10.9. LLM评估¶

10.9.1. AgentBench¶

https://github.com/THUDM/AgentBench
A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR’24)