|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.9999442741710785, |
|
"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": 1.4936, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4537, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4376, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 0.0001, |
|
"loss": 1.433, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4288, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4231, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4174, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 0.0001, |
|
"loss": 1.416, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4075, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3519, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 0.0001, |
|
"loss": 1.352, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3503, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3516, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3527, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3494, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3516, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 0.0001, |
|
"loss": 1.352, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3455, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2913, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2959, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2944, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 2.45, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2932, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3009, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2983, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3056, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 0.0001, |
|
"loss": 1.3026, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 13458, |
|
"total_flos": 3.24401732645145e+18, |
|
"train_loss": 1.3612319644696114, |
|
"train_runtime": 79001.0035, |
|
"train_samples_per_second": 10.903, |
|
"train_steps_per_second": 0.17 |
|
} |
|
], |
|
"max_steps": 13458, |
|
"num_train_epochs": 3, |
|
"total_flos": 3.24401732645145e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|