SverreNystad/whisper-tiny-english

Whisper Small En - Sverre Nystad

This model is a fine-tuned version of openai/whisper-tiny on the PolyAI/minds14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9366
• Wer Ortho: 0.3226
• Wer: 0.3241

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: 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: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.0001	17.8571	500	0.7530	0.3233	0.3235
0.0	35.7143	1000	0.7949	0.3183	0.3200
0.0	53.5714	1500	0.8247	0.3202	0.3217
0.0	71.4286	2000	0.8507	0.3245	0.3264
0.0	89.2857	2500	0.8748	0.3214	0.3235
0.0	107.1429	3000	0.8950	0.3208	0.3229
0.0	125.0	3500	0.9155	0.3165	0.3188
0.0	142.8571	4000	0.9366	0.3226	0.3241

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1