AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Context Tokens are Anchors: Understanding the Repetition Curse in Diffusion MLLMs from an Information Flow Perspective

Qiyan Zhao, Xiaofeng Zhang, Shuochen Chang, Qianyu Chen, Xiaosong Yuan, Xuhang Chen, Luoqi Liu, Jiajun Zhang, Xu-Yao Zhang, Da-Han Wang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding

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
enable dllm-cache on LLaDA-VLLaDA-V · COCO caption repetition (512)SRR: 6.9 -> 82.3Hurt
add CTEV to cached model (ablation)LLaDA-V + dllm-cache · COCO caption (512)ARR: 14.3 -> 3.2Improved
add CTAE to cached model (ablation)LLaDA-V + dllm-cache · COCO caption (512)ARR: 14.3 -> 2.9Improved
combine CTEV and CTAE (ablation)LLaDA-V + dllm-cache · COCO caption (512)ARR: 3.2/2.9 -> 1.2Improved
add CoTA to dllm-cacheLLaDA-V · MathVistascore: 54.9 -> 59.7Improved
add CoTA to dllm-cacheLLaDA-V · LLaVA-Wscore: 63.2 -> 69.9Improved
add CoTA to dllm-cacheLLaDA-V · DocVQAaccuracy: 82.1 -> 84.1Improved
add CoTA to dllm-cacheLLaDA-V · MMBenchaccuracy: 83.0 -> 83.1No clear effect
CoTA vs uncached LLaDA-VLLaDA-V · MMStar / SEED-Imageaccuracy: 60.1 -> 59.3; 74.8 -> 73.9Hurt
increase context tokens from 3 to 5 (ablation)CoTA · MathVerseARR: 1.2 -> 3.3Hurt
restrict entropy to layers 26-30 vs earlier layers (ablation)CTEV · COCO caption (512)SRR: 81% (layers 1-10) -> 10% (layers 26-30)Improved
use Gaussian decay instead of ALiBi bias in CTAE (ablation)CTAE · COCO captionARR: 3.9% -> 1.8%Improved

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