Rapid Diffusion: Building Domain-Specifc Text-to-Image Synthesizers with Fast Inference Speed

  • Bingyan Liu
  • , Weifeng Lin
  • , Zhongjie Duan
  • , Chengyu Wang*
  • , Ziheng Wu
  • , Zipeng Zhang
  • , Kui Jia*
  • , Lianwen Jin
  • , Cen Chen
  • , Jun Huang
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Text-to-Image Synthesis (TIS) aims to generate images based on textual inputs. Recently, several large pre-trained diffusion models have been released to create high-quality images with pre-trained text encoders and diffusion-based image synthesizers. However, popular diffusion-based models from the open-source community cannot support industrial domain-specifc applications due to the lack of entity knowledge and low inference speed. In this paper, we propose Rapid Diffusion, a novel framework for training and deploying super-resolution, text-to-image latent diffusion models with rich entity knowledge injected and optimized networks. Furthermore, we employ BladeDISC, an end-to-end Artifcial Intelligence (AI) compiler, and FlashAttention techniques to optimize computational graphs of the generated models for online deployment. Experiments verify the effectiveness of our approach in terms of image quality and inference speed. In addition, we present industrial use cases and integrate Rapid Diffusion to an AI platform to show its practical values.

Original languageEnglish
Title of host publicationIndustry Track
PublisherAssociation for Computational Linguistics (ACL)
Pages295-304
Number of pages10
ISBN (Electronic)9781959429685
DOIs
StatePublished - 2023
Event61st Annual Meeting of the Association for Computational Linguistics, ACL 2023 - Toronto, Canada
Duration: 9 Jul 202314 Jul 2023

Publication series

NameProceedings of the Annual Meeting of the Association for Computational Linguistics
Volume5
ISSN (Print)0736-587X

Conference

Conference61st Annual Meeting of the Association for Computational Linguistics, ACL 2023
Country/TerritoryCanada
CityToronto
Period9/07/2314/07/23

Fingerprint

Dive into the research topics of 'Rapid Diffusion: Building Domain-Specifc Text-to-Image Synthesizers with Fast Inference Speed'. Together they form a unique fingerprint.

Cite this