|
{ |
|
"best_metric": 0.7325565814971924, |
|
"best_model_checkpoint": "../../models/lora/codellama-13b-codtok-multitask_cot_reflection/checkpoint-200", |
|
"epoch": 12.5, |
|
"eval_steps": 200, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.9999999999999997e-05, |
|
"loss": 2.0085, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 5.9999999999999995e-05, |
|
"loss": 1.9902, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 8.999999999999999e-05, |
|
"loss": 1.8388, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 0.00011999999999999999, |
|
"loss": 1.5146, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 0.00015, |
|
"loss": 1.1563, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 3.75, |
|
"learning_rate": 0.00017999999999999998, |
|
"loss": 0.9863, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 4.38, |
|
"learning_rate": 0.00020999999999999998, |
|
"loss": 0.9537, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 0.00023999999999999998, |
|
"loss": 0.912, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 5.62, |
|
"learning_rate": 0.00027, |
|
"loss": 0.8958, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 6.25, |
|
"learning_rate": 0.0003, |
|
"loss": 0.8613, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 6.88, |
|
"learning_rate": 0.0002921052631578947, |
|
"loss": 0.8499, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 7.5, |
|
"learning_rate": 0.0002842105263157894, |
|
"loss": 0.8511, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 8.12, |
|
"learning_rate": 0.0002763157894736842, |
|
"loss": 0.8283, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 8.75, |
|
"learning_rate": 0.0002684210526315789, |
|
"loss": 0.812, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 9.38, |
|
"learning_rate": 0.0002605263157894737, |
|
"loss": 0.7994, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"learning_rate": 0.00025263157894736836, |
|
"loss": 0.8173, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 10.62, |
|
"learning_rate": 0.00024473684210526314, |
|
"loss": 0.8028, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 11.25, |
|
"learning_rate": 0.00023684210526315788, |
|
"loss": 0.7976, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 11.88, |
|
"learning_rate": 0.00022894736842105263, |
|
"loss": 0.783, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 12.5, |
|
"learning_rate": 0.00022105263157894733, |
|
"loss": 0.7765, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 12.5, |
|
"eval_loss": 0.7325565814971924, |
|
"eval_runtime": 1.7436, |
|
"eval_samples_per_second": 12.044, |
|
"eval_steps_per_second": 1.721, |
|
"step": 200 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 480, |
|
"num_train_epochs": 30, |
|
"save_steps": 200, |
|
"total_flos": 5.1901172112850944e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|