moved some preprocessing code inside a branch so it only executes if

theano/gradient.py

@@ -672,6 +672,13 @@ def _populate_grad_dict(var_to_node_to_idx,
 
             inputs = node.inputs
 
+            output_grads = [access_grad_cache(var) for var in node.outputs]
+
+            if False in [isinstance(g.type, DisconnectedType)
+                    for g in output_grads]:
+                # Some outputs of this op are connected to the cost so we must
+                # call the op's grad method
+
                 # Each Op's grad function requires inputs and output_grads
                 # If the Op destroys any input, but the grad expression uses it,
                 # then chances are the resulting graph will have a dependency
@@ -690,13 +697,6 @@ def _populate_grad_dict(var_to_node_to_idx,
 
                 inputs = [try_to_copy_if_needed(ipt) for ipt in inputs]
 
-            output_grads = [access_grad_cache(var) for var in node.outputs]
-
-            if False in [isinstance(g.type, DisconnectedType)
-                    for g in output_grads]:
-                # Some outputs of this op are connected to the cost so we must
-                # call the ops grad method
-
                 input_grads = node.op.grad(inputs, output_grads)
 
                 if input_grads is None: