|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 4.948148148148148, |
|
"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": 0.5518518518518518, |
|
"eval_f1": 0.5518518518518518, |
|
"eval_loss": 1.3629041910171509, |
|
"eval_precision": 0.5518518518518518, |
|
"eval_recall": 0.5518518518518518, |
|
"eval_runtime": 2.8033, |
|
"eval_samples_per_second": 288.949, |
|
"eval_steps_per_second": 18.193, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.7839506172839507, |
|
"eval_f1": 0.7839506172839507, |
|
"eval_loss": 0.7004565596580505, |
|
"eval_precision": 0.7839506172839507, |
|
"eval_recall": 0.7839506172839507, |
|
"eval_runtime": 2.7948, |
|
"eval_samples_per_second": 289.822, |
|
"eval_steps_per_second": 18.248, |
|
"step": 404 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 1.1221122112211222e-05, |
|
"loss": 1.2867, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.8049382716049382, |
|
"eval_f1": 0.8049382716049382, |
|
"eval_loss": 0.6230440735816956, |
|
"eval_precision": 0.8049382716049382, |
|
"eval_recall": 0.8049382716049382, |
|
"eval_runtime": 2.7886, |
|
"eval_samples_per_second": 290.468, |
|
"eval_steps_per_second": 18.289, |
|
"step": 606 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_accuracy": 0.8111111111111111, |
|
"eval_f1": 0.8111111111111111, |
|
"eval_loss": 0.5868209004402161, |
|
"eval_precision": 0.8111111111111111, |
|
"eval_recall": 0.8111111111111111, |
|
"eval_runtime": 2.7994, |
|
"eval_samples_per_second": 289.352, |
|
"eval_steps_per_second": 18.218, |
|
"step": 808 |
|
}, |
|
{ |
|
"epoch": 4.95, |
|
"learning_rate": 2.2002200220022005e-07, |
|
"loss": 0.6022, |
|
"step": 1000 |
|
} |
|
], |
|
"max_steps": 1010, |
|
"num_train_epochs": 5, |
|
"total_flos": 1656830648519424.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|