2 papers accepted to ACL 2025 🎉
May 16, 2025
·
1 min read
Great news! Both MultiAgentBench and EscapeBench have been officially accepted to ACL 2025 Main Conference (Vienna, July 2025).
Highlights
- MultiAgentBench received uniformly positive reviews — one reviewer called it “interesting, forward-looking, and visionary,” noting it “fills a critical gap in current multi-agent evaluation.”
- We will submit the camera-ready version soon and keep updating the codebase & data to make the framework a SWE-bench equivalent for LLM multi-agent research.
Stay tuned — this post will be updated with links to the camera-ready PDF and dataset once they are available.
