nohup: ignoring input [2023-02-21 19:51:35,707] [WARNING] [runner.py:186:fetch_hostfile] Unable to find hostfile, will proceed with training with local resources only. [2023-02-21 19:51:35,765] [INFO] [runner.py:548:main] cmd = /opt/conda/bin/python3 -u -m deepspeed.launcher.launch --world_info=eyJsb2NhbGhvc3QiOiBbMCwgMSwgMiwgMywgNCwgNSwgNiwgN119 --master_addr=127.0.0.1 --master_port=29500 --enable_each_rank_log=None tune_gpt.py --deepspeed deepspeed.json --upload-experiment /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index [2023-02-21 19:51:38,485] [INFO] [launch.py:135:main] 0 NCCL_VERSION=2.11.4 [2023-02-21 19:51:38,485] [INFO] [launch.py:142:main] WORLD INFO DICT: {'localhost': [0, 1, 2, 3, 4, 5, 6, 7]} [2023-02-21 19:51:38,485] [INFO] [launch.py:148:main] nnodes=1, num_local_procs=8, node_rank=0 [2023-02-21 19:51:38,485] [INFO] [launch.py:161:main] global_rank_mapping=defaultdict(, {'localhost': [0, 1, 2, 3, 4, 5, 6, 7]}) [2023-02-21 19:51:38,485] [INFO] [launch.py:162:main] dist_world_size=8 [2023-02-21 19:51:38,485] [INFO] [launch.py:164:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) No config specified, defaulting to: apps/all Found cached dataset apps (/home/.cache/huggingface/datasets/codeparrot___apps/all/0.0.0/04ac807715d07d6e5cc580f59cdc8213cd7dc4529d0bb819cca72c9f8e8c1aa5) Max length: 2048 PyTorch: setting up devices [2023-02-21 19:51:59,246] [INFO] [comm.py:657:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl Max length: 2048 PyTorch: setting up devices Max length: 2048 PyTorch: setting up devices Max length: 2048 PyTorch: setting up devices Max length: 2048Max length: 2048 PyTorch: setting up devices PyTorch: setting up devices Max length: 2048 PyTorch: setting up devices Max length: 2048 PyTorch: setting up devices The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). The default value for the training argument `--report_to` will change in v5 (from all installed integrations to none). In v5, you will need to use `--report_to all` to get the same behavior as now. You should start updating your code and make this info disappear :-). GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. [2023-02-21 19:51:59,917] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed info: version=0.8.1, git-hash=unknown, git-branch=unknown GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. GPU memory occupied: 7801 MB. [2023-02-21 19:52:02,386] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed Flops Profiler Enabled: False Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Installed CUDA version 11.6 does not match the version torch was compiled with 11.7 but since the APIs are compatible, accepting this combination Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Detected CUDA files, patching ldflags Emitting ninja build file /home/.cache/torch_extensions/py38_cu117/cpu_adam/build.ninja... Building extension module cpu_adam... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... ninja: no work to do. Loading extension module cpu_adam... Time to load cpu_adam op: 3.051539182662964 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.8587288856506348 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.8896634578704834 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.8358540534973145 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.7415549755096436 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.8722808361053467 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 2.8915891647338867 seconds Loading extension module cpu_adam... Time to load cpu_adam op: 3.0029048919677734 seconds Adam Optimizer #0 is created with AVX512 arithmetic capability. Config: alpha=0.000100, betas=(0.900000, 0.999000), weight_decay=0.100000, adam_w=1 [2023-02-21 19:52:09,340] [INFO] [logging.py:75:log_dist] [Rank 0] Using DeepSpeed Optimizer param name adamw as basic optimizer [2023-02-21 19:52:09,347] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed Basic Optimizer = DeepSpeedCPUAdam [2023-02-21 19:52:09,347] [INFO] [utils.py:53:is_zero_supported_optimizer] Checking ZeRO support for optimizer=DeepSpeedCPUAdam type= [2023-02-21 19:52:09,347] [INFO] [logging.py:75:log_dist] [Rank 0] Creating torch.float32 ZeRO stage 2 optimizer [2023-02-21 19:52:09,348] [INFO] [stage_1_and_2.py:144:__init__] Reduce bucket size 500000000 [2023-02-21 19:52:09,348] [INFO] [stage_1_and_2.py:145:__init__] Allgather bucket size 500000000 [2023-02-21 19:52:09,348] [INFO] [stage_1_and_2.py:146:__init__] CPU Offload: True [2023-02-21 19:52:09,348] [INFO] [stage_1_and_2.py:147:__init__] Round robin gradient partitioning: False Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Emitting ninja build file /home/.cache/torch_extensions/py38_cu117/utils/build.ninja... Building extension module utils... