Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on May 18, 2018

• Deprecation notice

  hollance

  committed
  04109c8

  Copy full SHA for 04109c8

Commits on Apr 4, 2018

• Fix compile errors in Xcode 9.3

  hollance

  committed
  feb1ef5

  Copy full SHA for feb1ef5

Commits on Dec 5, 2017

• Fix crash with biases being deallocated too soon (with optimizations enabled)

  hollance

  committed
  14a6d72

  Copy full SHA for 14a6d72

Commits on Sep 21, 2017

• Fix memory leak

  hollance

  committed
  4977cb8

  Copy full SHA for 4977cb8

Commits on Sep 3, 2017

• DepthwiseConvolutionKernel uses same weights layout as MPS depthwise conv

  hollance

  committed
  ed84326

  Copy full SHA for ed84326

• Always try to use maximum framerate

  hollance

  committed
  5195d6f

  Copy full SHA for 5195d6f

• Now runs on iOS 10 again

  hollance

  committed
  8bd655a

  Copy full SHA for 8bd655a

Commits on Aug 8, 2017

• Misc fixes + beta 5

  hollance

  committed
  5fad05f

  Copy full SHA for 5fad05f

Commits on Jul 25, 2017

• Changes for beta 4

  hollance

  committed
  79b6cbd

  Copy full SHA for 79b6cbd

Commits on Jul 18, 2017

• Improve FPS counters

  Show description for 1db190d
  hollance

  committed
  1db190d

  Copy full SHA for 1db190d

Commits on Jul 12, 2017

• Use MPS depthwise convolution in the MobileNets example app

  hollance

  committed
  912b46e

  Copy full SHA for 912b46e

• Add test for MPS depthwise convolution

  hollance

  committed
  042030d

  Copy full SHA for 042030d

Commits on Jun 30, 2017

• Can depthwise convolution be implemented with groups already?

  Show description for 948a13f
  hollance

  committed
  948a13f

  Copy full SHA for 948a13f

Commits on Jun 22, 2017

• Fixes for beta 2

  hollance

  committed
  8f669d7

  Copy full SHA for 8f669d7

Commits on Jun 14, 2017

• Fixes

  hollance

  committed
  e541708

  Copy full SHA for e541708

• Turn off optimization on Forge, as it crashes the app

  hollance

  committed
  bc70074

  Copy full SHA for bc70074

• Move debug image out of the video preview layer

  hollance

  committed
  5a85e4f

  Copy full SHA for 5a85e4f

• Descriptors for temporary images must have private storage mode

  hollance

  committed
  8a27e08

  Copy full SHA for 8a27e08

• Turn off @objc inference

  hollance

  committed
  029e78a

  Copy full SHA for 029e78a

Commits on Jun 13, 2017

• Top and bottom layout guides are deprecated

  hollance

  committed
  fc5b30c

  Copy full SHA for fc5b30c

• Swift 4 conversion, everything requires iOS 11 now

  hollance

  committed
  f7667a3

  Copy full SHA for f7667a3

Commits on Jun 6, 2017

• :fire: :fire: :fire:

  hollance

  committed
  46710db

  Copy full SHA for 46710db

• Add MobileNets weight conversion script

  hollance

  committed
  5a7d1b8

  Copy full SHA for 5a7d1b8

• Add working MobileNets demo

  hollance

  committed
  2f9e910

  Copy full SHA for 2f9e910

Commits on Jun 3, 2017

• Padding is now same/valid instead of true/false

  Show description for 5e04351
  hollance

  committed
  5e04351

  Copy full SHA for 5e04351

• Output shape for layers with padding and stride > 1 was wrong

  Show description for 9cc79c0
  hollance

  committed
  9cc79c0

  Copy full SHA for 9cc79c0

• Add applyPadding() method for pooling layers

  hollance

  committed
  8203d40

  Copy full SHA for 8203d40

Commits on Jun 2, 2017

• API cleanup

  hollance

  committed
  1b23618

  Copy full SHA for 1b23618

• Add argsort to Array

  hollance

  committed
  6a3002a

  Copy full SHA for 6a3002a

• Code cleanup

  hollance

  committed
  5327b57

  Copy full SHA for 5327b57

• API cleanup

  hollance

  committed
  473cad2

  Copy full SHA for 473cad2

• Merge branch 'master' of https://github.com/hollance/Forge

  hollance

  committed
  9f18923

  Copy full SHA for 9f18923

• Bias terms are optional

  hollance

  committed
  77ef253

  Copy full SHA for 77ef253

Commits on May 21, 2017

• Merge pull request #6 from marseel/transpose_fix

  Show description for 48020be
  hollance

  authored
  48020be

  Copy full SHA for 48020be

Commits on May 19, 2017

• Fixed allocating memory for buffer in TransposeChannelsKernel

  marseel

  committed
  446ca57

  Copy full SHA for 446ca57