neddamj/whisper-small-patois

Whisper Tiny - Patois

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

• Loss: 1.2488
• Wer: 120.2643

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: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.6086	1.9531	500	1.6058	158.7301
1.0993	3.9062	1000	1.3238	144.8173
0.8516	5.8594	1500	1.2432	137.7567
0.6354	7.8125	2000	1.2213	128.9479
0.522	9.7656	2500	1.2157	124.1094
0.4583	11.7188	3000	1.2275	120.0064
0.3882	13.6719	3500	1.2437	118.9618
0.367	15.625	4000	1.2488	120.2643

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1