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).
HELM was built during the era of few-shot in-context learning. The field is moving towards instruction-tuned models intended to be used in a zero-shot manner. We should update to HELM to support this.
HELM was built during the era of few-shot in-context learning. The field is moving towards instruction-tuned models intended to be used in a zero-shot manner. We should update to HELM to support this.