whisper_large_v2_thi_dataset-with-timestamps-phase2
This model was trained from scratch on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5050
- Cer: 13.9666
- Wer: 23.8423
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: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|
| 0.8271 | 0.9998 | 3132 | 16.4970 | 0.5218 | 28.6456 |
| 0.4996 | 2.0 | 6265 | 17.2551 | 0.4891 | 29.3134 |
| 0.4217 | 2.9995 | 9396 | 14.7599 | 0.4791 | 25.2521 |
| 0.4299 | 3.9998 | 12528 | 0.4967 | 15.5303 | 26.8188 |
| 0.3876 | 5.0 | 15661 | 0.4903 | 15.3058 | 26.2542 |
| 0.3522 | 5.9998 | 18793 | 0.5027 | 14.8503 | 25.4048 |
| 0.3236 | 7.0 | 21926 | 0.5040 | 14.1550 | 24.2675 |
| 0.2984 | 7.9998 | 25058 | 0.5050 | 13.9666 | 23.8423 |
| 0.2782 | 9.0 | 28191 | 0.5180 | 14.2619 | 24.1963 |
| 0.263 | 9.9989 | 31320 | 0.5199 | 14.2882 | 24.0673 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.1.2+cu118
- Datasets 2.19.0
- Tokenizers 0.19.1
