rag-eval — offline evaluation harness by Mark Mandrykrag-eval — offline evaluation harness by Mark Mandryk

rag-eval — offline evaluation harness

Mark Mandryk

Mark Mandryk

rag-eval — offline evaluation harness for RAG retrieval. Metrics computed against a golden set, no model API key required to run it. A compare command fails the build outright when recall regresses against the baseline. github.com/0mandrock1/rag-eval
Like this project

Posted Sep 11, 2026

rag-eval — offline evaluation harness for RAG retrieval. Metrics computed against a golden set, no model API key required to run it. A compare command fails ...