|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.999745123027154, |
|
"global_step": 9561, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 0.0001, |
|
"loss": 2.5416, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4758, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4544, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4235, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4206, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4084, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3928, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3703, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3678, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3605, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3572, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3473, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3415, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3158, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3191, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3028, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3086, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3124, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 2.98, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3034, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 9561, |
|
"total_flos": 1.5154927442907955e+17, |
|
"train_loss": 2.374575984090233, |
|
"train_runtime": 14184.3644, |
|
"train_samples_per_second": 43.15, |
|
"train_steps_per_second": 0.674 |
|
} |
|
], |
|
"max_steps": 9561, |
|
"num_train_epochs": 3, |
|
"total_flos": 1.5154927442907955e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|