rlad-discovering-abstractions

【效率与协作】Train LLMs to discover domain-specific abstractions (concise representations of procedural and factual knowledge) through two-player RL. An abstraction generator proposes key insights, a solution generator uses them to solve problems, and both receive RL rewards, learning structured reasoning that transfers to harder problems.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准