|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.0, |
|
"global_step": 4000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.99875e-05, |
|
"loss": 9.7964, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 4.375e-05, |
|
"loss": 0.7863, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 3.7500000000000003e-05, |
|
"loss": 0.4898, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.125e-05, |
|
"loss": 0.3452, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 2.5e-05, |
|
"loss": 0.3034, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_cer": 0.04716666666666667, |
|
"eval_loss": 0.2454013228416443, |
|
"eval_runtime": 671.7272, |
|
"eval_samples_per_second": 5.955, |
|
"eval_steps_per_second": 0.744, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 1.8750000000000002e-05, |
|
"loss": 0.2094, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 1.25e-05, |
|
"loss": 0.1765, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 6.25e-06, |
|
"loss": 0.1795, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 0.0, |
|
"loss": 0.1451, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_cer": 0.037, |
|
"eval_loss": 0.1549716591835022, |
|
"eval_runtime": 683.2682, |
|
"eval_samples_per_second": 5.854, |
|
"eval_steps_per_second": 0.732, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"step": 4000, |
|
"total_flos": 2.3945126111870976e+19, |
|
"train_loss": 0.3316297147274017, |
|
"train_runtime": 5518.3562, |
|
"train_samples_per_second": 5.799, |
|
"train_steps_per_second": 0.725 |
|
} |
|
], |
|
"max_steps": 4000, |
|
"num_train_epochs": 2, |
|
"total_flos": 2.3945126111870976e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|