Research map / Vision-language and multimodal / Multimodal LLMs
Efficient Multimodal Spatial Reasoning via Dynamic and Asymmetric Routing
Yixian Shen, Qi Bi, Zihan Wang, Zhiheng Yang, Changshuo Wang, Zhi Zhang, Prayag Tiwari, Andy D. Pimentel, Anuj Pathania
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes an efficient framework for multimodal spatial reasoning with visualization-of-thought that reduces the cost of lengthy, redundant visual reasoning tokens via dynamic and asymmetric routing.
Problem
Visualization-of-thought (VoT) improves spatial reasoning in MLLMs, but autoregressive accumulation of lengthy and redundant tokens greatly increases computation and memory costs.
Approach
Uses dynamic and asymmetric routing to make multimodal spatial reasoning more efficient, as named in the title; the abstract does not specify the mechanism in detail.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add DARE-LH hop- and layer-aware token routing to VolCano | VolCano · VSR | accuracy / FLOPs: 67.18 -> 68.09 acc; 19842.37 -> 11310.63 GFLOPs | Improved |
| add DARE-LH routing to VolCano | VolCano · Winoground | accuracy: 68.37 -> 68.31 | No clear effect |
| add DARE-LH routing to VolCano | VolCano · EmbSpatial | accuracy: 58.29 -> 68.09 | Improved |
| add DARE-LH routing to Anole-7B (vs VoT) | Anole-7B · MAZE | accuracy / FLOPs: 86.56 -> 93.32; 25640.14 -> 12739.67 GFLOPs | Improved |
| add DARE-L to VolCano | VolCano · GQA | accuracy: 64.40 -> 64.71 | No clear effect |
| add DARE-L to VolCano | VolCano · POPE | accuracy: 86.50 -> 87.27 | Improved |
| add hard visual pruning loss in deeper layers (DARE-L vs UniPrune) (ablation) | DARE · VSR | accuracy: 63.71 -> 67.13 | Improved |
| add hop-aware retention on top of layer-wise (DARE-LH vs DARE-L) (ablation) | DARE · VSR | accuracy: 67.13 -> 68.09 | Improved |
| add hop-aware retention (DARE-LH vs DARE-L) (ablation) | DARE · MAZE | accuracy: 89.78 -> 93.32 | Improved |
| use DARE-LH KV-cache retention instead of full cache | VolCano · VSR | KV cache: 5.45 GB -> 2.97GB | Improved |
| reserve prefix of 2 tokens per hop instead of 0 (ablation) | DARE-LH · VSR | accuracy: 64.12 -> 68.09 | Improved |
| increase prefix size beyond 2 (ablation) | DARE-LH · VSR | accuracy / KV cache: 68.09 -> 68.09 acc; KV 2.97 -> 3.40 (kappa=4) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SparseVLM — SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference (ICML 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Experimental setup
Builds on: visualization-of-thought
Closest papers by meaning
- Imagine While Reasoning in Space: Multimodal Visualization-of-Thought (ICML 2025 · similarity 0.91)
- Reinforcing Spatial Reasoning in Vision-Language Models with Interwoven Thinking and Visual Drawing (NeurIPS 2025 · similarity 0.90)
- AStar: Boosting Multimodal Reasoning with Automated Structured Thinking (AAAI 2026 · similarity 0.90)
- Spectral-Progressive Thought Flow for Lightweight Multimodal Reasoning (ICML 2026 · similarity 0.89)
- Coarse Correspondences Boost Spatial-Temporal Reasoning in Multimodal Language Model (CVPR 2025 · similarity 0.89)
- See&Trek: Training-Free Spatial Prompting for Multimodal Large Language Model (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