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) ninja: no work to do. Loading extension module utils... Time to load utils op: 0.5143241882324219 seconds Loading extension module utils... Time to load utils op: 0.4041626453399658 seconds Loading extension module utils... Time to load utils op: 0.40406346321105957 seconds Loading extension module utils... Time to load utils op: 0.40425753593444824 seconds Loading extension module utils... Time to load utils op: 0.40483736991882324 seconds Loading extension module utils... Loading extension module utils... Time to load utils op: 0.4042036533355713 seconds Time to load utils op: 0.4030454158782959 seconds Loading extension module utils... Time to load utils op: 0.4040055274963379 seconds Rank: 5 partition count [8] and sizes[(15649824, False)] Rank: 0 partition count [8] and sizes[(15649824, False)] Rank: 2 partition count [8] and sizes[(15649824, False)] Rank: 3 partition count [8] and sizes[(15649824, False)] Rank: 4 partition count [8] and sizes[(15649824, False)] Rank: 6 partition count [8] and sizes[(15649824, False)] Rank: 1 partition count [8] and sizes[(15649824, False)] Rank: 7 partition count [8] and sizes[(15649824, False)] Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.0007538795471191406 seconds [2023-02-21 19:52:14,395] [INFO] [utils.py:825:see_memory_usage] Before initializing optimizer states [2023-02-21 19:52:14,405] [INFO] [utils.py:826:see_memory_usage] MA 0.66 GB Max_MA 0.66 GB CA 0.85 GB Max_CA 1 GB [2023-02-21 19:52:14,405] [INFO] [utils.py:834:see_memory_usage] CPU Virtual Memory: used = 42.5 GB, percent = 16.9% Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.005951404571533203 seconds You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.0007529258728027344 seconds No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.020640850067138672 seconds Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.0007100105285644531 seconds Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.0004687309265136719 seconds Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.0029997825622558594 seconds You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. You're using a GPT2TokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. [2023-02-21 19:52:14,635] [INFO] [utils.py:825:see_memory_usage] After initializing optimizer states [2023-02-21 19:52:14,635] [INFO] [utils.py:826:see_memory_usage] MA 0.66 GB Max_MA 0.66 GB CA 0.85 GB Max_CA 1 GB [2023-02-21 19:52:14,636] [INFO] [utils.py:834:see_memory_usage] CPU Virtual Memory: used = 42.46 GB, percent = 16.9% [2023-02-21 19:52:14,636] [INFO] [stage_1_and_2.py:527:__init__] optimizer state initialized [2023-02-21 19:52:14,719] [INFO] [utils.py:825:see_memory_usage] After initializing ZeRO optimizer [2023-02-21 19:52:14,720] [INFO] [utils.py:826:see_memory_usage] MA 0.66 GB Max_MA 0.66 GB CA 0.85 GB Max_CA 1 GB [2023-02-21 19:52:14,720] [INFO] [utils.py:834:see_memory_usage] CPU Virtual Memory: used = 42.46 GB, percent = 16.9% [2023-02-21 19:52:14,721] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed Final Optimizer = adamw [2023-02-21 19:52:14,721] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed using configured LR scheduler = WarmupLR [2023-02-21 19:52:14,721] [INFO] [logging.py:75:log_dist] [Rank 0] DeepSpeed LR Scheduler = [2023-02-21 19:52:14,721] [INFO] [logging.py:75:log_dist] [Rank 0] step=0, skipped=0, lr=[0.0001], mom=[[0.9, 0.999]] [2023-02-21 19:52:14,722] [INFO] [config.py:1009:print] DeepSpeedEngine configuration: [2023-02-21 19:52:14,722] [INFO] [config.py:1013:print] activation_checkpointing_config { "partition_activations": false, "contiguous_memory_optimization": false, "cpu_checkpointing": false, "number_checkpoints": null, "synchronize_checkpoint_boundary": false, "profile": false } [2023-02-21 19:52:14,722] [INFO] [config.py:1013:print] aio_config ................... {'block_size': 1048576, 'queue_depth': 8, 'thread_count': 1, 'single_submit': False, 'overlap_events': True} [2023-02-21 19:52:14,722] [INFO] [config.py:1013:print] amp_enabled .................. False [2023-02-21 19:52:14,722] [INFO] [config.py:1013:print] amp_params ................... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] autotuning_config ............ { "enabled": false, "start_step": null, "end_step": null, "metric_path": null, "arg_mappings": null, "metric": "throughput", "model_info": null, "results_dir": "autotuning_results", "exps_dir": "autotuning_exps", "overwrite": true, "fast": true, "start_profile_step": 3, "end_profile_step": 5, "tuner_type": "gridsearch", "tuner_early_stopping": 5, "tuner_num_trials": 50, "model_info_path": null, "mp_size": 1, "max_train_batch_size": null, "min_train_batch_size": 1, "max_train_micro_batch_size_per_gpu": 1.024000e+03, "min_train_micro_batch_size_per_gpu": 1, "num_tuning_micro_batch_sizes": 3 } [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] bfloat16_enabled ............. False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] checkpoint_parallel_write_pipeline False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] checkpoint_tag_validation_enabled True [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] checkpoint_tag_validation_fail False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] comms_config ................. [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] communication_data_type ...... None [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] compression_config ........... {'weight_quantization': {'shared_parameters': {'enabled': False, 'quantizer_kernel': False, 'schedule_offset': 0, 'quantize_groups': 