|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 272886, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.3333333333333333e-05, |
|
"loss": 0.4574, |
|
"step": 90962 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.869403907989117, |
|
"eval_combined_score": 0.8495626351091097, |
|
"eval_f1": 0.8297213622291022, |
|
"eval_loss": 0.4945518374443054, |
|
"eval_runtime": 876.4633, |
|
"eval_samples_per_second": 46.129, |
|
"eval_steps_per_second": 5.766, |
|
"step": 90962 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 6.666666666666667e-06, |
|
"loss": 0.3387, |
|
"step": 181924 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.8873608706406134, |
|
"eval_combined_score": 0.8655189633993949, |
|
"eval_f1": 0.8436770561581766, |
|
"eval_loss": 0.47449207305908203, |
|
"eval_runtime": 876.5998, |
|
"eval_samples_per_second": 46.121, |
|
"eval_steps_per_second": 5.765, |
|
"step": 181924 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 0.0, |
|
"loss": 0.2029, |
|
"step": 272886 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.8943111550828593, |
|
"eval_combined_score": 0.8749838381906883, |
|
"eval_f1": 0.8556565212985171, |
|
"eval_loss": 0.5514675378799438, |
|
"eval_runtime": 879.9699, |
|
"eval_samples_per_second": 45.945, |
|
"eval_steps_per_second": 5.743, |
|
"step": 272886 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 272886, |
|
"total_flos": 6.828837458282865e+17, |
|
"train_loss": 0.33302516177772035, |
|
"train_runtime": 80424.3343, |
|
"train_samples_per_second": 13.572, |
|
"train_steps_per_second": 3.393 |
|
} |
|
], |
|
"max_steps": 272886, |
|
"num_train_epochs": 3, |
|
"total_flos": 6.828837458282865e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|