megvii-model/SinglePathOneShot

Issues

• What is the purpose of setting flops bins when training SuperNet?

  #11 opened 4 years ago by jianghaojun
  2

• Evolution search results

  #6 opened 5 years ago by ultmaster
  1

• Ckpt

  #19 opened 2 years ago by hanxingjia1
  0

• reproduce retrain result

  #18 opened 2 years ago by zhanglei1172
  0

• Evaluate pretrained model

  #17 opened 3 years ago by yaolu-zjut
  0

• One called function in EA search has no effect and should be removed

  #16 opened 4 years ago by AlbertiPot
  2

• Can not open the $Link

  #15 opened 4 years ago by KeyKy
  0

• Channel search

  #1 opened 4 years ago by pawopawo
  6

• Missing arguments in `validate`.

  #14 opened 4 years ago
  1

• how can I constraint FLOPs?

  #13 opened 4 years ago by cingyoung
  0

• 代码运行时间

  #12 opened 4 years ago by nbl97
  1

• Supernet accuracy

  #8 opened 5 years ago by xiaoshingshing
  1

• What's the minimum flops-limit in the search?

  #10 opened 5 years ago by betterhalfwzm
  1

• resource in onedrive missing

  #9 opened 5 years ago by hjyiloveyou1314
  1

• SuperNet Training 代码优化

  #7 opened 5 years ago by enduringstack
  1

• Use BGR tensor in data preprocessing

  #5 opened 5 years ago by ultmaster
  2

• Why not shuffle validation data during search

  #4 opened 5 years ago by ultmaster
  2

• Why set `affine=False` in all `nn.BatchNorm2d`

  #3 opened 5 years ago by kaleidoscopical
  1

• link 进不去

  #2 opened 5 years ago by tuofeilunhifi
  0