arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.16837cs.CL
收藏

大型语言模型是否潜在地执行多跳推理?

Do Large Language Models Latently Perform Multi-Hop Reasoning?

Google DeepMind(谷歌DeepMind);UCL(伦敦大学学院);Google Research(谷歌研究);Tel Aviv University(特拉维夫大学)

Sohee Yang, Elena Gribovskaya, Nora Kassner, Mor Geva, Sebastian Riedel

AI总结:

研究探讨LLMs是否在复杂提示下潜在执行多跳推理,发现特定关系类型提示中存在该推理路径,且第一跳表现显著,但整体利用高度依赖上下文。

摘要:

我们研究大型语言模型(LLMs)是否在复杂提示下潜在地执行多跳推理,例如“The mother of the singer of 'Superstition' is”。我们寻找潜在推理路径的证据,其中LLM(1)潜在地将“the singer of 'Superstition'”识别为桥梁实体Stevie Wonder,以及(2)利用其关于Stevie Wonder母亲的知识来完成提示。我们分别分析这两个跳,并将它们的共同出现视为潜在多跳推理的指示。对于第一跳,我们测试将提示更改为间接提及桥梁实体而非任何其他实体,是否会增加LLM对桥梁实体的内部回忆。对于第二跳,我们测试增加这种回忆是否会导致LLM更好地利用其关于桥梁实体的知识。我们发现,对于某些关系类型的提示,存在潜在多跳推理的有力证据,该推理路径在超过80%的提示中被使用。然而,这种利用高度依赖上下文,在不同类型的提示之间存在差异。此外,平均而言,第二跳和完整多跳遍历的证据相当温和,仅第一跳的证据较为显著。此外,我们发现随着模型规模增加,第一跳推理存在明显的扩展趋势,但第二跳则没有。我们的实验发现为LLMs的未来发展和应用提出了潜在的挑战和机遇。

英文摘要:

We study whether Large Language Models (LLMs) latently perform multi-hop reasoning with complex prompts such as "The mother of the singer of 'Superstition' is". We look for evidence of a latent reasoning pathway where an LLM (1) latently identifies "the singer of 'Superstition'" as Stevie Wonder, the bridge entity, and (2) uses its knowledge of Stevie Wonder's mother to complete the prompt. We analyze these two hops individually and consider their co-occurrence as indicative of latent multi-hop reasoning. For the first hop, we test if changing the prompt to indirectly mention the bridge entity instead of any other entity increases the LLM's internal recall of the bridge entity. For the second hop, we test if increasing this recall causes the LLM to better utilize what it knows about the bridge entity. We find strong evidence of latent multi-hop reasoning for the prompts of certain relation types, with the reasoning pathway used in more than 80% of the prompts. However, the utilization is highly contextual, varying across different types of prompts. Also, on average, the evidence for the second hop and the full multi-hop traversal is rather moderate and only substantial for the first hop. Moreover, we find a clear scaling trend with increasing model size for the first hop of reasoning but not for the second hop. Our experimental findings suggest potential challenges and opportunities for future development and applications of LLMs.

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。

↑