Metraipolis › archive › dc_33927333eece

Unlocking the Potential of Contrastive Regularization for Diffusion Transformer Training — arXiv:2611.02921

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02921
License (as archived)
CC0-1.0
Archived
2026-11-03T11:26:45.901Z
SHA-256 of the archived text
8a6255a1c3429e6701d23d1ef3155f030d3d54bfb07ca6cd10480eb95ddef443
Stable URL of this page
https://metraipolis.com/documento/dc_33927333eece

How to cite

Unlocking the Potential of Contrastive Regularization for Diffusion Transformer Training — arXiv:2611.02921 — arXiv (en), https://arxiv.org/abs/2611.02921 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_33927333eece (Metraipolis archive dc_33927333eece, sha256 8a6255a1c3429e6701d23d1ef3155f030d3d54bfb07ca6cd10480eb95ddef443)
BibTeX
@misc{dc33927333eece,
  title = {Unlocking the Potential of Contrastive Regularization for Diffusion Transformer Training — arXiv:2611.02921},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02921},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_33927333eece; sha256 8a6255a1c3429e6701d23d1ef3155f030d3d54bfb07ca6cd10480eb95ddef443},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02921. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Representation regularization has emerged as an effective way to improve diffusion transformer training. Yet one of the most canonical regularizers--InfoNCE-style contrastive learning--remains largely underexplored in this setting. We naturally ask a question: can standard contrastive learning serve as an effective regularizer for diffusion transformers? Surprisingly, we find that naive InfoNCE regularization brings only marginal gains. Our analysis reveals that the objective is easily dominated by a shortcut: near-orthogonal noise components in diffusion inputs provide a bypass for enlarging negative-pair distances while weakening positive-pair alignment, especially at high noise levels. To unlock the benefit of contrastive regularization, we propose Clean Representation-ANchored Contrastive (CRAN) regularization. Specifically, CRAN anchors contrastive dispersion with clean-sample representation distances, transferring clean relational semantic structure to noisy diffusion representations to suppress the noise shortcut. Extensive experiments on ImageNet show that CRAN consistently improves diffusion transformer training across model scales and training budgets, outperforming internal feature-alignment and dispersive regularization baselines while matching external DINO-guided REPA. These results re-establish InfoNCE-based contrastive learning as an effective regularizer for diffusion training, once its noise-driven over-dispersion is properly constrained.

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/2611.02921 — 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 2026-11-03T11:02:37.891Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).