Skip to content

DuoDecoding: Hardware-aware Heterogeneous Speculative Decoding with Dynamic Multi-Sequence Drafting

active 2025-03-042025-11-17 (UTC)

Partial coverage21,528 / 26,307 hourly files (82%) · 2 absent upstream · 4,778 failed, retryable2023-08-152026-08-15 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
30
Pushes
0
Pull requests
1
Issues
7
Stars
12
Forks
2

Activity over time

Daily event counts in the loaded window

Line chart, 259 days from 2025-03-04 to 2025-11-17. Pushes: 0 total, peak 0 in a day. Pull requests: 1 total, peak 1 in a day. Issues: 7 total, peak 2 in a day. Comments: 7 total, peak 3 in a day. Stars: 12 total, peak 4 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
KaiLv696005
Nanji-Huaji4010
theneildave2002
fousdfrf1000
shiyuetianqiang1000
baihuajun241000

Recent activity

Latest issues, pull requests and releases

  • Pull request#8Nanji-Huaji2025-11-17 05:25
  • Issue#7Nanji-Huaji2025-07-05 08:19
    The `para_sd` method seems to have an index out of bounds issue.
  • Issue#6Nanji-Huaji2025-07-05 04:31
    Can Duodecoding output the acceptance rate after the evaluation ends?
  • Issue#5Nanji-Huaji2025-06-26 19:04
    There is no problem when evaluating the Baseline, but when evaluating DuoDecoding, it shows that the model path does not exist.
  • Issue comment#3KaiLv692025-06-09 04:01
    Make it compatible with Llama-3.1-8B-Instruct
  • Issue comment#4KaiLv692025-06-09 03:49
    why normal spec decoding will increase ttft?
  • Issue#4shiyuetianqiang2025-06-06 09:41
    why normal spec decoding will increase ttft?
  • Issue#3baihuajun242025-05-10 08:35
    Make it compatible with Llama-3.1-8B-Instruct
  • Issue comment#2KaiLv692025-03-29 10:48
    Facing issue with the arg command of llama cpp installation
  • Issue comment#2theneildave2025-03-27 11:04
    Facing issue with the arg command of llama cpp installation
  • Issue comment#2KaiLv692025-03-27 10:27
    Facing issue with the arg command of llama cpp installation
  • Issue comment#2theneildave2025-03-27 08:55
    Facing issue with the arg command of llama cpp installation
  • Issue#1KaiLv692025-03-26 08:28
    the detailed configuration of the CPU
  • Issue comment#2KaiLv692025-03-26 08:07
    Facing issue with the arg command of llama cpp installation
  • Issue#1fousdfrf2025-03-08 14:50
    the detailed configuration of the CPU

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 12 stars here means stars gained during the window, not the repo's star count.