lucidrains/ema-pytorch

Issues

• Using EMA in a DDP setup

  #9 opened 2 years ago by JamesMcCullochDickens
  2

• How BatchNorm is handled?

  #29 opened 5 months ago by liqimai
  0

• FutureWarning with `torch.load` in `PostHocEMA` Example

  #26 opened 5 months ago by HiDolen
  3

• `PostHocEMA.copy_params_from_ema_to_model()` is not correctly named

  #28 opened 5 months ago by liqimai
  1

• about the training speed

  #5 opened 2 years ago by winnechan
  1

• torch.no_grad on net?

  #25 opened 7 months ago by TKassis
  2

• Leaf/computed variables requires_grad error

  #20 opened 10 months ago by crlandsc
  0

• Post-hoc ema?

  #14 opened a year ago by elyxlz
  11

• Function parameters confusing in inplace_copy and inplace_lerp

  #16 opened a year ago by sheshanshag
  1

• question about usage

  #11 opened a year ago by fire717
  4

• Support for non-deepcopyable models

  #6 opened a year ago by stefan-baumann
  5

• Implementation error

  #8 opened 2 years ago by sbl1996
  1

• Saving EMA also saves online model?

  #7 opened 2 years ago by Netruk44
  4

• Question about EMA

  #1 opened 3 years ago by slerman12
  1