Recent advances in vision-language models (VLMs) emphasize long chain-of-thought reasoning; yet, we find that their performance on visual tasks is primarily limited by a lack of visual perception as opposed to reasoning itself. In this work, we systematically study the interplay between perception and reasoning in VLM post-training by decomposing their capabilities into three separate training stages: visual perception, visual reasoning, and textual reasoning, incorporating specialized training data. We demonstrate that visual perception (a) requires targeted optimization with specialized data; (b) serves as a fundamental scaffold that should be solidified through staged training before refining visual reasoning; and (c) is more effectively learned via RL than caption-based SFT. Our experiments across multiple VLMs demonstrate that staged training consistently improves both visual perception and reasoning performance over merged training. Notably, models trained with our approach achieve 1.5% higher reasoning accuracy with 20.8% shorter reasoning traces, suggesting that superior perception reduces the need for excessive reasoning. Furthermore, we show that this capability-based staging represents a new curriculum dimension orthogonal to traditional difficulty-based curricula, and combining both yields further additive gains. Our staged-training models achieve superior performance among open-weight VLMs, establishing advanced results on several visual math and perception (e.g., +5.2% on WeMath and +3.7% on RealWorldQA) tasks compared with the base counterpart.

翻译：近年来，视觉语言模型（VLM）的进步强调长链思维推理；然而，我们发现其在视觉任务上的表现主要受限于视觉感知能力的不足，而非推理本身。本研究通过将VLM的能力分解为视觉感知、视觉推理和文本推理三个独立训练阶段并引入专门训练数据，系统探讨了VLM后训练中感知与推理的相互作用。我们证明：(a) 视觉感知需要针对专门数据的目标优化；(b) 视觉感知是基础框架，需通过分阶段训练巩固后方可优化视觉推理；(c) 采用强化学习比基于字幕的监督微调（SFT）更能有效习得视觉感知。在多个VLM上的实验表明，相较于混合训练，分阶段训练能持续提升视觉感知与推理性能。值得注意的是，采用本方法训练的模型在推理准确率提升1.5%的同时，推理轨迹缩短20.8%，表明更优的感知可减少冗余推理。此外，我们发现这种基于能力的阶段划分代表了与基于难度的传统课程正交的新课程维度，二者结合可产生额外的叠加增益。我们的分阶段训练模型在开源权重VLM中取得领先表现，相较于基础模型在多项视觉数学及感知任务（如WeMath提升5.2%，RealWorldQA提升3.7%）上实现了先进成果。