---
title: "SpecCTMC: Exact Speculative Simulation of Diffusion Language Model Continuous-Time Markov Chains"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.08362v1"
permalink: "https://feed.astralor.com/items/itm_162936968f34ac64e3e76769"
published: "2026-11-05T17:41:20.000Z"
score: 58
channel: "tech"
tags: ["推理效率","SpecCTMC","Exact Speculative Simulation","Diffusion Language Model","Continuous-Time Markov Chains","Masked","Exact"]
---

# SpecCTMC: Exact Speculative Simulation of Diffusion Language Model Continuous-Time Markov Chains

## 速览

Masked diffusion language models offer parallel text generation, but independently sampling multiple tokens can disrupt their dependencies and change the output distribution. Exact continuous-time sampling avoids this a…

## 关键点

- Masked diffusion language models offer parallel text generation, but independently sampling multiple tokens can disrupt their dependencies and change the output distribution.
- Exact continuous-time sampling avoids this approximation, yet typically requires one sequential model evaluation per token.
- We introduce SpecCTMC, a training-free speculative sampler that reduces this sequential bottleneck without a separate draft model.

## 全文

Masked diffusion language models offer parallel text generation, but independently sampling multiple tokens can disrupt their dependencies and change the output distribution. Exact continuous-time sampling avoids this approximation, yet typically requires one sequential model evaluation per token. We introduce SpecCTMC, a training-free speculative sampler that reduces this sequential bottleneck without a separate draft model. SpecCTMC proposes successive token updates using previously computed model predictions, evaluates their intermediate states in parallel, and applies an exact acceptance-and-correction procedure. We prove that SpecCTMC preserves the joint distribution of event times, generation order, and token values of the underlying continuous-time Markov chain. It therefore introduces neither time-discretization error nor additional factorization error from parallel decoding. Experiments on LLaDA-8B-Instruct and Dream-v0-Instruct-7B across five benchmarks demonstrate 1.73-4.38x fewer sequential model evaluations than direct continuous-time sampling, while preserving the reference sampling distribution. Our code is available at: https://github.com/kasakh/specctmc

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_162936968f34ac64e3e76769