Metraipolis › archive › dc_6a8ee136f3cc
Intrinsic Distribution Matching — arXiv:2610.39633
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.39633
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:13:21.112Z
- SHA-256 of the archived text
3ad06f826a3eb2cb0e8219b305f883cae1e570630e84546ad1d78cfdc272c061- Stable URL of this page
https://metraipolis.com/documento/dc_6a8ee136f3cc
How to cite
Intrinsic Distribution Matching — arXiv:2610.39633 — arXiv (en), https://arxiv.org/abs/2610.39633 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6a8ee136f3cc (Metraipolis archive dc_6a8ee136f3cc, sha256 3ad06f826a3eb2cb0e8219b305f883cae1e570630e84546ad1d78cfdc272c061)
BibTeX
@misc{dc6a8ee136f3cc,
title = {Intrinsic Distribution Matching — arXiv:2610.39633},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.39633},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6a8ee136f3cc; sha256 3ad06f826a3eb2cb0e8219b305f883cae1e570630e84546ad1d78cfdc272c061},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.39633. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
One-step and few-step text-to-image generation must accurately render the details specified in the prompt despite limited opportunities for refinement. Distribution matching distillation (DMD) uses score differences to bring the generated distribution closer to a reference distribution. Following prior work, we estimate this difference with internal velocity branches, avoiding a separate fake-score network. We ask whether an explicit moment constraint improves this learned score supervision. A decomposition of the reverse-KL field into a Gaussian moment component and a non-Gaussian residual identifies statistics that can be constrained explicitly. We introduce Intrinsic Distribution Matching (IDM), which retains the full DMD correction and matches feature means and per-coordinate standard deviations using the generator's exponential-moving-average (EMA) copy, requiring no external feature encoder. Experiments use full-parameter and low-rank adaptation (LoRA) tuning on Z-Image-Turbo 6B and Qwen-Image 20B. At two network evaluations, IDM reaches 94.25% and 90.20% on Chinese and English LongText-Bench, the highest scores among the compared methods at this budget. Controlled ablations show gains from the added moment loss, while generator EMA features perform competitively with an external DINOv2 encoder across the five reported metrics. Code is available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.39633 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T06:07:08.727Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).