010:0/16] 2024-02-10 12:59:02,167 (trainer:762) INFO: 40epoch:train:2401-2500batch: iter_time=8.054e-05, forward_time=0.291, loss_ctc=43.977, loss_att=44.243, acc=0.766, loss=44.163, backward_time=0.296, grad_norm=41.113, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.270
+[gpub010:0/16] 2024-02-10 12:59:22,233 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-10 12:59:41,691 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 12:59:45,101 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc0a1f75ab0>)
+[gpub010:0/16] 2024-02-10 12:59:45,101 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-10 12:59:45,143 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 13:06:42,136 (trainer:762) INFO: 40epoch:train:2501-2600batch: iter_time=3.083, forward_time=0.289, loss_ctc=45.848, loss_att=37.571, acc=0.778, loss=40.054, backward_time=0.294, grad_norm=41.291, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.598e-04, train_time=4.599
+[gpub010:0/16] 2024-02-10 13:09:14,573 (trainer:762) INFO: 40epoch:train:2601-2700batch: iter_time=7.959e-05, forward_time=0.406, loss_ctc=42.702, loss_att=45.976, acc=0.742, loss=44.994, backward_time=0.322, grad_norm=47.259, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.598e-04, train_time=1.524
+[gpub010:0/16] 2024-02-10 13:11:46,961 (trainer:762) INFO: 40epoch:train:2701-2800batch: iter_time=8.438e-05, forward_time=0.288, loss_ctc=39.356, loss_att=34.477, acc=0.782, loss=35.941, backward_time=0.290, grad_norm=37.401, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.524
+[gpub010:0/16] 2024-02-10 13:14:06,104 (trainer:762) INFO: 40epoch:train:2801-2900batch: iter_time=8.348e-05, forward_time=0.290, loss_ctc=53.434, loss_att=54.241, acc=0.733, loss=53.999, backward_time=0.295, grad_norm=52.235, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.598e-04, train_time=1.391
+[gpub010:0/16] 2024-02-10 13:16:37,986 (trainer:762) INFO: 40epoch:train:2901-3000batch: iter_time=2.428e-04, forward_time=0.403, loss_ctc=49.202, loss_att=47.599, acc=0.744, loss=48.080, backward_time=0.334, grad_norm=49.100, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.597e-04, train_time=1.518
+[gpub010:0/16] 2024-02-10 13:18:51,861 (trainer:762) INFO: 40epoch:train:3001-3100batch: iter_time=8.354e-05, forward_time=0.314, loss_ctc=45.527, loss_att=44.114, acc=0.781, loss=44.538, backward_time=0.297, grad_norm=39.483, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.340
+[gpub010:0/16] 2024-02-10 13:21:24,722 (trainer:762) INFO: 40epoch:train:3101-3200batch: iter_time=7.915e-05, forward_time=0.386, loss_ctc=49.102, loss_att=47.347, acc=0.760, loss=47.873, backward_time=0.340, grad_norm=44.324, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.597e-04, train_time=1.528
+[gpub010:0/16] 2024-02-10 13:23:50,889 (trainer:762) INFO: 40epoch:train:3201-3300batch: iter_time=8.745e-05, forward_time=0.291, loss_ctc=44.650, loss_att=44.041, acc=0.764, loss=44.224, backward_time=0.295, grad_norm=41.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.461
+[gpub010:0/16] 2024-02-10 13:26:01,543 (trainer:762) INFO: 40epoch:train:3301-3400batch: iter_time=8.466e-05, forward_time=0.290, loss_ctc=43.888, loss_att=39.398, acc=0.766, loss=40.745, backward_time=0.294, grad_norm=39.708, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.307
+[gpub010:0/16] 2024-02-10 13:28:13,243 (trainer:762) INFO: 40epoch:train:3401-3500batch: iter_time=1.630e-04, forward_time=0.317, loss_ctc=53.023, loss_att=52.142, acc=0.746, loss=52.406, backward_time=0.323, grad_norm=58.348, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.597e-04, train_time=1.317
+[gpub010:0/16] 2024-02-10 13:30:35,373 (trainer:762) INFO: 40epoch:train:3501-3600batch: iter_time=8.684e-05, forward_time=0.334, loss_ctc=50.079, loss_att=47.030, acc=0.751, loss=47.945, backward_time=0.306, grad_norm=47.320, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.597e-04, train_time=1.421
+[gpub010:0/16] 2024-02-10 13:32:46,610 (trainer:762) INFO: 40epoch:train:3601-3700batch: iter_time=8.294e-05, forward_time=0.293, loss_ctc=48.389, loss_att=46.672, acc=0.764, loss=47.187, backward_time=0.298, grad_norm=43.480, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.597e-04, train_time=1.311
+[gpub010:0/16] 2024-02-10 13:34:19,725 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-10 13:34:38,520 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 13:34:41,885 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbb9778f7f0>)
+[gpub010:0/16] 2024-02-10 13:34:41,885 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-10 13:34:41,942 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 13:44:06,410 (trainer:762) INFO: 40epoch:train:3701-3800batch: iter_time=3.316, forward_time=0.345, loss_ctc=42.132, loss_att=39.308, acc=0.773, loss=40.155, backward_time=0.335, grad_norm=42.612, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.596e-04, train_time=6.799
+[gpub010:0/16] 2024-02-10 13:46:35,787 (trainer:762) INFO: 40epoch:train:3801-3900batch: iter_time=8.472e-05, forward_time=0.292, loss_ctc=46.106, loss_att=46.493, acc=0.759, loss=46.377, backward_time=0.295, grad_norm=44.336, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.494
+[gpub010:0/16] 2024-02-10 13:48:36,105 (trainer:762) INFO: 40epoch:train:3901-4000batch: iter_time=8.231e-05, forward_time=0.289, loss_ctc=41.261, loss_att=39.442, acc=0.757, loss=39.988, backward_time=0.294, grad_norm=46.186, clip=100.000, loss_scale=7.788e+33, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.203
+[gpub010:0/16] 2024-02-10 13:50:29,817 (trainer:762) INFO: 40epoch:train:4001-4100batch: iter_time=8.879e-05, forward_time=0.290, loss_ctc=42.557, loss_att=39.819, acc=0.762, loss=40.640, backward_time=0.295, grad_norm=40.873, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.136
+[gpub010:0/16] 2024-02-10 13:52:51,995 (trainer:762) INFO: 40epoch:train:4101-4200batch: iter_time=1.596e-04, forward_time=0.377, loss_ctc=48.177, loss_att=50.963, acc=0.749, loss=50.127, backward_time=0.321, grad_norm=50.511, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.103, optim0_lr0=1.596e-04, train_time=1.422
+[gpub010:0/16] 2024-02-10 13:55:07,393 (trainer:762) INFO: 40epoch:train:4201-4300batch: iter_time=8.163e-05, forward_time=0.291, loss_ctc=52.233, loss_att=46.562, acc=0.752, loss=48.264, backward_time=0.296, grad_norm=46.136, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.354
+[gpub010:0/16] 2024-02-10 13:57:25,906 (trainer:762) INFO: 40epoch:train:4301-4400batch: iter_time=8.091e-05, forward_time=0.292, loss_ctc=48.809, loss_att=47.418, acc=0.770, loss=47.835, backward_time=0.295, grad_norm=45.091, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.596e-04, train_time=1.385
+[gpub010:0/16] 2024-02-10 13:59:37,022 (trainer:762) INFO: 40epoch:train:4401-4500batch: iter_time=8.286e-05, forward_time=0.291, loss_ctc=43.374, loss_att=42.115, acc=0.771, loss=42.493, backward_time=0.296, grad_norm=38.498, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=1.311
+[gpub010:0/16] 2024-02-10 14:01:55,072 (trainer:762) INFO: 40epoch:train:4501-4600batch: iter_time=8.672e-05, forward_time=0.394, loss_ctc=43.859, loss_att=39.505, acc=0.772, loss=40.811, backward_time=0.323, grad_norm=39.214, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.595e-04, train_time=1.379
+[gpub010:0/16] 2024-02-10 14:04:06,467 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 14:04:09,695 (trainer:762) INFO: 40epoch:train:4601-4700batch: iter_time=8.248e-05, forward_time=0.292, loss_ctc=47.303, loss_att=45.724, acc=0.760, loss=46.197, backward_time=0.296, grad_norm=46.321, clip=100.000, loss_scale=1.023e+34, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=1.347
+[gpub010:0/16] 2024-02-10 14:06:12,638 (trainer:762) INFO: 40epoch:train:4701-4800batch: iter_time=7.951e-05, forward_time=0.290, loss_ctc=51.608, loss_att=48.284, acc=0.741, loss=49.282, backward_time=0.295, grad_norm=55.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=1.229
+[gpub010:0/16] 2024-02-10 14:08:28,734 (trainer:762) INFO: 40epoch:train:4801-4900batch: iter_time=8.004e-05, forward_time=0.293, loss_ctc=51.254, loss_att=50.069, acc=0.756, loss=50.425, backward_time=0.299, grad_norm=46.201, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=1.361
+[gpub010:0/16] 2024-02-10 14:10:47,668 (trainer:762) INFO: 40epoch:train:4901-5000batch: iter_time=3.072e-04, forward_time=0.386, loss_ctc=43.229, loss_att=44.113, acc=0.768, loss=43.848, backward_time=0.318, grad_norm=40.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.595e-04, train_time=1.389
+[gpub010:0/16] 2024-02-10 14:11:07,735 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-10 14:11:26,345 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 14:11:29,757 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbb9a4b2fb0>)
+[gpub010:0/16] 2024-02-10 14:11:29,757 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-10 14:11:29,785 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 14:18:29,852 (trainer:762) INFO: 40epoch:train:5001-5100batch: iter_time=3.351, forward_time=0.290, loss_ctc=44.477, loss_att=38.680, acc=0.785, loss=40.419, backward_time=0.296, grad_norm=41.907, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.595e-04, train_time=4.621
+[gpub010:0/16] 2024-02-10 14:20:41,022 (trainer:762) INFO: 40epoch:train:5101-5200batch: iter_time=8.095e-05, forward_time=0.291, loss_ctc=42.272, loss_att=46.753, acc=0.755, loss=45.409, backward_time=0.295, grad_norm=48.111, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.312
+[gpub010:0/16] 2024-02-10 14:23:03,568 (trainer:762) INFO: 40epoch:train:5201-5300batch: iter_time=2.481e-04, forward_time=0.350, loss_ctc=39.160, loss_att=34.808, acc=0.789, loss=36.114, backward_time=0.316, grad_norm=38.199, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.594e-04, train_time=1.425
+[gpub010:0/16] 2024-02-10 14:25:09,635 (trainer:762) INFO: 40epoch:train:5301-5400batch: iter_time=8.953e-05, forward_time=0.292, loss_ctc=52.434, loss_att=53.356, acc=0.745, loss=53.079, backward_time=0.297, grad_norm=54.717, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.260
+[gpub010:0/16] 2024-02-10 14:27:16,433 (trainer:762) INFO: 40epoch:train:5401-5500batch: iter_time=8.393e-05, forward_time=0.292, loss_ctc=47.941, loss_att=48.895, acc=0.749, loss=48.609, backward_time=0.295, grad_norm=46.488, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.268
+[gpub010:0/16] 2024-02-10 14:29:30,518 (trainer:762) INFO: 40epoch:train:5501-5600batch: iter_time=8.202e-05, forward_time=0.363, loss_ctc=44.999, loss_att=43.800, acc=0.791, loss=44.160, backward_time=0.323, grad_norm=38.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.594e-04, train_time=1.341
+[gpub010:0/16] 2024-02-10 14:31:55,157 (trainer:762) INFO: 40epoch:train:5601-5700batch: iter_time=8.560e-05, forward_time=0.292, loss_ctc=48.878, loss_att=48.577, acc=0.767, loss=48.667, backward_time=0.295, grad_norm=47.269, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.594e-04, train_time=1.446
+[gpub010:0/16] 2024-02-10 14:33:55,703 (trainer:762) INFO: 40epoch:train:5701-5800batch: iter_time=8.530e-05, forward_time=0.292, loss_ctc=44.221, loss_att=44.004, acc=0.769, loss=44.069, backward_time=0.297, grad_norm=39.815, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.594e-04, train_time=1.205
+[gpub010:0/16] 2024-02-10 14:36:18,145 (trainer:762) INFO: 40epoch:train:5801-5900batch: iter_time=2.072e-04, forward_time=0.377, loss_ctc=43.578, loss_att=40.294, acc=0.775, loss=41.279, backward_time=0.343, grad_norm=39.532, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.594e-04, train_time=1.408
+[gpub010:0/16] 2024-02-10 14:40:23,959 (trainer:762) INFO: 40epoch:train:5901-6000batch: iter_time=0.145, forward_time=0.318, loss_ctc=52.603, loss_att=52.393, acc=0.748, loss=52.456, backward_time=0.305, grad_norm=60.238, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.593e-04, train_time=2.463
+[gpub010:0/16] 2024-02-10 14:42:54,148 (trainer:762) INFO: 40epoch:train:6001-6100batch: iter_time=0.063, forward_time=0.292, loss_ctc=50.086, loss_att=48.203, acc=0.762, loss=48.768, backward_time=0.296, grad_norm=45.856, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.513
+[gpub010:0/16] 2024-02-10 14:45:11,280 (trainer:762) INFO: 40epoch:train:6101-6200batch: iter_time=7.809e-05, forward_time=0.319, loss_ctc=47.861, loss_att=46.481, acc=0.766, loss=46.895, backward_time=0.299, grad_norm=40.735, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.593e-04, train_time=1.371
+[gpub010:0/16] 2024-02-10 14:46:37,576 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-10 14:46:56,122 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 14:46:59,951 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbb9774cf40>)
+[gpub010:0/16] 2024-02-10 14:46:59,951 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-10 14:47:00,021 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 14:53:00,994 (trainer:762) INFO: 40epoch:train:6201-6300batch: iter_time=3.342, forward_time=0.360, loss_ctc=41.527, loss_att=39.439, acc=0.781, loss=40.065, backward_time=0.317, grad_norm=40.996, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.593e-04, train_time=4.697
+[gpub010:0/16] 2024-02-10 14:55:03,522 (trainer:762) INFO: 40epoch:train:6301-6400batch: iter_time=8.387e-05, forward_time=0.292, loss_ctc=45.072, loss_att=45.905, acc=0.777, loss=45.655, backward_time=0.299, grad_norm=42.357, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.225
+[gpub010:0/16] 2024-02-10 14:57:05,966 (trainer:762) INFO: 40epoch:train:6401-6500batch: iter_time=8.277e-05, forward_time=0.290, loss_ctc=40.915, loss_att=40.981, acc=0.763, loss=40.961, backward_time=0.294, grad_norm=46.945, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.593e-04, train_time=1.224
+[gpub010:0/16] 2024-02-10 14:59:18,285 (trainer:762) INFO: 40epoch:train:6501-6600batch: iter_time=8.679e-05, forward_time=0.381, loss_ctc=42.181, loss_att=39.467, acc=0.770, loss=40.281, backward_time=0.322, grad_norm=41.552, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.593e-04, train_time=1.323
+[gpub010:0/16] 2024-02-10 15:01:21,439 (trainer:762) INFO: 40epoch:train:6601-6700batch: iter_time=9.148e-05, forward_time=0.290, loss_ctc=47.045, loss_att=52.893, acc=0.754, loss=51.139, backward_time=0.296, grad_norm=48.877, clip=100.000, loss_scale=5.348e+33, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.231
+[gpub010:0/16] 2024-02-10 15:03:20,121 (trainer:762) INFO: 40epoch:train:6701-6800batch: iter_time=8.964e-05, forward_time=0.292, loss_ctc=51.554, loss_att=46.123, acc=0.761, loss=47.752, backward_time=0.297, grad_norm=47.487, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.187
+[gpub010:0/16] 2024-02-10 15:05:16,244 (trainer:762) INFO: 40epoch:train:6801-6900batch: iter_time=8.359e-05, forward_time=0.334, loss_ctc=48.164, loss_att=46.949, acc=0.778, loss=47.313, backward_time=0.322, grad_norm=47.786, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.592e-04, train_time=1.161
+[gpub010:0/16] 2024-02-10 15:07:33,325 (trainer:762) INFO: 40epoch:train:6901-7000batch: iter_time=8.453e-05, forward_time=0.320, loss_ctc=43.582, loss_att=43.974, acc=0.775, loss=43.857, backward_time=0.301, grad_norm=38.080, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.370
+[gpub010:0/16] 2024-02-10 15:09:34,843 (trainer:762) INFO: 40epoch:train:7001-7100batch: iter_time=8.269e-05, forward_time=0.314, loss_ctc=43.814, loss_att=40.363, acc=0.777, loss=41.398, backward_time=0.297, grad_norm=40.037, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.215
+[gpub010:0/16] 2024-02-10 15:11:29,647 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 15:11:30,846 (trainer:762) INFO: 40epoch:train:7101-7200batch: iter_time=3.964e-04, forward_time=0.327, loss_ctc=46.801, loss_att=46.053, acc=0.766, loss=46.278, backward_time=0.328, grad_norm=41.385, clip=100.000, loss_scale=1.033e+34, optim_step_time=0.098, optim0_lr0=1.592e-04, train_time=1.160
+[gpub010:0/16] 2024-02-10 15:14:01,321 (trainer:762) INFO: 40epoch:train:7201-7300batch: iter_time=9.659e-05, forward_time=0.319, loss_ctc=50.113, loss_att=49.713, acc=0.751, loss=49.833, backward_time=0.305, grad_norm=57.163, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.592e-04, train_time=1.505
+[gpub010:0/16] 2024-02-10 15:15:57,965 (trainer:762) INFO: 40epoch:train:7301-7400batch: iter_time=9.279e-05, forward_time=0.296, loss_ctc=50.810, loss_att=50.470, acc=0.762, loss=50.572, backward_time=0.302, grad_norm=46.395, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.592e-04, train_time=1.166
+[gpub010:0/16] 2024-02-10 15:18:08,372 (trainer:762) INFO: 40epoch:train:7401-7500batch: iter_time=8.715e-05, forward_time=0.297, loss_ctc=42.877, loss_att=44.479, acc=0.771, loss=43.999, backward_time=0.300, grad_norm=41.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.591e-04, train_time=1.304
+[gpub010:0/16] 2024-02-10 15:18:28,430 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-10 15:18:47,367 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 15:18:50,923 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbbf9c991b0>)
+[gpub010:0/16] 2024-02-10 15:18:50,923 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-10 15:18:50,962 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 15:27:14,599 (trainer:762) INFO: 40epoch:train:7501-7600batch: iter_time=3.722, forward_time=0.382, loss_ctc=44.289, loss_att=36.895, acc=0.792, loss=39.113, backward_time=0.332, grad_norm=39.732, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.591e-04, train_time=5.462
+[gpub010:0/16] 2024-02-10 15:29:03,248 (trainer:762) INFO: 40epoch:train:7601-7700batch: iter_time=8.307e-05, forward_time=0.292, loss_ctc=42.205, loss_att=44.701, acc=0.763, loss=43.952, backward_time=0.300, grad_norm=48.562, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.591e-04, train_time=1.086
+[gpub010:0/16] 2024-02-10 15:31:47,986 (trainer:762) INFO: 40epoch:train:7701-7800batch: iter_time=0.090, forward_time=0.437, loss_ctc=38.840, loss_att=34.142, acc=0.791, loss=35.551, backward_time=0.319, grad_norm=36.524, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.108, optim0_lr0=1.591e-04, train_time=1.647
+[gpub010:0/16] 2024-02-10 15:34:23,380 (trainer:762) INFO: 40epoch:train:7801-7900batch: iter_time=0.057, forward_time=0.379, loss_ctc=52.483, loss_att=53.233, acc=0.747, loss=53.008, backward_time=0.327, grad_norm=52.529, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.591e-04, train_time=1.554
+[gpub010:0/16] 2024-02-10 15:37:03,248 (trainer:762) INFO: 40epoch:train:7901-8000batch: iter_time=0.170, forward_time=0.343, loss_ctc=47.881, loss_att=49.544, acc=0.747, loss=49.045, backward_time=0.304, grad_norm=46.938, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.591e-04, train_time=1.598
+[gpub010:0/16] 2024-02-10 15:40:03,127 (trainer:762) INFO: 40epoch:train:8001-8100batch: iter_time=8.422e-05, forward_time=0.469, loss_ctc=44.988, loss_att=43.473, acc=0.791, loss=43.928, backward_time=0.456, grad_norm=38.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.591e-04, train_time=1.799
+[gpub010:0/16] 2024-02-10 15:42:25,163 (trainer:762) INFO: 40epoch:train:8101-8200batch: iter_time=8.845e-05, forward_time=0.319, loss_ctc=48.932, loss_att=48.941, acc=0.766, loss=48.938, backward_time=0.305, grad_norm=43.673, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.590e-04, train_time=1.420
+[gpub010:0/16] 2024-02-10 15:44:22,517 (trainer:762) INFO: 40epoch:train:8201-8300batch: iter_time=8.670e-05, forward_time=0.329, loss_ctc=43.758, loss_att=43.289, acc=0.772, loss=43.430, backward_time=0.321, grad_norm=40.936, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.590e-04, train_time=1.173
+[gpub010:0/16] 2024-02-10 15:46:28,002 (trainer:762) INFO: 40epoch:train:8301-8400batch: iter_time=8.886e-05, forward_time=0.336, loss_ctc=43.223, loss_att=39.455, acc=0.777, loss=40.585, backward_time=0.305, grad_norm=39.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.590e-04, train_time=1.255
+[gpub010:0/16] 2024-02-10 15:49:24,549 (trainer:762) INFO: 40epoch:train:8401-8500batch: iter_time=2.120e-04, forward_time=0.363, loss_ctc=53.401, loss_att=52.658, acc=0.749, loss=52.881, backward_time=0.354, grad_norm=60.254, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.590e-04, train_time=1.764
+[gpub010:0/16] 2024-02-10 15:52:17,508 (trainer:762) INFO: 40epoch:train:8501-8600batch: iter_time=0.316, forward_time=0.301, loss_ctc=49.646, loss_att=47.276, acc=0.766, loss=47.987, backward_time=0.300, grad_norm=46.742, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.590e-04, train_time=1.730
+[gpub010:0/16] 2024-02-10 15:54:32,854 (trainer:762) INFO: 40epoch:train:8601-8700batch: iter_time=3.897e-04, forward_time=0.401, loss_ctc=47.556, loss_att=46.210, acc=0.768, loss=46.614, backward_time=0.333, grad_norm=41.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.590e-04, train_time=1.352
+[gpub010:0/16] 2024-02-10 15:55:51,818 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-10 15:56:11,111 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 15:56:14,657 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbc2f8e4ee0>)
+[gpub010:0/16] 2024-02-10 15:56:14,657 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-10 15:56:14,828 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 16:05:47,719 (trainer:762) INFO: 40epoch:train:8701-8800batch: iter_time=4.150, forward_time=0.358, loss_ctc=41.398, loss_att=39.924, acc=0.779, loss=40.366, backward_time=0.316, grad_norm=40.450, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.590e-04, train_time=6.749
+[gpub010:0/16] 2024-02-10 16:08:18,402 (trainer:762) INFO: 40epoch:train:8801-8900batch: iter_time=1.828e-04, forward_time=0.395, loss_ctc=45.110, loss_att=48.132, acc=0.758, loss=47.226, backward_time=0.340, grad_norm=44.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.590e-04, train_time=1.506
+[gpub010:0/16] 2024-02-10 16:10:47,719 (trainer:762) INFO: 40epoch:train:8901-9000batch: iter_time=7.883e-05, forward_time=0.357, loss_ctc=40.597, loss_att=39.638, acc=0.758, loss=39.926, backward_time=0.329, grad_norm=46.671, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.589e-04, train_time=1.492
+[gpub010:0/16] 2024-02-10 16:14:02,731 (trainer:762) INFO: 40epoch:train:9001-9100batch: iter_time=0.178, forward_time=0.487, loss_ctc=41.837, loss_att=39.321, acc=0.766, loss=40.076, backward_time=0.332, grad_norm=40.151, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.589e-04, train_time=1.950
+[gpub010:0/16] 2024-02-10 16:16:38,442 (trainer:762) INFO: 40epoch:train:9101-9200batch: iter_time=0.001, forward_time=0.349, loss_ctc=47.099, loss_att=51.183, acc=0.750, loss=49.957, backward_time=0.383, grad_norm=49.672, clip=100.000, loss_scale=5.244e+33, optim_step_time=0.099, optim0_lr0=1.589e-04, train_time=1.556
+[gpub010:0/16] 2024-02-10 16:19:15,352 (trainer:762) INFO: 40epoch:train:9201-9300batch: iter_time=3.146e-04, forward_time=0.422, loss_ctc=51.649, loss_att=46.330, acc=0.754, loss=47.926, backward_time=0.349, grad_norm=45.224, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.105, optim0_lr0=1.589e-04, train_time=1.569
+[gpub010:0/16] 2024-02-10 16:20:17,978 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 16:21:56,755 (trainer:762) INFO: 40epoch:train:9301-9400batch: iter_time=2.430e-04, forward_time=0.417, loss_ctc=48.893, loss_att=47.952, acc=0.771, loss=48.234, backward_time=0.370, grad_norm=46.859, clip=100.000, loss_scale=7.500e+33, optim_step_time=0.105, optim0_lr0=1.589e-04, train_time=1.614
+[gpub010:0/16] 2024-02-10 16:24:14,382 (trainer:762) INFO: 40epoch:train:9401-9500batch: iter_time=4.430e-04, forward_time=0.362, loss_ctc=43.546, loss_att=42.473, acc=0.771, loss=42.795, backward_time=0.348, grad_norm=37.420, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.589e-04, train_time=1.376
+[gpub010:0/16] 2024-02-10 16:26:27,585 (trainer:762) INFO: 40epoch:train:9501-9600batch: iter_time=1.857e-04, forward_time=0.364, loss_ctc=43.165, loss_att=39.572, acc=0.775, loss=40.650, backward_time=0.334, grad_norm=39.798, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.589e-04, train_time=1.331
+[gpub010:0/16] 2024-02-10 16:28:56,087 (trainer:762) INFO: 40epoch:train:9601-9700batch: iter_time=2.666e-04, forward_time=0.362, loss_ctc=46.256, loss_att=45.383, acc=0.762, loss=45.645, backward_time=0.331, grad_norm=43.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.588e-04, train_time=1.486
+[gpub010:0/16] 2024-02-10 16:31:04,930 (trainer:762) INFO: 40epoch:train:9701-9800batch: iter_time=0.001, forward_time=0.380, loss_ctc=49.120, loss_att=48.999, acc=0.742, loss=49.035, backward_time=0.347, grad_norm=60.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.588e-04, train_time=1.287
+[gpub010:0/16] 2024-02-10 16:33:39,061 (trainer:762) INFO: 40epoch:train:9801-9900batch: iter_time=9.320e-05, forward_time=0.446, loss_ctc=50.884, loss_att=50.389, acc=0.756, loss=50.537, backward_time=0.392, grad_norm=44.412, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.588e-04, train_time=1.542
+[gpub010:0/16] 2024-02-10 16:36:09,527 (trainer:762) INFO: 40epoch:train:9901-10000batch: iter_time=1.754e-04, forward_time=0.353, loss_ctc=42.773, loss_att=43.924, acc=0.770, loss=43.579, backward_time=0.338, grad_norm=42.171, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.588e-04, train_time=1.505
+[gpub010:0/16] 2024-02-10 16:36:29,786 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-10 16:36:48,847 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 16:36:52,360 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb22a6677c0>)
+[gpub010:0/16] 2024-02-10 16:36:52,360 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-10 16:36:52,402 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 16:47:20,178 (trainer:762) INFO: 40epoch:train:10001-10100batch: iter_time=5.478, forward_time=0.370, loss_ctc=44.174, loss_att=37.635, acc=0.790, loss=39.596, backward_time=0.313, grad_norm=40.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.588e-04, train_time=6.707
+[gpub010:0/16] 2024-02-10 16:50:00,581 (trainer:762) INFO: 40epoch:train:10101-10200batch: iter_time=9.127e-05, forward_time=0.472, loss_ctc=41.909, loss_att=44.989, acc=0.763, loss=44.065, backward_time=0.351, grad_norm=48.625, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.588e-04, train_time=1.603
+[gpub010:0/16] 2024-02-10 16:52:20,195 (trainer:762) INFO: 40epoch:train:10201-10300batch: iter_time=3.442e-04, forward_time=0.343, loss_ctc=38.641, loss_att=34.295, acc=0.790, loss=35.599, backward_time=0.331, grad_norm=36.113, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.588e-04, train_time=1.396
+[gpub010:0/16] 2024-02-10 16:54:27,011 (trainer:762) INFO: 40epoch:train:10301-10400batch: iter_time=2.526e-04, forward_time=0.315, loss_ctc=51.655, loss_att=52.621, acc=0.749, loss=52.331, backward_time=0.298, grad_norm=52.480, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.587e-04, train_time=1.268
+[gpub010:0/16] 2024-02-10 16:56:56,648 (trainer:762) INFO: 40epoch:train:10401-10500batch: iter_time=8.835e-05, forward_time=0.361, loss_ctc=47.462, loss_att=49.427, acc=0.748, loss=48.837, backward_time=0.325, grad_norm=46.564, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.587e-04, train_time=1.496
+[gpub010:0/16] 2024-02-10 16:59:15,198 (trainer:762) INFO: 40epoch:train:10501-10600batch: iter_time=2.003e-04, forward_time=0.366, loss_ctc=45.535, loss_att=43.643, acc=0.793, loss=44.211, backward_time=0.341, grad_norm=37.636, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.587e-04, train_time=1.385
+[gpub010:0/16] 2024-02-10 17:01:36,964 (trainer:762) INFO: 40epoch:train:10601-10700batch: iter_time=4.399e-04, forward_time=0.386, loss_ctc=48.143, loss_att=48.503, acc=0.767, loss=48.395, backward_time=0.343, grad_norm=45.643, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.587e-04, train_time=1.417
+[gpub010:0/16] 2024-02-10 17:03:58,965 (trainer:762) INFO: 40epoch:train:10701-10800batch: iter_time=3.182e-04, forward_time=0.356, loss_ctc=43.553, loss_att=42.817, acc=0.774, loss=43.038, backward_time=0.345, grad_norm=40.799, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.587e-04, train_time=1.420
+[gpub010:0/16] 2024-02-10 17:06:19,688 (trainer:762) INFO: 40epoch:train:10801-10900batch: iter_time=4.629e-04, forward_time=0.375, loss_ctc=42.799, loss_att=39.343, acc=0.777, loss=40.380, backward_time=0.354, grad_norm=38.152, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.587e-04, train_time=1.406
+[gpub010:0/16] 2024-02-10 17:08:42,205 (trainer:762) INFO: 40epoch:train:10901-11000batch: iter_time=4.537e-04, forward_time=0.383, loss_ctc=52.013, loss_att=54.351, acc=0.750, loss=53.650, backward_time=0.338, grad_norm=55.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.587e-04, train_time=1.426
+[gpub010:0/16] 2024-02-10 17:11:03,910 (trainer:762) INFO: 40epoch:train:11001-11100batch: iter_time=1.475e-04, forward_time=0.395, loss_ctc=49.129, loss_att=47.418, acc=0.765, loss=47.931, backward_time=0.357, grad_norm=47.058, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.587e-04, train_time=1.416
+[gpub010:0/16] 2024-02-10 17:13:19,822 (trainer:762) INFO: 40epoch:train:11101-11200batch: iter_time=1.864e-04, forward_time=0.410, loss_ctc=47.824, loss_att=47.038, acc=0.766, loss=47.274, backward_time=0.321, grad_norm=43.124, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.586e-04, train_time=1.360
+[gpub010:0/16] 2024-02-10 17:14:46,077 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-10 17:15:05,618 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 17:15:09,443 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1c2da2770>)
+[gpub010:0/16] 2024-02-10 17:15:09,443 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-10 17:15:09,446 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 17:22:11,986 (trainer:762) INFO: 40epoch:train:11201-11300batch: iter_time=3.258, forward_time=0.368, loss_ctc=41.236, loss_att=39.453, acc=0.778, loss=39.988, backward_time=0.312, grad_norm=41.067, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.111, optim0_lr0=1.586e-04, train_time=5.321
+[gpub010:0/16] 2024-02-10 17:26:24,944 (trainer:762) INFO: 40epoch:train:11301-11400batch: iter_time=8.216e-05, forward_time=0.311, loss_ctc=45.011, loss_att=47.261, acc=0.761, loss=46.586, backward_time=0.302, grad_norm=43.842, clip=100.000, loss_scale=8.048e+33, optim_step_time=0.094, optim0_lr0=1.586e-04, train_time=2.529
+[gpub010:0/16] 2024-02-10 17:28:33,376 (trainer:762) INFO: 40epoch:train:11401-11500batch: iter_time=8.034e-05, forward_time=0.335, loss_ctc=40.544, loss_att=39.970, acc=0.759, loss=40.142, backward_time=0.340, grad_norm=46.728, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.586e-04, train_time=1.284
+[gpub010:0/16] 2024-02-10 17:30:42,433 (trainer:762) INFO: 40epoch:train:11501-11600batch: iter_time=7.929e-05, forward_time=0.308, loss_ctc=41.782, loss_att=39.182, acc=0.766, loss=39.962, backward_time=0.300, grad_norm=40.028, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.586e-04, train_time=1.290
+[gpub010:0/16] 2024-02-10 17:31:36,459 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 17:32:32,172 (trainer:762) INFO: 40epoch:train:11601-11700batch: iter_time=8.085e-05, forward_time=0.295, loss_ctc=46.717, loss_att=51.090, acc=0.752, loss=49.778, backward_time=0.304, grad_norm=48.991, clip=100.000, loss_scale=7.762e+33, optim_step_time=0.095, optim0_lr0=1.586e-04, train_time=1.097
+[gpub010:0/16] 2024-02-10 17:34:53,722 (trainer:762) INFO: 40epoch:train:11701-11800batch: iter_time=8.168e-05, forward_time=0.375, loss_ctc=51.427, loss_att=45.907, acc=0.756, loss=47.563, backward_time=0.352, grad_norm=46.620, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.586e-04, train_time=1.415
+[gpub010:0/16] 2024-02-10 17:36:59,367 (trainer:762) INFO: 40epoch:train:11801-11900batch: iter_time=8.592e-05, forward_time=0.293, loss_ctc=48.126, loss_att=47.795, acc=0.771, loss=47.894, backward_time=0.296, grad_norm=46.817, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.586e-04, train_time=1.256
+[gpub010:0/16] 2024-02-10 17:39:02,882 (trainer:762) INFO: 40epoch:train:11901-12000batch: iter_time=8.024e-05, forward_time=0.399, loss_ctc=42.975, loss_att=42.424, acc=0.771, loss=42.589, backward_time=0.324, grad_norm=37.446, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.585e-04, train_time=1.235
+[gpub010:0/16] 2024-02-10 17:41:13,960 (trainer:762) INFO: 40epoch:train:12001-12100batch: iter_time=9.051e-05, forward_time=0.299, loss_ctc=43.395, loss_att=39.529, acc=0.775, loss=40.688, backward_time=0.297, grad_norm=40.735, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.585e-04, train_time=1.310
+[gpub010:0/16] 2024-02-10 17:43:37,332 (trainer:762) INFO: 40epoch:train:12101-12200batch: iter_time=5.464e-04, forward_time=0.347, loss_ctc=46.647, loss_att=45.426, acc=0.763, loss=45.792, backward_time=0.327, grad_norm=43.134, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.585e-04, train_time=1.434
+[gpub010:0/16] 2024-02-10 17:45:31,038 (trainer:762) INFO: 40epoch:train:12201-12300batch: iter_time=8.838e-05, forward_time=0.337, loss_ctc=49.065, loss_att=47.987, acc=0.743, loss=48.310, backward_time=0.309, grad_norm=57.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.585e-04, train_time=1.137
+[gpub010:0/16] 2024-02-10 17:47:31,297 (trainer:762) INFO: 40epoch:train:12301-12400batch: iter_time=8.019e-05, forward_time=0.320, loss_ctc=50.300, loss_att=50.391, acc=0.756, loss=50.364, backward_time=0.301, grad_norm=45.438, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.585e-04, train_time=1.202
+[gpub010:0/16] 2024-02-10 17:50:01,713 (trainer:762) INFO: 40epoch:train:12401-12500batch: iter_time=8.485e-05, forward_time=0.390, loss_ctc=42.557, loss_att=43.797, acc=0.770, loss=43.425, backward_time=0.334, grad_norm=41.418, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.585e-04, train_time=1.504
+[gpub010:0/16] 2024-02-10 17:50:21,742 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-10 17:50:41,076 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 17:50:44,617 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1ee470a00>)
+[gpub010:0/16] 2024-02-10 17:50:44,617 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-10 17:50:44,620 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 17:56:47,771 (trainer:762) INFO: 40epoch:train:12501-12600batch: iter_time=2.877, forward_time=0.289, loss_ctc=43.635, loss_att=37.420, acc=0.790, loss=39.284, backward_time=0.295, grad_norm=39.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.585e-04, train_time=4.060
+[gpub010:0/16] 2024-02-10 17:59:18,390 (trainer:762) INFO: 40epoch:train:12601-12700batch: iter_time=8.207e-05, forward_time=0.375, loss_ctc=41.666, loss_att=44.835, acc=0.763, loss=43.884, backward_time=0.321, grad_norm=48.591, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.584e-04, train_time=1.506
+[gpub010:0/16] 2024-02-10 18:01:14,110 (trainer:762) INFO: 40epoch:train:12701-12800batch: iter_time=8.467e-05, forward_time=0.307, loss_ctc=38.797, loss_att=34.183, acc=0.791, loss=35.567, backward_time=0.296, grad_norm=36.920, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.584e-04, train_time=1.157
+[gpub010:0/16] 2024-02-10 18:03:15,388 (trainer:762) INFO: 40epoch:train:12801-12900batch: iter_time=8.604e-05, forward_time=0.294, loss_ctc=51.567, loss_att=52.139, acc=0.750, loss=51.967, backward_time=0.300, grad_norm=51.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.584e-04, train_time=1.213
+[gpub010:0/16] 2024-02-10 18:05:39,460 (trainer:762) INFO: 40epoch:train:12901-13000batch: iter_time=8.697e-05, forward_time=0.354, loss_ctc=47.101, loss_att=49.117, acc=0.749, loss=48.512, backward_time=0.336, grad_norm=56.479, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.584e-04, train_time=1.440
+[gpub010:0/16] 2024-02-10 18:07:34,188 (trainer:762) INFO: 40epoch:train:13001-13100batch: iter_time=8.404e-05, forward_time=0.295, loss_ctc=44.802, loss_att=43.618, acc=0.792, loss=43.973, backward_time=0.300, grad_norm=39.404, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.584e-04, train_time=1.147
+[gpub010:0/16] 2024-02-10 18:10:12,587 (trainer:762) INFO: 40epoch:train:13101-13200batch: iter_time=8.857e-05, forward_time=0.378, loss_ctc=47.706, loss_att=48.642, acc=0.766, loss=48.362, backward_time=0.346, grad_norm=48.464, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.584e-04, train_time=1.584
+[gpub010:0/16] 2024-02-10 18:12:10,548 (trainer:762) INFO: 40epoch:train:13201-13300batch: iter_time=8.403e-05, forward_time=0.315, loss_ctc=43.591, loss_att=42.964, acc=0.774, loss=43.153, backward_time=0.300, grad_norm=39.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.584e-04, train_time=1.179
+[gpub010:0/16] 2024-02-10 18:14:32,783 (trainer:762) INFO: 40epoch:train:13301-13400batch: iter_time=8.826e-05, forward_time=0.292, loss_ctc=42.999, loss_att=39.433, acc=0.778, loss=40.503, backward_time=0.297, grad_norm=39.671, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.584e-04, train_time=1.422
+[gpub010:0/16] 2024-02-10 18:16:39,282 (trainer:762) INFO: 40epoch:train:13401-13500batch: iter_time=3.330e-04, forward_time=0.366, loss_ctc=51.891, loss_att=53.001, acc=0.751, loss=52.668, backward_time=0.328, grad_norm=63.029, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.583e-04, train_time=1.265
+[gpub010:0/16] 2024-02-10 18:18:53,545 (trainer:762) INFO: 40epoch:train:13501-13600batch: iter_time=8.326e-05, forward_time=0.293, loss_ctc=49.132, loss_att=47.299, acc=0.765, loss=47.849, backward_time=0.297, grad_norm=45.839, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.583e-04, train_time=1.342
+[gpub010:0/16] 2024-02-10 18:21:15,943 (trainer:762) INFO: 40epoch:train:13601-13700batch: iter_time=4.994e-04, forward_time=0.400, loss_ctc=47.607, loss_att=46.541, acc=0.768, loss=46.861, backward_time=0.334, grad_norm=43.434, clip=100.000, loss_scale=7.788e+33, optim_step_time=0.098, optim0_lr0=1.583e-04, train_time=1.423
+[gpub010:0/16] 2024-02-10 18:22:35,107 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-10 18:22:54,451 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 18:22:58,086 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1b0f1e1a0>)
+[gpub010:0/16] 2024-02-10 18:22:58,086 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-10 18:22:58,122 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 18:29:10,374 (trainer:762) INFO: 40epoch:train:13701-13800batch: iter_time=3.422, forward_time=0.308, loss_ctc=40.626, loss_att=39.595, acc=0.779, loss=39.904, backward_time=0.297, grad_norm=39.814, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.583e-04, train_time=4.745
+[gpub010:0/16] 2024-02-10 18:30:29,617 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 18:32:46,824 (trainer:762) INFO: 40epoch:train:13801-13900batch: iter_time=0.064, forward_time=0.292, loss_ctc=44.671, loss_att=46.689, acc=0.762, loss=46.083, backward_time=0.297, grad_norm=42.815, clip=100.000, loss_scale=7.395e+33, optim_step_time=0.093, optim0_lr0=1.583e-04, train_time=2.164
