This is the official repository for the paper:
Learning When to Translate for Multilingual Reasoning
Deokhyung Kang, Hyounghun Kim, Gary Geunbae Lee
POSTECH, South Korea
This repository provides the code and resources for Luar, a Language Understanding Boundary-aware Reinforcement learning framework for multilingual reasoning.
Luar trains reasoning language models to selectively invoke translation only when their direct understanding of the original input is unreliable, improving multilingual reasoning performance while avoiding unnecessary translation.
The codebase is currently being organized and will be released within June 2026. Stay tuned!