Curiousfox/whisper-small_new_ver1

Hokkien-to-Tai Lo Whisper ver 1.1

This model is a fine-tuned version of openai/whisper-small on the Common Voice 16.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7563
• Wer: 133.7227

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-06
• train_batch_size: 8
• eval_batch_size: 16
• 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: 1000
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.7198	0.5135	800	1.4813	136.8207
1.1293	1.0270	1600	1.0268	128.2885
0.9323	1.5404	2400	0.8802	131.1833
0.7797	2.0539	3200	0.8011	132.2499
0.6692	2.5674	4000	0.7563	133.7227

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0