|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"global_step": 1638, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.544097693351425e-05, |
|
"loss": 0.9985, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.656716417910447e-06, |
|
"loss": 0.7741, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 1.8724559023066488e-06, |
|
"loss": 0.7072, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 0.9316399097442627, |
|
"eval_runtime": 2.9494, |
|
"eval_samples_per_second": 71.541, |
|
"step": 1638 |
|
} |
|
], |
|
"max_steps": 1638, |
|
"num_train_epochs": 1, |
|
"total_flos": 1.006879961366784e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|