---
title: "The Arxiv for the new Decoupled DiLoCo paper is now up: https://t.co/z6MwS6Z4Ms"
source_name: "Jeff Dean(@JeffDean)"
original_url: "https://x.com/JeffDean/status/2047665074471387456"
canonical_url: "https://www.traeai.com/articles/2fc49968-8748-48bc-8610-c8134f04f00c"
content_type: "tweet"
language: "英文"
score: 7.5
tags: ["Decoupled DiLoCo","大规模语言模型","SPMD","分布式预训练","Jeff Dean"]
published_at: "2026-04-24T13:12:52+00:00"
created_at: "2026-05-01T01:58:08.43513+00:00"
---

# The Arxiv for the new Decoupled DiLoCo paper is now up: https://t.co/z6MwS6Z4Ms

Canonical URL: https://www.traeai.com/articles/2fc49968-8748-48bc-8610-c8134f04f00c
Original source: https://x.com/JeffDean/status/2047665074471387456

## Summary

Jeff Dean宣布关于Decoupled DiLoCo新论文的arXiv链接已发布，该论文探讨了用于弹性分布式预训练的解耦DiLoCo方法，旨在改进大规模语言模型训练中的SPMD范式。

## Key Takeaways

- Decoupled DiLoCo论文关注于提高大型语言模型预训练的效率与弹性。
- 通过解耦加速器间的紧密耦合，论文提出新方法以优化SPMD并行策略。
- Jeff Dean在个人社交媒体上分享，表明此研究可能对AI和机器学习社区有重要影响。

## Content

Title: Jeff Dean on X: "The Arxiv for the new Decoupled DiLoCo paper is now up: https://t.co/z6MwS6Z4Ms" / X

URL Source: http://x.com/JeffDean/status/2047665074471387456

Markdown Content:
# Jeff Dean on X: "The Arxiv for the new Decoupled DiLoCo paper is now up: https://t.co/z6MwS6Z4Ms" / X

Don’t miss what’s happening

People on X are the first to know.

[Log in](http://x.com/login)

[Sign up](http://x.com/i/flow/signup)

# [](http://x.com/)

## Post

See new posts

# Conversation

[![Image 1](https://pbs.twimg.com/profile_images/935325968280907776/AcBo6zJc_normal.jpg)](http://x.com/JeffDean)

[Jeff Dean](http://x.com/JeffDean)

[@JeffDean](http://x.com/JeffDean)

The Arxiv for the new Decoupled DiLoCo paper is now up:

[![Image 2: arXiv logo](https://pbs.twimg.com/card_img/2046703494661287936/aJp0Q4J7?format=jpg&name=240x240)](https://t.co/z6MwS6Z4Ms)

[arxiv.org Decoupled DiLoCo for Resilient Distributed Pre-training Modern large-scale language model pre-training relies heavily on the single program multiple data (SPMD) paradigm, which requires tight coupling across accelerators. Due to this coupling,...](https://t.co/z6MwS6Z4Ms)

[1:12 PM · Apr 24, 2026](http://x.com/JeffDean/status/2047665074471387456)

·

[9,860 Views](http://x.com/JeffDean/status/2047665074471387456/analytics)

3

4

72

43

## New to X?

Sign up now to get your own personalized timeline!

Sign up with Apple

[Create account](http://x.com/i/flow/signup)

By signing up, you agree to the [Terms of Service](https://x.com/tos) and [Privacy Policy](https://x.com/privacy), including [Cookie Use.](https://help.x.com/rules-and-policies/twitter-cookies)

## Relevant people

*     [![Image 3](https://pbs.twimg.com/profile_images/935325968280907776/AcBo6zJc_normal.jpg)](http://x.com/JeffDean)       [Jeff Dean](http://x.com/JeffDean) [@JeffDean](http://x.com/JeffDean)    Follow   Click to Follow JeffDean  Chief Scientist, Google DeepMind & Google Research. Gemini Lead. Opinions stated here are my own, not those of Google. TensorFlow, MapReduce, Bigtable, ...   

# Trending now

## What’s happening

Sports · Trending

Risacher

Only on X · Trending

#911onABC

Crime drama · Trending

#LawAndOrderSVU

Trending in United States

Fennell

[Show more](http://x.com/explore/tabs/for-you)

[Terms of Service](https://x.com/tos)

|

[Privacy Policy](https://x.com/privacy)

|

[Cookie Policy](https://support.x.com/articles/20170514)

|

[Accessibility](https://help.x.com/resources/accessibility)

|

[Ads info](https://business.x.com/en/help/troubleshooting/how-twitter-ads-work.html?ref=web-twc-ao-gbl-adsinfo&utm_source=twc&utm_medium=web&utm_campaign=ao&utm_content=adsinfo)

|

More

© 2026 X Corp.