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	Cer	Validation Loss
0.8207	16.6667	100	6.9620	0.9476
0.0353	33.3333	200	7.9114	0.8010
0.0002	50.0	300	7.9114	0.8538
0.0001	66.6667	400	7.9114	0.8916
0.0	83.3333	500	7.9114	0.9169
0.0	100.0	600	5.9072	0.9309
0.0	116.6667	700	5.8017	0.9514
0.0	133.3333	800	6.1181	0.9646
0.0	150.0	900	6.1181	0.9769
0.0	166.6667	1000	6.1181	0.9847
0.0	183.3333	1100	6.5401	1.0037
0.0	200.0	1200	6.5401	1.0220
0.0	216.6667	1300	6.6456	1.0314
0.0	233.3333	1400	7.1730	1.0465
0.0	250.0	1500	7.1730	1.0657
0.0	266.6667	1600	7.1730	1.0768
0.0	283.3333	1700	7.1730	1.0807
0.0	300.0	1800	7.5949	1.1066
0.0	316.6667	1900	7.5949	1.1196
0.0	333.3333	2000	7.5949	1.1301
0.0	350.0	2100	6.9620	1.1401
0.0	366.6667	2200	6.9620	1.1537
0.0	383.3333	2300	6.9620	1.1606
0.0	400.0	2400	6.9620	1.1631
0.0	416.6667	2500	6.3291	1.1769
0.0	433.3333	2600	6.3291	1.1780
0.0	450.0	2700	7.1730	1.1860
0.0	466.6667	2800	6.0127	1.1920
0.0	483.3333	2900	6.1181	1.1972
0.0	500.0	3000	6.3291	1.2019
0.0	516.6667	3100	6.5401	1.2073
0.0	533.3333	3200	6.7511	1.2118
0.0	550.0	3300	6.6456	1.2163
0.0	566.6667	3400	6.6456	1.2171
0.0	583.3333	3500	6.6456	1.2232
0.0	600.0	3600	1.2269	6.6456
0.0	616.6667	3700	1.2264	6.6456
0.0	633.3333	3800	1.2278	6.6456
0.0	650.0	3900	1.2286	6.6456
0.0	666.6667	4000	1.2284	6.6456

CianKim/whisper-tiny-kor_eng_tiny_ed_ob

whisper-tiny-kor_eng_tiny_ed_ob