large buffer size

config/sac_transfer/PushBlock.yaml

     trainer_type: sac_transfer
     hyperparameters:
       learning_rate: 0.0003
-      learning_rate_schedule: constant
+      learning_rate_schedule: linear
+      model_schedule: constant
-      buffer_size: 50000
+      buffer_size: 2000000
       buffer_init_steps: 0
       tau: 0.005
       steps_per_update: 10.0
       policy_layers: 2
       forward_layers: 2
       value_layers: 2
-      action_layers: -1
+      action_layers: 2
       feature_size: 128
       action_feature_size: 64
       separate_policy_train: true

config/sac_transfer/PushBlockTransfer.yaml

       learning_rate: 0.0003
       learning_rate_schedule: constant
       batch_size: 128
-      buffer_size: 50000
+      buffer_size: 2000000
       buffer_init_steps: 0
       tau: 0.005
       steps_per_update: 10.0
       policy_layers: 2
       forward_layers: 2
       value_layers: 2
-      action_layers: -1
+      action_layers: 2
       feature_size: 128
       action_feature_size: 64
       separate_policy_train: true
       use_transfer: true
       load_model: true
       train_model: false
-      # load_action: true
-      # train_action: false
+      load_action: true
+      train_action: false
       transfer_path: "results/block/PushBlock"
     network_settings:
       normalize: false

config/sac_transfer/Reacher.yaml

       learning_rate_schedule: linear
       model_schedule: constant
       batch_size: 128
-      buffer_size: 500000
+      buffer_size: 6000000
       buffer_init_steps: 0
       tau: 0.005
       steps_per_update: 20.0
       policy_layers: 2
       forward_layers: 0
       value_layers: 2
-      action_layers: 2
+      action_layers: 1
       feature_size: 64
       action_feature_size: 16
       separate_policy_train: true

config/sac_transfer/ReacherTransfer.yaml

       learning_rate_schedule: constant
       model_schedule: constant
       batch_size: 128
-      buffer_size: 500000
+      buffer_size: 6000000
       buffer_init_steps: 0
       tau: 0.005
       steps_per_update: 20.0
-      encoder_layers: 2
+      encoder_layers: 1
-      action_layers: 2
+      action_layers: 1
       feature_size: 64
       action_feature_size: 16
       separate_policy_train: true
       train_model: false
       load_action: true
       train_action: false
-      transfer_path: "results/reacher/Reacher"
+      transfer_path: "results/sacmod-reacher/Reacher"
     network_settings:
       normalize: true
       hidden_units: 128