nicolaakmal/whisper-tiny-med-en

Whisper Tiny En - Medical

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

• Loss: 0.4793
• Wer: 16.6120

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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0009	20.8333	1000	0.4294	16.4641
0.0002	41.6667	2000	0.4591	16.5027
0.0001	62.5	3000	0.4738	16.6136
0.0001	83.3333	4000	0.4793	16.6120

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0