whisper-tiny-kor_eng_tiny_pu_op

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

• Loss: 1.9500
• Cer: 43.5341

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
2.9775	0.2353	100	2.6186	49.4337
2.5067	0.4706	200	2.3685	48.1512
2.3456	0.7059	300	2.1790	47.0022
2.1658	0.9412	400	2.0273	46.6718
1.8787	1.1765	500	1.9346	45.9100
1.7335	1.4118	600	1.8634	45.4311
1.7722	1.6471	700	1.8020	45.7598
1.7048	1.8824	800	1.7389	45.0860
1.4935	2.1176	900	1.7298	44.2391
1.3435	2.3529	1000	1.7012	44.1586
1.2691	2.5882	1100	1.6800	43.7677
1.3402	2.8235	1200	1.6574	43.2258
1.2012	3.0588	1300	1.6678	43.5003
1.0158	3.2941	1400	1.6694	43.3113
1.0363	3.5294	1500	1.6404	43.2867
1.0837	3.7647	1600	1.6470	43.6563
1.0053	4.0	1700	1.6295	43.2120
0.826	4.2353	1800	1.6789	43.0088
0.7988	4.4706	1900	1.6878	43.3480
0.7784	4.7059	2000	1.6663	42.6200
0.8292	4.9412	2100	1.6695	43.1290
0.661	5.1765	2200	1.7367	42.9296
0.6549	5.4118	2300	1.7531	43.5070
0.6208	5.6471	2400	1.7310	42.8540
0.6689	5.8824	2500	1.7359	42.8862
0.5869	6.1176	2600	1.7936	43.9025
0.4903	6.3529	2700	1.8107	43.1336
0.4825	6.5882	2800	1.8106	42.7422
0.525	6.8235	2900	1.8052	43.2037
0.4975	7.0588	3000	1.8393	43.6400
0.4242	7.2941	3100	1.8622	43.2650
0.3849	7.5294	3200	1.8718	42.8411
0.4052	7.7647	3300	1.8784	43.2975
0.4213	8.0	3400	1.8789	43.4665
0.3175	8.2353	3500	1.9241	43.3201
0.3152	8.4706	3600	1.9257	43.0927
0.3611	8.7059	3700	1.9319	43.9033
0.3437	8.9412	3800	1.9332	43.9659
0.2935	9.1765	3900	1.9499	43.5616
0.2915	9.4118	4000	1.9500	43.5341

Framework versions

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