diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000000000000000000000000000000000000..c6a993a46f2636611e27f536fafc3aa81312bada --- /dev/null +++ b/all_results.json @@ -0,0 +1,14 @@ +{ + "epoch": 150.0, + "eval_loss": 0.4378179907798767, + "eval_runtime": 1.9751, + "eval_samples": 270, + "eval_samples_per_second": 136.702, + "eval_steps_per_second": 4.557, + "eval_wer": 0.16863370547581075, + "train_loss": 1.9741196327647943, + "train_runtime": 1808.1454, + "train_samples": 905, + "train_samples_per_second": 75.077, + "train_steps_per_second": 2.406 +} \ No newline at end of file diff --git a/checkpoint-3000/config.json b/checkpoint-3000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..966cd6c5b3d8cbae43fe6586aa218930bd060f97 --- /dev/null +++ b/checkpoint-3000/config.json @@ -0,0 +1,107 @@ +{ + "_name_or_path": "./", + "activation_dropout": 0.1, + "adapter_kernel_size": 3, + "adapter_stride": 2, + "add_adapter": false, + "apply_spec_augment": true, + "architectures": [ + "Wav2Vec2ForCTC" + ], + "attention_dropout": 0.0, + "bos_token_id": 1, + "classifier_proj_size": 256, + "codevector_dim": 768, + "contrastive_logits_temperature": 0.1, + "conv_bias": true, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "mean", + "ctc_zero_infinity": false, + "diversity_loss_weight": 0.1, + "do_stable_layer_norm": true, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_dropout": 0.0, + "feat_extract_norm": "layer", + "feat_proj_dropout": 0.0, + "feat_quantizer_dropout": 0.0, + "final_dropout": 0.0, + "hidden_act": "gelu", + "hidden_dropout": 0.0, + "hidden_size": 1024, + "initializer_range": 0.02, + "intermediate_size": 4096, + "layer_norm_eps": 1e-05, + "layerdrop": 0.0, + "mask_feature_length": 64, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.25, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.75, + "model_type": "wav2vec2", + "num_adapter_layers": 3, + "num_attention_heads": 16, + "num_codevector_groups": 2, + "num_codevectors_per_group": 320, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 24, + "num_negatives": 100, + "output_hidden_size": 1024, + "pad_token_id": 48, + "proj_codevector_dim": 768, + "tdnn_dilation": [ + 1, + 2, + 3, + 1, + 1 + ], + "tdnn_dim": [ + 512, + 512, + 512, + 512, + 1500 + ], + "tdnn_kernel": [ + 5, + 3, + 3, + 1, + 1 + ], + "torch_dtype": "float32", + "transformers_version": "4.15.0.dev0", + "use_weighted_layer_sum": false, + "vocab_size": 52, + "xvector_output_dim": 512 +} diff --git a/checkpoint-3000/optimizer.pt b/checkpoint-3000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..aff285364c0c4a861db284beb1336b3553581d51 --- /dev/null +++ b/checkpoint-3000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe99480dea945a1a380d12be09b6856fda798f10a34f11f503aab627a2fd9e54 +size 2490485393 diff --git a/checkpoint-3000/preprocessor_config.json b/checkpoint-3000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..36ebe8b7c1cc967b3059f0494ae8a1069dd67655 --- /dev/null +++ b/checkpoint-3000/preprocessor_config.json @@ -0,0 +1,9 @@ +{ + "do_normalize": true, + "feature_extractor_type": "Wav2Vec2FeatureExtractor", + "feature_size": 1, + "padding_side": "right", + "padding_value": 0, + "return_attention_mask": true, + "sampling_rate": 16000 +} diff --git a/checkpoint-3000/pytorch_model.bin b/checkpoint-3000/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..d66d1eaf09cefaddf46d679f84f18f756c10ae69 --- /dev/null +++ b/checkpoint-3000/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07d839729b1094bfc1dad9cf44e7a3baf0db79456bcca4a1c0e4c452c21bebc3 +size 1262136881 diff --git a/checkpoint-3000/rng_state_0.pth b/checkpoint-3000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..42f8a8542ae5b05cfede267294daf14ee70cabca --- /dev/null +++ b/checkpoint-3000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e48eaa59b156324d3c752bac6c979d12f1a4b38fc7d4e4f56ae9b260e174b2b5 +size 14567 diff --git a/checkpoint-3000/rng_state_1.pth b/checkpoint-3000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..7c1165804cca38bb26eb040d67dde857f3109a44 --- /dev/null +++ b/checkpoint-3000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f1ae93e2dd3cdba783c24773712aae4ddc605a264e2fa159af37ccaf70c21e2 +size 14503 diff --git a/checkpoint-3000/rng_state_2.pth b/checkpoint-3000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..1a4710976a244bf83a3c0757c7fb7f8a5b723afb --- /dev/null +++ b/checkpoint-3000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44cb9a419c137a795c59ada5bf76b4ae022572edd9068af0610737d45ab2090f +size 14567 diff --git a/checkpoint-3000/rng_state_3.pth b/checkpoint-3000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..2016612239db0d3d688fa8c7fad3cccfa131dde1 --- /dev/null +++ b/checkpoint-3000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05a9d8080a2008b1dfce194265d322150b21b366293a3020c93e0b227cb74efe +size 14503 diff --git a/checkpoint-3000/rng_state_4.pth b/checkpoint-3000/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..351b3683fb4d28f9ae2d28571cd4b332b874060d --- /dev/null +++ b/checkpoint-3000/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2eb127c2e2bd207cdfe9c6d2ee0ac5ae0e8fc5abb58b9c0391004b938a5aaafd +size 14567 diff --git a/checkpoint-3000/rng_state_5.pth b/checkpoint-3000/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..659995d3f9b464952e195c0dd39183ef9e8a3f94 --- /dev/null +++ b/checkpoint-3000/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b5a4fa54612b3e0e99132c4bbf6e5800eb2bc1e83300840263ff1aad4a17795 +size 14567 diff --git a/checkpoint-3000/rng_state_6.pth b/checkpoint-3000/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..91f1d1cac433a3cb4e75c8de9cd3f26ed1d7d896 --- /dev/null +++ b/checkpoint-3000/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ccba11918942a57e0c0cbc717f7f6cadbfd906edc19bc89af674b132ce7de846 +size 14567 diff --git a/checkpoint-3000/rng_state_7.pth b/checkpoint-3000/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..3cbad353f511995c41c7cae2d2b79ded4e6a2dcc --- /dev/null +++ b/checkpoint-3000/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76e0a7ce570523f52846def5a4e8fd2a312bdcc9d099b503d3eabc27e670477b +size 14631 diff --git a/checkpoint-3000/scaler.pt b/checkpoint-3000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..7bdc39056e3bcf2401ff4fc2fd663f76e7aa2c50 --- /dev/null +++ b/checkpoint-3000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e282386c88b92c8801eadb3ac3e2e1c684c2653af5b93ca015f68731ca894c0 +size 559 diff --git a/checkpoint-3000/scheduler.pt b/checkpoint-3000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..13ce3ce6e0f18beb2a5755db0cd680062055e4eb --- /dev/null +++ b/checkpoint-3000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9662679f467de6dc69408f77a06738ea4f16c091cbe75449af8e7685b8b8ab08 +size 623 diff --git a/checkpoint-3000/trainer_state.json b/checkpoint-3000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..2fc0a20e8a5cfb19ed9f27de85dc4d99379c63b6 --- /dev/null +++ b/checkpoint-3000/trainer_state.json @@ -0,0 +1,250 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 