DylanJHJ's picture
upload checkpoints
7333e13
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.018754186202277,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.33,
"learning_rate": 2.5e-05,
"loss": 0.1319,
"step": 500
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.0444,
"step": 1000
},
{
"epoch": 1.0,
"learning_rate": 4.722222222222222e-05,
"loss": 0.0401,
"step": 1500
},
{
"epoch": 1.34,
"learning_rate": 4.4444444444444447e-05,
"loss": 0.0367,
"step": 2000
},
{
"epoch": 1.67,
"learning_rate": 4.166666666666667e-05,
"loss": 0.0356,
"step": 2500
},
{
"epoch": 2.01,
"learning_rate": 3.888888888888889e-05,
"loss": 0.0352,
"step": 3000
},
{
"epoch": 2.01,
"eval_loss": 0.35025253891944885,
"eval_runtime": 43.6746,
"eval_samples_per_second": 75.055,
"eval_steps_per_second": 18.775,
"step": 3000
},
{
"epoch": 2.34,
"learning_rate": 3.611111111111111e-05,
"loss": 0.0332,
"step": 3500
},
{
"epoch": 2.68,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.0329,
"step": 4000
},
{
"epoch": 3.01,
"learning_rate": 3.055555555555556e-05,
"loss": 0.0329,
"step": 4500
},
{
"epoch": 3.35,
"learning_rate": 2.777777777777778e-05,
"loss": 0.0314,
"step": 5000
},
{
"epoch": 3.68,
"learning_rate": 2.5e-05,
"loss": 0.0315,
"step": 5500
},
{
"epoch": 4.02,
"learning_rate": 2.2222222222222223e-05,
"loss": 0.0311,
"step": 6000
},
{
"epoch": 4.02,
"eval_loss": 0.343203067779541,
"eval_runtime": 43.7614,
"eval_samples_per_second": 74.906,
"eval_steps_per_second": 18.738,
"step": 6000
}
],
"max_steps": 10000,
"num_train_epochs": 7,
"total_flos": 3.76173262867415e+16,
"trial_name": null,
"trial_params": null
}