Signals 4 · free daily AI digest

Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE

Paper recorded by Signals 4 on 2026-09-29 in cs.AI. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-29 on arXiv · recorded by Signals 4 on 2026-09-30

Category: cs.AI · 人工智能 · first seen 2026-09-30

Abstract

Mixture-of-Experts (MoE), popularized by large language models, is a promising paradigm for scaling visual generative models. However, conventional token-wise MoE routes tokens independently within a homogeneous expert pool and regularizes expert usage toward uniformity, making it poorly matched to video data that is spatiotemporally redundant and semantically long-tailed. We show that existing vi

Read on arXiv →

#8 most recent of 460 cs.AI papers we have recorded · ↑ newer: AdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self- · ↓ older: Stochastic World Models for Verifying Vision-Based Neural Feedback Sys
Cite this page: Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE: the #8 most recent of 460 cs.AI papers we have recorded (as of 2026-09-29). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/breaking-the-uniformity-trap-scaling-video-diffusion-model-via-splitmoe.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.AI papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions