Make lists modulelists

ml-agents/mlagents/trainers/torch/layers.py

         super().__init__()
         self.layers = []
         self.goal_encoders = []
-        prev_size = input_size
-        for i in range(num_layers):
+        prev_size = input_size + goal_size
+        for _ in range(num_layers):
             self.layers.append(
                 linear_layer(
                     prev_size,
                 )
             )
-            self.goal_encoders.append(LinearEncoder(goal_size, 2, hidden_size, final_activation=False))
+            self.goal_encoders.append(
+                LinearEncoder(goal_size, 2, hidden_size, final_activation=True)
+            )
+        self.layers = torch.nn.ModuleList(self.layers)
+        self.goal_encoders = torch.nn.ModuleList(self.goal_encoders)
-        activation = input_tensor
+        activation = torch.cat([input_tensor, goal_tensor], dim=-1)
-                activation = layer(activation) * self.goal_encoders[idx//2](goal_tensor)
+                activation = layer(activation) + self.goal_encoders[idx // 2](
+                    goal_tensor
+                )
         return activation
 
 
                     HyperNetwork(prev_size, hidden_size, goal_size, 2, hidden_size)
                 )
             self.layers.append(Swish())
+            self.layers = torch.nn.ModuleList(self.layers)
             prev_size = hidden_size
 
     def forward(

ml-agents/mlagents/trainers/torch/networks.py

 from mlagents.trainers.torch.layers import (
     LSTM,
     LinearEncoder,
-    HyperNetwork,
-    ConditionalEncoder, HyperEncoder,
+    ConditionalEncoder,
+    HyperEncoder,
 )
 from mlagents.trainers.torch.encoders import VectorInput
 from mlagents.trainers.buffer import AgentBuffer
                 total_goal_size,
                 network_settings.num_layers,
                 self.h_size,
-                num_hyper_layers=0
+                num_hyper_layers=1,
             )
         elif (
             ObservationType.GOAL in self.obs_types