|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.999958205046044, |
|
"global_step": 13458, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0001, |
|
"loss": 2.7217, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 0.0001, |
|
"loss": 2.5316, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4681, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4397, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 0.0001, |
|
"loss": 2.4168, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3957, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 0.0001, |
|
"loss": 2.372, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3612, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3433, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3225, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3121, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 0.0001, |
|
"loss": 2.3033, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2938, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2881, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2775, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2721, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2677, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2558, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2465, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2413, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 0.0001, |
|
"loss": 2.233, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 2.45, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2215, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2258, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2195, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 0.0001, |
|
"loss": 2.222, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 0.0001, |
|
"loss": 2.2149, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 13458, |
|
"total_flos": 1.9989187863597466e+17, |
|
"train_loss": 2.3215815350955866, |
|
"train_runtime": 22056.1585, |
|
"train_samples_per_second": 39.052, |
|
"train_steps_per_second": 0.61 |
|
} |
|
], |
|
"max_steps": 13458, |
|
"num_train_epochs": 3, |
|
"total_flos": 1.9989187863597466e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|