103.44827586206897, + "global_step": 3000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 3.45, + "learning_rate": 7.2e-06, + "loss": 16.2031, + "step": 100 + }, + { + "epoch": 6.9, + "learning_rate": 1.47e-05, + "loss": 6.6774, + "step": 200 + }, + { + "epoch": 10.34, + "learning_rate": 2.2199999999999998e-05, + "loss": 4.7603, + "step": 300 + }, + { + "epoch": 13.79, + "learning_rate": 2.97e-05, + "loss": 3.9198, + "step": 400 + }, + { + "epoch": 17.24, + "learning_rate": 3.7199999999999996e-05, + "loss": 3.5558, + "step": 500 + }, + { + "epoch": 17.24, + "eval_loss": 3.5216922760009766, + "eval_runtime": 2.4797, + "eval_samples_per_second": 108.882, + "eval_steps_per_second": 3.629, + "eval_wer": 0.9812865497076023, + "step": 500 + }, + { + "epoch": 20.69, + "learning_rate": 4.4699999999999996e-05, + "loss": 3.4799, + "step": 600 + }, + { + "epoch": 24.14, + "learning_rate": 5.2199999999999995e-05, + "loss": 3.4271, + "step": 700 + }, + { + "epoch": 27.59, + "learning_rate": 5.97e-05, + "loss": 3.2779, + "step": 800 + }, + { + "epoch": 31.03, + "learning_rate": 6.72e-05, + "loss": 2.9317, + "step": 900 + }, + { + "epoch": 34.48, + "learning_rate": 7.47e-05, + "loss": 2.2202, + "step": 1000 + }, + { + "epoch": 34.48, + "eval_loss": 0.9991578459739685, + "eval_runtime": 1.9665, + "eval_samples_per_second": 137.299, + "eval_steps_per_second": 4.577, + "eval_wer": 0.2637958532695375, + "step": 1000 + }, + { + "epoch": 37.93, + "learning_rate": 7.285074626865672e-05, + "loss": 1.835, + "step": 1100 + }, + { + "epoch": 41.38, + "learning_rate": 7.061194029850745e-05, + "loss": 1.653, + "step": 1200 + }, + { + "epoch": 44.83, + "learning_rate": 6.83731343283582e-05, + "loss": 1.5243, + "step": 1300 + }, + { + "epoch": 48.28, + "learning_rate": 6.613432835820895e-05, + "loss": 1.4343, + "step": 1400 + }, + { + "epoch": 51.72, + "learning_rate": 6.38955223880597e-05, + "loss": 1.376, + "step": 1500 + }, + { + "epoch": 51.72, + "eval_loss": 0.5396047830581665, + "eval_runtime": 1.8956, + "eval_samples_per_second": 142.435, + "eval_steps_per_second": 4.748, + "eval_wer": 0.1946836788942052, + "step": 1500 + }, + { + "epoch": 55.17, + "learning_rate": 6.165671641791044e-05, + "loss": 1.3168, + "step": 1600 + }, + { + "epoch": 58.62, + "learning_rate": 5.941791044776119e-05, + "loss": 1.2773, + "step": 1700 + }, + { + "epoch": 62.07, + "learning_rate": 5.717910447761193e-05, + "loss": 1.2455, + "step": 1800 + }, + { + "epoch": 65.52, + "learning_rate": 5.494029850746268e-05, + "loss": 1.1881, + "step": 1900 + }, + { + "epoch": 68.97, + "learning_rate": 5.272388059701492e-05, + "loss": 1.1565, + "step": 2000 + }, + { + "epoch": 68.97, + "eval_loss": 0.4707144498825073, + "eval_runtime": 1.8963, + "eval_samples_per_second": 142.381, + "eval_steps_per_second": 4.746, + "eval_wer": 0.1773524720893142, + "step": 2000 + }, + { + "epoch": 72.41, + "learning_rate": 5.048507462686567e-05, + "loss": 1.1295, + "step": 2100 + }, + { + "epoch": 75.86, + "learning_rate": 4.8268656716417906e-05, + "loss": 1.0983, + "step": 2200 + }, + { + "epoch": 79.31, + "learning_rate": 4.6029850746268655e-05, + "loss": 1.0716, + "step": 2300 + }, + { + "epoch": 82.76, + "learning_rate": 4.37910447761194e-05, + "loss": 1.0582, + "step": 2400 + }, + { + "epoch": 86.21, + "learning_rate": 4.155223880597015e-05, + "loss": 1.0112, + "step": 2500 + }, + { + "epoch": 86.21, + "eval_loss": 0.46029242873191833, + "eval_runtime": 1.9007, + "eval_samples_per_second": 142.05, + "eval_steps_per_second": 4.735, + "eval_wer": 0.17533227006911217, + "step": 2500 + }, + { + "epoch": 89.66, + "learning_rate": 3.931343283582089e-05, + "loss": 0.9962, + "step": 2600 + }, + { + "epoch": 93.1, + "learning_rate": 3.707462686567164e-05, + "loss": 0.9746, + "step": 2700 + }, + { + "epoch": 96.55, + "learning_rate": 3.483582089552238e-05, + "loss": 0.9592, + "step": 2800 + }, + { + "epoch": 100.0, + "learning_rate": 3.259701492537313e-05, + "loss": 0.9364, + "step": 2900 + }, + { + "epoch": 103.45, + "learning_rate": 3.0358208955223878e-05, + "loss": 0.9387, + "step": 3000 + }, + { + "epoch": 103.45, + "eval_loss": 0.4466392397880554, + "eval_runtime": 1.9017, + "eval_samples_per_second": 141.977, + "eval_steps_per_second": 4.733, + "eval_wer": 0.1721424774056353, + "step": 3000 + } + ], + "max_steps": 4350, + "num_train_epochs": 150, + "total_flos": 1.2910510097926455e+19, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-3000/training_args.bin b/checkpoint-3000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..90cacf509e616d5a83c4819d896dbf967f59e207 --- /dev/null +++ b/checkpoint-3000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6f38b03969c3e29b358a2037b4f61da5251f02d2d9436641d5a5d94788db60e +size 2927 diff --git a/checkpoint-3500/config.json b/checkpoint-3500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..966cd6c5b3d8cbae43fe6586aa218930bd060f97 --- /dev/null +++ b/checkpoint-3500/config.json @@ -0,0 +1,107 @@ +{ + "_name_or_path": "./", + "activation_dropout": 0.1, + "adapter_kernel_size": 3, + "adapter_stride": 2, + "add_adapter": false, + "apply_spec_augment": true, + "architectures": [ + "Wav2Vec2ForCTC" + ], + "attention_dropout": 0.0, + "bos_token_id": 1, + "classifier_proj_size": 256, + "codevector_dim": 768, + "contrastive_logits_temperature": 0.1, + "conv_bias": true, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "mean", + "ctc_zero_infinity": false, + "diversity_loss_weight": 0.1, + "do_stable_layer_norm": true, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_dropout": 0.0, + "feat_extract_norm": "layer", + "feat_proj_dropout": 0.0, + "feat_quantizer_dropout": 0.0, + "final_dropout": 0.0, + "hidden_act": "gelu", + "hidden_dropout": 0.0, + "hidden_size": 1024, + "initializer_range": 0.02, + "intermediate_size": 4096, + "layer_norm_eps": 1e-05, + "layerdrop": 0.0, + "mask_feature_length": 64, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.25, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.75, + "model_type": "wav2vec2", + "num_adapter_layers": 3, + "num_attention_heads": 16, + "num_codevector_groups": 2, + "num_codevectors_per_group": 320, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 24, + "num_negatives": 100, + "output_hidden_size": 1024, + "pad_token_id": 48, + "proj_codevector_dim": 768, + "tdnn_dilation": [ + 1, + 2, + 3, + 1, + 1 + ], + "tdnn_dim": [ + 512, + 512, + 512, + 512, + 1500 + ], + "tdnn_kernel": [ + 5, + 3, + 3, + 1, + 1 + ], + "torch_dtype": "float32", + "transformers_version": "4.15.0.dev0", + "use_weighted_layer_sum": false, + "vocab_size": 52, + "xvector_output_dim": 512 +} diff --git a/checkpoint-3500/optimizer.pt b/checkpoint-3500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..6cc3fff85fda8fee87643918b19c09a64e0c97e4 --- /dev/null +++ b/checkpoint-3500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f421d2b1e3f06b4cff391206ab29dcb32cf2aac5324f7d41ebf8aca6fada942 +size 2490485393 diff --git a/checkpoint-3500/preprocessor_config.json b/checkpoint-3500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..36ebe8b7c1cc967b3059f0494ae8a1069dd67655 --- /dev/null +++ b/checkpoint-3500/preprocessor_config.json @@ -0,0 +1,9 @@ +{ + "do_normalize": true, + "feature_extractor_type": "Wav2Vec2FeatureExtractor", + "feature_size": 1, + "padding_side": "right", + "padding_value": 0, + "return_attention_mask": true, + "sampling_rate": 16000 +} diff --git a/checkpoint-3500/pytorch_model.bin b/checkpoint-3500/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..6bc61ab82f812c0dde333cfd0238474851fcddbd --- /dev/null +++ b/checkpoint-3500/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8507f1fd4951af8ae7c4f333000a1b1ca0b8429480c2955ef2548d28c5793ab +size 1262136881 diff --git a/checkpoint-3500/rng_state_0.pth b/checkpoint-3500/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..b166211091fcbd1d71a8ec56adf33ed70069737f --- /dev/null +++ b/checkpoint-3500/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8788c98275b28b5b9ec4bfcd1f218ccda9c88b355673c8d00abef3cc369028cc +size 14567 diff --git a/checkpoint-3500/rng_state_1.pth b/checkpoint-3500/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..cdf667d920bb2ead90e013041beb99157267a477 --- /dev/null +++ b/checkpoint-3500/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3caaf21470c55a26a83a6db1f9e0b5e1adc0a4b02d7d570660a3bc828b3d8d6 +size 14503 diff --git a/checkpoint-3500/rng_state_2.pth b/checkpoint-3500/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f472dd999b1b51aa894a2f062d01093fcae3b05a --- /dev/null +++ b/checkpoint-3500/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77d386527eecef36648e701aca8cbe60644c0913930ddef107ad583bcd96ed69 +size 14567 diff --git a/checkpoint-3500/rng_state_3.pth b/checkpoint-3500/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..052fd06e61494765fe573957b5c4d7ee009faf72 --- /dev/null +++ b/checkpoint-3500/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0244895a4a72f32d72d09ba5f40370b298c8da4f5c7cd488b7c7aea9f0568ba0 +size 14567 diff --git a/checkpoint-3500/rng_state_4.pth b/checkpoint-3500/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..62b58752baa9984b7d6830785678a83aeade7b4f --- /dev/null +++ b/checkpoint-3500/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86af25e4eb41d0743b2678df4a77f634ab5b517e8354374f65f13d45d22cdbbf +size 14567 diff --git a/checkpoint-3500/rng_state_5.pth b/checkpoint-3500/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..9b3f144a538dd472bb98eca4c3b4ff35e63cd1a4 --- /dev/null +++ b/checkpoint-3500/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5457892e73154387ee22e28a873ddc0c92ba3c1ca39c4efef6d0bf475deaeeb7 +size 14567 diff --git a/checkpoint-3500/rng_state_6.pth b/checkpoint-3500/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..9af7c366e1864ff65ede0ab2e903b076c595e86d --- /dev/null +++ b/checkpoint-3500/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea5e60f39a6c296b52983d9a64b3924d0c3cf6d0930834aeba8d037f54e78a8e +size 14567 diff --git a/checkpoint-3500/rng_state_7.pth b/checkpoint-3500/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..da305c74c485f690748f9ebe5ba135cb5726eb98 --- /dev/null +++ b/checkpoint-3500/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5062184afe4371e809d6e392f3ce930b22620b6864cc1c51d79270e651eb17bb +size 14567 diff --git a/checkpoint-3500/scaler.pt b/checkpoint-3500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..940fe2e37f3e6ff0bcc57566d5e425ade21efdf5 --- /dev/null +++ b/checkpoint-3500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d09341b70ef57760e07267e67bbaee462df7adf8d23d92a20cc576715bd2b094 +size 559 diff --git a/checkpoint-3500/scheduler.pt b/checkpoint-3500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..542c1188dafa78eed4f4a87d531de2467e5a3ac9 --- /dev/null +++ b/checkpoint-3500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fbe3309fc6bfffbdf44f870e02817da89e5efe88906cbd19563e50da8624cfc +size 623 diff --git a/checkpoint-3500/trainer_state.json b/checkpoint-3500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..37df15096ec0e37741ab55c892781621ba99869f --- /dev/null +++ b/checkpoint-3500/trainer_state.json @@ -0,0 +1,289 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 120.6896551724138, + "global_step": 3500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 3.45, + "learning_rate": 7.2e-06, + "loss": 16.2031, + "step": 100 + }, + { + "epoch": 6.9, + "learning_rate": 1.47e-05, + "loss": 6.6774, + "step": 200 + }, + { + "epoch": 10.34, + "learning_rate": 2.2199999999999998e-05, + "loss": 4.7603, + "step": 300 + }, + { + "epoch": 13.79, + "learning_rate": 2.97e-05, + "loss": 3.9198, + "step": 400 + }, + { + "epoch": 17.24, + "learning_rate": 3.7199999999999996e-05, + "loss": 3.5558, + "step": 500 + }, + { + "epoch": 17.24, + "eval_loss": 3.5216922760009766, + "eval_runtime": 2.4797, + "eval_samples_per_second": 108.882, + "eval_steps_per_second": 3.629, + "eval_wer": 0.9812865497076023, + "step": 500 + }, + { + "epoch": 20.69, + "learning_rate": 4.4699999999999996e-05, + "loss": 3.4799, + "step": 600 + }, + { + "epoch": 24.14, + "learning_rate": 5.2199999999999995e-05, + "loss": 3.4271, + "step": 700 + }, + { + "epoch": 27.59, + "learning_rate": 5.97e-05, + "loss": 3.2779, + "step": 800 + }, + { + "epoch": 31.03, + "learning_rate": 6.72e-05, + "loss": 2.9317, + "step": 900 + }, + { + "epoch": 34.48, + "learning_rate": 7.47e-05, + "loss": 2.2202, + "step": 1000 + }, + { + "epoch": 34.48, + "eval_loss": 0.9991578459739685, + "eval_runtime": 1.9665, + "eval_samples_per_second": 137.299, + "eval_steps_per_second": 4.577, + "eval_wer": 0.2637958532695375, + "step": 1000 + }, + { + "epoch": 37.93, + "learning_rate": 7.285074626865672e-05, + "loss": 1.835, + "step": 1100 + }, + { + "epoch": 41.38, + "learning_rate": 7.061194029850745e-05, + "loss": 1.653, + "step": 1200 + }, + { + "epoch": 44.83, + "learning_rate": 6.83731343283582e-05, + "loss": 1.5243, + "step": 1300 + }, + { + "epoch": 48.28, + "learning_rate": 6.613432835820895e-05, + "loss": 1.4343, + "step": 1400 + }, + { + "epoch": 51.72, + "learning_rate": 6.38955223880597e-05, + "loss": 1.376, + "step": 1500 + }, + { + "epoch": 51.72, + "eval_loss": 0.5396047830581665, + "eval_runtime": 1.8956, + "eval_samples_per_second": 142.435, + "eval_steps_per_second": 4.748, + "eval_wer": 0.1946836788942052, + "step": 1500 + }, + { + "epoch": 55.17, + "learning_rate": 6.165671641791044e-05, + "loss": 1.3168, + "step": 1600 + }, + { + "epoch": 58.62, + "learning_rate": 5.941791044776119e-05, + "loss": 1.2773, + "step": 1700 + }, + { + "epoch": 62.07, + "learning_rate": 5.717910447761193e-05, + "loss": 1.2455, + "step": 1800 + }, + { + "epoch": 65.52, + "learning_rate": 5.494029850746268e-05, + "loss": 1.1881, + "step": 1900 + }, + { + "epoch": 68.97, + "learning_rate": 5.272388059701492e-05, + "loss": 1.1565, + "step": 2000 + }, + { + "epoch": 68.97, + "eval_loss": 0.4707144498825073, + "eval_runtime": 1.8963, + "eval_samples_per_second": 142.381, + "eval_steps_per_second": 4.746, + "eval_wer": 0.1773524720893142, + "step": 2000 + }, + { + "epoch": 72.41, + "learning_rate": 5.048507462686567e-05, + "loss": 1.1295, + "step": 2100 + }, + { + "epoch": 