whisper-small_vaani_62089_ratio_alldata_lr_1e-4_lgid_None_epochs_100_FT
This model is a fine-tuned version of openai/whisper-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5939
- Global Wer: 13.6581
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: 0.0001
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 500
- num_epochs: 100
Training results
| Training Loss | Epoch | Step | Validation Loss | Global Wer |
|---|---|---|---|---|
| 0.9965 | 1.0 | 540 | 0.4689 | 15.1661 |
| 0.6876 | 2.0 | 1080 | 0.4588 | 17.5198 |
| 0.4263 | 3.0 | 1620 | 0.4606 | 13.0594 |
| 0.3000 | 4.0 | 2160 | 0.4810 | 12.7750 |
| 0.2317 | 5.0 | 2700 | 0.5081 | 13.7405 |
| 0.1846 | 6.0 | 3240 | 0.5283 | 12.2250 |
| 0.1528 | 7.0 | 3780 | 0.5628 | 13.9500 |
| 0.1357 | 8.0 | 4320 | 0.5672 | 14.2756 |
| 0.1200 | 9.0 | 4860 | 0.5939 | 13.6581 |
Framework versions
- Transformers 5.0.0.dev0
- Pytorch 2.9.0+cu126
- Datasets 3.6.0
- Tokenizers 0.22.2