Whisper Medium Tajik
This model is a fine-tuned version of openai/whisper-medium on the google/fleurs tg_tj dataset. It achieves the following results on the evaluation set:
- Loss: 0.9217
- Wer: 23.1530
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: 32
- eval_batch_size: 16
- 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: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0016 | 66.0 | 1000 | 0.6929 | 24.2993 |
| 0.0001 | 133.0 | 2000 | 0.8054 | 23.3022 |
| 0.0001 | 199.0 | 3000 | 0.8652 | 23.2237 |
| 0.0 | 266.0 | 4000 | 0.9019 | 23.2394 |
| 0.0 | 333.0 | 5000 | 0.9217 | 23.1530 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2
- Downloads last month
- 11
Dataset used to train muhtasham/whisper-medium-tg_tj
Evaluation results
- Wer on google/fleurstest set self-reported23.153