mabrouk/whisper-small-yo

Whisper Small Yoruba - Mubarak Ganiyu

This model is a fine-tuned version of openai/whisper-small on the Common Voice 13 dataset. It achieves the following results on the evaluation set:

• Loss: 1.2013
• Wer Ortho: 71.5314
• Wer: 51.0134

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.1023	5.88	500	0.8498	73.4063	54.6298
0.0129	11.76	1000	1.0323	76.4662	55.2126
0.0074	17.65	1500	1.1289	71.2914	53.1064
0.005	23.53	2000	1.1822	71.2764	52.1923
0.0048	29.41	2500	1.1737	72.5664	51.9407
0.0044	35.29	3000	1.1774	76.2412	55.2524
0.0041	41.18	3500	1.2009	72.5364	51.3843
0.004	47.06	4000	1.2013	71.5314	51.0134

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3