Skip to content

[MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving

active 2024-01-022026-02-15 (UTC)

Partial coverage20,542 / 26,298 hourly files (78%) · 2 absent upstream · 5,753 failed, retryable2023-08-152026-08-14 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
306
Pushes
5
Pull requests
7
Issues
24
Stars
206
Forks
23

Activity over time

Daily event counts in the loaded window

Line chart, 776 days from 2024-01-02 to 2026-02-15. Pushes: 5 total, peak 1 in a day. Pull requests: 7 total, peak 2 in a day. Issues: 24 total, peak 2 in a day. Comments: 37 total, peak 9 in a day. Stars: 206 total, peak 32 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
happierpig364321
cylinbao8123
mxjyst6003
priscilla-pan5003
cat5383001
jimmy-adams3001
irasin2001
cokeshao2001
shadowpa03272020
wlll1234562000
lisuying2142001
rhway6661000
LapshinA1001
SherrySwift1001
FlyFoxPlayer1000
MrDoghead1000

Recent activity

Latest issues, pull requests and releases

  • Issue comment#26LapshinA2024-12-27 12:32
    Doing text generation with Atom quantized model
  • Issue#26rhway6662024-12-22 21:37
    Doing text generation with Atom quantized model
  • Issue comment#25happierpig2024-10-18 18:36
    kernel optimized for A100
  • Issue comment#25lisuying2142024-10-18 11:16
    kernel optimized for A100
  • Issue#25lisuying2142024-10-12 06:46
    kernel optimized for A100
  • Issue comment#24happierpig2024-09-27 18:25
    e2e demonstration for bigger models
  • Issue comment#23happierpig2024-09-07 05:28
    Question about KV Cache quantization
  • Issue comment#23SherrySwift2024-09-07 03:34
    Question about KV Cache quantization
  • Issue comment#23happierpig2024-09-06 16:28
    Question about KV Cache quantization
  • Issue comment#22cokeshao2024-08-02 11:11
    Quention about end-to-end efficiency evaluation of Atom
  • Issue#22cokeshao2024-08-02 11:11
    Quention about end-to-end efficiency evaluation of Atom
  • Issue#21wlll1234562024-07-30 06:40
    Is it possible to add support for other models?
  • Issue comment#21happierpig2024-07-30 06:37
    Is it possible to add support for other models?
  • Issue#21wlll1234562024-07-30 06:28
    Is it possible to add support for other models?
  • Issue#20cat5382024-07-18 06:07
    Question about the synchronazation in low-precision kernel
  • Issue comment#20cat5382024-07-18 06:06
    Question about the synchronazation in low-precision kernel
  • Issue comment#20happierpig2024-07-18 06:02
    Question about the synchronazation in low-precision kernel
  • Issue#20cat5382024-07-18 05:29
    Question about the synchronazation kernel
  • Issue comment#18jimmy-adams2024-07-17 05:59
    LLM model load hanging problem
  • Issue#18jimmy-adams2024-07-17 05:59
    LLM model load hanging problem
  • Issue comment#18happierpig2024-07-02 05:59
    LLM model load hanging problem
  • Issue comment#19happierpig2024-07-02 05:57
    TypeError: QLlamaDecoderLayer.forward() got an unexpected keyword argument 'cache_position'
  • Issue#18jimmy-adams2024-06-27 02:59
    LLM model load hanging problem
  • Issue comment#17happierpig2024-05-28 01:33
    Question regarding the efficiency evaluation
  • Issue#17happierpig2024-05-15 01:13
    Question regarding the efficiency evaluation

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 206 stars here means stars gained during the window, not the repo's star count.