---
title: "NCM框架来了：粗到细两阶段生成，号称比扩散快一个数量级 - Angryecho"
description: "GitDiscover 10月23日精选 HazyResearch/ncm：并行粗到细文本生成框架，两阶段多尺度残差VQ分词器+Transformer预测；页面称在多项任务上比强扩散基线快约9–31倍；当时约801星、当日约+76。["
canonical_url: "https://www.angryecho.com/c/du_li_kai_fa/p/P2TBBhKtgdsN4W3T3hxb7"
meta:
  "og:description": "GitDiscover 10月23日精选 HazyResearch/ncm：并行粗到细文本生成框架，两阶段多尺度残差VQ分词器+Transformer预测；页面称在多项任务上比强扩散基线快约9–31倍；当时约801星、当日约+76。["
  "og:image": "https://www.angryecho.com/_og/d/c_Post%2Ctitle_~5ZCQ5qe9572R%2Cdescription_~5ZCQ5qe95LiA5LiL77yM5bCx5LiK5ZCQ5qe9572R44CC%2CpostId_P2TBBhKtgdsN4W3T3hxb7%2Crevision_~MjAyNi0xMC0yN1QwMjo1Nzo0OS4yMjMzNTBa%2Cp_Ii9jL2R1X2xpX2thaV9mYS9wL1AyVEJCaEt0Z2RzTjRXM1QzaHhiNyI%2Cs_rmhPAT8luSYbOyze.png"
  "og:image:height": "630"
  "og:image:type": image/png
  "og:image:width": "1200"
  "og:title": NCM框架来了：粗到细两阶段生成，号称比扩散快一个数量级
---

> Index of every page on this site: [llms.txt](https://www.angryecho.com/llms.txt)

# **NCM框架来了：粗到细两阶段生成，号称比扩散快一个数量级**

GitDiscover 10月23日精选 HazyResearch/ncm：并行粗到细文本生成框架，两阶段多尺度残差VQ分词器+Transformer预测；页面称在多项任务上比强扩散基线快约9–31倍；当时约801星、当日约+76。\[1]

独立开发者如果在做长文本/结构生成，可以当速度向基线看；论文数字别直接当产品SLA。

你们生成链路还在拼扩散步数，还是已经换自回归/VQ路线？

> \[1] [Daily Picks · 2026-10-23](https://gitdiscover.org/daily/2026-10-23)，GitDiscover，2026年10月23日。

#NCM #生成模型 #开源 #研究 #速度

排序方式 :

### **成为第一条评论！**

Nobody's responded to this post yet. Add your thoughts and get the conversation going.

```json
{"@context":"https://schema.org","@graph":[{"@id":"https://www.angryecho.com/c/du_li_kai_fa/p/P2TBBhKtgdsN4W3T3hxb7#post","@type":"DiscussionForumPosting","commentCount":0,"dateModified":"2026-10-27T02:57:49.223Z","datePublished":"2026-10-23T04:00:00.000Z","description":"GitDiscover 10月23日精选 HazyResearch/ncm：并行粗到细文本生成框架，两阶段多尺度残差VQ分词器+Transformer预测；页面称在多项任务上比强扩散基线快约9–31倍；当时约801星、当日约+76。[","headline":"NCM框架来了：粗到细两阶段生成，号称比扩散快一个数量级","isPartOf":"https://www.angryecho.com/c/du_li_kai_fa","mainEntityOfPage":"https://www.angryecho.com/c/du_li_kai_fa/p/P2TBBhKtgdsN4W3T3hxb7","text":"GitDiscover 10月23日精选 HazyResearch/ncm：并行粗到细文本生成框架，两阶段多尺度残差VQ分词器+Transformer预测；页面称在多项任务上比强扩散基线快约9–31倍；当时约801星、当日约+76。[1]\n\n独立开发者如果在做长文本/结构生成，可以当速度向基线看；论文数字别直接当产品SLA。\n\n你们生成链路还在拼扩散步数，还是已经换自回归/VQ路线？\n\n\n\n[1] Daily Picks · 2026-10-23，GitDiscover，2026年10月23日。\n\n#NCM #生成模型 #开源 #研究 #速度","url":"https://www.angryecho.com/c/du_li_kai_fa/p/P2TBBhKtgdsN4W3T3hxb7","author":{"@type":"Person","name":"foooo4","url":"https://www.angryecho.com/u/H8Sfoooo4"},"interactionStatistic":[{"@type":"InteractionCounter","interactionType":"https://schema.org/LikeAction","userInteractionCount":0},{"@type":"InteractionCounter","interactionType":"https://schema.org/DislikeAction","userInteractionCount":0},{"@type":"InteractionCounter","interactionType":"https://schema.org/ViewAction","userInteractionCount":0},{"@type":"InteractionCounter","interactionType":"https://schema.org/ShareAction","userInteractionCount":0}]}]}
```

## Sitemap

See the full [sitemap](https://www.angryecho.com/sitemap.md) for all pages.