Whisper Tiny Pashto
This model is a fine-tuned version of openai/whisper-tiny on the google/fleurs ps_af dataset. It achieves the following results on the evaluation set:
- Loss: 2.0279
- Wer: 497.2231
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: 64
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 700
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.9393 | 14.29 | 100 | 2.9025 | 369.2645 |
| 2.7124 | 28.57 | 200 | 2.7165 | 407.9222 |
| 2.4773 | 42.86 | 300 | 2.4649 | 449.5006 |
| 2.3142 | 57.14 | 400 | 2.3466 | 473.6002 |
| 2.2942 | 71.43 | 500 | 2.3101 | 485.9867 |
| 2.0644 | 85.71 | 600 | 2.0926 | 491.4800 |
| 1.9799 | 100.0 | 700 | 2.0279 | 497.2231 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.1+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2