|
{ |
|
"best_metric": 0.6925490196078431, |
|
"best_model_checkpoint": "tiny-bert-sst2-distilled/run-23/checkpoint-320", |
|
"epoch": 2.0, |
|
"eval_steps": 500, |
|
"global_step": 320, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 1.446049451828003, |
|
"learning_rate": 0.0005691296983851737, |
|
"loss": 0.6049, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.6666666666666666, |
|
"eval_f1": 0.0, |
|
"eval_loss": 0.5720294713973999, |
|
"eval_mcc": 0.0, |
|
"eval_precision": 0.0, |
|
"eval_recall": 0.0, |
|
"eval_runtime": 1.9517, |
|
"eval_samples_per_second": 653.261, |
|
"eval_steps_per_second": 20.494, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 1.206958293914795, |
|
"learning_rate": 0.00047427474865431146, |
|
"loss": 0.5748, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.6925490196078431, |
|
"eval_f1": 0.312280701754386, |
|
"eval_loss": 0.5560094714164734, |
|
"eval_mcc": 0.21311872533733642, |
|
"eval_precision": 0.6137931034482759, |
|
"eval_recall": 0.20941176470588235, |
|
"eval_runtime": 1.9268, |
|
"eval_samples_per_second": 661.712, |
|
"eval_steps_per_second": 20.76, |
|
"step": 320 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 1120, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 7, |
|
"save_steps": 500, |
|
"total_flos": 583510875840.0, |
|
"train_batch_size": 32, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.9539104497371855, |
|
"learning_rate": 0.000663984648116036, |
|
"num_train_epochs": 7, |
|
"temperature": 6 |
|
} |
|
} |
|
|