|
{ |
|
"best_metric": 0.7866666666666666, |
|
"best_model_checkpoint": "Bert_tinybert-distilled/run-5/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": 0.9512056112289429, |
|
"learning_rate": 0.0001860785294707651, |
|
"loss": 0.5131, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_AUC": 0.835936, |
|
"eval_BAC": 0.7566666666666666, |
|
"eval_MCC": 0.513333789630238, |
|
"eval_PR_AUC": 0.7961733871717276, |
|
"eval_accuracy": 0.7566666666666667, |
|
"eval_loss": 0.45322349667549133, |
|
"eval_runtime": 6.3408, |
|
"eval_samples_per_second": 236.562, |
|
"eval_steps_per_second": 7.412, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 1.6722192764282227, |
|
"learning_rate": 0.00015949588240351295, |
|
"loss": 0.4704, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_AUC": 0.8548942222222222, |
|
"eval_BAC": 0.7866666666666666, |
|
"eval_MCC": 0.596741247380097, |
|
"eval_PR_AUC": 0.8018353018281653, |
|
"eval_accuracy": 0.7866666666666666, |
|
"eval_loss": 0.43870097398757935, |
|
"eval_runtime": 6.2585, |
|
"eval_samples_per_second": 239.673, |
|
"eval_steps_per_second": 7.51, |
|
"step": 344 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 1376, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 8, |
|
"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": 0.7927861374990359, |
|
"learning_rate": 0.00021266117653801728, |
|
"num_train_epochs": 8, |
|
"temperature": 5 |
|
} |
|
} |
|
|