|
{ |
|
"best_metric": 0.7773333333333333, |
|
"best_model_checkpoint": "Bert_tinybert-distilled/run-23/checkpoint-344", |
|
"epoch": 2.0, |
|
"eval_steps": 500, |
|
"global_step": 344, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 1.0031909942626953, |
|
"learning_rate": 0.0002164041823853993, |
|
"loss": 0.5574, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_AUC": 0.8248408888888888, |
|
"eval_BAC": 0.736, |
|
"eval_MCC": 0.4730523981849912, |
|
"eval_PR_AUC": 0.781847757181041, |
|
"eval_accuracy": 0.736, |
|
"eval_loss": 0.49910596013069153, |
|
"eval_runtime": 6.2575, |
|
"eval_samples_per_second": 239.711, |
|
"eval_steps_per_second": 7.511, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 2.0278003215789795, |
|
"learning_rate": 0.00018033681865449943, |
|
"loss": 0.5145, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_AUC": 0.8580764444444444, |
|
"eval_BAC": 0.7773333333333333, |
|
"eval_MCC": 0.582222053555848, |
|
"eval_PR_AUC": 0.8188026202087335, |
|
"eval_accuracy": 0.7773333333333333, |
|
"eval_loss": 0.4729117155075073, |
|
"eval_runtime": 6.2762, |
|
"eval_samples_per_second": 238.996, |
|
"eval_steps_per_second": 7.489, |
|
"step": 344 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 1204, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 7, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": false |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 900755460000.0, |
|
"train_batch_size": 32, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.868343761378926, |
|
"learning_rate": 0.0002524715461162992, |
|
"num_train_epochs": 7, |
|
"temperature": 5 |
|
} |
|
} |
|
|