[ACL 2024] A novel QAT with Self-Distillation framework to enhance ultra low-bit LLMs.
active 2024-02-16 → 2025-10-31 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 624 days from 2024-02-16 to 2025-10-31. Pushes: 6 total, peak 4 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 27 total, peak 4 in a day. Comments: 29 total, peak 5 in a day. Stars: 119 total, peak 3 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 |
|---|---|---|---|---|
| DD-DuDa | 24 | 6 | 0 | 14 |
| pwd11 | 8 | 0 | 0 | 1 |
| zjq0455 | 4 | 0 | 0 | 3 |
| SJ5817 | 3 | 0 | 0 | 2 |
| ChenMnZ | 3 | 0 | 0 | 1 |
| mustavikhan05 | 3 | 0 | 0 | 2 |
| dengzheng-cloud | 3 | 0 | 0 | 1 |
| Hon-Chen | 2 | 0 | 0 | 1 |
| brisker | 2 | 0 | 0 | 2 |
| Toughq | 2 | 0 | 0 | 1 |
| BrownianNotion | 2 | 0 | 0 | 1 |
| Saoyu99 | 1 | 0 | 0 | 0 |
| JJongyn | 1 | 0 | 0 | 0 |
| HsuWanTing | 1 | 0 | 0 | 0 |
| helloworld77 | 1 | 0 | 0 | 0 |
| ghostchoir | 1 | 0 | 0 | 0 |
| xiaoshideta | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#18DD-DuDa2025-10-31 12:31OOM
- Issue#18xiaoshideta2025-10-30 01:52OOM
- Issue comment#17SJ58172025-06-03 13:06Training issue
- Issue comment#17SJ58172025-06-03 12:03Training issue
- Issue#17SJ58172025-06-02 02:14Training issue
- Issue comment#10BrownianNotion2025-03-05 14:09Guidance/plans to add encoder/decoder model support for e.g T5 model?
- Issue#16BrownianNotion2025-03-01 18:18Incoherent repetitive data generated by teacher after clipping.
- Issue#15pwd112024-12-10 02:57version issue
- Issue#15pwd112024-12-10 02:56version issue
- Issue comment#2brisker2024-12-05 10:444-bit or higher?
- Issue comment#2DD-DuDa2024-12-05 10:404-bit or higher?
- Issue comment#2brisker2024-12-04 11:504-bit or higher?
- Issue#15pwd112024-11-30 12:32version issue
- Issue#14HsuWanTing2024-11-29 08:41Add masks when calculating the CAKLD coefficient
- Issue comment#10mustavikhan052024-11-28 20:48Guidance/plans to add encoder/decoder model support for e.g T5 model?
- Issue comment#13Toughq2024-11-28 14:37About the issue of training time
- Issue#13Toughq2024-11-28 14:36About the issue of training time
- Issue#12pwd112024-11-28 06:25about train
- Issue#12pwd112024-11-26 11:34about train
- Issue comment#11dengzheng-cloud2024-11-25 01:07score gap in gsm8k between llama2-7b 16bit and int2g128 model
- Issue#11dengzheng-cloud2024-11-25 01:07score gap in gsm8k between llama2-7b 16bit and int2g128 model
- Issue comment#9DD-DuDa2024-11-24 07:32about device
- Issue comment#10DD-DuDa2024-11-24 07:30Guidance/plans to add encoder/decoder model support for e.g T5 model?
- Issue comment#11DD-DuDa2024-11-24 07:26score gap in gsm8k between llama2-7b 16bit and int2g128 model
- Issue#11dengzheng-cloud2024-11-22 08:30score gap in gsm8k between llama2-7b 16bit and int2g128 model
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 119 stars here means stars gained during the window, not the repo's star count.