metadata
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper small TW - AlanDlink
results: []
Whisper small TW - AlanDlink
This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2367
- Wer: 149.6566
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 8000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.084 | 1.33 | 1000 | 0.1997 | 164.9495 |
| 0.0329 | 2.67 | 2000 | 0.1929 | 157.7172 |
| 0.0085 | 4.0 | 3000 | 0.2002 | 185.5758 |
| 0.0019 | 5.33 | 4000 | 0.2076 | 209.1717 |
| 0.0032 | 6.67 | 5000 | 0.2236 | 185.9394 |
| 0.0022 | 8.0 | 6000 | 0.2272 | 148.3434 |
| 0.0005 | 9.33 | 7000 | 0.2343 | 154.9495 |
| 0.0004 | 10.67 | 8000 | 0.2367 | 149.6566 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu121
- Datasets 2.16.0
- Tokenizers 0.15.0