WhisperTiny15hCommonVoice5hAugmentV2

This model is a fine-tuned version of OpenAI/whisper-tiny on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.9025
• Wer: 0.4359

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: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8054	0.6693	500	0.9866	0.4886
0.5822	1.3387	1000	0.9179	0.4688
0.4991	2.0080	1500	0.8808	0.4333
0.3864	2.6774	2000	0.8846	0.4295
0.2676	3.3467	2500	0.8738	0.4407
0.2598	4.0161	3000	0.8789	0.4324
0.1893	4.6854	3500	0.8829	0.4276
0.1514	5.3548	4000	0.9006	0.4256
0.1451	6.0241	4500	0.9021	0.4345
0.137	6.6934	5000	0.9025	0.4359

Framework versions

• Transformers 4.48.3
• Pytorch 2.6.0
• Datasets 3.3.0
• Tokenizers 0.21.0