Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

V-ABS: Action-Observer Driven Beam Search for Dynamic Visual Reasoning

About

Multimodal large language models (MLLMs) have achieved remarkable success in general perception, yet complex multi-step visual reasoning remains a persistent challenge. Although recent agentic approaches incorporate tool use, they often neglect critical execution feedback. Consequently, they suffer from the imagination-action-observer (IAO) bias, a misalignment between prior imagination and observer feedback that undermines reasoning stability and optimality. To bridge this gap, we introduce V-ABS, an action-observer driven beam search framework that enables deliberate reasoning through thinker-actor-observer iterations. We also propose an entropy-based adaptive weighting algorithm to mitigate the IAO bias by dynamically balancing the confidence scores between the policy priors and the observational feedback. Moreover, we construct a large-scale supervised fine-tuning (SFT) dataset comprising over 80k samples to guide the model to assign higher prior confidence to correct action paths. Extensive experiments across eight diverse benchmarks show that V-ABS achieves state-of-the-art performance, delivering an average improvement of 19.7% on the Qwen3-VL-8B baseline and consistent gains across both open-source and proprietary models.

Zhiwei Ning, Xuanang Gao, Jiaxi Cao, Gengming Zhang, Shengnan Ma, Wenwen Tong, Hanming Deng, Jie Yang, Wei Liu• 2026

Related benchmarks

TaskDatasetResultRank
High-resolution Visual UnderstandingHR-Bench-4K
FSP91
49
Fine-grained visual searchHR-Bench-8K
Overall Score78
24
SudokuSudoku
Accuracy37.8
19
Fine-grained visual searchV*
Overall Score91.1
14
Visual NavigationVisuoThink Level 3
Accuracy94.9
9
Visual NavigationVisuoThink (Level 4)
Accuracy74.1
9
Visual NavigationVisuoThink Level 5
Accuracy69.7
9
Visual NavigationFrozen-Lake 4x4 Grid
Accuracy55
9
Visual NavigationFrozen-Lake 6x6 Grid
Accuracy25
9
Visual NavigationFrozen-Lake 8x8 Grid
Accuracy12.5
9
Showing 10 of 14 rows

Other info

Follow for update