Whisper Small Vietnamese
This model is a fine-tuned version of openai/whisper-small on the google/fleurs vi_vn dataset. It achieves the following results on the evaluation set:
- Loss: 0.4476
- Wer: 18.3051
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-06
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0083 | 86.0 | 1000 | 0.4476 | 18.3051 |
| 0.0022 | 173.0 | 2000 | 0.4754 | 18.8086 |
| 0.001 | 260.0 | 3000 | 0.4970 | 18.8278 |
| 0.0006 | 347.0 | 4000 | 0.5153 | 19.5042 |
| 0.0004 | 434.0 | 5000 | 0.5331 | 19.4081 |
| 0.0003 | 521.0 | 6000 | 0.5482 | 19.5042 |
| 0.0002 | 608.0 | 7000 | 0.5638 | 19.3659 |
| 0.0001 | 695.0 | 8000 | 0.5755 | 19.6195 |
| 0.0001 | 782.0 | 9000 | 0.5862 | 19.6503 |
| 0.0001 | 869.0 | 10000 | 0.5902 | 19.6349 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.2.dev0
- Tokenizers 0.15.0