arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.22441cs.AIcs.CLcs.LG
收藏

在弱监督和强监督下,潜在推理方法的表现如何?

How Do Latent Reasoning Methods Perform Under Weak and Strong Supervision?

Amazon(亚马逊公司);Michigan State University(密歇根州立大学)

Yingqian Cui, Zhenwei Dai, Bing He, Zhan Shi, Hui Liu, Rui Sun, Zhiji Liu, Yue Xing, Jiliang Tang, Benoit Dumoulin

AI总结:

本文研究了在弱监督和强监督下潜在推理方法的表现,发现存在捷径行为和潜在空间探索问题,并揭示了监督强度与推理能力之间的权衡。

摘要:

潜在推理最近被提出作为一种推理范式,通过在潜在空间中生成步骤来进行多步推理,而不是在文本空间中。这种范式通过在连续潜在空间中执行多步计算,使推理超越离散语言标记。尽管已有大量研究致力于提高潜在推理的性能,但其内部机制仍未被完全探讨。在本工作中,我们对潜在推理方法进行了全面分析,以更好地理解潜在表示在过程中的作用和行为。我们发现潜在推理方法在不同监督水平下存在两个关键问题。首先,我们观察到普遍存在捷径行为,即它们在不依赖潜在推理的情况下实现高准确率。其次,我们检验了潜在推理在潜在空间中支持类似广度优先搜索(BFS)探索的假设,并发现虽然潜在表示可以编码多种可能性,但推理过程并未忠实实现结构化搜索,而是表现出隐含的剪枝和压缩。最后,我们的发现揭示了监督强度相关的权衡:更强的监督会减轻捷径行为,但会限制潜在表示维持多样化假设的能力;而较弱的监督则允许更丰富的潜在表示,但会增加捷径行为。

英文摘要:

Latent reasoning has been recently proposed as a reasoning paradigm and performs multi-step reasoning through generating steps in the latent space instead of the textual space. This paradigm enables reasoning beyond discrete language tokens by performing multi-step computation in continuous latent spaces. Although there have been numerous studies focusing on improving the performance of latent reasoning, its internal mechanisms remain not fully investigated. In this work, we conduct a comprehensive analysis of latent reasoning methods to better understand the role and behavior of latent representation in the process. We identify two key issues across latent reasoning methods with different levels of supervision. First, we observe pervasive shortcut behavior, where they achieve high accuracy without relying on latent reasoning. Second, we examine the hypothesis that latent reasoning supports BFS-like exploration in latent space, and find that while latent representations can encode multiple possibilities, the reasoning process does not faithfully implement structured search, but instead exhibits implicit pruning and compression. Finally, our findings reveal a trade-off associated with supervision strength: stronger supervision mitigates shortcut behavior but restricts the ability of latent representations to maintain diverse hypotheses, whereas weaker supervision allows richer latent representations at the cost of increased shortcut behavior.

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。

↑