Skip to content
#

opus-mt

Here are 33 public repositories matching this topic...

High-performance, asynchronous translation service leveraging the Helsinki-NLP OPUS-MT zh-en model optimized via CTranslate2. The solution provides a RESTful API endpoint that accepts Chinese text input and returns English translations with sub-second latency. Development supported by DeepSeek.

  • Updated Sep 19, 2026
  • HTML

A LoRA adapter that steers Helsinki-NLP/opus-mt-en-es toward literary prose. 589K trainable parameters — 0.75% of the model — trained on opus_books en-es. 19.99 sacreBLEU against 17.09 for the untouched base model. 2.4 MB; loads on top of the base with PEFT.

  • Updated Oct 25, 2026

Modular monolith serving a LoRA fine-tuned Helsinki-NLP opus-mt-en-es translation model, training about 0.38 percent of parameters on the opus_books corpus, with sentence-aware chunking to avoid silent truncation, deterministic beam-search decoding, a thread-safe TTL cache, proxy-aware rate limiting, and a 193-test CI suite on Docker and Render.

  • Updated Oct 14, 2026
  • Jupyter Notebook

Add this topic to your repo

To associate your repository with the opus-mt topic, visit your repo's landing page and select "manage topics."

Learn more