Holistic Evaluation of Text-to-Image Models: Datasheet | HackerNoon
Briefly

The HEIM benchmark was created to holistically evaluate text-to-image models across diverse aspects, ensuring comprehensive assessment beyond just alignment and quality.
HEIM evaluates across a spectrum of 12 crucial aspects for real-world deployment, including image alignment, quality, aesthetics, originality, reasoning, knowledge, bias, and more.
Read at Hackernoon
[
|
]