learning_rate: 3e-05
train_batch_size: 12
eval_batch_size: 6
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 4000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer
3.1262	4.0	100	2.6544	51.7475
2.073	8.0	200	2.1089	49.6658
1.1269	12.0	300	1.9373	49.1654
0.413	16.0	400	2.0504	50.4297
0.1382	20.0	500	2.0960	48.7606
0.062	24.0	600	2.1393	47.4313
0.0337	28.0	700	2.2064	50.1852
0.0243	32.0	800	2.1941	50.3151
0.0167	36.0	900	2.2504	49.0776
0.0148	40.0	1000	2.2040	47.5765
0.0125	44.0	1100	2.2400	48.8026
0.0086	48.0	1200	2.2466	47.7216
0.0073	52.0	1300	2.3227	47.7675
0.0055	56.0	1400	2.3057	48.6498
0.0039	60.0	1500	2.2848	47.7522
0.0035	64.0	1600	2.2913	46.6178
0.002	68.0	1700	2.3235	46.4421
0.0014	72.0	1800	2.3380	49.2647
0.002	76.0	1900	2.3131	47.8668
0.001	80.0	2000	2.3332	48.8102
0.0007	84.0	2100	2.3405	48.0578
0.0005	88.0	2200	2.3513	46.4841
0.0004	92.0	2300	2.3611	47.4313
0.0003	96.0	2400	2.3644	47.2213
0.0003	100.0	2500	2.3685	47.1945
0.0003	104.0	2600	2.3729	47.4695
0.0003	108.0	2700	2.3764	47.6491
0.0003	112.0	2800	2.3805	47.5688
0.0002	116.0	2900	2.3835	47.3779
0.0002	120.0	3000	2.3864	47.4122
0.0002	124.0	3100	2.3891	47.6032
0.0002	128.0	3200	2.3923	47.7904
0.0002	132.0	3300	2.3949	47.7904
0.0002	136.0	3400	2.3965	47.9011
0.0002	140.0	3500	2.3985	47.8439
0.0002	144.0	3600	2.3999	48.4092
0.0002	148.0	3700	2.4015	48.9974
0.0002	152.0	3800	2.4020	48.8675
0.0002	156.0	3900	2.4028	49.0126
0.0002	160.0	4000	2.4029	48.9821

CianKim/whisper-tiny-kor_eng_tiny_pu_pr

whisper-tiny-kor_eng_tiny_pu_pr