|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 324, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3e-05, |
|
"loss": 1.0211, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_exact_match": 72.6477789436866, |
|
"eval_f1": 75.7148640277524, |
|
"eval_runtime": 8.2879, |
|
"eval_samples_per_second": 1034.884, |
|
"eval_steps_per_second": 3.258, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 3e-05, |
|
"loss": 0.4627, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_exact_match": 71.69173370642416, |
|
"eval_f1": 75.8528410498886, |
|
"eval_runtime": 7.4899, |
|
"eval_samples_per_second": 1145.143, |
|
"eval_steps_per_second": 3.605, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 3e-05, |
|
"loss": 0.4159, |
|
"step": 324 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_exact_match": 73.55718782791186, |
|
"eval_f1": 76.3608510722886, |
|
"eval_runtime": 7.4798, |
|
"eval_samples_per_second": 1146.692, |
|
"eval_steps_per_second": 3.61, |
|
"step": 324 |
|
} |
|
], |
|
"max_steps": 2160, |
|
"num_train_epochs": 20, |
|
"total_flos": 102029926072320.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|