|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 5.0, |
|
"global_step": 32095, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 8.889735156655704e-06, |
|
"loss": 7.3112, |
|
"step": 6419 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 6.181427955627441, |
|
"eval_runtime": 64.709, |
|
"eval_samples_per_second": 198.396, |
|
"eval_steps_per_second": 9.921, |
|
"step": 6419 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 6.668513069066991e-06, |
|
"loss": 5.8524, |
|
"step": 12838 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 5.407498359680176, |
|
"eval_runtime": 64.7053, |
|
"eval_samples_per_second": 198.407, |
|
"eval_steps_per_second": 9.922, |
|
"step": 12838 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 4.446944781028216e-06, |
|
"loss": 5.3392, |
|
"step": 19257 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_loss": 5.080959320068359, |
|
"eval_runtime": 64.6173, |
|
"eval_samples_per_second": 198.677, |
|
"eval_steps_per_second": 9.935, |
|
"step": 19257 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 2.2257226934395014e-06, |
|
"loss": 5.0958, |
|
"step": 25676 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_loss": 4.901454448699951, |
|
"eval_runtime": 64.6233, |
|
"eval_samples_per_second": 198.659, |
|
"eval_steps_per_second": 9.934, |
|
"step": 25676 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 4.154405400727021e-09, |
|
"loss": 4.9897, |
|
"step": 32095 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_loss": 4.8497138023376465, |
|
"eval_runtime": 64.6531, |
|
"eval_samples_per_second": 198.568, |
|
"eval_steps_per_second": 9.93, |
|
"step": 32095 |
|
} |
|
], |
|
"max_steps": 32095, |
|
"num_train_epochs": 5, |
|
"total_flos": 4.22385209573376e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|