hyperparameters:
learning_rate: 0.0003
learning_rate_schedule: linear
model_schedule: linear
model_schedule: constant
buffer_size: 200000
buffer_size: 500000
buffer_init_steps: 0
tau: 0.005
steps_per_update: 10.0
policy_layers: 0
forward_layers: 0
value_layers: 1
feature_size: 16
action_layers: 1
feature_size: 64
action_feature_size: 16
# separate_value_net: true
separate_policy_train: true
# separate_value_train: true
use_bisim: false
use_transfer: true
load_model: true
load_value: true
transfer_path: "results/ball_f16_linear/3DBall"
transfer_path: "results/ball_f64_linear/3DBall"
network_settings:
normalize: true
hidden_units: 128
reuse_encoder: true
in_epoch_alter: false
in_batch_alter: false
in_batch_alter: true
use_op_buffer: false
use_var_predict: true
with_prior: false
train_model: false
load_policy: true
transfer_path: "results/ball-linear-s0/3DBall"
transfer_path: "results/ball-linear-s10/3DBall"
transfer_path: "results/ball-linear-s1/3DBall"
transfer_path: "results/ball-linear-s11/3DBall"
transfer_path: "results/ball-linear-s2/3DBall"
transfer_path: "results/ball-linear-s12/3DBall"
transfer_path: "results/ball-linear-s3/3DBall"
transfer_path: "results/ball-linear-s13/3DBall"
transfer_path: "results/ball-linear-s4/3DBall"
transfer_path: "results/ball-linear-s14/3DBall"