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

openai/whisper-small

This model is a fine-tuned version of openai/whisper-small on the pphuc25/VietMed-split-8-2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9256
• Wer: 24.4737
• Cer: 20.2163

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
0.5664	1.0	569	0.5651	25.4000	20.5826
0.3901	2.0	1138	0.5406	24.0271	19.6035
0.2343	3.0	1707	0.5885	23.4413	19.3336
0.145	4.0	2276	0.6539	23.9026	20.0783
0.0675	5.0	2845	0.7298	26.7875	22.1606
0.0549	6.0	3414	0.7806	27.5892	22.7204
0.0339	7.0	3983	0.7988	32.2350	31.3005
0.0249	8.0	4552	0.8216	32.5133	28.8528
0.016	9.0	5121	0.8464	27.9590	23.8089
0.0107	10.0	5690	0.8647	27.1060	21.7405
0.0112	11.0	6259	0.8650	28.1896	23.1761
0.0102	12.0	6828	0.8800	27.4831	23.1258
0.0025	13.0	7397	0.8910	25.2828	21.4575
0.0021	14.0	7966	0.8998	26.4397	21.8012
0.0049	15.0	8535	0.8959	26.6081	23.5693
0.0001	16.0	9104	0.9047	24.2907	19.6070
0.0001	17.0	9673	0.9104	25.1400	20.7102
0.0004	18.0	10242	0.9198	24.9350	20.5557
0.0	19.0	10811	0.9243	24.0893	19.7025
0.0	20.0	11380	0.9256	24.4737	20.2163

Framework versions

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