元认知行为调节用于多跳问答的大型语言模型
Distilling Metacognitive Behaviors through Agentic Editing
AI总结：
本文提出元认知行为调节（MBT）框架，通过五阶段结构提升多跳问答任务的准确性和效率，减少冗余并保持推理轨迹稳定。
摘要：
大型语言模型（LLMs）在多跳问答任务中即使推理轨迹包含正确中间结论仍常产生错误答案，本文归因于自我调节能力不足而非推理能力不足。提出元认知行为调节（MBT）框架，通过五阶段结构提升推理轨迹质量，MBT-S和MBT-R两种形式在HotpotQA、MuSiQue和2WikiMultiHopQA数据集上取得最高Accuracy-Efficiency Score（AES）。MBT在保持轨迹简短稳定的同时提升任务准确率，穆西克数据集上的平均响应长度比基线方法减少一个数量级，退化次数也相应减少。引入Reach-Redundancy Profile（RRP）和长度感知的元认知质量指数（MQI）两个新指标，MBT在所有模型规模下均实现最早答案到达、最低冗余和最丰富的阶段行为。
英文摘要：
Distilling long reasoning traces from a teacher model into a smaller student also passes on the teacher's reasoning habits. Prior work and our annotation of teacher traces show that current reasoning models often start without a plan, verify superficially, abandon promising solution paths, and take unchecked shortcuts. We propose Metacognitive Behavioral Tuning (MBT), which inserts seven metacognitive behaviors into teacher traces at the points where the reasoning lacks them and then trains the student on the edited traces. Because teacher traces are long, how the edits are made matters. Rewriting a whole trace at once summarizes long traces, and choosing all insertion points at once crowds them at the start and end. Our agentic method, MBT-A, edits the trace file line by line and records each edit as a git commit, which avoids summarizing long traces and spreads the insertions. On competition math, a Qwen3.5-4B-Base student trained on MBT-A traces scores 85.4 against 83.1 for SFT on the unedited traces and also outperforms a token-matched control and five prior methods. MBT-A improves two other students as well, and the inserted behaviors clearly transfer to the students' reasoning, with rare ones several times more frequent.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
