whisper-tiny-kor_eng_tiny_ed_lc

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

• Loss: 0.3587
• Cer: 14.8589

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	Cer	Validation Loss
1.6153	9.0909	100	5.2038	0.7524
0.2583	18.1818	200	5.9561	0.2177
0.0038	27.2727	300	7.0219	0.2586
0.0004	36.3636	400	0.2658	5.8934
0.0001	45.4545	500	0.2764	6.1442
0.0001	54.5455	600	0.2843	6.2696
0.0	63.6364	700	0.2902	6.2696
0.0	72.7273	800	0.2944	6.3323
0.0	81.8182	900	0.2990	6.3950
0.0	90.9091	1000	0.3028	7.0846
0.0	100.0	1100	0.3061	7.0846
0.0	109.0909	1200	0.3093	7.0846
0.0	118.1818	1300	0.3127	7.0846
0.0	127.2727	1400	0.3154	7.0846
0.0	136.3636	1500	0.3177	7.1473
0.0	145.4545	1600	0.3203	7.2727
0.0	154.5455	1700	0.3225	7.4608
0.0	163.6364	1800	0.3250	7.4608
0.0	172.7273	1900	0.3275	7.4608
0.0	181.8182	2000	0.3301	7.4608
0.0	190.9091	2100	0.3320	7.7116
0.0	200.0	2200	0.3348	7.7116
0.0	209.0909	2300	0.3375	7.7116
0.0	218.1818	2400	0.3390	7.7116
0.0	227.2727	2500	0.3411	7.7116
0.0	236.3636	2600	0.3432	7.7116
0.0	245.4545	2700	0.3450	7.7116
0.0	254.5455	2800	0.3469	7.7116
0.0	263.6364	2900	0.3484	14.5455
0.0	272.7273	3000	0.3503	14.5455
0.0	281.8182	3100	0.3518	14.8589
0.0	290.9091	3200	0.3529	14.8589
0.0	300.0	3300	0.3541	14.8589
0.0	309.0909	3400	0.3551	14.8589
0.0	318.1818	3500	0.3561	14.8589
0.0	327.2727	3600	0.3572	14.8589
0.0	336.3636	3700	0.3578	14.8589
0.0	345.4545	3800	0.3583	14.8589
0.0	354.5455	3900	0.3587	14.8589
0.0	363.6364	4000	0.3587	14.8589

Framework versions

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