Ψ-Bench: Evaluating Persona-Sensitive Influencing in Persuasive Dialogues
Benchmark for evaluating persona-sensitive influencing in persuasive dialogues; under review at EMNLP 2026.
May 15, 2026
Which LLM Multi-Agent Protocol to Choose?
First-authored benchmark for evaluating LLM multi-agent communication protocols; introduces ProtocolRouter for adaptive protocol selection. Accepted at ICML 2026.
Oct 20, 2025
ModelingAgent: Bridging LLMs and Mathematical Modeling for Real-World Challenges
Benchmark + agent framework for open-ended real-world mathematical modeling; Findings of EMNLP 2025.
May 21, 2025