jhjacobsen/pytorch-i-revnet

Issues

• Questions about using the i-resnet for other applications.

  #23 opened 3 years ago by xyIsHere
  1

• injective or bijective?

  #17 opened 6 years ago by yuffon
  14

• Some confusion for case 'stride = 2'

  #25 opened 2 years ago by shuizidesu
  0

• question about --nChannels

  #24 opened 3 years ago by eefnn
  0

• iRevNet.py's test code doesn't work

  #16 opened 4 years ago by SelimChraibi
  4

• Consider more efficient implementation of class psi

  #18 opened 5 years ago by spezold
  4

• Invertible up sampling

  #21 opened 5 years ago by VitorGuizilini
  1

• About model size of i-RevNet (b)

  #19 opened 5 years ago by djiajunustc
  2

• multi-GPU

  #20 opened 5 years ago by lucidrains
  0

• download pretrained model

  #11 opened 6 years ago by ssli23
  1

• Pretrained models cannot be loaded

  #15 opened 5 years ago by effendijohanes
  3

• Cifar exception in tensor size at beginning of training

  #13 opened 6 years ago by david-klindt
  1

• for "--nBlocks 18 18 18 --nStrides 1 2 2 --nChannels 16 64 256" architecture,Is there pretrained model ?

  #12 opened 6 years ago by ssli23
  2

• Have you implemented backprop without storing forward activations?

  #10 opened 6 years ago by makedede
  1

• is it fully invertible? last layer is pooling+ linear

  #9 opened 6 years ago by tsauri
  2

• Hi, I don't understand permute function

  #8 opened 6 years ago by MrLinNing
  1

• Inverse function has zero block size for final psi function initialization

  #7 opened 6 years ago by ThorJonsson
  1

• Potential bug in model_utils.py

  #6 opened 6 years ago by danfeiX
  1

• The size of tensor are not identifed

  #5 opened 7 years ago by chouqin3
  1

• Question: memory saving

  #4 opened 7 years ago by tartavull
  1

• I really don't understand what dose the psi class do, any explain would be helpful

  #3 opened 7 years ago by ChiHanfeng
  1

• Undefined name 'revnet' in CIFAR_main.py

  #2 opened 7 years ago by cclauss
  0