This dataset contains the official DMV driver's handbooks from all 50 US states, along with a set of question-answer pairs derived from each handbook.
It is designed to evaluate a system's ability to perform precise information retrieval and question answering across a collection of complex documents, serving as a realistic benchmark for Retrieval-Augmented Generation (RAG) systems.
Question Answering: The primary task is to answer… See the full description on the dataset page:
https://huggingface.co/datasets/corvicai/dmv-handbooks.
