Holistic Evaluation of Language Models (HELM), a framework to increase the transparency of language models (https://arxiv.org/abs/2211.09110). This framework is also used to evaluate text-to-image models in Holistic Evaluation of Text-to-Image Models (HEIM) (https://arxiv.org/abs/2311.04287).
This updates these open ended medical scenarios (liveQA and medicationQA) to use model as judge.
This implementation uses 4 buckets as human judges did in the LiveQA paper, and then 3 for MedicationQA (no recommended approach).