+[gpub010:0/16] 2024-02-10 18:35:07,666 (trainer:762) INFO: 40epoch:train:13901-14000batch: iter_time=8.613e-05, forward_time=0.358, loss_ctc=40.085, loss_att=38.978, acc=0.763, loss=39.310, backward_time=0.312, grad_norm=45.906, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.583e-04, train_time=1.408
+[gpub010:0/16] 2024-02-10 18:37:15,365 (trainer:762) INFO: 40epoch:train:14001-14100batch: iter_time=8.618e-05, forward_time=0.338, loss_ctc=41.667, loss_att=39.135, acc=0.766, loss=39.894, backward_time=0.302, grad_norm=41.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.583e-04, train_time=1.277
+[gpub010:0/16] 2024-02-10 18:39:09,220 (trainer:762) INFO: 40epoch:train:14101-14200batch: iter_time=8.847e-05, forward_time=0.295, loss_ctc=46.007, loss_att=50.889, acc=0.753, loss=49.425, backward_time=0.297, grad_norm=50.820, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.582e-04, train_time=1.138
+[gpub010:0/16] 2024-02-10 18:41:24,490 (trainer:762) INFO: 40epoch:train:14201-14300batch: iter_time=8.274e-05, forward_time=0.293, loss_ctc=50.698, loss_att=45.894, acc=0.757, loss=47.335, backward_time=0.296, grad_norm=46.856, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.582e-04, train_time=1.352
+[gpub010:0/16] 2024-02-10 18:43:59,082 (trainer:762) INFO: 40epoch:train:14301-14400batch: iter_time=0.007, forward_time=0.400, loss_ctc=48.065, loss_att=47.591, acc=0.771, loss=47.733, backward_time=0.323, grad_norm=47.556, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.582e-04, train_time=1.544
+[gpub010:0/16] 2024-02-10 18:46:14,024 (trainer:762) INFO: 40epoch:train:14401-14500batch: iter_time=8.048e-05, forward_time=0.294, loss_ctc=43.216, loss_att=42.211, acc=0.771, loss=42.513, backward_time=0.296, grad_norm=38.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.582e-04, train_time=1.351
+[gpub010:0/16] 2024-02-10 18:48:20,487 (trainer:762) INFO: 40epoch:train:14501-14600batch: iter_time=8.111e-05, forward_time=0.294, loss_ctc=43.108, loss_att=39.101, acc=0.776, loss=40.303, backward_time=0.296, grad_norm=39.622, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.582e-04, train_time=1.264
+[gpub010:0/16] 2024-02-10 18:50:21,813 (trainer:762) INFO: 40epoch:train:14601-14700batch: iter_time=8.259e-05, forward_time=0.323, loss_ctc=46.674, loss_att=45.844, acc=0.762, loss=46.093, backward_time=0.324, grad_norm=43.768, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.582e-04, train_time=1.213
+[gpub010:0/16] 2024-02-10 18:52:26,107 (trainer:762) INFO: 40epoch:train:14701-14800batch: iter_time=7.731e-05, forward_time=0.301, loss_ctc=48.759, loss_att=48.111, acc=0.743, loss=48.305, backward_time=0.301, grad_norm=62.194, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.582e-04, train_time=1.242
+[gpub010:0/16] 2024-02-10 18:54:28,600 (trainer:762) INFO: 40epoch:train:14801-14900batch: iter_time=7.851e-05, forward_time=0.296, loss_ctc=50.443, loss_att=50.178, acc=0.757, loss=50.257, backward_time=0.299, grad_norm=44.345, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.582e-04, train_time=1.226
+[gpub010:0/16] 2024-02-10 18:56:37,751 (trainer:762) INFO: 40epoch:train:14901-15000batch: iter_time=7.856e-05, forward_time=0.291, loss_ctc=42.289, loss_att=43.620, acc=0.771, loss=43.221, backward_time=0.295, grad_norm=41.412, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.581e-04, train_time=1.291
+[gpub010:0/16] 2024-02-10 19:34:17,972 (trainer:361) INFO: 40epoch results: [train] iter_time=0.280, forward_time=0.337, loss_ctc=46.283, loss_att=45.121, acc=0.764, loss=45.470, backward_time=0.316, grad_norm=45.290, clip=100.000, loss_scale=6.388e+33, optim_step_time=0.096, optim0_lr0=1.591e-04, train_time=1.696, time=7 hours, 4 minutes and 24.05 seconds, total_count=630000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=34.803, cer_ctc=0.181, loss_att=37.461, acc=0.688, cer=0.342, wer=0.990, loss=36.663, time=37 minutes and 16.1 seconds, total_count=196182, gpu_max_cached_mem_GB=42.092
+[gpub010:0/16] 2024-02-10 19:34:36,327 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-02-10 19:34:36,445 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/35epoch.pth
+[gpub010:0/16] 2024-02-10 19:34:36,445 (trainer:290) INFO: 41/45epoch started. Estimated time to finish: 1 day, 14 hours and 29 minutes
+[gpub010:0/16] 2024-02-10 19:34:36,456 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-10 19:34:54,761 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 19:34:58,140 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1ee465ed0>)
+[gpub010:0/16] 2024-02-10 19:34:58,140 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-10 19:34:58,144 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 19:41:04,647 (trainer:762) INFO: 41epoch:train:1-100batch: iter_time=2.688, forward_time=0.298, loss_ctc=46.593, loss_att=45.969, acc=0.743, loss=46.156, backward_time=0.303, grad_norm=43.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.581e-04, train_time=3.882
+[gpub010:0/16] 2024-02-10 19:43:07,382 (trainer:762) INFO: 41epoch:train:101-200batch: iter_time=3.777e-04, forward_time=0.338, loss_ctc=51.338, loss_att=47.456, acc=0.751, loss=48.620, backward_time=0.339, grad_norm=46.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.581e-04, train_time=1.227
+[gpub010:0/16] 2024-02-10 19:45:20,988 (trainer:762) INFO: 41epoch:train:201-300batch: iter_time=8.162e-05, forward_time=0.302, loss_ctc=44.194, loss_att=49.320, acc=0.751, loss=47.782, backward_time=0.303, grad_norm=43.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.581e-04, train_time=1.335
+[gpub010:0/16] 2024-02-10 19:47:27,534 (trainer:762) INFO: 41epoch:train:301-400batch: iter_time=0.001, forward_time=0.322, loss_ctc=43.535, loss_att=40.460, acc=0.766, loss=41.382, backward_time=0.316, grad_norm=40.311, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.581e-04, train_time=1.266
+[gpub010:0/16] 2024-02-10 19:49:28,184 (trainer:762) INFO: 41epoch:train:401-500batch: iter_time=7.992e-05, forward_time=0.337, loss_ctc=48.269, loss_att=46.888, acc=0.759, loss=47.302, backward_time=0.307, grad_norm=45.753, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.581e-04, train_time=1.206
+[gpub010:0/16] 2024-02-10 19:51:38,073 (trainer:762) INFO: 41epoch:train:501-600batch: iter_time=9.111e-05, forward_time=0.317, loss_ctc=48.053, loss_att=40.673, acc=0.767, loss=42.887, backward_time=0.325, grad_norm=48.486, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.581e-04, train_time=1.298
+[gpub010:0/16] 2024-02-10 19:53:34,919 (trainer:762) INFO: 41epoch:train:601-700batch: iter_time=1.001e-04, forward_time=0.331, loss_ctc=45.160, loss_att=47.571, acc=0.741, loss=46.848, backward_time=0.312, grad_norm=46.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.580e-04, train_time=1.168
+[gpub010:0/16] 2024-02-10 19:55:42,828 (trainer:762) INFO: 41epoch:train:701-800batch: iter_time=8.349e-05, forward_time=0.311, loss_ctc=44.801, loss_att=41.916, acc=0.774, loss=42.782, backward_time=0.311, grad_norm=41.548, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.580e-04, train_time=1.279
+[gpub010:0/16] 2024-02-10 19:57:46,980 (trainer:762) INFO: 41epoch:train:801-900batch: iter_time=8.284e-04, forward_time=0.317, loss_ctc=45.771, loss_att=47.817, acc=0.752, loss=47.203, backward_time=0.319, grad_norm=42.683, clip=100.000, loss_scale=8.152e+33, optim_step_time=0.096, optim0_lr0=1.580e-04, train_time=1.241
+[gpub010:0/16] 2024-02-10 19:59:52,137 (trainer:762) INFO: 41epoch:train:901-1000batch: iter_time=8.410e-05, forward_time=0.315, loss_ctc=45.355, loss_att=47.770, acc=0.759, loss=47.046, backward_time=0.317, grad_norm=40.960, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.580e-04, train_time=1.251
+[gpub010:0/16] 2024-02-10 20:02:07,931 (trainer:762) INFO: 41epoch:train:1001-1100batch: iter_time=4.428e-04, forward_time=0.330, loss_ctc=49.820, loss_att=49.985, acc=0.738, loss=49.936, backward_time=0.312, grad_norm=55.714, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.580e-04, train_time=1.357
+[gpub010:0/16] 2024-02-10 20:04:09,476 (trainer:762) INFO: 41epoch:train:1101-1200batch: iter_time=8.704e-05, forward_time=0.309, loss_ctc=52.324, loss_att=49.608, acc=0.746, loss=50.423, backward_time=0.305, grad_norm=59.066, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.580e-04, train_time=1.216
+[gpub010:0/16] 2024-02-10 20:05:32,320 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-10 20:05:51,140 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 20:05:54,696 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1e8398190>)
+[gpub010:0/16] 2024-02-10 20:05:54,696 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-10 20:05:54,716 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 20:11:26,965 (trainer:762) INFO: 41epoch:train:1201-1300batch: iter_time=3.111, forward_time=0.350, loss_ctc=48.751, loss_att=46.980, acc=0.754, loss=47.511, backward_time=0.305, grad_norm=45.995, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.580e-04, train_time=4.375
+[gpub010:0/16] 2024-02-10 20:13:28,308 (trainer:762) INFO: 41epoch:train:1301-1400batch: iter_time=8.254e-05, forward_time=0.312, loss_ctc=47.279, loss_att=44.675, acc=0.764, loss=45.456, backward_time=0.310, grad_norm=42.541, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.580e-04, train_time=1.212
+[gpub010:0/16] 2024-02-10 20:15:21,667 (trainer:762) INFO: 41epoch:train:1401-1500batch: iter_time=7.871e-05, forward_time=0.306, loss_ctc=42.538, loss_att=44.213, acc=0.767, loss=43.711, backward_time=0.309, grad_norm=40.087, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.579e-04, train_time=1.134
+[gpub010:0/16] 2024-02-10 20:17:35,093 (trainer:762) INFO: 41epoch:train:1501-1600batch: iter_time=9.267e-05, forward_time=0.331, loss_ctc=46.309, loss_att=50.954, acc=0.753, loss=49.561, backward_time=0.304, grad_norm=44.227, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.579e-04, train_time=1.334
+[gpub010:0/16] 2024-02-10 20:19:27,880 (trainer:762) INFO: 41epoch:train:1601-1700batch: iter_time=8.035e-05, forward_time=0.311, loss_ctc=46.487, loss_att=45.809, acc=0.767, loss=46.012, backward_time=0.305, grad_norm=44.420, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.579e-04, train_time=1.127
+[gpub010:0/16] 2024-02-10 20:21:34,654 (trainer:762) INFO: 41epoch:train:1701-1800batch: iter_time=8.103e-05, forward_time=0.317, loss_ctc=45.735, loss_att=43.344, acc=0.769, loss=44.061, backward_time=0.304, grad_norm=42.224, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.579e-04, train_time=1.268
+[gpub010:0/16] 2024-02-10 20:23:48,734 (trainer:762) INFO: 41epoch:train:1801-1900batch: iter_time=8.040e-05, forward_time=0.313, loss_ctc=48.677, loss_att=42.167, acc=0.769, loss=44.120, backward_time=0.315, grad_norm=44.792, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.579e-04, train_time=1.341
+[gpub010:0/16] 2024-02-10 20:25:41,536 (trainer:762) INFO: 41epoch:train:1901-2000batch: iter_time=8.510e-05, forward_time=0.307, loss_ctc=42.524, loss_att=42.001, acc=0.770, loss=42.157, backward_time=0.301, grad_norm=40.871, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.579e-04, train_time=1.128
+[gpub010:0/16] 2024-02-10 20:27:43,126 (trainer:762) INFO: 41epoch:train:2001-2100batch: iter_time=8.800e-05, forward_time=0.316, loss_ctc=46.311, loss_att=49.037, acc=0.773, loss=48.219, backward_time=0.305, grad_norm=44.391, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.579e-04, train_time=1.216
+[gpub010:0/16] 2024-02-10 20:29:45,269 (trainer:762) INFO: 41epoch:train:2101-2200batch: iter_time=8.739e-05, forward_time=0.321, loss_ctc=44.500, loss_att=43.160, acc=0.767, loss=43.562, backward_time=0.307, grad_norm=40.922, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.579e-04, train_time=1.221
+[gpub010:0/16] 2024-02-10 20:31:42,780 (trainer:762) INFO: 41epoch:train:2201-2300batch: iter_time=8.456e-05, forward_time=0.304, loss_ctc=51.680, loss_att=50.477, acc=0.760, loss=50.838, backward_time=0.302, grad_norm=43.841, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.578e-04, train_time=1.175
+[gpub010:0/16] 2024-02-10 20:33:45,802 (trainer:762) INFO: 41epoch:train:2301-2400batch: iter_time=8.162e-05, forward_time=0.307, loss_ctc=41.878, loss_att=45.258, acc=0.755, loss=44.244, backward_time=0.303, grad_norm=45.780, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.578e-04, train_time=1.230
+[gpub010:0/16] 2024-02-10 20:35:48,972 (trainer:762) INFO: 41epoch:train:2401-2500batch: iter_time=6.622e-04, forward_time=0.312, loss_ctc=53.112, loss_att=50.445, acc=0.760, loss=51.245, backward_time=0.310, grad_norm=58.830, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.578e-04, train_time=1.232
+[gpub010:0/16] 2024-02-10 20:36:09,000 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-10 20:36:28,226 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 20:36:31,969 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb22a3b3820>)
+[gpub010:0/16] 2024-02-10 20:36:31,969 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-10 20:36:31,972 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 20:43:18,675 (trainer:762) INFO: 41epoch:train:2501-2600batch: iter_time=3.293, forward_time=0.332, loss_ctc=45.650, loss_att=45.818, acc=0.746, loss=45.767, backward_time=0.310, grad_norm=43.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.578e-04, train_time=4.497
+[gpub010:0/16] 2024-02-10 20:45:13,905 (trainer:762) INFO: 41epoch:train:2601-2700batch: iter_time=8.104e-05, forward_time=0.311, loss_ctc=50.417, loss_att=47.147, acc=0.757, loss=48.128, backward_time=0.301, grad_norm=43.992, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.578e-04, train_time=1.151
+[gpub010:0/16] 2024-02-10 20:47:27,262 (trainer:762) INFO: 41epoch:train:2701-2800batch: iter_time=2.710e-04, forward_time=0.304, loss_ctc=43.034, loss_att=49.417, acc=0.753, loss=47.502, backward_time=0.349, grad_norm=43.210, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.578e-04, train_time=1.334
+[gpub010:0/16] 2024-02-10 20:49:20,456 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 20:49:46,708 (trainer:762) INFO: 41epoch:train:2801-2900batch: iter_time=8.805e-05, forward_time=0.323, loss_ctc=42.505, loss_att=40.764, acc=0.767, loss=41.286, backward_time=0.314, grad_norm=40.753, clip=100.000, loss_scale=1.395e+34, optim_step_time=0.096, optim0_lr0=1.578e-04, train_time=1.394
+[gpub010:0/16] 2024-02-10 20:51:44,672 (trainer:762) INFO: 41epoch:train:2901-3000batch: iter_time=8.462e-05, forward_time=0.299, loss_ctc=46.623, loss_att=46.310, acc=0.764, loss=46.404, backward_time=0.308, grad_norm=43.537, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.577e-04, train_time=1.179
+[gpub010:0/16] 2024-02-10 20:53:44,844 (trainer:762) INFO: 41epoch:train:3001-3100batch: iter_time=8.452e-05, forward_time=0.321, loss_ctc=46.776, loss_att=39.973, acc=0.772, loss=42.014, backward_time=0.298, grad_norm=42.600, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.577e-04, train_time=1.201
+[gpub010:0/16] 2024-02-10 20:56:04,028 (trainer:762) INFO: 41epoch:train:3101-3200batch: iter_time=8.046e-05, forward_time=0.326, loss_ctc=44.242, loss_att=47.014, acc=0.745, loss=46.183, backward_time=0.322, grad_norm=44.614, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.577e-04, train_time=1.392
+[gpub010:0/16] 2024-02-10 20:58:07,825 (trainer:762) INFO: 41epoch:train:3201-3300batch: iter_time=8.414e-05, forward_time=0.315, loss_ctc=44.014, loss_att=41.446, acc=0.779, loss=42.216, backward_time=0.299, grad_norm=37.812, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.577e-04, train_time=1.238
+[gpub010:0/16] 2024-02-10 21:00:15,707 (trainer:762) INFO: 41epoch:train:3301-3400batch: iter_time=1.652e-04, forward_time=0.329, loss_ctc=44.940, loss_att=47.470, acc=0.753, loss=46.711, backward_time=0.304, grad_norm=42.998, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.577e-04, train_time=1.279
+[gpub010:0/16] 2024-02-10 21:02:23,358 (trainer:762) INFO: 41epoch:train:3401-3500batch: iter_time=7.935e-05, forward_time=0.334, loss_ctc=44.517, loss_att=46.791, acc=0.764, loss=46.109, backward_time=0.313, grad_norm=40.666, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.577e-04, train_time=1.276
+[gpub010:0/16] 2024-02-10 21:04:21,738 (trainer:762) INFO: 41epoch:train:3501-3600batch: iter_time=8.275e-05, forward_time=0.297, loss_ctc=47.366, loss_att=48.313, acc=0.742, loss=48.029, backward_time=0.298, grad_norm=49.681, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.577e-04, train_time=1.184
+[gpub010:0/16] 2024-02-10 21:06:26,650 (trainer:762) INFO: 41epoch:train:3601-3700batch: iter_time=7.541e-04, forward_time=0.336, loss_ctc=50.316, loss_att=48.092, acc=0.746, loss=48.759, backward_time=0.308, grad_norm=55.777, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.577e-04, train_time=1.248
+[gpub010:0/16] 2024-02-10 21:07:54,402 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-10 21:08:13,604 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 21:08:17,101 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba27167e20>)
+[gpub010:0/16] 2024-02-10 21:08:17,101 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-10 21:08:17,129 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 21:13:36,899 (trainer:762) INFO: 41epoch:train:3701-3800batch: iter_time=3.025, forward_time=0.316, loss_ctc=48.053, loss_att=46.740, acc=0.745, loss=47.134, backward_time=0.305, grad_norm=44.964, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.576e-04, train_time=4.303
+[gpub010:0/16] 2024-02-10 21:13:46,422 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 21:15:37,289 (trainer:762) INFO: 41epoch:train:3801-3900batch: iter_time=7.835e-05, forward_time=0.313, loss_ctc=46.573, loss_att=44.651, acc=0.757, loss=45.227, backward_time=0.300, grad_norm=42.620, clip=100.000, loss_scale=5.612e+33, optim_step_time=0.093, optim0_lr0=1.576e-04, train_time=1.204
+[gpub010:0/16] 2024-02-10 21:17:46,402 (trainer:762) INFO: 41epoch:train:3901-4000batch: iter_time=8.574e-05, forward_time=0.333, loss_ctc=42.019, loss_att=42.629, acc=0.771, loss=42.446, backward_time=0.308, grad_norm=40.160, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.576e-04, train_time=1.290
+[gpub010:0/16] 2024-02-10 21:19:48,755 (trainer:762) INFO: 41epoch:train:4001-4100batch: iter_time=9.152e-05, forward_time=0.312, loss_ctc=45.509, loss_att=49.962, acc=0.753, loss=48.626, backward_time=0.309, grad_norm=46.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.576e-04, train_time=1.225
+[gpub010:0/16] 2024-02-10 21:21:42,396 (trainer:762) INFO: 41epoch:train:4101-4200batch: iter_time=8.879e-05, forward_time=0.306, loss_ctc=45.709, loss_att=45.090, acc=0.766, loss=45.276, backward_time=0.298, grad_norm=44.018, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.576e-04, train_time=1.136
+[gpub010:0/16] 2024-02-10 21:23:53,517 (trainer:762) INFO: 41epoch:train:4201-4300batch: iter_time=8.309e-05, forward_time=0.300, loss_ctc=44.767, loss_att=42.535, acc=0.763, loss=43.205, backward_time=0.311, grad_norm=41.316, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.576e-04, train_time=1.311
+[gpub010:0/16] 2024-02-10 21:25:54,647 (trainer:762) INFO: 41epoch:train:4301-4400batch: iter_time=8.033e-05, forward_time=0.341, loss_ctc=48.413, loss_att=42.025, acc=0.771, loss=43.941, backward_time=0.313, grad_norm=44.315, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.576e-04, train_time=1.210
+[gpub010:0/16] 2024-02-10 21:27:58,542 (trainer:762) INFO: 41epoch:train:4401-4500batch: iter_time=9.245e-05, forward_time=0.309, loss_ctc=42.534, loss_att=42.407, acc=0.763, loss=42.445, backward_time=0.301, grad_norm=41.493, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.576e-04, train_time=1.240
+[gpub010:0/16] 2024-02-10 21:30:16,294 (trainer:762) INFO: 41epoch:train:4501-4600batch: iter_time=9.159e-05, forward_time=0.321, loss_ctc=45.914, loss_att=47.697, acc=0.765, loss=47.162, backward_time=0.311, grad_norm=44.030, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.575e-04, train_time=1.377
+[gpub010:0/16] 2024-02-10 21:32:10,590 (trainer:762) INFO: 41epoch:train:4601-4700batch: iter_time=8.800e-05, forward_time=0.302, loss_ctc=44.030, loss_att=43.949, acc=0.754, loss=43.974, backward_time=0.300, grad_norm=40.223, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.575e-04, train_time=1.143
+[gpub010:0/16] 2024-02-10 21:34:20,917 (trainer:762) INFO: 41epoch:train:4701-4800batch: iter_time=8.566e-05, forward_time=0.308, loss_ctc=50.588, loss_att=50.407, acc=0.752, loss=50.462, backward_time=0.310, grad_norm=47.415, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.575e-04, train_time=1.302
+[gpub010:0/16] 2024-02-10 21:35:43,665 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 21:36:17,347 (trainer:762) INFO: 41epoch:train:4801-4900batch: iter_time=9.516e-05, forward_time=0.304, loss_ctc=41.912, loss_att=44.292, acc=0.749, loss=43.578, backward_time=0.314, grad_norm=48.422, clip=100.000, loss_scale=4.458e+33, optim_step_time=0.095, optim0_lr0=1.575e-04, train_time=1.165
+[gpub010:0/16] 2024-02-10 21:38:25,154 (trainer:762) INFO: 41epoch:train:4901-5000batch: iter_time=8.164e-05, forward_time=0.314, loss_ctc=51.818, loss_att=48.494, acc=0.750, loss=49.491, backward_time=0.304, grad_norm=59.340, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.575e-04, train_time=1.278
+[gpub010:0/16] 2024-02-10 21:38:45,182 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-10 21:39:04,477 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 21:39:08,051 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb22a66dba0>)
+[gpub010:0/16] 2024-02-10 21:39:08,051 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-10 21:39:08,056 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 21:45:53,719 (trainer:762) INFO: 41epoch:train:5001-5100batch: iter_time=3.179, forward_time=0.310, loss_ctc=45.499, loss_att=45.030, acc=0.751, loss=45.171, backward_time=0.305, grad_norm=43.447, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.575e-04, train_time=4.485
+[gpub010:0/16] 2024-02-10 21:47:55,390 (trainer:762) INFO: 41epoch:train:5101-5200batch: iter_time=8.054e-05, forward_time=0.318, loss_ctc=49.985, loss_att=46.346, acc=0.760, loss=47.438, backward_time=0.320, grad_norm=44.695, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.575e-04, train_time=1.216
+[gpub010:0/16] 2024-02-10 21:50:00,158 (trainer:762) INFO: 41epoch:train:5201-5300batch: iter_time=7.821e-05, forward_time=0.302, loss_ctc=42.526, loss_att=49.222, acc=0.755, loss=47.213, backward_time=0.303, grad_norm=42.114, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.574e-04, train_time=1.248
+[gpub010:0/16] 2024-02-10 21:51:57,031 (trainer:762) INFO: 41epoch:train:5301-5400batch: iter_time=7.551e-05, forward_time=0.292, loss_ctc=42.335, loss_att=40.132, acc=0.770, loss=40.793, backward_time=0.296, grad_norm=39.490, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.574e-04, train_time=1.168
+[gpub010:0/16] 2024-02-10 21:53:57,796 (trainer:762) INFO: 41epoch:train:5401-5500batch: iter_time=7.805e-05, forward_time=0.308, loss_ctc=45.837, loss_att=45.744, acc=0.769, loss=45.772, backward_time=0.312, grad_norm=42.509, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.574e-04, train_time=1.208
+[gpub010:0/16] 2024-02-10 21:56:08,746 (trainer:762) INFO: 41epoch:train:5501-5600batch: iter_time=8.206e-05, forward_time=0.347, loss_ctc=46.830, loss_att=39.784, acc=0.774, loss=41.898, backward_time=0.304, grad_norm=43.962, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.574e-04, train_time=1.309
+[gpub010:0/16] 2024-02-10 21:58:04,715 (trainer:762) INFO: 41epoch:train:5601-5700batch: iter_time=8.178e-05, forward_time=0.299, loss_ctc=43.787, loss_att=46.579, acc=0.747, loss=45.741, backward_time=0.298, grad_norm=43.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.574e-04, train_time=1.160
+[gpub010:0/16] 2024-02-10 22:00:06,630 (trainer:762) INFO: 41epoch:train:5701-5800batch: iter_time=8.269e-05, forward_time=0.291, loss_ctc=43.793, loss_att=41.052, acc=0.780, loss=41.874, backward_time=0.296, grad_norm=37.757, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.574e-04, train_time=1.219
+[gpub010:0/16] 2024-02-10 22:02:09,142 (trainer:762) INFO: 41epoch:train:5801-5900batch: iter_time=8.299e-05, forward_time=0.307, loss_ctc=44.735, loss_att=47.289, acc=0.754, loss=46.523, backward_time=0.313, grad_norm=43.293, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.574e-04, train_time=1.225
+[gpub010:0/16] 2024-02-10 22:04:17,904 (trainer:762) INFO: 41epoch:train:5901-6000batch: iter_time=2.121e-04, forward_time=0.319, loss_ctc=44.428, loss_att=46.530, acc=0.766, loss=45.899, backward_time=0.312, grad_norm=42.209, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.574e-04, train_time=1.286
+[gpub010:0/16] 2024-02-10 22:06:15,503 (trainer:762) INFO: 41epoch:train:6001-6100batch: iter_time=8.121e-05, forward_time=0.299, loss_ctc=48.001, loss_att=49.336, acc=0.744, loss=48.936, backward_time=0.302, grad_norm=50.070, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.573e-04, train_time=1.177
+[gpub010:0/16] 2024-02-10 22:08:24,152 (trainer:762) INFO: 41epoch:train:6101-6200batch: iter_time=7.943e-05, forward_time=0.291, loss_ctc=49.436, loss_att=47.068, acc=0.748, loss=47.778, backward_time=0.294, grad_norm=55.856, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.573e-04, train_time=1.286
+[gpub010:0/16] 2024-02-10 22:09:43,015 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-10 22:10:01,923 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 22:10:05,395 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba27180760>)
+[gpub010:0/16] 2024-02-10 22:10:05,395 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-10 22:10:05,400 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 22:14:43,658 (trainer:762) INFO: 41epoch:train:6201-6300batch: iter_time=2.586, forward_time=0.353, loss_ctc=48.059, loss_att=47.031, acc=0.755, loss=47.340, backward_time=0.303, grad_norm=45.940, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.573e-04, train_time=3.795
+[gpub010:0/16] 2024-02-10 22:16:52,560 (trainer:762) INFO: 41epoch:train:6301-6400batch: iter_time=8.033e-05, forward_time=0.301, loss_ctc=46.527, loss_att=44.584, acc=0.766, loss=45.167, backward_time=0.305, grad_norm=42.326, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.573e-04, train_time=1.289
+[gpub010:0/16] 2024-02-10 22:18:51,463 (trainer:762) INFO: 41epoch:train:6401-6500batch: iter_time=8.040e-05, forward_time=0.291, loss_ctc=41.920, loss_att=43.404, acc=0.771, loss=42.959, backward_time=0.296, grad_norm=40.313, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.573e-04, train_time=1.189
+[gpub010:0/16] 2024-02-10 22:21:04,165 (trainer:762) INFO: 41epoch:train:6501-6600batch: iter_time=9.686e-05, forward_time=0.345, loss_ctc=45.443, loss_att=49.579, acc=0.764, loss=48.338, backward_time=0.310, grad_norm=44.294, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.573e-04, train_time=1.327
+[gpub010:0/16] 2024-02-10 22:23:00,135 (trainer:762) INFO: 41epoch:train:6601-6700batch: iter_time=8.110e-05, forward_time=0.292, loss_ctc=45.260, loss_att=45.293, acc=0.772, loss=45.283, backward_time=0.296, grad_norm=40.949, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.573e-04, train_time=1.159
+[gpub010:0/16] 2024-02-10 22:25:10,022 (trainer:762) INFO: 41epoch:train:6701-6800batch: iter_time=8.471e-05, forward_time=0.306, loss_ctc=44.328, loss_att=43.353, acc=0.771, loss=43.646, backward_time=0.296, grad_norm=41.619, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.573e-04, train_time=1.299
+[gpub010:0/16] 2024-02-10 22:27:02,669 (trainer:762) INFO: 41epoch:train:6801-6900batch: iter_time=8.815e-05, forward_time=0.293, loss_ctc=47.286, loss_att=40.869, acc=0.778, loss=42.794, backward_time=0.299, grad_norm=43.604, clip=100.000, loss_scale=3.323e+33, optim_step_time=0.093, optim0_lr0=1.572e-04, train_time=1.126
+[gpub010:0/16] 2024-02-10 22:29:17,898 (trainer:762) INFO: 41epoch:train:6901-7000batch: iter_time=8.309e-05, forward_time=0.350, loss_ctc=42.270, loss_att=41.882, acc=0.773, loss=41.998, backward_time=0.327, grad_norm=40.371, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.572e-04, train_time=1.352
+[gpub010:0/16] 2024-02-10 22:31:17,208 (trainer:762) INFO: 41epoch:train:7001-7100batch: iter_time=8.327e-05, forward_time=0.307, loss_ctc=45.943, loss_att=49.107, acc=0.777, loss=48.157, backward_time=0.301, grad_norm=45.557, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.572e-04, train_time=1.193
+[gpub010:0/16] 2024-02-10 22:33:24,326 (trainer:762) INFO: 41epoch:train:7101-7200batch: iter_time=9.230e-05, forward_time=0.293, loss_ctc=43.691, loss_att=43.008, acc=0.770, loss=43.213, backward_time=0.297, grad_norm=39.281, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.572e-04, train_time=1.271
+[gpub010:0/16] 2024-02-10 22:35:20,535 (trainer:762) INFO: 41epoch:train:7201-7300batch: iter_time=8.235e-05, forward_time=0.297, loss_ctc=50.618, loss_att=50.749, acc=0.761, loss=50.710, backward_time=0.302, grad_norm=46.131, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.572e-04, train_time=1.162
+[gpub010:0/16] 2024-02-10 22:37:33,727 (trainer:762) INFO: 41epoch:train:7301-7400batch: iter_time=8.867e-05, forward_time=0.360, loss_ctc=41.300, loss_att=45.070, acc=0.758, loss=43.939, backward_time=0.314, grad_norm=45.536, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.572e-04, train_time=1.331
+[gpub010:0/16] 2024-02-10 22:39:45,841 (trainer:762) INFO: 41epoch:train:7401-7500batch: iter_time=8.441e-05, forward_time=0.293, loss_ctc=51.915, loss_att=49.113, acc=0.764, loss=49.954, backward_time=0.297, grad_norm=58.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.572e-04, train_time=1.321
+[gpub010:0/16] 2024-02-10 22:40:05,997 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-10 22:40:25,136 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 22:40:28,686 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba97d07790>)
+[gpub010:0/16] 2024-02-10 22:40:28,686 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-10 22:40:28,773 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 22:46:58,118 (trainer:762) INFO: 41epoch:train:7501-7600batch: iter_time=3.100, forward_time=0.344, loss_ctc=45.223, loss_att=46.047, acc=0.748, loss=45.800, backward_time=0.312, grad_norm=43.582, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.571e-04, train_time=4.323
+[gpub010:0/16] 2024-02-10 22:48:58,807 (trainer:762) INFO: 41epoch:train:7601-7700batch: iter_time=7.838e-05, forward_time=0.299, loss_ctc=50.055, loss_att=47.719, acc=0.758, loss=48.420, backward_time=0.303, grad_norm=44.562, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.571e-04, train_time=1.206
+[gpub010:0/16] 2024-02-10 22:51:02,950 (trainer:762) INFO: 41epoch:train:7701-7800batch: iter_time=8.245e-05, forward_time=0.291, loss_ctc=42.239, loss_att=48.801, acc=0.757, loss=46.832, backward_time=0.295, grad_norm=42.619, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.571e-04, train_time=1.241
+[gpub010:0/16] 2024-02-10 22:53:00,348 (trainer:762) INFO: 41epoch:train:7801-7900batch: iter_time=8.234e-05, forward_time=0.300, loss_ctc=42.042, loss_att=39.716, acc=0.772, loss=40.414, backward_time=0.306, grad_norm=39.709, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.571e-04, train_time=1.174
+[gpub010:0/16] 2024-02-10 22:55:10,683 (trainer:762) INFO: 41epoch:train:7901-8000batch: iter_time=8.010e-05, forward_time=0.333, loss_ctc=45.665, loss_att=45.803, acc=0.767, loss=45.761, backward_time=0.317, grad_norm=42.371, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.571e-04, train_time=1.303
+[gpub010:0/16] 2024-02-10 22:59:16,774 (trainer:762) INFO: 41epoch:train:8001-8100batch: iter_time=0.667, forward_time=0.301, loss_ctc=46.394, loss_att=39.329, acc=0.775, loss=41.449, backward_time=0.296, grad_norm=44.005, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.571e-04, train_time=2.460
+[gpub010:0/16] 2024-02-10 22:59:39,979 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-10 23:01:55,290 (trainer:762) INFO: 41epoch:train:8101-8200batch: iter_time=0.180, forward_time=0.297, loss_ctc=43.237, loss_att=46.092, acc=0.749, loss=45.236, backward_time=0.300, grad_norm=43.038, clip=100.000, loss_scale=2.963e+33, optim_step_time=0.094, optim0_lr0=1.571e-04, train_time=1.585
+[gpub010:0/16] 2024-02-10 23:05:53,360 (trainer:762) INFO: 41epoch:train:8201-8300batch: iter_time=8.282e-05, forward_time=0.340, loss_ctc=43.717, loss_att=41.222, acc=0.780, loss=41.970, backward_time=0.671, grad_norm=36.631, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.571e-04, train_time=2.380
+[gpub010:0/16] 2024-02-10 23:10:08,098 (trainer:762) INFO: 41epoch:train:8301-8400batch: iter_time=8.618e-05, forward_time=0.317, loss_ctc=44.493, loss_att=46.960, acc=0.757, loss=46.220, backward_time=0.301, grad_norm=41.041, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.570e-04, train_time=2.547
+[gpub010:0/16] 2024-02-10 23:12:10,092 (trainer:762) INFO: 41epoch:train:8401-8500batch: iter_time=8.465e-05, forward_time=0.292, loss_ctc=44.107, loss_att=46.685, acc=0.766, loss=45.912, backward_time=0.297, grad_norm=42.209, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.570e-04, train_time=1.219
+[gpub010:0/16] 2024-02-10 23:14:01,953 (trainer:762) INFO: 41epoch:train:8501-8600batch: iter_time=7.862e-05, forward_time=0.302, loss_ctc=46.911, loss_att=48.329, acc=0.745, loss=47.904, backward_time=0.303, grad_norm=48.536, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.570e-04, train_time=1.119
+[gpub010:0/16] 2024-02-10 23:16:28,280 (trainer:762) INFO: 41epoch:train:8601-8700batch: iter_time=4.394e-04, forward_time=0.357, loss_ctc=49.187, loss_att=46.495, acc=0.752, loss=47.303, backward_time=0.308, grad_norm=58.254, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.570e-04, train_time=1.463
+[gpub010:0/16] 2024-02-10 23:17:50,831 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-10 23:18:09,819 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 23:18:13,353 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba7f672e90>)
+[gpub010:0/16] 2024-02-10 23:18:13,353 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-10 23:18:13,402 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-10 23:23:46,845 (trainer:762) INFO: 41epoch:train:8701-8800batch: iter_time=3.202, forward_time=0.304, loss_ctc=47.601, loss_att=46.233, acc=0.759, loss=46.643, backward_time=0.297, grad_norm=46.308, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.570e-04, train_time=4.386
+[gpub010:0/16] 2024-02-10 23:25:53,432 (trainer:762) INFO: 41epoch:train:8801-8900batch: iter_time=8.452e-05, forward_time=0.349, loss_ctc=45.681, loss_att=43.820, acc=0.769, loss=44.378, backward_time=0.312, grad_norm=42.085, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.570e-04, train_time=1.266
+[gpub010:0/16] 2024-02-10 23:27:59,399 (trainer:762) INFO: 41epoch:train:8901-9000batch: iter_time=8.737e-05, forward_time=0.298, loss_ctc=42.084, loss_att=43.948, acc=0.772, loss=43.389, backward_time=0.302, grad_norm=42.344, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.570e-04, train_time=1.259
+[gpub010:0/16] 2024-02-10 23:29:58,208 (trainer:762) INFO: 41epoch:train:9001-9100batch: iter_time=8.736e-05, forward_time=0.307, loss_ctc=45.103, loss_att=49.841, acc=0.758, loss=48.420, backward_time=0.301, grad_norm=44.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.570e-04, train_time=1.188
+[gpub010:0/16] 2024-02-10 23:32:04,419 (trainer:762) INFO: 41epoch:train:9101-9200batch: iter_time=8.893e-05, forward_time=0.327, loss_ctc=44.622, loss_att=45.027, acc=0.771, loss=44.905, backward_time=0.328, grad_norm=42.667, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.569e-04, train_time=1.262
+[gpub010:0/16] 2024-02-10 23:37:05,940 (trainer:762) INFO: 41epoch:train:9201-9300batch: iter_time=0.383, forward_time=0.293, loss_ctc=44.558, loss_att=42.361, acc=0.773, loss=43.020, backward_time=0.299, grad_norm=41.601, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.569e-04, train_time=3.015
+[gpub010:0/16] 2024-02-10 23:39:03,542 (trainer:762) INFO: 41epoch:train:9301-9400batch: iter_time=8.445e-05, forward_time=0.301, loss_ctc=47.181, loss_att=41.155, acc=0.776, loss=42.963, backward_time=0.304, grad_norm=42.528, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.569e-04, train_time=1.176
+[gpub010:0/16] 2024-02-10 23:41:12,626 (trainer:762) INFO: 41epoch:train:9401-9500batch: iter_time=9.848e-05, forward_time=0.348, loss_ctc=42.262, loss_att=41.528, acc=0.775, loss=41.748, backward_time=0.323, grad_norm=40.446, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.569e-04, train_time=1.290
+[gpub010:0/16] 2024-02-10 23:43:17,703 (trainer:762) INFO: 41epoch:train:9501-9600batch: iter_time=9.281e-05, forward_time=0.296, loss_ctc=45.263, loss_att=48.378, acc=0.778, loss=47.444, backward_time=0.301, grad_norm=40.897, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.569e-04, train_time=1.250
+[gpub010:0/16] 2024-02-10 23:46:57,199 (trainer:762) INFO: 41epoch:train:9601-9700batch: iter_time=9.807e-05, forward_time=0.379, loss_ctc=43.394, loss_att=42.677, acc=0.772, loss=42.892, backward_time=0.503, grad_norm=41.035, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.157, optim0_lr0=1.569e-04, train_time=2.190
+[gpub010:0/16] 2024-02-10 23:49:27,632 (trainer:762) INFO: 41epoch:train:9701-9800batch: iter_time=0.005, forward_time=0.425, loss_ctc=50.146, loss_att=50.293, acc=0.763, loss=50.249, backward_time=0.310, grad_norm=44.605, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.569e-04, train_time=1.510
+[gpub010:0/16] 2024-02-10 23:51:20,252 (trainer:762) INFO: 41epoch:train:9801-9900batch: iter_time=8.064e-05, forward_time=0.297, loss_ctc=40.245, loss_att=44.392, acc=0.760, loss=43.148, backward_time=0.302, grad_norm=45.085, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.569e-04, train_time=1.126
+[gpub010:0/16] 2024-02-10 23:53:33,566 (trainer:762) INFO: 41epoch:train:9901-10000batch: iter_time=4.624e-04, forward_time=0.356, loss_ctc=51.038, loss_att=48.244, acc=0.765, loss=49.082, backward_time=0.337, grad_norm=55.557, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.568e-04, train_time=1.333
+[gpub010:0/16] 2024-02-10 23:53:53,645 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-10 23:54:12,795 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-10 23:54:16,365 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba327c2410>)
+[gpub010:0/16] 2024-02-10 23:54:16,365 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-10 23:54:16,370 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 00:00:53,277 (trainer:762) INFO: 41epoch:train:10001-10100batch: iter_time=3.245, forward_time=0.310, loss_ctc=45.086, loss_att=44.004, acc=0.767, loss=44.328, backward_time=0.298, grad_norm=40.928, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.568e-04, train_time=4.397
+[gpub010:0/16] 2024-02-11 00:03:02,825 (trainer:762) INFO: 41epoch:train:10101-10200batch: iter_time=8.898e-05, forward_time=0.321, loss_ctc=49.653, loss_att=47.143, acc=0.763, loss=47.896, backward_time=0.332, grad_norm=43.861, clip=100.000, loss_scale=4.803e+33, optim_step_time=0.097, optim0_lr0=1.568e-04, train_time=1.295
