metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- yashtiwari/PaulMooney-Medical-ASR-Data
metrics:
- wer
model-index:
- name: Whisper Medium Medical
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Medical ASR
type: yashtiwari/PaulMooney-Medical-ASR-Data
metrics:
- name: Wer
type: wer
value: 16.02703355056722
Whisper Medium Medical
This model is a fine-tuned version of openai/whisper-medium on the Medical ASR dataset. It achieves the following results on the evaluation set:
- Loss: 0.0577
- Wer: 16.0270
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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.4859 | 0.5405 | 100 | 0.1945 | 15.5926 |
| 0.1037 | 1.0811 | 200 | 0.0849 | 12.5754 |
| 0.0558 | 1.6216 | 300 | 0.0633 | 17.3787 |
| 0.0244 | 2.1622 | 400 | 0.0631 | 13.7581 |
| 0.0123 | 2.7027 | 500 | 0.0577 | 16.0270 |
Framework versions
- Transformers 4.45.1
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.0