思維鏈中無全域規劃：揭示大型語言模型的潛在規劃視野

摘要

本研究源於先前對思維鏈動力學的互補性觀察：研究顯示大型語言模型在思維鏈顯現前已存在潛在的後續推理規劃，從而削弱了顯性思維鏈的重要性；然而對於需要多步推理的任務，思維鏈仍具有關鍵作用。為深化對大型語言模型內部狀態與其言語化推理軌跡間關係的理解，我們透過探測方法Tele-Lens對跨領域任務的隱藏狀態進行分析，探究大型語言模型的潛在規劃能力。實證結果表明，大型語言模型呈現近視距特徵，主要進行增量式狀態轉換而缺乏精準的全局規劃。利用此特性，我們提出增強思維鏈不確定性估計的假設，並驗證僅需透過思維鏈中少量關鍵節點即可有效表徵整體路徑的不確定性。我們進一步強調開發思維鏈動力學的重要性，並展示無需犧牲性能即可實現思維鏈旁路自動識別。相關代碼、數據與模型已開源於：https://github.com/lxucs/tele-lens。

English

This work stems from prior complementary observations on the dynamics of Chain-of-Thought (CoT): Large Language Models (LLMs) is shown latent planning of subsequent reasoning prior to CoT emergence, thereby diminishing the significance of explicit CoT; whereas CoT remains critical for tasks requiring multi-step reasoning. To deepen the understanding between LLM's internal states and its verbalized reasoning trajectories, we investigate the latent planning strength of LLMs, through our probing method, Tele-Lens, applying to hidden states across diverse task domains. Our empirical results indicate that LLMs exhibit a myopic horizon, primarily conducting incremental transitions without precise global planning. Leveraging this characteristic, we propose a hypothesis on enhancing uncertainty estimation of CoT, which we validate that a small subset of CoT positions can effectively represent the uncertainty of the entire path. We further underscore the significance of exploiting CoT dynamics, and demonstrate that automatic recognition of CoT bypass can be achieved without performance degradation. Our code, data and models are released at https://github.com/lxucs/tele-lens.

思維鏈中無全域規劃：揭示大型語言模型的潛在規劃視野

No Global Plan in Chain-of-Thought: Uncover the Latent Planning Horizon of LLMs

摘要

Support