distance based penalty

Project/Assets/ML-Agents/Examples/Reacher/Scripts/NewReacherAgent.cs

         // {
         //     AddReward(-0.002f);
         // }
-        // AddReward( - 0.001f * (goal.transform.position - hand.transform.position).magnitude);
+        AddReward( - 0.001f * (goal.transform.position - hand.transform.position).magnitude);
         // Debug.Log((goal.transform.position - hand.transform.position).magnitude);
         var radians = m_GoalDegree * Mathf.PI / 180f;
         var goalX = 8f * Mathf.Cos(radians);

Project/Assets/ML-Agents/Examples/Reacher/Scripts/ReacherAgent.cs

         // {
         //     AddReward(-0.002f);
         // }
-        // AddReward( - 0.001f * (goal.transform.position - hand.transform.position).magnitude);
+        AddReward( - 0.001f * (goal.transform.position - hand.transform.position).magnitude);
         // Debug.Log((goal.transform.position - hand.transform.position).magnitude);
         var radians = m_GoalDegree * Mathf.PI / 180f;
         var goalX = 8f * Mathf.Cos(radians);

config/sac_transfer/ReacherTransfer.yaml

       train_model: false
       load_action: true
       train_action: false
-      transfer_path: "results/reacher-ori-sta/Reacher"
+      transfer_path: "results/reacher-stack/Reacher"
     network_settings:
       normalize: true
       hidden_units: 128