How well can Text-to-Image Generative Models understand Ethical Natural Language Interventions? arxiv: https://arxiv.org/abs/2210.15230
Our codebase is broadly similar to DallEval. However, we provide inference scripts to run DALL.E-mini and Stable Diffusion too. Additionally, we have the capability to intervene on the prompts.
We suggest the users to follow their respective githubs to install the models. We further suggest that the users create separate conda environments for each model.
The users can follow the this to setup minDALL-E. Our scripts are subset of this codebase.
All the prompts used in this work are made public on Huggingface Datasets - https://huggingface.co/datasets/hbXNov/entigen