Metraipolis › archive › dc_ea5b8d541d12
Self-Aligned Forcing: Streaming Video Diffusion with Differentiable Noisy History — arXiv:2609.38114
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.38114
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T18:41:42.566Z
- SHA-256 of the archived text
62293339b0212d4f9e5c0ace6cd7000136ac0b1ba4f06335ba511d24ff367b9d- Stable URL of this page
https://metraipolis.com/documento/dc_ea5b8d541d12
How to cite
Self-Aligned Forcing: Streaming Video Diffusion with Differentiable Noisy History — arXiv:2609.38114 — arXiv (en), https://arxiv.org/abs/2609.38114 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ea5b8d541d12 (Metraipolis archive dc_ea5b8d541d12, sha256 62293339b0212d4f9e5c0ace6cd7000136ac0b1ba4f06335ba511d24ff367b9d)
BibTeX
@misc{dcea5b8d541d12,
title = {Self-Aligned Forcing: Streaming Video Diffusion with Differentiable Noisy History — arXiv:2609.38114},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.38114},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ea5b8d541d12; sha256 62293339b0212d4f9e5c0ace6cd7000136ac0b1ba4f06335ba511d24ff367b9d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38114. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Autoregressive video diffusion enables interactive streaming generation, but suffers from error accumulation over long rollouts. Self-rollout training reduces exposure bias, yet finite rollouts leave long-range drift unresolved. We observe that the noise level of the history key-value (K/V) representations trades visual quality against motion, and that restoring gradients through the history aligns causal training far more closely with bidirectional training. Motivated by these observations, we introduce Self-Aligned Forcing (SAF), a training scheme that aligns the history of each block with the noise level of the block being denoised. Specifically, the history is the K/V produced by preceding blocks at the same denoising stage, so all blocks at a stage can be denoised in a single forward pass under a causal mask. This keeps the noisy history differentiable, allowing future losses to optimize how it is encoded. SAF therefore avoids a separate no-gradient rollout and per-block timestep-zero recaching, training up to 1.8x faster than prior methods with lower memory. At inference, SAF achieves the highest single-GPU throughput among existing methods and keeps one history bank per stage for a multi-GPU pipeline, reaching 49.1 FPS on 4 GPUs. Experiments show superior long-horizon generation with a better balance between visual quality and motion. Project page: this https URL . Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.38114 — 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 2026-10-07T18:04:24.940Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).