75.86, + "learning_rate": 4.8268656716417906e-05, + "loss": 1.0983, + "step": 2200 + }, + { + "epoch": 79.31, + "learning_rate": 4.6029850746268655e-05, + "loss": 1.0716, + "step": 2300 + }, + { + "epoch": 82.76, + "learning_rate": 4.37910447761194e-05, + "loss": 1.0582, + "step": 2400 + }, + { + "epoch": 86.21, + "learning_rate": 4.155223880597015e-05, + "loss": 1.0112, + "step": 2500 + }, + { + "epoch": 86.21, + "eval_loss": 0.46029242873191833, + "eval_runtime": 1.9007, + "eval_samples_per_second": 142.05, + "eval_steps_per_second": 4.735, + "eval_wer": 0.17533227006911217, + "step": 2500 + }, + { + "epoch": 89.66, + "learning_rate": 3.931343283582089e-05, + "loss": 0.9962, + "step": 2600 + }, + { + "epoch": 93.1, + "learning_rate": 3.707462686567164e-05, + "loss": 0.9746, + "step": 2700 + }, + { + "epoch": 96.55, + "learning_rate": 3.483582089552238e-05, + "loss": 0.9592, + "step": 2800 + }, + { + "epoch": 100.0, + "learning_rate": 3.259701492537313e-05, + "loss": 0.9364, + "step": 2900 + }, + { + "epoch": 103.45, + "learning_rate": 3.0358208955223878e-05, + "loss": 0.9387, + "step": 3000 + }, + { + "epoch": 103.45, + "eval_loss": 0.4466392397880554, + "eval_runtime": 1.9017, + "eval_samples_per_second": 141.977, + "eval_steps_per_second": 4.733, + "eval_wer": 0.1721424774056353, + "step": 3000 + }, + { + "epoch": 106.9, + "learning_rate": 2.811940298507462e-05, + "loss": 0.9223, + "step": 3100 + }, + { + "epoch": 110.34, + "learning_rate": 2.588059701492537e-05, + "loss": 0.8839, + "step": 3200 + }, + { + "epoch": 113.79, + "learning_rate": 2.364179104477612e-05, + "loss": 0.8845, + "step": 3300 + }, + { + "epoch": 117.24, + "learning_rate": 2.1402985074626863e-05, + "loss": 0.8604, + "step": 3400 + }, + { + "epoch": 120.69, + "learning_rate": 1.9164179104477612e-05, + "loss": 0.8526, + "step": 3500 + }, + { + "epoch": 120.69, + "eval_loss": 0.43604278564453125, + "eval_runtime": 1.8975, + "eval_samples_per_second": 142.29, + "eval_steps_per_second": 4.743, + "eval_wer": 0.17044125465178098, + "step": 3500 + } + ], + "max_steps": 4350, + "num_train_epochs": 150, + "total_flos": 1.5065166616188682e+19, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-3500/training_args.bin b/checkpoint-3500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..90cacf509e616d5a83c4819d896dbf967f59e207 --- /dev/null +++ b/checkpoint-3500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6f38b03969c3e29b358a2037b4f61da5251f02d2d9436641d5a5d94788db60e +size 2927 diff --git a/checkpoint-4000/config.json b/checkpoint-4000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..966cd6c5b3d8cbae43fe6586aa218930bd060f97 --- /dev/null +++ b/checkpoint-4000/config.json @@ -0,0 +1,107 @@ +{ + "_name_or_path": "./", + "activation_dropout": 0.1, + "adapter_kernel_size": 3, + "adapter_stride": 2, + "add_adapter": false, + "apply_spec_augment": true, + "architectures": [ + "Wav2Vec2ForCTC" + ], + "attention_dropout": 0.0, + "bos_token_id": 1, + "classifier_proj_size": 256, + "codevector_dim": 768, + "contrastive_logits_temperature": 0.1, + "conv_bias": true, + "conv_dim": [ + 512, + 512, + 512, + 512, + 512, + 512, + 512 + ], + "conv_kernel": [ + 10, + 3, + 3, + 3, + 3, + 2, + 2 + ], + "conv_stride": [ + 5, + 2, + 2, + 2, + 2, + 2, + 2 + ], + "ctc_loss_reduction": "mean", + "ctc_zero_infinity": false, + "diversity_loss_weight": 0.1, + "do_stable_layer_norm": true, + "eos_token_id": 2, + "feat_extract_activation": "gelu", + "feat_extract_dropout": 0.0, + "feat_extract_norm": "layer", + "feat_proj_dropout": 0.0, + "feat_quantizer_dropout": 0.0, + "final_dropout": 0.0, + "hidden_act": "gelu", + "hidden_dropout": 0.0, + "hidden_size": 1024, + "initializer_range": 0.02, + "intermediate_size": 4096, + "layer_norm_eps": 1e-05, + "layerdrop": 0.0, + "mask_feature_length": 64, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.25, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.75, + "model_type": "wav2vec2", + "num_adapter_layers": 3, + "num_attention_heads": 16, + "num_codevector_groups": 2, + "num_codevectors_per_group": 320, + "num_conv_pos_embedding_groups": 16, + "num_conv_pos_embeddings": 128, + "num_feat_extract_layers": 7, + "num_hidden_layers": 24, + "num_negatives": 100, + "output_hidden_size": 1024, + "pad_token_id": 48, + "proj_codevector_dim": 768, + "tdnn_dilation": [ + 1, + 2, + 3, + 1, + 1 + ], + "tdnn_dim": [ + 512, + 512, + 512, + 512, + 1500 + ], + "tdnn_kernel": [ + 5, + 3, + 3, + 1, + 1 + ], + "torch_dtype": "float32", + "transformers_version": "4.15.0.dev0", + "use_weighted_layer_sum": false, + "vocab_size": 52, + "xvector_output_dim": 512 +} diff --git a/checkpoint-4000/optimizer.pt b/checkpoint-4000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee9ab48a3a014efd2b92033fc3b95f0c214e9c94 --- /dev/null +++ b/checkpoint-4000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9b7156890d92f7e3e7d73a964ed92fda2116a2a252494fbe66415631ae94962 +size 2490485393 diff --git a/checkpoint-4000/preprocessor_config.json b/checkpoint-4000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..36ebe8b7c1cc967b3059f0494ae8a1069dd67655 --- /dev/null +++ b/checkpoint-4000/preprocessor_config.json @@ -0,0 +1,9 @@ +{ + "do_normalize": true, + "feature_extractor_type": "Wav2Vec2FeatureExtractor", + "feature_size": 1, + "padding_side": "right", + "padding_value": 0, + "return_attention_mask": true, + "sampling_rate": 16000 +} diff --git a/checkpoint-4000/pytorch_model.bin b/checkpoint-4000/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..dd9f0e3d0911e1a79629b545a8156ed04e8fe7c4 --- /dev/null +++ b/checkpoint-4000/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b00fd64e7e95f566a2fae32794ebbe237a19914930bcdccec19c516a3fca89e7 +size 1262136881 diff --git a/checkpoint-4000/rng_state_0.pth b/checkpoint-4000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..04654321ff31f19864aaa8c4e086370649605d30 --- /dev/null +++ b/checkpoint-4000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c27c8e6d4c33919229c7d4240a66b6e288e3a01b097eba84264ed3974d264a01 +size 14567 diff --git a/checkpoint-4000/rng_state_1.pth b/checkpoint-4000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..6286bf1e2bb14cc65d6e04029ccbd686dd18276a --- /dev/null +++ b/checkpoint-4000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72ec11b564dc40fad0622c5eec645266b0d2893c3bfeae1edb8d94e3d7be93e4 +size 14567 diff --git a/checkpoint-4000/rng_state_2.pth b/checkpoint-4000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..2bf6cdd2b4a520255968c07e947a020465c78f63 --- /dev/null +++ b/checkpoint-4000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ab56d59383d5509dfa905bff398502522b42d99e2170d36e951a07fea75222d +size 14567 diff --git a/checkpoint-4000/rng_state_3.pth b/checkpoint-4000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4c17b137143d281942416bbc5eae44a86c0056d1 --- /dev/null +++ b/checkpoint-4000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:640572036962b8c35ede1eb981f6ccf5c6e4c045f16a43fdbac641b47b9a7b3c +size 14567 diff --git a/checkpoint-4000/rng_state_4.pth