whisper-tiny-kor_eng_tiny_ps_lc

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

• Loss: 0.9590
• Cer: 16.4797

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
0.6706	8.3333	100	0.7326	16.5498
0.0538	16.6667	200	0.7100	13.6746
0.0024	25.0	300	0.7662	14.5161
0.0006	33.3333	400	0.7955	14.8668
0.0003	41.6667	500	0.8159	13.6746
0.0002	50.0	600	0.8301	14.0252
0.0001	58.3333	700	0.8450	13.1837
0.0001	66.6667	800	0.8559	13.1136
0.0001	75.0	900	0.8623	12.9032
0.0001	83.3333	1000	0.8691	12.9734
0.0001	91.6667	1100	0.8746	13.3941
0.0001	100.0	1200	0.8802	14.2356
0.0	108.3333	1300	0.8844	14.7966
0.0	116.6667	1400	0.8892	13.6045
0.0	125.0	1500	0.8936	15.4979
0.0	133.3333	1600	0.8976	15.4278
0.0	141.6667	1700	0.8996	14.9369
0.0	150.0	1800	0.9035	14.6564
0.0	158.3333	1900	0.9075	15.0771
0.0	166.6667	2000	0.9139	14.7966
0.0	175.0	2100	0.9200	16.6900
0.0	183.3333	2200	0.9247	15.0771
0.0	191.6667	2300	0.9280	15.4979
0.0	200.0	2400	0.9312	15.9888
0.0	208.3333	2500	0.9350	15.9888
0.0	216.6667	2600	0.9365	15.9888
0.0	225.0	2700	0.9394	15.3576
0.0	233.3333	2800	0.9420	16.4797
0.0	241.6667	2900	0.9447	17.9523
0.0	250.0	3000	0.9462	18.1627
0.0	258.3333	3100	0.9496	16.9004
0.0	266.6667	3200	0.9510	16.4095
0.0	275.0	3300	0.9530	16.4095
0.0	283.3333	3400	0.9540	16.4095
0.0	291.6667	3500	0.9547	16.4095
0.0	300.0	3600	0.9560	16.4797
0.0	308.3333	3700	0.9570	16.4797
0.0	316.6667	3800	0.9581	16.4797
0.0	325.0	3900	0.9588	16.4797
0.0	333.3333	4000	0.9590	16.4797

Framework versions

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