LightReasoner
https://github.com/hkuds/lightreasonerAbout
ACL 2026 paper exploring whether small language models can teach large language models to reason, with code and training recipes.
Why it made the leaderboard
ACL 2026 work asking whether small models can teach large models to reason — comes with code and training recipes, so you can test the counterintuitive direction of knowledge transfer yourself.
Tags
llmreasoningpost-trainingresearchsmall-models
Tech Stack
Python
Media
Comments (0)
No comments yet
Indexed by a proprietary survey. Corrections welcome.