load_model: false
encoder_layers: 2
action_layers: 2
policy_layers: 2
forward_layers: 0
value_layers: 2
feature_size: 128
policy_layers: 1
forward_layers: 1
value_layers: 1
feature_size: 64
reuse_encoder: false
reuse_encoder: true
in_epoch_alter: false
use_op_buffer: false
use_var_predict: true
separate_value_train: true
separate_value_train: false
in_batch_alter: true
network_settings:
normalize: true
model_schedule: linear
train_model: false
load_model: true
train_action: false
Andrew Cohen