whisper-small-finetuned-iwslt-irish-transfer-welsh-synthetic
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: 3.5157
- Wer: 95.2566
- Bleu: 8.6311
- Chrf++: 28.9864
- Comet: 0.5371
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- training_steps: 15000
- mixed_precision_training: Native AMP
- label_smoothing_factor: 0.1
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Bleu | Chrf++ | Comet |
|---|---|---|---|---|---|---|---|
| 3.3397 | 0.4264 | 2500 | 3.5318 | 106.3878 | 5.8424 | 23.6528 | 0.4968 |
| 2.5971 | 0.8528 | 5000 | 3.4333 | 105.3939 | 6.4818 | 25.8206 | 0.5089 |
| 2.3176 | 1.2792 | 7500 | 3.4041 | 99.2411 | 8.2264 | 28.0766 | 0.5317 |
| 2.1728 | 1.7056 | 10000 | 3.4221 | 94.1724 | 8.2101 | 27.7234 | 0.5297 |
| 2.0561 | 2.1320 | 12500 | 3.5006 | 93.9827 | 8.6366 | 29.1896 | 0.5328 |
| 1.9125 | 2.5584 | 15000 | 3.5157 | 95.2566 | 8.6311 | 28.9864 | 0.5371 |
Framework versions
- Transformers 4.52.0
- Pytorch 2.12.1+cu130
- Datasets 2.21.0
- Tokenizers 0.21.4