Muon is Scalable for LLM Training
active 2025-03-11 → 2026-05-21 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 437 days from 2025-03-11 to 2026-05-21. Pushes: 2 total, peak 1 in a day. Pull requests: 2 total, peak 1 in a day. Issues: 21 total, peak 3 in a day. Comments: 27 total, peak 5 in a day. Stars: 237 total, peak 16 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
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| toothacher17 | 25 | 2 | 1 | 14 |
| jie040109 | 4 | 0 | 0 | 3 |
| bugm | 3 | 0 | 0 | 3 |
| weiyikang | 3 | 0 | 0 | 0 |
| Junyu-Z | 2 | 0 | 0 | 1 |
| Wongboo | 2 | 0 | 0 | 1 |
| airlsyn | 2 | 0 | 0 | 2 |
| SeunghyunSEO | 2 | 0 | 0 | 2 |
| casinca | 1 | 0 | 1 | 0 |
| liangtea | 1 | 0 | 0 | 0 |
| lk137095576 | 1 | 0 | 0 | 0 |
| jiuzhuanzhuan | 1 | 0 | 0 | 0 |
| Sun2018421 | 1 | 0 | 0 | 0 |
| yuanwujun | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue#44yuanwujun2026-05-15 05:29Muon is Scalable for LLM Training 论文分布式问题
- Issue#41lk1370955762025-11-18 03:01what is the dtype of muon？
- Issue comment#36toothacher172025-08-18 03:58Muon训练效果问题
- Issue#36Sun20184212025-08-08 03:07Muon训练效果问题
- Issue comment#33toothacher172025-08-03 06:33Training time on Qwen2: Muon is slower than AdamW
- Pull request#35toothacher172025-08-03 06:31
- Pull request#35casinca2025-07-27 19:13
- Issue#34rishabh-ranjan2025-07-21 04:18lr scheduling with muon
- Issue comment#34MarktHart2025-07-19 11:59lr scheduling with muon
- Issue#34rishabh-ranjan2025-07-18 17:28lr scheduling with muon
- Issue#33weiyikang2025-07-11 08:13Training time on Qwen2: Muon is slower than AdamW
- Issue#33weiyikang2025-07-10 14:34Training time on Qwen2: Muon is slower than AdamW
- Issue#32weiyikang2025-07-09 05:45推理代码报错
- Issue comment#31Junyu-Z2025-06-15 15:03a question about scale update
- Issue comment#31toothacher172025-06-14 03:42a question about scale update
- Issue#31toothacher172025-06-14 03:42a question about scale update
- Issue#31Junyu-Z2025-06-04 05:42a question about scale update
- Issue#30jiuzhuanzhuan2025-05-22 05:54有多轮对话的测试脚本么？
- Issue comment#27airlsyn2025-05-04 04:13[Bug] zeropower_via_newtonschulz5 will get -inf value when using torch FSDP
- Issue comment#28toothacher172025-04-27 23:41用Kimi开发项目很香。
- Issue#28toothacher172025-04-27 23:41用Kimi开发项目很香。
- Issue#27toothacher172025-04-27 23:40[Bug] zeropower_via_newtonschulz5 will get -inf value when using torch FSDP
- Issue#28liangtea2025-04-27 09:39用Kimi开发项目很香。
- Issue#27toothacher172025-04-18 05:38[Bug] zeropower_via_newtonschulz5 will get -inf value when using torch FSDP
- Issue comment#27toothacher172025-04-18 02:50[Bug] zeropower_via_newtonschulz5 will get -inf value when using torch FSDP
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 237 stars here means stars gained during the window, not the repo's star count.