Wave2Vec2-Bert2.0 - Kiran Pantha
This model is a fine-tuned version of kiranpantha/w2v-bert-2.0-nepali on the OpenSLR54 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4058
- Wer: 0.4307
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.7246 | 0.15 | 300 | 0.5189 | 0.5402 |
| 0.6721 | 0.3 | 600 | 0.6084 | 0.5423 |
| 0.6956 | 0.45 | 900 | 0.5712 | 0.5412 |
| 0.6341 | 0.6 | 1200 | 0.4997 | 0.5105 |
| 0.6119 | 0.75 | 1500 | 0.5008 | 0.5148 |
| 0.564 | 0.9 | 1800 | 0.4627 | 0.4793 |
| 0.5416 | 1.05 | 2100 | 0.4767 | 0.4734 |
| 0.4569 | 1.2 | 2400 | 0.4754 | 0.4651 |
| 0.4768 | 1.35 | 2700 | 0.4420 | 0.4702 |
| 0.438 | 1.5 | 3000 | 0.4563 | 0.4614 |
| 0.4337 | 1.65 | 3300 | 0.4290 | 0.4543 |
| 0.447 | 1.8 | 3600 | 0.4081 | 0.4392 |
| 0.4108 | 1.95 | 3900 | 0.4058 | 0.4307 |
Framework versions
- Transformers 4.45.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 2.21.0
- Tokenizers 0.19.1