fvgp version with fix for relative entropy

examples/kernel_function.py

@@ -216,6 +216,21 @@ def forward(self, x):
         x = self.fc3(x)  # Hidden layer 2 to output layer
         return x
 
+    def set_weights_biases(self, weights, biases):
+        weights = [torch.Tensor(entry) for entry in weights]
+        biases = [torch.Tensor(entry) for entry in biases]
+        with torch.no_grad():
+            model.fc1.weight = nn.Parameter(weights[0])
+            model.fc1.bias = nn.Parameter(biases[0])     
+
+            model.fc2.weight = nn.Parameter(weights[1])
+            model.fc2.bias = nn.Parameter(biases[1])  
+
+            model.fc3.weight = nn.Parameter(weights[2])
+            model.fc3.bias = nn.Parameter(biases[2])
+
+
+
 
 # Initialize the network
 #model = WarpNet()