|
{ |
|
"best_metric": 1.6233268976211548, |
|
"best_model_checkpoint": "AlexWang99/byt5_add/checkpoint-100", |
|
"epoch": 5.0, |
|
"eval_steps": 500, |
|
"global_step": 100, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 1.6871496438980103, |
|
"eval_runtime": 12.4033, |
|
"eval_samples_per_second": 806.234, |
|
"eval_steps_per_second": 1.612, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 1.6486749649047852, |
|
"eval_runtime": 12.4805, |
|
"eval_samples_per_second": 801.251, |
|
"eval_steps_per_second": 1.603, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_loss": 1.6341347694396973, |
|
"eval_runtime": 12.3858, |
|
"eval_samples_per_second": 807.376, |
|
"eval_steps_per_second": 1.615, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_loss": 1.6265267133712769, |
|
"eval_runtime": 12.382, |
|
"eval_samples_per_second": 807.624, |
|
"eval_steps_per_second": 1.615, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_loss": 1.6233268976211548, |
|
"eval_runtime": 12.4067, |
|
"eval_samples_per_second": 806.016, |
|
"eval_steps_per_second": 1.612, |
|
"step": 100 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 100, |
|
"num_train_epochs": 5, |
|
"save_steps": 500, |
|
"total_flos": 1435546214400000.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|