AgentBench: Evaluating LLMs as Agents

Evaluating Large Language Models (LLMs) as agents in interactive environments, highlighting the performance gap between API-based and open-source models, and introducing the AgentBench benchmark.

arxiv.org

Saved by Darren LI

RelatedCollectionsHighlightsNotes

Our extensive test over 25 LLMs (including APIs and open-sourced models) shows that, while top commercial LLMs present a strong ability of acting as agents in complex environments, there is a significant disparity in performance between them and open-sourced competitors.

Xiao Liu • AgentBench: Evaluating LLMs as Agents

AgentBench: Evaluating LLMs as Agents