timgaripov/swa

Issues

• Figure 3 plot

  #24 opened 2 years ago by byan19
  1

• SWA with Torchbearer

  #23 opened 2 years ago by AnabetsyR
  3

• SWA with distributed training

  #22 opened 4 years ago by milliema
  1

• SWA compared to Exponential Moving Average (EMA)

  #16 opened 5 years ago by glenn-jocher
  4

• Can we use Adam or other optimizer instead of SGD to train the network?

  #6 opened 6 years ago by ShadowLau
  3

• Finetuning model with SWA

  #20 opened 4 years ago by mrgloom
  1

• How many weights instances stored in memory simultaneously?

  #21 opened 4 years ago by mrgloom
  1

• release code for training imagenet

  #19 opened 4 years ago by Haijunlv
  1

• Isn't the weight learnt from each minibatchs are supposed to be averaged?

  #18 opened 4 years ago by feather0011
  5

• Codes for figure 1,3,4, and 5

  #17 opened 5 years ago by yswang1717
  1

• [ Question ] Isn't the mid-training evaluation of the SWA performance corrupting the batchnorm running averages ?

  #15 opened 5 years ago by alexandrebone
  4

• Do you have a keras version?

  #1 opened 7 years ago by xuzhang5788
  12

• Unable to re-produce SGD numbers

  #14 opened 5 years ago by amitchandak
  7

• How to reproduce result of Fig.1 in the paper?

  #5 opened 6 years ago by Haijunlv
  5

• How about sgdr?

  #9 opened 5 years ago by twmht
  1

• About calculating parameters of BN

  #7 opened 5 years ago by AITTSMD
  1

• performance drop due to batch norm params recalculation

  #8 opened 5 years ago
  1

• About bn_update

  #10 opened 5 years ago by zllrunning
  5

• CUDA out of memory

  #12 opened 5 years ago by 5059
  1

• About conv_init(m)

  #11 opened 6 years ago by DCC-lzhy
  0

• BrokenPipeError: [Errno 32] Broken pipe

  #2 opened 6 years ago by videetnimsarkar21
  2

• About preresnet results

  #4 opened 6 years ago by BIGBALLON
  2

• Incorrect model parameter update?

  #3 opened 6 years ago by hahnicity
  2