AI Paper Atlas

Research map / Video / Video generation

Recurrent Autoregressive Diffusion: Global Memory Meets Local Attention

Taiye Chen, Zihan Ding, Anjian Li, Christina Zhang, Zeqi Xiao, Yisen Wang, Chi Jin

ECCV 2026 · Video › Video generation · approach: Efficient video diffusion models · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use clean instead of noised frames as RNN memory input (ablation)RAD LSTM-f · LSTM-f memory-noise ablationPSNR: 14.70 -> 15.30Improved
use clean instead of noised frames as RNN memory input (ablation)RAD LSTM-f · LSTM-f memory-noise ablationSSIM: 0.38 -> 0.41Improved
use frame-wise (stride 1) instead of chunk-wise autoregressionRAD TTT · Memory MazePSNR: 14.36 -> 15.50Improved
use frame-wise instead of chunk-wise autoregressionRAD Mamba2 · Memory MazePSNR: 13.80 -> 15.35Improved
use frame-wise instead of chunk-wise autoregressionRAD LSTM · Memory MazePSNR: 15.64 -> 15.50Mixed
use frame-wise instead of chunk-wise autoregressionRAD Mamba2 · MinecraftPSNR: 12.72 -> 16.70Improved
use frame-wise instead of chunk-wise autoregressionRAD TTT · MinecraftPSNR: 14.04 -> 16.72Improved
use frame-wise instead of chunk-wise autoregressionRAD LSTM · MinecraftPSNR: 14.24 -> 16.59Improved
use LSTM instead of Mamba2 in chunk-wise RADRAD chunk-wise · Memory MazePSNR: 13.80 -> 15.64Improved
use LSTM instead of TTT in frame-wise RADRAD frame-wise · Memory MazePSNR: 15.50 -> 15.50No clear effect
use partial-overlap stride 10 instead of chunk-wise stride 20 for LSTM (ablation)RAD LSTM · Memory MazePSNR: 15.64 -> 15.06Hurt
add frame-wise RAD with LSTM over Diffusion ForcingRAD LSTM-f · Memory MazePSNR: 14.73 -> 15.50Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy