Skip to content

MoonshotAI/checkpoint-engine

View on GitHub ↗Related repositories →

Checkpoint-engine is a simple middleware to update model weights in LLM inference engines

active 2025-09-102026-06-30 (UTC)

Partial coverage11,206 / 11,874 hourly files (94%) · 2 absent upstream · 665 failed, retryable2025-04-032026-08-10 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
778
Pushes
41
Pull requests
30
Issues
20
Stars
476
Forks
36

Activity over time

Daily event counts in the loaded window

Line chart, 294 days from 2025-09-10 to 2026-06-30. Pushes: 41 total, peak 8 in a day. Pull requests: 30 total, peak 7 in a day. Issues: 20 total, peak 2 in a day. Comments: 114 total, peak 8 in a day. Stars: 476 total, peak 167 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

ContributorContributionsPushesPRsComments
weixiao-huang92291534
specture7244021119
Copilot300018
blahgeek17616
HubertZhang17309
kip-cxj7005
BraveY6003
ZSL984002
dakwang4001
x1314aq3003
SongXiaoXi3003
XiaotaoChen3003
hanhan-networking2001
morluto2010
alogfans2001
wpf199111182002
GeeeekExplorer1001
chestnut-Q1001
ppwwyyxx1100
big-ZWZ1000

Recent activity

Latest issues, pull requests and releases

  • Issue#94morluto2026-06-30 22:50
    Clean up failed TransferEngine instances during P2PStore retries
  • Pull request#91morluto2026-06-30 22:03
  • Issue#85big-ZWZ2026-02-06 08:39
    请问可以提供训推分离场景,p2p方式将分片数据从训练进程传输到推理进程的用例吗?
  • Issue comment#66HubertZhang2026-01-30 05:39
    feat: add StatelessProcessGroup to extend collective library
  • Pull request#81youzhedian2026-01-26 15:08
  • Issue comment#66weixiao-huang2026-01-23 01:30
    feat: add StatelessProcessGroup to extend collective library
  • Issue comment#66kip-cxj2026-01-22 06:46
    feat: add StatelessProcessGroup to extend collective library
  • Releaseblahgeek2026-01-20 11:52
    v0.3.3
  • Issue comment#66kip-cxj2026-01-19 08:08
    feat: add StatelessProcessGroup to extend collective library
  • Issue comment#66x1314aq2026-01-13 08:07
    feat: add StatelessProcessGroup to extend collective library
  • Issue#72ZSL982026-01-06 00:52
    [Question] Pipeline Design for Communication Overlapping
  • Issue comment#72ZSL982026-01-06 00:52
    [Question] Pipeline Design for Communication Overlapping
  • Releaseblahgeek2026-01-05 09:00
    v0.3.1
  • Issue comment#72weixiao-huang2026-01-05 07:49
    [Question] Pipeline Design for Communication Overlapping
  • Issue comment#64blahgeek2026-01-04 07:43
    misc: split `ps.py` file into multiple files
  • Pull request#75blahgeek2026-01-04 07:42
  • Issue comment#71hewr20102025-12-29 13:03
    是否可以接受引入torch.distributed以外的集合通信库?
  • Issue comment#71x1314aq2025-12-26 01:54
    是否可以接受引入torch.distributed以外的集合通信库?
  • Issue comment#71x1314aq2025-12-24 09:24
    是否可以接受引入torch.distributed以外的集合通信库?
  • Issue comment#72specture7242025-12-23 14:07
    [Question] Pipeline Design for Communication Overlapping
  • Issue#72ZSL982025-12-23 12:24
    [Question] Pipeline Design for Communication Overlapping
  • Issue comment#50weixiao-huang2025-12-22 14:31
    传输一个558GB的模型,申请H卡4卡进行部署,注册的时候直接OOM导致容器OOMKilled,4卡机器申请的容器内存是800G,注册文件的时候是有写放大吗?实际内存占用和模型的关系是什么?
  • Issue#50weixiao-huang2025-12-22 14:31
    传输一个558GB的模型,申请H卡4卡进行部署,注册的时候直接OOM导致容器OOMKilled,4卡机器申请的容器内存是800G,注册文件的时候是有写放大吗?实际内存占用和模型的关系是什么?
  • Issue comment#71weixiao-huang2025-12-22 09:32
    是否可以接受引入torch.distributed以外的集合通信库?
  • Issue#71hanhan-networking2025-12-20 08:26
    是否可以接受引入torch.distributed以外的集合通信库?

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 476 stars here means stars gained during the window, not the repo's star count.