|
{ |
|
"best_metric": 0.6673280721533259, |
|
"best_model_checkpoint": "final_models/glue_models/structroberta_s1_final//finetune/main_verb_lexical_content_the/checkpoint-400", |
|
"epoch": 10.0, |
|
"global_step": 610, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 6.56, |
|
"eval_accuracy": 0.6661840081214905, |
|
"eval_f1": 0.6673280721533259, |
|
"eval_loss": 3.184373378753662, |
|
"eval_mcc": 0.3323634262546941, |
|
"eval_runtime": 43.3539, |
|
"eval_samples_per_second": 509.735, |
|
"eval_steps_per_second": 63.731, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 8.2, |
|
"learning_rate": 9.016393442622952e-06, |
|
"loss": 0.0192, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"step": 610, |
|
"total_flos": 6096792351836160.0, |
|
"train_loss": 0.01572355615341517, |
|
"train_runtime": 383.5462, |
|
"train_samples_per_second": 190.486, |
|
"train_steps_per_second": 1.59 |
|
} |
|
], |
|
"max_steps": 610, |
|
"num_train_epochs": 10, |
|
"total_flos": 6096792351836160.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|