AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

DreamOn: Diffusion Language Models For Code Infilling Beyond Fixed-size Canvas

Zirui Wu, Lin Zheng, Zhihui Xie, Jiacheng Ye, Jiahui Gao, Shansan Gong, Yansong Feng, Zhenguo Li, Wei Bi, Guorui Zhou, Lingpeng Kong

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding · 12 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
add DreamOn length-control states to fine-tune Dream-7BDream-7B · HumanEval-Infilling single-linepass@1: 48.2 -> 88.6Improved
add DreamOn to DreamCoder-7B (mask length 64)DreamCoder-7B · HumanEval-Infilling multi-linepass@1: 43.2 -> 63.8Improved
add DreamOn to DreamCoder-7BDreamCoder-7B · SantaCoder-FIM Pythonexact match: 59.3 -> 79.0Improved
add [delete] state (contraction) to DreamOn (ablation)DreamOn · HumanEval-Infilling single-line (avg over mask lengths)pass@1: 67.4 -> 90.8Improved
add [expand] state to DreamOn (ablation)DreamOn · HumanEval-Infilling single-line (avg over mask lengths)pass@1: 73.4 -> 90.8Improved
add [expand] state to DreamOn (ablation)DreamOn · HumanEval-Infilling multi-line (avg over mask lengths)pass@1: 32.4 -> 57.1Improved
add [delete] state to DreamOn (ablation)DreamOn · SantaCoder-FIM Python (avg over mask lengths)exact match: 53.0 -> 77.4Improved
down-weight loss on [delete] tokens (loss balancing) (ablation)DreamOn · HumanEval-Infilling single-line (avg over mask lengths)pass@1: 84.6 -> 90.8Improved
remove [delete] tokens in place instead of keeping placeholders (ablation)DreamOn · HumanEval-Infilling single-line (avg over mask lengths)pass@1: 84.6 -> 90.8Improved
add inference-time deletion broadcasting (ablation)DreamOn · HumanEval-Infilling single-line (avg over mask lengths)pass@1: 90.2 -> 90.8Improved
add deletion broadcasting for efficiency (ablation)DreamOn · generation steps at mask length 64inference steps: 122.8 -> 52.4Improved
use DreamOn vs SFT on ROCStories story ending at mask length 8Dream-7B · ROCStories story endingRouge-L: 25.1 -> 24.6No clear effect

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