ihanif's picture
add best WER checkpoint at step 1200
93bff9a
{
"best_metric": 50.544794188861985,
"best_model_checkpoint": "./checkpoint-1200",
"epoch": 171.42857142857142,
"global_step": 1200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 3.57,
"learning_rate": 4.4e-06,
"loss": 2.1832,
"step": 25
},
{
"epoch": 7.14,
"learning_rate": 9.4e-06,
"loss": 0.7236,
"step": 50
},
{
"epoch": 10.71,
"learning_rate": 9.76842105263158e-06,
"loss": 0.2181,
"step": 75
},
{
"epoch": 14.29,
"learning_rate": 9.505263157894738e-06,
"loss": 0.0334,
"step": 100
},
{
"epoch": 14.29,
"eval_loss": 1.0348163843154907,
"eval_runtime": 1390.6808,
"eval_samples_per_second": 0.368,
"eval_steps_per_second": 0.023,
"eval_wer": 50.090799031477,
"step": 100
},
{
"epoch": 17.86,
"learning_rate": 9.242105263157896e-06,
"loss": 0.0077,
"step": 125
},
{
"epoch": 21.43,
"learning_rate": 8.978947368421055e-06,
"loss": 0.0041,
"step": 150
},
{
"epoch": 25.0,
"learning_rate": 8.715789473684211e-06,
"loss": 0.0026,
"step": 175
},
{
"epoch": 28.57,
"learning_rate": 8.45263157894737e-06,
"loss": 0.0021,
"step": 200
},
{
"epoch": 28.57,
"eval_loss": 1.1971338987350464,
"eval_runtime": 1356.432,
"eval_samples_per_second": 0.377,
"eval_steps_per_second": 0.024,
"eval_wer": 49.48547215496368,
"step": 200
},
{
"epoch": 32.14,
"learning_rate": 8.189473684210527e-06,
"loss": 0.0014,
"step": 225
},
{
"epoch": 35.71,
"learning_rate": 7.926315789473686e-06,
"loss": 0.0009,
"step": 250
},
{
"epoch": 39.29,
"learning_rate": 7.663157894736842e-06,
"loss": 0.0008,
"step": 275
},
{
"epoch": 42.86,
"learning_rate": 7.4e-06,
"loss": 0.0007,
"step": 300
},
{
"epoch": 42.86,
"eval_loss": 1.265063762664795,
"eval_runtime": 1370.312,
"eval_samples_per_second": 0.374,
"eval_steps_per_second": 0.023,
"eval_wer": 49.73516949152542,
"step": 300
},
{
"epoch": 46.43,
"learning_rate": 7.1368421052631585e-06,
"loss": 0.0007,
"step": 325
},
{
"epoch": 50.0,
"learning_rate": 6.873684210526317e-06,
"loss": 0.0006,
"step": 350
},
{
"epoch": 53.57,
"learning_rate": 6.610526315789474e-06,
"loss": 0.0006,
"step": 375
},
{
"epoch": 57.14,
"learning_rate": 6.347368421052632e-06,
"loss": 0.0006,
"step": 400
},
{
"epoch": 57.14,
"eval_loss": 1.3083866834640503,
"eval_runtime": 1377.4534,
"eval_samples_per_second": 0.372,
"eval_steps_per_second": 0.023,
"eval_wer": 49.969733656174334,
"step": 400
},
{
"epoch": 60.71,
"learning_rate": 6.08421052631579e-06,
"loss": 0.0005,
"step": 425
},
{
"epoch": 64.29,
"learning_rate": 5.8210526315789486e-06,
"loss": 0.0005,
"step": 450
},
{
"epoch": 67.86,
"learning_rate": 5.557894736842105e-06,
"loss": 0.0005,
"step": 475
},
{
"epoch": 71.43,
"learning_rate": 5.294736842105263e-06,
"loss": 0.0005,
"step": 500
},
{
"epoch": 71.43,
"eval_loss": 1.3478939533233643,
"eval_runtime": 1368.2052,
"eval_samples_per_second": 0.374,
"eval_steps_per_second": 0.023,
"eval_wer": 50.060532687651325,
"step": 500
},
{
"epoch": 75.0,
"learning_rate": 5.0315789473684214e-06,
"loss": 0.0005,
"step": 525
},
{
"epoch": 78.57,
"learning_rate": 4.76842105263158e-06,
"loss": 0.0005,
"step": 550
},
{
"epoch": 82.14,
"learning_rate": 4.505263157894737e-06,
"loss": 0.0004,
"step": 575
},
{
"epoch": 85.71,
"learning_rate": 4.242105263157895e-06,
"loss": 0.0004,
"step": 600
},
{
"epoch": 85.71,
"eval_loss": 1.3834712505340576,
"eval_runtime": 1377.4466,
"eval_samples_per_second": 0.372,
"eval_steps_per_second": 0.023,
"eval_wer": 50.30266343825666,
"step": 600