+[gpub010:0/16] 2024-02-11 00:05:18,245 (trainer:762) INFO: 41epoch:train:10201-10300batch: iter_time=8.261e-05, forward_time=0.310, loss_ctc=42.112, loss_att=47.902, acc=0.770, loss=46.165, backward_time=0.307, grad_norm=41.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.568e-04, train_time=1.355
+[gpub010:0/16] 2024-02-11 00:07:09,778 (trainer:762) INFO: 41epoch:train:10301-10400batch: iter_time=8.437e-05, forward_time=0.291, loss_ctc=41.667, loss_att=39.203, acc=0.779, loss=39.943, backward_time=0.296, grad_norm=38.940, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.568e-04, train_time=1.115
+[gpub010:0/16] 2024-02-11 00:09:23,679 (trainer:762) INFO: 41epoch:train:10401-10500batch: iter_time=8.692e-05, forward_time=0.315, loss_ctc=45.210, loss_att=45.081, acc=0.778, loss=45.120, backward_time=0.312, grad_norm=43.209, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.568e-04, train_time=1.339
+[gpub010:0/16] 2024-02-11 00:11:32,567 (trainer:762) INFO: 41epoch:train:10501-10600batch: iter_time=8.851e-05, forward_time=0.327, loss_ctc=46.827, loss_att=39.731, acc=0.777, loss=41.860, backward_time=0.306, grad_norm=44.179, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.568e-04, train_time=1.289
+[gpub010:0/16] 2024-02-11 00:13:47,942 (trainer:762) INFO: 41epoch:train:10601-10700batch: iter_time=9.932e-05, forward_time=0.292, loss_ctc=43.220, loss_att=45.785, acc=0.757, loss=45.016, backward_time=0.297, grad_norm=43.973, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.567e-04, train_time=1.354
+[gpub010:0/16] 2024-02-11 00:16:27,207 (trainer:762) INFO: 41epoch:train:10701-10800batch: iter_time=9.607e-05, forward_time=0.521, loss_ctc=43.345, loss_att=40.940, acc=0.789, loss=41.662, backward_time=0.327, grad_norm=36.552, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.567e-04, train_time=1.592
+[gpub010:0/16] 2024-02-11 00:18:40,457 (trainer:762) INFO: 41epoch:train:10801-10900batch: iter_time=8.367e-05, forward_time=0.316, loss_ctc=44.090, loss_att=46.403, acc=0.771, loss=45.709, backward_time=0.305, grad_norm=40.801, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.567e-04, train_time=1.332
+[gpub010:0/16] 2024-02-11 00:20:40,551 (trainer:762) INFO: 41epoch:train:10901-11000batch: iter_time=8.828e-05, forward_time=0.293, loss_ctc=43.641, loss_att=46.221, acc=0.778, loss=45.447, backward_time=0.299, grad_norm=39.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.567e-04, train_time=1.202
+[gpub010:0/16] 2024-02-11 00:22:49,656 (trainer:762) INFO: 41epoch:train:11001-11100batch: iter_time=1.961e-04, forward_time=0.343, loss_ctc=46.557, loss_att=48.799, acc=0.751, loss=48.127, backward_time=0.339, grad_norm=47.692, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.567e-04, train_time=1.291
+[gpub010:0/16] 2024-02-11 00:25:02,573 (trainer:762) INFO: 41epoch:train:11101-11200batch: iter_time=8.271e-05, forward_time=0.307, loss_ctc=48.371, loss_att=46.361, acc=0.764, loss=46.964, backward_time=0.301, grad_norm=58.228, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.567e-04, train_time=1.329
+[gpub010:0/16] 2024-02-11 00:26:31,648 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-11 00:26:50,856 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 00:26:54,442 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba3cf0ef20>)
+[gpub010:0/16] 2024-02-11 00:26:54,442 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-11 00:26:54,445 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 00:32:46,058 (trainer:762) INFO: 41epoch:train:11201-11300batch: iter_time=3.291, forward_time=0.363, loss_ctc=47.302, loss_att=45.570, acc=0.768, loss=46.089, backward_time=0.313, grad_norm=44.329, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.567e-04, train_time=4.635
+[gpub010:0/16] 2024-02-11 00:34:57,105 (trainer:762) INFO: 41epoch:train:11301-11400batch: iter_time=8.118e-05, forward_time=0.320, loss_ctc=46.014, loss_att=43.358, acc=0.771, loss=44.155, backward_time=0.314, grad_norm=40.914, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.567e-04, train_time=1.310
+[gpub010:0/16] 2024-02-11 00:36:58,958 (trainer:762) INFO: 41epoch:train:11401-11500batch: iter_time=8.171e-05, forward_time=0.289, loss_ctc=41.474, loss_att=42.628, acc=0.776, loss=42.281, backward_time=0.295, grad_norm=39.888, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.566e-04, train_time=1.219
+[gpub010:0/16] 2024-02-11 00:39:05,053 (trainer:762) INFO: 41epoch:train:11501-11600batch: iter_time=8.172e-05, forward_time=0.333, loss_ctc=45.057, loss_att=49.099, acc=0.760, loss=47.887, backward_time=0.316, grad_norm=44.724, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.566e-04, train_time=1.261
+[gpub010:0/16] 2024-02-11 00:41:07,551 (trainer:762) INFO: 41epoch:train:11601-11700batch: iter_time=8.651e-05, forward_time=0.311, loss_ctc=44.551, loss_att=44.779, acc=0.772, loss=44.710, backward_time=0.304, grad_norm=41.629, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.566e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 00:43:26,535 (trainer:762) INFO: 41epoch:train:11701-11800batch: iter_time=8.332e-05, forward_time=0.311, loss_ctc=44.416, loss_att=42.337, acc=0.775, loss=42.961, backward_time=0.305, grad_norm=40.164, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.566e-04, train_time=1.390
+[gpub010:0/16] 2024-02-11 00:45:21,448 (trainer:762) INFO: 41epoch:train:11801-11900batch: iter_time=9.315e-05, forward_time=0.309, loss_ctc=47.161, loss_att=41.428, acc=0.776, loss=43.148, backward_time=0.307, grad_norm=43.989, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.566e-04, train_time=1.148
+[gpub010:0/16] 2024-02-11 00:47:27,373 (trainer:762) INFO: 41epoch:train:11901-12000batch: iter_time=8.813e-05, forward_time=0.321, loss_ctc=41.682, loss_att=40.991, acc=0.776, loss=41.198, backward_time=0.305, grad_norm=39.336, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.566e-04, train_time=1.259
+[gpub010:0/16] 2024-02-11 00:49:46,909 (trainer:762) INFO: 41epoch:train:12001-12100batch: iter_time=8.831e-05, forward_time=0.315, loss_ctc=45.493, loss_att=48.231, acc=0.778, loss=47.409, backward_time=0.327, grad_norm=43.442, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.566e-04, train_time=1.396
+[gpub010:0/16] 2024-02-11 00:51:39,051 (trainer:762) INFO: 41epoch:train:12101-12200batch: iter_time=8.498e-05, forward_time=0.307, loss_ctc=43.511, loss_att=42.400, acc=0.773, loss=42.733, backward_time=0.310, grad_norm=38.933, clip=100.000, loss_scale=9.606e+33, optim_step_time=0.094, optim0_lr0=1.566e-04, train_time=1.121
+[gpub010:0/16] 2024-02-11 00:52:45,447 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 00:53:40,910 (trainer:762) INFO: 41epoch:train:12201-12300batch: iter_time=8.215e-05, forward_time=0.293, loss_ctc=49.488, loss_att=49.655, acc=0.765, loss=49.605, backward_time=0.299, grad_norm=43.264, clip=100.000, loss_scale=8.024e+33, optim_step_time=0.093, optim0_lr0=1.565e-04, train_time=1.217
+[gpub010:0/16] 2024-02-11 00:55:57,350 (trainer:762) INFO: 41epoch:train:12301-12400batch: iter_time=9.075e-05, forward_time=0.339, loss_ctc=39.951, loss_att=44.184, acc=0.762, loss=42.914, backward_time=0.312, grad_norm=44.250, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.565e-04, train_time=1.365
+[gpub010:0/16] 2024-02-11 00:57:52,183 (trainer:762) INFO: 41epoch:train:12401-12500batch: iter_time=8.563e-05, forward_time=0.309, loss_ctc=50.484, loss_att=49.381, acc=0.763, loss=49.712, backward_time=0.308, grad_norm=60.607, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.565e-04, train_time=1.148
+[gpub010:0/16] 2024-02-11 00:58:12,222 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-11 00:58:31,639 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 00:58:35,161 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba84838940>)
+[gpub010:0/16] 2024-02-11 00:58:35,161 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-11 00:58:35,217 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 01:05:52,457 (trainer:762) INFO: 41epoch:train:12501-12600batch: iter_time=3.463, forward_time=0.351, loss_ctc=44.787, loss_att=46.147, acc=0.750, loss=45.739, backward_time=0.303, grad_norm=42.408, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.565e-04, train_time=4.802
+[gpub010:0/16] 2024-02-11 01:07:50,456 (trainer:762) INFO: 41epoch:train:12601-12700batch: iter_time=7.935e-05, forward_time=0.321, loss_ctc=49.537, loss_att=47.301, acc=0.760, loss=47.971, backward_time=0.306, grad_norm=44.834, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.565e-04, train_time=1.180
+[gpub010:0/16] 2024-02-11 01:10:10,142 (trainer:762) INFO: 41epoch:train:12701-12800batch: iter_time=8.520e-05, forward_time=0.316, loss_ctc=42.348, loss_att=49.760, acc=0.755, loss=47.536, backward_time=0.316, grad_norm=43.690, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.565e-04, train_time=1.397
+[gpub010:0/16] 2024-02-11 01:12:07,835 (trainer:762) INFO: 41epoch:train:12801-12900batch: iter_time=8.305e-05, forward_time=0.301, loss_ctc=41.717, loss_att=39.855, acc=0.773, loss=40.414, backward_time=0.306, grad_norm=37.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.565e-04, train_time=1.176
+[gpub010:0/16] 2024-02-11 01:14:18,372 (trainer:762) INFO: 41epoch:train:12901-13000batch: iter_time=8.827e-05, forward_time=0.306, loss_ctc=45.529, loss_att=45.959, acc=0.770, loss=45.830, backward_time=0.305, grad_norm=42.638, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.565e-04, train_time=1.305
+[gpub010:0/16] 2024-02-11 01:16:31,159 (trainer:762) INFO: 41epoch:train:13001-13100batch: iter_time=8.153e-05, forward_time=0.325, loss_ctc=46.154, loss_att=39.685, acc=0.776, loss=41.626, backward_time=0.308, grad_norm=43.496, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.564e-04, train_time=1.328
+[gpub010:0/16] 2024-02-11 01:18:29,571 (trainer:762) INFO: 41epoch:train:13101-13200batch: iter_time=8.392e-05, forward_time=0.305, loss_ctc=43.214, loss_att=46.714, acc=0.747, loss=45.664, backward_time=0.300, grad_norm=42.539, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.564e-04, train_time=1.183
+[gpub010:0/16] 2024-02-11 01:20:51,976 (trainer:762) INFO: 41epoch:train:13201-13300batch: iter_time=7.984e-05, forward_time=0.333, loss_ctc=43.169, loss_att=41.122, acc=0.781, loss=41.736, backward_time=0.327, grad_norm=38.062, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.564e-04, train_time=1.424
+[gpub010:0/16] 2024-02-11 01:22:54,492 (trainer:762) INFO: 41epoch:train:13301-13400batch: iter_time=8.221e-05, forward_time=0.311, loss_ctc=44.292, loss_att=47.416, acc=0.755, loss=46.479, backward_time=0.310, grad_norm=43.672, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.564e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 01:24:56,566 (trainer:762) INFO: 41epoch:train:13401-13500batch: iter_time=8.359e-05, forward_time=0.308, loss_ctc=43.805, loss_att=46.313, acc=0.767, loss=45.561, backward_time=0.299, grad_norm=39.608, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.564e-04, train_time=1.220
+[gpub010:0/16] 2024-02-11 01:27:06,751 (trainer:762) INFO: 41epoch:train:13501-13600batch: iter_time=9.652e-05, forward_time=0.329, loss_ctc=45.821, loss_att=48.672, acc=0.745, loss=47.817, backward_time=0.307, grad_norm=50.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.564e-04, train_time=1.302
+[gpub010:0/16] 2024-02-11 01:29:17,830 (trainer:762) INFO: 41epoch:train:13601-13700batch: iter_time=8.404e-05, forward_time=0.311, loss_ctc=48.331, loss_att=46.286, acc=0.751, loss=46.899, backward_time=0.297, grad_norm=54.203, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.564e-04, train_time=1.311
+[gpub010:0/16] 2024-02-11 01:30:39,725 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-11 01:30:58,886 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 01:31:02,344 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba801dfb80>)
+[gpub010:0/16] 2024-02-11 01:31:02,344 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-11 01:31:02,388 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 01:36:51,855 (trainer:762) INFO: 41epoch:train:13701-13800batch: iter_time=3.328, forward_time=0.309, loss_ctc=47.008, loss_att=46.423, acc=0.757, loss=46.598, backward_time=0.297, grad_norm=44.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.564e-04, train_time=4.540
+[gpub010:0/16] 2024-02-11 01:39:01,683 (trainer:762) INFO: 41epoch:train:13801-13900batch: iter_time=8.658e-05, forward_time=0.336, loss_ctc=46.132, loss_att=43.991, acc=0.768, loss=44.634, backward_time=0.309, grad_norm=41.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.563e-04, train_time=1.297
+[gpub010:0/16] 2024-02-11 01:41:04,104 (trainer:762) INFO: 41epoch:train:13901-14000batch: iter_time=8.672e-05, forward_time=0.308, loss_ctc=41.416, loss_att=42.779, acc=0.777, loss=42.370, backward_time=0.302, grad_norm=39.875, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.563e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 01:43:00,220 (trainer:762) INFO: 41epoch:train:14001-14100batch: iter_time=8.793e-05, forward_time=0.312, loss_ctc=44.840, loss_att=49.054, acc=0.766, loss=47.790, backward_time=0.302, grad_norm=43.976, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.563e-04, train_time=1.161
+[gpub010:0/16] 2024-02-11 01:45:21,437 (trainer:762) INFO: 41epoch:train:14101-14200batch: iter_time=9.610e-05, forward_time=0.321, loss_ctc=44.379, loss_att=44.426, acc=0.776, loss=44.412, backward_time=0.312, grad_norm=42.471, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.563e-04, train_time=1.412
+[gpub010:0/16] 2024-02-11 01:47:19,765 (trainer:762) INFO: 41epoch:train:14201-14300batch: iter_time=9.408e-05, forward_time=0.309, loss_ctc=44.048, loss_att=42.652, acc=0.772, loss=43.071, backward_time=0.304, grad_norm=40.971, clip=100.000, loss_scale=7.529e+33, optim_step_time=0.094, optim0_lr0=1.563e-04, train_time=1.183
+[gpub010:0/16] 2024-02-11 01:49:19,621 (trainer:762) INFO: 41epoch:train:14301-14400batch: iter_time=9.441e-05, forward_time=0.326, loss_ctc=46.776, loss_att=40.493, acc=0.780, loss=42.378, backward_time=0.302, grad_norm=42.716, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.563e-04, train_time=1.199
+[gpub010:0/16] 2024-02-11 01:51:42,883 (trainer:762) INFO: 41epoch:train:14401-14500batch: iter_time=9.767e-05, forward_time=0.329, loss_ctc=41.367, loss_att=41.241, acc=0.775, loss=41.279, backward_time=0.310, grad_norm=41.128, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.563e-04, train_time=1.432
+[gpub010:0/16] 2024-02-11 01:53:51,810 (trainer:762) INFO: 41epoch:train:14501-14600batch: iter_time=8.899e-05, forward_time=0.310, loss_ctc=45.379, loss_att=48.013, acc=0.777, loss=47.223, backward_time=0.306, grad_norm=43.069, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.563e-04, train_time=1.289
+[gpub010:0/16] 2024-02-11 01:56:04,822 (trainer:762) INFO: 41epoch:train:14601-14700batch: iter_time=8.628e-05, forward_time=0.300, loss_ctc=43.290, loss_att=42.411, acc=0.773, loss=42.675, backward_time=0.298, grad_norm=38.228, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.562e-04, train_time=1.328
+[gpub010:0/16] 2024-02-11 01:58:15,932 (trainer:762) INFO: 41epoch:train:14701-14800batch: iter_time=9.261e-05, forward_time=0.315, loss_ctc=49.901, loss_att=50.069, acc=0.763, loss=50.018, backward_time=0.302, grad_norm=47.890, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.562e-04, train_time=1.313
+[gpub010:0/16] 2024-02-11 01:59:33,266 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 02:00:24,563 (trainer:762) INFO: 41epoch:train:14801-14900batch: iter_time=8.792e-05, forward_time=0.333, loss_ctc=39.536, loss_att=44.222, acc=0.761, loss=42.816, backward_time=0.322, grad_norm=43.411, clip=100.000, loss_scale=8.077e+33, optim_step_time=0.096, optim0_lr0=1.562e-04, train_time=1.286
+[gpub010:0/16] 2024-02-11 02:00:47,454 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 02:02:33,332 (trainer:762) INFO: 41epoch:train:14901-15000batch: iter_time=9.346e-05, forward_time=0.315, loss_ctc=51.170, loss_att=49.525, acc=0.767, loss=50.018, backward_time=0.304, grad_norm=63.571, clip=100.000, loss_scale=3.042e+33, optim_step_time=0.098, optim0_lr0=1.562e-04, train_time=1.288
+[gpub010:0/16] 2024-02-11 02:40:52,914 (trainer:361) INFO: 41epoch results: [train] iter_time=0.258, forward_time=0.318, loss_ctc=45.625, loss_att=45.425, acc=0.764, loss=45.485, backward_time=0.311, grad_norm=44.193, clip=100.000, loss_scale=5.791e+33, optim_step_time=0.096, optim0_lr0=1.572e-04, train_time=1.552, time=6 hours, 28 minutes and 21.76 seconds, total_count=645000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=35.638, cer_ctc=0.182, loss_att=37.594, acc=0.693, cer=0.317, wer=0.986, loss=37.007, time=37 minutes and 54.45 seconds, total_count=200853, gpu_max_cached_mem_GB=42.092
+[gpub010:0/16] 2024-02-11 02:41:01,736 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-02-11 02:41:01,827 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/36epoch.pth
+[gpub010:0/16] 2024-02-11 02:41:01,828 (trainer:290) INFO: 42/45epoch started. Estimated time to finish: 1 day, 5 hours and 36 minutes
+[gpub010:0/16] 2024-02-11 02:41:01,838 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-11 02:41:20,365 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 02:41:23,734 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1e83ee560>)
+[gpub010:0/16] 2024-02-11 02:41:23,734 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-11 02:41:23,737 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 02:47:54,564 (trainer:762) INFO: 42epoch:train:1-100batch: iter_time=2.999, forward_time=0.319, loss_ctc=50.314, loss_att=43.194, acc=0.754, loss=45.330, backward_time=0.300, grad_norm=49.466, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.562e-04, train_time=4.127
+[gpub010:0/16] 2024-02-11 02:49:57,121 (trainer:762) INFO: 42epoch:train:101-200batch: iter_time=8.272e-05, forward_time=0.297, loss_ctc=48.437, loss_att=50.456, acc=0.743, loss=49.851, backward_time=0.301, grad_norm=47.423, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.562e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 02:52:03,884 (trainer:762) INFO: 42epoch:train:201-300batch: iter_time=8.204e-05, forward_time=0.304, loss_ctc=44.721, loss_att=44.783, acc=0.774, loss=44.764, backward_time=0.301, grad_norm=40.985, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.562e-04, train_time=1.267
+[gpub010:0/16] 2024-02-11 02:54:14,857 (trainer:762) INFO: 42epoch:train:301-400batch: iter_time=9.269e-05, forward_time=0.368, loss_ctc=47.482, loss_att=50.788, acc=0.746, loss=49.796, backward_time=0.316, grad_norm=48.065, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.561e-04, train_time=1.310
+[gpub010:0/16] 2024-02-11 02:56:22,632 (trainer:762) INFO: 42epoch:train:401-500batch: iter_time=8.626e-05, forward_time=0.310, loss_ctc=47.250, loss_att=52.238, acc=0.733, loss=50.742, backward_time=0.296, grad_norm=42.694, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.561e-04, train_time=1.277
+[gpub010:0/16] 2024-02-11 02:58:26,427 (trainer:762) INFO: 42epoch:train:501-600batch: iter_time=8.388e-05, forward_time=0.291, loss_ctc=43.298, loss_att=46.578, acc=0.756, loss=45.594, backward_time=0.295, grad_norm=40.967, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.561e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 03:00:35,306 (trainer:762) INFO: 42epoch:train:601-700batch: iter_time=9.109e-05, forward_time=0.351, loss_ctc=34.458, loss_att=29.856, acc=0.799, loss=31.237, backward_time=0.330, grad_norm=34.214, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.107, optim0_lr0=1.561e-04, train_time=1.288
+[gpub010:0/16] 2024-02-11 03:02:41,397 (trainer:762) INFO: 42epoch:train:701-800batch: iter_time=8.150e-05, forward_time=0.303, loss_ctc=50.324, loss_att=48.098, acc=0.758, loss=48.766, backward_time=0.299, grad_norm=47.062, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.561e-04, train_time=1.261
+[gpub010:0/16] 2024-02-11 03:04:55,746 (trainer:762) INFO: 42epoch:train:801-900batch: iter_time=9.491e-05, forward_time=0.302, loss_ctc=51.431, loss_att=50.549, acc=0.736, loss=50.813, backward_time=0.303, grad_norm=52.228, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.561e-04, train_time=1.343
+[gpub010:0/16] 2024-02-11 03:06:59,181 (trainer:762) INFO: 42epoch:train:901-1000batch: iter_time=8.663e-05, forward_time=0.300, loss_ctc=38.532, loss_att=37.955, acc=0.772, loss=38.128, backward_time=0.309, grad_norm=39.989, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.561e-04, train_time=1.234
+[gpub010:0/16] 2024-02-11 03:09:11,478 (trainer:762) INFO: 42epoch:train:1001-1100batch: iter_time=8.417e-05, forward_time=0.337, loss_ctc=39.960, loss_att=38.206, acc=0.774, loss=38.732, backward_time=0.300, grad_norm=39.844, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.561e-04, train_time=1.323
+[gpub010:0/16] 2024-02-11 03:11:15,513 (trainer:762) INFO: 42epoch:train:1101-1200batch: iter_time=8.882e-05, forward_time=0.307, loss_ctc=48.728, loss_att=49.477, acc=0.750, loss=49.252, backward_time=0.299, grad_norm=55.643, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.560e-04, train_time=1.240
+[gpub010:0/16] 2024-02-11 03:12:37,892 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-11 03:12:56,981 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 03:13:00,765 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc216c0f760>)
+[gpub010:0/16] 2024-02-11 03:13:00,765 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-11 03:13:00,768 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 03:18:43,523 (trainer:762) INFO: 42epoch:train:1201-1300batch: iter_time=3.233, forward_time=0.346, loss_ctc=51.192, loss_att=43.441, acc=0.759, loss=45.766, backward_time=0.303, grad_norm=47.487, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.560e-04, train_time=4.480
+[gpub010:0/16] 2024-02-11 03:20:43,686 (trainer:762) INFO: 42epoch:train:1301-1400batch: iter_time=7.756e-05, forward_time=0.292, loss_ctc=48.970, loss_att=41.152, acc=0.768, loss=43.497, backward_time=0.296, grad_norm=45.512, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.560e-04, train_time=1.201
+[gpub010:0/16] 2024-02-11 03:22:52,033 (trainer:762) INFO: 42epoch:train:1401-1500batch: iter_time=8.211e-05, forward_time=0.303, loss_ctc=47.151, loss_att=52.923, acc=0.762, loss=51.191, backward_time=0.309, grad_norm=42.886, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.560e-04, train_time=1.283
+[gpub010:0/16] 2024-02-11 03:25:02,225 (trainer:762) INFO: 42epoch:train:1501-1600batch: iter_time=8.445e-05, forward_time=0.367, loss_ctc=42.021, loss_att=45.394, acc=0.769, loss=44.382, backward_time=0.321, grad_norm=40.097, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.560e-04, train_time=1.301
+[gpub010:0/16] 2024-02-11 03:27:01,932 (trainer:762) INFO: 42epoch:train:1601-1700batch: iter_time=8.498e-05, forward_time=0.296, loss_ctc=44.888, loss_att=49.039, acc=0.751, loss=47.794, backward_time=0.304, grad_norm=44.513, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.560e-04, train_time=1.197
+[gpub010:0/16] 2024-02-11 03:29:16,554 (trainer:762) INFO: 42epoch:train:1701-1800batch: iter_time=8.540e-05, forward_time=0.307, loss_ctc=45.215, loss_att=48.510, acc=0.747, loss=47.522, backward_time=0.300, grad_norm=41.629, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.560e-04, train_time=1.346
+[gpub010:0/16] 2024-02-11 03:31:20,429 (trainer:762) INFO: 42epoch:train:1801-1900batch: iter_time=8.687e-05, forward_time=0.331, loss_ctc=41.624, loss_att=42.568, acc=0.798, loss=42.285, backward_time=0.335, grad_norm=37.431, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.100, optim0_lr0=1.560e-04, train_time=1.239
+[gpub010:0/16] 2024-02-11 03:33:22,624 (trainer:762) INFO: 42epoch:train:1901-2000batch: iter_time=8.936e-05, forward_time=0.292, loss_ctc=41.095, loss_att=37.136, acc=0.775, loss=38.324, backward_time=0.299, grad_norm=39.318, clip=100.000, loss_scale=4.725e+33, optim_step_time=0.093, optim0_lr0=1.559e-04, train_time=1.222
+[gpub010:0/16] 2024-02-11 03:35:30,488 (trainer:762) INFO: 42epoch:train:2001-2100batch: iter_time=8.591e-05, forward_time=0.317, loss_ctc=49.682, loss_att=46.817, acc=0.764, loss=47.677, backward_time=0.307, grad_norm=46.150, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.559e-04, train_time=1.279
+[gpub010:0/16] 2024-02-11 03:37:40,549 (trainer:762) INFO: 42epoch:train:2101-2200batch: iter_time=8.035e-05, forward_time=0.293, loss_ctc=44.471, loss_att=46.130, acc=0.752, loss=45.632, backward_time=0.294, grad_norm=41.521, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.559e-04, train_time=1.301
+[gpub010:0/16] 2024-02-11 03:39:51,140 (trainer:762) INFO: 42epoch:train:2201-2300batch: iter_time=8.357e-05, forward_time=0.354, loss_ctc=40.292, loss_att=40.981, acc=0.776, loss=40.775, backward_time=0.306, grad_norm=39.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.559e-04, train_time=1.306
+[gpub010:0/16] 2024-02-11 03:41:57,860 (trainer:762) INFO: 42epoch:train:2301-2400batch: iter_time=9.241e-05, forward_time=0.302, loss_ctc=42.379, loss_att=42.500, acc=0.768, loss=42.464, backward_time=0.319, grad_norm=43.345, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.559e-04, train_time=1.267
+[gpub010:0/16] 2024-02-11 03:44:04,902 (trainer:762) INFO: 42epoch:train:2401-2500batch: iter_time=8.448e-05, forward_time=0.292, loss_ctc=51.180, loss_att=45.769, acc=0.769, loss=47.392, backward_time=0.296, grad_norm=46.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.559e-04, train_time=1.270
+[gpub010:0/16] 2024-02-11 03:44:25,001 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-11 03:44:44,029 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 03:44:47,536 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1aeaf4070>)
+[gpub010:0/16] 2024-02-11 03:44:47,536 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-11 03:44:47,600 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 03:51:20,028 (trainer:762) INFO: 42epoch:train:2501-2600batch: iter_time=3.185, forward_time=0.325, loss_ctc=49.205, loss_att=42.128, acc=0.759, loss=44.251, backward_time=0.301, grad_norm=46.789, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.559e-04, train_time=4.351
+[gpub010:0/16] 2024-02-11 03:53:26,309 (trainer:762) INFO: 42epoch:train:2601-2700batch: iter_time=8.243e-05, forward_time=0.315, loss_ctc=47.777, loss_att=50.194, acc=0.746, loss=49.469, backward_time=0.303, grad_norm=48.858, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.559e-04, train_time=1.263
+[gpub010:0/16] 2024-02-11 03:55:35,577 (trainer:762) INFO: 42epoch:train:2701-2800batch: iter_time=8.294e-05, forward_time=0.292, loss_ctc=44.218, loss_att=44.519, acc=0.776, loss=44.429, backward_time=0.297, grad_norm=40.413, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.558e-04, train_time=1.292
+[gpub010:0/16] 2024-02-11 03:57:40,795 (trainer:762) INFO: 42epoch:train:2801-2900batch: iter_time=3.083e-04, forward_time=0.366, loss_ctc=46.653, loss_att=50.416, acc=0.749, loss=49.287, backward_time=0.314, grad_norm=45.420, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.558e-04, train_time=1.252
+[gpub010:0/16] 2024-02-11 03:59:47,224 (trainer:762) INFO: 42epoch:train:2901-3000batch: iter_time=8.432e-05, forward_time=0.306, loss_ctc=46.383, loss_att=51.283, acc=0.738, loss=49.813, backward_time=0.320, grad_norm=42.252, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.558e-04, train_time=1.264
+[gpub010:0/16] 2024-02-11 04:01:55,914 (trainer:762) INFO: 42epoch:train:3001-3100batch: iter_time=8.559e-05, forward_time=0.325, loss_ctc=42.667, loss_att=45.776, acc=0.762, loss=44.843, backward_time=0.294, grad_norm=39.557, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.558e-04, train_time=1.286
+[gpub010:0/16] 2024-02-11 04:04:00,437 (trainer:762) INFO: 42epoch:train:3101-3200batch: iter_time=8.702e-05, forward_time=0.321, loss_ctc=34.550, loss_att=29.720, acc=0.802, loss=31.169, backward_time=0.316, grad_norm=33.823, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.558e-04, train_time=1.245
+[gpub010:0/16] 2024-02-11 04:06:11,351 (trainer:762) INFO: 42epoch:train:3201-3300batch: iter_time=8.947e-05, forward_time=0.311, loss_ctc=49.700, loss_att=47.598, acc=0.760, loss=48.228, backward_time=0.317, grad_norm=45.046, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.558e-04, train_time=1.309
+[gpub010:0/16] 2024-02-11 04:08:01,436 (trainer:762) INFO: 42epoch:train:3301-3400batch: iter_time=8.307e-05, forward_time=0.291, loss_ctc=50.695, loss_att=49.884, acc=0.740, loss=50.127, backward_time=0.297, grad_norm=48.657, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.558e-04, train_time=1.101
+[gpub010:0/16] 2024-02-11 04:10:19,818 (trainer:762) INFO: 42epoch:train:3401-3500batch: iter_time=9.563e-05, forward_time=0.323, loss_ctc=37.926, loss_att=37.779, acc=0.773, loss=37.823, backward_time=0.333, grad_norm=39.266, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.558e-04, train_time=1.384
+[gpub010:0/16] 2024-02-11 04:12:20,507 (trainer:762) INFO: 42epoch:train:3501-3600batch: iter_time=8.939e-05, forward_time=0.289, loss_ctc=39.495, loss_att=37.880, acc=0.777, loss=38.365, backward_time=0.294, grad_norm=39.211, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.557e-04, train_time=1.206
+[gpub010:0/16] 2024-02-11 04:14:19,756 (trainer:762) INFO: 42epoch:train:3601-3700batch: iter_time=9.394e-05, forward_time=0.304, loss_ctc=47.380, loss_att=48.540, acc=0.753, loss=48.192, backward_time=0.317, grad_norm=50.563, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.557e-04, train_time=1.193
+[gpub010:0/16] 2024-02-11 04:15:33,721 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-11 04:15:52,813 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 04:15:56,340 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbccf4cb7f0>)
+[gpub010:0/16] 2024-02-11 04:15:56,340 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-11 04:15:56,359 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 04:21:45,916 (trainer:762) INFO: 42epoch:train:3701-3800batch: iter_time=3.241, forward_time=0.344, loss_ctc=49.291, loss_att=42.052, acc=0.766, loss=44.223, backward_time=0.306, grad_norm=46.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.557e-04, train_time=4.461
+[gpub010:0/16] 2024-02-11 04:23:46,830 (trainer:762) INFO: 42epoch:train:3801-3900batch: iter_time=8.732e-05, forward_time=0.317, loss_ctc=48.113, loss_att=40.915, acc=0.773, loss=43.074, backward_time=0.301, grad_norm=42.324, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.557e-04, train_time=1.208
+[gpub010:0/16] 2024-02-11 04:26:03,676 (trainer:762) INFO: 42epoch:train:3901-4000batch: iter_time=8.250e-05, forward_time=0.294, loss_ctc=46.197, loss_att=52.586, acc=0.765, loss=50.669, backward_time=0.299, grad_norm=42.762, clip=100.000, loss_scale=9.450e+33, optim_step_time=0.093, optim0_lr0=1.557e-04, train_time=1.369
+[gpub010:0/16] 2024-02-11 04:28:06,306 (trainer:762) INFO: 42epoch:train:4001-4100batch: iter_time=8.376e-05, forward_time=0.327, loss_ctc=41.915, loss_att=45.119, acc=0.772, loss=44.158, backward_time=0.346, grad_norm=38.663, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.557e-04, train_time=1.226
+[gpub010:0/16] 2024-02-11 04:30:09,221 (trainer:762) INFO: 42epoch:train:4101-4200batch: iter_time=8.087e-05, forward_time=0.294, loss_ctc=44.360, loss_att=48.660, acc=0.755, loss=47.370, backward_time=0.298, grad_norm=43.623, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.557e-04, train_time=1.229
+[gpub010:0/16] 2024-02-11 04:32:28,571 (trainer:762) INFO: 42epoch:train:4201-4300batch: iter_time=8.293e-05, forward_time=0.311, loss_ctc=44.750, loss_att=48.182, acc=0.751, loss=47.152, backward_time=0.298, grad_norm=40.771, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.557e-04, train_time=1.392
+[gpub010:0/16] 2024-02-11 04:34:19,733 (trainer:762) INFO: 42epoch:train:4301-4400batch: iter_time=8.100e-05, forward_time=0.294, loss_ctc=41.547, loss_att=42.701, acc=0.800, loss=42.355, backward_time=0.300, grad_norm=35.804, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.556e-04, train_time=1.113
+[gpub010:0/16] 2024-02-11 04:36:37,663 (trainer:762) INFO: 42epoch:train:4401-4500batch: iter_time=5.485e-04, forward_time=0.339, loss_ctc=41.069, loss_att=37.180, acc=0.777, loss=38.346, backward_time=0.338, grad_norm=40.610, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.556e-04, train_time=1.379
+[gpub010:0/16] 2024-02-11 04:38:48,497 (trainer:762) INFO: 42epoch:train:4501-4600batch: iter_time=8.402e-05, forward_time=0.290, loss_ctc=48.961, loss_att=45.912, acc=0.766, loss=46.827, backward_time=0.296, grad_norm=45.106, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.556e-04, train_time=1.308
+[gpub010:0/16] 2024-02-11 04:40:50,313 (trainer:762) INFO: 42epoch:train:4601-4700batch: iter_time=8.635e-05, forward_time=0.313, loss_ctc=43.750, loss_att=46.043, acc=0.755, loss=45.355, backward_time=0.309, grad_norm=41.169, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.556e-04, train_time=1.217
+[gpub010:0/16] 2024-02-11 04:42:18,223 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 04:43:17,144 (trainer:762) INFO: 42epoch:train:4701-4800batch: iter_time=8.091e-05, forward_time=0.370, loss_ctc=40.369, loss_att=41.159, acc=0.775, loss=40.922, backward_time=0.314, grad_norm=39.008, clip=100.000, loss_scale=8.182e+33, optim_step_time=0.096, optim0_lr0=1.556e-04, train_time=1.469
+[gpub010:0/16] 2024-02-11 04:45:14,184 (trainer:762) INFO: 42epoch:train:4801-4900batch: iter_time=7.948e-05, forward_time=0.291, loss_ctc=41.546, loss_att=41.892, acc=0.772, loss=41.788, backward_time=0.296, grad_norm=43.341, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.556e-04, train_time=1.170
+[gpub010:0/16] 2024-02-11 04:47:39,119 (trainer:762) INFO: 42epoch:train:4901-5000batch: iter_time=7.799e-05, forward_time=0.304, loss_ctc=50.367, loss_att=45.149, acc=0.770, loss=46.714, backward_time=0.309, grad_norm=46.946, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.556e-04, train_time=1.449
+[gpub010:0/16] 2024-02-11 04:47:59,149 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-11 04:48:18,084 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 04:48:21,588 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbc20b9ac80>)
+[gpub010:0/16] 2024-02-11 04:48:21,588 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-11 04:48:21,625 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 04:55:04,851 (trainer:762) INFO: 42epoch:train:5001-5100batch: iter_time=3.305, forward_time=0.337, loss_ctc=49.484, loss_att=41.131, acc=0.769, loss=43.637, backward_time=0.303, grad_norm=47.107, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.556e-04, train_time=4.457
+[gpub010:0/16] 2024-02-11 04:57:10,456 (trainer:762) INFO: 42epoch:train:5101-5200batch: iter_time=7.956e-05, forward_time=0.292, loss_ctc=46.950, loss_att=51.058, acc=0.751, loss=49.825, backward_time=0.298, grad_norm=46.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.555e-04, train_time=1.256
+[gpub010:0/16] 2024-02-11 04:59:36,225 (trainer:762) INFO: 42epoch:train:5201-5300batch: iter_time=7.806e-05, forward_time=0.375, loss_ctc=43.880, loss_att=43.683, acc=0.786, loss=43.742, backward_time=0.326, grad_norm=40.582, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.555e-04, train_time=1.457
+[gpub010:0/16] 2024-02-11 05:01:29,164 (trainer:762) INFO: 42epoch:train:5301-5400batch: iter_time=7.960e-05, forward_time=0.293, loss_ctc=46.075, loss_att=50.092, acc=0.765, loss=48.887, backward_time=0.299, grad_norm=44.731, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.555e-04, train_time=1.129
+[gpub010:0/16] 2024-02-11 05:03:35,133 (trainer:762) INFO: 42epoch:train:5401-5500batch: iter_time=8.054e-05, forward_time=0.316, loss_ctc=46.238, loss_att=52.396, acc=0.743, loss=50.549, backward_time=0.308, grad_norm=41.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.555e-04, train_time=1.260
+[gpub010:0/16] 2024-02-11 05:05:38,847 (trainer:762) INFO: 42epoch:train:5501-5600batch: iter_time=2.459e-04, forward_time=0.329, loss_ctc=42.086, loss_att=46.765, acc=0.766, loss=45.361, backward_time=0.326, grad_norm=39.032, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.555e-04, train_time=1.236
+[gpub010:0/16] 2024-02-11 05:07:47,077 (trainer:762) INFO: 42epoch:train:5601-5700batch: iter_time=8.277e-05, forward_time=0.290, loss_ctc=34.663, loss_att=29.760, acc=0.805, loss=31.231, backward_time=0.293, grad_norm=33.840, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.555e-04, train_time=1.283
+[gpub010:0/16] 2024-02-11 05:09:59,878 (trainer:762) INFO: 42epoch:train:5701-5800batch: iter_time=8.219e-05, forward_time=0.306, loss_ctc=49.123, loss_att=46.509, acc=0.774, loss=47.293, backward_time=0.317, grad_norm=45.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.555e-04, train_time=1.328
+[gpub010:0/16] 2024-02-11 05:12:01,835 (trainer:762) INFO: 42epoch:train:5801-5900batch: iter_time=2.047e-04, forward_time=0.358, loss_ctc=49.686, loss_att=49.997, acc=0.742, loss=49.904, backward_time=0.323, grad_norm=47.300, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.555e-04, train_time=1.219
+[gpub010:0/16] 2024-02-11 05:14:05,038 (trainer:762) INFO: 42epoch:train:5901-6000batch: iter_time=8.616e-05, forward_time=0.290, loss_ctc=37.468, loss_att=37.237, acc=0.786, loss=37.306, backward_time=0.294, grad_norm=37.125, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.554e-04, train_time=1.232
+[gpub010:0/16] 2024-02-11 05:16:13,814 (trainer:762) INFO: 42epoch:train:6001-6100batch: iter_time=9.314e-05, forward_time=0.300, loss_ctc=39.507, loss_att=38.095, acc=0.780, loss=38.518, backward_time=0.311, grad_norm=40.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.554e-04, train_time=1.288
+[gpub010:0/16] 2024-02-11 05:18:13,574 (trainer:762) INFO: 42epoch:train:6101-6200batch: iter_time=2.558e-04, forward_time=0.344, loss_ctc=46.923, loss_att=48.418, acc=0.759, loss=47.970, backward_time=0.328, grad_norm=48.594, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.554e-04, train_time=1.197
+[gpub010:0/16] 2024-02-11 05:19:27,765 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-11 05:19:47,320 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 05:19:50,909 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba410a75b0>)
+[gpub010:0/16] 2024-02-11 05:19:50,909 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-11 05:19:50,914 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 05:24:51,756 (trainer:762) INFO: 42epoch:train:6201-6300batch: iter_time=2.836, forward_time=0.300, loss_ctc=49.705, loss_att=42.853, acc=0.761, loss=44.908, backward_time=0.297, grad_norm=45.960, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.554e-04, train_time=3.982
+[gpub010:0/16] 2024-02-11 05:27:06,351 (trainer:762) INFO: 42epoch:train:6301-6400batch: iter_time=8.228e-05, forward_time=0.341, loss_ctc=48.421, loss_att=41.899, acc=0.766, loss=43.856, backward_time=0.336, grad_norm=45.536, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.554e-04, train_time=1.346
