d-li14/octconv.pytorch

Issues

• Could you please provide the training recipe file?

  #23 opened 3 years ago by wenyuqing
  0

• Could you add a example of resnet18?

  #19 opened 4 years ago by mozpp
  1

• Why only return high frequency features (x_h)

  #22 opened 4 years ago by zhaoyanjoin
  1

• BatchNorm before activation vs BatchNorm after activation

  #21 opened 4 years ago by AlexanderHustinx
  0

• Difference between your implementation and paper.

  #20 opened 4 years ago by shoutOutYangJie
  3

• about the pre-trained model of the OctResNet-101

  #18 opened 5 years ago by Marcus208
  3

• train coed

  #17 opened 5 years ago by Sangkwun
  1

• redundant x_h in oct_resnet ?

  #14 opened 5 years ago by Khoa-NT
  1

• basic_block in oct_resnet

  #13 opened 5 years ago by Khoa-NT
  1

• There is no main script to run the network ?

  #15 opened 5 years ago by polyrhythML
  0

• There are implementation with UNet?

  #16 opened 5 years ago by David-19940718
  1

• about the pre-trained model

  #11 opened 5 years ago by Marcus208
  1

• Initialisation of layers and weights not available

  #10 opened 5 years ago by aakash-saboo
  1

• Computation time around 3 times longer with OctConv

  #9 opened 5 years ago by niam06
  3

• octave conv cannot support input at any size

  #7 opened 5 years ago by Stinky-Tofu
  3

• Why is original resnet50 faster than octave-resnet50?

  #4 opened 5 years ago by BlueAnthony
  4

• hi,Why the accuracy of original resnet26 is higher than octave-resnet26 used in cifar10 and cifar100?

  #5 opened 5 years ago by gogo03
  2

• Hi, why not adopting pre-actice version?

  #8 opened 5 years ago by vMagina
  1

• Good work , will you provide more pretrained models?

  #6 opened 5 years ago by nemonameless
  5

• Training code

  #3 opened 6 years ago by miguelvr
  2

• about plot

  #2 opened 6 years ago by Git-Fastai
  3

• About loading pre-training errors

  #1 opened 6 years ago by linyi785
  7