一作ACL 2026Findings· Findings of ACL
Eval- RAR : Evaluation-Driven Retrieval-Augmented Reasoning via Reinforcement Learning
Heng Yu · Rui Li · Qi Liu · Wenjun Feng · Junfeng Kang · Yi Zhan
摘要
Retrieval-augmented generation (RAG) effectively extends the knowledge boundaries of large language models (LLMs) for complex tasks, yet current paradigms typically optimize for an interleaving of reasoning and retrieval, where models fail to critically evaluate retrieved information against the...