metadata
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_6_1
metrics:
- wer
model-index:
- name: Whisper Small Frisian 10m
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 6.1
type: mozilla-foundation/common_voice_6_1
args: 'config: frisian, split: test'
metrics:
- name: Wer
type: wer
value: 75.45431645615172
Whisper Small Frisian 10m
This model is a fine-tuned version of openai/whisper-small on the Common Voice 6.1 dataset. It achieves the following results on the evaluation set:
- Loss: 1.7704
- Wer: 75.4543
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-06
- 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: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0024 | 25.0 | 200 | 1.6444 | 72.9821 |
| 0.0008 | 50.0 | 400 | 1.6951 | 73.5507 |
| 0.0004 | 75.0 | 600 | 1.7404 | 73.4895 |
| 0.0003 | 100.0 | 800 | 1.7631 | 74.6374 |
| 0.0002 | 125.0 | 1000 | 1.7704 | 75.4543 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1