1, 'quantize_verbose': False, 'quantization_type': 'symmetric', 'quantize_weight_in_forward': False, 'rounding': 'nearest', 'fp16_mixed_quantize': False, 'quantize_change_ratio': 0.001}, 'different_groups': {}}, 'activation_quantization': {'shared_parameters': {'enabled': False, 'quantization_type': 'symmetric', 'range_calibration': 'dynamic', 'schedule_offset': 1000}, 'different_groups': {}}, 'sparse_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'row_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'head_pruning': {'shared_parameters': {'enabled': False, 'method': 'topk', 'schedule_offset': 1000}, 'different_groups': {}}, 'channel_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'layer_reduction': {'enabled': False}} [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] curriculum_enabled_legacy .... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] curriculum_params_legacy ..... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] data_efficiency_config ....... {'enabled': False, 'seed': 1234, 'data_sampling': {'enabled': False, 'num_epochs': 1000, 'num_workers': 0, 'curriculum_learning': {'enabled': False}}, 'data_routing': {'enabled': False, 'random_ltd': {'enabled': False, 'layer_token_lr_schedule': {'enabled': False}}}} [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] data_efficiency_enabled ...... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] dataloader_drop_last ......... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] disable_allgather ............ False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] dump_state ................... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] dynamic_loss_scale_args ...... None [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_enabled ........... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_gas_boundary_resolution 1 [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_layer_name ........ bert.encoder.layer [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_layer_num ......... 0 [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_max_iter .......... 100 [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_stability ......... 1e-06 [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_tol ............... 0.01 [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] eigenvalue_verbose ........... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] elasticity_enabled ........... False [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] flops_profiler_config ........ { "enabled": false, "profile_step": 1, "module_depth": -1, "top_modules": 1, "detailed": true, "output_file": null } [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] fp16_auto_cast ............... None [2023-02-21 19:52:14,723] [INFO] [config.py:1013:print] fp16_enabled ................. False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] fp16_master_weights_and_gradients False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] global_rank .................. 0 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] grad_accum_dtype ............. None [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] gradient_accumulation_steps .. 4 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] gradient_clipping ............ 1.0 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] gradient_predivide_factor .... 1.0 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] initial_dynamic_scale ........ 65536 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] load_universal_checkpoint .... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] loss_scale ................... 0 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] memory_breakdown ............. False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] monitor_config ............... tensorboard=TensorBoardConfig(enabled=True, output_path='logs/', job_name='train_neo') wandb=WandbConfig(enabled=False, group=None, team=None, project='deepspeed') csv_monitor=CSVConfig(enabled=False, output_path='', job_name='DeepSpeedJobName') enabled=True [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] nebula_config ................ { "enabled": false, "persistent_storage_path": null, "persistent_time_interval": 100, "num_of_version_in_retention": 2, "enable_nebula_load": true, "load_path": null } [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] optimizer_legacy_fusion ...... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] optimizer_name ............... adamw [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] optimizer_params ............. {'lr': 0.0001, 'betas': [0.9, 0.999], 'eps': 1e-08, 'weight_decay': 0.1} [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] pipeline ..................... {'stages': 'auto', 'partition': 'best', 'seed_layers': False, 'activation_checkpoint_interval': 0} [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] pld_enabled .................. False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] pld_params ................... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] prescale_gradients ........... