update activations.py

utils/activations.py

@@ -3,8 +3,21 @@ import torch.nn as nn
 import torch.nn.functional as F
 
 
-# Swish ------------------------------------------------------------------------
-class SwishImplementation(torch.autograd.Function):
+# Swish https://arxiv.org/pdf/1905.02244.pdf ---------------------------------------------------------------------------
+class Swish(nn.Module):  #
+    @staticmethod
+    def forward(x):
+        return x * torch.sigmoid(x)
+
+
+class HardSwish(nn.Module):
+    @staticmethod
+    def forward(x):
+        return x * F.hardtanh(x + 3, 0., 6., True) / 6.
+
+
+class MemoryEfficientSwish(nn.Module):
+    class F(torch.autograd.Function):
         @staticmethod
         def forward(ctx, x):
             ctx.save_for_backward(x)
@@ -16,27 +29,19 @@ class SwishImplementation(torch.autograd.Function):
             sx = torch.sigmoid(x)
             return grad_output * (sx * (1 + x * (1 - sx)))
 
-
-class MemoryEfficientSwish(nn.Module):
-    @staticmethod
-    def forward(x):
-        return SwishImplementation.apply(x)
-
-
-class HardSwish(nn.Module):  # https://arxiv.org/pdf/1905.02244.pdf
-    @staticmethod
-    def forward(x):
-        return x * F.hardtanh(x + 3, 0., 6., True) / 6.
+    def forward(self, x):
+        return self.F.apply(x)
 
 
-class Swish(nn.Module):
+# Mish https://github.com/digantamisra98/Mish --------------------------------------------------------------------------
+class Mish(nn.Module):
     @staticmethod
     def forward(x):
-        return x * torch.sigmoid(x)
+        return x * F.softplus(x).tanh()
 
 
-# Mish ------------------------------------------------------------------------
-class MishImplementation(torch.autograd.Function):
+class MemoryEfficientMish(nn.Module):
+    class F(torch.autograd.Function):
         @staticmethod
         def forward(ctx, x):
             ctx.save_for_backward(x)
@@ -49,23 +54,14 @@ class MishImplementation(torch.autograd.Function):
             fx = F.softplus(x).tanh()
             return grad_output * (fx + x * sx * (1 - fx * fx))
 
-
-class MemoryEfficientMish(nn.Module):
-    @staticmethod
-    def forward(x):
-        return MishImplementation.apply(x)
-
-
-class Mish(nn.Module):  # https://github.com/digantamisra98/Mish
-    @staticmethod
-    def forward(x):
-        return x * F.softplus(x).tanh()
+    def forward(self, x):
+        return self.F.apply(x)
 
 
-# FReLU https://arxiv.org/abs/2007.11824 --------------------------------------
+# FReLU https://arxiv.org/abs/2007.11824 -------------------------------------------------------------------------------
 class FReLU(nn.Module):
     def __init__(self, c1, k=3):  # ch_in, kernel
-        super(FReLU, self).__init__()
+        super().__init__()
         self.conv = nn.Conv2d(c1, c1, k, 1, 1, groups=c1)
         self.bn = nn.BatchNorm2d(c1)