arXiv cs.AI· Jiaoyang Ruan, Xin Gao, Yinda Chen, Hengyu Zeng, Liang Du, Guanghao Li, Jie Fu, Jian Pu·· 4 小时前AI 评分43
扩散语言模型流形上的推理：双向一致性用于自验证（ICML 2026）
Reasoning on the Manifold: Bidirectional Consistency for Self-Verification in Diffusion Language Models
AI 导读
一篇 ICML 2026 论文提出双向流形一致性（BMC），一种免训练、无监督的指标，通过前向遮蔽与反向重构循环量化扩散大语言模型（dLLM）生成序列的稳定性。该指标可用于无需标准答案的有效性判别、推理阶段拒识重采样，以及作为密集几何奖励替代稀疏结果监督进行对齐，表明内在几何稳定性可作为 dLLM 正确性的可靠指标。
正文 · 原文
Abstract:While Diffusion Large Language Models (dLLMs) offer structural advantages for global planning, efficiently verifying that they arrive at correct answers via valid reasoning traces remains a critical challenge. In this work, we propose a geometric perspective: Reasoning on the Manifold. We hypothesize that valid generation trajectories reside as stable attractors on the high-density manifold of the learned distribution, whereas invalid paths exhibit off-manifold drift. To operationalize this, we introduce Bidirectional Manifold Consistency (BMC), a training-free, unsupervised metric that quantifies the stability of the generated sequence through a forward-masking and backward-reconstruction cycle. Empirically, we demonstrate BMC's versatility across the full reasoning lifecycle: (1) in Diagnosis, it serves as a robust discriminator of solution validity without ground truth answer; (2) in Inference, it enables rejection resampling to effectively concentrate computational resources on complex reasoning tasks; and (3) in Alignment, it functions as a dense geometric reward that transforms sparse outcome supervision into fine-grained guidance, empowering models to self-evolve beyond standard baselines. Our results establish intrinsic geometric stability as a robust indicator of correctness for dLLMs.
| Comments: | 31 pages, 7 figures. Accepted to the 43rd International Conference on Machine Learning (ICML 2026). Camera-ready version |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2604.16565 [cs.LG] |
| (or arXiv:2604.16565v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2604.16565 arXiv-issued DOI via DataCite |
|
| Journal reference: | Proceedings of the 43rd International Conference on Machine Learning, PMLR 306, 2026 |
Submission history
From: Jiaoyang Ruan [view email]
[v1]
Fri, 17 Apr 2026 10:17:16 UTC (788 KB)
[v2]
Thu, 23 Apr 2026 12:41:25 UTC (788 KB)
[v3]
Wed, 27 May 2026 14:55:17 UTC (789 KB)
[v4]
Thu, 5 Nov 2026 12:30:15 UTC (786 KB)
来源：arXiv cs.AI · arxiv.org