Structural Process Supervision for Latent Chain-of-Thought Reasoning

2026-09-09

AI Summary

本文针对潜变量推理中的表示崩溃和信息分布不均问题，提出了一种基于原型介导的过程监督方法PMPS，并通过实验验证了其有效性。

Abstract

Latent reasoning approaches enhance token-level efficiency and robustness by replacing verbose, explicit chain-of-thought (CoT) tokens with compact continuous-space embeddings. However, existing methods lack direct process supervision over these latent embeddings, which often leads to representation collapse and uneven information distribution. To address this, we propose Prototype-Mediated Process Supervision (PMPS), which introduces learnable reasoning prototypes as semantic anchors to provide structural process-level supervision for latent reasoning. PMPS projects latent embeddings and explicit CoT embeddings into a shared prototype space, achieving many-to-many soft alignment between unequal-length representations through prototype assignment. Meanwhile, we introduce a Progressive Sequential Alignment (PSA) module to further guide training: positional priors initially encourage sequential alignment structure, then gradually relax to permit adaptive matching. Experimental results show that PMPS compresses output token length to under 50% of explicit CoT on GSM8K-Aug. Compared to leading baseline SIM-CoT, our method achieves average accuracy gains of 2.08% across different model families. On GPT-2, PMPS even surpasses CoT-SFT. On larger models and a more challenging task, PMPS consistently attains the highest accuracy among all latent reasoning methods with comparable output length.

Problem

• latent reasoning
• process supervision
• representation collapse
• information distribution

Innovation

• Prototype-Mediated Process Supervision
• Latent Reasoning
• Progressive Sequential Alignment
• Shared Prototype Space
• Soft Alignment