whisper-tiny-kor_eng_tiny_pu_lc

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.1931
• Cer: 45.4569

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• 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 results

Training Loss	Epoch	Step	Validation Loss	Cer
4.5801	0.3215	100	3.5304	61.3564
3.2109	0.6431	200	2.8469	56.6179
2.7285	0.9646	300	2.4536	52.9449
2.2208	1.2862	400	2.2358	51.0614
2.1218	1.6077	500	2.0682	50.2996
1.9401	1.9293	600	1.9574	48.7154
1.6592	2.2508	700	1.9034	47.5534
1.5294	2.5723	800	1.8213	47.4325
1.4396	2.8939	900	1.7628	46.5537
1.191	3.2154	1000	1.7707	45.7587
1.0942	3.5370	1100	1.7421	46.6279
1.0841	3.8585	1200	1.7177	45.2688
0.9389	4.1801	1300	1.7614	45.3705
0.7875	4.5016	1400	1.7524	46.9368
0.8226	4.8232	1500	1.7437	45.6014
0.7305	5.1447	1600	1.8109	45.4933
0.5954	5.4662	1700	1.8032	45.2931
0.6285	5.7878	1800	1.8162	45.1064
0.5397	6.1093	1900	1.8557	45.5554
0.4319	6.4309	2000	1.8992	44.9983
0.4295	6.7524	2100	1.8797	44.8263
0.4061	7.0740	2200	1.9398	45.1838
0.3088	7.3955	2300	1.9597	45.2317
0.3164	7.7170	2400	1.9753	45.4102
0.3024	8.0386	2500	2.0079	45.6730
0.2242	8.3601	2600	2.0153	46.7756
0.2374	8.6817	2700	2.0264	45.4825
0.2135	9.0032	2800	2.0365	45.9212
0.1569	9.3248	2900	2.0829	45.9333
0.1689	9.6463	3000	2.0868	45.7210
0.1625	9.9678	3100	2.0860	45.2637
0.1194	10.2894	3200	2.1307	46.3491
0.1197	10.6109	3300	2.1322	45.3565
0.1175	10.9325	3400	2.1421	45.8943
0.0931	11.2540	3500	2.1646	45.7709
0.086	11.5756	3600	2.1691	45.9423
0.0931	11.8971	3700	2.1769	45.3987
0.0822	12.2186	3800	2.1896	45.9858
0.0729	12.5402	3900	2.1911	45.2791
0.0819	12.8617	4000	2.1931	45.4569

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.1