arXiv 2312.046842025-06-20cs.CLcs.AI

LaRS：用于思维链推理的潜在推理技能

LaRS: Latent Reasoning Skills for Chain-of-Thought Reasoning

The University of Texas at Austin（德克萨斯大学奥斯汀分校）；Amazon Web Service（亚马逊网络服务）；Sony AI（索尼人工智能）

Zifan Xu, Haozhu Wang, Dmitriy Bespalov, Xian Wu, Peter Stone, Yanjun Qi

发表信息：: Findings of Empirical Methods in Natural Language Processing 2024

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出LaRS方法，利用无监督学习构建推理技能的潜在空间，通过对齐示例与问题的推理技能来选择思维链示例，无需人工或额外LLM，计算高效且性能优于现有方法。

摘要：

思维链（CoT）提示是大语言模型（LLMs）中一种流行的上下文学习（ICL）方法，尤其是在处理复杂推理任务时。传统的ICL方法使用包含与输入问题相似问题的示例来构建提示。然而，CoT提示在其示例中包含了关键的中间推理步骤（理由），因此需要基于这些理由而非问题本身来选择示例。现有方法需要人类专家或预训练的LLM来描述技能（即理由的高级抽象）以指导选择。然而，这些方法通常成本高昂且难以扩展。相反，本文提出了一种名为潜在推理技能（LaRS）的新方法，该方法采用无监督学习来创建理由的潜在空间表示，其中潜在变量称为推理技能。同时，LaRS学习一种推理策略，以确定给定问题所需的推理技能。然后，通过对齐过去示例与问题之间的推理技能来选择ICL示例。该方法具有理论基础且计算效率高，无需辅助LLM推理或手动提示设计。实验结果表明，LaRS在技能选择方法上持续优于最先进的（SOTA）方法，处理示例库的速度快四倍，在选择阶段将LLM推理次数减少一半，并且对次优示例库表现出更强的鲁棒性。

英文摘要：

Chain-of-thought (CoT) prompting is a popular in-context learning (ICL) approach for large language models (LLMs), especially when tackling complex reasoning tasks. Traditional ICL approaches construct prompts using examples that contain questions similar to the input question. However, CoT prompting, which includes crucial intermediate reasoning steps (rationales) within its examples, necessitates selecting examples based on these rationales rather than the questions themselves. Existing methods require human experts or pre-trained LLMs to describe the skill, a high-level abstraction of rationales, to guide the selection. These methods, however, are often costly and difficult to scale. Instead, this paper introduces a new approach named Latent Reasoning Skills (LaRS) that employs unsupervised learning to create a latent space representation of rationales, with a latent variable called a reasoning skill. Concurrently, LaRS learns a reasoning policy to determine the required reasoning skill for a given question. Then the ICL examples are selected by aligning the reasoning skills between past examples and the question. This approach is theoretically grounded and compute-efficient, eliminating the need for auxiliary LLM inference or manual prompt design. Empirical results demonstrate that LaRS consistently outperforms SOTA skill-based selection methods, processing example banks four times faster, reducing LLM inferences during the selection stage by half, and showing greater robustness to sub-optimal example banks.

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。