mt5-eng2nep
Model description
This model is a fine-tuned version of google/mt5-base trained on "Nepali Translator" dataset. This fine-tuned model is for translating English to Nepali language (with limited capability)
Training and evaluation data
- "Nepali Translator" dataset: (https://github.com/sharad461/nepali-translator)
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 16
- 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
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.8424 | 0.26 | 500 | 2.9765 |
| 3.2655 | 0.53 | 1000 | 2.5221 |
| 2.9425 | 0.79 | 1500 | 2.3160 |
| 2.7533 | 1.05 | 2000 | 2.2170 |
| 2.7147 | 1.32 | 2500 | 2.1432 |
| 2.6331 | 1.58 | 3000 | 2.0872 |
| 2.5863 | 1.84 | 3500 | 2.0526 |
| 2.492 | 2.11 | 4000 | 2.0274 |
| 2.4757 | 2.37 | 4500 | 1.9948 |
| 2.4423 | 2.63 | 5000 | 1.9855 |
| 2.4339 | 2.9 | 5500 | 1.9645 |
| 2.3955 | 3.16 | 6000 | 1.9530 |
| 2.372 | 3.42 | 6500 | 1.9454 |
| 2.3661 | 3.69 | 7000 | 1.9410 |
| 2.3553 | 3.95 | 7500 | 1.9387 |
Framework versions
- Transformers 4.13.0
- Pytorch 1.11.0
- Datasets 1.16.1
- Tokenizers 0.10.3