Whisper Large Nepali - Kiran Pantha
This model is a fine-tuned version of openai/whisper-small on the OpenSLR54 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2112
- Wer: 30.2546
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
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2615 | 0.5995 | 500 | 0.2454 | 47.2685 |
| 0.123 | 1.1990 | 1000 | 0.1994 | 39.3287 |
| 0.1145 | 1.7986 | 1500 | 0.1835 | 36.1574 |
| 0.0547 | 2.3981 | 2000 | 0.1813 | 33.7037 |
| 0.0506 | 2.9976 | 2500 | 0.1730 | 32.2454 |
| 0.0204 | 3.5971 | 3000 | 0.1911 | 32.2454 |
| 0.0079 | 4.1966 | 3500 | 0.2009 | 31.6667 |
| 0.0061 | 4.7962 | 4000 | 0.2022 | 30.0926 |
| 0.0022 | 5.3957 | 4500 | 0.2097 | 30.2546 |
| 0.0022 | 5.9952 | 5000 | 0.2112 | 30.2546 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 2.21.0
- Tokenizers 0.19.1