DiGraphHal-Bench: Evaluating Multimodal Large Language Models on Complex Directed Graphs
Yixin Fan, Zhao He, Yuxin Hou, Changhua Zhou, Zihao Liu, Peng Wang, Chenglong Lu, Xu Zhang, Wei Wang
摘要
Element-Centric Class 1 -What / Which Q (Non-Semantic): Develop Solution -> 4. According to the graph, find the 1st blue-filled with pink-bordered node in order in the subpath. Q (Semantic): I have successfully reached Deploy Solution, what is the first critical milestone I must encounter in the four steps after this node? All the nodes in the picture that are pink-bordered and have blue fill-color represent critical milestones. Q (Semantic-Rewrite): I'm at Solution Deployment now, what is the first key milestone I must have passed within its four preceding steps? All the nodes in the picture that have blue fill and pink outline represent key milestones. A: Evaluate Performance Path-Centric Class 2 -How Q (Non-Semantic): Start Project -> -> Test and Validate. According to the graph, find at least 1 paths that include most yellow-bordered solid edges. Q (Semantic): ..(Similar to Semantic-Rewrite.) Q (Semantic-Rewrite): I'm at Beginning Project now, how can I reach Evaluate and Verify? Show me at least one possible path that covers the most important steps marked with yellow-bordered solid edges? A: Gather Resources -> Develop Solution Semantic Category Q (Semantic): Imagine you're working on a project where you need to Refine Solution before clearly Deploy Solution, however, it seems some steps are missing in between. What are they? Q (Semantic-Rewrite): Picture you're handling a project where Refining an Initial Solution is required prior to precisely Implementing Solutions. What are the steps in between? A
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
它引用的顶会 Paper19
- Evaluating Object Hallucination in Large Vision-Language ModelsYifan Li, Yifan Du, Kun Zhou, Jinpeng Wang 等EMNLP 2023 · 被引用 344 次
- TableBench: A Comprehensive and Complex Benchmark for Table Question AnsweringXianjie Wu, Jian Yang, Linzheng Chai, Ge Zhang 等AAAI 2025 · 被引用 138 次
- Multi-Object Hallucination in Vision Language ModelsXuweiyi Chen, Ziqiao Ma, Xuejun Zhang, Sihan Xu 等NeurIPS 2024 · 被引用 77 次
- GITA: Graph to Visual and Textual Integration for Vision-Language Graph ReasoningYanbin Wei, Shuai Fu, Weisen Jiang, Zejian Zhang 等NeurIPS 2024 · 被引用 56 次
- VisionGraph: Leveraging Large Multimodal Models for Graph Theory Problems in Visual ContextYunxin Li, Baotian Hu, Haoyuan Shi, Wei Wang 等ICML 2024 · 被引用 33 次
相关 Paper
- PromotionLens: Inspecting Promotion Strategies of Online E-commerce via Visual AnalyticsChenyang Zhang, Xiyuan Wang, Chuyi Zhao, Yijing Ren 等IEEE VIS 2022 · 被引用 15 次
- EditDuet: A Multi-Agent System for Video Non-Linear EditingMarcelo Sandoval-Castañeda, Bryan C. Russell, Josef Sivic, Gregory Shakhnarovich 等SIGGRAPH 2025 · 被引用 6 次
- Benchmark Platform for Ultra-Fine-Grained Visual Categorization Beyond Human PerformanceXiaohan Yu, Yang Zhao, Yongsheng Gao, Xiaohui Yuan 等ICCV 2021 · 被引用 37 次
- GRES: Generalized Referring Expression SegmentationChang Liu, Henghui Ding, Xudong JiangCVPR 2023
- Efficient Incremental GR(1) Synthesis via Monotonic Fixed-Point ReuseSirui Liu, Wei Dong, Yijie Zheng, Haonan GuoOOPSLA 2026