+[gpub010:0/16] 2024-02-11 05:29:00,583 (trainer:762) INFO: 42epoch:train:6401-6500batch: iter_time=8.160e-05, forward_time=0.293, loss_ctc=46.482, loss_att=51.548, acc=0.760, loss=50.028, backward_time=0.300, grad_norm=45.632, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.554e-04, train_time=1.142
+[gpub010:0/16] 2024-02-11 05:31:15,952 (trainer:762) INFO: 42epoch:train:6501-6600batch: iter_time=8.499e-05, forward_time=0.303, loss_ctc=41.496, loss_att=45.187, acc=0.768, loss=44.080, backward_time=0.308, grad_norm=40.406, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.554e-04, train_time=1.353
+[gpub010:0/16] 2024-02-11 05:33:16,879 (trainer:762) INFO: 42epoch:train:6601-6700batch: iter_time=8.320e-05, forward_time=0.353, loss_ctc=43.837, loss_att=47.683, acc=0.751, loss=46.529, backward_time=0.316, grad_norm=46.509, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.554e-04, train_time=1.209
+[gpub010:0/16] 2024-02-11 05:35:30,389 (trainer:762) INFO: 42epoch:train:6701-6800batch: iter_time=8.088e-05, forward_time=0.302, loss_ctc=44.214, loss_att=47.995, acc=0.742, loss=46.861, backward_time=0.296, grad_norm=41.079, clip=100.000, loss_scale=7.373e+33, optim_step_time=0.094, optim0_lr0=1.553e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 05:37:23,030 (trainer:762) INFO: 42epoch:train:6801-6900batch: iter_time=8.060e-05, forward_time=0.302, loss_ctc=41.111, loss_att=42.107, acc=0.799, loss=41.808, backward_time=0.304, grad_norm=38.131, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.553e-04, train_time=1.127
+[gpub010:0/16] 2024-02-11 05:39:36,790 (trainer:762) INFO: 42epoch:train:6901-7000batch: iter_time=8.056e-05, forward_time=0.343, loss_ctc=40.584, loss_att=36.992, acc=0.775, loss=38.070, backward_time=0.337, grad_norm=41.373, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.553e-04, train_time=1.337
+[gpub010:0/16] 2024-02-11 05:41:43,260 (trainer:762) INFO: 42epoch:train:7001-7100batch: iter_time=7.913e-05, forward_time=0.296, loss_ctc=48.599, loss_att=45.668, acc=0.762, loss=46.547, backward_time=0.305, grad_norm=44.498, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.553e-04, train_time=1.265
+[gpub010:0/16] 2024-02-11 05:43:45,445 (trainer:762) INFO: 42epoch:train:7101-7200batch: iter_time=7.876e-05, forward_time=0.290, loss_ctc=43.681, loss_att=45.981, acc=0.748, loss=45.291, backward_time=0.295, grad_norm=42.485, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.553e-04, train_time=1.222
+[gpub010:0/16] 2024-02-11 05:46:07,864 (trainer:762) INFO: 42epoch:train:7201-7300batch: iter_time=1.979e-04, forward_time=0.331, loss_ctc=39.949, loss_att=41.565, acc=0.767, loss=41.080, backward_time=0.329, grad_norm=39.206, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.553e-04, train_time=1.424
+[gpub010:0/16] 2024-02-11 05:48:12,569 (trainer:762) INFO: 42epoch:train:7301-7400batch: iter_time=7.602e-05, forward_time=0.298, loss_ctc=41.923, loss_att=40.998, acc=0.776, loss=41.275, backward_time=0.307, grad_norm=43.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.553e-04, train_time=1.246
+[gpub010:0/16] 2024-02-11 05:49:57,678 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 05:50:21,467 (trainer:762) INFO: 42epoch:train:7401-7500batch: iter_time=7.452e-05, forward_time=0.295, loss_ctc=50.276, loss_att=44.425, acc=0.769, loss=46.180, backward_time=0.298, grad_norm=45.691, clip=100.000, loss_scale=9.283e+33, optim_step_time=0.093, optim0_lr0=1.553e-04, train_time=1.289
+[gpub010:0/16] 2024-02-11 05:50:41,517 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-11 05:51:00,743 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 05:51:04,535 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba2a3abf40>)
+[gpub010:0/16] 2024-02-11 05:51:04,535 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-11 05:51:04,539 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 05:57:41,673 (trainer:762) INFO: 42epoch:train:7501-7600batch: iter_time=3.215, forward_time=0.360, loss_ctc=48.410, loss_att=42.548, acc=0.768, loss=44.307, backward_time=0.305, grad_norm=46.542, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.552e-04, train_time=4.402
+[gpub010:0/16] 2024-02-11 05:59:36,016 (trainer:762) INFO: 42epoch:train:7601-7700batch: iter_time=7.929e-05, forward_time=0.295, loss_ctc=46.803, loss_att=51.277, acc=0.751, loss=49.935, backward_time=0.311, grad_norm=44.120, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.552e-04, train_time=1.143
+[gpub010:0/16] 2024-02-11 06:01:05,960 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 06:02:01,400 (trainer:762) INFO: 42epoch:train:7701-7800batch: iter_time=8.301e-05, forward_time=0.370, loss_ctc=43.674, loss_att=44.064, acc=0.786, loss=43.947, backward_time=0.316, grad_norm=40.280, clip=100.000, loss_scale=4.196e+33, optim_step_time=0.110, optim0_lr0=1.552e-04, train_time=1.454
+[gpub010:0/16] 2024-02-11 06:04:05,221 (trainer:762) INFO: 42epoch:train:7801-7900batch: iter_time=8.108e-05, forward_time=0.294, loss_ctc=45.849, loss_att=49.981, acc=0.764, loss=48.742, backward_time=0.299, grad_norm=45.696, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.552e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 06:06:11,443 (trainer:762) INFO: 42epoch:train:7901-8000batch: iter_time=8.316e-05, forward_time=0.299, loss_ctc=46.312, loss_att=52.562, acc=0.743, loss=50.687, backward_time=0.307, grad_norm=42.625, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.552e-04, train_time=1.262
+[gpub010:0/16] 2024-02-11 06:08:25,498 (trainer:762) INFO: 42epoch:train:8001-8100batch: iter_time=8.532e-05, forward_time=0.368, loss_ctc=42.183, loss_att=46.741, acc=0.768, loss=45.374, backward_time=0.334, grad_norm=41.502, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.552e-04, train_time=1.340
+[gpub010:0/16] 2024-02-11 06:10:35,873 (trainer:762) INFO: 42epoch:train:8101-8200batch: iter_time=8.269e-05, forward_time=0.290, loss_ctc=34.234, loss_att=29.560, acc=0.805, loss=30.962, backward_time=0.294, grad_norm=32.657, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.552e-04, train_time=1.304
+[gpub010:0/16] 2024-02-11 06:12:35,121 (trainer:762) INFO: 42epoch:train:8201-8300batch: iter_time=8.328e-05, forward_time=0.299, loss_ctc=49.012, loss_att=46.333, acc=0.774, loss=47.137, backward_time=0.302, grad_norm=45.920, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.552e-04, train_time=1.192
+[gpub010:0/16] 2024-02-11 06:15:08,797 (trainer:762) INFO: 42epoch:train:8301-8400batch: iter_time=8.762e-05, forward_time=0.400, loss_ctc=49.613, loss_att=50.942, acc=0.741, loss=50.543, backward_time=0.326, grad_norm=48.122, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.551e-04, train_time=1.537
+[gpub010:0/16] 2024-02-11 06:16:58,982 (trainer:762) INFO: 42epoch:train:8401-8500batch: iter_time=8.822e-05, forward_time=0.295, loss_ctc=37.470, loss_att=37.232, acc=0.787, loss=37.303, backward_time=0.304, grad_norm=37.752, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.551e-04, train_time=1.102
+[gpub010:0/16] 2024-02-11 06:19:12,166 (trainer:762) INFO: 42epoch:train:8501-8600batch: iter_time=8.885e-05, forward_time=0.292, loss_ctc=39.233, loss_att=37.952, acc=0.782, loss=38.337, backward_time=0.297, grad_norm=39.874, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.551e-04, train_time=1.330
+[gpub010:0/16] 2024-02-11 06:21:30,211 (trainer:762) INFO: 42epoch:train:8601-8700batch: iter_time=9.088e-05, forward_time=0.344, loss_ctc=46.425, loss_att=48.240, acc=0.759, loss=47.696, backward_time=0.335, grad_norm=49.762, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.551e-04, train_time=1.382
+[gpub010:0/16] 2024-02-11 06:22:47,070 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-11 06:23:06,342 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 06:23:10,104 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba8cafb610>)
+[gpub010:0/16] 2024-02-11 06:23:10,105 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-11 06:23:10,108 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 06:28:47,593 (trainer:762) INFO: 42epoch:train:8701-8800batch: iter_time=3.095, forward_time=0.298, loss_ctc=49.195, loss_att=42.239, acc=0.763, loss=44.325, backward_time=0.296, grad_norm=44.998, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.551e-04, train_time=4.374
+[gpub010:0/16] 2024-02-11 06:31:06,526 (trainer:762) INFO: 42epoch:train:8801-8900batch: iter_time=8.241e-05, forward_time=0.353, loss_ctc=47.899, loss_att=40.821, acc=0.770, loss=42.944, backward_time=0.315, grad_norm=45.185, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.551e-04, train_time=1.389
+[gpub010:0/16] 2024-02-11 06:32:58,061 (trainer:762) INFO: 42epoch:train:8901-9000batch: iter_time=8.349e-05, forward_time=0.295, loss_ctc=45.751, loss_att=50.194, acc=0.763, loss=48.861, backward_time=0.300, grad_norm=42.858, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.551e-04, train_time=1.115
+[gpub010:0/16] 2024-02-11 06:34:59,404 (trainer:762) INFO: 42epoch:train:9001-9100batch: iter_time=8.162e-05, forward_time=0.301, loss_ctc=41.264, loss_att=44.689, acc=0.768, loss=43.661, backward_time=0.299, grad_norm=41.389, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.551e-04, train_time=1.214
+[gpub010:0/16] 2024-02-11 06:37:14,096 (trainer:762) INFO: 42epoch:train:9101-9200batch: iter_time=8.253e-05, forward_time=0.291, loss_ctc=43.569, loss_att=47.148, acc=0.753, loss=46.074, backward_time=0.296, grad_norm=44.175, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.550e-04, train_time=1.347
+[gpub010:0/16] 2024-02-11 06:39:25,949 (trainer:762) INFO: 42epoch:train:9201-9300batch: iter_time=8.378e-05, forward_time=0.397, loss_ctc=44.426, loss_att=47.966, acc=0.742, loss=46.904, backward_time=0.312, grad_norm=42.337, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.550e-04, train_time=1.318
+[gpub010:0/16] 2024-02-11 06:41:50,028 (trainer:762) INFO: 42epoch:train:9301-9400batch: iter_time=8.283e-05, forward_time=0.299, loss_ctc=41.096, loss_att=41.755, acc=0.800, loss=41.557, backward_time=0.301, grad_norm=35.569, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.550e-04, train_time=1.440
+[gpub010:0/16] 2024-02-11 06:43:40,929 (trainer:762) INFO: 42epoch:train:9401-9500batch: iter_time=8.540e-05, forward_time=0.290, loss_ctc=40.374, loss_att=37.104, acc=0.776, loss=38.085, backward_time=0.296, grad_norm=42.572, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.550e-04, train_time=1.110
+[gpub010:0/16] 2024-02-11 06:45:52,293 (trainer:762) INFO: 42epoch:train:9501-9600batch: iter_time=8.351e-05, forward_time=0.365, loss_ctc=48.441, loss_att=45.259, acc=0.765, loss=46.214, backward_time=0.310, grad_norm=44.214, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.550e-04, train_time=1.313
+[gpub010:0/16] 2024-02-11 06:48:13,076 (trainer:762) INFO: 42epoch:train:9601-9700batch: iter_time=8.158e-05, forward_time=0.297, loss_ctc=43.404, loss_att=45.322, acc=0.751, loss=44.747, backward_time=0.299, grad_norm=44.128, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.550e-04, train_time=1.407
+[gpub010:0/16] 2024-02-11 06:50:14,393 (trainer:762) INFO: 42epoch:train:9701-9800batch: iter_time=1.003e-04, forward_time=0.291, loss_ctc=39.867, loss_att=40.924, acc=0.770, loss=40.607, backward_time=0.295, grad_norm=40.776, clip=100.000, loss_scale=3.583e+33, optim_step_time=0.094, optim0_lr0=1.550e-04, train_time=1.213
+[gpub010:0/16] 2024-02-11 06:52:16,372 (trainer:762) INFO: 42epoch:train:9801-9900batch: iter_time=8.008e-05, forward_time=0.290, loss_ctc=41.044, loss_att=40.621, acc=0.778, loss=40.748, backward_time=0.295, grad_norm=43.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.550e-04, train_time=1.220
+[gpub010:0/16] 2024-02-11 06:54:40,359 (trainer:762) INFO: 42epoch:train:9901-10000batch: iter_time=8.256e-05, forward_time=0.408, loss_ctc=49.397, loss_att=44.200, acc=0.769, loss=45.759, backward_time=0.312, grad_norm=46.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.114, optim0_lr0=1.549e-04, train_time=1.439
+[gpub010:0/16] 2024-02-11 06:55:00,387 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-11 06:55:19,588 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 06:55:23,142 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbaab49ba60>)
+[gpub010:0/16] 2024-02-11 06:55:23,142 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-11 06:55:23,187 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 07:01:34,296 (trainer:762) INFO: 42epoch:train:10001-10100batch: iter_time=2.960, forward_time=0.372, loss_ctc=48.058, loss_att=41.876, acc=0.769, loss=43.731, backward_time=0.308, grad_norm=45.292, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.549e-04, train_time=4.139
+[gpub010:0/16] 2024-02-11 07:03:47,641 (trainer:762) INFO: 42epoch:train:10101-10200batch: iter_time=9.043e-05, forward_time=0.292, loss_ctc=46.409, loss_att=51.173, acc=0.752, loss=49.744, backward_time=0.296, grad_norm=45.531, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.549e-04, train_time=1.333
+[gpub010:0/16] 2024-02-11 07:06:03,958 (trainer:762) INFO: 42epoch:train:10201-10300batch: iter_time=9.577e-05, forward_time=0.389, loss_ctc=43.332, loss_att=43.887, acc=0.787, loss=43.720, backward_time=0.314, grad_norm=40.342, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.549e-04, train_time=1.363
+[gpub010:0/16] 2024-02-11 07:08:14,179 (trainer:762) INFO: 42epoch:train:10301-10400batch: iter_time=8.928e-05, forward_time=0.295, loss_ctc=45.559, loss_att=50.332, acc=0.765, loss=48.900, backward_time=0.299, grad_norm=44.836, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.549e-04, train_time=1.302
+[gpub010:0/16] 2024-02-11 07:10:18,435 (trainer:762) INFO: 42epoch:train:10401-10500batch: iter_time=8.901e-05, forward_time=0.373, loss_ctc=46.297, loss_att=52.218, acc=0.743, loss=50.441, backward_time=0.337, grad_norm=43.676, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.549e-04, train_time=1.242
+[gpub010:0/16] 2024-02-11 07:12:35,080 (trainer:762) INFO: 42epoch:train:10501-10600batch: iter_time=9.574e-05, forward_time=0.301, loss_ctc=41.768, loss_att=46.147, acc=0.769, loss=44.833, backward_time=0.301, grad_norm=41.149, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.549e-04, train_time=1.366
+[gpub010:0/16] 2024-02-11 07:14:37,335 (trainer:762) INFO: 42epoch:train:10601-10700batch: iter_time=8.656e-05, forward_time=0.294, loss_ctc=33.884, loss_att=29.326, acc=0.806, loss=30.694, backward_time=0.304, grad_norm=31.487, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.549e-04, train_time=1.222
+[gpub010:0/16] 2024-02-11 07:16:52,273 (trainer:762) INFO: 42epoch:train:10701-10800batch: iter_time=8.846e-05, forward_time=0.335, loss_ctc=48.838, loss_att=46.716, acc=0.773, loss=47.353, backward_time=0.334, grad_norm=43.846, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.548e-04, train_time=1.349
+[gpub010:0/16] 2024-02-11 07:19:00,692 (trainer:762) INFO: 42epoch:train:10801-10900batch: iter_time=8.598e-05, forward_time=0.306, loss_ctc=49.117, loss_att=50.024, acc=0.743, loss=49.752, backward_time=0.301, grad_norm=48.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.548e-04, train_time=1.284
+[gpub010:0/16] 2024-02-11 07:21:14,239 (trainer:762) INFO: 42epoch:train:10901-11000batch: iter_time=9.952e-05, forward_time=0.367, loss_ctc=37.245, loss_att=37.388, acc=0.787, loss=37.345, backward_time=0.324, grad_norm=35.528, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.548e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 07:23:16,235 (trainer:762) INFO: 42epoch:train:11001-11100batch: iter_time=8.587e-05, forward_time=0.289, loss_ctc=38.943, loss_att=37.486, acc=0.783, loss=37.923, backward_time=0.294, grad_norm=37.268, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.548e-04, train_time=1.220
+[gpub010:0/16] 2024-02-11 07:25:35,555 (trainer:762) INFO: 42epoch:train:11101-11200batch: iter_time=8.667e-05, forward_time=0.307, loss_ctc=46.463, loss_att=48.492, acc=0.760, loss=47.883, backward_time=0.307, grad_norm=50.647, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.548e-04, train_time=1.393
+[gpub010:0/16] 2024-02-11 07:27:05,374 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-11 07:27:24,816 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 07:27:28,333 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbd54f35900>)
+[gpub010:0/16] 2024-02-11 07:27:28,333 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-11 07:27:28,346 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 07:33:05,852 (trainer:762) INFO: 42epoch:train:11201-11300batch: iter_time=3.090, forward_time=0.435, loss_ctc=49.628, loss_att=42.224, acc=0.763, loss=44.445, backward_time=0.314, grad_norm=46.950, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.548e-04, train_time=4.503
+[gpub010:0/16] 2024-02-11 07:35:15,215 (trainer:762) INFO: 42epoch:train:11301-11400batch: iter_time=7.898e-05, forward_time=0.329, loss_ctc=47.632, loss_att=40.497, acc=0.771, loss=42.638, backward_time=0.351, grad_norm=46.409, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.548e-04, train_time=1.293
+[gpub010:0/16] 2024-02-11 07:37:40,597 (trainer:762) INFO: 42epoch:train:11401-11500batch: iter_time=7.855e-05, forward_time=0.312, loss_ctc=46.106, loss_att=50.813, acc=0.762, loss=49.400, backward_time=0.301, grad_norm=43.231, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.548e-04, train_time=1.454
+[gpub010:0/16] 2024-02-11 07:39:39,830 (trainer:762) INFO: 42epoch:train:11501-11600batch: iter_time=8.264e-05, forward_time=0.373, loss_ctc=40.676, loss_att=44.726, acc=0.768, loss=43.511, backward_time=0.309, grad_norm=41.084, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.547e-04, train_time=1.193
+[gpub010:0/16] 2024-02-11 07:41:55,067 (trainer:762) INFO: 42epoch:train:11601-11700batch: iter_time=8.957e-05, forward_time=0.297, loss_ctc=43.551, loss_att=47.193, acc=0.754, loss=46.100, backward_time=0.300, grad_norm=45.360, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.547e-04, train_time=1.351
+[gpub010:0/16] 2024-02-11 07:43:25,835 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 07:43:51,331 (trainer:762) INFO: 42epoch:train:11701-11800batch: iter_time=8.434e-05, forward_time=0.290, loss_ctc=44.364, loss_att=47.982, acc=0.741, loss=46.897, backward_time=0.294, grad_norm=41.136, clip=100.000, loss_scale=6.084e+33, optim_step_time=0.093, optim0_lr0=1.547e-04, train_time=1.163
+[gpub010:0/16] 2024-02-11 07:46:07,045 (trainer:762) INFO: 42epoch:train:11801-11900batch: iter_time=1.010e-04, forward_time=0.396, loss_ctc=40.906, loss_att=41.835, acc=0.800, loss=41.556, backward_time=0.309, grad_norm=36.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.547e-04, train_time=1.357
+[gpub010:0/16] 2024-02-11 07:48:22,687 (trainer:762) INFO: 42epoch:train:11901-12000batch: iter_time=8.774e-05, forward_time=0.306, loss_ctc=40.438, loss_att=36.347, acc=0.779, loss=37.574, backward_time=0.295, grad_norm=40.619, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.547e-04, train_time=1.355
+[gpub010:0/16] 2024-02-11 07:50:26,632 (trainer:762) INFO: 42epoch:train:12001-12100batch: iter_time=8.441e-05, forward_time=0.358, loss_ctc=48.195, loss_att=45.097, acc=0.764, loss=46.027, backward_time=0.363, grad_norm=42.681, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.547e-04, train_time=1.241
+[gpub010:0/16] 2024-02-11 07:52:38,629 (trainer:762) INFO: 42epoch:train:12101-12200batch: iter_time=8.538e-05, forward_time=0.300, loss_ctc=43.145, loss_att=44.979, acc=0.752, loss=44.429, backward_time=0.299, grad_norm=42.788, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.547e-04, train_time=1.318
+[gpub010:0/16] 2024-02-11 07:55:01,761 (trainer:762) INFO: 42epoch:train:12201-12300batch: iter_time=8.731e-05, forward_time=0.384, loss_ctc=39.482, loss_att=40.546, acc=0.771, loss=40.227, backward_time=0.306, grad_norm=40.642, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.547e-04, train_time=1.433
+[gpub010:0/16] 2024-02-11 07:56:50,466 (trainer:762) INFO: 42epoch:train:12301-12400batch: iter_time=8.194e-05, forward_time=0.296, loss_ctc=41.335, loss_att=40.576, acc=0.777, loss=40.804, backward_time=0.297, grad_norm=45.060, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.546e-04, train_time=1.086
+[gpub010:0/16] 2024-02-11 07:59:16,063 (trainer:762) INFO: 42epoch:train:12401-12500batch: iter_time=7.984e-05, forward_time=0.318, loss_ctc=50.235, loss_att=44.072, acc=0.769, loss=45.921, backward_time=0.298, grad_norm=45.234, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.546e-04, train_time=1.457
+[gpub010:0/16] 2024-02-11 07:59:36,090 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-11 07:59:55,675 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 07:59:59,979 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba70203610>)
+[gpub010:0/16] 2024-02-11 08:00:00,009 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-11 08:00:00,026 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 08:06:25,753 (trainer:762) INFO: 42epoch:train:12501-12600batch: iter_time=3.115, forward_time=0.345, loss_ctc=48.401, loss_att=40.172, acc=0.766, loss=42.641, backward_time=0.304, grad_norm=46.207, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.546e-04, train_time=4.295
+[gpub010:0/16] 2024-02-11 08:08:54,317 (trainer:762) INFO: 42epoch:train:12601-12700batch: iter_time=8.373e-05, forward_time=0.301, loss_ctc=46.257, loss_att=48.171, acc=0.755, loss=47.597, backward_time=0.299, grad_norm=44.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.546e-04, train_time=1.487
+[gpub010:0/16] 2024-02-11 08:10:55,771 (trainer:762) INFO: 42epoch:train:12701-12800batch: iter_time=8.105e-05, forward_time=0.363, loss_ctc=43.675, loss_att=43.884, acc=0.780, loss=43.822, backward_time=0.313, grad_norm=39.208, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.546e-04, train_time=1.213
+[gpub010:0/16] 2024-02-11 08:13:25,027 (trainer:762) INFO: 42epoch:train:12801-12900batch: iter_time=4.204e-04, forward_time=0.306, loss_ctc=45.633, loss_att=49.595, acc=0.754, loss=48.406, backward_time=0.297, grad_norm=44.603, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.546e-04, train_time=1.494
+[gpub010:0/16] 2024-02-11 08:15:29,873 (trainer:762) INFO: 42epoch:train:12901-13000batch: iter_time=8.256e-05, forward_time=0.340, loss_ctc=45.713, loss_att=50.318, acc=0.745, loss=48.937, backward_time=0.321, grad_norm=42.261, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.546e-04, train_time=1.247
+[gpub010:0/16] 2024-02-11 08:17:53,004 (trainer:762) INFO: 42epoch:train:13001-13100batch: iter_time=8.087e-05, forward_time=0.290, loss_ctc=41.981, loss_att=45.691, acc=0.763, loss=44.578, backward_time=0.294, grad_norm=39.275, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.546e-04, train_time=1.432
+[gpub010:0/16] 2024-02-11 08:20:07,883 (trainer:762) INFO: 42epoch:train:13101-13200batch: iter_time=4.479e-04, forward_time=0.375, loss_ctc=33.974, loss_att=29.447, acc=0.806, loss=30.805, backward_time=0.308, grad_norm=32.416, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.546e-04, train_time=1.349
+[gpub010:0/16] 2024-02-11 08:21:56,713 (trainer:762) INFO: 42epoch:train:13201-13300batch: iter_time=8.439e-05, forward_time=0.292, loss_ctc=48.340, loss_att=45.890, acc=0.769, loss=46.625, backward_time=0.298, grad_norm=43.949, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.545e-04, train_time=1.087
+[gpub010:0/16] 2024-02-11 08:24:38,975 (trainer:762) INFO: 42epoch:train:13301-13400batch: iter_time=2.736e-04, forward_time=0.358, loss_ctc=49.127, loss_att=49.800, acc=0.743, loss=49.598, backward_time=0.305, grad_norm=47.411, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.545e-04, train_time=1.623
+[gpub010:0/16] 2024-02-11 08:26:26,473 (trainer:762) INFO: 42epoch:train:13401-13500batch: iter_time=8.902e-05, forward_time=0.288, loss_ctc=37.070, loss_att=36.875, acc=0.779, loss=36.933, backward_time=0.294, grad_norm=37.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.545e-04, train_time=1.075
+[gpub010:0/16] 2024-02-11 08:28:50,435 (trainer:762) INFO: 42epoch:train:13501-13600batch: iter_time=9.414e-05, forward_time=0.314, loss_ctc=38.511, loss_att=38.146, acc=0.778, loss=38.255, backward_time=0.296, grad_norm=37.554, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.545e-04, train_time=1.439
+[gpub010:0/16] 2024-02-11 08:31:09,742 (trainer:762) INFO: 42epoch:train:13601-13700batch: iter_time=8.544e-05, forward_time=0.385, loss_ctc=46.614, loss_att=47.629, acc=0.756, loss=47.324, backward_time=0.313, grad_norm=49.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.545e-04, train_time=1.393
+[gpub010:0/16] 2024-02-11 08:32:32,735 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-11 08:32:51,974 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 08:32:55,817 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba8fa44a90>)
+[gpub010:0/16] 2024-02-11 08:32:55,817 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-11 08:32:55,821 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 08:38:19,755 (trainer:762) INFO: 42epoch:train:13701-13800batch: iter_time=3.027, forward_time=0.303, loss_ctc=49.156, loss_att=42.337, acc=0.767, loss=44.383, backward_time=0.296, grad_norm=44.144, clip=100.000, loss_scale=6.231e+33, optim_step_time=0.094, optim0_lr0=1.545e-04, train_time=4.300
+[gpub010:0/16] 2024-02-11 08:40:44,740 (trainer:762) INFO: 42epoch:train:13801-13900batch: iter_time=8.189e-05, forward_time=0.339, loss_ctc=47.315, loss_att=40.464, acc=0.776, loss=42.519, backward_time=0.314, grad_norm=44.844, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.545e-04, train_time=1.449
+[gpub010:0/16] 2024-02-11 08:42:57,938 (trainer:762) INFO: 42epoch:train:13901-14000batch: iter_time=7.927e-05, forward_time=0.295, loss_ctc=45.818, loss_att=52.049, acc=0.767, loss=50.180, backward_time=0.300, grad_norm=42.110, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.545e-04, train_time=1.332
+[gpub010:0/16] 2024-02-11 08:45:03,359 (trainer:762) INFO: 42epoch:train:14001-14100batch: iter_time=8.363e-05, forward_time=0.313, loss_ctc=41.116, loss_att=44.630, acc=0.775, loss=43.575, backward_time=0.299, grad_norm=40.632, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.544e-04, train_time=1.253
+[gpub010:0/16] 2024-02-11 08:47:20,993 (trainer:762) INFO: 42epoch:train:14101-14200batch: iter_time=8.389e-05, forward_time=0.332, loss_ctc=43.411, loss_att=48.672, acc=0.757, loss=47.094, backward_time=0.319, grad_norm=45.079, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.544e-04, train_time=1.377
+[gpub010:0/16] 2024-02-11 08:49:23,560 (trainer:762) INFO: 42epoch:train:14201-14300batch: iter_time=8.162e-05, forward_time=0.289, loss_ctc=43.827, loss_att=47.959, acc=0.754, loss=46.719, backward_time=0.295, grad_norm=39.243, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.544e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 08:51:54,360 (trainer:762) INFO: 42epoch:train:14301-14400batch: iter_time=7.847e-05, forward_time=0.303, loss_ctc=40.883, loss_att=42.208, acc=0.802, loss=41.810, backward_time=0.302, grad_norm=36.839, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.544e-04, train_time=1.508
+[gpub010:0/16] 2024-02-11 08:53:57,110 (trainer:762) INFO: 42epoch:train:14401-14500batch: iter_time=2.534e-04, forward_time=0.356, loss_ctc=40.400, loss_att=36.813, acc=0.779, loss=37.889, backward_time=0.305, grad_norm=40.210, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.544e-04, train_time=1.227
+[gpub010:0/16] 2024-02-11 08:55:54,049 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 08:56:12,229 (trainer:762) INFO: 42epoch:train:14501-14600batch: iter_time=8.017e-05, forward_time=0.292, loss_ctc=48.189, loss_att=45.944, acc=0.769, loss=46.617, backward_time=0.297, grad_norm=43.000, clip=100.000, loss_scale=9.493e+33, optim_step_time=0.093, optim0_lr0=1.544e-04, train_time=1.352
+[gpub010:0/16] 2024-02-11 08:58:20,077 (trainer:762) INFO: 42epoch:train:14601-14700batch: iter_time=8.376e-05, forward_time=0.298, loss_ctc=42.983, loss_att=45.943, acc=0.757, loss=45.055, backward_time=0.304, grad_norm=44.344, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.544e-04, train_time=1.278
+[gpub010:0/16] 2024-02-11 09:00:38,038 (trainer:762) INFO: 42epoch:train:14701-14800batch: iter_time=8.362e-05, forward_time=0.357, loss_ctc=39.602, loss_att=40.371, acc=0.779, loss=40.140, backward_time=0.315, grad_norm=38.721, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.544e-04, train_time=1.380
+[gpub010:0/16] 2024-02-11 09:02:31,826 (trainer:762) INFO: 42epoch:train:14801-14900batch: iter_time=8.223e-05, forward_time=0.291, loss_ctc=41.232, loss_att=42.084, acc=0.773, loss=41.829, backward_time=0.297, grad_norm=43.124, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.543e-04, train_time=1.137
+[gpub010:0/16] 2024-02-11 09:04:50,158 (trainer:762) INFO: 42epoch:train:14901-15000batch: iter_time=8.725e-05, forward_time=0.307, loss_ctc=49.686, loss_att=44.464, acc=0.774, loss=46.031, backward_time=0.300, grad_norm=43.708, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.543e-04, train_time=1.383
+[gpub010:0/16] 2024-02-11 09:44:28,288 (trainer:361) INFO: 42epoch results: [train] iter_time=0.249, forward_time=0.320, loss_ctc=44.479, loss_att=44.367, acc=0.767, loss=44.400, backward_time=0.307, grad_norm=42.822, clip=100.000, loss_scale=5.337e+33, optim_step_time=0.095, optim0_lr0=1.553e-04, train_time=1.535, time=6 hours, 24 minutes and 12.71 seconds, total_count=660000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=35.666, cer_ctc=0.178, loss_att=37.943, acc=0.698, cer=0.292, wer=0.989, loss=37.260, time=39 minutes and 13.53 seconds, total_count=205524, gpu_max_cached_mem_GB=42.092
+[gpub010:0/16] 2024-02-11 09:44:37,474 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-02-11 09:44:37,569 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/37epoch.pth
+[gpub010:0/16] 2024-02-11 09:44:37,570 (trainer:290) INFO: 43/45epoch started. Estimated time to finish: 21 hours, 52 minutes and 0.11 seconds
+[gpub010:0/16] 2024-02-11 09:44:37,580 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-11 09:44:56,225 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 09:45:00,443 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc0a1f18310>)
+[gpub010:0/16] 2024-02-11 09:45:00,443 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-11 09:45:00,446 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 09:51:13,658 (trainer:762) INFO: 43epoch:train:1-100batch: iter_time=2.750, forward_time=0.310, loss_ctc=56.909, loss_att=59.341, acc=0.739, loss=58.611, backward_time=0.305, grad_norm=57.169, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.543e-04, train_time=3.960
+[gpub010:0/16] 2024-02-11 09:53:20,747 (trainer:762) INFO: 43epoch:train:101-200batch: iter_time=1.078e-04, forward_time=0.348, loss_ctc=50.376, loss_att=47.152, acc=0.754, loss=48.119, backward_time=0.311, grad_norm=48.572, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.543e-04, train_time=1.271
+[gpub010:0/16] 2024-02-11 09:55:30,443 (trainer:762) INFO: 43epoch:train:201-300batch: iter_time=2.329e-04, forward_time=0.304, loss_ctc=45.980, loss_att=43.168, acc=0.764, loss=44.012, backward_time=0.309, grad_norm=42.119, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.543e-04, train_time=1.296
+[gpub010:0/16] 2024-02-11 09:58:49,479 (trainer:762) INFO: 43epoch:train:301-400batch: iter_time=0.078, forward_time=0.304, loss_ctc=57.625, loss_att=49.287, acc=0.749, loss=51.789, backward_time=0.308, grad_norm=53.419, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.543e-04, train_time=1.991
+[gpub010:0/16] 2024-02-11 10:00:57,617 (trainer:762) INFO: 43epoch:train:401-500batch: iter_time=9.189e-05, forward_time=0.315, loss_ctc=54.170, loss_att=47.018, acc=0.761, loss=49.164, backward_time=0.326, grad_norm=46.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.543e-04, train_time=1.281
+[gpub010:0/16] 2024-02-11 10:03:09,341 (trainer:762) INFO: 43epoch:train:501-600batch: iter_time=2.010e-04, forward_time=0.306, loss_ctc=50.374, loss_att=50.500, acc=0.769, loss=50.462, backward_time=0.305, grad_norm=51.148, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.543e-04, train_time=1.316
+[gpub010:0/16] 2024-02-11 10:05:16,876 (trainer:762) INFO: 43epoch:train:601-700batch: iter_time=8.759e-05, forward_time=0.328, loss_ctc=42.486, loss_att=38.187, acc=0.792, loss=39.477, backward_time=0.329, grad_norm=40.331, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.542e-04, train_time=1.276
+[gpub010:0/16] 2024-02-11 10:07:25,710 (trainer:762) INFO: 43epoch:train:701-800batch: iter_time=8.193e-05, forward_time=0.303, loss_ctc=41.074, loss_att=39.698, acc=0.766, loss=40.111, backward_time=0.308, grad_norm=43.228, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.542e-04, train_time=1.288
+[gpub010:0/16] 2024-02-11 10:09:49,977 (trainer:762) INFO: 43epoch:train:801-900batch: iter_time=3.606e-04, forward_time=0.339, loss_ctc=57.079, loss_att=51.466, acc=0.774, loss=53.150, backward_time=0.321, grad_norm=45.499, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.542e-04, train_time=1.442
+[gpub010:0/16] 2024-02-11 10:11:52,765 (trainer:762) INFO: 43epoch:train:901-1000batch: iter_time=8.068e-05, forward_time=0.302, loss_ctc=43.300, loss_att=42.403, acc=0.767, loss=42.672, backward_time=0.312, grad_norm=42.359, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.542e-04, train_time=1.228
+[gpub010:0/16] 2024-02-11 10:14:04,418 (trainer:762) INFO: 43epoch:train:1001-1100batch: iter_time=7.525e-05, forward_time=0.325, loss_ctc=48.423, loss_att=50.032, acc=0.760, loss=49.549, backward_time=0.301, grad_norm=45.787, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.542e-04, train_time=1.316
+[gpub010:0/16] 2024-02-11 10:16:17,873 (trainer:762) INFO: 43epoch:train:1101-1200batch: iter_time=7.998e-05, forward_time=0.311, loss_ctc=42.509, loss_att=38.152, acc=0.774, loss=39.459, backward_time=0.322, grad_norm=43.998, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.542e-04, train_time=1.334
+[gpub010:0/16] 2024-02-11 10:17:41,422 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-11 10:18:01,134 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 10:18:04,636 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb229ecb730>)
+[gpub010:0/16] 2024-02-11 10:18:04,636 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-11 10:18:04,684 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 10:23:25,934 (trainer:762) INFO: 43epoch:train:1201-1300batch: iter_time=3.018, forward_time=0.360, loss_ctc=46.543, loss_att=49.842, acc=0.742, loss=48.853, backward_time=0.310, grad_norm=48.490, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.542e-04, train_time=4.279
+[gpub010:0/16] 2024-02-11 10:25:35,154 (trainer:762) INFO: 43epoch:train:1301-1400batch: iter_time=8.851e-05, forward_time=0.312, loss_ctc=53.325, loss_att=55.271, acc=0.731, loss=54.687, backward_time=0.316, grad_norm=56.449, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.542e-04, train_time=1.293
+[gpub010:0/16] 2024-02-11 10:27:44,652 (trainer:762) INFO: 43epoch:train:1401-1500batch: iter_time=8.081e-05, forward_time=0.295, loss_ctc=50.103, loss_att=46.074, acc=0.761, loss=47.282, backward_time=0.310, grad_norm=45.891, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.541e-04, train_time=1.295
+[gpub010:0/16] 2024-02-11 10:29:49,749 (trainer:762) INFO: 43epoch:train:1501-1600batch: iter_time=9.256e-05, forward_time=0.346, loss_ctc=44.242, loss_att=38.129, acc=0.757, loss=39.963, backward_time=0.300, grad_norm=41.403, clip=100.000, loss_scale=6.075e+33, optim_step_time=0.094, optim0_lr0=1.541e-04, train_time=1.251
+[gpub010:0/16] 2024-02-11 10:32:08,659 (trainer:762) INFO: 43epoch:train:1601-1700batch: iter_time=8.725e-05, forward_time=0.301, loss_ctc=62.761, loss_att=51.546, acc=0.731, loss=54.911, backward_time=0.314, grad_norm=54.125, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.541e-04, train_time=1.389
+[gpub010:0/16] 2024-02-11 10:34:02,673 (trainer:762) INFO: 43epoch:train:1701-1800batch: iter_time=9.123e-05, forward_time=0.292, loss_ctc=49.336, loss_att=51.266, acc=0.762, loss=50.687, backward_time=0.298, grad_norm=47.969, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.541e-04, train_time=1.140
+[gpub010:0/16] 2024-02-11 10:36:15,780 (trainer:762) INFO: 43epoch:train:1801-1900batch: iter_time=8.360e-05, forward_time=0.384, loss_ctc=42.617, loss_att=40.996, acc=0.776, loss=41.482, backward_time=0.305, grad_norm=41.121, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.541e-04, train_time=1.331
+[gpub010:0/16] 2024-02-11 10:38:25,021 (trainer:762) INFO: 43epoch:train:1901-2000batch: iter_time=7.772e-05, forward_time=0.304, loss_ctc=42.341, loss_att=37.944, acc=0.776, loss=39.263, backward_time=0.302, grad_norm=42.985, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.541e-04, train_time=1.292
+[gpub010:0/16] 2024-02-11 10:40:36,223 (trainer:762) INFO: 43epoch:train:2001-2100batch: iter_time=8.095e-05, forward_time=0.300, loss_ctc=46.492, loss_att=41.478, acc=0.773, loss=42.982, backward_time=0.309, grad_norm=42.132, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.541e-04, train_time=1.312
+[gpub010:0/16] 2024-02-11 10:42:42,610 (trainer:762) INFO: 43epoch:train:2101-2200batch: iter_time=9.058e-05, forward_time=0.352, loss_ctc=52.011, loss_att=50.441, acc=0.755, loss=50.912, backward_time=0.319, grad_norm=47.976, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.541e-04, train_time=1.264
+[gpub010:0/16] 2024-02-11 10:44:50,945 (trainer:762) INFO: 43epoch:train:2201-2300batch: iter_time=8.971e-05, forward_time=0.301, loss_ctc=42.058, loss_att=43.750, acc=0.768, loss=43.243, backward_time=0.305, grad_norm=42.870, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.540e-04, train_time=1.283
+[gpub010:0/16] 2024-02-11 10:46:55,246 (trainer:762) INFO: 43epoch:train:2301-2400batch: iter_time=8.806e-05, forward_time=0.301, loss_ctc=45.561, loss_att=43.685, acc=0.758, loss=44.247, backward_time=0.316, grad_norm=45.809, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.540e-04, train_time=1.242
+[gpub010:0/16] 2024-02-11 10:49:06,935 (trainer:762) INFO: 43epoch:train:2401-2500batch: iter_time=8.497e-05, forward_time=0.340, loss_ctc=41.626, loss_att=37.077, acc=0.778, loss=38.441, backward_time=0.308, grad_norm=38.840, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.540e-04, train_time=1.317
+[gpub010:0/16] 2024-02-11 10:49:26,962 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-11 10:49:46,395 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 10:49:49,940 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbcd434d4b0>)
+[gpub010:0/16] 2024-02-11 10:49:49,940 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-11 10:49:50,034 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 10:58:15,443 (trainer:762) INFO: 43epoch:train:2501-2600batch: iter_time=3.984, forward_time=0.356, loss_ctc=54.897, loss_att=59.625, acc=0.724, loss=58.207, backward_time=0.309, grad_norm=65.082, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.540e-04, train_time=5.485
+[gpub010:0/16] 2024-02-11 11:00:04,290 (trainer:762) INFO: 43epoch:train:2601-2700batch: iter_time=7.661e-05, forward_time=0.290, loss_ctc=48.301, loss_att=45.696, acc=0.753, loss=46.478, backward_time=0.295, grad_norm=44.750, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.540e-04, train_time=1.088
