whisper-small-En

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

• Loss: 0.7718
• Wer: 16.6597

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_FUSED 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: 5
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.8550	0.1923	10	2.0652	23.5170
1.5730	0.3846	20	1.2286	21.5398
1.1471	0.5769	30	1.0885	18.7631
0.9979	0.7692	40	0.9908	17.5431
0.9493	0.9615	50	0.9206	17.8376
0.7381	1.1538	60	0.8674	17.7955
0.6440	1.3462	70	0.8274	17.0383
0.6470	1.5385	80	0.7976	16.7438
0.6241	1.7308	90	0.7785	16.7438
0.6161	1.9231	100	0.7718	16.6597

Framework versions

• Transformers 5.0.0.dev0
• Pytorch 2.9.0+cu126
• Datasets 2.16.0
• Tokenizers 0.22.1