Agents2 min read
Evaluating OpenWiki with WikiBench: A New Agent Assessment Framework
LangChain’s WikiBench benchmark assesses OpenWiki’s effectiveness in generating codebase documentation by evaluating generated wikis using questions grounded in the underlying codebase. The framework uses a reader agent to judge the wiki’s quality, revealing performance differences across models and highlighting the value of a wiki as an index for agents.
From LangChain blog






