---
title: "Measured Coupling for Parallel Decoding in Diffusion Language Models"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.08307v1"
permalink: "https://feed.astralor.com/items/itm_15218e7c0aa6c71d81570234"
published: "2026-11-05T17:10:03.000Z"
score: 55
channel: "tech"
tags: ["RAG","Measured Coupling","Parallel Decoding","Diffusion Language Models","Diffusion","Mean-field","Cheap"]
---

# Measured Coupling for Parallel Decoding in Diffusion Language Models

## At a glance

Diffusion language models can write many tokens per step, but must choose which tokens are safe to commit together. Mean-field decoding penalizes pairs of positions whose predictions look alike. We replace this proxy wi…

## Key points

- Diffusion language models can write many tokens per step, but must choose which tokens are safe to commit together.
- Mean-field decoding penalizes pairs of positions whose predictions look alike.
- We replace this proxy with the coupling itself: how much committing one token changes the model's confidence at another.

## Full text

Diffusion language models can write many tokens per step, but must choose which tokens are safe to commit together. Mean-field decoding penalizes pairs of positions whose predictions look alike. We replace this proxy with the coupling itself: how much committing one token changes the model's confidence at another. Cheap probes inside the main forward pass measure it, with no training or extra model, and it plugs into the same mean-field rule. On three models and four benchmarks, our method needs far fewer steps than mean-field decoding and is faster in most settings, also with a key-value cache, at an average accuracy cost of 1.1 points. Shuffling the couplings or replacing them with a uniform bonus loses this gain.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_15218e7c0aa6c71d81570234