RL/breakout.ini at master · dgtgrade/RL · GitHub

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
[Atari]
SCREEN_Y            = 210
SCREEN_X            = 160
SCREEN_Z            = 3
SCREEN_LOW_Y        = 42
SCREEN_LOW_X        = 32
SCREEN_LOW_Z        = 3

[Breakout]
ACTION_N            = 3
ACTION_OFFSET       = 1
ACTION_NOISE_0      = 0.1
ACTION_NOISE_1      = 0.0

[PolicyNetwork]
N_CONV_1            = 16
N_CONV_2            = 16
N_CONV_3            = 32
N_CONV_4            = 32
N_FC_1              = 512
N_FC_2              = 128
LEARNING_RATE       = 1e-3
BN_DECAY            = 0.97

[Trainer]
PLAY_ONLY           = yes
;
RUNS                = 10000
EPISODES_PER_RUN    = 2
SAVE_MODEL_PER_RUNS = 10
;
MINI_BATCH_SIZE     = 2000
EPOCHS_PER_LEARN    = 2
PRINT_PER_EPOCHS    = 1
;
PLAYER_N            = 20
;
PAST_EX_USE         = yes
PAST_EX_MM          = 5
PAST_EX_MAX         = 500000

[Player]
T_MAX               = 5000
IMPACT_DECAY        = 0.98

[Tensorflow]
MODEL_DIR           = ckpt/breakout/
MODEL_FILE          = model.ckpt
MODEL_LOAD          = yes