Metraipolis › archive › dc_23bd4500065b
CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation — arXiv:2609.14725
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.14725
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T01:53:18.091Z
- SHA-256 of the archived text
49bb6d2843497e0e6b550e177b265ce1b99d495f49c36ac5b2949ceb8efb3d60- Stable URL of this page
https://metraipolis.com/documento/dc_23bd4500065b
How to cite
CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation — arXiv:2609.14725 — arXiv (en), https://arxiv.org/abs/2609.14725 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_23bd4500065b (Metraipolis archive dc_23bd4500065b, sha256 49bb6d2843497e0e6b550e177b265ce1b99d495f49c36ac5b2949ceb8efb3d60)
BibTeX
@misc{dc23bd4500065b,
title = {CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation — arXiv:2609.14725},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.14725},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_23bd4500065b; sha256 49bb6d2843497e0e6b550e177b265ce1b99d495f49c36ac5b2949ceb8efb3d60},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.14725. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Few-step distillation accelerates diffusion models but must balance diversity and fidelity: trajectory-based distillation preserves mode coverage, while distribution matching sharpens samples but can reduce diversity. We show that this tension can be exploited in a noise-regime-dependent way: high-noise steps largely determine global modes, whereas low-noise steps refine local details. We propose CrossDistill, a trajectory-level hybrid distillation framework that splits the sampling trajectory at a crossover point, applies a trajectory-preserving objective on the high-noise interval and a distribution-matching objective on the low-noise interval, and couples the two stages through the crossover state. In contrast to loss-level mixing, and complementarily to training-time two-stage recipes, CrossDistill explicitly assigns complementary objectives along the noise axis, so that global branching is preserved before local statistics are sharpened. CrossDistill is a noise-level scheduling policy: PCM and DMD are plug-in instantiations, while the noise partition, crossover coupling, and objective ordering are the key design elements. Experiments on text-to-video diffusion models and qualitative image-to-video results show that CrossDistill expands the few-step quality-diversity frontier, retaining seed-level variation while achieving competitive visual fidelity. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-15 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.14725 — 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 2026-10-16T01:25:23.621Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).