|
{ |
|
"best_metric": 0.0, |
|
"best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-2/checkpoint-107", |
|
"epoch": 3.0, |
|
"eval_steps": 500, |
|
"global_step": 321, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 0.6115646362304688, |
|
"eval_matthews_correlation": 0.0, |
|
"eval_runtime": 0.9988, |
|
"eval_samples_per_second": 1044.286, |
|
"eval_steps_per_second": 131.162, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 0.6053324937820435, |
|
"eval_matthews_correlation": 0.0, |
|
"eval_runtime": 1.2151, |
|
"eval_samples_per_second": 858.345, |
|
"eval_steps_per_second": 107.807, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_loss": 0.5962837338447571, |
|
"eval_matthews_correlation": 0.0, |
|
"eval_runtime": 1.3302, |
|
"eval_samples_per_second": 784.082, |
|
"eval_steps_per_second": 98.48, |
|
"step": 321 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 535, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 5, |
|
"save_steps": 500, |
|
"total_flos": 0, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": { |
|
"learning_rate": 3.065563813356405e-06, |
|
"num_train_epochs": 5, |
|
"per_device_train_batch_size": 8, |
|
"seed": 29 |
|
} |
|
} |
|
|