WhisperTiny20hCommonVoice10hAugment

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: 0.8526
• Wer: 0.4173

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
0.8867	0.4444	500	1.0354	0.4892
0.7794	0.8889	1000	0.9536	0.4646
0.6075	1.3333	1500	0.9055	0.4392
0.5813	1.7778	2000	0.8904	0.4338
0.4598	2.2222	2500	0.8724	0.4262
0.4365	2.6667	3000	0.8647	0.4259
0.3813	3.1111	3500	0.8568	0.4188
0.3842	3.5556	4000	0.8605	0.4251
0.3735	4.0	4500	0.8560	0.4221
0.3411	4.4444	5000	0.8526	0.4173

Framework versions

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