Agoyr/output_dir

Whisper Small Ja Custom

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

• Loss: 0.4312
• Wer: 88.0342

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
0.1175	2.7322	1000	0.3352	84.7578
0.0102	5.4645	2000	0.3849	84.3305
0.0019	8.1967	3000	0.4231	87.3219
0.0014	10.9290	4000	0.4312	88.0342

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0