b/checkpoint-4000/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..120883cb7e8b3a7a40589d77bec920bb190812e0 --- /dev/null +++ b/checkpoint-4000/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dfdc4dd455cb23066cb93b1612945ab0a244a95ac4350fb7a8ff9d70e0222ed1 +size 14567 diff --git a/checkpoint-4000/rng_state_5.pth b/checkpoint-4000/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..9babbece02b613d813a90a71383a6b5183cf1650 --- /dev/null +++ b/checkpoint-4000/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6bc63bed88616cba15942352fff38898fc793302db4c0bc3055c5d6f63b1c01 +size 14567 diff --git a/checkpoint-4000/rng_state_6.pth b/checkpoint-4000/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..9cb405d1a3bf794d452a72892eaa23442c70ade5 --- /dev/null +++ b/checkpoint-4000/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2abd59205bcb4e28532b2fda4cb473371a64026fe2a28d95a5f767ad422f5a1 +size 14503 diff --git a/checkpoint-4000/rng_state_7.pth b/checkpoint-4000/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..5d3f679f248bd318cb1750ac9eabf920090bf44c --- /dev/null +++ b/checkpoint-4000/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f9f72a8b6e105d83979c9e66f262d8265bf1b4ae2763d5a63dbde9b12fa6a07 +size 14567 diff --git a/checkpoint-4000/scaler.pt b/checkpoint-4000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4c21617a534d101985327ce3c9910f0f901de0d --- /dev/null +++ b/checkpoint-4000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75dc1f3df92aaeaa0e2e26fd93ca618cb37b77b3395d56a08ef59bb2bb1f2047 +size 559 diff --git a/checkpoint-4000/scheduler.pt b/checkpoint-4000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..caf08a59c52751150c32324b9f675a723b0812db --- /dev/null +++ b/checkpoint-4000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e415907330c3a0a40649be82e9bcced404f2cb43e4bba013232cd32680765684 +size 623 diff --git a/checkpoint-4000/trainer_state.json b/checkpoint-4000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..0a8e0c4250592141fe92fafc12792d8208d5b82a --- /dev/null +++ b/checkpoint-4000/trainer_state.json @@ -0,0 +1,328 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 137.93103448275863, + "global_step": 4000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 3.45, + "learning_rate": 7.2e-06, + "loss": 16.2031, + "step": 100 + }, + { + "epoch": 6.9, + "learning_rate": 1.47e-05, + "loss": 6.6774, + "step": 200 + }, + { + "epoch": 10.34, + "learning_rate": 2.2199999999999998e-05, + "loss": 4.7603, + "step": 300 + }, + { + "epoch": 13.79, + "learning_rate": 2.97e-05, + "loss": 3.9198, + "step": 400 + }, + { + "epoch": 17.24, + "learning_rate": 3.7199999999999996e-05, + "loss": 3.5558, + "step": 500 + }, + { + "epoch": 17.24, + "eval_loss": 3.5216922760009766, + "eval_runtime": 2.4797, + "eval_samples_per_second": 108.882, + "eval_steps_per_second": 3.629, + "eval_wer": 0.9812865497076023, + "step": 500 + }, + { + "epoch": 20.69, + "learning_rate": 4.4699999999999996e-05, + "loss": 3.4799, + "step": 600 + }, + { + "epoch": 24.14, + "learning_rate": 5.2199999999999995e-05, + "loss": 3.4271, + "step": 700 + }, + { + "epoch": 27.59, + "learning_rate": 5.97e-05, + "loss": 3.2779, + "step": 800 + }, + { + "epoch": 31.03, + "learning_rate": 6.72e-05, + "loss": 2.9317, + "step": 900 + }, + { + "epoch": 34.48, + "learning_rate": 7.47e-05, + "loss": 2.2202, + "step": 1000 + }, + { + "epoch": 34.48, + "eval_loss": 0.9991578459739685, + "eval_runtime": 1.9665, + "eval_samples_per_second": 137.299, + "eval_steps_per_second": 4.577, + "eval_wer": 0.2637958532695375, + "step": 1000 + }, + { + "epoch": 37.93, + "learning_rate": 7.285074626865672e-05, + "loss": 1.835, + "step": 1100 + }, + { + "epoch": 41.38, + "learning_rate": 7.061194029850745e-05, + "loss": 1.653, + "step": 1200 + }, + { + "epoch": 44.83, + "learning_rate": 6.83731343283582e-05, + "loss": 1.5243, + "step": 1300 + }, + { + "epoch": 48.28, + "learning_rate": 6.613432835820895e-05, + "loss": 1.4343, + "step": 1400 + }, + { + "epoch": 51.72, + "learning_rate": 6.38955223880597e-05, + "loss": 1.376, + "step": 1500 + }, + { + "epoch": 51.72, + "eval_loss": 0.5396047830581665, + "eval_runtime": 1.8956, + "eval_samples_per_second": 142.435, + "eval_steps_per_second": 4.748, + "eval_wer": 0.1946836788942052, + "step": 1500 + }, + { + "epoch": 55.17, + "learning_rate": 6.165671641791044e-05, + "loss": 1.3168, + "step": 1600 + }, + { + "epoch": 58.62, + "learning_rate": 5.941791044776119e-05, + "loss": 1.2773, + "step": 1700 + }, + { + "epoch": 62.07, + "learning_rate": 5.717910447761193e-05, + "loss": 1.2455, + "step": 1800 + }, + { + "epoch": 65.52, + "learning_rate": 5.494029850746268e-05, + "loss": 1.1881, + "step": 1900 + }, + { + "epoch": 68.97, + "learning_rate": 5.272388059701492e-05, + "loss": 1.1565, + "step": 2000 + }, + { + "epoch": 68.97, + "eval_loss": 0.4707144498825073, + "eval_runtime": 1.8963, + "eval_samples_per_second": 142.381, + "eval_steps_per_second": 4.746, + "eval_wer": 0.1773524720893142, + "step": 2000 + }, + { + "epoch": 72.41, + "learning_rate": 5.048507462686567e-05, + "loss": 1.1295, + "step": 2100 + }, + { + "epoch": 75.86, + "learning_rate": 4.8268656716417906e-05, + "loss": 1.0983, + "step": 2200 + }, + { + "epoch": 79.31, + "learning_rate": 4.6029850746268655e-05, + "loss": 1.0716, + "step": 2300 + }, + { + "epoch": 82.76, + "learning_rate": 4.37910447761194e-05, + "loss": 1.0582, + "step": 2400 + }, + { + "epoch": 86.21, + "learning_rate": 4.155223880597015e-05, + "loss": 1.0112, + "step": 2500 + }, + { + "epoch": 86.21, + "eval_loss": 0.46029242873191833, + "eval_runtime": 1.9007, + "eval_samples_per_second": 142.05, + "eval_steps_per_second": 4.735, + "eval_wer": 0.17533227006911217, + "step": 2500 + }, + { + "epoch": 89.66, + "learning_rate": 3.931343283582089e-05, + "loss": 0.9962, + "step": 2600 + }, + { + "epoch": 93.1, + "learning_rate": 3.707462686567164e-05, + "loss": 0.9746, + "step": 2700 + }, + { + "epoch": 96.55, + "learning_rate": 3.483582089552238e-05, + "loss": 0.9592, + "step": 2800 + }, + { + "epoch": 100.0, + "learning_rate": 3.259701492537313e-05, + "loss": 0.9364, + "step": 2900 + }, + { + "epoch": 103.45, + "learning_rate": 3.0358208955223878e-05, + "loss": 0.9387, + "step": 3000 + }, + { + "epoch": 103.45, + "eval_loss": 0.4466392397880554, + "eval_runtime": 1.9017, + "eval_samples_per_second": 141.977, + "eval_steps_per_second": 4.733, + "eval_wer": 0.1721424774056353, + "step": 3000 + }, + { + "epoch": 106.9, + "learning_rate": 2.811940298507462e-05, + "loss": 0.9223, + "step": 3100 + }, + { + "epoch": 110.34, + "learning_rate": 2.588059701492537e-05, + "loss": 0.8839, + "step": 3200 + }, + { + "epoch": 113.79, + "learning_rate": 2.364179104477612e-05, + "loss": 0.8845, + "step": 3300 + }, + { + "epoch": 117.24, + "learning_rate": 2.1402985074626863e-05, + "loss": 0.8604, + "step": 3400 + }, + { + "epoch": 120.69, + "learning_rate": 1.9164179104477612e-05, + "loss": 0.8526, + "step": 3500 + }, + { + "epoch": 120.69, + "eval_loss": 0.43604278564453125, + "eval_runtime": 1.8975, + "eval_samples_per_second": 142.29, + "eval_steps_per_second": 