whisper-tiny-kor_eng_tiny_pu_ob

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

• Loss: 1.9687
• Cer: 39.2172

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
2.988	0.4292	100	2.5841	48.9257
2.6577	0.8584	200	2.2528	47.3699
2.2158	1.2876	300	1.9927	45.9930
1.9737	1.7167	400	1.8063	43.4010
1.6909	2.1459	500	1.6837	42.3744
1.4011	2.5751	600	1.5948	40.3783
1.4575	3.0043	700	1.5332	41.1149
1.0689	3.4335	800	1.5208	41.0101
1.0004	3.8627	900	1.5151	41.1276
0.8509	4.2918	1000	1.5661	39.5824
0.7546	4.7210	1100	1.5217	40.2735
0.6646	5.1502	1200	1.5983	40.1359
0.5452	5.5794	1300	1.5892	40.2470
0.5251	6.0086	1400	1.6012	38.6901
0.3632	6.4378	1500	1.6482	38.2668
0.3644	6.8670	1600	1.6629	39.3707
0.2652	7.2961	1700	1.7181	39.6490
0.2686	7.7253	1800	1.7140	40.3476
0.224	8.1545	1900	1.7408	40.1221
0.1659	8.5837	2000	1.7560	39.5263
0.1652	9.0129	2100	1.7754	38.5716
0.1101	9.4421	2200	1.7880	38.3451
0.1125	9.8712	2300	1.7915	38.9537
0.0911	10.3004	2400	1.8303	38.4584
0.0742	10.7296	2500	1.8278	39.1357
0.0652	11.1588	2600	1.8553	39.6914
0.0496	11.5880	2700	1.8616	39.1294
0.0505	12.0172	2800	1.8687	38.3663
0.0335	12.4464	2900	1.8867	39.0796
0.034	12.8755	3000	1.8918	38.8129
0.0262	13.3047	3100	1.9148	39.5887
0.0227	13.7339	3200	1.9158	38.6468
0.021	14.1631	3300	1.9206	38.5441
0.017	14.5923	3400	1.9390	39.3093
0.0156	15.0215	3500	1.9407	38.6277
0.0127	15.4506	3600	1.9545	38.8214
0.0118	15.8798	3700	1.9610	39.0892
0.0111	16.3090	3800	1.9660	38.6478
0.0096	16.7382	3900	1.9681	39.4606
0.0092	17.1674	4000	1.9687	39.2172

Framework versions

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