opus-mt-tc-big-he-en
| Property | Value |
|---|---|
| Model Architecture | Transformer-big |
| Source Language | Hebrew (he) |
| Target Language | English (en) |
| Release Date | 2022-03-13 |
| BLEU Score | 53.8 (Tatoeba test set) |
| Paper | OPUS-MT Paper |
What is opus-mt-tc-big-he-en?
opus-mt-tc-big-he-en is a state-of-the-art neural machine translation model designed specifically for translating Hebrew text to English. Developed by Helsinki-NLP as part of the OPUS-MT project, this model leverages the transformer-big architecture and achieves impressive performance metrics, including a 53.8 BLEU score on the Tatoeba test set.
Implementation Details
The model implements a transformer-big architecture and utilizes SentencePiece tokenization with 32k vocabulary size for both source and target languages. It was trained on the opusTCv20210807+bt dataset and has been converted from Marian NMT to PyTorch using the Hugging Face transformers library.
- Built using Marian NMT framework
- Implements SentencePiece tokenization (spm32k)
- Converted to PyTorch for wider accessibility
- Trained on comprehensive OPUS dataset
Core Capabilities
- High-quality Hebrew to English translation
- Achieves 68.565 chr-F score on Tatoeba test
- Handles diverse text inputs effectively
- Easily integrable with Hugging Face transformers
- Supports batch translation
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its impressive performance metrics and specific optimization for Hebrew-to-English translation, achieving a BLEU score of 53.8 on the Tatoeba test set. It's part of the larger OPUS-MT initiative to democratize machine translation across many languages.
Q: What are the recommended use cases?
The model is ideal for applications requiring Hebrew to English translation, including content localization, document translation, and automated translation services. It's particularly suitable for production environments due to its implementation in PyTorch and integration with the transformers library.
