SenseSearch: Empowering Vision-Language Models with High-Resolution Agentic Search-Reasoning via Reinforcement Learning
Yong Xien Chng, Tao Hu, Wenwen Tong, Xueheng Li, Jiandong Chen, Haojia Yu, Jiefan Lu, Hewei Guo, Hanming Deng, Chengjun Xie, Gao Huang, Lewei Lu
Abstract
How many years separate the founding year of the company indicated by the black logo on the lower left of the racer's suit and the racer's birth year? <think> First, I need to identify the company name indicated that I must zoom in to see the logo clearly. I will utilize the image crop tool. </think> Tool: Image Crop, bbox: [0.35, 0.45, 0.38, 0.48] <think> Next, I need to use image search to identify this racer. </think> Tool: Image Search, image: <image> <think> then I need to search Max Verstappen birth year </think> Tool: Text Search, query" "Max Verstappen birth" <think> I found the company was founded in 2015 and the driver was born in 1997. 2015 minus 1997 is 18. </think> <answer> 18 </answer> Title 1: Max Verstappen 💪 What a performace Thumbnail 1: … Max Verstappen was born on September 30, 1997, in Hasselt, Belgium … <think> now I can see clearly that the company is castore. I will search to find out the founding year of castore. </think> … Castore was founded by Tomas and Philip Beahon and incorporated on July 5, 2015…
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Builds on22
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni et al.NeurIPS 2020 · 19,162 citations
- Toolformer: Language Models Can Teach Themselves to Use ToolsTimo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu et al.NeurIPS 2023 · 5,989 citations
- DAPO: An Open-Source LLM Reinforcement Learning System at ScaleQiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan et al.NeurIPS 2025 · 2,828 citations
- Large Language Models Struggle to Learn Long-Tail KnowledgeNikhil Kandpal, Haikang Deng, Adam Roberts, Eric Wallace et al.ICML 2023 · 623 citations
- ReTool: Reinforcement Learning for Strategic Tool Use in LLMsJiazhan Feng, Shijue Huang, Xingwei Qu, Ge Zhang et al.ICLR 2026 · 406 citations
Related papers
- Visual-Aware Testing and Debugging for Web Performance OptimizationXinlei Yang, Wei Liu, Hao Lin, Zhenhua Li et al.WWW 2023 · 4 citations
- PEER: A Collaborative Language ModelTimo Schick, Jane A. Yu, Zhengbao Jiang, Fabio Petroni et al.ICLR 2023 · 44 citations
- SplitNet: A Reinforcement Learning Based Sequence Splitting Method for the MinMax Multiple Travelling Salesman ProblemHebin Liang, Yi Ma, Zilin Cao, Tianyang Liu et al.AAAI 2023 · 13 citations
- SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and TrainingJierun Chen, Dongting Hu, Xijie Huang, Huseyin Coskun et al.CVPR 2025
- Easy Regional Contrastive Learning of Expressive Fashion RepresentationsDaiqing Qi, Handong Zhao, Sheng LiNeurIPS 2024 · 4 citations
