ReFACT: Updating Text-to-Image Models by Editing the Text Encoder

arXiv:2306.00738v2 Announce Type: replace-cross
Abstract: Our world is marked by unprecedented technological, global, and socio-political transformations, posing a significant challenge to text-to-image generative models. These models encode factual associations within their parameters that can quickly become outdated, diminishing their utility for end-users. To that end, we introduce ReFACT, a novel approach for editing factual associations in text-to-image models without relaying on explicit input from end-users or costly re-training. ReFACT updates the weights of a specific layer in the text encoder, …

abstract arxiv become challenge cs.cl cs.cv editing encode encoder generative generative models global image novel parameters political text text-to-image type utility world