| { | |
| "best_metric": 0.09755547344684601, | |
| "best_model_checkpoint": "dslim/bert-base-NER-model/checkpoint-4390", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 4390, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.22779043280182232, | |
| "grad_norm": 1.933945894241333, | |
| "learning_rate": 1.7722095671981778e-05, | |
| "loss": 0.6767, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45558086560364464, | |
| "grad_norm": 2.780022144317627, | |
| "learning_rate": 1.5444191343963555e-05, | |
| "loss": 0.2153, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.683371298405467, | |
| "grad_norm": 1.4999059438705444, | |
| "learning_rate": 1.3166287015945332e-05, | |
| "loss": 0.1472, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9111617312072893, | |
| "grad_norm": 0.7693917155265808, | |
| "learning_rate": 1.0888382687927108e-05, | |
| "loss": 0.1308, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9589862807506537, | |
| "eval_f1": 0.8543397844850693, | |
| "eval_loss": 0.1143401563167572, | |
| "eval_precision": 0.8349575605155611, | |
| "eval_recall": 0.8746432491767289, | |
| "eval_runtime": 105.1757, | |
| "eval_samples_per_second": 83.451, | |
| "eval_steps_per_second": 5.22, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 1.1389521640091116, | |
| "grad_norm": 0.9619026184082031, | |
| "learning_rate": 8.610478359908885e-06, | |
| "loss": 0.1068, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.366742596810934, | |
| "grad_norm": 0.6502276659011841, | |
| "learning_rate": 6.3325740318906616e-06, | |
| "loss": 0.1, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.5945330296127562, | |
| "grad_norm": 0.8502880930900574, | |
| "learning_rate": 4.054669703872437e-06, | |
| "loss": 0.0963, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.8223234624145785, | |
| "grad_norm": 1.196458101272583, | |
| "learning_rate": 1.7767653758542143e-06, | |
| "loss": 0.0965, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.963877173871149, | |
| "eval_f1": 0.8823518854209665, | |
| "eval_loss": 0.09755547344684601, | |
| "eval_precision": 0.8659245429245781, | |
| "eval_recall": 0.8994145627515551, | |
| "eval_runtime": 97.9073, | |
| "eval_samples_per_second": 89.646, | |
| "eval_steps_per_second": 5.607, | |
| "step": 4390 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 4390, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8407997307385200.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |