|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 8.875739644970414, |
|
"global_step": 6000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 4.96e-05, |
|
"loss": 1.8134, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 4.996328645447816e-05, |
|
"loss": 1.3742, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 4.992627683197632e-05, |
|
"loss": 1.122, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 4.988934122871947e-05, |
|
"loss": 0.8645, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 3.7, |
|
"learning_rate": 4.9852553663952626e-05, |
|
"loss": 0.6174, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 4.9815618060695786e-05, |
|
"loss": 0.5355, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 5.18, |
|
"learning_rate": 4.977868245743893e-05, |
|
"loss": 0.4336, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 5.92, |
|
"learning_rate": 4.974167283493709e-05, |
|
"loss": 0.3695, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 6.66, |
|
"learning_rate": 4.970466321243524e-05, |
|
"loss": 0.3013, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 7.4, |
|
"learning_rate": 4.966772760917839e-05, |
|
"loss": 0.2954, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 8.14, |
|
"learning_rate": 4.963071798667654e-05, |
|
"loss": 0.2812, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 8.88, |
|
"learning_rate": 4.959370836417469e-05, |
|
"loss": 0.2718, |
|
"step": 6000 |
|
} |
|
], |
|
"max_steps": 676000, |
|
"num_train_epochs": 1000, |
|
"total_flos": 1.2118642683671347e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|