|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 1458, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 1.9986282578875174e-05, |
|
"loss": 2.1431, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 1.3141289437585736e-05, |
|
"loss": 1.061, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 6.282578875171468e-06, |
|
"loss": 0.6752, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 1458, |
|
"total_flos": 8.661283434292838e+16, |
|
"train_loss": 0.7548384267443329, |
|
"train_runtime": 6087.8768, |
|
"train_samples_per_second": 1.916, |
|
"train_steps_per_second": 0.239 |
|
} |
|
], |
|
"max_steps": 1458, |
|
"num_train_epochs": 3, |
|
"total_flos": 8.661283434292838e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|