-
Notifications
You must be signed in to change notification settings - Fork 0
/
config.py
56 lines (41 loc) · 1.27 KB
/
config.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
"""This file is to put any configuration to our classes"""
class EnvConfig(object):
env_name = 'Breakout-v0'
state_shape = [84, 84, 4]
state_processor_params = {"resize_shape": (84, 84),
"crop_box": (34, 0, 160, 160),
"gray": True,
"frames_num": 4}
record_video_every = 5
class AgentConfig(object):
initial_epsilon = 1.0
final_epsilon = 0.1
exploration_steps = 750000
policy_fn = 'epsilon_greedy'
discount_factor = 0.99
evaluate_every = 25
evaluation_episodes = 5
class ReplayMemoryConfig(object):
rep_max_size = 500000
replay_memory_init_size = 50000
class PRMConfig(object):
prm = True
prm_max_size = 500000
prm_init_size = 50000
initial_alpha = 0.5
initial_beta = 0.0
beta_grad = 0
alpha_grad = 0.5 / 5000000
class EstimatorConfig(object):
name = "DQN_Dragon"
learning_rate = 0.00025
class Experiment1(EnvConfig, AgentConfig, ReplayMemoryConfig, EstimatorConfig, PRMConfig):
is_train = True
cont_training = True
is_play = False
num_episodes = 10000
update_target_estimator_every = 10000
batch_size = 32
experiment_dir = "./experiment_1/"
def get_config():
return Experiment1