+[gpub010:0/16] 2024-02-11 11:02:28,905 (trainer:762) INFO: 43epoch:train:2701-2800batch: iter_time=8.802e-05, forward_time=0.408, loss_ctc=45.176, loss_att=41.393, acc=0.768, loss=42.528, backward_time=0.334, grad_norm=40.319, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.540e-04, train_time=1.446
+[gpub010:0/16] 2024-02-11 11:04:41,154 (trainer:762) INFO: 43epoch:train:2801-2900batch: iter_time=7.479e-05, forward_time=0.293, loss_ctc=56.105, loss_att=47.869, acc=0.744, loss=50.340, backward_time=0.297, grad_norm=50.685, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.540e-04, train_time=1.322
+[gpub010:0/16] 2024-02-11 11:05:04,721 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 11:06:43,685 (trainer:762) INFO: 43epoch:train:2901-3000batch: iter_time=7.793e-05, forward_time=0.290, loss_ctc=49.994, loss_att=44.527, acc=0.760, loss=46.167, backward_time=0.296, grad_norm=43.151, clip=100.000, loss_scale=6.294e+33, optim_step_time=0.093, optim0_lr0=1.540e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 11:09:00,883 (trainer:762) INFO: 43epoch:train:3001-3100batch: iter_time=4.025e-04, forward_time=0.405, loss_ctc=48.152, loss_att=48.244, acc=0.761, loss=48.216, backward_time=0.314, grad_norm=49.441, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.540e-04, train_time=1.372
+[gpub010:0/16] 2024-02-11 11:13:04,241 (trainer:762) INFO: 43epoch:train:3101-3200batch: iter_time=7.974e-05, forward_time=0.290, loss_ctc=41.407, loss_att=37.957, acc=0.786, loss=38.992, backward_time=0.294, grad_norm=41.279, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.539e-04, train_time=2.433
+[gpub010:0/16] 2024-02-11 11:15:00,844 (trainer:762) INFO: 43epoch:train:3201-3300batch: iter_time=7.574e-05, forward_time=0.290, loss_ctc=40.609, loss_att=39.614, acc=0.764, loss=39.913, backward_time=0.294, grad_norm=41.708, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.539e-04, train_time=1.166
+[gpub010:0/16] 2024-02-11 11:17:06,423 (trainer:762) INFO: 43epoch:train:3301-3400batch: iter_time=7.830e-05, forward_time=0.376, loss_ctc=56.399, loss_att=50.099, acc=0.770, loss=51.989, backward_time=0.361, grad_norm=44.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.539e-04, train_time=1.255
+[gpub010:0/16] 2024-02-11 11:19:16,902 (trainer:762) INFO: 43epoch:train:3401-3500batch: iter_time=8.792e-05, forward_time=0.290, loss_ctc=41.809, loss_att=40.984, acc=0.771, loss=41.231, backward_time=0.294, grad_norm=41.336, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.539e-04, train_time=1.305
+[gpub010:0/16] 2024-02-11 11:21:25,094 (trainer:762) INFO: 43epoch:train:3501-3600batch: iter_time=8.018e-05, forward_time=0.292, loss_ctc=46.994, loss_att=47.456, acc=0.765, loss=47.317, backward_time=0.298, grad_norm=43.290, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.539e-04, train_time=1.282
+[gpub010:0/16] 2024-02-11 11:23:38,215 (trainer:762) INFO: 43epoch:train:3601-3700batch: iter_time=9.136e-05, forward_time=0.398, loss_ctc=40.888, loss_att=36.875, acc=0.773, loss=38.079, backward_time=0.316, grad_norm=40.514, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.539e-04, train_time=1.331
+[gpub010:0/16] 2024-02-11 11:25:01,357 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-11 11:25:59,743 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 11:26:10,097 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbcd4ff0f40>)
+[gpub010:0/16] 2024-02-11 11:26:10,097 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-11 11:26:10,884 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 11:32:03,712 (trainer:762) INFO: 43epoch:train:3701-3800batch: iter_time=3.756, forward_time=0.293, loss_ctc=45.570, loss_att=48.032, acc=0.745, loss=47.294, backward_time=0.296, grad_norm=43.470, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.539e-04, train_time=5.055
+[gpub010:0/16] 2024-02-11 11:34:12,286 (trainer:762) INFO: 43epoch:train:3801-3900batch: iter_time=7.963e-05, forward_time=0.412, loss_ctc=51.765, loss_att=53.265, acc=0.737, loss=52.815, backward_time=0.313, grad_norm=52.158, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.539e-04, train_time=1.285
+[gpub010:0/16] 2024-02-11 11:36:08,092 (trainer:762) INFO: 43epoch:train:3901-4000batch: iter_time=8.051e-05, forward_time=0.293, loss_ctc=49.178, loss_att=44.282, acc=0.768, loss=45.751, backward_time=0.298, grad_norm=46.045, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.538e-04, train_time=1.158
+[gpub010:0/16] 2024-02-11 11:38:25,290 (trainer:762) INFO: 43epoch:train:4001-4100batch: iter_time=8.277e-05, forward_time=0.291, loss_ctc=43.756, loss_att=37.279, acc=0.762, loss=39.222, backward_time=0.295, grad_norm=42.347, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.538e-04, train_time=1.372
+[gpub010:0/16] 2024-02-11 11:40:35,740 (trainer:762) INFO: 43epoch:train:4101-4200batch: iter_time=8.067e-05, forward_time=0.403, loss_ctc=60.736, loss_att=50.052, acc=0.738, loss=53.257, backward_time=0.314, grad_norm=52.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.538e-04, train_time=1.304
+[gpub010:0/16] 2024-02-11 11:42:35,544 (trainer:762) INFO: 43epoch:train:4201-4300batch: iter_time=8.026e-05, forward_time=0.293, loss_ctc=48.217, loss_att=50.266, acc=0.766, loss=49.651, backward_time=0.297, grad_norm=49.704, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.538e-04, train_time=1.198
+[gpub010:0/16] 2024-02-11 11:44:48,995 (trainer:762) INFO: 43epoch:train:4301-4400batch: iter_time=8.009e-05, forward_time=0.291, loss_ctc=42.469, loss_att=40.789, acc=0.779, loss=41.293, backward_time=0.296, grad_norm=40.742, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.538e-04, train_time=1.334
+[gpub010:0/16] 2024-02-11 11:46:56,363 (trainer:762) INFO: 43epoch:train:4401-4500batch: iter_time=7.767e-05, forward_time=0.377, loss_ctc=41.397, loss_att=36.928, acc=0.782, loss=38.268, backward_time=0.323, grad_norm=43.811, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.538e-04, train_time=1.273
+[gpub010:0/16] 2024-02-11 11:49:04,312 (trainer:762) INFO: 43epoch:train:4501-4600batch: iter_time=8.170e-05, forward_time=0.292, loss_ctc=46.696, loss_att=40.998, acc=0.777, loss=42.707, backward_time=0.296, grad_norm=43.004, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.538e-04, train_time=1.279
+[gpub010:0/16] 2024-02-11 11:51:09,796 (trainer:762) INFO: 43epoch:train:4601-4700batch: iter_time=8.195e-05, forward_time=0.293, loss_ctc=50.954, loss_att=49.553, acc=0.757, loss=49.973, backward_time=0.299, grad_norm=44.894, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.538e-04, train_time=1.255
+[gpub010:0/16] 2024-02-11 11:53:15,683 (trainer:762) INFO: 43epoch:train:4701-4800batch: iter_time=8.407e-05, forward_time=0.376, loss_ctc=41.592, loss_att=43.409, acc=0.771, loss=42.864, backward_time=0.322, grad_norm=41.719, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.104, optim0_lr0=1.537e-04, train_time=1.254
+[gpub010:0/16] 2024-02-11 11:55:24,170 (trainer:762) INFO: 43epoch:train:4801-4900batch: iter_time=8.375e-05, forward_time=0.296, loss_ctc=45.084, loss_att=42.353, acc=0.763, loss=43.172, backward_time=0.295, grad_norm=45.219, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.537e-04, train_time=1.289
+[gpub010:0/16] 2024-02-11 11:57:25,773 (trainer:762) INFO: 43epoch:train:4901-5000batch: iter_time=8.380e-05, forward_time=0.291, loss_ctc=41.433, loss_att=36.209, acc=0.783, loss=37.776, backward_time=0.296, grad_norm=37.976, clip=100.000, loss_scale=9.242e+33, optim_step_time=0.093, optim0_lr0=1.537e-04, train_time=1.216
+[gpub010:0/16] 2024-02-11 11:57:45,896 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-11 11:58:04,996 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 11:58:08,525 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba73692b00>)
+[gpub010:0/16] 2024-02-11 11:58:08,525 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-11 11:58:08,530 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 12:03:55,687 (trainer:762) INFO: 43epoch:train:5001-5100batch: iter_time=2.710, forward_time=0.349, loss_ctc=54.394, loss_att=60.884, acc=0.737, loss=58.937, backward_time=0.313, grad_norm=54.518, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.537e-04, train_time=3.899
+[gpub010:0/16] 2024-02-11 12:06:03,818 (trainer:762) INFO: 43epoch:train:5101-5200batch: iter_time=8.386e-05, forward_time=0.290, loss_ctc=47.508, loss_att=46.010, acc=0.762, loss=46.459, backward_time=0.296, grad_norm=43.999, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.537e-04, train_time=1.281
+[gpub010:0/16] 2024-02-11 12:08:05,485 (trainer:762) INFO: 43epoch:train:5201-5300batch: iter_time=8.296e-05, forward_time=0.339, loss_ctc=44.853, loss_att=42.699, acc=0.770, loss=43.345, backward_time=0.351, grad_norm=39.345, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.537e-04, train_time=1.216
+[gpub010:0/16] 2024-02-11 12:10:18,290 (trainer:762) INFO: 43epoch:train:5301-5400batch: iter_time=8.560e-05, forward_time=0.294, loss_ctc=56.471, loss_att=48.701, acc=0.756, loss=51.032, backward_time=0.298, grad_norm=50.992, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.537e-04, train_time=1.327
+[gpub010:0/16] 2024-02-11 12:12:22,414 (trainer:762) INFO: 43epoch:train:5401-5500batch: iter_time=8.518e-05, forward_time=0.291, loss_ctc=49.615, loss_att=45.855, acc=0.768, loss=46.983, backward_time=0.297, grad_norm=46.842, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.537e-04, train_time=1.241
+[gpub010:0/16] 2024-02-11 12:14:44,986 (trainer:762) INFO: 43epoch:train:5501-5600batch: iter_time=8.595e-05, forward_time=0.412, loss_ctc=47.548, loss_att=49.820, acc=0.775, loss=49.138, backward_time=0.316, grad_norm=48.200, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.536e-04, train_time=1.426
+[gpub010:0/16] 2024-02-11 12:16:40,409 (trainer:762) INFO: 43epoch:train:5601-5700batch: iter_time=8.089e-05, forward_time=0.294, loss_ctc=41.246, loss_att=37.495, acc=0.798, loss=38.621, backward_time=0.297, grad_norm=38.507, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.536e-04, train_time=1.154
+[gpub010:0/16] 2024-02-11 12:18:46,910 (trainer:762) INFO: 43epoch:train:5701-5800batch: iter_time=8.125e-05, forward_time=0.291, loss_ctc=40.218, loss_att=39.268, acc=0.772, loss=39.553, backward_time=0.295, grad_norm=41.749, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.536e-04, train_time=1.265
+[gpub010:0/16] 2024-02-11 12:20:52,986 (trainer:762) INFO: 43epoch:train:5801-5900batch: iter_time=8.528e-05, forward_time=0.403, loss_ctc=55.366, loss_att=51.238, acc=0.777, loss=52.476, backward_time=0.316, grad_norm=46.831, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.536e-04, train_time=1.260
+[gpub010:0/16] 2024-02-11 12:22:53,081 (trainer:762) INFO: 43epoch:train:5901-6000batch: iter_time=8.017e-05, forward_time=0.294, loss_ctc=41.733, loss_att=41.860, acc=0.771, loss=41.822, backward_time=0.298, grad_norm=40.910, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.536e-04, train_time=1.201
+[gpub010:0/16] 2024-02-11 12:25:08,246 (trainer:762) INFO: 43epoch:train:6001-6100batch: iter_time=8.122e-05, forward_time=0.293, loss_ctc=46.584, loss_att=49.202, acc=0.766, loss=48.417, backward_time=0.298, grad_norm=47.305, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.536e-04, train_time=1.352
+[gpub010:0/16] 2024-02-11 12:26:48,777 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 12:27:14,630 (trainer:762) INFO: 43epoch:train:6101-6200batch: iter_time=6.166e-04, forward_time=0.373, loss_ctc=41.011, loss_att=37.524, acc=0.779, loss=38.570, backward_time=0.314, grad_norm=41.405, clip=100.000, loss_scale=9.126e+33, optim_step_time=0.108, optim0_lr0=1.536e-04, train_time=1.264
+[gpub010:0/16] 2024-02-11 12:28:35,252 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-11 12:28:54,252 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 12:28:58,112 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba8cc3b8b0>)
+[gpub010:0/16] 2024-02-11 12:28:58,112 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-11 12:28:58,116 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 12:34:20,094 (trainer:762) INFO: 43epoch:train:6201-6300batch: iter_time=2.985, forward_time=0.291, loss_ctc=45.484, loss_att=47.959, acc=0.755, loss=47.216, backward_time=0.297, grad_norm=44.998, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.536e-04, train_time=4.254
+[gpub010:0/16] 2024-02-11 12:36:41,870 (trainer:762) INFO: 43epoch:train:6301-6400batch: iter_time=7.763e-05, forward_time=0.386, loss_ctc=51.560, loss_att=53.611, acc=0.751, loss=52.996, backward_time=0.321, grad_norm=48.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.536e-04, train_time=1.417
+[gpub010:0/16] 2024-02-11 12:38:51,252 (trainer:762) INFO: 43epoch:train:6401-6500batch: iter_time=8.027e-05, forward_time=0.292, loss_ctc=49.237, loss_att=44.477, acc=0.777, loss=45.905, backward_time=0.298, grad_norm=45.151, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.535e-04, train_time=1.294
+[gpub010:0/16] 2024-02-11 12:40:55,836 (trainer:762) INFO: 43epoch:train:6501-6600batch: iter_time=8.443e-05, forward_time=0.291, loss_ctc=43.694, loss_att=37.904, acc=0.767, loss=39.641, backward_time=0.297, grad_norm=41.267, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.535e-04, train_time=1.245
+[gpub010:0/16] 2024-02-11 12:43:09,731 (trainer:762) INFO: 43epoch:train:6601-6700batch: iter_time=9.294e-05, forward_time=0.378, loss_ctc=60.931, loss_att=52.669, acc=0.745, loss=55.148, backward_time=0.318, grad_norm=55.935, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.535e-04, train_time=1.339
+[gpub010:0/16] 2024-02-11 12:45:28,822 (trainer:762) INFO: 43epoch:train:6701-6800batch: iter_time=8.945e-05, forward_time=0.293, loss_ctc=47.924, loss_att=50.907, acc=0.771, loss=50.012, backward_time=0.299, grad_norm=46.766, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.535e-04, train_time=1.391
+[gpub010:0/16] 2024-02-11 12:47:30,972 (trainer:762) INFO: 43epoch:train:6801-6900batch: iter_time=7.684e-05, forward_time=0.291, loss_ctc=41.580, loss_att=40.574, acc=0.796, loss=40.876, backward_time=0.297, grad_norm=37.252, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.535e-04, train_time=1.221
+[gpub010:0/16] 2024-02-11 12:49:50,963 (trainer:762) INFO: 43epoch:train:6901-7000batch: iter_time=7.985e-05, forward_time=0.367, loss_ctc=41.543, loss_att=37.129, acc=0.791, loss=38.453, backward_time=0.336, grad_norm=42.900, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.535e-04, train_time=1.400
+[gpub010:0/16] 2024-02-11 12:51:57,247 (trainer:762) INFO: 43epoch:train:7001-7100batch: iter_time=8.043e-05, forward_time=0.291, loss_ctc=45.514, loss_att=41.652, acc=0.780, loss=42.810, backward_time=0.297, grad_norm=41.771, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.535e-04, train_time=1.263
+[gpub010:0/16] 2024-02-11 12:54:32,630 (trainer:762) INFO: 43epoch:train:7101-7200batch: iter_time=9.482e-05, forward_time=0.363, loss_ctc=49.847, loss_att=49.208, acc=0.767, loss=49.400, backward_time=0.359, grad_norm=44.118, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.535e-04, train_time=1.553
+[gpub010:0/16] 2024-02-11 12:56:50,410 (trainer:762) INFO: 43epoch:train:7201-7300batch: iter_time=8.408e-05, forward_time=0.293, loss_ctc=41.010, loss_att=44.006, acc=0.774, loss=43.107, backward_time=0.298, grad_norm=42.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.534e-04, train_time=1.378
+[gpub010:0/16] 2024-02-11 12:58:54,311 (trainer:762) INFO: 43epoch:train:7301-7400batch: iter_time=3.015e-04, forward_time=0.369, loss_ctc=45.167, loss_att=43.443, acc=0.764, loss=43.960, backward_time=0.333, grad_norm=44.195, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.534e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 13:01:18,495 (trainer:762) INFO: 43epoch:train:7401-7500batch: iter_time=8.887e-05, forward_time=0.291, loss_ctc=41.190, loss_att=37.652, acc=0.786, loss=38.713, backward_time=0.295, grad_norm=38.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.534e-04, train_time=1.443
+[gpub010:0/16] 2024-02-11 13:01:38,522 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-11 13:01:58,156 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 13:02:01,952 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbaa4fe7610>)
+[gpub010:0/16] 2024-02-11 13:02:01,952 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-11 13:02:01,955 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 13:08:41,680 (trainer:762) INFO: 43epoch:train:7501-7600batch: iter_time=3.084, forward_time=0.291, loss_ctc=54.097, loss_att=61.622, acc=0.721, loss=59.364, backward_time=0.298, grad_norm=53.928, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.534e-04, train_time=4.431
+[gpub010:0/16] 2024-02-11 13:11:05,158 (trainer:762) INFO: 43epoch:train:7601-7700batch: iter_time=8.142e-05, forward_time=0.359, loss_ctc=47.382, loss_att=45.720, acc=0.753, loss=46.218, backward_time=0.358, grad_norm=47.839, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.534e-04, train_time=1.434
+[gpub010:0/16] 2024-02-11 13:13:18,515 (trainer:762) INFO: 43epoch:train:7701-7800batch: iter_time=8.174e-05, forward_time=0.290, loss_ctc=44.882, loss_att=41.567, acc=0.770, loss=42.562, backward_time=0.294, grad_norm=39.603, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.534e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 13:15:36,159 (trainer:762) INFO: 43epoch:train:7801-7900batch: iter_time=8.273e-05, forward_time=0.329, loss_ctc=55.118, loss_att=47.601, acc=0.748, loss=49.856, backward_time=0.384, grad_norm=49.993, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.534e-04, train_time=1.376
+[gpub010:0/16] 2024-02-11 13:17:51,302 (trainer:762) INFO: 43epoch:train:7901-8000batch: iter_time=8.742e-05, forward_time=0.291, loss_ctc=49.470, loss_att=44.028, acc=0.764, loss=45.661, backward_time=0.296, grad_norm=43.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.534e-04, train_time=1.351
+[gpub010:0/16] 2024-02-11 13:19:54,438 (trainer:762) INFO: 43epoch:train:8001-8100batch: iter_time=8.489e-05, forward_time=0.292, loss_ctc=47.556, loss_att=48.542, acc=0.762, loss=48.246, backward_time=0.297, grad_norm=49.126, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.533e-04, train_time=1.232
+[gpub010:0/16] 2024-02-11 13:22:02,270 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 13:22:07,854 (trainer:762) INFO: 43epoch:train:8101-8200batch: iter_time=8.971e-05, forward_time=0.413, loss_ctc=41.156, loss_att=37.970, acc=0.787, loss=38.926, backward_time=0.323, grad_norm=40.059, clip=100.000, loss_scale=6.136e+33, optim_step_time=0.095, optim0_lr0=1.533e-04, train_time=1.333
+[gpub010:0/16] 2024-02-11 13:24:19,263 (trainer:762) INFO: 43epoch:train:8201-8300batch: iter_time=8.300e-05, forward_time=0.289, loss_ctc=39.966, loss_att=38.775, acc=0.770, loss=39.132, backward_time=0.293, grad_norm=40.780, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.533e-04, train_time=1.315
+[gpub010:0/16] 2024-02-11 13:26:23,972 (trainer:762) INFO: 43epoch:train:8301-8400batch: iter_time=8.078e-05, forward_time=0.295, loss_ctc=55.063, loss_att=49.548, acc=0.772, loss=51.202, backward_time=0.301, grad_norm=44.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.533e-04, train_time=1.246
+[gpub010:0/16] 2024-02-11 13:28:37,460 (trainer:762) INFO: 43epoch:train:8401-8500batch: iter_time=8.068e-05, forward_time=0.352, loss_ctc=41.015, loss_att=41.487, acc=0.769, loss=41.345, backward_time=0.308, grad_norm=43.104, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.533e-04, train_time=1.336
+[gpub010:0/16] 2024-02-11 13:30:38,849 (trainer:762) INFO: 43epoch:train:8501-8600batch: iter_time=8.645e-05, forward_time=0.295, loss_ctc=46.737, loss_att=47.536, acc=0.764, loss=47.297, backward_time=0.299, grad_norm=44.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.533e-04, train_time=1.213
+[gpub010:0/16] 2024-02-11 13:32:40,310 (trainer:762) INFO: 43epoch:train:8601-8700batch: iter_time=7.861e-05, forward_time=0.310, loss_ctc=39.954, loss_att=35.966, acc=0.777, loss=37.162, backward_time=0.294, grad_norm=42.263, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.533e-04, train_time=1.215
+[gpub010:0/16] 2024-02-11 13:34:14,869 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-11 13:34:34,249 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 13:34:37,816 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba8c9df610>)
+[gpub010:0/16] 2024-02-11 13:34:37,816 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-11 13:34:37,846 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 13:40:35,499 (trainer:762) INFO: 43epoch:train:8701-8800batch: iter_time=3.363, forward_time=0.367, loss_ctc=44.844, loss_att=47.172, acc=0.749, loss=46.474, backward_time=0.306, grad_norm=42.462, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.533e-04, train_time=4.751
+[gpub010:0/16] 2024-02-11 13:43:05,284 (trainer:762) INFO: 43epoch:train:8801-8900batch: iter_time=5.359e-04, forward_time=0.364, loss_ctc=50.490, loss_att=53.078, acc=0.739, loss=52.301, backward_time=0.314, grad_norm=50.053, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.117, optim0_lr0=1.533e-04, train_time=1.498
+[gpub010:0/16] 2024-02-11 13:45:47,310 (trainer:762) INFO: 43epoch:train:8901-9000batch: iter_time=8.018e-05, forward_time=0.353, loss_ctc=48.959, loss_att=44.257, acc=0.767, loss=45.668, backward_time=0.322, grad_norm=45.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.532e-04, train_time=1.620
+[gpub010:0/16] 2024-02-11 13:47:47,071 (trainer:762) INFO: 43epoch:train:9001-9100batch: iter_time=7.977e-05, forward_time=0.351, loss_ctc=43.533, loss_att=37.491, acc=0.765, loss=39.303, backward_time=0.310, grad_norm=42.035, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.532e-04, train_time=1.196
+[gpub010:0/16] 2024-02-11 13:50:17,499 (trainer:762) INFO: 43epoch:train:9101-9200batch: iter_time=8.371e-05, forward_time=0.330, loss_ctc=59.268, loss_att=49.419, acc=0.741, loss=52.373, backward_time=0.309, grad_norm=51.879, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.532e-04, train_time=1.504
+[gpub010:0/16] 2024-02-11 13:52:31,817 (trainer:762) INFO: 43epoch:train:9201-9300batch: iter_time=2.250e-04, forward_time=0.329, loss_ctc=47.759, loss_att=50.374, acc=0.768, loss=49.589, backward_time=0.338, grad_norm=49.933, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.532e-04, train_time=1.343
+[gpub010:0/16] 2024-02-11 13:54:40,095 (trainer:762) INFO: 43epoch:train:9301-9400batch: iter_time=3.844e-04, forward_time=0.363, loss_ctc=41.803, loss_att=40.573, acc=0.779, loss=40.942, backward_time=0.326, grad_norm=40.301, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.532e-04, train_time=1.283
+[gpub010:0/16] 2024-02-11 13:56:55,757 (trainer:762) INFO: 43epoch:train:9401-9500batch: iter_time=7.291e-04, forward_time=0.349, loss_ctc=41.059, loss_att=36.875, acc=0.783, loss=38.130, backward_time=0.310, grad_norm=41.575, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.532e-04, train_time=1.357
+[gpub010:0/16] 2024-02-11 13:59:17,691 (trainer:762) INFO: 43epoch:train:9501-9600batch: iter_time=5.063e-04, forward_time=0.414, loss_ctc=45.795, loss_att=40.710, acc=0.779, loss=42.236, backward_time=0.324, grad_norm=41.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.532e-04, train_time=1.418
+[gpub010:0/16] 2024-02-11 14:01:46,148 (trainer:762) INFO: 43epoch:train:9601-9700batch: iter_time=3.546e-04, forward_time=0.385, loss_ctc=50.329, loss_att=49.478, acc=0.758, loss=49.734, backward_time=0.358, grad_norm=46.418, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.532e-04, train_time=1.485
+[gpub010:0/16] 2024-02-11 14:04:11,459 (trainer:762) INFO: 43epoch:train:9701-9800batch: iter_time=9.004e-05, forward_time=0.378, loss_ctc=40.950, loss_att=42.857, acc=0.773, loss=42.285, backward_time=0.327, grad_norm=41.904, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.531e-04, train_time=1.453
+[gpub010:0/16] 2024-02-11 14:06:28,336 (trainer:762) INFO: 43epoch:train:9801-9900batch: iter_time=3.197e-04, forward_time=0.348, loss_ctc=44.302, loss_att=42.338, acc=0.764, loss=42.927, backward_time=0.350, grad_norm=45.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.531e-04, train_time=1.369
+[gpub010:0/16] 2024-02-11 14:08:42,983 (trainer:762) INFO: 43epoch:train:9901-10000batch: iter_time=2.464e-04, forward_time=0.393, loss_ctc=40.749, loss_att=36.139, acc=0.783, loss=37.522, backward_time=0.319, grad_norm=40.458, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.116, optim0_lr0=1.531e-04, train_time=1.347
+[gpub010:0/16] 2024-02-11 14:09:03,036 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-11 14:09:22,743 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 14:09:26,260 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbac6cbbc10>)
+[gpub010:0/16] 2024-02-11 14:09:26,260 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-11 14:09:26,410 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 14:16:19,719 (trainer:762) INFO: 43epoch:train:10001-10100batch: iter_time=3.177, forward_time=0.296, loss_ctc=52.810, loss_att=58.503, acc=0.728, loss=56.795, backward_time=0.298, grad_norm=57.860, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.531e-04, train_time=4.567
+[gpub010:0/16] 2024-02-11 14:18:48,269 (trainer:762) INFO: 43epoch:train:10101-10200batch: iter_time=7.840e-05, forward_time=0.353, loss_ctc=47.229, loss_att=44.550, acc=0.757, loss=45.354, backward_time=0.330, grad_norm=46.732, clip=100.000, loss_scale=5.452e+33, optim_step_time=0.098, optim0_lr0=1.531e-04, train_time=1.485
+[gpub010:0/16] 2024-02-11 14:20:44,569 (trainer:762) INFO: 43epoch:train:10201-10300batch: iter_time=8.095e-05, forward_time=0.292, loss_ctc=44.798, loss_att=41.051, acc=0.772, loss=42.175, backward_time=0.298, grad_norm=39.806, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.531e-04, train_time=1.162
+[gpub010:0/16] 2024-02-11 14:23:20,905 (trainer:762) INFO: 43epoch:train:10301-10400batch: iter_time=2.228e-04, forward_time=0.354, loss_ctc=54.897, loss_att=47.210, acc=0.749, loss=49.516, backward_time=0.310, grad_norm=51.494, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.113, optim0_lr0=1.531e-04, train_time=1.564
+[gpub010:0/16] 2024-02-11 14:25:23,298 (trainer:762) INFO: 43epoch:train:10401-10500batch: iter_time=8.126e-05, forward_time=0.354, loss_ctc=49.056, loss_att=44.022, acc=0.764, loss=45.532, backward_time=0.304, grad_norm=42.870, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.531e-04, train_time=1.224
+[gpub010:0/16] 2024-02-11 14:27:35,109 (trainer:762) INFO: 43epoch:train:10501-10600batch: iter_time=1.953e-04, forward_time=0.301, loss_ctc=47.095, loss_att=47.403, acc=0.766, loss=47.311, backward_time=0.297, grad_norm=46.773, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.530e-04, train_time=1.318
+[gpub010:0/16] 2024-02-11 14:30:01,265 (trainer:762) INFO: 43epoch:train:10601-10700batch: iter_time=8.199e-05, forward_time=0.415, loss_ctc=40.908, loss_att=37.498, acc=0.790, loss=38.521, backward_time=0.316, grad_norm=39.851, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.108, optim0_lr0=1.530e-04, train_time=1.461
+[gpub010:0/16] 2024-02-11 14:30:21,698 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 14:32:03,834 (trainer:762) INFO: 43epoch:train:10701-10800batch: iter_time=8.536e-05, forward_time=0.289, loss_ctc=40.184, loss_att=39.146, acc=0.767, loss=39.457, backward_time=0.294, grad_norm=40.517, clip=100.000, loss_scale=6.136e+33, optim_step_time=0.093, optim0_lr0=1.530e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 14:34:28,733 (trainer:762) INFO: 43epoch:train:10801-10900batch: iter_time=1.002e-04, forward_time=0.337, loss_ctc=54.714, loss_att=49.208, acc=0.773, loss=50.860, backward_time=0.300, grad_norm=43.513, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.530e-04, train_time=1.448
+[gpub010:0/16] 2024-02-11 14:36:35,568 (trainer:762) INFO: 43epoch:train:10901-11000batch: iter_time=8.688e-05, forward_time=0.351, loss_ctc=41.409, loss_att=40.741, acc=0.773, loss=40.941, backward_time=0.314, grad_norm=42.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.530e-04, train_time=1.269
+[gpub010:0/16] 2024-02-11 14:38:44,133 (trainer:762) INFO: 43epoch:train:11001-11100batch: iter_time=9.318e-05, forward_time=0.292, loss_ctc=46.437, loss_att=47.892, acc=0.766, loss=47.455, backward_time=0.297, grad_norm=44.185, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.530e-04, train_time=1.285
+[gpub010:0/16] 2024-02-11 14:40:46,539 (trainer:762) INFO: 43epoch:train:11101-11200batch: iter_time=1.062e-04, forward_time=0.305, loss_ctc=40.246, loss_att=36.221, acc=0.776, loss=37.428, backward_time=0.294, grad_norm=39.763, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.530e-04, train_time=1.223
+[gpub010:0/16] 2024-02-11 14:42:40,049 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-11 14:42:59,327 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 14:43:02,872 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba91d7e110>)
+[gpub010:0/16] 2024-02-11 14:43:02,872 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-11 14:43:02,919 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 14:48:57,058 (trainer:762) INFO: 43epoch:train:11201-11300batch: iter_time=3.260, forward_time=0.382, loss_ctc=44.245, loss_att=48.067, acc=0.752, loss=46.920, backward_time=0.327, grad_norm=43.746, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.530e-04, train_time=4.906
+[gpub010:0/16] 2024-02-11 14:51:16,086 (trainer:762) INFO: 43epoch:train:11301-11400batch: iter_time=2.465e-04, forward_time=0.304, loss_ctc=50.569, loss_att=55.055, acc=0.749, loss=53.710, backward_time=0.298, grad_norm=48.658, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.530e-04, train_time=1.390
+[gpub010:0/16] 2024-02-11 14:53:35,857 (trainer:762) INFO: 43epoch:train:11401-11500batch: iter_time=7.773e-05, forward_time=0.385, loss_ctc=48.873, loss_att=44.778, acc=0.778, loss=46.007, backward_time=0.315, grad_norm=45.256, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.529e-04, train_time=1.397
+[gpub010:0/16] 2024-02-11 14:56:00,752 (trainer:762) INFO: 43epoch:train:11501-11600batch: iter_time=8.311e-05, forward_time=0.290, loss_ctc=43.341, loss_att=38.331, acc=0.768, loss=39.834, backward_time=0.294, grad_norm=41.379, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.529e-04, train_time=1.449
+[gpub010:0/16] 2024-02-11 14:58:04,203 (trainer:762) INFO: 43epoch:train:11601-11700batch: iter_time=5.015e-04, forward_time=0.305, loss_ctc=59.402, loss_att=52.667, acc=0.745, loss=54.688, backward_time=0.298, grad_norm=53.867, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.529e-04, train_time=1.234
+[gpub010:0/16] 2024-02-11 15:00:14,703 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 15:00:35,436 (trainer:762) INFO: 43epoch:train:11701-11800batch: iter_time=8.530e-05, forward_time=0.404, loss_ctc=47.252, loss_att=50.426, acc=0.771, loss=49.474, backward_time=0.325, grad_norm=44.031, clip=100.000, loss_scale=4.930e+33, optim_step_time=0.105, optim0_lr0=1.529e-04, train_time=1.512
+[gpub010:0/16] 2024-02-11 15:02:32,619 (trainer:762) INFO: 43epoch:train:11801-11900batch: iter_time=8.440e-05, forward_time=0.294, loss_ctc=41.904, loss_att=40.998, acc=0.796, loss=41.270, backward_time=0.297, grad_norm=39.374, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.529e-04, train_time=1.172
+[gpub010:0/16] 2024-02-11 15:04:39,049 (trainer:762) INFO: 43epoch:train:11901-12000batch: iter_time=7.954e-05, forward_time=0.303, loss_ctc=41.216, loss_att=37.099, acc=0.791, loss=38.334, backward_time=0.296, grad_norm=41.301, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.529e-04, train_time=1.264
+[gpub010:0/16] 2024-02-11 15:07:01,804 (trainer:762) INFO: 43epoch:train:12001-12100batch: iter_time=8.587e-05, forward_time=0.313, loss_ctc=45.317, loss_att=41.667, acc=0.781, loss=42.762, backward_time=0.323, grad_norm=42.379, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.529e-04, train_time=1.426
+[gpub010:0/16] 2024-02-11 15:09:11,809 (trainer:762) INFO: 43epoch:train:12101-12200batch: iter_time=8.249e-05, forward_time=0.316, loss_ctc=49.948, loss_att=49.739, acc=0.769, loss=49.802, backward_time=0.300, grad_norm=44.000, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.529e-04, train_time=1.300
+[gpub010:0/16] 2024-02-11 15:11:36,769 (trainer:762) INFO: 43epoch:train:12201-12300batch: iter_time=8.425e-05, forward_time=0.346, loss_ctc=40.700, loss_att=44.380, acc=0.774, loss=43.276, backward_time=0.338, grad_norm=42.684, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.528e-04, train_time=1.450
+[gpub010:0/16] 2024-02-11 15:13:49,628 (trainer:762) INFO: 43epoch:train:12301-12400batch: iter_time=8.638e-05, forward_time=0.291, loss_ctc=43.909, loss_att=42.779, acc=0.767, loss=43.118, backward_time=0.296, grad_norm=45.126, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.528e-04, train_time=1.327
+[gpub010:0/16] 2024-02-11 15:16:12,462 (trainer:762) INFO: 43epoch:train:12401-12500batch: iter_time=7.940e-05, forward_time=0.418, loss_ctc=40.503, loss_att=37.471, acc=0.788, loss=38.381, backward_time=0.336, grad_norm=38.052, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.528e-04, train_time=1.428
+[gpub010:0/16] 2024-02-11 15:16:32,693 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-11 15:16:51,753 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 15:16:55,399 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba2ebfbf70>)
+[gpub010:0/16] 2024-02-11 15:16:55,399 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-11 15:16:55,444 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 15:23:53,647 (trainer:762) INFO: 43epoch:train:12501-12600batch: iter_time=3.177, forward_time=0.293, loss_ctc=52.550, loss_att=58.162, acc=0.744, loss=56.478, backward_time=0.299, grad_norm=52.239, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.528e-04, train_time=4.612
+[gpub010:0/16] 2024-02-11 15:25:53,540 (trainer:762) INFO: 43epoch:train:12601-12700batch: iter_time=7.888e-05, forward_time=0.309, loss_ctc=47.079, loss_att=45.264, acc=0.765, loss=45.808, backward_time=0.303, grad_norm=46.695, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.528e-04, train_time=1.198
+[gpub010:0/16] 2024-02-11 15:28:38,280 (trainer:762) INFO: 43epoch:train:12701-12800batch: iter_time=8.230e-05, forward_time=0.385, loss_ctc=44.504, loss_att=41.584, acc=0.774, loss=42.460, backward_time=0.310, grad_norm=38.637, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.101, optim0_lr0=1.528e-04, train_time=1.647
+[gpub010:0/16] 2024-02-11 15:30:49,703 (trainer:762) INFO: 43epoch:train:12801-12900batch: iter_time=8.731e-05, forward_time=0.296, loss_ctc=55.273, loss_att=47.818, acc=0.759, loss=50.055, backward_time=0.307, grad_norm=48.947, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.528e-04, train_time=1.314
+[gpub010:0/16] 2024-02-11 15:33:28,559 (trainer:762) INFO: 43epoch:train:12901-13000batch: iter_time=8.248e-05, forward_time=0.387, loss_ctc=48.376, loss_att=45.835, acc=0.766, loss=46.597, backward_time=0.318, grad_norm=48.444, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.528e-04, train_time=1.588
+[gpub010:0/16] 2024-02-11 15:35:19,185 (trainer:762) INFO: 43epoch:train:13001-13100batch: iter_time=8.095e-05, forward_time=0.293, loss_ctc=46.610, loss_att=48.332, acc=0.780, loss=47.815, backward_time=0.300, grad_norm=48.308, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.527e-04, train_time=1.106
+[gpub010:0/16] 2024-02-11 15:37:33,398 (trainer:762) INFO: 43epoch:train:13101-13200batch: iter_time=8.158e-05, forward_time=0.328, loss_ctc=40.635, loss_att=36.780, acc=0.800, loss=37.937, backward_time=0.306, grad_norm=39.018, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.527e-04, train_time=1.342
+[gpub010:0/16] 2024-02-11 15:40:06,763 (trainer:762) INFO: 43epoch:train:13201-13300batch: iter_time=8.348e-05, forward_time=0.372, loss_ctc=39.934, loss_att=38.714, acc=0.775, loss=39.080, backward_time=0.307, grad_norm=40.410, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.527e-04, train_time=1.533
+[gpub010:0/16] 2024-02-11 15:42:06,880 (trainer:762) INFO: 43epoch:train:13301-13400batch: iter_time=1.345e-04, forward_time=0.317, loss_ctc=54.195, loss_att=50.400, acc=0.779, loss=51.539, backward_time=0.302, grad_norm=44.121, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.527e-04, train_time=1.200
+[gpub010:0/16] 2024-02-11 15:44:47,291 (trainer:762) INFO: 43epoch:train:13401-13500batch: iter_time=8.669e-05, forward_time=0.398, loss_ctc=40.975, loss_att=40.782, acc=0.776, loss=40.839, backward_time=0.338, grad_norm=41.049, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.527e-04, train_time=1.605
+[gpub010:0/16] 2024-02-11 15:47:03,319 (trainer:762) INFO: 43epoch:train:13501-13600batch: iter_time=3.467e-04, forward_time=0.298, loss_ctc=46.054, loss_att=49.200, acc=0.767, loss=48.256, backward_time=0.299, grad_norm=44.749, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.095, optim0_lr0=1.527e-04, train_time=1.360
+[gpub010:0/16] 2024-02-11 15:49:18,289 (trainer:762) INFO: 43epoch:train:13601-13700batch: iter_time=8.107e-05, forward_time=0.294, loss_ctc=39.811, loss_att=36.613, acc=0.784, loss=37.573, backward_time=0.296, grad_norm=40.619, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.527e-04, train_time=1.349
+[gpub010:0/16] 2024-02-11 15:50:59,435 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-11 15:51:18,836 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 15:51:22,405 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba8ebb9fc0>)
+[gpub010:0/16] 2024-02-11 15:51:22,405 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-11 15:51:22,453 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 15:57:31,001 (trainer:762) INFO: 43epoch:train:13701-13800batch: iter_time=3.324, forward_time=0.376, loss_ctc=44.512, loss_att=47.556, acc=0.757, loss=46.643, backward_time=0.313, grad_norm=44.703, clip=100.000, loss_scale=2.856e+33, optim_step_time=0.102, optim0_lr0=1.527e-04, train_time=4.928
+[gpub010:0/16] 2024-02-11 15:59:27,863 (trainer:762) INFO: 43epoch:train:13801-13900batch: iter_time=8.479e-05, forward_time=0.302, loss_ctc=50.409, loss_att=52.817, acc=0.755, loss=52.095, backward_time=0.298, grad_norm=47.985, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.527e-04, train_time=1.168
+[gpub010:0/16] 2024-02-11 16:01:58,944 (trainer:762) INFO: 43epoch:train:13901-14000batch: iter_time=8.288e-05, forward_time=0.344, loss_ctc=48.536, loss_att=44.250, acc=0.778, loss=45.536, backward_time=0.338, grad_norm=45.660, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.526e-04, train_time=1.511
+[gpub010:0/16] 2024-02-11 16:04:06,731 (trainer:762) INFO: 43epoch:train:14001-14100batch: iter_time=8.913e-05, forward_time=0.291, loss_ctc=43.377, loss_att=38.426, acc=0.765, loss=39.911, backward_time=0.296, grad_norm=42.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.526e-04, train_time=1.278