4.743, + "eval_wer": 0.17044125465178098, + "step": 3500 + }, + { + "epoch": 124.14, + "learning_rate": 1.6925373134328355e-05, + "loss": 0.8311, + "step": 3600 + }, + { + "epoch": 127.59, + "learning_rate": 1.4686567164179104e-05, + "loss": 0.832, + "step": 3700 + }, + { + "epoch": 131.03, + "learning_rate": 1.244776119402985e-05, + "loss": 0.812, + "step": 3800 + }, + { + "epoch": 134.48, + "learning_rate": 1.0208955223880597e-05, + "loss": 0.8016, + "step": 3900 + }, + { + "epoch": 137.93, + "learning_rate": 7.970149253731343e-06, + "loss": 0.8018, + "step": 4000 + }, + { + "epoch": 137.93, + "eval_loss": 0.437362939119339, + "eval_runtime": 1.9635, + "eval_samples_per_second": 137.506, + "eval_steps_per_second": 4.584, + "eval_wer": 0.1720361509835194, + "step": 4000 + } + ], + "max_steps": 4350, + "num_train_epochs": 150, + "total_flos": 1.7217444822080553e+19, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-4000/training_args.bin b/checkpoint-4000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..90cacf509e616d5a83c4819d896dbf967f59e207 --- /dev/null +++ b/checkpoint-4000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6f38b03969c3e29b358a2037b4f61da5251f02d2d9436641d5a5d94788db60e +size 2927 diff --git a/config.json b/config.json index b7b39db6bafe47ff0d8022a79aa6abae08de381a..966cd6c5b3d8cbae43fe6586aa218930bd060f97 100644 --- a/config.json +++ b/config.json @@ -1,6 +1,6 @@ { - "_name_or_path": "facebook/wav2vec2-xls-r-300m", - "activation_dropout": 0.0, + "_name_or_path": "./", + "activation_dropout": 0.1, "adapter_kernel_size": 3, "adapter_stride": 2, "add_adapter": false, @@ -8,7 +8,7 @@ "architectures": [ "Wav2Vec2ForCTC" ], - "attention_dropout": 0.1, + "attention_dropout": 0.0, "bos_token_id": 1, "classifier_proj_size": 256, "codevector_dim": 768, @@ -41,7 +41,7 @@ 2, 2 ], - "ctc_loss_reduction": "sum", + "ctc_loss_reduction": "mean", "ctc_zero_infinity": false, "diversity_loss_weight": 0.1, "do_stable_layer_norm": true, @@ -49,23 +49,22 @@ "feat_extract_activation": "gelu", "feat_extract_dropout": 0.0, "feat_extract_norm": "layer", - "feat_proj_dropout": 0.1, + "feat_proj_dropout": 0.0, "feat_quantizer_dropout": 0.0, "final_dropout": 0.0, - "gradient_checkpointing": false, "hidden_act": "gelu", - "hidden_dropout": 0.1, + "hidden_dropout": 0.0, "hidden_size": 1024, "initializer_range": 0.02, "intermediate_size": 4096, "layer_norm_eps": 1e-05, - "layerdrop": 0.1, - "mask_feature_length": 10, + "layerdrop": 0.0, + "mask_feature_length": 64, "mask_feature_min_masks": 0, - "mask_feature_prob": 0.0, + "mask_feature_prob": 0.25, "mask_time_length": 10, "mask_time_min_masks": 2, - "mask_time_prob": 0.075, + "mask_time_prob": 0.75, "model_type": "wav2vec2", "num_adapter_layers": 3, "num_attention_heads": 16, @@ -77,7 +76,7 @@ "num_hidden_layers": 24, "num_negatives": 100, "output_hidden_size": 1024, - "pad_token_id": 0, + "pad_token_id": 48, "proj_codevector_dim": 768, "tdnn_dilation": [ 1, diff --git a/emissions.csv b/emissions.csv new file mode 100644 index 0000000000000000000000000000000000000000..258e4c2a1f9f3a212ab68eaef7e77c70f96a4245 --- /dev/null +++ b/emissions.csv @@ -0,0 +1,2 @@ +timestamp,experiment_id,project_name,duration,emissions,energy_consumed,country_name,country_iso_code,region,on_cloud,cloud_provider,cloud_region +2021-12-21T00:10:30,8748a241-eb86-410e-bd9b-5434a78c42ab,codecarbon,1805.1241974830627,0.2642598757127812,0.7184879709428527,USA,USA,nan,Y,aws,us-east-1 diff --git a/eval_result.txt b/eval_result.txt new file mode 100644 index 0000000000000000000000000000000000000000..80f7cc01ccd23418639385f5b2c15eb22f3ef3e4 --- /dev/null +++ b/eval_result.txt @@ -0,0 +1 @@ +wer 0.09936981963803433 diff --git a/eval_results.json b/eval_results.json new file mode 100644 index 0000000000000000000000000000000000000000..4ba0382b43504bda736a14f175e83b5e17123024 --- /dev/null +++ b/eval_results.json @@ -0,0 +1,9 @@ +{ + "epoch": 150.0, + "eval_loss": 0.4378179907798767, + "eval_runtime": 1.9751, + "eval_samples": 270, + "eval_samples_per_second": 136.702, + "eval_steps_per_second": 4.557, + "eval_wer": 0.16863370547581075 +} \ No newline at end of file diff --git a/pytorch_model.bin b/pytorch_model.bin index 6745de9094bd2b59677d7aa306a1c77b3269cbdc..cbe7cb4c829ba37434c3be2c1c9efb238a4ef8ff 100644 --- a/pytorch_model.bin +++ b/pytorch_model.bin @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:960d49e43ad9640ba3b5f517d99c95661af6c65d2eececb50e7182d27f730437 -size 1262130993 +oid sha256:b6f7446584f03b6250f9367f6cf819ff172b7663bf46ca909353a04e610feb67 +size 1262136881 diff --git a/run.sh b/run.sh index dfd186e5d3cb55f7cee11d7ee72201270c23933b..cad15cefa472225521a903ed66303c57a6364979 100755 --- a/run.sh +++ b/run.sh @@ -5,8 +5,8 @@ python -m torch.distributed.launch \ --model_name_or_path="./" \ --tokenizer_name_or_path="./" \ --dataset_name="mozilla-foundation/common_voice_3_0" \ - --path_to_eval_file="./valSeqs_1.0_uniform_new_version.text" \ - --path_to_train_file="./trainSeqs_1.0_uniform_new_version.text" \ + --path_to_eval_file="./valSeqs_1.0_uniform_new_version.txt" \ + --path_to_train_file="./trainSeqs_1.0_uniform_new_version.txt" \ --dataset_config_name="tr" \ --text_column_name="sentence" \ --output_dir="./" \ @@ -19,7 +19,7 @@ python -m torch.distributed.launch \ --warmup_steps="1000" \ --evaluation_strategy="steps" \ --save_steps="500" \ - --eval_steps="2" \ + --eval_steps="500" \ --logging_steps="100" \ --layerdrop="0.0" \ --activation_dropout="0.1" \ diff --git a/run_eval.py b/run_eval.py new file mode 100755 index 0000000000000000000000000000000000000000..21d501b047e3119563ed9e6cf8966027338e3cb1 --- /dev/null +++ b/run_eval.py @@ -0,0 +1,51 @@ +#!/usr/bin/env python3 +from datasets import load_dataset, load_metric +import datasets +import torch +from transformers import AutoModelForCTC, AutoProcessor + + +device = "cuda:0" if torch.cuda.is_available() else "cpu" + + +ds = load_dataset("mozilla-foundation/common_voice_3_0", "tr", split="train+validation+test+other") +wer = load_metric("wer") + +model = AutoModelForCTC.from_pretrained("./") +model = model.to(device) +processor = AutoProcessor.from_pretrained("./") + + +# taken from +# https://github.com/microsoft/UniSpeech/blob/main/UniSpeech/examples/unispeech/data/it/phonesMatches_reduced.json +with open("./testSeqs_uniform_new_version.txt", "r") as f: + lines = f.readlines() + + +# retrieve ids model is evaluated on +ids = [x.strip() for x in lines] + + +ds = ds.filter(lambda p: p.split("/")[-1].split(".")