2017-fall-DL-training-program/ConvNetwork

Issues

• Learning rate warm up

  #13 opened 7 years ago by jessejchuang
  1

• HW 1 . Demo volunteer

  #8 opened 7 years ago by mtkmark
  5

• Lab Demo

  #14 opened 7 years ago by ChengLung-Jen
  1

• What is "nn.BatchNorm2d" for?

  #3 opened 7 years ago by ghbcpeng
  3

• The definition of "Test Error"

  #12 opened 7 years ago by bcpenggh
  1

• Why the mini-batch sizes of training and testing are different?

  #11 opened 7 years ago by bcpenggh
  1

• Weight Init Gain

  #4 opened 7 years ago by pandasfang
  6

• Do we need pooling between convolution and linear layers?

  #10 opened 7 years ago by bcpenggh
  3

• Does we need to build validation set ?

  #6 opened 7 years ago by doom8199
  2

• Could we use Bottleneck to achieve 50 layers network even 110 layers network?

  #9 opened 7 years ago by fansia
  1

• execution order of transform.Compose

  #1 opened 7 years ago by mryzlee
  1

• Is start_epoch 0 or 1?

  #2 opened 7 years ago by mryzlee
  1

• What is Plain network in Lab01?

  #7 opened 7 years ago by fansia
  2

• Architecture & Scoring Question

  #5 opened 7 years ago by c31601
  1