Skip to content

image upscale HyperTile optimizes the self-attention layer within the Stable-Diffusion U-Net and VAE models, resulting in a reduction in computation time ranging from 1 to 4 times, depending on the initial resolution and tile size. The implementation is exceptionally straightforward.

active 2023-10-012026-02-12 (UTC)

Partial coverage19,602 / 25,244 hourly files (78%) · 2 absent upstream · 5,639 failed, retryable2023-09-262026-08-12 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
107
Pushes
8
Pull requests
1
Issues
8
Stars
74
Forks
8

Activity over time

Daily event counts in the loaded window

Line chart, 866 days from 2023-10-01 to 2026-02-12. Pushes: 8 total, peak 4 in a day. Pull requests: 1 total, peak 1 in a day. Issues: 8 total, peak 1 in a day. Comments: 6 total, peak 2 in a day. Stars: 74 total, peak 8 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
tfernd11802
aria1th2011
JohnnyStreet2001
lhucklen1000
brknsoul1000
gohan20911001
huchenlei1000
sashaok1231000
Gothmoth881001
Astropulse1000
maniac-0s1000

Recent activity

Latest issues, pull requests and releases

  • Issue#11maniac-0s2024-04-06 17:30
    HOW do I actually use this?
  • Issue comment#8gohan20912024-03-25 03:26
    Feature Request: Auto select hypertile size
  • Issue comment#9Gothmoth882024-02-18 20:13
    Explanation not really a problem
  • Issue#10huchenlei2024-01-15 05:15
    [Bug] Potential interference with ControlNet reference
  • Issue#9lhucklen2023-12-31 19:18
    Explanation not really a problem
  • Issue comment#7JohnnyStreet2023-12-23 01:44
    wait for model to move back to device when using .enable_model_cpu_offload()
  • Issue#8brknsoul2023-12-23 01:19
    Feature Request: Auto select hypertile size
  • Issue#7JohnnyStreet2023-12-22 16:36
    wait for model to move back to device when using .enable_model_cpu_offload()
  • Issue comment#6aria1th2023-11-12 15:49
    Update ReadME / avoid chances for same attribute names in hijack / add h*w assertion
  • Pull request#6aria1th2023-11-11 14:27
  • Issue comment#3tfernd2023-10-11 19:22
    HyperTile x ToMe
  • Issue#3tfernd2023-10-11 19:22
    HyperTile x ToMe
  • Issue comment#3tfernd2023-10-11 16:27
    HyperTile x ToMe
  • Issue#3Astropulse2023-10-08 15:02
    HyperTile x ToMe
  • Issue#1sashaok1232023-10-02 05:03
    Support for newer diffusion models (2.1 and XL)

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 74 stars here means stars gained during the window, not the repo's star count.