一作ACL 2026Poster· Natural Language Processing
Beyond Experience Retrieval: Learning to Generate Utility-Optimized Structured Experience for Frozen LLM s
Xuancheng Li · Haitao Li · Yujia Zhou · Yiqun LIU · Qingyao Ai
摘要
Large language models (LLMs) are largely static and often redo reasoning or repeat mistakes. Prior experience reuse typically relies on external retrieval, which is similarity-based, can introduce noise, and adds latency. We introduce SEAM ( S tructured E xperience A dapter M odule), a lightweigh...