+[gpub010:0/16] 2024-02-11 16:06:49,738 (trainer:762) INFO: 43epoch:train:14101-14200batch: iter_time=8.701e-05, forward_time=0.366, loss_ctc=59.084, loss_att=51.203, acc=0.750, loss=53.567, backward_time=0.324, grad_norm=54.702, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.526e-04, train_time=1.630
+[gpub010:0/16] 2024-02-11 16:08:46,862 (trainer:762) INFO: 43epoch:train:14201-14300batch: iter_time=8.378e-05, forward_time=0.293, loss_ctc=47.286, loss_att=50.218, acc=0.773, loss=49.339, backward_time=0.299, grad_norm=47.369, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.526e-04, train_time=1.171
+[gpub010:0/16] 2024-02-11 16:11:13,702 (trainer:762) INFO: 43epoch:train:14301-14400batch: iter_time=8.008e-05, forward_time=0.291, loss_ctc=41.679, loss_att=40.994, acc=0.795, loss=41.200, backward_time=0.297, grad_norm=40.776, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.526e-04, train_time=1.468
+[gpub010:0/16] 2024-02-11 16:13:49,405 (trainer:762) INFO: 43epoch:train:14401-14500batch: iter_time=4.736e-04, forward_time=0.394, loss_ctc=41.189, loss_att=36.767, acc=0.793, loss=38.094, backward_time=0.338, grad_norm=42.406, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.526e-04, train_time=1.557
+[gpub010:0/16] 2024-02-11 16:16:00,845 (trainer:762) INFO: 43epoch:train:14501-14600batch: iter_time=8.321e-05, forward_time=0.291, loss_ctc=45.793, loss_att=41.779, acc=0.782, loss=42.983, backward_time=0.296, grad_norm=41.756, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.526e-04, train_time=1.314
+[gpub010:0/16] 2024-02-11 16:18:44,557 (trainer:762) INFO: 43epoch:train:14601-14700batch: iter_time=8.082e-05, forward_time=0.376, loss_ctc=49.636, loss_att=48.926, acc=0.770, loss=49.139, backward_time=0.313, grad_norm=44.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.526e-04, train_time=1.637
+[gpub010:0/16] 2024-02-11 16:20:48,374 (trainer:762) INFO: 43epoch:train:14701-14800batch: iter_time=8.814e-05, forward_time=0.294, loss_ctc=40.631, loss_att=45.063, acc=0.770, loss=43.733, backward_time=0.299, grad_norm=39.913, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.525e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 16:23:08,456 (trainer:762) INFO: 43epoch:train:14801-14900batch: iter_time=8.024e-05, forward_time=0.296, loss_ctc=44.375, loss_att=42.433, acc=0.768, loss=43.016, backward_time=0.296, grad_norm=43.389, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.525e-04, train_time=1.400
+[gpub010:0/16] 2024-02-11 16:25:28,929 (trainer:762) INFO: 43epoch:train:14901-15000batch: iter_time=0.020, forward_time=0.340, loss_ctc=40.728, loss_att=37.536, acc=0.787, loss=38.493, backward_time=0.365, grad_norm=37.445, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.525e-04, train_time=1.405
+[gpub010:0/16] 2024-02-11 17:03:32,602 (trainer:361) INFO: 43epoch results: [train] iter_time=0.258, forward_time=0.329, loss_ctc=46.884, loss_att=44.899, acc=0.767, loss=45.494, backward_time=0.310, grad_norm=44.884, clip=100.000, loss_scale=5.931e+33, optim_step_time=0.096, optim0_lr0=1.534e-04, train_time=1.603, time=6 hours, 41 minutes and 15.29 seconds, total_count=675000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=35.196, cer_ctc=0.178, loss_att=36.829, acc=0.707, cer=0.251, wer=0.984, loss=36.339, time=37 minutes and 39.48 seconds, total_count=210195, gpu_max_cached_mem_GB=42.092
+[gpub010:0/16] 2024-02-11 17:03:42,418 (trainer:416) INFO: The best model has been updated: valid.acc, valid.total_count
+[gpub010:0/16] 2024-02-11 17:03:42,478 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/32epoch.pth, exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/38epoch.pth
+[gpub010:0/16] 2024-02-11 17:03:42,479 (trainer:290) INFO: 44/45epoch started. Estimated time to finish: 14 hours, 35 minutes and 32.51 seconds
+[gpub010:0/16] 2024-02-11 17:03:42,498 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-11 17:04:01,008 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 17:04:04,401 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc0a1f3d990>)
+[gpub010:0/16] 2024-02-11 17:04:04,401 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-11 17:04:04,404 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 17:10:31,840 (trainer:762) INFO: 44epoch:train:1-100batch: iter_time=2.917, forward_time=0.362, loss_ctc=42.799, loss_att=39.156, acc=0.773, loss=40.249, backward_time=0.303, grad_norm=42.901, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.525e-04, train_time=4.093
+[gpub010:0/16] 2024-02-11 17:12:42,915 (trainer:762) INFO: 44epoch:train:101-200batch: iter_time=8.346e-05, forward_time=0.292, loss_ctc=45.987, loss_att=45.411, acc=0.768, loss=45.584, backward_time=0.298, grad_norm=42.098, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.525e-04, train_time=1.311
+[gpub010:0/16] 2024-02-11 17:15:02,702 (trainer:762) INFO: 44epoch:train:201-300batch: iter_time=8.391e-05, forward_time=0.298, loss_ctc=46.829, loss_att=50.097, acc=0.750, loss=49.117, backward_time=0.301, grad_norm=44.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.525e-04, train_time=1.398
+[gpub010:0/16] 2024-02-11 17:17:22,740 (trainer:762) INFO: 44epoch:train:301-400batch: iter_time=8.153e-05, forward_time=0.358, loss_ctc=46.632, loss_att=52.906, acc=0.741, loss=51.023, backward_time=0.330, grad_norm=46.320, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.525e-04, train_time=1.400
+[gpub010:0/16] 2024-02-11 17:19:35,133 (trainer:762) INFO: 44epoch:train:401-500batch: iter_time=8.032e-05, forward_time=0.290, loss_ctc=40.703, loss_att=42.654, acc=0.756, loss=42.068, backward_time=0.294, grad_norm=49.156, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.525e-04, train_time=1.324
+[gpub010:0/16] 2024-02-11 17:21:44,184 (trainer:762) INFO: 44epoch:train:501-600batch: iter_time=9.087e-05, forward_time=0.297, loss_ctc=49.250, loss_att=49.794, acc=0.758, loss=49.631, backward_time=0.298, grad_norm=49.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.525e-04, train_time=1.290
+[gpub010:0/16] 2024-02-11 17:24:00,084 (trainer:762) INFO: 44epoch:train:601-700batch: iter_time=8.215e-05, forward_time=0.352, loss_ctc=49.503, loss_att=47.287, acc=0.760, loss=47.951, backward_time=0.313, grad_norm=48.567, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.524e-04, train_time=1.359
+[gpub010:0/16] 2024-02-11 17:26:28,665 (trainer:762) INFO: 44epoch:train:701-800batch: iter_time=8.546e-05, forward_time=0.304, loss_ctc=46.366, loss_att=41.617, acc=0.760, loss=43.042, backward_time=0.302, grad_norm=55.972, clip=100.000, loss_scale=5.712e+33, optim_step_time=0.093, optim0_lr0=1.524e-04, train_time=1.485
+[gpub010:0/16] 2024-02-11 17:27:27,894 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 17:28:23,593 (trainer:762) INFO: 44epoch:train:801-900batch: iter_time=8.765e-05, forward_time=0.313, loss_ctc=55.443, loss_att=56.391, acc=0.743, loss=56.106, backward_time=0.300, grad_norm=55.095, clip=100.000, loss_scale=7.972e+33, optim_step_time=0.093, optim0_lr0=1.524e-04, train_time=1.149
+[gpub010:0/16] 2024-02-11 17:30:44,173 (trainer:762) INFO: 44epoch:train:901-1000batch: iter_time=8.753e-05, forward_time=0.292, loss_ctc=45.339, loss_att=41.765, acc=0.761, loss=42.837, backward_time=0.300, grad_norm=42.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.524e-04, train_time=1.406
+[gpub010:0/16] 2024-02-11 17:32:55,772 (trainer:762) INFO: 44epoch:train:1001-1100batch: iter_time=5.823e-04, forward_time=0.374, loss_ctc=42.082, loss_att=35.638, acc=0.779, loss=37.571, backward_time=0.338, grad_norm=41.359, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.524e-04, train_time=1.316
+[gpub010:0/16] 2024-02-11 17:35:14,097 (trainer:762) INFO: 44epoch:train:1101-1200batch: iter_time=9.392e-05, forward_time=0.291, loss_ctc=44.691, loss_att=46.995, acc=0.734, loss=46.304, backward_time=0.295, grad_norm=49.025, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.524e-04, train_time=1.383
+[gpub010:0/16] 2024-02-11 17:36:41,300 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-11 17:37:00,238 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 17:37:03,768 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1c9852cb0>)
+[gpub010:0/16] 2024-02-11 17:37:03,768 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-11 17:37:03,771 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 17:42:28,974 (trainer:762) INFO: 44epoch:train:1201-1300batch: iter_time=2.931, forward_time=0.295, loss_ctc=47.750, loss_att=44.035, acc=0.768, loss=45.150, backward_time=0.298, grad_norm=44.451, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.524e-04, train_time=4.349
+[gpub010:0/16] 2024-02-11 17:45:03,029 (trainer:762) INFO: 44epoch:train:1301-1400batch: iter_time=8.486e-05, forward_time=0.376, loss_ctc=44.576, loss_att=44.946, acc=0.769, loss=44.835, backward_time=0.312, grad_norm=44.280, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.524e-04, train_time=1.540
+[gpub010:0/16] 2024-02-11 17:47:32,783 (trainer:762) INFO: 44epoch:train:1401-1500batch: iter_time=8.059e-05, forward_time=0.292, loss_ctc=42.577, loss_att=41.445, acc=0.762, loss=41.785, backward_time=0.296, grad_norm=37.994, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.523e-04, train_time=1.497
+[gpub010:0/16] 2024-02-11 17:49:45,235 (trainer:762) INFO: 44epoch:train:1501-1600batch: iter_time=9.362e-05, forward_time=0.296, loss_ctc=47.906, loss_att=56.842, acc=0.739, loss=54.161, backward_time=0.302, grad_norm=46.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.523e-04, train_time=1.324
+[gpub010:0/16] 2024-02-11 17:51:47,834 (trainer:762) INFO: 44epoch:train:1601-1700batch: iter_time=8.386e-05, forward_time=0.290, loss_ctc=42.679, loss_att=44.629, acc=0.752, loss=44.044, backward_time=0.297, grad_norm=76.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.523e-04, train_time=1.226
+[gpub010:0/16] 2024-02-11 17:54:01,403 (trainer:762) INFO: 44epoch:train:1701-1800batch: iter_time=8.625e-05, forward_time=0.372, loss_ctc=45.292, loss_att=47.686, acc=0.766, loss=46.968, backward_time=0.323, grad_norm=46.834, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.523e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 17:56:18,209 (trainer:762) INFO: 44epoch:train:1801-1900batch: iter_time=9.177e-05, forward_time=0.291, loss_ctc=49.701, loss_att=46.187, acc=0.759, loss=47.241, backward_time=0.295, grad_norm=50.355, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.523e-04, train_time=1.369
+[gpub010:0/16] 2024-02-11 17:58:20,093 (trainer:762) INFO: 44epoch:train:1901-2000batch: iter_time=8.726e-05, forward_time=0.296, loss_ctc=42.088, loss_att=39.677, acc=0.773, loss=40.400, backward_time=0.299, grad_norm=44.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.523e-04, train_time=1.219
+[gpub010:0/16] 2024-02-11 18:00:30,257 (trainer:762) INFO: 44epoch:train:2001-2100batch: iter_time=9.044e-05, forward_time=0.318, loss_ctc=47.438, loss_att=45.740, acc=0.760, loss=46.249, backward_time=0.319, grad_norm=47.917, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.523e-04, train_time=1.301
+[gpub010:0/16] 2024-02-11 18:02:44,246 (trainer:762) INFO: 44epoch:train:2101-2200batch: iter_time=8.739e-05, forward_time=0.293, loss_ctc=53.148, loss_att=54.325, acc=0.745, loss=53.972, backward_time=0.297, grad_norm=50.109, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.523e-04, train_time=1.338
+[gpub010:0/16] 2024-02-11 18:04:38,013 (trainer:762) INFO: 44epoch:train:2201-2300batch: iter_time=8.497e-05, forward_time=0.294, loss_ctc=42.643, loss_att=36.868, acc=0.774, loss=38.600, backward_time=0.300, grad_norm=40.058, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.523e-04, train_time=1.138
+[gpub010:0/16] 2024-02-11 18:06:05,730 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 18:07:25,447 (trainer:762) INFO: 44epoch:train:2301-2400batch: iter_time=8.712e-05, forward_time=0.340, loss_ctc=44.025, loss_att=41.447, acc=0.751, loss=42.220, backward_time=0.354, grad_norm=49.500, clip=100.000, loss_scale=3.960e+33, optim_step_time=0.099, optim0_lr0=1.522e-04, train_time=1.675
+[gpub010:0/16] 2024-02-11 18:09:46,693 (trainer:762) INFO: 44epoch:train:2401-2500batch: iter_time=8.172e-05, forward_time=0.291, loss_ctc=47.522, loss_att=49.157, acc=0.754, loss=48.666, backward_time=0.297, grad_norm=44.302, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.522e-04, train_time=1.412
+[gpub010:0/16] 2024-02-11 18:10:06,722 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-11 18:10:26,049 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 18:10:29,612 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1ee6c37c0>)
+[gpub010:0/16] 2024-02-11 18:10:29,613 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-11 18:10:29,616 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 18:17:42,114 (trainer:762) INFO: 44epoch:train:2501-2600batch: iter_time=3.261, forward_time=0.295, loss_ctc=41.218, loss_att=40.836, acc=0.765, loss=40.951, backward_time=0.296, grad_norm=42.098, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.522e-04, train_time=4.754
+[gpub010:0/16] 2024-02-11 18:20:11,651 (trainer:762) INFO: 44epoch:train:2601-2700batch: iter_time=7.940e-05, forward_time=0.427, loss_ctc=44.991, loss_att=43.456, acc=0.772, loss=43.916, backward_time=0.320, grad_norm=40.824, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.522e-04, train_time=1.495
+[gpub010:0/16] 2024-02-11 18:22:19,656 (trainer:762) INFO: 44epoch:train:2701-2800batch: iter_time=7.918e-05, forward_time=0.291, loss_ctc=45.769, loss_att=50.305, acc=0.746, loss=48.944, backward_time=0.297, grad_norm=45.665, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.522e-04, train_time=1.280
+[gpub010:0/16] 2024-02-11 18:24:42,117 (trainer:762) INFO: 44epoch:train:2801-2900batch: iter_time=8.352e-05, forward_time=0.296, loss_ctc=45.617, loss_att=51.307, acc=0.741, loss=49.600, backward_time=0.299, grad_norm=44.035, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.522e-04, train_time=1.423
+[gpub010:0/16] 2024-02-11 18:27:07,580 (trainer:762) INFO: 44epoch:train:2901-3000batch: iter_time=8.320e-05, forward_time=0.381, loss_ctc=39.300, loss_att=42.394, acc=0.755, loss=41.466, backward_time=0.344, grad_norm=45.668, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.104, optim0_lr0=1.522e-04, train_time=1.456
+[gpub010:0/16] 2024-02-11 18:29:34,692 (trainer:762) INFO: 44epoch:train:3001-3100batch: iter_time=7.869e-05, forward_time=0.295, loss_ctc=47.739, loss_att=48.956, acc=0.755, loss=48.591, backward_time=0.296, grad_norm=49.806, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.522e-04, train_time=1.471
+[gpub010:0/16] 2024-02-11 18:31:26,690 (trainer:762) INFO: 44epoch:train:3101-3200batch: iter_time=8.261e-05, forward_time=0.293, loss_ctc=48.161, loss_att=46.557, acc=0.757, loss=47.038, backward_time=0.299, grad_norm=46.853, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.521e-04, train_time=1.119
+[gpub010:0/16] 2024-02-11 18:34:08,198 (trainer:762) INFO: 44epoch:train:3201-3300batch: iter_time=8.379e-05, forward_time=0.425, loss_ctc=44.824, loss_att=40.770, acc=0.756, loss=41.987, backward_time=0.312, grad_norm=50.429, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.521e-04, train_time=1.615
+[gpub010:0/16] 2024-02-11 18:36:26,662 (trainer:762) INFO: 44epoch:train:3301-3400batch: iter_time=8.308e-05, forward_time=0.297, loss_ctc=53.826, loss_att=52.892, acc=0.744, loss=53.172, backward_time=0.303, grad_norm=50.520, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.521e-04, train_time=1.384
+[gpub010:0/16] 2024-02-11 18:38:15,147 (trainer:762) INFO: 44epoch:train:3401-3500batch: iter_time=8.779e-05, forward_time=0.292, loss_ctc=44.093, loss_att=41.983, acc=0.753, loss=42.616, backward_time=0.300, grad_norm=40.714, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.521e-04, train_time=1.084
+[gpub010:0/16] 2024-02-11 18:40:57,219 (trainer:762) INFO: 44epoch:train:3501-3600batch: iter_time=8.502e-05, forward_time=0.381, loss_ctc=41.201, loss_att=35.902, acc=0.772, loss=37.492, backward_time=0.320, grad_norm=41.683, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.105, optim0_lr0=1.521e-04, train_time=1.621
+[gpub010:0/16] 2024-02-11 18:42:45,122 (trainer:762) INFO: 44epoch:train:3601-3700batch: iter_time=8.350e-05, forward_time=0.290, loss_ctc=43.440, loss_att=45.885, acc=0.738, loss=45.151, backward_time=0.297, grad_norm=50.452, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.521e-04, train_time=1.079
+[gpub010:0/16] 2024-02-11 18:44:18,107 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-11 18:44:37,233 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 18:44:40,779 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbcd87ab250>)
+[gpub010:0/16] 2024-02-11 18:44:40,779 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-11 18:44:40,835 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 18:50:13,936 (trainer:762) INFO: 44epoch:train:3701-3800batch: iter_time=3.099, forward_time=0.397, loss_ctc=47.618, loss_att=44.151, acc=0.766, loss=45.191, backward_time=0.316, grad_norm=43.564, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.521e-04, train_time=4.488
+[gpub010:0/16] 2024-02-11 18:52:20,818 (trainer:762) INFO: 44epoch:train:3801-3900batch: iter_time=8.016e-05, forward_time=0.294, loss_ctc=44.758, loss_att=46.625, acc=0.766, loss=46.065, backward_time=0.298, grad_norm=46.345, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.521e-04, train_time=1.269
+[gpub010:0/16] 2024-02-11 18:54:37,314 (trainer:762) INFO: 44epoch:train:3901-4000batch: iter_time=8.166e-05, forward_time=0.292, loss_ctc=42.532, loss_att=42.262, acc=0.764, loss=42.343, backward_time=0.297, grad_norm=40.798, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.521e-04, train_time=1.365
+[gpub010:0/16] 2024-02-11 18:56:40,051 (trainer:762) INFO: 44epoch:train:4001-4100batch: iter_time=8.024e-05, forward_time=0.383, loss_ctc=47.644, loss_att=56.955, acc=0.742, loss=54.161, backward_time=0.336, grad_norm=46.220, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.098, optim0_lr0=1.520e-04, train_time=1.227
+[gpub010:0/16] 2024-02-11 18:58:42,501 (trainer:762) INFO: 44epoch:train:4101-4200batch: iter_time=7.985e-05, forward_time=0.295, loss_ctc=41.862, loss_att=44.252, acc=0.754, loss=43.535, backward_time=0.296, grad_norm=45.679, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.520e-04, train_time=1.224
+[gpub010:0/16] 2024-02-11 19:01:11,875 (trainer:762) INFO: 44epoch:train:4201-4300batch: iter_time=8.403e-05, forward_time=0.292, loss_ctc=44.785, loss_att=47.015, acc=0.770, loss=46.346, backward_time=0.296, grad_norm=45.993, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.520e-04, train_time=1.494
+[gpub010:0/16] 2024-02-11 19:03:03,340 (trainer:762) INFO: 44epoch:train:4301-4400batch: iter_time=8.141e-05, forward_time=0.293, loss_ctc=48.798, loss_att=45.922, acc=0.761, loss=46.785, backward_time=0.301, grad_norm=45.628, clip=100.000, loss_scale=3.816e+33, optim_step_time=0.095, optim0_lr0=1.520e-04, train_time=1.114
+[gpub010:0/16] 2024-02-11 19:05:04,615 (trainer:762) INFO: 44epoch:train:4401-4500batch: iter_time=8.044e-05, forward_time=0.367, loss_ctc=42.249, loss_att=39.725, acc=0.774, loss=40.482, backward_time=0.320, grad_norm=47.080, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.520e-04, train_time=1.213
+[gpub010:0/16] 2024-02-11 19:07:36,420 (trainer:762) INFO: 44epoch:train:4501-4600batch: iter_time=8.873e-05, forward_time=0.295, loss_ctc=47.347, loss_att=45.756, acc=0.761, loss=46.234, backward_time=0.298, grad_norm=46.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.520e-04, train_time=1.518
+[gpub010:0/16] 2024-02-11 19:09:28,377 (trainer:762) INFO: 44epoch:train:4601-4700batch: iter_time=8.076e-05, forward_time=0.318, loss_ctc=51.897, loss_att=53.880, acc=0.747, loss=53.285, backward_time=0.305, grad_norm=52.430, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.520e-04, train_time=1.119
+[gpub010:0/16] 2024-02-11 19:11:26,968 (trainer:762) INFO: 44epoch:train:4701-4800batch: iter_time=9.281e-05, forward_time=0.354, loss_ctc=42.171, loss_att=36.797, acc=0.776, loss=38.409, backward_time=0.327, grad_norm=40.970, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.520e-04, train_time=1.185
+[gpub010:0/16] 2024-02-11 19:14:03,063 (trainer:762) INFO: 44epoch:train:4801-4900batch: iter_time=8.766e-05, forward_time=0.290, loss_ctc=43.599, loss_att=40.698, acc=0.754, loss=41.568, backward_time=0.294, grad_norm=45.796, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.561
+[gpub010:0/16] 2024-02-11 19:15:58,141 (trainer:762) INFO: 44epoch:train:4901-5000batch: iter_time=8.259e-05, forward_time=0.293, loss_ctc=47.202, loss_att=48.794, acc=0.755, loss=48.317, backward_time=0.300, grad_norm=46.474, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.151
+[gpub010:0/16] 2024-02-11 19:16:18,210 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-11 19:16:38,027 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 19:16:41,547 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba717437f0>)
+[gpub010:0/16] 2024-02-11 19:16:41,547 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-11 19:16:41,551 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 19:23:50,231 (trainer:762) INFO: 44epoch:train:5001-5100batch: iter_time=3.289, forward_time=0.385, loss_ctc=40.913, loss_att=39.938, acc=0.768, loss=40.230, backward_time=0.314, grad_norm=43.466, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.519e-04, train_time=4.721
+[gpub010:0/16] 2024-02-11 19:26:09,021 (trainer:762) INFO: 44epoch:train:5101-5200batch: iter_time=8.061e-05, forward_time=0.292, loss_ctc=44.807, loss_att=42.674, acc=0.775, loss=43.314, backward_time=0.297, grad_norm=41.255, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.388
+[gpub010:0/16] 2024-02-11 19:28:11,265 (trainer:762) INFO: 44epoch:train:5201-5300batch: iter_time=8.461e-05, forward_time=0.354, loss_ctc=45.268, loss_att=49.139, acc=0.750, loss=47.978, backward_time=0.332, grad_norm=44.110, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.519e-04, train_time=1.222
+[gpub010:0/16] 2024-02-11 19:30:29,776 (trainer:762) INFO: 44epoch:train:5301-5400batch: iter_time=8.461e-05, forward_time=0.290, loss_ctc=44.984, loss_att=50.273, acc=0.744, loss=48.686, backward_time=0.295, grad_norm=43.653, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.384
+[gpub010:0/16] 2024-02-11 19:32:33,709 (trainer:762) INFO: 44epoch:train:5401-5500batch: iter_time=8.879e-05, forward_time=0.290, loss_ctc=38.881, loss_att=41.408, acc=0.761, loss=40.650, backward_time=0.294, grad_norm=45.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.239
+[gpub010:0/16] 2024-02-11 19:34:59,972 (trainer:762) INFO: 44epoch:train:5501-5600batch: iter_time=3.777e-04, forward_time=0.407, loss_ctc=47.549, loss_att=49.155, acc=0.755, loss=48.673, backward_time=0.320, grad_norm=47.685, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.109, optim0_lr0=1.519e-04, train_time=1.462
+[gpub010:0/16] 2024-02-11 19:37:02,010 (trainer:762) INFO: 44epoch:train:5601-5700batch: iter_time=8.248e-05, forward_time=0.292, loss_ctc=47.594, loss_att=46.285, acc=0.759, loss=46.678, backward_time=0.297, grad_norm=49.103, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.519e-04, train_time=1.220
+[gpub010:0/16] 2024-02-11 19:39:05,930 (trainer:762) INFO: 44epoch:train:5701-5800batch: iter_time=8.280e-05, forward_time=0.290, loss_ctc=45.003, loss_att=40.656, acc=0.757, loss=41.960, backward_time=0.294, grad_norm=46.761, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.518e-04, train_time=1.239
+[gpub010:0/16] 2024-02-11 19:41:26,115 (trainer:762) INFO: 44epoch:train:5801-5900batch: iter_time=8.710e-05, forward_time=0.383, loss_ctc=52.728, loss_att=53.076, acc=0.745, loss=52.972, backward_time=0.337, grad_norm=48.805, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.518e-04, train_time=1.402
+[gpub010:0/16] 2024-02-11 19:43:41,112 (trainer:762) INFO: 44epoch:train:5901-6000batch: iter_time=8.114e-05, forward_time=0.291, loss_ctc=43.874, loss_att=41.821, acc=0.754, loss=42.437, backward_time=0.294, grad_norm=44.065, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.518e-04, train_time=1.349
+[gpub010:0/16] 2024-02-11 19:45:40,085 (trainer:762) INFO: 44epoch:train:6001-6100batch: iter_time=8.066e-05, forward_time=0.341, loss_ctc=40.776, loss_att=35.924, acc=0.773, loss=37.380, backward_time=0.337, grad_norm=41.156, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.518e-04, train_time=1.190
+[gpub010:0/16] 2024-02-11 19:46:50,409 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 19:47:52,069 (trainer:762) INFO: 44epoch:train:6101-6200batch: iter_time=8.522e-05, forward_time=0.291, loss_ctc=42.988, loss_att=45.035, acc=0.743, loss=44.421, backward_time=0.295, grad_norm=49.889, clip=100.000, loss_scale=4.222e+33, optim_step_time=0.093, optim0_lr0=1.518e-04, train_time=1.320
+[gpub010:0/16] 2024-02-11 19:49:24,862 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-11 19:49:44,273 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 19:49:47,797 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba82125000>)
+[gpub010:0/16] 2024-02-11 19:49:47,797 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-11 19:49:47,852 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 19:54:39,844 (trainer:762) INFO: 44epoch:train:6201-6300batch: iter_time=2.757, forward_time=0.291, loss_ctc=46.826, loss_att=43.552, acc=0.765, loss=44.534, backward_time=0.297, grad_norm=41.854, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.518e-04, train_time=4.078
+[gpub010:0/16] 2024-02-11 19:56:44,559 (trainer:762) INFO: 44epoch:train:6301-6400batch: iter_time=8.142e-05, forward_time=0.406, loss_ctc=44.126, loss_att=43.065, acc=0.768, loss=43.384, backward_time=0.333, grad_norm=44.976, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.518e-04, train_time=1.247
+[gpub010:0/16] 2024-02-11 19:59:11,816 (trainer:762) INFO: 44epoch:train:6401-6500batch: iter_time=8.124e-05, forward_time=0.291, loss_ctc=42.132, loss_att=40.602, acc=0.765, loss=41.061, backward_time=0.295, grad_norm=40.235, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.518e-04, train_time=1.472
+[gpub010:0/16] 2024-02-11 20:01:08,518 (trainer:762) INFO: 44epoch:train:6501-6600batch: iter_time=8.241e-05, forward_time=0.315, loss_ctc=47.045, loss_att=54.292, acc=0.737, loss=52.118, backward_time=0.300, grad_norm=44.244, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.167
+[gpub010:0/16] 2024-02-11 20:03:43,139 (trainer:762) INFO: 44epoch:train:6601-6700batch: iter_time=8.460e-05, forward_time=0.372, loss_ctc=41.707, loss_att=43.752, acc=0.752, loss=43.139, backward_time=0.325, grad_norm=45.756, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.517e-04, train_time=1.546
+[gpub010:0/16] 2024-02-11 20:05:46,103 (trainer:762) INFO: 44epoch:train:6701-6800batch: iter_time=8.220e-05, forward_time=0.291, loss_ctc=44.429, loss_att=46.567, acc=0.767, loss=45.925, backward_time=0.297, grad_norm=46.559, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.230
+[gpub010:0/16] 2024-02-11 20:07:46,469 (trainer:762) INFO: 44epoch:train:6801-6900batch: iter_time=8.345e-05, forward_time=0.289, loss_ctc=48.759, loss_att=46.349, acc=0.754, loss=47.072, backward_time=0.295, grad_norm=47.456, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.203
+[gpub010:0/16] 2024-02-11 20:09:50,816 (trainer:762) INFO: 44epoch:train:6901-7000batch: iter_time=7.807e-05, forward_time=0.290, loss_ctc=41.365, loss_att=38.848, acc=0.768, loss=39.603, backward_time=0.295, grad_norm=46.870, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.243
+[gpub010:0/16] 2024-02-11 20:12:12,444 (trainer:762) INFO: 44epoch:train:7001-7100batch: iter_time=2.265e-04, forward_time=0.335, loss_ctc=46.462, loss_att=44.047, acc=0.755, loss=44.771, backward_time=0.366, grad_norm=45.647, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.096, optim0_lr0=1.517e-04, train_time=1.415
+[gpub010:0/16] 2024-02-11 20:14:01,670 (trainer:762) INFO: 44epoch:train:7101-7200batch: iter_time=8.109e-05, forward_time=0.293, loss_ctc=51.395, loss_att=51.802, acc=0.741, loss=51.680, backward_time=0.298, grad_norm=49.783, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.093
+[gpub010:0/16] 2024-02-11 20:16:33,616 (trainer:762) INFO: 44epoch:train:7201-7300batch: iter_time=8.478e-05, forward_time=0.290, loss_ctc=41.686, loss_att=37.734, acc=0.763, loss=38.919, backward_time=0.294, grad_norm=41.276, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.517e-04, train_time=1.519
+[gpub010:0/16] 2024-02-11 20:18:39,552 (trainer:762) INFO: 44epoch:train:7301-7400batch: iter_time=8.977e-05, forward_time=0.356, loss_ctc=43.236, loss_att=39.933, acc=0.755, loss=40.924, backward_time=0.351, grad_norm=45.539, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.099, optim0_lr0=1.517e-04, train_time=1.259
+[gpub010:0/16] 2024-02-11 20:20:53,811 (trainer:762) INFO: 44epoch:train:7401-7500batch: iter_time=8.131e-05, forward_time=0.292, loss_ctc=47.019, loss_att=48.163, acc=0.750, loss=47.820, backward_time=0.297, grad_norm=43.323, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.516e-04, train_time=1.341
+[gpub010:0/16] 2024-02-11 20:21:14,012 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-11 20:21:33,384 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 20:21:36,888 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba3179fcd0>)
+[gpub010:0/16] 2024-02-11 20:21:36,889 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-11 20:21:36,892 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 20:28:51,873 (trainer:762) INFO: 44epoch:train:7501-7600batch: iter_time=3.454, forward_time=0.318, loss_ctc=40.588, loss_att=38.209, acc=0.773, loss=38.923, backward_time=0.344, grad_norm=42.350, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.097, optim0_lr0=1.516e-04, train_time=4.781
+[gpub010:0/16] 2024-02-11 20:31:18,258 (trainer:762) INFO: 44epoch:train:7601-7700batch: iter_time=7.848e-05, forward_time=0.337, loss_ctc=44.652, loss_att=41.969, acc=0.777, loss=42.774, backward_time=0.297, grad_norm=38.356, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.516e-04, train_time=1.464
+[gpub010:0/16] 2024-02-11 20:33:11,322 (trainer:762) INFO: 44epoch:train:7701-7800batch: iter_time=8.135e-05, forward_time=0.294, loss_ctc=45.047, loss_att=48.156, acc=0.753, loss=47.223, backward_time=0.301, grad_norm=42.935, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.093, optim0_lr0=1.516e-04, train_time=1.129
+[gpub010:0/16] 2024-02-11 20:36:03,576 (trainer:762) INFO: 44epoch:train:7801-7900batch: iter_time=8.231e-05, forward_time=0.369, loss_ctc=44.782, loss_att=50.438, acc=0.744, loss=48.741, backward_time=0.321, grad_norm=43.646, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.103, optim0_lr0=1.516e-04, train_time=1.724
+[gpub010:0/16] 2024-02-11 20:37:51,002 (trainer:762) INFO: 44epoch:train:7901-8000batch: iter_time=8.450e-05, forward_time=0.290, loss_ctc=38.907, loss_att=41.679, acc=0.760, loss=40.848, backward_time=0.296, grad_norm=45.288, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.516e-04, train_time=1.074
+[gpub010:0/16] 2024-02-11 20:39:59,587 (trainer:762) INFO: 44epoch:train:8001-8100batch: iter_time=8.760e-05, forward_time=0.299, loss_ctc=46.833, loss_att=48.227, acc=0.758, loss=47.809, backward_time=0.298, grad_norm=48.038, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.094, optim0_lr0=1.516e-04, train_time=1.285
+[gpub010:0/16] 2024-02-11 20:42:37,939 (trainer:762) INFO: 44epoch:train:8101-8200batch: iter_time=8.052e-05, forward_time=0.401, loss_ctc=48.116, loss_att=45.807, acc=0.762, loss=46.500, backward_time=0.338, grad_norm=45.296, clip=100.000, loss_scale=3.557e+33, optim_step_time=0.099, optim0_lr0=1.516e-04, train_time=1.584
+[gpub010:0/16] 2024-02-11 20:44:26,836 (trainer:762) INFO: 44epoch:train:8201-8300batch: iter_time=8.072e-05, forward_time=0.290, loss_ctc=44.586, loss_att=40.029, acc=0.760, loss=41.396, backward_time=0.296, grad_norm=52.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.516e-04, train_time=1.089
+[gpub010:0/16] 2024-02-11 20:46:30,668 (trainer:762) INFO: 44epoch:train:8301-8400batch: iter_time=8.539e-05, forward_time=0.296, loss_ctc=52.588, loss_att=51.954, acc=0.748, loss=52.145, backward_time=0.301, grad_norm=47.934, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.515e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 20:49:00,595 (trainer:762) INFO: 44epoch:train:8401-8500batch: iter_time=1.922e-04, forward_time=0.344, loss_ctc=43.677, loss_att=41.182, acc=0.756, loss=41.931, backward_time=0.344, grad_norm=41.574, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.515e-04, train_time=1.499
+[gpub010:0/16] 2024-02-11 20:51:08,247 (trainer:762) INFO: 44epoch:train:8501-8600batch: iter_time=8.380e-05, forward_time=0.291, loss_ctc=40.639, loss_att=35.476, acc=0.776, loss=37.025, backward_time=0.295, grad_norm=38.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.515e-04, train_time=1.276
+[gpub010:0/16] 2024-02-11 20:53:16,099 (trainer:762) INFO: 44epoch:train:8601-8700batch: iter_time=8.092e-05, forward_time=0.296, loss_ctc=42.875, loss_att=45.431, acc=0.743, loss=44.664, backward_time=0.298, grad_norm=48.761, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.515e-04, train_time=1.278
+[gpub010:0/16] 2024-02-11 20:54:33,023 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-11 20:54:52,818 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 20:54:56,412 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbd2155ffd0>)
+[gpub010:0/16] 2024-02-11 20:54:56,412 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-11 20:54:56,415 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 21:00:20,937 (trainer:762) INFO: 44epoch:train:8701-8800batch: iter_time=3.000, forward_time=0.390, loss_ctc=47.027, loss_att=43.790, acc=0.765, loss=44.761, backward_time=0.314, grad_norm=41.678, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.515e-04, train_time=4.249
+[gpub010:0/16] 2024-02-11 21:02:08,634 (trainer:762) INFO: 44epoch:train:8801-8900batch: iter_time=8.142e-05, forward_time=0.291, loss_ctc=43.452, loss_att=42.450, acc=0.772, loss=42.750, backward_time=0.297, grad_norm=43.419, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.515e-04, train_time=1.076
+[gpub010:0/16] 2024-02-11 21:04:37,115 (trainer:762) INFO: 44epoch:train:8901-9000batch: iter_time=2.377e-04, forward_time=0.375, loss_ctc=41.521, loss_att=40.303, acc=0.767, loss=40.668, backward_time=0.309, grad_norm=38.710, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.515e-04, train_time=1.485
+[gpub010:0/16] 2024-02-11 21:06:41,702 (trainer:762) INFO: 44epoch:train:9001-9100batch: iter_time=8.478e-05, forward_time=0.291, loss_ctc=47.450, loss_att=54.385, acc=0.738, loss=52.305, backward_time=0.297, grad_norm=46.816, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.515e-04, train_time=1.246
+[gpub010:0/16] 2024-02-11 21:08:51,024 (trainer:762) INFO: 44epoch:train:9101-9200batch: iter_time=8.314e-05, forward_time=0.291, loss_ctc=41.226, loss_att=43.952, acc=0.752, loss=43.134, backward_time=0.300, grad_norm=46.937, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.514e-04, train_time=1.293
+[gpub010:0/16] 2024-02-11 21:11:02,319 (trainer:762) INFO: 44epoch:train:9201-9300batch: iter_time=8.300e-05, forward_time=0.343, loss_ctc=43.777, loss_att=46.101, acc=0.769, loss=45.404, backward_time=0.356, grad_norm=45.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.514e-04, train_time=1.313
+[gpub010:0/16] 2024-02-11 21:13:14,205 (trainer:762) INFO: 44epoch:train:9301-9400batch: iter_time=8.237e-05, forward_time=0.290, loss_ctc=48.021, loss_att=45.403, acc=0.758, loss=46.188, backward_time=0.294, grad_norm=48.104, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.514e-04, train_time=1.319
+[gpub010:0/16] 2024-02-11 21:15:05,573 (trainer:762) INFO: 44epoch:train:9401-9500batch: iter_time=8.137e-05, forward_time=0.295, loss_ctc=41.624, loss_att=38.564, acc=0.772, loss=39.482, backward_time=0.302, grad_norm=47.132, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.514e-04, train_time=1.113
+[gpub010:0/16] 2024-02-11 21:17:28,664 (trainer:762) INFO: 44epoch:train:9501-9600batch: iter_time=2.759e-04, forward_time=0.326, loss_ctc=46.236, loss_att=43.265, acc=0.759, loss=44.157, backward_time=0.324, grad_norm=44.294, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.514e-04, train_time=1.431
+[gpub010:0/16] 2024-02-11 21:19:32,020 (trainer:762) INFO: 44epoch:train:9601-9700batch: iter_time=8.067e-05, forward_time=0.292, loss_ctc=51.175, loss_att=51.973, acc=0.740, loss=51.733, backward_time=0.296, grad_norm=49.000, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.514e-04, train_time=1.234
+[gpub010:0/16] 2024-02-11 21:22:07,433 (trainer:762) INFO: 44epoch:train:9701-9800batch: iter_time=9.051e-05, forward_time=0.411, loss_ctc=41.541, loss_att=37.018, acc=0.767, loss=38.375, backward_time=0.319, grad_norm=40.895, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.514e-04, train_time=1.554
+[gpub010:0/16] 2024-02-11 21:23:59,287 (trainer:762) INFO: 44epoch:train:9801-9900batch: iter_time=7.926e-05, forward_time=0.289, loss_ctc=42.966, loss_att=39.543, acc=0.757, loss=40.570, backward_time=0.295, grad_norm=49.893, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.514e-04, train_time=1.118
+[gpub010:0/16] 2024-02-11 21:26:17,040 (trainer:762) INFO: 44epoch:train:9901-10000batch: iter_time=8.059e-05, forward_time=0.294, loss_ctc=46.641, loss_att=48.180, acc=0.749, loss=47.718, backward_time=0.304, grad_norm=44.384, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.514e-04, train_time=1.378
+[gpub010:0/16] 2024-02-11 21:26:37,085 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-11 21:26:56,451 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 21:27:00,015 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbab4920a00>)
+[gpub010:0/16] 2024-02-11 21:27:00,015 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-11 21:27:00,056 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 21:33:55,199 (trainer:762) INFO: 44epoch:train:10001-10100batch: iter_time=3.209, forward_time=0.390, loss_ctc=40.285, loss_att=39.358, acc=0.781, loss=39.636, backward_time=0.310, grad_norm=41.063, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.513e-04, train_time=4.581
+[gpub010:0/16] 2024-02-11 21:35:58,968 (trainer:762) INFO: 44epoch:train:10101-10200batch: iter_time=8.533e-05, forward_time=0.292, loss_ctc=44.321, loss_att=45.139, acc=0.774, loss=44.893, backward_time=0.297, grad_norm=39.648, clip=100.000, loss_scale=7.113e+33, optim_step_time=0.093, optim0_lr0=1.513e-04, train_time=1.238
+[gpub010:0/16] 2024-02-11 21:38:13,897 (trainer:762) INFO: 44epoch:train:10201-10300batch: iter_time=8.043e-05, forward_time=0.344, loss_ctc=45.231, loss_att=50.387, acc=0.754, loss=48.840, backward_time=0.390, grad_norm=44.830, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.513e-04, train_time=1.349
+[gpub010:0/16] 2024-02-11 21:40:15,252 (trainer:762) INFO: 44epoch:train:10301-10400batch: iter_time=8.699e-05, forward_time=0.292, loss_ctc=44.762, loss_att=53.109, acc=0.745, loss=50.605, backward_time=0.297, grad_norm=46.679, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.513e-04, train_time=1.213
