disable bisim

config/ppo_transfer/TransferCrawlerStatic.yaml

       use_var_predict: true
       with_prior: false
       predict_return: true
-      use_bisim: true
+      use_bisim: false
       separate_value_net: true
       use_transfer: true
       load_policy: false

ml-agents/mlagents/trainers/sac_transfer/trainer.py

 
         # Update the normalization
         if self.is_training:
-            self.policy.update_normalization(agent_buffer_trajectory["vector_obs"])
+            self.policy.update_normalization(
+                agent_buffer_trajectory["vector_obs"],
+                agent_buffer_trajectory["next_vector_in"],
+                agent_buffer_trajectory["vector_obs"],
+            )
 
         # Evaluate all reward functions for reporting purposes
         self.collected_rewards["environment"][agent_id] += np.sum(