Last active
May 3, 2018 02:00
-
-
Save ajbrock/04e2586ef0bc8965e4f57d0de8e8ee8a to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# A simple Layer Norm implementation | |
# Andy Brock, March 2017 | |
# | |
# Andy's Notes: | |
# -This is sort of hacky but it seems to work. | |
# -You may also want an affine transform in there. | |
# -Note the .cuda() call on the dummys! | |
class LayerNorm(nn.Module): | |
def __init__(self): | |
super(LayerNorm, self).__init__() | |
def forward(self, input,dummy=None): | |
if dummy is None: | |
dummy = torch.zeros(input.size(0)).cuda() | |
dummy_var = torch.ones(input.size(0)).cuda() # These may need to be Variables | |
x = input.transpose(0,1).contiguous() | |
x = F.batch_norm(x,running_mean=dummy,running_var=dummy,weight=None,bias=None,training=True, momentum=0.1,eps=1e-5) | |
return x.transpose(0,1) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment