Reproduce “Intrinsic Distribution Matching” 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
Deyuan Liu, Yansen Han, Kaiwen Zheng, Jun Xie, Jiacheng Liu, Zhenglin Cheng, Peng Sun, Jintao Zhang and 2 more
Intrinsic Distribution Matching (IDM) adds pooled feature-mean and standard-deviation matching to distribution-matching distillation while retaining its score-difference correction and short inference path. In a matched two-evaluation ablation, it raises GenEval from 77.46% to 77.81%. On Z-Image-Turbo 6B, it scores 94.25% on LongText-CN and 90.20% on LongText-EN, highest among compared methods at that budget.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.