opus-mt-gl-pt
| Property | Value |
|---|---|
| License | Apache 2.0 |
| Developer | Helsinki-NLP |
| BLEU Score | 57.9 |
| chrF Score | 0.758 |
What is opus-mt-gl-pt?
opus-mt-gl-pt is a specialized neural machine translation model designed to translate text from Galician (gl) to Portuguese (pt). Developed by Helsinki-NLP, this model employs a transformer-align architecture and has demonstrated impressive performance with a BLEU score of 57.9 and a chrF score of 0.758.
Implementation Details
The model utilizes a transformer-align architecture with specific pre-processing steps including normalization and SentencePiece tokenization (spm4k,spm4k). It was trained on the OPUS dataset and released on June 16, 2020.
- Pre-processing: Normalization + SentencePiece (spm4k,spm4k)
- Architecture: Transformer-align
- Training Data: OPUS dataset
- Evaluation Metrics: BLEU (57.9) and chrF (0.758)
Core Capabilities
- High-quality translation from Galician to Portuguese
- Optimized for accuracy with strong evaluation metrics
- Supports inference endpoints for deployment
- Compatible with PyTorch and TensorFlow frameworks
Frequently Asked Questions
Q: What makes this model unique?
This model specializes in Galician to Portuguese translation with exceptional accuracy, demonstrated by its high BLEU score of 57.9. It's particularly notable for using the transformer-align architecture and specialized tokenization.
Q: What are the recommended use cases?
The model is ideal for translating Galician text to Portuguese, particularly useful for content localization, academic translations, and general text translation between these closely related languages. The high BLEU score suggests it's suitable for professional translation tasks.
