reward_signal_steps_per_update: 20.0
encoder_layers: 1
policy_layers: 2
forward_layers: 0
forward_layers: 1
feature_size: 128
action_feature_size: 32
feature_size: 64
action_feature_size: 16
separate_policy_train: true
separate_policy_net: true
separate_model_train: true
reuse_encoder: false
train_model: false
load_action: true
train_action: false
transfer_path: "results/reacher-stack/Reacher"
transfer_path: "results/reacher-fix/Reacher"
network_settings:
normalize: true
hidden_units: 128