Get Started
Home
Topics
Search
Library
Research questionHow should multilingual models choose the amount of English context used for reasoning in lower-resource languages?English context can compensate for weaker representations in lower-resource languages, but redundant translation may introduce interference or overconfidence. The useful context choice can therefore vary by language and input.
AI
Evaluation & Benchmarks
Inference Optimization
Natural Language Processing
Reasoning
Latest papersRecent research connected to this question, newest first.Choosing the Right Language Mode at Inference Time for Multilingual ReliabilityThe source evaluates LLaMA and Qwen models across low- to mid-resource languages using target-only, English-only, and bilingual settings, measuring accuracy and calibration. It also tests a training-free adaptive inference approach; the evidence is limited to the studied models, languages, and tasks.research paper · Sep 4, 2026
Related questions
How can large language models allocate reasoning computation to preserve accuracy under limited training and inference budgets?Do multilingual attention heads that retrieve context also control transitions into the target language during reasoning?How can models adapt to low-resource languages without damaging source-language and related-task performance?How should open reasoning language models be selected under prompt and deployment-resource trade-offs?