← Search

Junzhi Chen

2 accepted papers

2026

AppWorld-UL: Benchmarking Diverse Agent-User Interactions for Tool-Use

ICML 2026poster

Tool-use agents that address day-to-day digital tasks such as ordering groceries must not only operate applications, but also interact with the user, e.g., to ask clarification questions, prompt for confirmation, and inform the user when the instruction is infeasible. However, current benchmarks for…

Cited by 0SourceScholar
2025

Smurfs: Multi-Agent System using Context-Efficient DFSDT for Tool Planning

NAACL 2025long

Teaching large language models (LLMs) to use tools for solving complex problems can grant them human-like reasoning abilities. ReAct and its variants are popular frameworks for tool use in both single-agent and multi-agent systems. To address issues like error propagation and limited exploration in…