SGF+: Decoupling Gradient Flows for Autoregressive Video Generation

Hugging Face Daily Papers Papers
autoregressive-video-generation video-diffusion long-horizon-generation gradient-flow self-supervised-learning

Summary

论文提出 Self Gradient Forcing Plus（SGF+），通过为上下文写入与去噪角色解耦参数并保留因果注意力交互，显著提升自回归视频生成的视觉质量与长时一致性，仅用5秒训练片段即可支持长达24小时的连续生成。

Autoregressive video generation requires denoising the current frames while writing their key-value representations as context for future predictions. However, these two roles typically share parameters, and we find that their gradients exhibit distinct patterns and systematic negative alignment, hindering the joint optimization of visual quality and temporal consistency. We introduce Self Gradient Forcing Plus (SGF+), which assigns separate parameters to context writing and denoising while preserving their interaction through causal attention. Both roles are jointly optimized using the original generation objective without auxiliary losses, with context writing supervised through its contribution to future predictions. This simple change improves visual quality and long-horizon consistency over the evaluated baselines in both framewise and chunkwise generation, without additional video training data or a longer training horizon. Trained on only 5s rollouts, SGF+ supports continuous generation for up to 24 hours without long-video fine-tuning. These results highlight role-specific parameterization as an effective design principle for high-quality autoregressive video generation and native long-horizon extrapolation.
Original Article
View Cached Full Text

Cached at: 10/08/26, 12:26 PM

Paper page - SGF+: Decoupling Gradient Flows for Autoregressive Video Generation

Source: https://huggingface.co/papers/2610.10429 Authors:

,

,

,

,

,

,

,

,

,

Abstract

Autoregressivevideogenerationrequiresdenoisingthecurrentframeswhilewritingtheirkey-valuerepresentationsascontextforfuturepredictions.However,thesetworolestypicallyshareparameters,andwefindthattheirgradientsexhibitdistinctpatternsandsystematicnegativealignment,hinderingthejointoptimizationofvisualqualityandtemporalconsistency.WeintroduceSelfGradientForcingPlus(SGF+),whichassignsseparateparameterstocontextwritinganddenoisingwhilepreservingtheirinteractionthroughcausalattention.Bothrolesarejointlyoptimizedusingtheoriginalgenerationobjectivewithoutauxiliarylosses,withcontextwritingsupervisedthroughitscontributiontofuturepredictions.Thissimplechangeimprovesvisualqualityandlong-horizonconsistencyovertheevaluatedbaselinesinbothframewiseandchunkwisegeneration,withoutadditionalvideotrainingdataoralongertraininghorizon.Trainedononly5srollouts,SGF+supportscontinuousgenerationforupto24hourswithoutlong-videofine-tuning.Theseresultshighlightrole-specificparameterizationasaneffectivedesignprincipleforhigh-qualityautoregressivevideogenerationandnativelong-horizonextrapolation.

View arXiv pageView PDFProject pageGitHub15Add to collection

Get this paper in your agent:

hf papers read 2610\.10429

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper1

#### ZihanSu/Self_Gradient_Forcing_Plus Text-to-Video• Updatedabout 10 hours ago • 6

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.10429 in a dataset README.md to link it from this page.

Spaces citing this paper2

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Self Gradient Forcing: Native Long Video Extrapolation

Papers with Code Trending

Proposes Self Gradient Forcing (SGF), a two-pass training strategy for autoregressive video diffusion models that provides missing supervision for writing useful context memory, enabling strong long-video extrapolation even from short training windows.

Speculative Forcing for Efficient Autoregressive Video Generation

Hugging Face Daily Papers

Speculative Forcing 通过将中间 clean estimates 视为草稿并提前接受 token、跳过不必要的去噪步骤，加速少步自回归视频生成，在 LongLive 上以约 1.7 步/ token 匹配四步生成器质量，速度提升超过 1.5 倍。