Change final value network activation to identity

config/value_dice.json5

@@ -65,7 +65,7 @@
             hidden_n: 3,
             hidden_dim: 50,
             output_dim: 1, // number of outputs e.g. number of actions, or just one
-            final_activation: 'sigmoid',
+            final_activation: 'id',
         },
     },
     policy_optim: {

src/nets/util.py

@@ -1,5 +1,5 @@
 import torch
-from torch.nn import functional
+from torch.nn import functional, Identity
 
 def parse_functional(functional_config):
     if isinstance(functional_config, str):
@@ -9,4 +9,6 @@ def parse_functional(functional_config):
             return torch.sigmoid
         elif functional_config == 'softmax':
             return functional.softmax
+        elif functional_config == 'id':
+            return Identity()
     return None