|
{ |
|
"best_metric": 0.31137627363204956, |
|
"best_model_checkpoint": "gal_enptsp_mBERT/checkpoint-750", |
|
"epoch": 4.0, |
|
"global_step": 1000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.8996258586849852, |
|
"eval_f1": 0.6356504001255296, |
|
"eval_loss": 0.3629606366157532, |
|
"eval_precision": 0.6960481099656357, |
|
"eval_recall": 0.584897487727404, |
|
"eval_runtime": 5.3665, |
|
"eval_samples_per_second": 185.782, |
|
"eval_steps_per_second": 11.739, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 1.9600000000000002e-05, |
|
"loss": 0.4395, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.9107580961727183, |
|
"eval_f1": 0.6848413974565122, |
|
"eval_loss": 0.3252100646495819, |
|
"eval_precision": 0.6934576672587329, |
|
"eval_recall": 0.6764366156511695, |
|
"eval_runtime": 5.417, |
|
"eval_samples_per_second": 184.051, |
|
"eval_steps_per_second": 11.63, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.914622178606477, |
|
"eval_f1": 0.7038397809956056, |
|
"eval_loss": 0.31137627363204956, |
|
"eval_precision": 0.702372393961179, |
|
"eval_recall": 0.7053133121570893, |
|
"eval_runtime": 5.3805, |
|
"eval_samples_per_second": 185.297, |
|
"eval_steps_per_second": 11.709, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 1.9200000000000003e-05, |
|
"loss": 0.22, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_accuracy": 0.9169835623159961, |
|
"eval_f1": 0.7169057303932063, |
|
"eval_loss": 0.32163381576538086, |
|
"eval_precision": 0.7087625229293072, |
|
"eval_recall": 0.7252382327461738, |
|
"eval_runtime": 5.4697, |
|
"eval_samples_per_second": 182.278, |
|
"eval_steps_per_second": 11.518, |
|
"step": 1000 |
|
} |
|
], |
|
"max_steps": 25000, |
|
"num_train_epochs": 100, |
|
"total_flos": 623488737039000.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|