MCPcopy
hub / github.com/openai/guided-diffusion / Upsample

Class Upsample

guided_diffusion/unet.py:81–110  ·  view source on GitHub ↗

An upsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determining if a convolution is applied. :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then upsampling occurs in the inne

Source from the content-addressed store, hash-verified

79
80
81class Upsample(nn.Module):
82 """
83 An upsampling layer with an optional convolution.
84
85 :param channels: channels in the inputs and outputs.
86 :param use_conv: a bool determining if a convolution is applied.
87 :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
88 upsampling occurs in the inner-two dimensions.
89 """
90
91 def __init__(self, channels, use_conv, dims=2, out_channels=None):
92 super().__init__()
93 self.channels = channels
94 self.out_channels = out_channels or channels
95 self.use_conv = use_conv
96 self.dims = dims
97 if use_conv:
98 self.conv = conv_nd(dims, self.channels, self.out_channels, 3, padding=1)
99
100 def forward(self, x):
101 assert x.shape[1] == self.channels
102 if self.dims == 3:
103 x = F.interpolate(
104 x, (x.shape[2], x.shape[3] * 2, x.shape[4] * 2), mode="nearest"
105 )
106 else:
107 x = F.interpolate(x, scale_factor=2, mode="nearest")
108 if self.use_conv:
109 x = self.conv(x)
110 return x
111
112
113class Downsample(nn.Module):

Callers 2

__init__Method · 0.85
__init__Method · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected