diff --git a/training/scripts/train_agent.py b/training/scripts/train_agent.py
old mode 100755
new mode 100644
index 447787e2f..c03cc9f19
--- a/training/scripts/train_agent.py
+++ b/training/scripts/train_agent.py
@@ -133,7 +133,7 @@ def main():
             verbose=1,
         )
     elif args.agent is not None:
-        agent: Union[Type[BaseAgent], Type[ActorCriticPolicy]] = AgentFactory.instantiate(args.agent)
+        agent: Union[Type[BaseAgent], Type[ActorCriticPolicy]] = AgentFactory.instantiate(args.agent, path = args.path)
         if isinstance(agent, BaseAgent):
             model = PPO(
                 agent.type.value,
diff --git a/training/tools/argsparser.py b/training/tools/argsparser.py
index 46645e7e2..cb9110a97 100644
--- a/training/tools/argsparser.py
+++ b/training/tools/argsparser.py
@@ -23,6 +23,7 @@ def training_args(parser):
     import rosnav.model.custom_policy
     import rosnav.model.custom_sb3_policy
     from rosnav.model.agent_factory import AgentFactory
+    import rosnav.model.custom_policy_from_json
 
     group.add_argument(
         "--agent",
@@ -60,6 +61,11 @@ def training_args(parser):
     parser.add_argument(
         "--tb", action="store_true", help="enables tensorboard logging"
     )
+    parser.add_argument(
+        "--path",
+        type=str,
+        help="path to the json file containing" "the neural network",
+    )
 
 
 def run_agent_args(parser):