AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Sparse-LaViDa: Sparse Multimodal Discrete Diffusion Language Models

Shufan Li, Jiuxiang Gu, Kangning Liu, Zhe Lin, Zijun Wei, Aditya Grover, Jason Kuen

CVPR 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding · 10 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
replace dense MDM parameterization with sparse truncation, register tokens and cachingLaViDa-O · GenEvallatency (s/image): 21.27 -> 10.86Improved
replace dense MDM parameterization with sparse oneLaViDa-O · GenEvaloverall score: 0.77 -> 0.78No clear effect
use sparse parameterizationLaViDa-O · ImgEditlatency (s): 63.98 -> 22.55Improved
use sparse parameterizationLaViDa-O · ImgEditoverall score: 3.71 -> 3.79Improved
use sparse parameterizationLaViDa-O · MathVistalatency (s): 10.41s -> 3.72sImproved
use sparse parameterization instead of Fast-dLLM cachingLaViDa-O · MathVistaaccuracy: 56.1 -> 56.7Improved
use sparse parameterizationLaViDa-O · image understanding suite (Table 5)accuracy: MME 488 -> 450; MMMU 45.1 -> 43.6; ChartQA 80.0 -> 82.0Mixed
use sparse parameterizationLaViDa-O · MJHQ-30kFID: 6.68 -> 7.63Hurt
train baseline on same 20M data subset vs sparseSparse-LaViDa · MJHQ-30kFID: 8.11 -> 7.63Improved
add 64 register tokens (vs 0) (ablation)Sparse-LaViDa · DPG-Benchscore: 80.3 -> 82.4Improved
add 64 register tokens (vs 0) (ablation)Sparse-LaViDa · MJHQ-30kFID: 9.32 -> 7.63Improved
add 64 register tokens (vs 0) (ablation)Sparse-LaViDa · GenEvaloverall score: 0.76 -> 0.78No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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