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] scheduler_name ............... WarmupLR [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] scheduler_params ............. {'warmup_min_lr': 1e-09, 'warmup_max_lr': 0.0001, 'warmup_num_steps': 1000} [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] sparse_attention ............. None [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] sparse_gradients_enabled ..... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] steps_per_print .............. 2000 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] train_batch_size ............. 192 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] train_micro_batch_size_per_gpu 6 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] use_node_local_storage ....... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] wall_clock_breakdown ......... False [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] world_size ................... 8 [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] zero_allow_untested_optimizer True [2023-02-21 19:52:14,724] [INFO] [config.py:1013:print] zero_config .................. stage=2 contiguous_gradients=True reduce_scatter=True reduce_bucket_size=500000000 allgather_partitions=True allgather_bucket_size=500000000 overlap_comm=True load_from_fp32_weights=True elastic_checkpoint=False offload_param=None offload_optimizer=DeepSpeedZeroOffloadOptimizerConfig(device='cpu', nvme_path=None, buffer_count=4, pin_memory=True, pipeline=False, pipeline_read=False, pipeline_write=False, fast_init=False) sub_group_size=1,000,000,000 cpu_offload_param=None cpu_offload_use_pin_memory=None cpu_offload=None prefetch_bucket_size=50,000,000 param_persistence_threshold=100,000 model_persistence_threshold=sys.maxsize max_live_parameters=1,000,000,000 max_reuse_distance=1,000,000,000 gather_16bit_weights_on_model_save=False stage3_gather_fp16_weights_on_model_save=False ignore_unused_parameters=True legacy_stage1=False round_robin_gradients=False [2023-02-21 19:52:14,725] [INFO] [config.py:1013:print] zero_enabled ................. True [2023-02-21 19:52:14,725] [INFO] [config.py:1013:print] zero_optimization_stage ...... 2 [2023-02-21 19:52:14,725] [INFO] [config.py:998:print_user_config] json = { "optimizer": { "type": "AdamW", "params": { "lr": 0.0001, "betas": [0.9, 0.999], "eps": 1e-08, "weight_decay": 0.1 } }, "scheduler": { "type": "WarmupLR", "params": { "warmup_min_lr": 1e-09, "warmup_max_lr": 0.0001, "warmup_num_steps": 1000 } }, "zero_optimization": { "stage": 2, "offload_optimizer": { "device": "cpu", "pin_memory": true }, "allgather_partitions": true, "allgather_bucket_size": 5.000000e+08, "overlap_comm": true, "reduce_scatter": true, "reduce_bucket_size": 5.000000e+08, "contiguous_gradients": true }, "tensorboard": { "enabled": true, "output_path": "logs/", "job_name": "train_neo" }, "zero_allow_untested_optimizer": true, "gradient_accumulation_steps": 4, "gradient_clipping": 1.0, "steps_per_print": 2.000000e+03, "train_batch_size": 192, "train_micro_batch_size_per_gpu": 6, "wall_clock_breakdown": false } Using /home/.cache/torch_extensions/py38_cu117 as PyTorch extensions root... No modifications detected for re-loaded extension module utils, skipping build step... Loading extension module utils... Time to load utils op: 0.00039505958557128906 seconds ***** Running training ***** Num examples = 117232 Num Epochs = 10 Instantaneous batch size per device = 6 Total train batch size (w. parallel, distributed & accumulation) = 192 Gradient Accumulation steps = 4 Total optimization steps = 6100 Number of trainable parameters = 125198592 0%| | 0/6100 [00:00 trainer.save_state(trainer_save_dir) TypeError: save_state() takes 1 positional argument but 2 were given [2023-02-22 05:05:17,992] [INFO] [launch.py:350:main] Process 11659 exits successfully. [2023-02-22 05:05:17,993] [INFO] [launch.py:350:main] Process 11661 exits successfully. [2023-02-22 05:05:17,994] [INFO] [launch.py:350:main] Process 11655 exits successfully. [2023-02-22 05:05:18,995] [INFO] [launch.py:350:main] Process 11654 exits successfully. [2023-02-22 05:05:18,996] [INFO] [launch.py:350:main] Process 11656 exits successfully. [2023-02-22 05:05:18,996] [INFO] [launch.py:350:main] Process 11653 exits successfully. [2023-02-22 05:05:18,997] [INFO] [launch.py:350:main] Process 11657 exits successfully. [2023-02-22 05:05:19,998] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11652 [2023-02-22 05:05:20,000] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11653 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11654 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11655 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11656 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11657 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11659 [2023-02-22 05:05:20,001] [INFO] [launch.py:318:sigkill_handler] Killing subprocess 11661 [2023-02-22 05:05:20,002] [ERROR] [launch.py:324:sigkill_handler] ['/opt/conda/bin/python3', '-u', 'tune_gpt.py', '--local_rank=7', '--deepspeed', 'deepspeed.json', '--upload-experiment'] exits with return code = 1 /opt/conda/lib/python3.8/site-packages/xgboost/compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead. from pandas import MultiIndex, Int64Index