AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Locally Coherent Parallel Decoding in Diffusion Language Models

Michael Hersche, Nicolas Menet, Ronan Tanios, Abbas Rahimi

ICML 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding · 0 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 CoDiLA local AR block to Dream-Coder (K=B=4)Dream-Coder-Instruct-7B · HumanEvalPass@1: 24.4 -> 51.8Improved
add CoDiLA local AR block to Dream-Coder (K=B=8)Dream-Coder-Instruct-7B · HumanEvalPass@1: 5.5 -> 29.9Improved
add CoDiLA to Dream-CoderDream-Coder-Instruct-7B · HumanEval syntax errorssyntax error rate: 70 -> 16Improved
use soft-conditioning instead of Top-1 (ablation)CoDiLA (B=4) · HumanEvalPass@1: 36.6 -> 51.8Improved
use soft-conditioning instead of Top-1 (ablation)CoDiLA (B=4) · MBPPPass@1: 33.5 -> 55.6Improved
use soft-conditioning instead of Top-5 (ablation)CoDiLA (B=4) · HumanEvalPass@1: 56.1 -> 51.8Mixed
add boundary tokens around soft embeddings (ablation)CoDiLA (B=4) · NELBONELBO: 15.5 -> 13.6Improved
increase block size B (ablation)CoDiLA · training losslossImproved
scale AR model to 1.7B or 4B (ablation)CoDiLA · HumanEval/MBPPPass@1No clear effect
extend candidate scope from 10 to 50 blocks (ablation)CoDiLA (B=4) · HumanEvalPass@1: 51.8% -> 42.1%Hurt
use global scope with EOS penalty instead of local scope (ablation)CoDiLA · code benchmarksPass@1No clear effect
decode one block instead of two blocks in parallel (B=8) (ablation)CoDiLA · HumanEvalPass@1: 22.3 -> 39.3Improved

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