Hanhpt23/whisper-small-GermanMed-full

openai/whisper-small

This model is a fine-tuned version of openai/whisper-small on the Hanhpt23/GermanMed-full dataset. It achieves the following results on the evaluation set:

• Loss: 0.6437
• Wer: 21.9994

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
0.5314	1.0	194	0.5553	45.6238
0.2492	2.0	388	0.5421	25.9694
0.1477	3.0	582	0.5588	33.7962
0.0954	4.0	776	0.5956	27.4915
0.0734	5.0	970	0.5882	24.2312
0.0494	6.0	1164	0.6253	25.1774
0.0334	7.0	1358	0.6412	26.2676
0.026	8.0	1552	0.6175	23.3158
0.0149	9.0	1746	0.6484	22.3696
0.0101	10.0	1940	0.6391	23.1102
0.0083	11.0	2134	0.6371	22.2668
0.0078	12.0	2328	0.6486	22.2154
0.002	13.0	2522	0.6499	22.4725
0.0004	14.0	2716	0.6438	22.4313
0.0019	15.0	2910	0.6381	22.0508
0.0011	16.0	3104	0.6343	22.0817
0.0002	17.0	3298	0.6412	21.6806
0.0001	18.0	3492	0.6428	21.9274
0.0001	19.0	3686	0.6435	22.0200
0.0002	20.0	3880	0.6437	21.9994

Framework versions

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