metadata
language:
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- divakaivan/glaswegian_audio
metrics:
- wer
model-index:
- name: Glaswegian_Whisper_001
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Glaswegian audio
type: divakaivan/glaswegian_audio
config: default
split: train
args: default
metrics:
- name: Wer
type: wer
value: 24.51154529307282
Glaswegian_Whisper_001
This model is a fine-tuned version of openai/whisper-small on the Glaswegian audio dataset. It achieves the following results on the evaluation set:
- Loss: 0.9462
- Wer: 24.5115
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 |
|---|---|---|---|---|
| 0.003 | 21.2766 | 1000 | 0.8488 | 31.4387 |
| 0.0029 | 42.5532 | 2000 | 0.9056 | 30.7282 |
| 0.0001 | 63.8298 | 3000 | 0.9364 | 24.4227 |
| 0.0001 | 85.1064 | 4000 | 0.9462 | 24.5115 |
Framework versions
- Transformers 4.42.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1