},
{
"epoch": 89.29,
"learning_rate": 3.9789473684210525e-06,
"loss": 0.0004,
"step": 625
},
{
"epoch": 92.86,
"learning_rate": 3.715789473684211e-06,
"loss": 0.0004,
"step": 650
},
{
"epoch": 96.43,
"learning_rate": 3.4526315789473684e-06,
"loss": 0.0004,
"step": 675
},
{
"epoch": 100.0,
"learning_rate": 3.1894736842105266e-06,
"loss": 0.0004,
"step": 700
},
{
"epoch": 100.0,
"eval_loss": 1.4139070510864258,
"eval_runtime": 1374.8999,
"eval_samples_per_second": 0.372,
"eval_steps_per_second": 0.023,
"eval_wer": 50.45399515738499,
"step": 700
},
{
"epoch": 103.57,
"learning_rate": 2.9263157894736844e-06,
"loss": 0.0004,
"step": 725
},
{
"epoch": 107.14,
"learning_rate": 2.6631578947368426e-06,
"loss": 0.0004,
"step": 750
},
{
"epoch": 110.71,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.0004,
"step": 775
},
{
"epoch": 114.29,
"learning_rate": 2.136842105263158e-06,
"loss": 0.0004,
"step": 800
},
{
"epoch": 114.29,
"eval_loss": 1.4381563663482666,
"eval_runtime": 1378.8054,
"eval_samples_per_second": 0.371,
"eval_steps_per_second": 0.023,
"eval_wer": 50.46156174334141,
"step": 800
},
{
"epoch": 117.86,
"learning_rate": 1.8736842105263158e-06,
"loss": 0.0004,
"step": 825
},
{
"epoch": 121.43,
"learning_rate": 1.6105263157894738e-06,
"loss": 0.0004,
"step": 850
},
{
"epoch": 125.0,
"learning_rate": 1.3473684210526316e-06,
"loss": 0.0004,
"step": 875
},
{
"epoch": 128.57,
"learning_rate": 1.0842105263157895e-06,
"loss": 0.0004,
"step": 900
},
{
"epoch": 128.57,
"eval_loss": 1.454498291015625,
"eval_runtime": 1378.7878,
"eval_samples_per_second": 0.371,
"eval_steps_per_second": 0.023,
"eval_wer": 50.529661016949156,
"step": 900
},
{
"epoch": 132.14,
"learning_rate": 8.210526315789474e-07,
"loss": 0.0003,
"step": 925
},
{
"epoch": 135.71,
"learning_rate": 5.578947368421053e-07,
"loss": 0.0003,
"step": 950
},
{
"epoch": 139.29,
"learning_rate": 2.9473684210526315e-07,
"loss": 0.0003,
"step": 975
},
{
"epoch": 142.86,
"learning_rate": 3.157894736842106e-08,
"loss": 0.0003,
"step": 1000
},
{
"epoch": 142.86,
"eval_loss": 1.4602879285812378,
"eval_runtime": 1395.1761,
"eval_samples_per_second": 0.367,
"eval_steps_per_second": 0.023,
"eval_wer": 50.56749394673123,
"step": 1000
},
{
"epoch": 146.43,
"learning_rate": 1.4957983193277311e-06,
"loss": 0.0003,
"step": 1025
},
{
"epoch": 150.0,
"learning_rate": 1.2857142857142856e-06,
"loss": 0.0003,
"step": 1050
},
{
"epoch": 153.57,
"learning_rate": 1.0756302521008405e-06,
"loss": 0.0003,
"step": 1075
},
{
"epoch": 157.14,
"learning_rate": 8.65546218487395e-07,
"loss": 0.0003,
"step": 1100
},
{
"epoch": 157.14,
"eval_loss": 1.4750263690948486,
"eval_runtime": 1385.2093,
"eval_samples_per_second": 0.37,
"eval_steps_per_second": 0.023,
"eval_wer": 50.559927360774815,
"step": 1100
},
{
"epoch": 160.71,
"learning_rate": 6.554621848739497e-07,
"loss": 0.0003,
"step": 1125
},
{
"epoch": 164.29,
"learning_rate": 4.4537815126050424e-07,
"loss": 0.0003,
"step": 1150
},
{
"epoch": 167.86,
"learning_rate": 2.3529411764705883e-07,
"loss": 0.0003,
"step": 1175
},
{
"epoch": 171.43,
"learning_rate": 2.521008403361345e-08,
"loss": 0.0003,
"step": 1200
},
{
"epoch": 171.43,
"eval_loss": 1.4806574583053589,
"eval_runtime": 1382.0563,
"eval_samples_per_second": 0.37,
"eval_steps_per_second": 0.023,
"eval_wer": 50.544794188861985,
"step": 1200
}
],
"max_steps": 1200,
"num_train_epochs": 172,
"total_flos": 7.541555843137536e+19,
"trial_name": null,
"trial_params": null
}