Reinforcement-Learning-Pacman/test_generalisation.log at main · JeremyMaille/Reinforcement-Learning-Pacman · 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
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
=== Entraînement sur smallClassic contre RandomGhost ===
Descente du gradientBeginning 1000 episodes of Training
Reinforcement Learning Status:
	Completed 100 out of 1000 training episodes
	Average Rewards over all training: -318.31
	Average Rewards for last 100 episodes: -318.31
	Episode took 4.47 seconds
Reinforcement Learning Status:
	Completed 200 out of 1000 training episodes
	Average Rewards over all training: -253.29
	Average Rewards for last 100 episodes: -188.27
	Episode took 4.98 seconds
Reinforcement Learning Status:
	Completed 300 out of 1000 training episodes
	Average Rewards over all training: -194.83
	Average Rewards for last 100 episodes: -77.91
	Episode took 5.41 seconds
Reinforcement Learning Status:
	Completed 400 out of 1000 training episodes
	Average Rewards over all training: -180.83
	Average Rewards for last 100 episodes: -138.84
	Episode took 4.19 seconds
Reinforcement Learning Status:
	Completed 500 out of 1000 training episodes
	Average Rewards over all training: -173.19
	Average Rewards for last 100 episodes: -142.61
	Episode took 4.53 seconds
Reinforcement Learning Status:
	Completed 600 out of 1000 training episodes
	Average Rewards over all training: -160.62
	Average Rewards for last 100 episodes: -97.78
	Episode took 4.52 seconds
Reinforcement Learning Status:
	Completed 700 out of 1000 training episodes
	Average Rewards over all training: -157.64
	Average Rewards for last 100 episodes: -139.75
	Episode took 4.95 seconds
Reinforcement Learning Status:
	Completed 800 out of 1000 training episodes
	Average Rewards over all training: -159.59
	Average Rewards for last 100 episodes: -173.27
	Episode took 3.95 seconds
Reinforcement Learning Status:
	Completed 900 out of 1000 training episodes
	Average Rewards over all training: -165.63
	Average Rewards for last 100 episodes: -213.90
	Episode took 3.99 seconds
Reinforcement Learning Status:
	Completed 1000 out of 1000 training episodes
	Average Rewards over all training: -148.56
	Average Rewards for last 100 episodes: 5.04
	Episode took 4.82 seconds
Training Done (turning off epsilon and alpha)
---------------------------------------------
=== Test sur mediumClassic contre DirectionalGhost ===
Beginning 0 episodes of Training
Pacman died! Score: -444
Pacman died! Score: -513
Pacman died! Score: -530
Pacman died! Score: -530
Pacman died! Score: -518
Pacman died! Score: -513
Pacman died! Score: -510
Pacman died! Score: -504
Pacman died! Score: -514
Pacman died! Score: -519
Pacman died! Score: -526
Pacman died! Score: -510
Pacman died! Score: -529
Pacman died! Score: -505
Pacman died! Score: -535
Pacman died! Score: -520
Pacman died! Score: -522
Pacman died! Score: -540
Pacman died! Score: -525
Pacman died! Score: -513
Average Score: -516.0
Scores:        -444.0, -513.0, -530.0, -530.0, -518.0, -513.0, -510.0, -504.0, -514.0, -519.0, -526.0, -510.0, -529.0, -505.0, -535.0, -520.0, -522.0, -540.0, -525.0, -513.0
Win Rate:      0/20 (0.00)
Record:        Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss, Loss
 Loss