arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2405.12939cs.CL
收藏

推理聚合:一种增强大型语言模型答案选择的分层框架

Aggregation of Reasoning: A Hierarchical Framework for Enhancing Answer Selection in Large Language Models

School of Computer Science, Fudan University(复旦大学计算机科学技术学院);National University of Singapore(新加坡国立大学);Shanghai AI Laboratory(上海人工智能实验室);Midea AI Research Center(美的AI研究中心)

Zhangyue Yin, Qiushi Sun, Qipeng Guo, Zhiyuan Zeng, Xiaonan Li, Tianxiang Sun, Cheng Chang, Qinyuan Cheng, Ding Wang, Xiaofeng Mou, Xipeng Qiu, XuanJing Huang

AI总结:

针对多数投票在正确答案为少数时失效的问题,本文提出分层推理聚合框架AoR,通过评估推理链选择答案并动态调整采样数量,在复杂推理任务上优于主流集成方法。

摘要:

近期,Chain-of-Thought(思维链)提示技术的进展推动了大型语言模型(LLM)在复杂推理任务上的重大突破。当前研究通过采样多条推理链并依据答案频率进行集成,以提升LLM的推理性能。然而,当正确答案处于少数派情形时,这种方法会失效。我们认为这是制约LLM推理能力的主要因素,且仅依据预测答案无法消除该局限。为弥补这一不足,我们提出分层推理聚合框架AoR(Aggregation of Reasoning),其基于对推理链的评估来选择答案。此外,AoR引入动态采样机制,可根据任务复杂度调整推理链数量。在一系列复杂推理任务上的实验结果表明,AoR优于主流集成方法。进一步分析显示,AoR不仅能够适配多种LLM,而且与现有方法相比可达到更高的性能上限。

英文摘要:

Recent advancements in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks. Current research enhances the reasoning performance of LLMs by sampling multiple reasoning chains and ensembling based on the answer frequency. However, this approach fails in scenarios where the correct answers are in the minority. We identify this as a primary factor constraining the reasoning capabilities of LLMs, a limitation that cannot be resolved solely based on the predicted answers. To address this shortcoming, we introduce a hierarchical reasoning aggregation framework AoR (Aggregation of Reasoning), which selects answers based on the evaluation of reasoning chains. Additionally, AoR incorporates dynamic sampling, adjusting the number of reasoning chains in accordance with the complexity of the task. Experimental results on a series of complex reasoning tasks show that AoR outperforms prominent ensemble methods. Further analysis reveals that AoR not only adapts various LLMs but also achieves a superior performance ceiling when compared to current methods.

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。

↑