Swecha Gonthuka - Limited Release
This model is a fine-tuned version of openai/whisper-small on the Crowed sourced dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.0768
- Wer: 28.5976
Collaborators
Trained by Naga Budigam
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: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 3000
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0729 | 1.08 | 5000 | 0.0934 | 33.3306 |
| 0.0519 | 2.16 | 10000 | 0.0768 | 28.5976 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0
- Datasets 2.7.1
- Tokenizers 0.13.2