Metraipolis › archive › dc_bb513196a66d
An End-to-End Latent-Rollout Approach for Pushing Few-Step ImageNet-$256$ Generation to FID $1.11$ without Fréchet Losses — arXiv:2609.32376
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32376
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T23:50:52.399Z
- SHA-256 of the archived text
5e57ad577cd6a8f5dd3b05c5688f01b7d7cd26b28c465a617da538b2faca32aa- Stable URL of this page
https://metraipolis.com/documento/dc_bb513196a66d
How to cite
An End-to-End Latent-Rollout Approach for Pushing Few-Step ImageNet-$256$ Generation to FID $1.11$ without Fréchet Losses — arXiv:2609.32376 — arXiv (en), https://arxiv.org/abs/2609.32376 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bb513196a66d (Metraipolis archive dc_bb513196a66d, sha256 5e57ad577cd6a8f5dd3b05c5688f01b7d7cd26b28c465a617da538b2faca32aa)
BibTeX
@misc{dcbb513196a66d,
title = {An End-to-End Latent-Rollout Approach for Pushing Few-Step ImageNet-$256$ Generation to FID $1.11$ without Fréchet Losses — arXiv:2609.32376},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32376},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bb513196a66d; sha256 5e57ad577cd6a8f5dd3b05c5688f01b7d7cd26b28c465a617da538b2faca32aa},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32376. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Iterative generation poses a joint optimization problem across steps, as intermediate predictions shape subsequent computations and ultimately determine the final output distribution. Few-step generators distilled from pretrained diffusion and flow-matching models make such optimization computationally practical end to end. We build on this opportunity with a distill-then-refine approach that uses teacher imitation to establish a strong initialization for a few-step rollout in latent space, then shifts to end-to-end refinement of the complete latent rollout against real data. We introduce FiST (Flow-in-Stage Transformer), an architecture that composes learned latent-state transitions in a few stages using a shared Transformer, with optional cross-stage hidden communication. Distillation applies teacher-forced regression to selected states along teacher trajectories; refinement replaces this supervision with adversarial and auxiliary classification objectives on the final latent output. A trainable discriminator module operates on semantically rich features extracted from clean real and generated latents by a frozen SiT backbone pretrained with REPA. All training takes place in latent space, without image decoding. During refinement, FiST consumes its own intermediate predictions, and endpoint gradients pass through every generation stage. For class-conditional generation on ImageNet at $256\times256$, our approach achieves FID 1.11 (IS 282) with three stages and FID 1.15 (IS 280) with two. These results demonstrate competitive few-step generation through learned distribution-level supervision, without explicit Fréchet-distance minimization. Ablations characterize how distillation, pretrained checkpoint choices, refinement supervision, and cross-stage hidden communication affect generation quality. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.32376 — 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 2026-10-07T23:29:27.026Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).