[0] in ids, input_columns=["path"]) + +ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16_000)) + + +def decode(batch): + input_values = processor(batch["audio"]["array"], return_tensors="pt", sampling_rate=16_000).input_values + input_values = input_values.to(device) + logits = model(input_values).logits + + pred_ids = torch.argmax(logits, axis=-1) + + batch["id"] = batch["path"] + batch["prediction"] = processor.batch_decode(pred_ids)[0] + batch["target"] = processor.tokenizer.phonemize(batch["sentence"]) + + return batch + + +out = ds.map(decode, remove_columns=ds.column_names) +wer_out = wer.compute(predictions=out["prediction"], references=out["target"]) + +print("wer", wer_out) diff --git a/run_speech_recognition_ctc.py b/run_speech_recognition_ctc.py index 7f229fe0583276c4c4a0546f47291ecf2eb32523..88ed5db0ec16d28471bec65c5660cd3ab4b85564 100755 --- a/run_speech_recognition_ctc.py +++ b/run_speech_recognition_ctc.py @@ -357,7 +357,7 @@ def filter_dataset(ds, path_to_file): lines = f.readlines() # retrieve ids model is evaluated on - ids = [x.split("\t")[0] for x in lines] + ids = [x.strip() for x in lines] ds = ds.filter(lambda p: p.split("/")[-1].split(".")[0] in ids, input_columns=["path"]) return ds @@ -652,7 +652,6 @@ def main(): # we do not want to group tokens when computing the metrics label_str = tokenizer.batch_decode(pred.label_ids, group_tokens=False) - import ipdb; ipdb.set_trace() metrics = {k: v.compute(predictions=pred_str, references=label_str) for k, v in eval_metrics.items()} return metrics diff --git a/runs/Dec20_22-34-59_ip-172-31-1-149/1640040770.1765845/events.out.tfevents.1640040770.ip-172-31-1-149.384883.1 b/runs/Dec20_22-34-59_ip-172-31-1-149/1640040770.1765845/events.out.tfevents.1640040770.ip-172-31-1-149.384883.1 new file mode 100644 index 0000000000000000000000000000000000000000..ac5ea2a1c8a6ebd1ebe5a37eb2354bda6538bf46 --- /dev/null +++ b/runs/Dec20_22-34-59_ip-172-31-1-149/1640040770.1765845/events.out.tfevents.1640040770.ip-172-31-1-149.384883.1 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f9f613d74edcd2669bd0026e4cf926992425b2ea62a78fbd4565f3c9289ada4 +size 4661 diff --git a/runs/Dec20_22-34-59_ip-172-31-1-149/events.out.tfevents.1640040770.ip-172-31-1-149.384883.0 b/runs/Dec20_22-34-59_ip-172-31-1-149/events.out.tfevents.1640040770.ip-172-31-1-149.384883.0 new file mode 100644 index 0000000000000000000000000000000000000000..98a787915ad1f28b62a9a428eed986e0951ef5bc --- /dev/null +++ b/runs/Dec20_22-34-59_ip-172-31-1-149/events.out.tfevents.1640040770.ip-172-31-1-149.384883.0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:191ceaf9c7d5cff6d79041034bb8e39d065fa758b38942b3665dbed003e83d16 +size 4596 diff --git a/runs/Dec20_23-02-28_ip-172-31-1-149/1640042398.7985415/events.out.tfevents.1640042398.ip-172-31-1-149.400297.1 b/runs/Dec20_23-02-28_ip-172-31-1-149/1640042398.7985415/events.out.tfevents.1640042398.ip-172-31-1-149.400297.1 new file mode 100644 index 0000000000000000000000000000000000000000..538452a4d2153bb62465fd9a5521daaefc76a7ec --- /dev/null +++ b/runs/Dec20_23-02-28_ip-172-31-1-149/1640042398.7985415/events.out.tfevents.1640042398.ip-172-31-1-149.400297.1 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d11a9c245609bbc7382b0802f6774c99fd9dd014eb5ad61a908dd9254f5aeb5b +size 4661 diff --git a/runs/Dec20_23-02-28_ip-172-31-1-149/events.out.tfevents.1640042398.ip-172-31-1-149.400297.0 b/runs/Dec20_23-02-28_ip-172-31-1-149/events.out.tfevents.1640042398.ip-172-31-1-149.400297.0 new file mode 100644 index 0000000000000000000000000000000000000000..63350ad069b38118c29e5f1c567e38b6cd20670c --- /dev/null +++ b/runs/Dec20_23-02-28_ip-172-31-1-149/events.out.tfevents.1640042398.ip-172-31-1-149.400297.0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11dd554f7b18beabd7a3dd94a9a7db7e461e1befc5f187a98c3b7ae70c441977 +size 4596 diff --git a/runs/Dec20_23-33-24_ip-172-31-1-149/1640043497.5574217/events.out.tfevents.1640043497.ip-172-31-1-149.415042.1 b/runs/Dec20_23-33-24_ip-172-31-1-149/1640043497.5574217/events.out.tfevents.1640043497.ip-172-31-1-149.415042.1 new file mode 100644 index 0000000000000000000000000000000000000000..0b4a1c68f3a981ab0845889ad3655c9c816093a6 --- /dev/null +++ b/runs/Dec20_23-33-24_ip-172-31-1-149/1640043497.5574217/events.out.tfevents.1640043497.ip-172-31-1-149.415042.1 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83918deb0c8a89ae0a0cb51fe8bf0da729f9a1b64edcd2bb0241be62538750dd +size 4661 diff --git a/runs/Dec20_23-33-24_ip-172-31-1-149/events.out.tfevents.1640043497.ip-172-31-1-149.415042.0 b/runs/Dec20_23-33-24_ip-172-31-1-149/events.out.tfevents.1640043497.ip-172-31-1-149.415042.0 new file mode 100644 index 0000000000000000000000000000000000000000..9f8c76b0a828ccb63aee6a969930ae8eac476901 --- /dev/null +++ b/runs/Dec20_23-33-24_ip-172-31-1-149/events.out.tfevents.1640043497.ip-172-31-1-149.415042.0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f21a2efc02232142c724d31b4abf45fc2658117ddf0f8532281f99127478f137 +size 9899 diff --git a/runs/Dec20_23-39-35_ip-172-31-1-149/1640043622.4880335/events.out.tfevents.1640043622.ip-172-31-1-149.418884.1 b/runs/Dec20_23-39-35_ip-172-31-1-149/1640043622.4880335/events.out.tfevents.1640043622.ip-172-31-1-149.418884.1 new file mode 100644 index 0000000000000000000000000000000000000000..0cf1c0e499869fd8f10be3ee0cc3d55b6cc3abda --- /dev/null +++ b/runs/Dec20_23-39-35_ip-172-31-1-149/1640043622.4880335/events.out.tfevents.1640043622.ip-172-31-1-149.418884.1 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:847bdaff4a79d1259d414b26ee50c0ac3d67cf8f6e80d38316d5728ff78ecb95 +size 4661 diff --git a/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640043622.ip-172-31-1-149.418884.0 b/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640043622.ip-172-31-1-149.418884.0 new file mode 100644 index 0000000000000000000000000000000000000000..b9c370fcb33c8522c90a0988f8b04954fe237974 --- /dev/null +++ b/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640043622.ip-172-31-1-149.418884.0 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44c48e48c7411c73a48c35c32831cbfb965cd1b1545b76525445c312336e41eb +size 14243 diff --git a/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640045438.ip-172-31-1-149.418884.2 b/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640045438.ip-172-31-1-149.418884.2 new file mode 100644 index 0000000000000000000000000000000000000000..1ae5b4557194727fa153e971eba6ca1d2fcc044d --- /dev/null +++ b/runs/Dec20_23-39-35_ip-172-31-1-149/events.out.tfevents.1640045438.ip-172-31-1-149.418884.2 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b73fe96faaade9d2839b4e8028c8c6e895321aaeef91b6248d9d5b7ddf17d72d +size 358 diff --git a/tokenizer_config.json b/tokenizer_config.json index c963b8a84d739dd93cbb93df3c425e946c9b8b6e..5f9961b3d05484876119e811e48c516b17c2d3e0 100644 --- a/tokenizer_config.json +++ b/tokenizer_config.json @@ -1 +1 @@ -{"unk_token": "", "bos_token": "", "eos_token": "", "pad_token": "", "word_delimiter_token": null, "phone_delimiter_token": " ", "do_phonemize": true, "phonemizer_lang": "en-us", "phonemizer_backend": "espeak", "tokenizer_class": "Wav2Vec2PhonemeCTCTokenizer"} \ No newline at end of file +{"unk_token": "", "bos_token": "", "eos_token": "", "pad_token": "", "word_delimiter_token": null, "phone_delimiter_token": " ", "do_phonemize": true, "phonemizer_lang": "tr", "phonemizer_backend": "espeak", "special_tokens_map_file": "./special_tokens_map.json", "tokenizer_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2PhonemeCTCTokenizer"} \ No newline at end of file diff --git a/train_results.json b/train_results.json new file mode 100644 index 