Pith. sign in

REVIEW 13 cited by

VMoBA: Mixture-of-Block Attention for Video Diffusion Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.23858 v1 pith:5UWBZ4KN submitted 2025-06-30 cs.CV

VMoBA: Mixture-of-Block Attention for Video Diffusion Models

classification cs.CV
keywords attentionvideoblockvmobaspatio-temporalvdmsdiffusiondynamically
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The quadratic complexity of full attention mechanisms poses a significant bottleneck for Video Diffusion Models (VDMs) aiming to generate long-duration, high-resolution videos. While various sparse attention methods have been proposed, many are designed as training-free inference accelerators or do not optimally capture the unique spatio-temporal characteristics inherent in video data when trained natively. This paper introduces Video Mixture of Block Attention (VMoBA), a novel sparse attention mechanism specifically adapted for VDMs. Motivated by an in-depth analysis of attention patterns within pre-trained video transformers, which revealed strong spatio-temporal locality, varying query importance, and head-specific concentration levels, VMoBA enhances the original MoBA framework with three key modifications: (1) a layer-wise recurrent block partition scheme (1D-2D-3D) to dynamically adapt to diverse spatio-temporal attention patterns and improve efficiency; (2) global block selection to prioritize the most salient query-key block interactions across an entire attention head; and (3) threshold-based block selection to dynamically determine the number of attended blocks based on their cumulative similarity. Extensive experiments demonstrate that VMoBA significantly accelerates the training of VDMs on longer sequences, achieving 2.92x FLOPs and 1.48x latency speedup, while attaining comparable or even superior generation quality to full attention. Furthermore, VMoBA exhibits competitive performance in training-free inference, offering 2.40x FLOPs and 1.35x latency speedup for high-res video generation.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 13 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. HASTE: Training-Free Video Diffusion Acceleration via Head-Wise Adaptive Sparse Attention

    cs.CV 2026-05 unverdicted novelty 7.0

    HASTE delivers up to 1.93x speedup on Wan2.1 video DiTs via head-wise adaptive sparse attention using temporal mask reuse and error-guided per-head calibration while preserving video quality.

  2. Efficient Video Diffusion Models: Advancements and Challenges

    cs.CV 2026-04 unverdicted novelty 7.0

    A survey that groups efficient video diffusion methods into four paradigms—step distillation, efficient attention, model compression, and cache/trajectory optimization—and outlines open challenges for practical use.

  3. Attention Sparsity is Input-Stable: Training-Free Sparse Attention for Video Generation via Offline Sparsity Profiling and Online QK Co-Clustering

    cs.CV 2026-03 conditional novelty 7.0

    Attention sparsity in video DiTs is an input-stable layer-wise property, enabling offline profiling and online bidirectional QK co-clustering for up to 1.93x speedup with PSNR up to 29 dB.

  4. Chimera: Designing and Chinchilla-Scaling Hybrid Visual Diffusion Transformers

    cs.CV 2026-07 conditional novelty 6.5

    A hybrid NoPE visual diffusion backbone with HeteroP scaling yields ~7× pretraining compute efficiency versus matched full attention and near-stable zero-shot 5s→30s video extrapolation.

  5. Light Interaction: Training-Free Inference Acceleration for Interactive Video World Models

    cs.CV 2026-05 unverdicted novelty 6.0

    Light Interaction accelerates interactive video world models up to 2.59x via adaptive context management, denoising cache acceleration, and 3D block sparse attention without retraining.

  6. Veda: Scalable Video Diffusion via Distilled Sparse Attention

    cs.CV 2026-05 unverdicted novelty 6.0

    Veda formulates tile selection in video diffusion attention as a reconstruction problem from full attention maps, using statistics-aware and head-aware scoring to enable high sparsity with maintained quality and hardw...

  7. Rethinking Position Embedding as a Context Controller for Multi-Reference and Multi-Shot Video Generation

    cs.CV 2026-04 conditional novelty 6.0

    SideInfo-RoPE encodes reference-identity agreement as an extra rotary axis, disambiguating similar characters in multi-reference multi-shot video generation while keeping full semantic attention.

  8. SVG-EAR: Parameter-Free Linear Compensation for Sparse Video Generation via Error-aware Routing

    cs.CV 2026-03 conditional novelty 6.0

    Parameter-free centroid compensation plus error-aware block routing yields a better quality–density Pareto frontier for sparse attention in video DiTs than score-based sparsification.

  9. Light Forcing: Accelerating Autoregressive Video Diffusion via Sparse Attention

    cs.CV 2026-02 conditional novelty 6.0

    Light Forcing accelerates autoregressive video diffusion with a chunk-aware sparsity schedule and hierarchical frame-level and block-level attention selection, matching dense-attention quality at 1.2–1.3× speedup.

  10. Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers

    cs.CV 2025-12 conditional novelty 6.0

    LLSA uses hierarchical top-K selection plus coarse-token enrichment to make diffusion-transformer attention O(N log N), giving about 6x faster training at 256x256 pixel resolution with FID comparable to full attention.

  11. End-to-End Training for Autoregressive Video Diffusion via Self-Resampling

    cs.CV 2025-12 conditional novelty 6.0

    Resampling Forcing trains autoregressive video diffusion models on self-resampled degraded histories with a causal mask, achieving stable long-horizon generation without a teacher or discriminator.

  12. Ride the Wave: Precision-Allocated Sparse Attention for Smooth Video Generation

    cs.CV 2026-04 unverdicted novelty 5.0

    PASA uses curvature-aware dynamic budgeting, grouped approximations, and stochastic attention routing to accelerate video diffusion transformers while eliminating temporal flickering from sparse patterns.

  13. Light Forcing: Accelerating Autoregressive Video Diffusion via Sparse Attention

    cs.CV 2026-02 conditional novelty 5.0

    Sparse attention with chunk-aware sparsity growth and hierarchical frame/block selection accelerates autoregressive video diffusion at ~1.3x with VBench quality on par with dense attention.