arXiv：cs.LG· Zeyang Zhang, Linxiang Gao, Meiqi Gu, Xingyan Chen, Minrui Luo, Kaifeng Lyu, Tianxing He·· 3 小时前
自条件（Self-Conditioning）如何让 Flow Matching 实现深度序列推理
Self-Conditioning Enables Deep Sequential Reasoning in Flow Matching
AI 导读
一项理论研究证明，vanilla flow-matching 采样器的轨迹可由常数轮并行模型评估模拟，因此增加采样步数无法提升常深度骨干模型的表达能力。而自条件（self-conditioning）能打破这一瓶颈：在多种视觉推理任务上，带自条件的 flow matching 准确率更高、推理深度更深，且性能随采样步数增加而提升。理论上，带自条件时，常深度骨干可达到空间有界图灵机同等的表达能力。
正文
Abstract:Recent work has explored visual reasoning through image generation, where models solve visual-reasoning problems by directly generating solution images. Intuitively, iterative samplers such as flow matching are well suited to this task, since each sampling step may resemble one reasoning step. However, in this work, from a circuit complexity perspective, we show that this intuition is misleading. We prove that as long as the model outputs satisfy certain mild regularity conditions, the trajectory of the vanilla flow-matching sampler can be simulated by a constant number of sequential rounds of parallel model evaluations. Consequently, additional sampling steps cannot increase the expressive power of a constant-depth backbone model under the standard complexity-theoretic model of transformers. We then reveal that self-conditioning, which conditions the model on the previous step's estimate of the final image, plays a previously unrecognized role in removing this bottleneck. Controlled experiments on diverse visual-reasoning tasks show that flow matching with self-conditioning achieves higher accuracy and greater reasoning depth than vanilla flow matching, and its performance improves with more sampling steps. Theoretically, we prove that with self-conditioning, a constant-depth backbone can achieve expressive power equivalent to that of space-bounded Turing machines. Our study suggests a design principle for future visual reasoning research: carry more information between sampling steps, so that additional steps translate into deeper reasoning.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.19628 [cs.CV] |
| (or arXiv:2610.19628v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.19628 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zeyang Zhang [view email]
[v1]
Wed, 14 Oct 2026 05:56:19 UTC (8,427 KB)
来源：arXiv：cs.LG · arxiv.org