<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
whisper-tiny-v2
This model is a fine-tuned version of roseman/whisper-tiny-ckb on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.0055
- Wer: 50.9922
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: 64
- 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: 70000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0181 | 4.0 | 10000 | 0.6723 | 54.6795 |
| 0.0039 | 8.01 | 20000 | 0.7814 | 52.9841 |
| 0.0025 | 12.01 | 30000 | 0.8667 | 52.7750 |
| 0.0002 | 16.02 | 40000 | 0.9316 | 52.5431 |
| 0.0009 | 20.02 | 50000 | 0.9454 | 51.6042 |
| 0.0 | 24.02 | 60000 | 0.9938 | 51.0872 |
| 0.0 | 28.03 | 70000 | 1.0055 | 50.9922 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.7.1
- Tokenizers 0.13.2
