|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 3723, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.32849852269675e-05, |
|
"loss": 3.6989, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 3.6569970453935e-05, |
|
"loss": 3.0253, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 2.9854955680902503e-05, |
|
"loss": 2.7328, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 2.313994090787e-05, |
|
"loss": 2.5327, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 1.6424926134837498e-05, |
|
"loss": 2.4709, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 9.709911361804996e-06, |
|
"loss": 2.1255, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 2.9948965887724956e-06, |
|
"loss": 2.078, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 3723, |
|
"total_flos": 873975932264448.0, |
|
"train_loss": 2.628729448054639, |
|
"train_runtime": 573.403, |
|
"train_samples_per_second": 25.966, |
|
"train_steps_per_second": 6.493 |
|
} |
|
], |
|
"max_steps": 3723, |
|
"num_train_epochs": 3, |
|
"total_flos": 873975932264448.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|