Hanhpt23/whisper-base-vietmed-v1

openai/whisper-base

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

• Loss: 1.0352
• Wer: 27.1975

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
0.7081	1.0	569	0.7147	32.6304
0.5097	2.0	1138	0.6779	30.7670
0.3642	3.0	1707	0.6890	30.5144
0.2242	4.0	2276	0.7389	31.4662
0.1221	5.0	2845	0.7970	32.5828
0.07	6.0	3414	0.8480	30.3240
0.0411	7.0	3983	0.8862	29.4380
0.0288	8.0	4552	0.9171	29.9066
0.0199	9.0	5121	0.9572	29.6321
0.0105	10.0	5690	0.9698	28.6473
0.0068	11.0	6259	0.9811	29.5881
0.0084	12.0	6828	0.9985	28.7424
0.0024	13.0	7397	0.9903	29.3355
0.003	14.0	7966	1.0112	27.6588
0.0017	15.0	8535	1.0137	28.7205
0.0004	16.0	9104	1.0185	27.2305
0.0002	17.0	9673	1.0257	27.2964
0.0006	18.0	10242	1.0282	27.2817
0.0002	19.0	10811	1.0336	27.1609
0.0001	20.0	11380	1.0352	27.1975

Framework versions

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