w2v-bert-2.0-krd-colab-CV16.0
This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2704
- Wer: 0.2306
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.283 | 0.7979 | 300 | 0.3271 | 0.3871 |
| 0.2931 | 1.5957 | 600 | 0.2957 | 0.3468 |
| 0.2358 | 2.3936 | 900 | 0.2746 | 0.3299 |
| 0.1842 | 3.1915 | 1200 | 0.2473 | 0.2846 |
| 0.1532 | 3.9894 | 1500 | 0.2257 | 0.2632 |
| 0.1198 | 4.7872 | 1800 | 0.2403 | 0.2600 |
| 0.1027 | 5.5851 | 2100 | 0.2239 | 0.2513 |
| 0.0837 | 6.3830 | 2400 | 0.2310 | 0.2591 |
| 0.0678 | 7.1809 | 2700 | 0.2295 | 0.2402 |
| 0.0527 | 7.9787 | 3000 | 0.2428 | 0.2334 |
| 0.0374 | 8.7766 | 3300 | 0.2448 | 0.2347 |
| 0.0298 | 9.5745 | 3600 | 0.2704 | 0.2306 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.1+cu118
- Datasets 2.19.2
- Tokenizers 0.19.1