⚗️ | Experimenting, again.

discriminator.py

@@ -1,30 +1,31 @@
+import torch
 import torch.nn as nn
+import torch.nn.utils as utils
 
 class SISUDiscriminator(nn.Module):
     def __init__(self):
         super(SISUDiscriminator, self).__init__()
-        layers = 32
+        layers = 8
         self.model = nn.Sequential(
-            nn.Conv1d(1, layers, kernel_size=5, stride=2, padding=2),
+            utils.spectral_norm(nn.Conv1d(1, layers, kernel_size=7, stride=2, padding=3)),
             nn.BatchNorm1d(layers),
-            nn.LeakyReLU(0.2, inplace=True),
-            nn.Conv1d(layers, layers * 2, kernel_size=5, stride=2, padding=2),
+            nn.PReLU(),
+            nn.Conv1d(layers, layers * 2, kernel_size=7, padding=3),
             nn.BatchNorm1d(layers * 2),
-            nn.LeakyReLU(0.2, inplace=True),
-            nn.Conv1d(layers * 2, layers * 4, kernel_size=5, stride=2, padding=2),
+            nn.PReLU(),
+            nn.Conv1d(layers * 2, layers * 4, kernel_size=5, padding=2),
             nn.BatchNorm1d(layers * 4),
-            nn.LeakyReLU(0.2, inplace=True),
-            nn.Conv1d(layers * 4, layers * 8, kernel_size=5, stride=2, padding=2),
+            nn.PReLU(),
+            nn.Conv1d(layers * 4, layers * 8, kernel_size=3, padding=1),
             nn.BatchNorm1d(layers * 8),
-            nn.LeakyReLU(0.2, inplace=True),
+            nn.PReLU(),
             nn.Conv1d(layers * 8, 1, kernel_size=3, padding=1),
         )
         self.global_avg_pool = nn.AdaptiveAvgPool1d(1)
-        self.sigmoid = nn.Sigmoid()
 
     def forward(self, x):
+        x = x + 0.01 * torch.randn_like(x)
         x = self.model(x)
         x = self.global_avg_pool(x)
         x = x.view(-1, 1)
-        x = self.sigmoid(x)
         return x