About MCP Server
#4
by xiaoyaojun - opened
I am interested in reproducing your SimpleQA benchmark results, especially as reported for the MCP-enabled mode. Could you kindly share which external tools, plugins, or data sources were configured for the MCP server during your evaluation (e.g. specific retrievers, databases, or search APIs)?
Thanks a lot for your excellent work!
we use this https://github.com/marcopesani/mcp-server-serper
we use langgraph agent loop with no tool, basically just the same as normal chat