linear value, no target

config/sac_transfer/3DBall.yaml

       init_entcoef: 0.5
       reward_signal_steps_per_update: 10.0
       encoder_layers: 2
-      policy_layers: 0
+      policy_layers: 1
-      value_layers: 2
-      feature_size: 32
-      separate_value_net: true
-      # separate_policy_train: true
-      reuse_encoder: false
+      value_layers: 0
+      feature_size: 64
+      # separate_value_net: true
+      separate_policy_train: true
+      # separate_value_train: true
+      separate_model_train: true
+      reuse_encoder: true
       in_epoch_alter: false
       in_batch_alter: true
       use_op_buffer: false

config/sac_transfer/3DBallHard.yaml

       init_entcoef: 1.0
       reward_signal_steps_per_update: 10.0
       encoder_layers: 2
-      policy_layers: 0
+      policy_layers: 1
-      value_layers: 2
-      feature_size: 32
-      separate_value_net: true
-      # separate_policy_train: true
-      reuse_encoder: false
+      value_layers: 0
+      feature_size: 64
+      # separate_value_net: true
+      separate_policy_train: true
+      # separate_value_train: true
+      separate_model_train: true
+      reuse_encoder: true
       in_epoch_alter: false
       in_batch_alter: true
       use_op_buffer: false

config/sac_transfer/3DBallHardTransfer.yaml

       save_replay_buffer: false
       init_entcoef: 1.0
       reward_signal_steps_per_update: 10.0
-      encoder_layers: 2
+      encoder_layers: 1
-      forward_layers: 0
+      forward_layers: 1
+      # separate_value_train: true
       reuse_encoder: false
       in_epoch_alter: false
       in_batch_alter: false
       use_transfer: true
       load_model: true
       train_model: false
-      transfer_path: "results/sac-ball-lintest/3DBall"
+      transfer_path: "results/sac_model_ball_ml1/3DBall"
     network_settings:
       normalize: true
       hidden_units: 64

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

             self.processed_vector_in, 
             vis_encode_type,
             encoder_layers=encoder_layers,
-            scope="target_enc",
+            scope="encoding",
             reuse=True
         )
         if separate_train:

config/sac_transfer/3DBallHardTransfer1.yaml

+behaviors:
+  3DBallHard:
+    trainer_type: sac_transfer
+    hyperparameters:
+      learning_rate: 0.0003
+      learning_rate_schedule: linear
+      batch_size: 256
+      buffer_size: 50000
+      buffer_init_steps: 0
+      tau: 0.005
+      steps_per_update: 10.0
+      save_replay_buffer: false
+      init_entcoef: 1.0
+      reward_signal_steps_per_update: 10.0
+      encoder_layers: 2
+      policy_layers: 1
+      forward_layers: 0
+      value_layers: 0
+      feature_size: 64
+      # separate_value_net: true
+      # separate_model_train: true
+      separate_policy_train: true
+      reuse_encoder: true
+      in_epoch_alter: false
+      in_batch_alter: false
+      use_op_buffer: false
+      use_var_predict: true
+      with_prior: false
+      predict_return: true
+      use_bisim: false
+      use_transfer: true
+      load_model: true
+      load_policy: true
+      train_policy: false
+      train_model: false
+      transfer_path: "results/sacmodel-ball-v0/3DBall"
+    network_settings:
+      normalize: true
+      hidden_units: 64
+      num_layers: 2
+      vis_encode_type: simple
+    reward_signals:
+      extrinsic:
+        gamma: 0.99
+        strength: 1.0
+    keep_checkpoints: 5
+    max_steps: 500000
+    time_horizon: 1000
+    summary_freq: 12000
+    threaded: true