Reproduce “Four-Step Text-Bearing Image Generation via Distillation with Local Response Regularization and Teacher-Guided Training” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Kairui Wen, Sheng Ye, Yong-Jin Liu
The authors distill GLM-Image’s denoiser to four steps while retaining its autoregressive visual prior, glyph encoder, and VAE. They add a local-response regularizer and train with synthetic text and teacher-generated scene images. On GLM-Image, the four-step student scores 0.84 GenEval Overall and 0.5045 word accuracy, versus 0.81 and 0.1690 for four-step TDM, respectively; results do not guarantee exact text.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.