+[gpub010:0/16] 2024-02-11 21:42:31,199 (trainer:762) INFO: 44epoch:train:10401-10500batch: iter_time=8.482e-05, forward_time=0.299, loss_ctc=38.166, loss_att=41.618, acc=0.766, loss=40.582, backward_time=0.300, grad_norm=43.592, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.513e-04, train_time=1.359
+[gpub010:0/16] 2024-02-11 21:45:00,159 (trainer:762) INFO: 44epoch:train:10501-10600batch: iter_time=0.003, forward_time=0.321, loss_ctc=47.337, loss_att=48.620, acc=0.766, loss=48.235, backward_time=0.334, grad_norm=45.864, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.513e-04, train_time=1.490
+[gpub010:0/16] 2024-02-11 21:46:48,353 (trainer:762) INFO: 44epoch:train:10601-10700batch: iter_time=8.956e-05, forward_time=0.292, loss_ctc=47.165, loss_att=46.337, acc=0.770, loss=46.585, backward_time=0.298, grad_norm=44.970, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.513e-04, train_time=1.082
+[gpub010:0/16] 2024-02-11 21:49:02,672 (trainer:762) INFO: 44epoch:train:10701-10800batch: iter_time=9.196e-05, forward_time=0.298, loss_ctc=44.306, loss_att=40.686, acc=0.769, loss=41.772, backward_time=0.295, grad_norm=47.200, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.513e-04, train_time=1.342
+[gpub010:0/16] 2024-02-11 21:51:22,138 (trainer:762) INFO: 44epoch:train:10801-10900batch: iter_time=8.867e-05, forward_time=0.347, loss_ctc=52.144, loss_att=55.224, acc=0.752, loss=54.300, backward_time=0.369, grad_norm=51.073, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.513e-04, train_time=1.395
+[gpub010:0/16] 2024-02-11 21:52:51,467 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 21:53:12,933 (trainer:762) INFO: 44epoch:train:10901-11000batch: iter_time=8.591e-05, forward_time=0.292, loss_ctc=43.408, loss_att=41.811, acc=0.764, loss=42.290, backward_time=0.298, grad_norm=42.023, clip=100.000, loss_scale=9.336e+33, optim_step_time=0.094, optim0_lr0=1.512e-04, train_time=1.108
+[gpub010:0/16] 2024-02-11 21:55:50,359 (trainer:762) INFO: 44epoch:train:11001-11100batch: iter_time=8.454e-05, forward_time=0.304, loss_ctc=40.677, loss_att=35.633, acc=0.783, loss=37.146, backward_time=0.296, grad_norm=39.579, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.512e-04, train_time=1.573
+[gpub010:0/16] 2024-02-11 21:57:51,064 (trainer:762) INFO: 44epoch:train:11101-11200batch: iter_time=2.969e-04, forward_time=0.362, loss_ctc=42.053, loss_att=45.017, acc=0.747, loss=44.128, backward_time=0.318, grad_norm=49.062, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.512e-04, train_time=1.207
+[gpub010:0/16] 2024-02-11 21:59:05,433 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-11 21:59:25,152 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 21:59:28,771 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbd216ec130>)
+[gpub010:0/16] 2024-02-11 21:59:28,771 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-11 21:59:28,777 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 22:04:49,982 (trainer:762) INFO: 44epoch:train:11201-11300batch: iter_time=2.873, forward_time=0.400, loss_ctc=46.945, loss_att=44.299, acc=0.771, loss=45.093, backward_time=0.321, grad_norm=40.679, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.512e-04, train_time=4.189
+[gpub010:0/16] 2024-02-11 22:07:03,446 (trainer:762) INFO: 44epoch:train:11301-11400batch: iter_time=8.298e-05, forward_time=0.290, loss_ctc=43.544, loss_att=44.054, acc=0.767, loss=43.901, backward_time=0.294, grad_norm=42.783, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.512e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 22:09:15,643 (trainer:762) INFO: 44epoch:train:11401-11500batch: iter_time=8.097e-05, forward_time=0.371, loss_ctc=41.675, loss_att=40.651, acc=0.767, loss=40.959, backward_time=0.332, grad_norm=39.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.512e-04, train_time=1.322
+[gpub010:0/16] 2024-02-11 22:11:33,505 (trainer:762) INFO: 44epoch:train:11501-11600batch: iter_time=8.396e-05, forward_time=0.292, loss_ctc=46.359, loss_att=54.436, acc=0.740, loss=52.013, backward_time=0.298, grad_norm=46.232, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.512e-04, train_time=1.379
+[gpub010:0/16] 2024-02-11 22:13:49,929 (trainer:762) INFO: 44epoch:train:11601-11700batch: iter_time=8.539e-05, forward_time=0.374, loss_ctc=41.326, loss_att=43.098, acc=0.755, loss=42.566, backward_time=0.318, grad_norm=45.776, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.512e-04, train_time=1.364
+[gpub010:0/16] 2024-02-11 22:16:04,200 (trainer:762) INFO: 44epoch:train:11701-11800batch: iter_time=8.902e-05, forward_time=0.320, loss_ctc=43.637, loss_att=46.774, acc=0.766, loss=45.833, backward_time=0.309, grad_norm=47.677, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.511e-04, train_time=1.343
+[gpub010:0/16] 2024-02-11 22:18:12,425 (trainer:762) INFO: 44epoch:train:11801-11900batch: iter_time=9.082e-05, forward_time=0.355, loss_ctc=48.160, loss_att=46.200, acc=0.757, loss=46.788, backward_time=0.307, grad_norm=45.989, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.106, optim0_lr0=1.511e-04, train_time=1.282
+[gpub010:0/16] 2024-02-11 22:20:32,782 (trainer:762) INFO: 44epoch:train:11901-12000batch: iter_time=8.431e-05, forward_time=0.382, loss_ctc=41.410, loss_att=38.732, acc=0.771, loss=39.535, backward_time=0.328, grad_norm=46.521, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.511e-04, train_time=1.403
+[gpub010:0/16] 2024-02-11 22:22:51,627 (trainer:762) INFO: 44epoch:train:12001-12100batch: iter_time=8.327e-05, forward_time=0.294, loss_ctc=46.625, loss_att=44.009, acc=0.757, loss=44.794, backward_time=0.297, grad_norm=44.219, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.511e-04, train_time=1.388
+[gpub010:0/16] 2024-02-11 22:25:06,948 (trainer:762) INFO: 44epoch:train:12101-12200batch: iter_time=8.816e-05, forward_time=0.381, loss_ctc=50.655, loss_att=52.167, acc=0.743, loss=51.713, backward_time=0.324, grad_norm=47.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.103, optim0_lr0=1.511e-04, train_time=1.353
+[gpub010:0/16] 2024-02-11 22:27:06,321 (trainer:762) INFO: 44epoch:train:12201-12300batch: iter_time=8.483e-05, forward_time=0.291, loss_ctc=41.840, loss_att=37.826, acc=0.764, loss=39.031, backward_time=0.295, grad_norm=42.943, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.511e-04, train_time=1.193
+[gpub010:0/16] 2024-02-11 22:29:42,534 (trainer:762) INFO: 44epoch:train:12301-12400batch: iter_time=8.661e-05, forward_time=0.372, loss_ctc=42.835, loss_att=39.795, acc=0.759, loss=40.707, backward_time=0.316, grad_norm=43.655, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.511e-04, train_time=1.562
+[gpub010:0/16] 2024-02-11 22:31:45,933 (trainer:762) INFO: 44epoch:train:12401-12500batch: iter_time=7.760e-05, forward_time=0.293, loss_ctc=46.318, loss_att=47.830, acc=0.752, loss=47.377, backward_time=0.298, grad_norm=42.670, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.511e-04, train_time=1.234
+[gpub010:0/16] 2024-02-11 22:32:06,012 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-11 22:32:25,321 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 22:32:28,898 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbac6310820>)
+[gpub010:0/16] 2024-02-11 22:32:28,898 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-11 22:32:28,901 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 22:39:36,402 (trainer:762) INFO: 44epoch:train:12501-12600batch: iter_time=3.184, forward_time=0.385, loss_ctc=40.140, loss_att=38.574, acc=0.783, loss=39.044, backward_time=0.315, grad_norm=41.045, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.511e-04, train_time=4.704
+[gpub010:0/16] 2024-02-11 22:41:49,040 (trainer:762) INFO: 44epoch:train:12601-12700batch: iter_time=8.483e-05, forward_time=0.302, loss_ctc=44.305, loss_att=44.463, acc=0.775, loss=44.416, backward_time=0.299, grad_norm=39.882, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.510e-04, train_time=1.326
+[gpub010:0/16] 2024-02-11 22:43:52,535 (trainer:762) INFO: 44epoch:train:12701-12800batch: iter_time=8.735e-05, forward_time=0.317, loss_ctc=44.842, loss_att=49.896, acc=0.757, loss=48.380, backward_time=0.300, grad_norm=45.610, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.510e-04, train_time=1.235
+[gpub010:0/16] 2024-02-11 22:46:23,782 (trainer:762) INFO: 44epoch:train:12801-12900batch: iter_time=8.737e-05, forward_time=0.358, loss_ctc=44.791, loss_att=53.293, acc=0.745, loss=50.742, backward_time=0.318, grad_norm=47.293, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.510e-04, train_time=1.511
+[gpub010:0/16] 2024-02-11 22:48:34,994 (trainer:762) INFO: 44epoch:train:12901-13000batch: iter_time=9.141e-05, forward_time=0.307, loss_ctc=38.273, loss_att=41.675, acc=0.767, loss=40.655, backward_time=0.305, grad_norm=44.406, clip=100.000, loss_scale=6.231e+33, optim_step_time=0.106, optim0_lr0=1.510e-04, train_time=1.312
+[gpub010:0/16] 2024-02-11 22:50:47,821 (trainer:762) INFO: 44epoch:train:13001-13100batch: iter_time=8.706e-05, forward_time=0.297, loss_ctc=46.635, loss_att=48.165, acc=0.766, loss=47.706, backward_time=0.300, grad_norm=47.557, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.510e-04, train_time=1.328
+[gpub010:0/16] 2024-02-11 22:53:02,127 (trainer:762) INFO: 44epoch:train:13101-13200batch: iter_time=1.238e-04, forward_time=0.332, loss_ctc=47.468, loss_att=45.782, acc=0.770, loss=46.288, backward_time=0.320, grad_norm=47.193, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.510e-04, train_time=1.342
+[gpub010:0/16] 2024-02-11 22:54:59,657 (trainer:762) INFO: 44epoch:train:13201-13300batch: iter_time=8.985e-05, forward_time=0.306, loss_ctc=44.084, loss_att=40.741, acc=0.769, loss=41.744, backward_time=0.305, grad_norm=46.608, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.510e-04, train_time=1.175
+[gpub010:0/16] 2024-02-11 22:57:36,683 (trainer:762) INFO: 44epoch:train:13301-13400batch: iter_time=1.857e-04, forward_time=0.384, loss_ctc=52.365, loss_att=54.789, acc=0.753, loss=54.062, backward_time=0.315, grad_norm=48.220, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.510e-04, train_time=1.571
+[gpub010:0/16] 2024-02-11 22:59:29,508 (trainer:762) INFO: 44epoch:train:13401-13500batch: iter_time=8.450e-05, forward_time=0.295, loss_ctc=42.846, loss_att=41.353, acc=0.767, loss=41.801, backward_time=0.312, grad_norm=40.556, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.510e-04, train_time=1.128
+[gpub010:0/16] 2024-02-11 23:01:50,237 (trainer:762) INFO: 44epoch:train:13501-13600batch: iter_time=8.394e-05, forward_time=0.301, loss_ctc=40.315, loss_att=35.387, acc=0.782, loss=36.866, backward_time=0.302, grad_norm=38.729, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.509e-04, train_time=1.407
+[gpub010:0/16] 2024-02-11 23:04:03,710 (trainer:762) INFO: 44epoch:train:13601-13700batch: iter_time=9.865e-05, forward_time=0.339, loss_ctc=42.009, loss_att=45.023, acc=0.748, loss=44.119, backward_time=0.335, grad_norm=48.181, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.509e-04, train_time=1.335
+[gpub010:0/16] 2024-02-11 23:05:27,313 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-11 23:05:46,761 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-11 23:05:50,390 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba5838ff40>)
+[gpub010:0/16] 2024-02-11 23:05:50,390 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-11 23:05:50,394 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-11 23:11:46,461 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-11 23:12:02,535 (trainer:762) INFO: 44epoch:train:13701-13800batch: iter_time=3.530, forward_time=0.334, loss_ctc=46.759, loss_att=42.978, acc=0.779, loss=44.112, backward_time=0.308, grad_norm=39.947, clip=100.000, loss_scale=9.598e+33, optim_step_time=0.095, optim0_lr0=1.509e-04, train_time=4.788
+[gpub010:0/16] 2024-02-11 23:14:12,127 (trainer:762) INFO: 44epoch:train:13801-13900batch: iter_time=7.978e-05, forward_time=0.309, loss_ctc=43.105, loss_att=44.062, acc=0.776, loss=43.775, backward_time=0.299, grad_norm=41.674, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.509e-04, train_time=1.296
+[gpub010:0/16] 2024-02-11 23:16:35,739 (trainer:762) INFO: 44epoch:train:13901-14000batch: iter_time=8.466e-05, forward_time=0.341, loss_ctc=41.397, loss_att=41.429, acc=0.767, loss=41.419, backward_time=0.340, grad_norm=39.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.105, optim0_lr0=1.509e-04, train_time=1.436
+[gpub010:0/16] 2024-02-11 23:18:27,665 (trainer:762) INFO: 44epoch:train:14001-14100batch: iter_time=8.525e-05, forward_time=0.293, loss_ctc=46.723, loss_att=56.273, acc=0.743, loss=53.408, backward_time=0.300, grad_norm=46.331, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.509e-04, train_time=1.119
+[gpub010:0/16] 2024-02-11 23:20:56,467 (trainer:762) INFO: 44epoch:train:14101-14200batch: iter_time=4.150e-04, forward_time=0.355, loss_ctc=40.847, loss_att=44.632, acc=0.756, loss=43.496, backward_time=0.344, grad_norm=46.728, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.100, optim0_lr0=1.509e-04, train_time=1.488
+[gpub010:0/16] 2024-02-11 23:22:56,966 (trainer:762) INFO: 44epoch:train:14201-14300batch: iter_time=8.550e-05, forward_time=0.291, loss_ctc=43.756, loss_att=45.919, acc=0.775, loss=45.270, backward_time=0.296, grad_norm=45.444, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.509e-04, train_time=1.205
+[gpub010:0/16] 2024-02-11 23:24:59,544 (trainer:762) INFO: 44epoch:train:14301-14400batch: iter_time=8.318e-05, forward_time=0.313, loss_ctc=48.093, loss_att=46.228, acc=0.762, loss=46.788, backward_time=0.297, grad_norm=47.309, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.508e-04, train_time=1.225
+[gpub010:0/16] 2024-02-11 23:27:32,625 (trainer:762) INFO: 44epoch:train:14401-14500batch: iter_time=2.350e-04, forward_time=0.415, loss_ctc=40.957, loss_att=39.029, acc=0.780, loss=39.608, backward_time=0.322, grad_norm=42.860, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.508e-04, train_time=1.529
+[gpub010:0/16] 2024-02-11 23:29:26,193 (trainer:762) INFO: 44epoch:train:14501-14600batch: iter_time=8.052e-05, forward_time=0.293, loss_ctc=46.493, loss_att=44.862, acc=0.767, loss=45.351, backward_time=0.298, grad_norm=44.646, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.508e-04, train_time=1.137
+[gpub010:0/16] 2024-02-11 23:31:54,074 (trainer:762) INFO: 44epoch:train:14601-14700batch: iter_time=7.973e-05, forward_time=0.311, loss_ctc=50.494, loss_att=53.127, acc=0.751, loss=52.337, backward_time=0.308, grad_norm=48.463, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.508e-04, train_time=1.478
+[gpub010:0/16] 2024-02-11 23:34:08,506 (trainer:762) INFO: 44epoch:train:14701-14800batch: iter_time=0.001, forward_time=0.420, loss_ctc=41.356, loss_att=36.947, acc=0.775, loss=38.270, backward_time=0.330, grad_norm=39.112, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.508e-04, train_time=1.343
+[gpub010:0/16] 2024-02-11 23:36:13,813 (trainer:762) INFO: 44epoch:train:14801-14900batch: iter_time=8.372e-05, forward_time=0.304, loss_ctc=42.749, loss_att=40.399, acc=0.757, loss=41.104, backward_time=0.296, grad_norm=45.499, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.508e-04, train_time=1.254
+[gpub010:0/16] 2024-02-11 23:38:34,046 (trainer:762) INFO: 44epoch:train:14901-15000batch: iter_time=3.777e-04, forward_time=0.366, loss_ctc=46.031, loss_att=47.283, acc=0.763, loss=46.907, backward_time=0.312, grad_norm=42.696, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.124, optim0_lr0=1.508e-04, train_time=1.402
+[gpub010:0/16] 2024-02-12 00:16:51,582 (trainer:361) INFO: 44epoch results: [train] iter_time=0.250, forward_time=0.324, loss_ctc=44.941, loss_att=44.930, acc=0.760, loss=44.933, backward_time=0.310, grad_norm=45.340, clip=100.000, loss_scale=5.083e+33, optim_step_time=0.096, optim0_lr0=1.516e-04, train_time=1.579, time=6 hours, 35 minutes and 15.43 seconds, total_count=690000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=34.331, cer_ctc=0.174, loss_att=36.037, acc=0.702, cer=0.295, wer=0.989, loss=35.525, time=37 minutes and 53.4 seconds, total_count=214866, gpu_max_cached_mem_GB=42.092
+[gpub010:0/16] 2024-02-12 00:17:01,726 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-02-12 00:17:01,804 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/39epoch.pth
+[gpub010:0/16] 2024-02-12 00:17:01,804 (trainer:290) INFO: 45/45epoch started. Estimated time to finish: 7 hours, 16 minutes and 52.87 seconds
+[gpub010:0/16] 2024-02-12 00:17:01,815 (multiple_iter_factory:32) INFO: Building 0th iter-factory...
+[gpub010:0/16] 2024-02-12 00:17:20,448 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 00:17:23,878 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.8", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.8", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.8", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.8", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb22a3e95a0>)
+[gpub010:0/16] 2024-02-12 00:17:23,878 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.8, 
+[gpub010:0/16] 2024-02-12 00:17:23,881 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 00:24:14,665 (trainer:762) INFO: 45epoch:train:1-100batch: iter_time=3.087, forward_time=0.330, loss_ctc=45.604, loss_att=48.084, acc=0.760, loss=47.340, backward_time=0.304, grad_norm=43.861, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.508e-04, train_time=4.328
+[gpub010:0/16] 2024-02-12 00:26:09,193 (trainer:762) INFO: 45epoch:train:101-200batch: iter_time=8.028e-05, forward_time=0.297, loss_ctc=44.787, loss_att=39.089, acc=0.778, loss=40.798, backward_time=0.296, grad_norm=42.092, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.508e-04, train_time=1.145
+[gpub010:0/16] 2024-02-12 00:28:30,854 (trainer:762) INFO: 45epoch:train:201-300batch: iter_time=8.308e-05, forward_time=0.333, loss_ctc=49.519, loss_att=45.390, acc=0.762, loss=46.629, backward_time=0.316, grad_norm=43.125, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.507e-04, train_time=1.416
+[gpub010:0/16] 2024-02-12 00:30:40,845 (trainer:762) INFO: 45epoch:train:301-400batch: iter_time=4.609e-04, forward_time=0.314, loss_ctc=37.336, loss_att=34.083, acc=0.779, loss=35.059, backward_time=0.308, grad_norm=40.330, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.507e-04, train_time=1.300
+[gpub010:0/16] 2024-02-12 00:32:55,496 (trainer:762) INFO: 45epoch:train:401-500batch: iter_time=7.967e-05, forward_time=0.307, loss_ctc=48.241, loss_att=47.416, acc=0.790, loss=47.664, backward_time=0.301, grad_norm=42.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.507e-04, train_time=1.346
+[gpub010:0/16] 2024-02-12 00:35:16,885 (trainer:762) INFO: 45epoch:train:501-600batch: iter_time=7.871e-05, forward_time=0.364, loss_ctc=62.130, loss_att=51.874, acc=0.763, loss=54.951, backward_time=0.334, grad_norm=51.393, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.107, optim0_lr0=1.507e-04, train_time=1.413
+[gpub010:0/16] 2024-02-12 00:37:21,283 (trainer:762) INFO: 45epoch:train:601-700batch: iter_time=8.397e-05, forward_time=0.303, loss_ctc=53.023, loss_att=46.606, acc=0.770, loss=48.531, backward_time=0.299, grad_norm=49.542, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.507e-04, train_time=1.244
+[gpub010:0/16] 2024-02-12 00:39:42,005 (trainer:762) INFO: 45epoch:train:701-800batch: iter_time=8.500e-05, forward_time=0.298, loss_ctc=51.596, loss_att=41.695, acc=0.768, loss=44.665, backward_time=0.299, grad_norm=48.494, clip=100.000, loss_scale=5.971e+33, optim_step_time=0.094, optim0_lr0=1.507e-04, train_time=1.407
+[gpub010:0/16] 2024-02-12 00:41:53,495 (trainer:762) INFO: 45epoch:train:801-900batch: iter_time=8.319e-05, forward_time=0.348, loss_ctc=46.238, loss_att=44.107, acc=0.760, loss=44.746, backward_time=0.330, grad_norm=43.670, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.507e-04, train_time=1.314
+[gpub010:0/16] 2024-02-12 00:44:05,405 (trainer:762) INFO: 45epoch:train:901-1000batch: iter_time=7.940e-05, forward_time=0.297, loss_ctc=44.598, loss_att=46.158, acc=0.774, loss=45.690, backward_time=0.297, grad_norm=40.350, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.507e-04, train_time=1.319
+[gpub010:0/16] 2024-02-12 00:46:10,212 (trainer:762) INFO: 45epoch:train:1001-1100batch: iter_time=2.265e-04, forward_time=0.315, loss_ctc=40.559, loss_att=44.004, acc=0.757, loss=42.970, backward_time=0.318, grad_norm=43.678, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.507e-04, train_time=1.248
+[gpub010:0/16] 2024-02-12 00:48:28,455 (trainer:762) INFO: 45epoch:train:1101-1200batch: iter_time=8.106e-05, forward_time=0.331, loss_ctc=51.074, loss_att=46.655, acc=0.744, loss=47.980, backward_time=0.316, grad_norm=55.772, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.506e-04, train_time=1.382
+[gpub010:0/16] 2024-02-12 00:49:49,681 (multiple_iter_factory:32) INFO: Building 1th iter-factory...
+[gpub010:0/16] 2024-02-12 00:50:08,823 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 00:50:12,410 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.5", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.5", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.5", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.5", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fc0e8b7e290>)
+[gpub010:0/16] 2024-02-12 00:50:12,410 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.5, 
+[gpub010:0/16] 2024-02-12 00:50:12,432 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 00:55:10,674 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 00:56:03,765 (trainer:762) INFO: 45epoch:train:1201-1300batch: iter_time=3.296, forward_time=0.333, loss_ctc=43.705, loss_att=49.027, acc=0.754, loss=47.430, backward_time=0.299, grad_norm=47.848, clip=100.000, loss_scale=7.815e+33, optim_step_time=0.094, optim0_lr0=1.506e-04, train_time=4.553
+[gpub010:0/16] 2024-02-12 00:58:15,896 (trainer:762) INFO: 45epoch:train:1301-1400batch: iter_time=8.014e-05, forward_time=0.326, loss_ctc=44.310, loss_att=43.992, acc=0.755, loss=44.088, backward_time=0.335, grad_norm=41.488, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.506e-04, train_time=1.321
+[gpub010:0/16] 2024-02-12 01:00:31,681 (trainer:762) INFO: 45epoch:train:1401-1500batch: iter_time=7.964e-05, forward_time=0.304, loss_ctc=46.981, loss_att=40.713, acc=0.770, loss=42.593, backward_time=0.294, grad_norm=43.719, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.506e-04, train_time=1.357
+[gpub010:0/16] 2024-02-12 01:02:46,070 (trainer:762) INFO: 45epoch:train:1501-1600batch: iter_time=8.389e-05, forward_time=0.294, loss_ctc=44.054, loss_att=39.542, acc=0.767, loss=40.896, backward_time=0.296, grad_norm=40.805, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.506e-04, train_time=1.344
+[gpub010:0/16] 2024-02-12 01:04:55,501 (trainer:762) INFO: 45epoch:train:1601-1700batch: iter_time=8.300e-05, forward_time=0.348, loss_ctc=41.551, loss_att=40.595, acc=0.776, loss=40.882, backward_time=0.315, grad_norm=40.095, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.506e-04, train_time=1.294
+[gpub010:0/16] 2024-02-12 01:07:09,510 (trainer:762) INFO: 45epoch:train:1701-1800batch: iter_time=8.059e-05, forward_time=0.297, loss_ctc=47.712, loss_att=49.038, acc=0.773, loss=48.640, backward_time=0.296, grad_norm=42.040, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.506e-04, train_time=1.340
+[gpub010:0/16] 2024-02-12 01:09:14,156 (trainer:762) INFO: 45epoch:train:1801-1900batch: iter_time=8.301e-05, forward_time=0.294, loss_ctc=59.287, loss_att=48.075, acc=0.761, loss=51.438, backward_time=0.297, grad_norm=49.589, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.506e-04, train_time=1.245
+[gpub010:0/16] 2024-02-12 01:11:33,817 (trainer:762) INFO: 45epoch:train:1901-2000batch: iter_time=8.374e-05, forward_time=0.324, loss_ctc=60.295, loss_att=47.483, acc=0.765, loss=51.327, backward_time=0.319, grad_norm=50.435, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.099, optim0_lr0=1.506e-04, train_time=1.397
+[gpub010:0/16] 2024-02-12 01:13:52,799 (trainer:762) INFO: 45epoch:train:2001-2100batch: iter_time=7.829e-05, forward_time=0.330, loss_ctc=46.907, loss_att=45.648, acc=0.738, loss=46.026, backward_time=0.323, grad_norm=45.888, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.505e-04, train_time=1.390
+[gpub010:0/16] 2024-02-12 01:16:17,791 (trainer:762) INFO: 45epoch:train:2101-2200batch: iter_time=7.858e-05, forward_time=0.300, loss_ctc=40.683, loss_att=39.308, acc=0.773, loss=39.721, backward_time=0.295, grad_norm=39.895, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.505e-04, train_time=1.450
+[gpub010:0/16] 2024-02-12 01:18:33,818 (trainer:762) INFO: 45epoch:train:2201-2300batch: iter_time=8.078e-05, forward_time=0.311, loss_ctc=42.477, loss_att=50.323, acc=0.749, loss=47.969, backward_time=0.303, grad_norm=41.856, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.505e-04, train_time=1.360
+[gpub010:0/16] 2024-02-12 01:20:53,167 (trainer:762) INFO: 45epoch:train:2301-2400batch: iter_time=7.921e-05, forward_time=0.353, loss_ctc=44.709, loss_att=40.898, acc=0.769, loss=42.041, backward_time=0.306, grad_norm=43.138, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.505e-04, train_time=1.394
+[gpub010:0/16] 2024-02-12 01:22:59,959 (trainer:762) INFO: 45epoch:train:2401-2500batch: iter_time=7.740e-05, forward_time=0.301, loss_ctc=47.786, loss_att=46.282, acc=0.742, loss=46.733, backward_time=0.298, grad_norm=55.217, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.505e-04, train_time=1.268
+[gpub010:0/16] 2024-02-12 01:23:19,987 (multiple_iter_factory:32) INFO: Building 2th iter-factory...
+[gpub010:0/16] 2024-02-12 01:23:39,266 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 01:23:42,779 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.2", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.2", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.2", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.2", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fb1ee6483d0>)
+[gpub010:0/16] 2024-02-12 01:23:42,779 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.2, 
+[gpub010:0/16] 2024-02-12 01:23:42,782 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 01:30:44,781 (trainer:762) INFO: 45epoch:train:2501-2600batch: iter_time=3.226, forward_time=0.321, loss_ctc=44.656, loss_att=48.345, acc=0.761, loss=47.239, backward_time=0.304, grad_norm=44.202, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.505e-04, train_time=4.648
+[gpub010:0/16] 2024-02-12 01:33:06,218 (trainer:762) INFO: 45epoch:train:2601-2700batch: iter_time=8.637e-05, forward_time=0.340, loss_ctc=42.463, loss_att=39.171, acc=0.781, loss=40.159, backward_time=0.299, grad_norm=40.676, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.113, optim0_lr0=1.505e-04, train_time=1.415
+[gpub010:0/16] 2024-02-12 01:35:15,702 (trainer:762) INFO: 45epoch:train:2701-2800batch: iter_time=8.265e-05, forward_time=0.319, loss_ctc=48.966, loss_att=44.679, acc=0.766, loss=45.965, backward_time=0.317, grad_norm=43.126, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.505e-04, train_time=1.294
+[gpub010:0/16] 2024-02-12 01:37:23,031 (trainer:762) INFO: 45epoch:train:2801-2900batch: iter_time=8.311e-05, forward_time=0.303, loss_ctc=36.657, loss_att=33.432, acc=0.783, loss=34.400, backward_time=0.338, grad_norm=38.932, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.505e-04, train_time=1.273
+[gpub010:0/16] 2024-02-12 01:39:28,922 (trainer:762) INFO: 45epoch:train:2901-3000batch: iter_time=9.182e-05, forward_time=0.304, loss_ctc=47.543, loss_att=46.937, acc=0.793, loss=47.119, backward_time=0.301, grad_norm=43.351, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.504e-04, train_time=1.259
+[gpub010:0/16] 2024-02-12 01:41:44,881 (trainer:762) INFO: 45epoch:train:3001-3100batch: iter_time=8.848e-05, forward_time=0.306, loss_ctc=59.059, loss_att=50.490, acc=0.766, loss=53.061, backward_time=0.319, grad_norm=50.087, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.504e-04, train_time=1.359
+[gpub010:0/16] 2024-02-12 01:43:43,171 (trainer:762) INFO: 45epoch:train:3101-3200batch: iter_time=8.549e-05, forward_time=0.360, loss_ctc=51.540, loss_att=46.209, acc=0.770, loss=47.808, backward_time=0.302, grad_norm=47.351, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.504e-04, train_time=1.183
+[gpub010:0/16] 2024-02-12 01:45:50,316 (trainer:762) INFO: 45epoch:train:3201-3300batch: iter_time=8.191e-05, forward_time=0.301, loss_ctc=50.793, loss_att=41.909, acc=0.770, loss=44.574, backward_time=0.298, grad_norm=52.275, clip=100.000, loss_scale=7.737e+33, optim_step_time=0.093, optim0_lr0=1.504e-04, train_time=1.271
+[gpub010:0/16] 2024-02-12 01:48:08,638 (trainer:762) INFO: 45epoch:train:3301-3400batch: iter_time=8.353e-05, forward_time=0.309, loss_ctc=44.417, loss_att=44.162, acc=0.761, loss=44.238, backward_time=0.310, grad_norm=42.970, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.504e-04, train_time=1.383
+[gpub010:0/16] 2024-02-12 01:50:24,184 (trainer:762) INFO: 45epoch:train:3401-3500batch: iter_time=7.980e-05, forward_time=0.346, loss_ctc=43.814, loss_att=45.296, acc=0.779, loss=44.851, backward_time=0.325, grad_norm=40.057, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.504e-04, train_time=1.355
+[gpub010:0/16] 2024-02-12 01:52:24,471 (trainer:762) INFO: 45epoch:train:3501-3600batch: iter_time=8.131e-05, forward_time=0.296, loss_ctc=40.614, loss_att=44.727, acc=0.757, loss=43.493, backward_time=0.295, grad_norm=42.164, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.504e-04, train_time=1.203
+[gpub010:0/16] 2024-02-12 01:54:23,085 (trainer:762) INFO: 45epoch:train:3601-3700batch: iter_time=8.287e-05, forward_time=0.312, loss_ctc=49.919, loss_att=45.299, acc=0.750, loss=46.685, backward_time=0.331, grad_norm=51.689, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.504e-04, train_time=1.186
+[gpub010:0/16] 2024-02-12 01:56:05,424 (multiple_iter_factory:32) INFO: Building 3th iter-factory...
+[gpub010:0/16] 2024-02-12 01:56:25,251 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 01:56:28,775 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.9", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.9", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.9", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.9", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbcdbdd77c0>)
+[gpub010:0/16] 2024-02-12 01:56:28,775 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.9, 
+[gpub010:0/16] 2024-02-12 01:56:28,779 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 02:02:04,929 (trainer:762) INFO: 45epoch:train:3701-3800batch: iter_time=3.122, forward_time=0.355, loss_ctc=42.378, loss_att=46.988, acc=0.762, loss=45.605, backward_time=0.333, grad_norm=44.322, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.503e-04, train_time=4.619
+[gpub010:0/16] 2024-02-12 02:03:58,344 (trainer:762) INFO: 45epoch:train:3801-3900batch: iter_time=7.555e-05, forward_time=0.310, loss_ctc=44.280, loss_att=43.136, acc=0.759, loss=43.479, backward_time=0.304, grad_norm=41.027, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.503e-04, train_time=1.133
+[gpub010:0/16] 2024-02-12 02:06:23,452 (trainer:762) INFO: 45epoch:train:3901-4000batch: iter_time=4.247e-04, forward_time=0.338, loss_ctc=46.335, loss_att=40.160, acc=0.775, loss=42.013, backward_time=0.298, grad_norm=43.416, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.503e-04, train_time=1.452
+[gpub010:0/16] 2024-02-12 02:08:23,796 (trainer:762) INFO: 45epoch:train:4001-4100batch: iter_time=9.074e-05, forward_time=0.311, loss_ctc=43.299, loss_att=38.859, acc=0.773, loss=40.191, backward_time=0.311, grad_norm=41.349, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.503e-04, train_time=1.203
+[gpub010:0/16] 2024-02-12 02:10:23,909 (trainer:762) INFO: 45epoch:train:4101-4200batch: iter_time=8.300e-05, forward_time=0.365, loss_ctc=41.384, loss_att=40.581, acc=0.779, loss=40.822, backward_time=0.304, grad_norm=39.885, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.503e-04, train_time=1.201
+[gpub010:0/16] 2024-02-12 02:12:47,652 (trainer:762) INFO: 45epoch:train:4201-4300batch: iter_time=8.994e-05, forward_time=0.309, loss_ctc=47.280, loss_att=49.115, acc=0.776, loss=48.565, backward_time=0.308, grad_norm=42.302, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.503e-04, train_time=1.437
+[gpub010:0/16] 2024-02-12 02:14:44,508 (trainer:762) INFO: 45epoch:train:4301-4400batch: iter_time=8.348e-05, forward_time=0.303, loss_ctc=58.154, loss_att=47.639, acc=0.763, loss=50.793, backward_time=0.303, grad_norm=51.677, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.503e-04, train_time=1.168
+[gpub010:0/16] 2024-02-12 02:16:48,736 (trainer:762) INFO: 45epoch:train:4401-4500batch: iter_time=8.182e-05, forward_time=0.359, loss_ctc=59.124, loss_att=47.175, acc=0.767, loss=50.759, backward_time=0.333, grad_norm=53.827, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.503e-04, train_time=1.242
+[gpub010:0/16] 2024-02-12 02:18:58,823 (trainer:762) INFO: 45epoch:train:4501-4600batch: iter_time=8.133e-05, forward_time=0.297, loss_ctc=45.551, loss_att=45.253, acc=0.741, loss=45.342, backward_time=0.302, grad_norm=45.268, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.503e-04, train_time=1.301
+[gpub010:0/16] 2024-02-12 02:21:11,802 (trainer:762) INFO: 45epoch:train:4601-4700batch: iter_time=8.518e-05, forward_time=0.341, loss_ctc=40.152, loss_att=38.786, acc=0.776, loss=39.195, backward_time=0.370, grad_norm=40.190, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.502e-04, train_time=1.329
+[gpub010:0/16] 2024-02-12 02:23:02,583 (trainer:762) INFO: 45epoch:train:4701-4800batch: iter_time=8.384e-05, forward_time=0.292, loss_ctc=42.123, loss_att=49.637, acc=0.754, loss=47.383, backward_time=0.298, grad_norm=44.351, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.502e-04, train_time=1.108
+[gpub010:0/16] 2024-02-12 02:24:54,083 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 02:25:22,351 (trainer:762) INFO: 45epoch:train:4801-4900batch: iter_time=3.309e-04, forward_time=0.360, loss_ctc=43.438, loss_att=39.841, acc=0.773, loss=40.920, backward_time=0.321, grad_norm=43.323, clip=100.000, loss_scale=9.021e+33, optim_step_time=0.099, optim0_lr0=1.502e-04, train_time=1.397
+[gpub010:0/16] 2024-02-12 02:27:29,667 (trainer:762) INFO: 45epoch:train:4901-5000batch: iter_time=8.339e-05, forward_time=0.297, loss_ctc=47.415, loss_att=45.326, acc=0.745, loss=45.953, backward_time=0.304, grad_norm=54.482, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.502e-04, train_time=1.273
+[gpub010:0/16] 2024-02-12 02:27:49,697 (multiple_iter_factory:32) INFO: Building 4th iter-factory...
+[gpub010:0/16] 2024-02-12 02:28:09,198 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 02:28:12,687 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.7", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.7", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.7", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.7", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbaad79f220>)
+[gpub010:0/16] 2024-02-12 02:28:12,687 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.7, 
+[gpub010:0/16] 2024-02-12 02:28:12,690 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 02:35:02,151 (trainer:762) INFO: 45epoch:train:5001-5100batch: iter_time=3.218, forward_time=0.338, loss_ctc=44.368, loss_att=45.195, acc=0.760, loss=44.947, backward_time=0.300, grad_norm=41.737, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.502e-04, train_time=4.524
+[gpub010:0/16] 2024-02-12 02:37:09,467 (trainer:762) INFO: 45epoch:train:5101-5200batch: iter_time=7.788e-05, forward_time=0.301, loss_ctc=41.909, loss_att=38.559, acc=0.775, loss=39.564, backward_time=0.314, grad_norm=39.786, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.502e-04, train_time=1.272
+[gpub010:0/16] 2024-02-12 02:39:19,409 (trainer:762) INFO: 45epoch:train:5201-5300batch: iter_time=7.843e-05, forward_time=0.334, loss_ctc=48.495, loss_att=44.980, acc=0.757, loss=46.034, backward_time=0.300, grad_norm=41.957, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.502e-04, train_time=1.300
+[gpub010:0/16] 2024-02-12 02:41:30,132 (trainer:762) INFO: 45epoch:train:5301-5400batch: iter_time=8.952e-05, forward_time=0.344, loss_ctc=36.429, loss_att=32.640, acc=0.788, loss=33.777, backward_time=0.301, grad_norm=38.535, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.502e-04, train_time=1.307
+[gpub010:0/16] 2024-02-12 02:43:43,731 (trainer:762) INFO: 45epoch:train:5401-5500batch: iter_time=8.200e-05, forward_time=0.294, loss_ctc=46.866, loss_att=46.437, acc=0.786, loss=46.566, backward_time=0.299, grad_norm=40.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.502e-04, train_time=1.336
+[gpub010:0/16] 2024-02-12 02:45:54,426 (trainer:762) INFO: 45epoch:train:5501-5600batch: iter_time=8.267e-05, forward_time=0.299, loss_ctc=58.393, loss_att=50.163, acc=0.763, loss=52.632, backward_time=0.301, grad_norm=50.926, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.501e-04, train_time=1.307
+[gpub010:0/16] 2024-02-12 02:48:03,621 (trainer:762) INFO: 45epoch:train:5601-5700batch: iter_time=8.668e-05, forward_time=0.381, loss_ctc=51.081, loss_att=45.773, acc=0.769, loss=47.365, backward_time=0.328, grad_norm=47.626, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.501e-04, train_time=1.292
+[gpub010:0/16] 2024-02-12 02:50:31,399 (trainer:762) INFO: 45epoch:train:5701-5800batch: iter_time=9.716e-05, forward_time=0.294, loss_ctc=50.090, loss_att=41.729, acc=0.767, loss=44.237, backward_time=0.297, grad_norm=49.147, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.501e-04, train_time=1.478
+[gpub010:0/16] 2024-02-12 02:52:38,933 (trainer:762) INFO: 45epoch:train:5801-5900batch: iter_time=9.016e-05, forward_time=0.292, loss_ctc=43.553, loss_att=42.748, acc=0.755, loss=42.990, backward_time=0.302, grad_norm=42.194, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.501e-04, train_time=1.275
+[gpub010:0/16] 2024-02-12 02:54:40,850 (trainer:762) INFO: 45epoch:train:5901-6000batch: iter_time=5.547e-04, forward_time=0.366, loss_ctc=43.742, loss_att=45.437, acc=0.771, loss=44.929, backward_time=0.306, grad_norm=41.358, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.101, optim0_lr0=1.501e-04, train_time=1.219
+[gpub010:0/16] 2024-02-12 02:57:09,201 (trainer:762) INFO: 45epoch:train:6001-6100batch: iter_time=8.671e-05, forward_time=0.294, loss_ctc=40.051, loss_att=42.942, acc=0.757, loss=42.075, backward_time=0.300, grad_norm=40.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.501e-04, train_time=1.484
+[gpub010:0/16] 2024-02-12 02:58:59,460 (trainer:762) INFO: 45epoch:train:6101-6200batch: iter_time=9.323e-05, forward_time=0.292, loss_ctc=49.614, loss_att=45.619, acc=0.745, loss=46.818, backward_time=0.298, grad_norm=54.248, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.501e-04, train_time=1.102
+[gpub010:0/16] 2024-02-12 03:00:32,023 (multiple_iter_factory:32) INFO: Building 5th iter-factory...
