|
{ |
|
"best_metric": 73.85502277098726, |
|
"best_model_checkpoint": "outputs/banglabert/checkpoint-348", |
|
"epoch": 2.9941081949651847, |
|
"global_step": 348, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.4285714285714286e-06, |
|
"loss": 6.0104, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 3.7060702875399364e-05, |
|
"loss": 2.8926, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"eval_HasAns_exact": 54.036770583533176, |
|
"eval_HasAns_f1": 67.53514335825187, |
|
"eval_HasAns_total": 1251, |
|
"eval_NoAns_exact": 77.61790567545964, |
|
"eval_NoAns_f1": 77.61790567545964, |
|
"eval_NoAns_total": 1251, |
|
"eval_best_exact": 65.78737010391687, |
|
"eval_best_exact_thresh": 0.0, |
|
"eval_best_f1": 72.53655649127667, |
|
"eval_best_f1_thresh": 0.0, |
|
"eval_exact": 65.8273381294964, |
|
"eval_f1": 72.5765245168559, |
|
"eval_total": 2502, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.8530351437699682e-05, |
|
"loss": 1.5349, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"eval_HasAns_exact": 57.95363709032774, |
|
"eval_HasAns_f1": 71.37777157674782, |
|
"eval_HasAns_total": 1251, |
|
"eval_NoAns_exact": 74.74020783373301, |
|
"eval_NoAns_f1": 74.74020783373301, |
|
"eval_NoAns_total": 1251, |
|
"eval_best_exact": 66.30695443645084, |
|
"eval_best_exact_thresh": 0.0, |
|
"eval_best_f1": 73.01902167966134, |
|
"eval_best_f1_thresh": 0.0, |
|
"eval_exact": 66.34692246203038, |
|
"eval_f1": 73.05898970524055, |
|
"eval_total": 2502, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 0.0, |
|
"loss": 1.3596, |
|
"step": 348 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"eval_HasAns_exact": 58.113509192645886, |
|
"eval_HasAns_f1": 71.8507329920142, |
|
"eval_HasAns_total": 1251, |
|
"eval_NoAns_exact": 75.85931254996004, |
|
"eval_NoAns_f1": 75.85931254996004, |
|
"eval_NoAns_total": 1251, |
|
"eval_best_exact": 66.94644284572342, |
|
"eval_best_exact_thresh": 0.0, |
|
"eval_best_f1": 73.81505474540803, |
|
"eval_best_f1_thresh": 0.0, |
|
"eval_exact": 66.98641087130295, |
|
"eval_f1": 73.85502277098726, |
|
"eval_total": 2502, |
|
"step": 348 |
|
} |
|
], |
|
"max_steps": 348, |
|
"num_train_epochs": 3, |
|
"total_flos": 7.010369880983654e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|