arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2311.06754cs.CL
收藏

从复杂到简单:揭示小型语言模型推理的认知树

From Complex to Simple: Unraveling the Cognitive Tree for Reasoning with Small Language Models

East China Normal University(华东师范大学);Alibaba Group(阿里巴巴集团);Shanghai Institute for AI Education(上海人工智能教育研究院)

Junbing Yan, Chengyu Wang, Taolin Zhang, Xiaofeng He, Jun Huang, Wei Zhang

AI总结:

本研究基于双过程理论提出认知树框架,通过直觉与反思系统迭代分解复杂问题,使小于GPT-3.5参数5%的小模型在推理任务上达到与其相当的性能。

摘要:

推理是人类独特的能力,使我们能够通过将复杂问题分解为一系列可管理的认知步骤来解决它们。然而,复杂的逻辑推理对语言模型来说仍然很困难。基于认知科学中的双过程理论,我们首次揭示了语言模型的认知推理能力。我们的框架采用迭代方法构建认知树(CogTree)。该树的根节点代表初始查询,而叶节点由可以直接回答的简单问题组成。该构建涉及两个主要组件:隐式提取模块(称为直觉系统)和显式推理模块(称为反思系统)。直觉系统通过利用上下文示例快速生成多个响应,而反思系统使用比较学习对这些响应进行评分。评分指导直觉系统进行后续生成步骤。我们在两个流行且具有挑战性的推理任务上的实验结果表明,使用一个显著更小的语言模型(参数不超过7B,少于GPT-3.5参数的5%)可以达到与GPT-3.5(175B参数)相当的性能水平。

英文摘要:

Reasoning is a distinctive human capacity, enabling us to address complex problems by breaking them down into a series of manageable cognitive steps. Yet, complex logical reasoning is still cumbersome for language models. Based on the dual process theory in cognitive science, we are the first to unravel the cognitive reasoning abilities of language models. Our framework employs an iterative methodology to construct a Cognitive Tree (CogTree). The root node of this tree represents the initial query, while the leaf nodes consist of straightforward questions that can be answered directly. This construction involves two main components: the implicit extraction module (referred to as the intuitive system) and the explicit reasoning module (referred to as the reflective system). The intuitive system rapidly generates multiple responses by utilizing in-context examples, while the reflective system scores these responses using comparative learning. The scores guide the intuitive system in its subsequent generation step. Our experimental results on two popular and challenging reasoning tasks indicate that it is possible to achieve a performance level comparable to that of GPT-3.5 (with 175B parameters), using a significantly smaller language model that contains fewer parameters (<=7B) than 5% of GPT-3.5.

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。

↑