|
{ |
|
"best_metric": 0.9334564208984375, |
|
"best_model_checkpoint": "outputs/checkpoint-138", |
|
"epoch": 2.979757085020243, |
|
"eval_steps": 500, |
|
"global_step": 138, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.4318488529014845, |
|
"grad_norm": 0.5840024948120117, |
|
"learning_rate": 4e-05, |
|
"loss": 1.9547, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.863697705802969, |
|
"grad_norm": 0.4871944487094879, |
|
"learning_rate": 8e-05, |
|
"loss": 1.8203, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.9932523616734144, |
|
"eval_loss": 1.5045615434646606, |
|
"eval_runtime": 13.3054, |
|
"eval_samples_per_second": 27.959, |
|
"eval_steps_per_second": 3.532, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 1.2955465587044535, |
|
"grad_norm": 0.6788378953933716, |
|
"learning_rate": 0.00012, |
|
"loss": 1.6913, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 1.7273954116059378, |
|
"grad_norm": 0.8430602550506592, |
|
"learning_rate": 0.00016, |
|
"loss": 1.4654, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 1.9865047233468287, |
|
"eval_loss": 1.10453200340271, |
|
"eval_runtime": 13.2625, |
|
"eval_samples_per_second": 28.049, |
|
"eval_steps_per_second": 3.544, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 2.1592442645074224, |
|
"grad_norm": 1.1823046207427979, |
|
"learning_rate": 0.0002, |
|
"loss": 1.3454, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 2.591093117408907, |
|
"grad_norm": 1.2263983488082886, |
|
"learning_rate": 0.00019957591414302984, |
|
"loss": 1.1674, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 2.979757085020243, |
|
"eval_loss": 0.9334564208984375, |
|
"eval_runtime": 13.257, |
|
"eval_samples_per_second": 28.061, |
|
"eval_steps_per_second": 3.545, |
|
"step": 138 |
|
} |
|
], |
|
"logging_steps": 20, |
|
"max_steps": 782, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 17, |
|
"save_steps": 500, |
|
"total_flos": 1.0633812858433536e+16, |
|
"train_batch_size": 2, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|