Hanhpt23/whisper-small-engmed-free_E0-8D0-8

openai/whisper-small

This model is a fine-tuned version of openai/whisper-small on the pphuc25/EngMed dataset. It achieves the following results on the evaluation set:

• Loss: 0.0002
• Wer: 18.0420
• Cer: 13.7394

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.7947	1.0	386	0.3297	31.2590	24.7156
0.4419	2.0	772	0.1631	46.7252	46.6811
0.2375	3.0	1158	0.1046	36.4448	33.2804
0.1131	4.0	1544	0.0643	35.7395	32.0761
0.0974	5.0	1930	0.0458	32.6501	28.0269
0.0582	6.0	2316	0.0345	32.8258	26.8794
0.049	7.0	2702	0.0251	30.1177	25.6116
0.0551	8.0	3088	0.0171	26.8918	22.6404
0.0219	9.0	3474	0.0133	24.6637	19.5561
0.0183	10.0	3860	0.0092	21.5865	16.2971
0.0097	11.0	4246	0.0066	27.2316	23.4164
0.0122	12.0	4632	0.0044	18.0929	14.6218
0.0049	13.0	5018	0.0019	19.5776	16.1094
0.0034	14.0	5404	0.0011	17.0281	13.6385
0.001	15.0	5790	0.0006	17.5356	13.4349
0.0031	16.0	6176	0.0004	18.5909	13.6354
0.0006	17.0	6562	0.0003	16.7808	12.7811
0.0002	18.0	6948	0.0002	18.2195	13.7038
0.0006	19.0	7334	0.0002	18.3330	13.9521
0.0001	20.0	7720	0.0002	18.0420	13.7394

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.19.1