Add files via upload

functions/SoftmaxPower.py

@@ -0,0 +1,29 @@
+import torch
+
+
+class SoftmaxPower(torch.nn.Module):
+    dim: int | None
+    power: float
+
+    def __init__(self, power: float = 2.0, dim: int | None = None) -> None:
+        super().__init__()
+        self.dim = dim
+        self.power = power
+
+    def __setstate__(self, state):
+        super().__setstate__(state)
+        if not hasattr(self, "dim"):
+            self.dim = None
+        if not hasattr(self, "power"):
+            self.power = 2.0
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        output: torch.Tensor = torch.abs(input).pow(self.power)
+        if self.dim is None:
+            output = output / output.sum()
+        else:
+            output = output / output.sum(dim=self.dim, keepdim=True)
+        return output
+
+    def extra_repr(self) -> str:
+        return f"dim={self.dim} ; power={self.power}"

functions/make_cnn.py

@@ -1,5 +1,6 @@
 import torch
 import numpy as np
+from functions.SoftmaxPower import SoftmaxPower
 
 
 def make_cnn(
@@ -14,6 +15,7 @@ def make_cnn(
     mp_1_stride: int,
     pooling_type: str,
     conv_0_enable_softmax: bool,
+    conv_0_power_softmax: float,
     l_relu_negative_slope: float,
 ) -> torch.nn.Sequential:
     assert len(conv_out_channels_list) >= 1
@@ -58,6 +60,9 @@ def make_cnn(
     assert setting_understood
 
     if conv_0_enable_softmax:
+        if conv_0_power_softmax != 0.0:
+            cnn.append(SoftmaxPower(dim=1, power=conv_0_power_softmax))
+        else:
             cnn.append(torch.nn.Softmax(dim=1))
 
     # Changing structure