No AI summary available for this article.
Why It Matters
\emph{Distributional training} provides collective supervision for one-step visual generation by matching real and generated features in frozen representation spaces.
Provenance
Discovered via ArXiv and published by ArXiv.
Key Claims
Original description
\emph{Distributional training} provides collective supervision for one-step visual generation by matching real and generated features in frozen representation spaces. We introduce \emph{a unified theoretical framework} that separates distribution modeling from matching discrepancy and connects global objectives to pointwise feature updates through Wasserstein gradient flow. Under this framework, FD-Loss and Gaussian-kernel Drifting are recovered through Gaussian optimal transport and kernel-density-based KL matching, respectively. The framework motivates \textbf{MGFlow}, which models feature d...
Discovered via ArXiv
Research papers and preprints from arXiv.
Publisher: arxiv.org
ID: http://arxiv.org/abs/2609.35763v1 · Indexed 39 minutes ago