topsky2222's picture
Release v0.1: eval set with human/LLM kappas, cross-retriever metrics, judge rubric, and leakage audit
76bb000 verified