🚀🚀 Efficient implementations of Native Sparse Attention
active 2025-08-19 → 2026-05-19 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 274 days from 2025-08-19 to 2026-05-19. Pushes: 15 total, peak 4 in a day. Pull requests: 7 total, peak 3 in a day. Issues: 1 total, peak 1 in a day. Comments: 8 total, peak 3 in a day. Stars: 589 total, peak 61 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 |
|---|---|---|---|---|
| Ran-Yan-HK | 25 | 14 | 5 | 4 |
| yiakwy-xpu-ml-framework-team | 4 | 0 | 1 | 3 |
| Youhe-Jiang | 2 | 1 | 1 | 0 |
| PXHnb666 | 1 | 0 | 0 | 0 |
| EricLina | 1 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue#12PXHnb6662025-11-01 07:00关于run_unit_test.sh测试结果咨询
- Pull request#10Youhe-Jiang2025-09-29 10:36
- Issue comment#6Ran-Yan-HK2025-09-02 08:37Speed comparison with FlashAttention2/3 CUDA backbend.
- Pull request#9Ran-Yan-HK2025-09-02 04:28
- Pull request#8Ran-Yan-HK2025-09-02 04:09
- Pull request#8Ran-Yan-HK2025-09-02 04:04
- Issue comment#6yiakwy-xpu-ml-framework-team2025-09-02 02:26Speed comparison with FlashAttention2/3 CUDA backbend.
- Issue comment#7Ran-Yan-HK2025-09-01 14:36add Fa3 baseline
- Pull request#7yiakwy-xpu-ml-framework-team2025-09-01 06:45
- Issue comment#6yiakwy-xpu-ml-framework-team2025-09-01 05:04Speed comparison with FlashAttention2/3 CUDA backbend.
- Issue comment#6yiakwy-xpu-ml-framework-team2025-09-01 02:28Speed comparison with FlashAttention2/3 CUDA backbend.
- Issue comment#6EricLina2025-08-28 06:21Speed comparison with FlashAttention2/3 CUDA backbend.
- Issue comment#6Ran-Yan-HK2025-08-27 18:20Speed comparison with FlashAttention2/3 CUDA backbend.
- Pull request#5Ran-Yan-HK2025-08-26 12:31
- Pull request#4Ran-Yan-HK2025-08-26 09:12
- Issue comment#2Ran-Yan-HK2025-08-22 04:15Question about performance at different seqlen
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 589 stars here means stars gained during the window, not the repo's star count.