MoGA: Mixture-of-Groups Attention for End-to-End Long Video Generation

Weinan Jia; Yuning Lu; Mengqi Huang; Hualiang Wang; Binyuan Huang; Nan Chen; Mu Liu; Jidong Jiang; Zhendong Mao

返回文献探索

Paper · arXiv 2510.18692

MoGA: Mixture-of-Groups Attention for End-to-End Long Video Generation

Weinan Jia, Yuning Lu, Mengqi Huang, Hualiang Wang, Binyuan Huang, Nan Chen, Mu Liu, Jidong Jiang, Zhendong Mao

41 upvotesOctober 21, 2025arXiv 预印本

AI 摘要

Mixture-of-Groups Attention (MoGA) enables efficient long video generation by addressing the quadratic scaling issue of full attention in Diffusion Transformers.

Diffusion TransformersDiTsfull attentionsequence lengthsparse attentiontoken routersemantic-aware routingkernel-free methodFlashAttentionsequence parallelismlong video generationminute-levelmulti-shot480p24 fpscontext length

Abstract

Long video generation with Diffusion Transformers (DiTs) is bottlenecked by the quadratic scaling of full attention with sequence length. Since attention is highly redundant, outputs are dominated by a small subset of query-key pairs. Existing sparse methods rely on blockwise coarse estimation, whose accuracy-efficiency trade-offs are constrained by block size. This paper introduces Mixture-of-Groups Attention (MoGA), an efficient sparse attention that uses a lightweight, learnable token router to precisely match tokens without blockwise estimation. Through semantic-aware routing, MoGA enables effective long-range interactions. As a kernel-free method, MoGA integrates seamlessly with modern attention stacks, including FlashAttention and sequence parallelism. Building on MoGA, we develop an efficient long video generation model that end-to-end produces minute-level, multi-shot, 480p videos at 24 fps, with a context length of approximately 580k. Comprehensive experiments on various video generation tasks validate the effectiveness of our approach.