REVIEW 35 cited by
V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs
read the original abstract
When we look around and perform complex tasks, how we see and selectively process what we see is crucial. However, the lack of this visual search mechanism in current multimodal LLMs (MLLMs) hinders their ability to focus on important visual details, especially when handling high-resolution and visually crowded images. To address this, we introduce V*, an LLM-guided visual search mechanism that employs the world knowledge in LLMs for efficient visual querying. When combined with an MLLM, this mechanism enhances collaborative reasoning, contextual understanding, and precise targeting of specific visual elements. This integration results in a new MLLM meta-architecture, named Show, sEArch, and TelL (SEAL). We further create V*Bench, a benchmark specifically designed to evaluate MLLMs in their ability to process high-resolution images and focus on visual details. Our study highlights the necessity of incorporating visual search capabilities into multimodal systems. The code is available https://github.com/penghao-wu/vstar.
Forward citations
Cited by 35 Pith papers
-
GazeVLM: Active Vision via Internal Attention Control for Multimodal Reasoning
GazeVLM introduces internal gaze tokens that allow VLMs to dynamically suppress irrelevant visual features and simulate foveal attention for improved high-resolution multimodal reasoning.
-
Improving Vision-language Models with Perception-centric Process Reward Models
Perceval is a perception-centric PRM that detects token-level perceptual errors in VLMs, supporting token-advantage RL training and iterative test-time scaling for improved reasoning.
-
SketchVLM: Vision language models can annotate images to explain thoughts and guide users
SketchVLM lets VLMs generate non-destructive SVG annotations on input images to visually explain answers, raising visual reasoning accuracy by up to 28.5 points and annotation quality by 1.48x over baselines.
-
AnchorSeg: Language Grounded Query Banks for Reasoning Segmentation
AnchorSeg uses ordered query banks of latent reasoning tokens plus a spatial anchor token and a Token-Mask Cycle Consistency loss to achieve 67.7% gIoU and 68.1% cIoU on the ReasonSeg benchmark.
-
SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning
A small tool-free MLLM plus answer-separability gating bypasses agentic tool loops for many queries, yielding 1.1–3.35× speedup with preserved or higher accuracy.
-
DLEBench: Evaluating Small-scale Object Editing Ability for Instruction-based Image Editing Model
DLEBench is the first benchmark for small-scale object editing in instruction-based image editing models, using 1889 samples, seven instruction types, and a dual-mode evaluation protocol to reveal performance gaps in ...
-
Visual Para-Thinker: Divide-and-Conquer Reasoning for Visual Comprehension
Visual Para-Thinker is the first parallel reasoning framework for MLLMs that uses visual partitioning strategies, Pa-Attention, and LPRoPE to extend test-time scaling benefits to visual comprehension tasks.
-
HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling
HiDe is a training-free hierarchical decoupling method that separates key visual tokens from background interference in high-resolution MLLMs to achieve new state-of-the-art results on V*Bench, HRBench4K, and HRBench8K.
-
VAD: Attributing Visual Evidence for Target Reconstruction in Multimodal On-Policy Distillation
Counterfactual present/removed teacher views attribute visually supported corrections and reconstruct student-anchored distillation targets that beat source-mixed multimodal OPD.
-
ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding
ST-Veto improves reasoning in diffusion MLLMs by vetoing temporally unstable tokens and tokens with weak image grounding, swapping in safer near-boundary candidates.
-
What Transfers from Text to Vision? Capability Scaling Laws and Transfer Dynamics for VLMs
VLM accuracy can be predicted from a scalar capability score derived from LLM text benchmarks plus multimodal data volume via a fitted transfer-absorption scaling law.
-
Kamera: Unified Position-Invariant Multimodal KV Cache for Training-Free Reuse
Kamera stores a low-rank patch with each position-free KV chunk to restore cross-chunk conditioning lost in naive reuse, enabling cheap reordering, sliding windows, and recall across attention mechanisms.
-
Look Before You Zoom: Adaptive Routing for the Resolution-Context Trade-off in Visual RAG
ViRGo adaptively routes visual retrieval decisions in VLMs by estimating object scale from intrinsic localization heads combined with token confidence, matching patch retrieval on small objects, attention retrieval on...
-
Visual-OPSD: Cross-Modal On-Policy Self-Distillation for Efficient Unified Multimodal Reasoning
Visual-OPSD distills reasoning from a privileged visual-thought teacher to a text-only student using on-policy JSD, delivering +3.40pp accuracy gain and 14.3x speedup over the generative teacher on nine benchmarks.
-
MOSS-Video-Preview: Toward Real-Time Video Understanding via Cross-Attention
MOSS-Video-Preview introduces a cross-attention architecture and synthesized real-time QA data to enable continuous perception, answer revision, and faster inference in video-language models compared to decoder-only designs.
