Skip to content
AI News HubLIVE
Original source2 min read

InstaGen: Group-Shared Importance Sampling for Fast Video Generation

Summary

InstaGen recasts block selection in permuted block-sparse attention (PBSA) as importance sampling: top-scoring keys are computed exactly, the rest are sampled in proportion to estimated attention and reweighted by inverse sampling probability, with one sample shared across each query cluster. Parameterized by SVG2, it improves PSNR by 3.4 dB and 2 dB over SVG2 on Wan2.1-T2V-1.3B and 14B at matched attention density, and by 1.6 dB and 2.2 dB at matched wall-clock time; on the 14B model it also beats SVG-EAR by 1.1 dB at similar density while running faster.

SourcearXiv Computer VisionAuthor: Kislay Aditya Oj, Kumar Krishna Agrawal, Sri Surya Varma Datla, Aditya Desai
InstaGen: Group-Shared Importance Sampling for Fast Video Generation
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 5 Nov 2026]

Title:InstaGen: Group-Shared Importance Sampling for Fast Video Generation

View a PDF of the paper titled InstaGen: Group-Shared Importance Sampling for Fast Video Generation, by Kislay Aditya Oj and 3 other authors

View PDF HTML (experimental)

Abstract:Diffusion Transformers (DiTs) are the dominant architecture for video generation, but their quadratic attention complexity results in substantial computational cost and latency. Permuted block-sparse attention (PBSA) offers a promising approach to reducing this cost by selectively computing blocks of a semantically permuted attention matrix. However, existing approaches typically select only the top few key blocks for each query block, following a top-$p$ paradigm. We find that top-$p$ selection can be highly sample inefficient in high-entropy attention regimes, which are common in video generation. In these regimes, we show that sampling can provide substantially better approximations under the same computational budget. More generally, we cast block selection as importance sampling, where the top-scoring keys form an exact set that is computed in full and the remaining keys are represented by a sample drawn in proportion to their estimated attention and reweighted by inverse sampling probability. Building on this formulation, we present InstaGen, which can be parameterized by any PBSA method: it replaces the method's block selection with an exact set plus a weighted sample. Each query cluster draws one sample shared by all its queries, so the sampled keys form blocks amenable to efficient GPU execution. In this paper, we take the SVG family of PBSA methods and show that, on Wan2.1-T2V-1.3B and Wan2.1-T2V-14B, InstaGen parameterized by SVG2 improves PSNR over SVG2 by 3.4 dB and 2 dB, respectively, at matched attention density, and by 1.6 dB and 2.2 dB at matched wall-clock time. On Wan2.1-T2V-14B, InstaGen also beats SVG-EAR by 1.1 dB PSNR at nearly the same density while running faster (1.65x against 1.48x over dense).

Comments: 13 pages, 6 figures, 2 tables

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2611.08610 [cs.CV]

(or arXiv:2611.08610v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2611.08610

arXiv-issued DOI via DataCite

Submission history

From: Kislay Aditya Oj [view email] [v1] Thu, 5 Nov 2026 19:04:14 UTC (4,599 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled InstaGen: Group-Shared Importance Sampling for Fast Video Generation, by Kislay Aditya Oj and 3 other authors

View PDF

HTML (experimental)

TeX Source

view license

Additional Features

Audio Summary

Current browse context:

cs.CV

new | recent | 2026-11

Change to browse by:

cs

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • Generalizes PBSA top-p block selection to importance sampling with an exact top set plus inverse-probability-weighted samples.
  • Query clusters share a single sample so sampled keys form GPU-friendly blocks, compatible with any PBSA method.
  • With SVG2 parameterization, InstaGen improves PSNR by 3.4 dB/2 dB at matched density and 1.6 dB/2.2 dB at matched time on Wan2.1-T2V-1.3B/14B.
  • On Wan2.1-T2V-14B it beats SVG-EAR by 1.1 dB PSNR at nearly the same density and runs at 1.65x versus 1.48x over dense.

Highlights and analysis are generated automatically and may contain errors. Check the original source.