metadata
language:
- en
license: apache-2.0
base_model: openai/whisper-small.en
tags:
- generated_from_trainer
datasets:
- Hani89/medical_asr_recording_dataset
metrics:
- wer
model-index:
- name: English Whisper Model
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Medical
type: Hani89/medical_asr_recording_dataset
args: 'split: test'
metrics:
- name: Wer
type: wer
value: 6.681238615664845
English Whisper Model
This model is a fine-tuned version of openai/whisper-small.en on the Medical dataset. It achieves the following results on the evaluation set:
- Loss: 0.1085
- Wer: 6.6812
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- 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 |
|---|---|---|---|---|
| 0.0268 | 3.0030 | 1000 | 0.1019 | 6.4189 |
| 0.0017 | 6.0060 | 2000 | 0.1010 | 5.6903 |
| 0.0012 | 9.0090 | 3000 | 0.1064 | 6.6302 |
| 0.0001 | 12.0120 | 4000 | 0.1085 | 6.6812 |
Framework versions
- Transformers 4.42.2
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1