Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Jan 1, 2026

• Problem: on Blackwell GPUs, functions such as `cudnnFindConvolutionBackwardDataAlgorithm` may run out of stack space on Windows (leading to a stack overflow error) (#3130)

  Show description for 2a878a6
  reunanen

  authored
  2a878a6

  Copy full SHA for 2a878a6

Commits on Dec 28, 2025

• Fix cuDNN convolution precision on Ampere+ GPUs (#3127)

  Show description for 07c1e73
  joelnn

  authored
  07c1e73

  Copy full SHA for 07c1e73

• Set test threshold more appropriately so it doesn't randomly fail.

  davisking

  committed
  60adc65

  Copy full SHA for 60adc65

• Fix flaky holder table test by increasing max function calls (#3129)

  Show description for 98016e8
  joelnn

  authored
  98016e8

  Copy full SHA for 98016e8

• Fix ACT layer gradient computation on CUDA (#3128)

  Show description for 0f86c88
  joelnn

  authored
  0f86c88

  Copy full SHA for 0f86c88

Commits on Nov 28, 2025

• Accept CMake args as env vars for Python setup (#3123)

  turicas

  authored
  49b7cba

  Copy full SHA for 49b7cba

Commits on Nov 22, 2025

• Fix callable detection to support functools.partial and all callable objects (#3121)

  Show description for 23d73be
  ssam18

  authored
  23d73be

  Copy full SHA for 23d73be

Commits on Oct 11, 2025

• Implementation of ARC-AGI dataset loader (#3115)

  Show description for 20b2172
  

  4 peopleauthored
  20b2172

  Copy full SHA for 20b2172

Commits on Sep 29, 2025

• Improve numerical stability of batch normalization on CPUs (#3113)

  reunanen

  authored
  2ac06f9

  Copy full SHA for 2ac06f9

Commits on Sep 28, 2025

• Add Adaptive Computation Time (ACT) layer (#3111)

  Show description for d890f7d
  

  4 peopleauthored
  d890f7d

  Copy full SHA for d890f7d

• Improve the somewhat flaky `test_loss_multibinary_log` by avoiding samples very close to class boundaries (#3112)

  reunanen

  authored
  35a8e1f

  Copy full SHA for 35a8e1f

Commits on Sep 6, 2025

• Replace FindCUDA with FindCUDAToolkit (fix #2833) (#3090)

  Show description for 5ce4891
  

  3 peopleauthored
  5ce4891

  Copy full SHA for 5ce4891

Commits on Aug 22, 2025

• [unicode] fixed warning in convert_mbstring_to_wstring() (#3110)

  Show description for ffe7881
  

  pfeatherstone

  and

  pfeatherstone

  authored
  ffe7881

  Copy full SHA for ffe7881

Commits on Aug 13, 2025

• FFMpeg 7.1 API (#3105)

  Show description for 0d2153a
  

  pfeatherstone

  and

  me

  authored
  0d2153a

  Copy full SHA for 0d2153a

Commits on Aug 11, 2025

• fix -Wcast-user-defined (#3103)

  Show description for fb0865b
  

  ribbon-otter

  and

  ribbon-otter

  authored
  fb0865b

  Copy full SHA for fb0865b

• update python packaging to work with newer toolchains (#3102)

  davisking

  authored
  4cf8f57

  Copy full SHA for 4cf8f57

Commits on Aug 10, 2025

• Suppress a spurious warning in unit tests and make colors not always output

  Show description for 561759c
  davisking

  committed
  561759c

  Copy full SHA for 561759c

Commits on Aug 9, 2025

• fix up build errors with the regular makefile and non-cmake builds.

  davisking

  committed
  56b806d

  Copy full SHA for 56b806d

• std::aligned_storage_t is deprecated in c++23 (#3100)

  Show description for e93a0ef
  

  ribbon-otter

  and

  ribbon-otter

  authored
  e93a0ef

  Copy full SHA for e93a0ef

Commits on Aug 4, 2025

• Account for change in ref count in Python 3.14+ (#3098)

  Show description for 043bed8
  penguinpee

  authored
  043bed8

  Copy full SHA for 043bed8

Commits on Jul 25, 2025

• windows server 2019 isn't supported on github actions anymore

  davisking

  committed
  fa163cb

  Copy full SHA for fa163cb

• cleanup

  davisking

  committed
  36f4226

  Copy full SHA for 36f4226

• full_object_detection: API Support for dpoint (#2183) (#3095)

  Show description for c1fbfb5
  Infinage

  authored
  c1fbfb5

  Copy full SHA for c1fbfb5

• more doc updates

  davisking

  committed
  57547b9

  Copy full SHA for 57547b9

• updated docs

  davisking

  committed
  52d38c7

  Copy full SHA for 52d38c7

• Add transformer example with RoPE and MoE-like mechanisms (#3078)

  Show description for 7889cf3
  

  3 peopleauthored
  7889cf3

  Copy full SHA for 7889cf3

Commits on Jun 28, 2025

• `throw()` -> `noexcept` (#3092)

  r-barnes

  authored
  131e46e

  Copy full SHA for 131e46e

Commits on Jun 21, 2025

• Problem: reading large images (in my case, it was a 40792x20480 RGBA PNG one) causes a crash, when the product row-number * width-step overflows a 32-bit integer (#3091)

  Show description for 3cb1f35
  reunanen

  authored
  3cb1f35

  Copy full SHA for 3cb1f35

Commits on Jun 17, 2025

• fix the miss of including matlab ut library (#3089)

  gmm76

  authored
  d0f4179

  Copy full SHA for d0f4179

Commits on Jun 13, 2025

• Fix: Correct YOLO loss gradient for x,y coordinates (#3088)

  Show description for 79e2d13
  

  arrufat

  and

  google-labs-jules[bot]

  authored
  79e2d13

  Copy full SHA for 79e2d13

Commits on Jun 9, 2025

• Minor optimization: `emplace_back` neighbors (#3086)

  reunanen

  authored
  81cd1f1

  Copy full SHA for 81cd1f1

Commits on Jun 2, 2025

• remove std::move where clang warns of pessimizing-move (#3080)

  Show description for a078e81
  

  ribbon-otter

  and

  ribbon-otter

  authored
  a078e81

  Copy full SHA for a078e81

Commits on May 30, 2025

• docs: dlib requires c++14 (#3079)

  Show description for e529d2a
  

  ribbon-otter

  and

  ribbon-otter

  authored
  e529d2a

  Copy full SHA for e529d2a

Commits on May 28, 2025

• Record last changeset and set PATCH version to 99

  davisking

  committed
  278ac4c

  Copy full SHA for 278ac4c

• Created release v20.0

  davisking

  committed
  2e0b384

  Copy full SHA for 2e0b384

Pagination

Previous

Next