Hanhpt23/whisper-medium-vietmed-v1

openai/whisper-medium

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

• Loss: 0.9216
• Wer: 20.0549

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
0.5881	1.0	569	0.5795	27.0181
0.3919	2.0	1138	0.5413	21.8451
0.2496	3.0	1707	0.6104	22.6213
0.1324	4.0	2276	0.6854	23.5695
0.0898	5.0	2845	0.7554	22.0245
0.0642	6.0	3414	0.7739	21.6804
0.0366	7.0	3983	0.8119	21.2667
0.026	8.0	4552	0.8180	21.4058
0.0243	9.0	5121	0.8560	22.1893
0.0168	10.0	5690	0.8636	21.3326
0.0203	11.0	6259	0.8314	20.8091
0.0101	12.0	6828	0.8892	21.3106
0.0093	13.0	7397	0.8793	21.2594
0.0046	14.0	7966	0.8985	20.4613
0.0026	15.0	8535	0.8907	20.5272
0.0021	16.0	9104	0.9033	20.3807
0.0029	17.0	9673	0.9153	20.2014
0.0005	18.0	10242	0.9080	20.2746
0.0001	19.0	10811	0.9195	20.1135
0.0002	20.0	11380	0.9216	20.0549

Framework versions

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