mamabench
Domain:
healthcarenatural language processingRecord type:
dataset
Creator:
nmr
Host:
GitHub · 25,949 normalized rows · schema v0.4 · release v0.2.1
A normalized OBGYN / pediatrics / reproductive-health benchmark for evaluating end-to-end medical question-answering systems. Originally built to evaluate MAMAI, a Gemma 4 E4B + RAG medical-advice chatbot for nurses and midwives in Zanzibar.
Release v0.2.1