|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.903694144216809, |
|
"eval_steps": 2000, |
|
"global_step": 6000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.3865763149403035e-05, |
|
"loss": 0.0452, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"eval_loss": 0.0001345932250842452, |
|
"eval_runtime": 53.5976, |
|
"eval_samples_per_second": 157.899, |
|
"eval_steps_per_second": 19.74, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.7731526298806067e-05, |
|
"loss": 0.0006, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"eval_loss": 0.00010070026473840699, |
|
"eval_runtime": 54.3295, |
|
"eval_samples_per_second": 155.772, |
|
"eval_steps_per_second": 19.474, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 1.5972894482090998e-06, |
|
"loss": 0.0004, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"eval_loss": 8.544776937924325e-05, |
|
"eval_runtime": 54.82, |
|
"eval_samples_per_second": 154.378, |
|
"eval_steps_per_second": 19.3, |
|
"step": 6000 |
|
} |
|
], |
|
"logging_steps": 2000, |
|
"max_steps": 6198, |
|
"num_train_epochs": 3, |
|
"save_steps": 2000, |
|
"total_flos": 5.754652065792e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|