取其精华,去其糟粕:面向自由形式语言生成的细粒度自洽性
Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language Generation
AI总结:
本文针对自洽性难以聚合自由形式生成答案的问题,提出 FSC,通过提取并融合候选样本的片段级共性,并结合候选过滤与合并策略,显著提升多类任务表现。
摘要:
自洽性(Self-consistency,SC)利用来自 LLM 的多个样本,在各类推理任务上展现出显著提升,但由于答案聚合困难,其在自由形式生成中表现不佳。其变体 UCS 和 USC 依赖样本选择或投票机制来提升输出质量。然而,这些方法因无法充分利用多个候选样本中存在的细微共识知识而存在局限,往往导致次优输出。我们提出细粒度自洽性(Fine-Grained Self-Consistency,FSC),通过提取并整合候选样本中的片段级共性来解决这些局限,从而提升 LLM 在开放式任务和推理任务中的性能。在此基础上,我们提出两种额外策略:候选过滤,通过识别高度相似的候选集合来提升整体质量;以及合并,通过组合相似样本来降低输入 token 需求。通过使用 GPT-3.5-turbo 和 GPT-4 在摘要、代码生成和数学推理等多种任务上开展大量实验,FSC 的有效性得到验证。结果表明,其相较基线方法有显著提升,展示了 FSC 通过有效综合来自多个样本的细粒度共识知识来优化输出质量的潜力。
英文摘要:
Self-consistency (SC), leveraging multiple samples from LLMs, shows significant gains on various reasoning tasks but struggles with free-form generation due to the difficulty of aggregating answers. Its variants, UCS and USC, rely on sample selection or voting mechanisms to improve output quality. These methods, however, face limitations due to their inability to fully utilize the nuanced consensus knowledge present within multiple candidate samples, often resulting in suboptimal outputs. We propose Fine-Grained Self-Consistency (FSC) to addresses these limitations by extracting and integrating segment-level commonalities from candidate samples, enhancing the performance of LLMs both in open-ended and reasoning tasks. Based on this, we present two additional strategies: candidate filtering, which enhances overall quality by identifying highly similar candidate sets, and merging, which reduces input token requirements by combining similar samples. The effectiveness of FSC is demonstrated through extensive experiments on various tasks, including summarization, code generation, and mathematical reasoning, using GPT-3.5-turbo and GPT-4. The results indicate significant improvements over baseline methods, showcasing the potential of FSC to optimize output quality by effectively synthesizing fine-grained consensus knowledge from multiple samples.
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。