0000000000000000000000000000000000000000..14065b6463932bf93b29e9b6c7ccdec1fdf0b057 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "epoch": 150.0, + "train_loss": 1.9741196327647943, + "train_runtime": 1808.1454, + "train_samples": 905, + "train_samples_per_second": 75.077, + "train_steps_per_second": 2.406 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..f88ddeb2e06c29e6408b4b11084779d00737f6ff --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,355 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 150.0, + "global_step": 4350, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 3.45, + "learning_rate": 7.2e-06, + "loss": 16.2031, + "step": 100 + }, + { + "epoch": 6.9, + "learning_rate": 1.47e-05, + "loss": 6.6774, + "step": 200 + }, + { + "epoch": 10.34, + "learning_rate": 2.2199999999999998e-05, + "loss": 4.7603, + "step": 300 + }, + { + "epoch": 13.79, + "learning_rate": 2.97e-05, + "loss": 3.9198, + "step": 400 + }, + { + "epoch": 17.24, + "learning_rate": 3.7199999999999996e-05, + "loss": 3.5558, + "step": 500 + }, + { + "epoch": 17.24, + "eval_loss": 3.5216922760009766, + "eval_runtime": 2.4797, + "eval_samples_per_second": 108.882, + "eval_steps_per_second": 3.629, + "eval_wer": 0.9812865497076023, + "step": 500 + }, + { + "epoch": 20.69, + "learning_rate": 4.4699999999999996e-05, + "loss": 3.4799, + "step": 600 + }, + { + "epoch": 24.14, + "learning_rate": 5.2199999999999995e-05, + "loss": 3.4271, + "step": 700 + }, + { + "epoch": 27.59, + "learning_rate": 5.97e-05, + "loss": 3.2779, + "step": 800 + }, + { + "epoch": 31.03, + "learning_rate": 6.72e-05, + "loss": 2.9317, + "step": 900 + }, + { + "epoch": 34.48, + "learning_rate": 7.47e-05, + "loss": 2.2202, + "step": 1000 + }, + { + "epoch": 34.48, + "eval_loss": 0.9991578459739685, + "eval_runtime": 1.9665, + "eval_samples_per_second": 137.299, + "eval_steps_per_second": 4.577, + "eval_wer": 0.2637958532695375, + "step": 1000 + }, + { + "epoch": 37.93, + "learning_rate": 7.285074626865672e-05, + "loss": 1.835, + "step": 1100 + }, + { + "epoch": 41.38, + "learning_rate": 7.061194029850745e-05, + "loss": 1.653, + "step": 1200 + }, + { + "epoch": 44.83, + "learning_rate": 6.83731343283582e-05, + "loss": 1.5243, + "step": 1300 + }, + { + "epoch": 48.28, + "learning_rate": 6.613432835820895e-05, + "loss": 1.4343, + "step": 1400 + }, + { + "epoch": 51.72, + "learning_rate": 6.38955223880597e-05, + "loss": 1.376, + "step": 1500 + }, + { + "epoch": 51.72, + "eval_loss": 0.5396047830581665, + "eval_runtime": 1.8956, + "eval_samples_per_second": 142.435, + "eval_steps_per_second": 4.748, + "eval_wer": 0.1946836788942052, + "step": 1500 + }, + { + "epoch": 55.17, + "learning_rate": 6.165671641791044e-05, + "loss": 1.3168, + "step": 1600 + }, + { + "epoch": 58.62, + "learning_rate": 5.941791044776119e-05, + "loss": 1.2773, + "step": 1700 + }, + { + "epoch": 62.07, + "learning_rate": 5.717910447761193e-05, + "loss": 1.2455, + "step": 1800 + }, + { + "epoch": 65.52, + "learning_rate": 5.494029850746268e-05, + "loss": 1.1881, + "step": 1900 + }, + { + "epoch": 68.97, + "learning_rate": 5.272388059701492e-05, + "loss": 1.1565, + "step": 2000 + }, + { + "epoch": 68.97, + "eval_loss": 0.4707144498825073, + "eval_runtime": 1.8963, + "eval_samples_per_second": 142.381, + "eval_steps_per_second": 4.746, + "eval_wer": 0.1773524720893142, + "step": 2000 + }, + { + "epoch": 72.41, + "learning_rate": 5.048507462686567e-05, + "loss": 1.1295, + "step": 2100 + }, + { + "epoch": 75.86, + "learning_rate": 4.8268656716417906e-05, + "loss": 1.0983, + "step": 2200 + }, + { + "epoch": 79.31, + "learning_rate": 4.6029850746268655e-05, + "loss": 1.0716, + "step": 2300 + }, + { + "epoch": 82.76, + "learning_rate": 4.37910447761194e-05, + "loss": 1.0582, + "step": 2400 + }, + { + "epoch": 86.21, + "learning_rate": 4.155223880597015e-05, + "loss": 1.0112, + "step": 2500 + }, + { + "epoch": 86.21, + "eval_loss": 0.46029242873191833, + "eval_runtime": 1.9007, + "eval_samples_per_second": 142.05, + "eval_steps_per_second": 4.735, + "eval_wer": 0.17533227006911217, + "step": 2500 + }, + { + "epoch": 89.66, + "learning_rate": 3.931343283582089e-05, + "loss": 0.9962, + "step": 2600 + }, + { + "epoch": 93.1, + "learning_rate": 3.707462686567164e-05, + "loss": 0.9746, + "step": 2700 + }, + { + "epoch": 96.55, + "learning_rate": 3.483582089552238e-05, + "loss": 0.9592, + "step": 2800 + }, + { + "epoch": 100.0, + "learning_rate": 3.259701492537313e-05, + "loss": 0.9364, + "step": 2900 + }, + { + "epoch": 103.45, + "learning_rate": 3.0358208955223878e-05, + "loss": 0.9387, + "step": 3000 + }, + { + "epoch": 103.45, + "eval_loss": 0.4466392397880554, + "eval_runtime": 1.9017, + "eval_samples_per_second": 141.977, + "eval_steps_per_second": 4.733, + "eval_wer": 0.1721424774056353, + "step": 3000 + }, + { + "epoch": 106.9, + "learning_rate": 2.811940298507462e-05, + "loss": 0.9223, + "step": 3100 + }, + { + "epoch": 110.34, + "learning_rate": 2.588059701492537e-05, + "loss": 0.8839, + "step": 3200 + }, + { + "epoch": 113.79, + "learning_rate": 2.364179104477612e-05, + "loss": 0.8845, + "step": 3300 + }, + { + "epoch": 117.24, + "learning_rate": 2.1402985074626863e-05, + "loss": 0.8604, + "step": 3400 + }, + { + "epoch": 120.69, + "learning_rate": 1.9164179104477612e-05, + "loss": 0.8526, + "step": 3500 + }, + { + "epoch": 120.69, + "eval_loss": 0.43604278564453125, + "eval_runtime": 1.8975, + "eval_samples_per_second": 142.29, + "eval_steps_per_second": 4.743, + "eval_wer": 0.17044125465178098, + "step": 3500 + }, + { + "epoch": 124.14, + "learning_rate": 1.6925373134328355e-05, + "loss": 0.8311, + "step": 3600 + }, + { + "epoch": 127.59, + "learning_rate": 1.4686567164179104e-05, + "loss": 0.832, + "step": 3700 + }, + { + "epoch": 131.03, + "learning_rate": 1.244776119402985e-05, + "loss": 0.812, + "step": 3800 + }, + { + "epoch": 134.48, + "learning_rate": 1.0208955223880597e-05, + "loss": 0.8016, + "step": 3900 + }, + { + "epoch": 137.93, + "learning_rate": 7.970149253731343e-06, + "loss": 0.8018, + "step": 4000 + }, + { + "epoch": 137.93, + "eval_loss": 0.437362939119339, + "eval_runtime": 1.9635, + "eval_samples_per_second": 137.506, + "eval_steps_per_second": 4.584, + "eval_wer": 0.1720361509835194, + "step": 4000 + }, + { + "epoch": 141.38, + "learning_rate": 5.731343283582089e-06, + "loss": 0.7863, + "step": 4100 + }, + { + "epoch": 144.83, + "learning_rate": 3.4925373134328353e-06, + "loss": 0.7862, + "step": 4200 + }, + { + "epoch": 148.28, + "learning_rate": 1.2537313432835818e-06, + "loss": 0.7983, + "step": 4300 + }, + { + "epoch": 150.0, + "step": 4350, + "total_flos": 1.8721826713154617e+19, + "train_loss": 1.9741196327647943, + "train_runtime": 1808.1454, + "train_samples_per_second": 75.077, + "train_steps_per_second": 2.406 + } + ], + "max_steps": 4350, + "num_train_epochs": 150, + "total_flos": 1.8721826713154617e+19, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..90cacf509e616d5a83c4819d896dbf967f59e207 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6f38b03969c3e29b358a2037b4f61da5251f02d2d9436641d5a5d94788db60e +size 2927