What does a discrete diffusion model learn: a denoiser, a score ratio, or a bridge plug-in predictor? At the level of jump rates, these are one object in different coordinates, and reading a neural network in the wrong coordinate changes the process being trained and sampled. Starting with a rigorous derivation of the continuous-time Markov chain (CTMC) ELBO for any noising process, boundary terms
---
**📖 中文解读**
以上内容由AI翻译自英文原文,可能存在不准确之处。建议阅读[原文](https://arxiv.org/abs/2607.05381v1)获取最准确的信息。
---
🔗 **原文链接**: [What Does a Discrete Diffusion Model Learn?](https://arxiv.org/abs/2607.05381v1)
🏷️ **转载来源**: ArXiv cs.AI
> 本文由小九AI技术站翻译整理,内容版权归原作者所有。
👤 作者: Rodrigo Casado Noguerales, Bernhard Schölkopf, Thomas Hofmann, Aran Raoufi
---
🐾 **小九锐评**
这篇论文来自arXiv预印本,虽然还没有经过同行评审,但选题方向值得关注。
建议先读中文摘要判断是否相关,再看全文细节。
你对这个话题有什么看法?欢迎在评论区讨论 💬
> _转载自 ArXiv cs.AI,内容版权归原作者所有_
---
⏱️ 2026-07-07 14:02
news
What Does a Discrete Diffusion Model Learn?
加载回复中...