whisper-tiny-kor_eng_tiny_pu_is

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

• Loss: 3.4002
• Cer: 58.0671

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
5.0851	16.6667	100	3.7426	64.2745
1.3343	33.3333	200	2.4317	57.6480
0.1055	50.0	300	2.5515	53.7192
0.0176	66.6667	400	2.4245	56.8622
0.0138	83.3333	500	2.3513	48.4809
0.0076	100.0	600	2.4115	50.1048
0.0062	116.6667	700	2.4019	51.1262
0.0009	133.3333	800	2.4536	50.4453
0.0004	150.0	900	2.5163	53.8240
0.0003	166.6667	1000	2.6062	54.6621
0.0003	183.3333	1100	2.6740	51.9906
0.0003	200.0	1200	2.7592	48.7428
0.0003	216.6667	1300	2.8464	52.3049
0.0003	233.3333	1400	2.9159	50.3667
0.0002	250.0	1500	3.0110	53.1168
0.0002	266.6667	1600	3.1017	55.9193
0.0002	283.3333	1700	3.1303	50.7334
0.0002	300.0	1800	3.2130	51.4144
0.0002	316.6667	1900	3.2193	59.0885
0.0002	333.3333	2000	3.2941	52.0430
0.0001	350.0	2100	3.2525	52.2787
0.0001	366.6667	2200	3.2935	53.5621
0.0001	383.3333	2300	3.3069	54.1121
0.0001	400.0	2400	3.2937	54.5312
0.0001	416.6667	2500	3.3400	59.5862
0.0001	433.3333	2600	3.3268	53.6668
0.0001	450.0	2700	3.3293	55.4741
0.0001	466.6667	2800	3.3424	54.9240
0.0001	483.3333	2900	3.3723	53.7454
0.0001	500.0	3000	3.3718	56.7313
0.0001	516.6667	3100	3.3704	57.0456
0.0001	533.3333	3200	3.3880	56.0765
0.0001	550.0	3300	3.3886	56.3646
0.0001	566.6667	3400	3.3961	55.3693
0.0001	583.3333	3500	3.4012	55.2383
0.0001	600.0	3600	3.3952	56.8360
0.0001	616.6667	3700	3.4087	56.4955
0.0001	633.3333	3800	3.4031	58.0671
0.0001	650.0	3900	3.4041	58.3028
0.0001	666.6667	4000	3.4002	58.0671

Framework versions

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