metadata
language:
- nan
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_16_1
- mozilla-foundation/common_voice_15_0
model-index:
- name: Whisper Small Taiwanese
results: []
metrics:
- cer
pipeline_tag: automatic-speech-recognition
Whisper Small Taiwanese
This model is a fine-tuned version of openai/whisper-small on the Common Voice 16.1 and the Common Voice 15.0 datasets. It achieves the following results on the evaluation set:
- Loss: 0.3658
- Cer: 29.0572
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: 100
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss |
|---|---|---|---|---|
| 0.2295 | 1.13 | 3500 | 32.4577 | 0.4268 |
| 0.179 | 1.29 | 4000 | 32.1109 | 0.4088 |
| 0.161 | 1.45 | 4500 | 30.9241 | 0.3923 |
| 0.1607 | 1.61 | 5000 | 30.1640 | 0.3840 |
| 0.1336 | 1.77 | 5500 | 29.7573 | 0.3783 |
| 0.1258 | 1.93 | 6000 | 29.5906 | 0.3736 |
| 0.0725 | 2.09 | 6500 | 30.3507 | 0.3842 |
| 0.0692 | 2.25 | 7000 | 30.0973 | 0.3776 |
| 0.0635 | 2.41 | 7500 | 29.5106 | 0.3740 |
| 0.053 | 2.57 | 8000 | 29.0772 | 0.3706 |
| 0.0441 | 2.73 | 8500 | 28.5238 | 0.3656 |
| 0.0427 | 2.89 | 9000 | 29.0572 | 0.3658 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2