opus-mt-vi-en Translation Model
Property | Value |
---|---|
License | Apache 2.0 |
Developer | Helsinki-NLP |
BLEU Score | 42.8 |
chrF Score | 0.608 |
What is opus-mt-vi-en?
opus-mt-vi-en is a specialized machine translation model developed by Helsinki-NLP for translating Vietnamese text to English. Built on the transformer-align architecture, this model has demonstrated impressive performance with a BLEU score of 42.8 on the Tatoeba test set.
Implementation Details
The model employs a transformer-align architecture with SentencePiece tokenization (spm32k,spm32k) and includes preprocessing normalization. It was trained on Vietnamese (including Vietnamese Hani script) to English translations and was released on June 17, 2020.
- Pre-processing: Normalization + SentencePiece tokenization
- Source languages supported: Vietnamese (vie, vie_Hani)
- Target language: English (eng)
- Training framework: Marian MT
Core Capabilities
- High-quality Vietnamese to English translation
- Support for both standard Vietnamese and Hani script
- Competitive performance metrics (42.8 BLEU, 0.608 chrF)
- Suitable for production deployment via Inference Endpoints
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its specific optimization for Vietnamese-to-English translation, achieving strong performance metrics while maintaining compatibility with both standard Vietnamese and Hani script inputs.
Q: What are the recommended use cases?
The model is ideal for automated translation of Vietnamese content to English, particularly suitable for applications requiring high-quality translations such as document translation, content localization, and cross-lingual information processing.