QA eval pipeline for retrieval#1754

Open

KyleZheng1284 wants to merge 17 commits intoNVIDIA:mainfrom

KyleZheng1284:feature/qa-harness-fullpage-pipeline

Commits on Apr 2, 2026

QA eval pipeline with full-page markdown and multi-tier scoring
Kyle Zheng
committed
removed stale ref
Kyle Zheng
committed
fixing ci/cd issues
Kyle Zheng
committed
style: black formatting for QA harness files
Kyle Zheng
committed
update readme
Kyle Zheng
committed

Commits on Apr 3, 2026

migrate eval framework to graph pipeline and also added new changes with better scoring naming, more generic env naming for api keys for multi model support
Kyle Zheng
committed

Commits on Apr 5, 2026

add support for multi run sweep w/ support for multiple different model configs
Kyle Zheng
committed
updated eval sweep to correct model name
Kyle Zheng
committed

Commits on Apr 7, 2026

add scripts for running retrieval_bench
Kyle Zheng
committed
priortize bo767 as dataset in retrieval bench in readme
Kyle Zheng
committed
refactor scripts
Kyle Zheng
committed
add agentic retrieval example
Kyle Zheng
committed

Commits on Apr 9, 2026

refactor and consolidate into retriever cli
Kyle Zheng
committed
update to support only plural + remove cli entry point for retrievalbench
Kyle Zheng
committed

Commits on Apr 10, 2026

bug fixes
Kyle Zheng
committed
remove case
Kyle Zheng
committed
restored singular column names for test
Kyle Zheng
committed