ras0k/whisper-rap-queb-v10

Whisper Quebec Rap - ras0k

This model is a fine-tuned version of openai/whisper-large-v3 on the Genius-Quebec-Rap-Top-500 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8482
• Wer: 29.8209

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: 4e-05
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 200
• training_steps: 1200

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0	0	0.7656	58.2465
0.2658	5.8824	200	0.5356	31.8419
0.0230	11.7647	400	0.7306	30.2070
0.0051	17.6471	600	0.8482	29.8209
0.0044	23.5294	800	0.8627	29.9070
0.0040	29.4118	1000	0.8684	30.0140
0.0042	35.2941	1200	0.8689	29.8837

Framework versions

• Transformers 5.3.0
• Pytorch 2.10.0+cu128
• Datasets 4.6.1
• Tokenizers 0.22.2