|
{ |
|
"best_metric": 0.7533333333333333, |
|
"best_model_checkpoint": "Bert_tinybert-distilled/run-18/checkpoint-172", |
|
"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": 0.5358241200447083, |
|
"learning_rate": 0.00015590369179509114, |
|
"loss": 0.3755, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_AUC": 0.8250026666666667, |
|
"eval_BAC": 0.7533333333333333, |
|
"eval_MCC": 0.5072513572061289, |
|
"eval_PR_AUC": 0.7777873442469706, |
|
"eval_accuracy": 0.7533333333333333, |
|
"eval_loss": 0.34898021817207336, |
|
"eval_runtime": 6.2992, |
|
"eval_samples_per_second": 238.127, |
|
"eval_steps_per_second": 7.461, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 0.8845176100730896, |
|
"learning_rate": 0.00013641573032070476, |
|
"loss": 0.3585, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_AUC": 0.8418488888888889, |
|
"eval_BAC": 0.7506666666666666, |
|
"eval_MCC": 0.5416697930019726, |
|
"eval_PR_AUC": 0.8061916006845603, |
|
"eval_accuracy": 0.7506666666666667, |
|
"eval_loss": 0.3457224667072296, |
|
"eval_runtime": 6.2869, |
|
"eval_samples_per_second": 238.591, |
|
"eval_steps_per_second": 7.476, |
|
"step": 344 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 1548, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 9, |
|
"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": 0.5621897712480096, |
|
"learning_rate": 0.00017539165326947755, |
|
"num_train_epochs": 9, |
|
"temperature": 10 |
|
} |
|
} |
|
|