|
{ |
|
"best_metric": 0.6792929292929293, |
|
"best_model_checkpoint": "tinybert-toxigen-bothpretrained\\run-3\\checkpoint-165", |
|
"epoch": 3.0, |
|
"eval_steps": 500, |
|
"global_step": 165, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.4720964669022145e-05, |
|
"loss": 0.3636, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.6378787878787879, |
|
"eval_loss": 0.3626602292060852, |
|
"eval_runtime": 3.3658, |
|
"eval_samples_per_second": 588.264, |
|
"eval_steps_per_second": 4.754, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 1.2267470557518455e-05, |
|
"loss": 0.3604, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.6393939393939394, |
|
"eval_loss": 0.35877418518066406, |
|
"eval_runtime": 3.34, |
|
"eval_samples_per_second": 592.817, |
|
"eval_steps_per_second": 4.79, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 9.813976446014764e-06, |
|
"loss": 0.356, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.6792929292929293, |
|
"eval_loss": 0.3534165918827057, |
|
"eval_runtime": 3.437, |
|
"eval_samples_per_second": 576.084, |
|
"eval_steps_per_second": 4.655, |
|
"step": 165 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 385, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 7, |
|
"save_steps": 500, |
|
"total_flos": 2524547079600.0, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.524441928071262, |
|
"learning_rate": 1.7174458780525836e-05, |
|
"num_train_epochs": 7, |
|
"temperature": 11 |
|
} |
|
} |
|
|