Hanhpt23/whisper-tiny-chinesemed-v1

openai/whisper-tiny

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

• Loss: 4.4161
• Wer: 116.6667

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
2.0985	1.0	371	2.8659	188.8889
1.6605	2.0	742	2.8279	102.2222
1.2794	3.0	1113	3.0839	110.0000
0.7344	4.0	1484	3.4284	107.7778
0.4201	5.0	1855	3.6794	142.2222
0.1934	6.0	2226	3.8480	111.1111
0.0892	7.0	2597	4.0445	112.2222
0.0609	8.0	2968	4.1135	107.7778
0.0421	9.0	3339	4.2117	116.6667
0.0264	10.0	3710	4.3500	118.8889
0.0168	11.0	4081	4.2340	110.0000
0.0098	12.0	4452	4.3287	115.5556
0.0053	13.0	4823	4.3447	111.1111
0.0012	14.0	5194	4.3370	118.8889
0.0031	15.0	5565	4.3509	118.8889
0.0012	16.0	5936	4.3898	115.5556
0.0003	17.0	6307	4.3852	115.5556
0.0002	18.0	6678	4.4175	116.6667
0.0002	19.0	7049	4.4100	116.6667
0.0002	20.0	7420	4.4161	116.6667

Framework versions

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