-
Self-Prophetic Decoding to Unlock Visual Search in LVLMs
SeProD is a plug-and-play self-prophetic decoding framework that combines pre- and post-training LVLM capabilities via probability-based sampling to improve coherent visual search and multi-step reasoning.
-
Starve to Perceive: Taming Lazy Perception in VLMs with Constrained Visual Bandwidth
Constraining visual token budgets during SFT and RL forces VLMs to learn functional active perception, yielding ~5% relative gains and strong transfer to unconstrained evaluation.
-
Cognitive Pivot Points and Visual Anchoring: Unveiling and Rectifying Hallucinations in Multimodal Reasoning Models
Hallucinations in multimodal reasoners cluster at high-entropy pivot tokens due to intermediate-layer visual anchoring failure, and V-STAR mitigates this with attention rewards plus forced reflection.
-
Entropy-Gradient Grounding: Training-Free Evidence Retrieval in Vision-Language Models
Entropy-gradient grounding uses model uncertainty to retrieve evidence regions in VLMs, improving performance on detail-critical and compositional tasks across multiple architectures.
-
Multimodal Latent Reasoning via Predictive Embeddings
Pearl learns predictive embeddings from multimodal tool trajectories in latent space to enable efficient reasoning that matches or exceeds supervised fine-tuning and reconstruction-based methods without explicit tool ...
-
Q-Zoom: Query-Aware Adaptive Perception for Efficient Multimodal Large Language Models
Q-Zoom achieves up to 4.39x inference speedup in high-resolution MLLM scenarios via query-aware gating and region localization, matching or exceeding baseline accuracy on document and high-res benchmarks.
-
Blink: Dynamic Visual Token Resolution for Enhanced Multimodal Understanding
Blink dynamically expands high-saliency visual tokens and drops them when attention shifts, improving LLaVA-1.5 and LLaVA-NeXT across seven multimodal benchmarks.
-
Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models
A dynamic replay and reweighting scheduler (RECAP) preserves general capabilities during RLVR while keeping reasoning performance at least as good as reasoning-only finetuning.
-
HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling
HiDe uses token-wise attention decoupling and layout-preserving decoupling to build compact crops that push Qwen2.5-VL and InternVL3 to state-of-the-art scores on high-resolution VQA benchmarks.
-
Grounded Reinforcement Learning for Visual Reasoning
ViGoRL introduces visually grounded RL that anchors reasoning steps to image coordinates and uses multi-turn zooming to outperform standard RL and supervised baselines on spatial and GUI reasoning benchmarks.
-
ToolGate: Token-Efficient Pre-Call Control for Tool-Augmented Vision-Language Agents
ToolGate is a pre-call controller that reduces token usage in tool-augmented VLM agents to 64-69% of baseline while preserving or slightly improving accuracy on benchmarks.
-
Starve to Perceive: Taming Lazy Perception in VLMs with Constrained Visual Bandwidth
Constraining visual token budget per observation during VLM training forces genuine active perception and delivers 5% average relative improvement without auxiliary losses or architecture changes.
-
What's Holding Back Latent Visual Reasoning?
Latent visual reasoning fails in current models because standard datasets make oracle latents uninformative and inference-time latents collapse away from useful representations.
-
Cognitive Pivot Points and Visual Anchoring: Unveiling and Rectifying Hallucinations in Multimodal Reasoning Models
Multimodal reasoning models hallucinate at high-entropy cognitive bifurcation points due to loss of visual semantic anchoring, and the V-STAR training paradigm with HVAR rewards and FRM reflection mitigates this by re...
-
MAG-3D: Multi-Agent Grounded Reasoning for 3D Understanding
MAG-3D is a training-free multi-agent framework that coordinates planning, grounding, and coding agents with off-the-shelf VLMs to achieve grounded 3D reasoning and state-of-the-art benchmark results.
-
LanteRn: Latent Visual Structured Reasoning
A 3B vision-language model trained to emit latent visual thought tokens interleaved with text, then refined by reinforcement learning, outperforms a matched text-only baseline on several visual reasoning benchmarks.
-
Mixture-of-Visual-Thoughts: Exploring Context-Adaptive Reasoning Mode Selection for General Visual Reasoning
MoVT unifies different visual reasoning modes in a single model and uses the AdaVaR two-stage framework with supervised cold-start and RL via AdaGRPO to enable context-adaptive mode selection, yielding consistent gain...
-
Seed1.5-VL Technical Report
Seed1.5-VL is a compact multimodal model that sets new records on dozens of vision-language benchmarks and outperforms prior systems on agent-style tasks.
-
PaliGemma: A versatile 3B VLM for transfer
PaliGemma is an open 3B VLM based on SigLIP and Gemma that achieves strong performance on nearly 40 diverse open-world tasks including benchmarks, remote-sensing, and segmentation.
-
A Survey on Multimodal Large Language Models
This survey organizes the architectures, training strategies, data, evaluation methods, extensions, and challenges of Multimodal Large Language Models.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.