Tag
1 article
Researchers from PSU and Duke University develop a framework to automatically identify which agent in an LLM multi-agent system causes task failures and when the failure occurs.