whisper-v3-large-turbo-jp

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0682
• Wer: 75.0139

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
0.1733	0.2476	1000	0.2069	91.7688
0.0947	0.4953	2000	0.1115	84.9304
0.0829	0.7429	3000	0.0811	78.1616
0.0667	0.9906	4000	0.0682	75.0139

Framework versions

• Transformers 4.52.3
• Pytorch 2.8.0.dev20250518+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1