policy_layers: 1
forward_layers: 1
value_layers: 1
feature_size: 32
feature_size: 16
action_feature_size: 32
reuse_encoder: true
in_epoch_alter: false
in_batch_alter: true
network_settings:
normalize: true
hidden_units: 128
hidden_units: 64
num_layers: 1
vis_encode_type: simple
reward_signals: