DCGC: Draft-Conditioned Global Correction for Complex Reasoning with Masked Diffusion Models

2026-08-26

AI Summary

为解决大型语言模型推理错误传播问题，提出DCGC方法，利用掩码扩散模型和动态双CFG机制，基于不完美解草稿进行全局修正。

Abstract

Correcting flawed reasoning traces remains a significant challenge for Large Language Models (LLMs), whose autoregressive generation can propagate early mistakes into subsequent reasoning. We introduce DCGC, a Masked Diffusion Model (MDM) framework for global correction that uses an imperfect solution draft from an upstream solver as auxiliary context. DCGC combines task-specific Supervised Fine-Tuning (SFT) with a novel inference-time mechanism called Dynamic Dual-CFG. This mechanism separates problem-only and joint problem-draft branches and scales the draft-conditioned residual using a relative confidence gap. Across math, code, and knowledge reasoning benchmarks, DCGC outperforms standard sampling and simpler CFG variants, with additional results suggesting transfer to different diffusion backbones. In test-time setting where ground-truth failure labels are unavailable, DCGC improves full test set accuracy by correcting low-consensus upstream outputs, highlighting its utility as a verifier-free global correction module for difficult reasoning instances.

Problem

• Large Language Models
• Reasoning Traces
• Autoregressive Generation
• Global Correction
• Masked Diffusion Models

Innovation

• Masked Diffusion Model
• Supervised Fine-Tuning
• Dynamic Dual-CFG
• Global Correction