en_nso_ukuxhumana_model
| Property | Value |
|---|---|
| Base Model | Helsinki-NLP/opus-mt-en-nso |
| BLEU Score | 18.9287 |
| Author | kabelomalapane |
| Model Link | Hugging Face |
What is en_nso_ukuxhumana_model?
The en_nso_ukuxhumana_model is a specialized machine translation model fine-tuned from Helsinki-NLP's opus-mt-en-nso base model. It's designed specifically for English to Northern Sotho (Sepedi) translation, showing significant improvements with a BLEU score increase from 12.23 to 18.93.
Implementation Details
The model was trained using carefully selected hyperparameters, including Adam optimizer with a learning rate of 2e-05, batch sizes of 32 for training and 64 for evaluation, and a linear learning rate scheduler. The training process ran for 3 epochs with a seed value of 42.
- Fine-tuned using PyTorch 1.10.2 and Transformers 4.16.2
- Implements linear learning rate scheduling
- Achieved a final loss of 2.8482
Core Capabilities
- English to Northern Sotho translation
- Improved translation quality compared to base model
- Optimized for batch processing
- Suitable for production deployment
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its significant improvement in BLEU score (from 12.23 to 18.93) for English to Northern Sotho translation, making it particularly effective for this specific language pair.
Q: What are the recommended use cases?
The model is best suited for English to Northern Sotho translation tasks, particularly in scenarios requiring batch processing with its optimized evaluation batch size of 64.
