|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 6.289308176100629, |
|
"eval_steps": 500, |
|
"global_step": 2000, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.6535483870967742, |
|
"eval_loss": 0.28503966331481934, |
|
"eval_runtime": 2.1194, |
|
"eval_samples_per_second": 1462.691, |
|
"eval_steps_per_second": 30.669, |
|
"step": 318 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"grad_norm": 0.7276282906532288, |
|
"learning_rate": 1.685534591194969e-05, |
|
"loss": 0.4585, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.8422580645161291, |
|
"eval_loss": 0.1152644231915474, |
|
"eval_runtime": 2.111, |
|
"eval_samples_per_second": 1468.52, |
|
"eval_steps_per_second": 30.792, |
|
"step": 636 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.8938709677419355, |
|
"eval_loss": 0.06576532125473022, |
|
"eval_runtime": 2.1204, |
|
"eval_samples_per_second": 1462.014, |
|
"eval_steps_per_second": 30.655, |
|
"step": 954 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"grad_norm": 0.6005321741104126, |
|
"learning_rate": 1.371069182389937e-05, |
|
"loss": 0.1354, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_accuracy": 0.9119354838709678, |
|
"eval_loss": 0.04726028814911842, |
|
"eval_runtime": 2.0744, |
|
"eval_samples_per_second": 1494.437, |
|
"eval_steps_per_second": 31.335, |
|
"step": 1272 |
|
}, |
|
{ |
|
"epoch": 4.72, |
|
"grad_norm": 0.449545294046402, |
|
"learning_rate": 1.0566037735849058e-05, |
|
"loss": 0.0753, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_accuracy": 0.92, |
|
"eval_loss": 0.038489848375320435, |
|
"eval_runtime": 4.3863, |
|
"eval_samples_per_second": 706.75, |
|
"eval_steps_per_second": 14.819, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_accuracy": 0.9261290322580645, |
|
"eval_loss": 0.03405739739537239, |
|
"eval_runtime": 2.3565, |
|
"eval_samples_per_second": 1315.523, |
|
"eval_steps_per_second": 27.584, |
|
"step": 1908 |
|
}, |
|
{ |
|
"epoch": 6.29, |
|
"grad_norm": 0.3650922179222107, |
|
"learning_rate": 7.421383647798742e-06, |
|
"loss": 0.0567, |
|
"step": 2000 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 3180, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 10, |
|
"save_steps": 500, |
|
"total_flos": 567980855842764.0, |
|
"train_batch_size": 48, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.2856698138299478, |
|
"num_train_epochs": 10, |
|
"temperature": 3 |
|
} |
|
} |
|
|