InfoMosaic-Bench: Evaluating Multi-Source Information Seeking in Tool-Augmented Agents

Yaxin Du, Yuanshuo Zhang, Xiyuan Yang, Yifan Zhou, Cheng Wang, Gongyi Zou, Xianghe Pang, Wenhao Wang, Menglan Chen, Shuo Tang, Zhiyu Li, Feiyu Xiong, Siheng Chen

Code   PDF

• Role: co-author.
• We introduce InfoMosaic-Bench to evaluate multi-source information seeking for tool-augmented agents.
• Accepted by ICLR 2026.

FLORA: GNNs as Predictors of Agentic Workflow Performances

Yuanshuo Zhang, Yuchen Hou, Bohan Tang, Shuo Chen, Muhan Zhang, Xiaowen Dong, Siheng Chen

Code   PDF

• Role: first author.
• We propose FLORA to use GNNs as predictors of agentic workflow performance, reducing repeated LLM invocations during evaluation.
• Accepted by LoG 2025.

ERASE: Error-Resilient Representation Learning on Graphs for Label Noise Tolerance

Ling-Hao Chen^*, Yuanshuo Zhang^*, et al.

Project   Code   Video

• Role: co-first author.
• We introduce an error-resilient mechanism for graph representation learning under label noise.
• Accepted by CIKM 2024.

EmboCoach-Bench: Benchmarking AI Agents on Developing Embodied Robots

Zixing Lei^*, Genjia Liu^*, Yuanshuo Zhang^*, Qipeng Liu, Chuan Wen, Shanghang Zhang, Wenzhao Lian, Siheng Chen

PDF

• Role: core contributor.
• We benchmark AI agents on embodied robot development tasks, highlighting end-to-end tool use and iteration.

AgentIF-OneDay: A Task-level Instruction-Following Benchmark for General AI Agents in Daily Scenarios

Code   PDF

• Role: contributor.
• We provide a task-level instruction-following benchmark for daily scenarios to evaluate general AI agents.