AI Paper Atlas

Research map / Video / Video generation

Lumos-1: On Autoregressive Video Generation with Discrete Diffusion from a Unified Model Perspective

Hangjie Yuan, Weihua Chen, Jun CEN, Hu Yu, Jingyun Liang, Shuning Chang, Zhihui Lin, Tao Feng, Pengwei Liu, Jiazheng Xing, Hao Luo, Jiasheng Tang and 2 more

ICLR 2026 · Video › Video generation · approach: Autoregressive long video generation

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 temporal tube masks (AR-DF) instead of global random masks (ablation)Lumos-1 1B · VBench Imaging QualityIQ: 0.424 -> 0.559Improved
use temporal tube masks instead of random masks (ablation)Lumos-1 1B · GenEvalGenEval: 0.593 -> 0.591No clear effect
use loss reweighting (step/linear decay) instead of tube masks (ablation)Lumos-1 1B · VBench IQIQ: 0.559 -> 0.464 / 0.410Hurt
add distributed frequency allocation to M-RoPE (ablation)Lumos-1 1B · GenEvalGenEval: 0.414 -> 0.591 (with scaling)Improved
add scaled 3D positions (ablation)Lumos-1 1B · GenEvalGenEval: 0.566 -> 0.591Improved
scale factors (4,8,8) vs (1,1,1) (ablation)MM-RoPE · GenEvalGenEval: 0.566 -> 0.591Improved
scale factors (8,16,16) vs (4,8,8) (ablation)MM-RoPE · VBench IQIQ: 0.559 -> 0.553No clear effect
use MM-RoPE instead of HoPE (ablation)Lumos-1 1B · GenEvalGenEval: 0.570 -> 0.591Improved
omit inference-time masks in AR-DF (ablation)Lumos-1 · video qualityartifactsHurt
small-scale SFT on high quality data (ablation)Lumos-1 · GenEvaloverall: 0.664 -> 0.791 (3.6B, with resolution 352 -> 512)Improved
use mask prediction with KV cache instead of next-token prediction (ablation)Lumos-1 · video generation latencyseconds: 960.0 -> 77.8Improved
add 3D RoPE priors versus 1D RoPE (ablation)Lumos-1 · inference latencylatency: 3.5%-4.1% overheadHurt

Later papers that compared with this method

1 later accepted paper reports Lumos-1 as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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