看得更少，看得更准：面向多模态推理的双向感知塑形
See Less, See Right: Bi-directional Perceptual Shaping For Multimodal Reasoning
BiPS 用「保留问题相关区域」与「抹除关键像素」两种掩码视图构造双向 KL 约束，可直接作为 VLM 微调时的训练正则项，迫使模型依赖细粒度视觉证据而非文本捷径，且不增加推理开销。
原文依据 transforms question-conditioned masked views into bidirectional where-to-look signals that shape perception during training
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论大型视觉语言模型（VLM）往往受益于中间视觉线索——这些线索或由外部工具注入，或在推理过程中以潜变量视觉 token 的形式生成——但此类机制仍然忽略了细粒度视觉证据（例如图表中的折线），跨域泛化能力差，且带来高昂的推理时开销。本文提出双向感知塑形（BiPS），将问题条件化的掩码视图转化为双向的「看哪里」信号，从而在训练阶段塑造感知。BiPS 首先在原始图像与仅保留问题相关区域的证据保留视图之间施加 KL 一致性约束，鼓励对支撑像素形成粗略但完整的覆盖。随后，它在原始图像与证据消融视图之间施加 KL 分离约束，后者遮蔽关键像素，使图像不再支持原答案，从而抑制纯文本捷径（即仅凭文本作答）并强制模型依赖细粒度视觉信息。在八个基准上，BiPS 使 Qwen2.5-VL-7B 平均提升 8.2%，并对未见数据集与图像类型展现出较强的域外泛化能力。
英文原文摘要 1187 字符
Large vision-language models (VLMs) often benefit from intermediate visual cues, either injected via external tools or generated as latent visual tokens during reasoning, but these mechanisms still overlook fine-grained visual evidence (e.g., polylines in charts), generalize poorly across domains, and incur high inference-time cost. In this paper, we propose Bi-directional Perceptual Shaping (BiPS), which transforms question-conditioned masked views into bidirectional where-to-look signals that shape perception during training. BiPS first applies a KL-consistency constraint between the original image and an evidence-preserving view that keeps only question-relevant regions, encouraging coarse but complete coverage of supporting pixels. It then applies a KL-separation constraint between the original and an evidence-ablated view where critical pixels are masked so the image no longer supports the original answer, discouraging text-only shortcuts (i.e., answering from text alone) and enforcing fine-grained visual reliance. Across eight benchmarks, BiPS boosts Qwen2.5-VL-7B by 8.2% on average and shows strong out-of-domain generalization to unseen datasets and image types.