Tutorials
Visual Saliency Steering Distillation for Multimodal Chain-of-Thought Reasoning
arXiv:2607.22013v1 Announce Type: new Abstract: Multimodal chain-of-thought (CoT) reasoning integrates visual and textual cues through step-by-step inference. In small models with limited token budget
arXiv:2607.22013v1 Announce Type: new Abstract: Multimodal chain-of-thought (CoT) reasoning integrates visual and textual cues through step-by-step inference. In small models with limited token budgets, modality-interaction fusion often suppresses tiny cross-modal differences. In particular, multimodal CoT often struggles when different images pair with identical text or different texts pair with an identical image, making such inputs nearly indistinguishable after fusion. This study proposes Visual Saliency Steering Distillation (VSSD). VSSD leverages the attention maps of multimodal large language models to generate perturbed images that capture task-sensitive feature directions, and then applies singular value decomposition to extract dominant steering vectors to guide inter-layer distillation. Experiments on ScienceQA and M^3CoT demonstrate that VSSD improves rationale generation and answer inference. The code is available at https://github.com/BGWH123/VSSD.
Related
- Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot Do
- PointVG-R: Internalizing Geometric Reasoning in MLLMs for Precise Pointing Localization via Visual Chain of Thought
- Look Less, Think Faster: Joint Token-Compute Adaptation for Multimodal LLMs
Source: arXiv cs.CV | 2026-07-27