|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"global_step": 130, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.00013669982876312649, |
|
"loss": 2.5046, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.00013520660867542716, |
|
"loss": 2.1374, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.00013274211424821946, |
|
"loss": 2.2201, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 0.00012934228335981018, |
|
"loss": 1.9548, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 0.00012505669320030482, |
|
"loss": 2.0656, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 0.00011994783732453754, |
|
"loss": 1.9398, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 0.00011409021435531856, |
|
"loss": 2.1725, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 0.00010756924162575728, |
|
"loss": 1.8872, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 0.00010048000960220252, |
|
"loss": 1.9796, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.292589525111794e-05, |
|
"loss": 1.8577, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 8.501705457012647e-05, |
|
"loss": 1.7766, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 7.686881626551515e-05, |
|
"loss": 1.8008, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 6.86e-05, |
|
"loss": 1.8354, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 6.033118373448485e-05, |
|
"loss": 1.9892, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 5.218294542987356e-05, |
|
"loss": 1.8204, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 4.427410474888205e-05, |
|
"loss": 1.8009, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 3.671999039779748e-05, |
|
"loss": 1.7768, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 2.963075837424272e-05, |
|
"loss": 1.7438, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.3109785644681465e-05, |
|
"loss": 1.7175, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 1.725216267546246e-05, |
|
"loss": 1.9227, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.2143306799695174e-05, |
|
"loss": 1.9182, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 7.857716640189809e-06, |
|
"loss": 1.8767, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 4.45788575178055e-06, |
|
"loss": 1.8175, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 1.993391324572832e-06, |
|
"loss": 1.727, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 5.001712368734975e-07, |
|
"loss": 1.6829, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.0, |
|
"loss": 1.8023, |
|
"step": 130 |
|
} |
|
], |
|
"max_steps": 130, |
|
"num_train_epochs": 1, |
|
"total_flos": 135610564608000.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|