whisper-tiny-aug-1-april-v1

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5096
• Wer: 89.9851

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.7333	1.0	62	1.5522	103.8030
1.4558	2.0	124	1.4240	106.6315
1.3396	3.0	186	1.3440	105.6706
1.2598	4.0	248	1.2716	116.6870
1.1714	5.0	310	1.1962	110.5021
1.0583	6.0	372	1.0536	115.0223
0.8981	7.0	434	0.8571	100.0812
0.7261	8.0	496	0.6891	98.2812
0.5917	9.0	558	0.5762	93.4091
0.5038	9.8455	610	0.5096	89.9851

Framework versions

• Transformers 4.50.3
• Pytorch 2.5.1+cu121
• Datasets 3.5.0
• Tokenizers 0.21.0