init_entcoef: 1.0
reward_signal_steps_per_update: 20.0
encoder_layers: 2
policy_layers: 2
forward_layers: 0
policy_layers: 1
forward_layers: 1
feature_size: 256
feature_size: 128
action_layers: 2
action_feature_size: 128
separate_policy_train: true
trainer_type: sac_transfer
hyperparameters:
learning_rate: 0.0003
learning_rate_schedule: constant
learning_rate_schedule: linear
batch_size: 256
buffer_size: 3000000
buffer_init_steps: 2000
use_transfer: true
load_model: true
train_model: false
load_action: true
train_action: false
transfer_path: "results/oldcs/CrawlerStatic"
network_settings:
normalize: true
yanchaosun