Metraipolis › archive › dc_ec5cc234f208
A Better Critic Matters for Distribution Matching Distillation — arXiv:2610.42135
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.42135
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-02T07:52:45.780Z
- SHA-256 of the archived text
b56c9af2e95738a66d11389df3bb4b50c61a6b535225301d9117423539955c0c- Stable URL of this page
https://metraipolis.com/documento/dc_ec5cc234f208
How to cite
A Better Critic Matters for Distribution Matching Distillation — arXiv:2610.42135 — arXiv (en), https://arxiv.org/abs/2610.42135 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ec5cc234f208 (Metraipolis archive dc_ec5cc234f208, sha256 b56c9af2e95738a66d11389df3bb4b50c61a6b535225301d9117423539955c0c)
BibTeX
@misc{dcec5cc234f208,
title = {A Better Critic Matters for Distribution Matching Distillation — arXiv:2610.42135},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.42135},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ec5cc234f208; sha256 b56c9af2e95738a66d11389df3bb4b50c61a6b535225301d9117423539955c0c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.42135. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Distribution Matching Distillation (DMD) trains a few-step student using the difference between a teacher's score and an online critic's estimate of the student distribution's score. We investigate what the critic needs, in capacity and in information, to provide a reliable training signal. On the capacity side, controlled experiments across seven SiT model sizes on ImageNet show that increasing teacher size while keeping a small critic fixed can degrade student quality and cause training collapse. Scaling the critic alongside the teacher, a strategy we call joint scaling, prevents collapse in our experiments and allows the student to benefit from larger teachers. On the information side, rather than forcing the critic to model a step-agnostic mixture of student outputs, we introduce a step-aware critic (STAC) that conditions on the student's generation step to estimate step-specific scores. Together, joint scaling and STAC consistently improve few-step distillation across class-conditional ImageNet generation, text-to-image generation with SDXL and FLUX.2, and text-to-video generation with Wan2.1. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.42135 — 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 2026-11-02T07:40:00.283Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).