arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.17174cs.CLcs.AI
收藏

CSCE:通过同时增强因果显著性与一致性提升大语言模型推理能力

CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency

School of Computer and Communication Engineering(计算机与通信工程学院);University of Science and Technology Beijing(北京科技大学)

Kangsheng Wang, Xiao Zhang, Juntao Lyu, Tianyu Hu, Huimin Ma

AI总结:

针对大语言模型链式推理中存在的因果幻觉问题,本文提出非链式推理框架CSCE,通过定制损失函数增强因果显著性与一致性,将级联推理改为一次性输出,同时提升了推理成功率与速度。

摘要:

诸如思维链(Chain of Thought, CoT)之类的链式推理方法在解决大语言模型(Large Language Models, LLMs)的推理任务中发挥着日益重要的作用。然而,推理步骤与对应状态转换之间的因果幻觉正成为提升LLMs推理能力的重大障碍,在长程推理任务中尤为突出。本文提出了一种同时考量因果显著性与一致性的非链式推理框架,即因果显著性与一致性增强器(Causal Significance and Consistency Enhancer, CSCE)。我们利用处理效应评估定制LLM的损失函数,从因果显著性和一致性两个方面增强其推理能力,确保模型捕捉关键因果关系,并在各类场景中保持稳健一致的性能。此外,我们将推理过程从CoT等链式方法中常用的级联多步单步推理,转变为一次性输出完整推理过程的因果增强方法,进一步提升了模型的推理效率。大量实验表明,我们的方法同时提升了推理成功率和推理速度。这些改进进一步证明,非链式方法同样能够辅助LLMs完成推理任务。

英文摘要:

Chain-based reasoning methods like chain of thought (CoT) play a rising role in solving reasoning tasks for large language models (LLMs). However, the causal hallucinations between a step of reasoning and corresponding state transitions are becoming a significant obstacle to advancing LLMs' reasoning capabilities, especially in long-range reasoning tasks. This paper proposes a non-chain-based reasoning framework for simultaneous consideration of causal significance and consistency, i.e., the Causal Significance and Consistency Enhancer (CSCE). We customize LLM's loss function utilizing treatment effect assessments to enhance its reasoning ability from two aspects: causal significance and consistency. This ensures that the model captures essential causal relationships and maintains robust and consistent performance across various scenarios. Additionally, we transform the reasoning process from the cascading multiple one-step reasoning commonly used in Chain-Based methods, like CoT, to a causal-enhanced method that outputs the entire reasoning process in one go, further improving the model's reasoning efficiency. Extensive experiments show that our method improves both the reasoning success rate and speed. These improvements further demonstrate that non-chain-based methods can also aid LLMs in completing reasoning tasks.

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。

↑