+[gpub010:0/16] 2024-02-12 03:00:51,088 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 03:00:54,893 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.4", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.4", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.4", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.4", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba4c9175e0>)
+[gpub010:0/16] 2024-02-12 03:00:54,893 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.4, 
+[gpub010:0/16] 2024-02-12 03:00:54,896 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 03:06:34,043 (trainer:762) INFO: 45epoch:train:6201-6300batch: iter_time=3.142, forward_time=0.401, loss_ctc=41.822, loss_att=46.113, acc=0.768, loss=44.826, backward_time=0.318, grad_norm=46.624, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.110, optim0_lr0=1.501e-04, train_time=4.546
+[gpub010:0/16] 2024-02-12 03:08:34,640 (trainer:762) INFO: 45epoch:train:6301-6400batch: iter_time=7.568e-05, forward_time=0.359, loss_ctc=43.848, loss_att=44.916, acc=0.769, loss=44.595, backward_time=0.335, grad_norm=40.345, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.098, optim0_lr0=1.501e-04, train_time=1.205
+[gpub010:0/16] 2024-02-12 03:10:38,158 (trainer:762) INFO: 45epoch:train:6401-6500batch: iter_time=8.220e-05, forward_time=0.290, loss_ctc=46.049, loss_att=39.605, acc=0.785, loss=41.538, backward_time=0.296, grad_norm=61.644, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.500e-04, train_time=1.235
+[gpub010:0/16] 2024-02-12 03:12:58,182 (trainer:762) INFO: 45epoch:train:6501-6600batch: iter_time=9.014e-05, forward_time=0.295, loss_ctc=42.878, loss_att=38.479, acc=0.784, loss=39.799, backward_time=0.299, grad_norm=40.260, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.500e-04, train_time=1.400
+[gpub010:0/16] 2024-02-12 03:15:12,351 (trainer:762) INFO: 45epoch:train:6601-6700batch: iter_time=9.129e-05, forward_time=0.343, loss_ctc=41.141, loss_att=41.423, acc=0.778, loss=41.338, backward_time=0.318, grad_norm=40.224, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.500e-04, train_time=1.342
+[gpub010:0/16] 2024-02-12 03:17:06,135 (trainer:762) INFO: 45epoch:train:6701-6800batch: iter_time=9.508e-05, forward_time=0.337, loss_ctc=46.509, loss_att=48.883, acc=0.786, loss=48.170, backward_time=0.306, grad_norm=41.011, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.500e-04, train_time=1.137
+[gpub010:0/16] 2024-02-12 03:19:42,989 (trainer:762) INFO: 45epoch:train:6801-6900batch: iter_time=9.225e-05, forward_time=0.298, loss_ctc=57.698, loss_att=47.979, acc=0.767, loss=50.895, backward_time=0.307, grad_norm=50.647, clip=100.000, loss_scale=6.542e+33, optim_step_time=0.094, optim0_lr0=1.500e-04, train_time=1.568
+[gpub010:0/16] 2024-02-12 03:21:44,423 (trainer:762) INFO: 45epoch:train:6901-7000batch: iter_time=9.352e-05, forward_time=0.348, loss_ctc=58.003, loss_att=46.978, acc=0.777, loss=50.285, backward_time=0.333, grad_norm=51.568, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.500e-04, train_time=1.215
+[gpub010:0/16] 2024-02-12 03:23:46,021 (trainer:762) INFO: 45epoch:train:7001-7100batch: iter_time=8.826e-05, forward_time=0.291, loss_ctc=45.457, loss_att=44.914, acc=0.756, loss=45.077, backward_time=0.297, grad_norm=44.693, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.500e-04, train_time=1.215
+[gpub010:0/16] 2024-02-12 03:25:56,755 (trainer:762) INFO: 45epoch:train:7101-7200batch: iter_time=8.404e-05, forward_time=0.295, loss_ctc=39.979, loss_att=38.677, acc=0.783, loss=39.067, backward_time=0.301, grad_norm=37.970, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.500e-04, train_time=1.307
+[gpub010:0/16] 2024-02-12 03:27:58,759 (trainer:762) INFO: 45epoch:train:7201-7300batch: iter_time=8.761e-05, forward_time=0.394, loss_ctc=42.135, loss_att=50.604, acc=0.761, loss=48.064, backward_time=0.307, grad_norm=41.956, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.500e-04, train_time=1.220
+[gpub010:0/16] 2024-02-12 03:30:17,118 (trainer:762) INFO: 45epoch:train:7301-7400batch: iter_time=8.176e-05, forward_time=0.385, loss_ctc=43.875, loss_att=40.204, acc=0.775, loss=41.305, backward_time=0.325, grad_norm=40.959, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.099, optim0_lr0=1.499e-04, train_time=1.383
+[gpub010:0/16] 2024-02-12 03:31:26,202 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 03:32:19,378 (trainer:762) INFO: 45epoch:train:7401-7500batch: iter_time=8.081e-05, forward_time=0.316, loss_ctc=46.583, loss_att=47.023, acc=0.750, loss=46.891, backward_time=0.297, grad_norm=53.488, clip=100.000, loss_scale=7.920e+33, optim_step_time=0.093, optim0_lr0=1.499e-04, train_time=1.222
+[gpub010:0/16] 2024-02-12 03:32:39,405 (multiple_iter_factory:32) INFO: Building 6th iter-factory...
+[gpub010:0/16] 2024-02-12 03:32:59,061 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 03:33:02,528 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.1", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.1", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.1", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.1", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba39863fa0>)
+[gpub010:0/16] 2024-02-12 03:33:02,528 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.1, 
+[gpub010:0/16] 2024-02-12 03:33:02,532 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 03:39:06,453 (trainer:762) INFO: 45epoch:train:7501-7600batch: iter_time=2.685, forward_time=0.295, loss_ctc=44.453, loss_att=46.779, acc=0.756, loss=46.081, backward_time=0.301, grad_norm=42.706, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.499e-04, train_time=4.070
+[gpub010:0/16] 2024-02-12 03:41:07,561 (trainer:762) INFO: 45epoch:train:7601-7700batch: iter_time=7.896e-05, forward_time=0.395, loss_ctc=41.892, loss_att=38.991, acc=0.776, loss=39.861, backward_time=0.310, grad_norm=40.024, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.499e-04, train_time=1.211
+[gpub010:0/16] 2024-02-12 03:43:10,533 (trainer:762) INFO: 45epoch:train:7701-7800batch: iter_time=8.248e-05, forward_time=0.295, loss_ctc=48.246, loss_att=44.687, acc=0.759, loss=45.755, backward_time=0.304, grad_norm=44.018, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.499e-04, train_time=1.230
+[gpub010:0/16] 2024-02-12 03:45:32,836 (trainer:762) INFO: 45epoch:train:7801-7900batch: iter_time=4.234e-04, forward_time=0.368, loss_ctc=36.236, loss_att=32.597, acc=0.788, loss=33.689, backward_time=0.308, grad_norm=36.871, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.499e-04, train_time=1.423
+[gpub010:0/16] 2024-02-12 03:47:40,583 (trainer:762) INFO: 45epoch:train:7901-8000batch: iter_time=8.598e-05, forward_time=0.295, loss_ctc=46.828, loss_att=46.379, acc=0.787, loss=46.514, backward_time=0.299, grad_norm=41.492, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.499e-04, train_time=1.277
+[gpub010:0/16] 2024-02-12 03:49:30,199 (trainer:762) INFO: 45epoch:train:8001-8100batch: iter_time=8.353e-05, forward_time=0.297, loss_ctc=59.193, loss_att=49.914, acc=0.763, loss=52.698, backward_time=0.307, grad_norm=54.663, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.499e-04, train_time=1.096
+[gpub010:0/16] 2024-02-12 03:51:56,831 (trainer:762) INFO: 45epoch:train:8101-8200batch: iter_time=1.910e-04, forward_time=0.378, loss_ctc=50.856, loss_att=45.275, acc=0.771, loss=46.949, backward_time=0.305, grad_norm=44.831, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.499e-04, train_time=1.466
+[gpub010:0/16] 2024-02-12 03:54:10,482 (trainer:762) INFO: 45epoch:train:8201-8300batch: iter_time=8.386e-05, forward_time=0.298, loss_ctc=49.806, loss_att=41.642, acc=0.769, loss=44.091, backward_time=0.305, grad_norm=42.964, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.498e-04, train_time=1.336
+[gpub010:0/16] 2024-02-12 03:56:06,724 (trainer:762) INFO: 45epoch:train:8301-8400batch: iter_time=8.366e-05, forward_time=0.318, loss_ctc=43.266, loss_att=43.805, acc=0.752, loss=43.643, backward_time=0.333, grad_norm=44.671, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.498e-04, train_time=1.163
+[gpub010:0/16] 2024-02-12 03:58:33,088 (trainer:762) INFO: 45epoch:train:8401-8500batch: iter_time=4.734e-04, forward_time=0.355, loss_ctc=43.401, loss_att=45.480, acc=0.771, loss=44.856, backward_time=0.313, grad_norm=39.812, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.498e-04, train_time=1.463
+[gpub010:0/16] 2024-02-12 04:00:37,449 (trainer:762) INFO: 45epoch:train:8501-8600batch: iter_time=8.098e-05, forward_time=0.295, loss_ctc=39.562, loss_att=42.600, acc=0.759, loss=41.688, backward_time=0.301, grad_norm=41.459, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.498e-04, train_time=1.244
+[gpub010:0/16] 2024-02-12 04:02:58,316 (trainer:762) INFO: 45epoch:train:8601-8700batch: iter_time=1.590e-04, forward_time=0.421, loss_ctc=49.072, loss_att=45.664, acc=0.748, loss=46.686, backward_time=0.307, grad_norm=53.244, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.498e-04, train_time=1.408
+[gpub010:0/16] 2024-02-12 04:04:12,325 (multiple_iter_factory:32) INFO: Building 7th iter-factory...
+[gpub010:0/16] 2024-02-12 04:04:31,635 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 04:04:35,221 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.0", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.0", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.0", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.0", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba95476950>)
+[gpub010:0/16] 2024-02-12 04:04:35,221 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.0, 
+[gpub010:0/16] 2024-02-12 04:04:35,227 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 04:10:10,338 (trainer:762) INFO: 45epoch:train:8701-8800batch: iter_time=3.138, forward_time=0.294, loss_ctc=41.481, loss_att=45.336, acc=0.769, loss=44.179, backward_time=0.304, grad_norm=44.443, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.498e-04, train_time=4.321
+[gpub010:0/16] 2024-02-12 04:12:31,647 (trainer:762) INFO: 45epoch:train:8801-8900batch: iter_time=7.367e-05, forward_time=0.292, loss_ctc=43.974, loss_att=44.971, acc=0.769, loss=44.672, backward_time=0.296, grad_norm=42.296, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.498e-04, train_time=1.413
+[gpub010:0/16] 2024-02-12 04:14:32,794 (trainer:762) INFO: 45epoch:train:8901-9000batch: iter_time=2.431e-04, forward_time=0.386, loss_ctc=45.660, loss_att=39.165, acc=0.786, loss=41.113, backward_time=0.316, grad_norm=41.486, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.498e-04, train_time=1.210
+[gpub010:0/16] 2024-02-12 04:16:35,195 (trainer:762) INFO: 45epoch:train:9001-9100batch: iter_time=8.290e-05, forward_time=0.297, loss_ctc=42.757, loss_att=38.221, acc=0.786, loss=39.582, backward_time=0.300, grad_norm=42.274, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.498e-04, train_time=1.224
+[gpub010:0/16] 2024-02-12 04:19:01,487 (trainer:762) INFO: 45epoch:train:9101-9200batch: iter_time=8.178e-05, forward_time=0.321, loss_ctc=40.908, loss_att=41.421, acc=0.780, loss=41.267, backward_time=0.352, grad_norm=40.717, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.497e-04, train_time=1.464
+[gpub010:0/16] 2024-02-12 04:20:56,670 (trainer:762) INFO: 45epoch:train:9201-9300batch: iter_time=8.080e-05, forward_time=0.298, loss_ctc=46.506, loss_att=49.250, acc=0.787, loss=48.427, backward_time=0.304, grad_norm=38.892, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.497e-04, train_time=1.151
+[gpub010:0/16] 2024-02-12 04:23:15,804 (trainer:762) INFO: 45epoch:train:9301-9400batch: iter_time=1.398e-04, forward_time=0.329, loss_ctc=57.662, loss_att=47.835, acc=0.767, loss=50.783, backward_time=0.342, grad_norm=50.669, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.497e-04, train_time=1.391
+[gpub010:0/16] 2024-02-12 04:25:22,248 (trainer:762) INFO: 45epoch:train:9401-9500batch: iter_time=8.094e-05, forward_time=0.295, loss_ctc=57.482, loss_att=46.739, acc=0.779, loss=49.962, backward_time=0.300, grad_norm=48.031, clip=100.000, loss_scale=7.633e+33, optim_step_time=0.093, optim0_lr0=1.497e-04, train_time=1.264
+[gpub010:0/16] 2024-02-12 04:27:38,226 (trainer:762) INFO: 45epoch:train:9501-9600batch: iter_time=8.205e-05, forward_time=0.355, loss_ctc=44.597, loss_att=44.008, acc=0.756, loss=44.184, backward_time=0.351, grad_norm=42.529, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.497e-04, train_time=1.360
+[gpub010:0/16] 2024-02-12 04:29:42,494 (trainer:762) INFO: 45epoch:train:9601-9700batch: iter_time=8.247e-05, forward_time=0.289, loss_ctc=39.998, loss_att=38.536, acc=0.783, loss=38.975, backward_time=0.295, grad_norm=39.547, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.497e-04, train_time=1.242
+[gpub010:0/16] 2024-02-12 04:31:59,698 (trainer:762) INFO: 45epoch:train:9701-9800batch: iter_time=8.262e-05, forward_time=0.297, loss_ctc=41.710, loss_att=50.194, acc=0.762, loss=47.649, backward_time=0.305, grad_norm=42.940, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.497e-04, train_time=1.372
+[gpub010:0/16] 2024-02-12 04:34:02,628 (trainer:762) INFO: 45epoch:train:9801-9900batch: iter_time=2.855e-04, forward_time=0.395, loss_ctc=43.258, loss_att=39.635, acc=0.778, loss=40.722, backward_time=0.318, grad_norm=41.369, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.497e-04, train_time=1.229
+[gpub010:0/16] 2024-02-12 04:35:56,724 (trainer:762) INFO: 45epoch:train:9901-10000batch: iter_time=7.898e-05, forward_time=0.293, loss_ctc=46.100, loss_att=45.426, acc=0.751, loss=45.628, backward_time=0.298, grad_norm=60.826, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.497e-04, train_time=1.141
+[gpub010:0/16] 2024-02-12 04:36:16,794 (multiple_iter_factory:32) INFO: Building 8th iter-factory...
+[gpub010:0/16] 2024-02-12 04:36:36,442 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 04:36:39,976 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.3", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.3", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.3", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.3", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba41c48d90>)
+[gpub010:0/16] 2024-02-12 04:36:39,976 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.3, 
+[gpub010:0/16] 2024-02-12 04:36:39,981 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 04:43:29,780 (trainer:762) INFO: 45epoch:train:10001-10100batch: iter_time=3.182, forward_time=0.395, loss_ctc=44.188, loss_att=46.926, acc=0.757, loss=46.105, backward_time=0.315, grad_norm=43.552, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.496e-04, train_time=4.530
+[gpub010:0/16] 2024-02-12 04:45:22,590 (trainer:762) INFO: 45epoch:train:10101-10200batch: iter_time=8.230e-05, forward_time=0.290, loss_ctc=41.208, loss_att=38.961, acc=0.776, loss=39.635, backward_time=0.295, grad_norm=41.383, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.496e-04, train_time=1.128
+[gpub010:0/16] 2024-02-12 04:47:42,921 (trainer:762) INFO: 45epoch:train:10201-10300batch: iter_time=8.219e-05, forward_time=0.299, loss_ctc=48.102, loss_att=45.110, acc=0.756, loss=46.008, backward_time=0.304, grad_norm=42.688, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.095, optim0_lr0=1.496e-04, train_time=1.403
+[gpub010:0/16] 2024-02-12 04:49:50,132 (trainer:762) INFO: 45epoch:train:10301-10400batch: iter_time=8.410e-05, forward_time=0.366, loss_ctc=35.840, loss_att=32.405, acc=0.790, loss=33.436, backward_time=0.302, grad_norm=38.203, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.127, optim0_lr0=1.496e-04, train_time=1.272
+[gpub010:0/16] 2024-02-12 04:51:46,902 (trainer:762) INFO: 45epoch:train:10401-10500batch: iter_time=8.295e-05, forward_time=0.293, loss_ctc=46.768, loss_att=46.340, acc=0.787, loss=46.468, backward_time=0.299, grad_norm=41.969, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.496e-04, train_time=1.168
+[gpub010:0/16] 2024-02-12 04:54:16,668 (trainer:762) INFO: 45epoch:train:10501-10600batch: iter_time=8.207e-05, forward_time=0.413, loss_ctc=58.667, loss_att=50.090, acc=0.763, loss=52.663, backward_time=0.326, grad_norm=49.831, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.097, optim0_lr0=1.496e-04, train_time=1.497
+[gpub010:0/16] 2024-02-12 04:56:15,404 (trainer:762) INFO: 45epoch:train:10601-10700batch: iter_time=7.990e-05, forward_time=0.293, loss_ctc=50.221, loss_att=45.508, acc=0.771, loss=46.922, backward_time=0.298, grad_norm=44.310, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.496e-04, train_time=1.187
+[gpub010:0/16] 2024-02-12 04:58:22,656 (trainer:762) INFO: 45epoch:train:10701-10800batch: iter_time=7.994e-05, forward_time=0.292, loss_ctc=49.347, loss_att=41.526, acc=0.769, loss=43.872, backward_time=0.297, grad_norm=44.498, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.496e-04, train_time=1.272
+[gpub010:0/16] 2024-02-12 05:00:39,312 (trainer:762) INFO: 45epoch:train:10801-10900batch: iter_time=8.086e-05, forward_time=0.373, loss_ctc=42.852, loss_att=42.567, acc=0.757, loss=42.652, backward_time=0.313, grad_norm=42.444, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.104, optim0_lr0=1.496e-04, train_time=1.366
+[gpub010:0/16] 2024-02-12 05:02:39,259 (trainer:762) INFO: 45epoch:train:10901-11000batch: iter_time=8.017e-05, forward_time=0.291, loss_ctc=43.513, loss_att=45.559, acc=0.771, loss=44.945, backward_time=0.297, grad_norm=41.297, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.495e-04, train_time=1.200
+[gpub010:0/16] 2024-02-12 05:05:11,880 (trainer:762) INFO: 45epoch:train:11001-11100batch: iter_time=8.317e-05, forward_time=0.388, loss_ctc=39.618, loss_att=42.671, acc=0.759, loss=41.755, backward_time=0.309, grad_norm=40.005, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.121, optim0_lr0=1.495e-04, train_time=1.526
+[gpub010:0/16] 2024-02-12 05:07:07,432 (trainer:762) INFO: 45epoch:train:11101-11200batch: iter_time=8.189e-05, forward_time=0.293, loss_ctc=48.708, loss_att=45.579, acc=0.746, loss=46.518, backward_time=0.298, grad_norm=51.810, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.495e-04, train_time=1.155
+[gpub010:0/16] 2024-02-12 05:08:23,668 (multiple_iter_factory:32) INFO: Building 9th iter-factory...
+[gpub010:0/16] 2024-02-12 05:08:42,935 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 05:08:46,543 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.10", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.10", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.10", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.10", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbd6b023790>)
+[gpub010:0/16] 2024-02-12 05:08:46,543 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.10, 
+[gpub010:0/16] 2024-02-12 05:08:46,592 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 05:14:22,409 (trainer:762) INFO: 45epoch:train:11201-11300batch: iter_time=3.097, forward_time=0.291, loss_ctc=41.152, loss_att=44.999, acc=0.773, loss=43.845, backward_time=0.330, grad_norm=43.553, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.098, optim0_lr0=1.495e-04, train_time=4.350
+[gpub010:0/16] 2024-02-12 05:16:17,945 (trainer:762) INFO: 45epoch:train:11301-11400batch: iter_time=7.961e-05, forward_time=0.342, loss_ctc=43.688, loss_att=44.400, acc=0.772, loss=44.186, backward_time=0.303, grad_norm=40.601, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.495e-04, train_time=1.155
+[gpub010:0/16] 2024-02-12 05:18:26,178 (trainer:762) INFO: 45epoch:train:11401-11500batch: iter_time=7.933e-05, forward_time=0.292, loss_ctc=45.695, loss_att=39.715, acc=0.784, loss=41.509, backward_time=0.296, grad_norm=43.768, clip=100.000, loss_scale=1.527e+34, optim_step_time=0.093, optim0_lr0=1.495e-04, train_time=1.282
+[gpub010:0/16] 2024-02-12 05:19:07,259 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 05:20:50,259 (trainer:762) INFO: 45epoch:train:11501-11600batch: iter_time=8.375e-05, forward_time=0.395, loss_ctc=42.429, loss_att=37.675, acc=0.788, loss=39.101, backward_time=0.334, grad_norm=38.844, clip=100.000, loss_scale=1.227e+34, optim_step_time=0.098, optim0_lr0=1.495e-04, train_time=1.441
+[gpub010:0/16] 2024-02-12 05:22:47,525 (trainer:762) INFO: 45epoch:train:11601-11700batch: iter_time=8.243e-05, forward_time=0.291, loss_ctc=41.122, loss_att=41.647, acc=0.779, loss=41.489, backward_time=0.296, grad_norm=40.200, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.495e-04, train_time=1.173
+[gpub010:0/16] 2024-02-12 05:25:21,380 (trainer:762) INFO: 45epoch:train:11701-11800batch: iter_time=8.552e-05, forward_time=0.331, loss_ctc=46.821, loss_att=48.236, acc=0.788, loss=47.811, backward_time=0.308, grad_norm=40.173, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.106, optim0_lr0=1.495e-04, train_time=1.538
+[gpub010:0/16] 2024-02-12 05:27:27,972 (trainer:762) INFO: 45epoch:train:11801-11900batch: iter_time=8.550e-05, forward_time=0.340, loss_ctc=57.414, loss_att=47.912, acc=0.768, loss=50.762, backward_time=0.308, grad_norm=47.521, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.094, optim0_lr0=1.494e-04, train_time=1.266
+[gpub010:0/16] 2024-02-12 05:29:37,212 (trainer:762) INFO: 45epoch:train:11901-12000batch: iter_time=8.168e-05, forward_time=0.295, loss_ctc=57.579, loss_att=47.109, acc=0.777, loss=50.250, backward_time=0.301, grad_norm=47.345, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.494e-04, train_time=1.292
+[gpub010:0/16] 2024-02-12 05:32:01,793 (trainer:762) INFO: 45epoch:train:12001-12100batch: iter_time=9.433e-05, forward_time=0.403, loss_ctc=44.864, loss_att=44.261, acc=0.754, loss=44.441, backward_time=0.325, grad_norm=43.116, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.494e-04, train_time=1.446
+[gpub010:0/16] 2024-02-12 05:33:53,178 (trainer:762) INFO: 45epoch:train:12101-12200batch: iter_time=8.586e-05, forward_time=0.291, loss_ctc=39.958, loss_att=38.934, acc=0.783, loss=39.241, backward_time=0.297, grad_norm=38.948, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.494e-04, train_time=1.113
+[gpub010:0/16] 2024-02-12 05:36:09,378 (trainer:762) INFO: 45epoch:train:12201-12300batch: iter_time=8.327e-05, forward_time=0.292, loss_ctc=41.876, loss_att=50.556, acc=0.762, loss=47.952, backward_time=0.297, grad_norm=42.614, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.093, optim0_lr0=1.494e-04, train_time=1.362
+[gpub010:0/16] 2024-02-12 05:38:29,648 (trainer:762) INFO: 45epoch:train:12301-12400batch: iter_time=8.122e-05, forward_time=0.371, loss_ctc=42.631, loss_att=40.140, acc=0.775, loss=40.888, backward_time=0.325, grad_norm=40.617, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.100, optim0_lr0=1.494e-04, train_time=1.402
+[gpub010:0/16] 2024-02-12 05:38:46,194 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 05:40:30,875 (trainer:762) INFO: 45epoch:train:12401-12500batch: iter_time=8.061e-05, forward_time=0.292, loss_ctc=46.425, loss_att=45.914, acc=0.754, loss=46.067, backward_time=0.297, grad_norm=54.397, clip=100.000, loss_scale=5.769e+33, optim_step_time=0.093, optim0_lr0=1.494e-04, train_time=1.212
+[gpub010:0/16] 2024-02-12 05:40:50,904 (multiple_iter_factory:32) INFO: Building 10th iter-factory...
+[gpub010:0/16] 2024-02-12 05:41:10,481 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 05:41:14,090 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.11", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.11", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.11", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.11", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fbacfe52dd0>)
+[gpub010:0/16] 2024-02-12 05:41:14,090 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.11, 
+[gpub010:0/16] 2024-02-12 05:41:14,093 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 05:47:51,983 (trainer:762) INFO: 45epoch:train:12501-12600batch: iter_time=3.123, forward_time=0.348, loss_ctc=43.642, loss_att=46.387, acc=0.758, loss=45.563, backward_time=0.312, grad_norm=43.506, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.494e-04, train_time=4.411
+[gpub010:0/16] 2024-02-12 05:50:01,581 (trainer:762) INFO: 45epoch:train:12601-12700batch: iter_time=8.391e-05, forward_time=0.334, loss_ctc=41.529, loss_att=38.979, acc=0.776, loss=39.744, backward_time=0.299, grad_norm=42.564, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.494e-04, train_time=1.296
+[gpub010:0/16] 2024-02-12 05:52:15,918 (trainer:762) INFO: 45epoch:train:12701-12800batch: iter_time=8.069e-05, forward_time=0.292, loss_ctc=47.600, loss_att=44.513, acc=0.759, loss=45.439, backward_time=0.297, grad_norm=43.597, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.493e-04, train_time=1.343
+[gpub010:0/16] 2024-02-12 05:54:19,321 (trainer:762) INFO: 45epoch:train:12801-12900batch: iter_time=8.528e-05, forward_time=0.330, loss_ctc=35.876, loss_att=32.332, acc=0.790, loss=33.395, backward_time=0.303, grad_norm=36.259, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.102, optim0_lr0=1.493e-04, train_time=1.234
+[gpub010:0/16] 2024-02-12 05:56:30,847 (trainer:762) INFO: 45epoch:train:12901-13000batch: iter_time=3.633e-04, forward_time=0.338, loss_ctc=46.780, loss_att=46.185, acc=0.787, loss=46.363, backward_time=0.304, grad_norm=40.795, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.493e-04, train_time=1.315
+[gpub010:0/16] 2024-02-12 05:58:49,987 (trainer:762) INFO: 45epoch:train:13001-13100batch: iter_time=8.637e-05, forward_time=0.347, loss_ctc=58.710, loss_att=49.968, acc=0.764, loss=52.591, backward_time=0.312, grad_norm=47.039, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.493e-04, train_time=1.391
+[gpub010:0/16] 2024-02-12 06:01:00,452 (trainer:762) INFO: 45epoch:train:13101-13200batch: iter_time=8.214e-05, forward_time=0.350, loss_ctc=50.462, loss_att=45.540, acc=0.771, loss=47.016, backward_time=0.308, grad_norm=45.662, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.493e-04, train_time=1.305
+[gpub010:0/16] 2024-02-12 06:03:19,804 (trainer:762) INFO: 45epoch:train:13201-13300batch: iter_time=8.480e-05, forward_time=0.291, loss_ctc=50.045, loss_att=41.988, acc=0.768, loss=44.405, backward_time=0.297, grad_norm=44.962, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.493e-04, train_time=1.393
+[gpub010:0/16] 2024-02-12 06:05:32,838 (trainer:762) INFO: 45epoch:train:13301-13400batch: iter_time=8.491e-05, forward_time=0.351, loss_ctc=43.071, loss_att=42.682, acc=0.758, loss=42.799, backward_time=0.309, grad_norm=43.911, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.493e-04, train_time=1.330
+[gpub010:0/16] 2024-02-12 06:07:43,012 (trainer:762) INFO: 45epoch:train:13401-13500batch: iter_time=2.334e-04, forward_time=0.344, loss_ctc=43.105, loss_att=45.103, acc=0.773, loss=44.504, backward_time=0.302, grad_norm=38.807, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.493e-04, train_time=1.302
+[gpub010:0/16] 2024-02-12 06:10:03,286 (trainer:762) INFO: 45epoch:train:13501-13600batch: iter_time=8.416e-05, forward_time=0.352, loss_ctc=38.843, loss_att=42.438, acc=0.762, loss=41.359, backward_time=0.301, grad_norm=40.501, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.493e-04, train_time=1.402
+[gpub010:0/16] 2024-02-12 06:12:01,094 (trainer:762) INFO: 45epoch:train:13601-13700batch: iter_time=1.550e-04, forward_time=0.375, loss_ctc=47.941, loss_att=44.988, acc=0.747, loss=45.874, backward_time=0.304, grad_norm=51.786, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.492e-04, train_time=1.178
+[gpub010:0/16] 2024-02-12 06:13:32,068 (multiple_iter_factory:32) INFO: Building 11th iter-factory...
+[gpub010:0/16] 2024-02-12 06:13:51,489 (s2t:445) INFO: Optional Data Names: ('text_prev', 'text_ctc', 'text_spk2', 'text_spk3', 'text_spk4')
+[gpub010:0/16] 2024-02-12 06:13:54,973 (abs_task:1663) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "exp/s2t_stats_raw_bpe50000/splits12/wav.scp/split.6", "type": "kaldi_ark"}
+  text_prev: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.prev/split.6", "type": "text"}
+  text_ctc: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text.ctc/split.6", "type": "text"}
+  text: {"path": "exp/s2t_stats_raw_bpe50000/splits12/text/split.6", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.S2TPreprocessor object at 0x7fba98c8bfa0>)
+[gpub010:0/16] 2024-02-12 06:13:54,973 (abs_task:1664) INFO: [train] Batch sampler: UnsortedBatchSampler(N-batch=19027, batch_size=256, key_file=exp/s2t_stats_raw_bpe50000/splits12/speech_shape/split.6, 
+[gpub010:0/16] 2024-02-12 06:13:55,029 (abs_task:1665) INFO: [train] mini-batch sizes summary: N-batch=19027, mean=256.0, min=256, max=257
+[gpub010:0/16] 2024-02-12 06:19:25,843 (trainer:762) INFO: 45epoch:train:13701-13800batch: iter_time=3.055, forward_time=0.324, loss_ctc=41.222, loss_att=45.338, acc=0.772, loss=44.103, backward_time=0.301, grad_norm=45.596, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.492e-04, train_time=4.447
+[gpub010:0/16] 2024-02-12 06:21:19,568 (trainer:762) INFO: 45epoch:train:13801-13900batch: iter_time=7.946e-05, forward_time=0.342, loss_ctc=43.645, loss_att=44.179, acc=0.773, loss=44.019, backward_time=0.303, grad_norm=40.311, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.094, optim0_lr0=1.492e-04, train_time=1.137
+[gpub010:0/16] 2024-02-12 06:23:41,287 (trainer:762) INFO: 45epoch:train:13901-14000batch: iter_time=8.008e-05, forward_time=0.319, loss_ctc=45.620, loss_att=39.810, acc=0.783, loss=41.553, backward_time=0.322, grad_norm=41.655, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.492e-04, train_time=1.417
+[gpub010:0/16] 2024-02-12 06:25:35,826 (trainer:762) INFO: 45epoch:train:14001-14100batch: iter_time=8.331e-05, forward_time=0.292, loss_ctc=42.676, loss_att=38.137, acc=0.788, loss=39.499, backward_time=0.298, grad_norm=39.276, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.492e-04, train_time=1.145
+[gpub010:0/16] 2024-02-12 06:27:28,874 (trainer:762) INFO: 45epoch:train:14101-14200batch: iter_time=8.667e-05, forward_time=0.320, loss_ctc=40.883, loss_att=41.290, acc=0.781, loss=41.168, backward_time=0.319, grad_norm=39.384, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.095, optim0_lr0=1.492e-04, train_time=1.130
+[gpub010:0/16] 2024-02-12 06:30:09,375 (trainer:762) INFO: 45epoch:train:14201-14300batch: iter_time=9.141e-05, forward_time=0.349, loss_ctc=46.300, loss_att=48.069, acc=0.789, loss=47.538, backward_time=0.318, grad_norm=40.512, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.492e-04, train_time=1.605
+[gpub010:0/16] 2024-02-12 06:32:03,860 (trainer:762) INFO: 45epoch:train:14301-14400batch: iter_time=8.755e-05, forward_time=0.344, loss_ctc=57.358, loss_att=47.813, acc=0.768, loss=50.677, backward_time=0.308, grad_norm=51.025, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.492e-04, train_time=1.145
+[gpub010:0/16] 2024-02-12 06:34:21,430 (trainer:762) INFO: 45epoch:train:14401-14500batch: iter_time=8.070e-05, forward_time=0.296, loss_ctc=57.971, loss_att=47.067, acc=0.776, loss=50.338, backward_time=0.301, grad_norm=49.884, clip=100.000, loss_scale=9.762e+33, optim_step_time=0.093, optim0_lr0=1.492e-04, train_time=1.375
+[gpub010:0/16] 2024-02-12 06:36:32,335 (trainer:762) INFO: 45epoch:train:14501-14600batch: iter_time=8.012e-05, forward_time=0.327, loss_ctc=44.461, loss_att=44.121, acc=0.757, loss=44.223, backward_time=0.320, grad_norm=45.405, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.096, optim0_lr0=1.491e-04, train_time=1.309
+[gpub010:0/16] 2024-02-12 06:36:42,092 (trainer:693) WARNING: The grad norm is nan. Skipping updating the model.
+[gpub010:0/16] 2024-02-12 06:38:29,059 (trainer:762) INFO: 45epoch:train:14601-14700batch: iter_time=2.559e-04, forward_time=0.350, loss_ctc=39.460, loss_att=38.487, acc=0.785, loss=38.779, backward_time=0.303, grad_norm=38.826, clip=100.000, loss_scale=5.612e+33, optim_step_time=0.093, optim0_lr0=1.491e-04, train_time=1.168
+[gpub010:0/16] 2024-02-12 06:40:41,551 (trainer:762) INFO: 45epoch:train:14701-14800batch: iter_time=8.601e-05, forward_time=0.291, loss_ctc=41.600, loss_att=50.541, acc=0.763, loss=47.859, backward_time=0.296, grad_norm=41.954, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.093, optim0_lr0=1.491e-04, train_time=1.325
+[gpub010:0/16] 2024-02-12 06:43:00,382 (trainer:762) INFO: 45epoch:train:14801-14900batch: iter_time=8.265e-05, forward_time=0.333, loss_ctc=42.811, loss_att=40.067, acc=0.775, loss=40.890, backward_time=0.322, grad_norm=42.227, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.097, optim0_lr0=1.491e-04, train_time=1.388
+[gpub010:0/16] 2024-02-12 06:45:08,079 (trainer:762) INFO: 45epoch:train:14901-15000batch: iter_time=8.281e-05, forward_time=0.341, loss_ctc=45.669, loss_att=45.008, acc=0.757, loss=45.207, backward_time=0.308, grad_norm=52.642, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.096, optim0_lr0=1.491e-04, train_time=1.277
+[gpub010:0/16] 2024-02-12 07:22:53,575 (trainer:361) INFO: 45epoch results: [train] iter_time=0.249, forward_time=0.326, loss_ctc=46.133, loss_att=43.984, acc=0.769, loss=44.629, backward_time=0.309, grad_norm=44.287, clip=100.000, loss_scale=7.252e+33, optim_step_time=0.096, optim0_lr0=1.499e-04, train_time=1.552, time=6 hours, 28 minutes and 31.74 seconds, total_count=705000, gpu_max_cached_mem_GB=42.092, [valid] loss_ctc=33.517, cer_ctc=0.172, loss_att=37.876, acc=0.692, cer=0.310, wer=0.991, loss=36.569, time=37 minutes and 19.84 seconds, total_count=219537, gpu_max_cached_mem_GB=42.092
+gpub037:2479731:2479788 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+gpub037:2479732:2479785 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+gpub037:2479733:2479786 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+gpub076:3268962:3269017 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+gpub037:2479733:2479733 [2] NCCL INFO comm 0x55c8aa09ba70 rank 6 nranks 16 cudaDev 2 busId 85000 - Abort COMPLETE
+gpub076:3268963:3269016 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+gpub076:3268960:3269018 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+gpub037:2479734:2479787 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+gpub010:2415461:2415514 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+gpub010:2415460:2415516 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+gpub076:3268962:3268962 [2] NCCL INFO comm 0x561f7f19bb70 rank 10 nranks 16 cudaDev 2 busId 85000 - Abort COMPLETE
+gpub076:3268961:3269019 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+gpub010:2415460:2415460 [2] NCCL INFO comm 0x560007013060 rank 2 nranks 16 cudaDev 2 busId 85000 - Abort COMPLETE
+gpub010:2415459:2415517 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+gpub010:2415459:2415459 [1] NCCL INFO comm 0x5576d44db4b0 rank 1 nranks 16 cudaDev 1 busId 46000 - Abort COMPLETE
+gpub082:3674493:3674551 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+gpub082:3674496:3674549 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+gpub082:3674494:3674550 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+gpub082:3674495:3674552 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+gpub082:3674494:3674494 [1] NCCL INFO comm 0x5605a17f6130 rank 13 nranks 16 cudaDev 1 busId 46000 - Abort COMPLETE
+gpub082:3674495:3674495 [2] NCCL INFO comm 0x556db2db58a0 rank 14 nranks 16 cudaDev 2 busId 85000 - Abort COMPLETE
+gpub076:3268963:3268963 [3] NCCL INFO comm 0x56218608af50 rank 11 nranks 16 cudaDev 3 busId c7000 - Abort COMPLETE
+gpub076:3268961:3268961 [1] NCCL INFO comm 0x558968564390 rank 9 nranks 16 cudaDev 1 busId 46000 - Abort COMPLETE
+gpub076:3268960:3268960 [0] NCCL INFO comm 0x55f0b7620d20 rank 8 nranks 16 cudaDev 0 busId 7000 - Abort COMPLETE
+gpub010:2415461:2415461 [3] NCCL INFO comm 0x56066bfecd40 rank 3 nranks 16 cudaDev 3 busId c7000 - Abort COMPLETE
+gpub082:3674496:3674496 [3] NCCL INFO comm 0x55cb4fbfa630 rank 15 nranks 16 cudaDev 3 busId c7000 - Abort COMPLETE
+gpub037:2479731:2479731 [0] NCCL INFO comm 0x556289adbe40 rank 4 nranks 16 cudaDev 0 busId 7000 - Abort COMPLETE
+gpub037:2479734:2479734 [3] NCCL INFO comm 0x560fe900a3e0 rank 7 nranks 16 cudaDev 3 busId c7000 - Abort COMPLETE
+gpub082:3674493:3674493 [0] NCCL INFO comm 0x5595c37939f0 rank 12 nranks 16 cudaDev 0 busId 7000 - Abort COMPLETE
+gpub037:2479732:2479732 [1] NCCL INFO comm 0x5641d0bbff50 rank 5 nranks 16 cudaDev 1 busId 46000 - Abort COMPLETE
+[gpub010:0/16] 2024-02-12 07:23:02,819 (trainer:416) INFO: The best model has been updated: valid.total_count
+[gpub010:0/16] 2024-02-12 07:23:02,932 (trainer:470) INFO: The model files were removed: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/40epoch.pth
+[gpub010:0/16] 2024-02-12 07:23:02,932 (trainer:488) INFO: The training was finished at 45 epochs 
+[gpub010:0/16] 2024-02-12 07:23:02,959 (average_nbest_models:69) INFO: Averaging 5best models: criterion="valid.acc": exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.acc.ave_5best.pth
+[gpub010:0/16] 2024-02-12 07:23:21,911 (average_nbest_models:69) INFO: Averaging 5best models: criterion="valid.total_count": exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.pth
+gpub010:2415458:2415515 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+gpub010:2415458:2415458 [0] NCCL INFO comm 0x5627bbb736f0 rank 0 nranks 16 cudaDev 0 busId 7000 - Abort COMPLETE
+# Accounting: begin_time=1707587278
+# Accounting: end_time=1707744214
+# Accounting: time=156936 threads=1
+# Finished at Mon Feb 12 07:23:34 CST 2024 with status 0
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.acc.ave_5best.pth b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.acc.ave_5best.pth
new file mode 100644
index 0000000000000000000000000000000000000000..85f667630bbd7a8c101ee571d31f89335fa720f9
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.acc.ave_5best.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f15bc3dde14fa43e480e58e4f563bb6e1463bfd1f5960f1135ab21e007fbb11
+size 1466924749
diff --git a/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.pth b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.pth
new file mode 100644
index 0000000000000000000000000000000000000000..81580fa6f8ea714ffb8b6968023efc51ce30f991
--- /dev/null
+++ b/exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d51c19d07c251b1bfdea44a3c84a0d1a189a8e64a7bf8b364c955806ecb0247c
+size 1466929645
diff --git a/meta.yaml b/meta.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..f906ae310ec32d6d4a990b2b665e8807a5dccead
--- /dev/null
+++ b/meta.yaml
@@ -0,0 +1,8 @@
+espnet: '202310'
+files:
+  s2t_model_file: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/valid.acc.ave_5best.pth
+python: 3.10.13 | packaged by conda-forge | (main, Dec 23 2023, 15:36:39) [GCC 12.3.0]
+timestamp: 1725245981.515642
+torch: 1.13.1
+yaml_files:
+  s2t_train_config: exp/s2t_train_s2t_ebf_conv2d_size768_e9_d9_piecewise_lr5e-4_warmup60k_flashattn_raw_bpe50000/config.yaml