|
{ |
|
"best_metric": 20.932719441556515, |
|
"best_model_checkpoint": "./checkpoint-1000", |
|
"epoch": 83.33333333333333, |
|
"global_step": 1000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 5.000000000000001e-07, |
|
"loss": 0.5839, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 4.17, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 0.5075, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 6.25, |
|
"learning_rate": 1.5e-06, |
|
"loss": 0.4123, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 8.33, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 0.3349, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 10.42, |
|
"learning_rate": 2.5e-06, |
|
"loss": 0.2786, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 12.5, |
|
"learning_rate": 3e-06, |
|
"loss": 0.2203, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 14.58, |
|
"learning_rate": 3.5e-06, |
|
"loss": 0.1776, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 16.67, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 0.1259, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 18.75, |
|
"learning_rate": 4.5e-06, |
|
"loss": 0.0858, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 20.83, |
|
"learning_rate": 5e-06, |
|
"loss": 0.0538, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 22.92, |
|
"learning_rate": 5.500000000000001e-06, |
|
"loss": 0.0303, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 25.0, |
|
"learning_rate": 6e-06, |
|
"loss": 0.0177, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 27.08, |
|
"learning_rate": 6.5000000000000004e-06, |
|
"loss": 0.0105, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 29.17, |
|
"learning_rate": 7e-06, |
|
"loss": 0.0068, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 31.25, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 0.0044, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 33.33, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 0.0033, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 35.42, |
|
"learning_rate": 8.5e-06, |
|
"loss": 0.0025, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 37.5, |
|
"learning_rate": 9e-06, |
|
"loss": 0.002, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 39.58, |
|
"learning_rate": 9.5e-06, |
|
"loss": 0.0016, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 41.67, |
|
"learning_rate": 1e-05, |
|
"loss": 0.0013, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 43.75, |
|
"learning_rate": 9.5e-06, |
|
"loss": 0.0011, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 45.83, |
|
"learning_rate": 9e-06, |
|
"loss": 0.001, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 47.92, |
|
"learning_rate": 8.5e-06, |
|
"loss": 0.0008, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 50.0, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 0.0007, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 52.08, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 0.0007, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 54.17, |
|
"learning_rate": 7e-06, |
|
"loss": 0.0006, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 56.25, |
|
"learning_rate": 6.5000000000000004e-06, |
|
"loss": 0.0006, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 58.33, |
|
"learning_rate": 6e-06, |
|
"loss": 0.0005, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 60.42, |
|
"learning_rate": 5.500000000000001e-06, |
|
"loss": 0.0005, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 62.5, |
|
"learning_rate": 5e-06, |
|
"loss": 0.0005, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 64.58, |
|
"learning_rate": 4.5e-06, |
|
"loss": 0.0005, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 66.67, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 0.0004, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 68.75, |
|
"learning_rate": 3.5e-06, |
|
"loss": 0.0004, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 70.83, |
|
"learning_rate": 3e-06, |
|
"loss": 0.0004, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 72.92, |
|
"learning_rate": 2.5e-06, |
|
"loss": 0.0004, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 75.0, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 0.0004, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 77.08, |
|
"learning_rate": 1.5e-06, |
|
"loss": 0.0004, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 79.17, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 0.0004, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 81.25, |
|
"learning_rate": 5.000000000000001e-07, |
|
"loss": 0.0004, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 83.33, |
|
"learning_rate": 0.0, |
|
"loss": 0.0004, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 83.33, |
|
"eval_loss": 0.5390146970748901, |
|
"eval_runtime": 793.4997, |
|
"eval_samples_per_second": 0.745, |
|
"eval_steps_per_second": 0.047, |
|
"eval_wer": 20.932719441556515, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 83.33, |
|
"step": 1000, |
|
"total_flos": 1.79904138559488e+19, |
|
"train_loss": 0.07179988482128828, |
|
"train_runtime": 4551.8374, |
|
"train_samples_per_second": 14.06, |
|
"train_steps_per_second": 0.22 |
|
} |
|
], |
|
"max_steps": 1000, |
|
"num_train_epochs": 84, |
|
"total_flos": 1.79904138559488e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|