whisper-small-finetuned-iwslt-irish-welsh-curriculum-no-frozen-decoder
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: 5.7829
- Wer: 101.3191
- Bleu: 0.8463
- Chrf++: 13.0678
- Comet: 0.4192
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: 5000
- mixed_precision_training: Native AMP
- label_smoothing_factor: 0.1
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Bleu | Chrf++ | Comet |
|---|---|---|---|---|---|---|---|
| 1.8291 | 2.6709 | 1250 | 4.9555 | 104.6440 | 1.0968 | 13.9231 | 0.4142 |
| 1.5586 | 5.3419 | 2500 | 5.2521 | 105.5295 | 0.8431 | 14.0858 | 0.4130 |
| 1.4879 | 8.0128 | 3750 | 5.5397 | 104.8699 | 0.7676 | 13.4909 | 0.4183 |
| 1.4306 | 10.6838 | 5000 | 5.7829 | 101.3191 | 0.8463 | 13.0678 | 0.4192 |
Framework versions
- Transformers 4.52.0
- Pytorch 2.12.1+cu130
- Datasets 2.21.0
- Tokenizers 0.21.4