reward_signal_steps_per_update: 20.0
encoder_layers: 2
policy_layers: 1
forward_layers: 1
value_layers: 2
feature_size: 128
forward_layers: 0
value_layers: 3
feature_size: 64
action_feature_size: 128
action_feature_size: 32
separate_policy_train: true
separate_model_train: true
# separate_value_net: true
action_feature_size: 64
# separate_model_train: true
train_model: false
load_action: true
train_action: false
transfer_path: "results/oldcs-m1/CrawlerStatic"
transfer_path: "results/oldcs-f64/CrawlerStatic"
network_settings:
normalize: true
hidden_units: 512