forked from karinemiras/evoman_framework
-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_gatherer_PPO.py
210 lines (186 loc) · 8.42 KB
/
data_gatherer_PPO.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
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
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
import csv
import os
import subprocess
import sys
import cv2
import numpy as np
from stable_baselines3 import PPO, A2C
from stable_baselines3.common.callbacks import BaseCallback
from stable_baselines3.common.monitor import Monitor
from map_enemy_id_to_name import id_to_name
sys.path.insert(0, 'evoman')
from gym_environment import Evoman
algorithm = 'PPO'
runs = int(sys.argv[1])
runs_start = int(sys.argv[2])
randomini = sys.argv[3]
if randomini != 'yes' and randomini != 'no':
raise EnvironmentError()
if runs < 0:
runs = sys.maxsize
environments = [
(
n,
[(
Monitor(Evoman(
enemyn=str(n),
weight_player_hitpoint=weight_player_hitpoint,
weight_enemy_hitpoint=1.0 - weight_player_hitpoint,
randomini=randomini,
)),
Monitor(Evoman(
enemyn=str(n),
weight_player_hitpoint=1,
weight_enemy_hitpoint=1,
randomini=randomini,
))
) for weight_player_hitpoint in [0.5]]
)
for n in [1, 2, 4, 7]
]
class EvalEnvCallback(BaseCallback):
def __init__(
self,
eval_env,
lengths_path,
rewards_path,
models_dir = None,
video_dir = None,
raw_data_dir = None,
verbose: int = 0,
lengths_prepend: list = [],
rewards_prepend: list = [],
n_eval_episodes: int = 5,
eval_freq: int = 10000,
model_freq: int = 100000,
video_freq: int = 25000,
):
super(EvalEnvCallback, self).__init__(verbose=verbose)
if not os.path.exists(models_dir) and models_dir is not None:
os.makedirs(models_dir)
if not os.path.exists(video_dir) and video_dir is not None:
os.makedirs(video_dir)
if not os.path.exists(raw_data_dir) and raw_data_dir is not None:
os.makedirs(raw_data_dir)
self.eval_env = eval_env
self.lengths_path = lengths_path
self.rewards_path = rewards_path
self.lengths_prepend = lengths_prepend
self.rewards_prepend = rewards_prepend
self.video_dir = video_dir
self.model_freq = model_freq
self.video_freq = video_freq
self.n_eval_episodes = n_eval_episodes
self.eval_freq = eval_freq
self.lengths = []
self.rewards = []
self.models_dir = models_dir
self.raw_data_dir = raw_data_dir
def _on_step(self) -> bool:
if self.n_calls % self.model_freq == 0:
if self.models_dir is not None:
self.model.save(f'{self.models_dir}/{self.n_calls}.model')
if self.n_calls % self.video_freq == 0:
if self.video_dir is not None:
obs = self.eval_env.reset()
fourcc = cv2.VideoWriter_fourcc('M', 'J', 'P', 'G')
fps = 30
video_filename = f'{self.video_dir}/{self.n_calls}-temp.avi'
video_filename_compresed = f'{self.video_dir}/{self.n_calls}.avi'
out = cv2.VideoWriter(video_filename, fourcc, fps, (self.eval_env.WIDTH, self.eval_env.HEIGHT))
for _ in range(3500):
action, _state = model.predict(obs, deterministic=False)
obs, reward, done, info = self.eval_env.step(action)
out.write(self.eval_env.render("bgr"))
if done:
break
out.release()
compac = f'ffmpeg -i "{video_filename}" -vcodec h264 "{video_filename_compresed}" -y ; rm "{video_filename}"'
print(compac)
os.system(compac)
if self.n_calls % self.eval_freq == 0:
self.evaluate()
return True
def evaluate(self):
wins = []
rs = []
ls = []
for j in range(self.n_eval_episodes):
obs = self.eval_env.reset()
rew = 0
for s in range(3500):
action, _state = self.model.predict(obs, deterministic=False)
obs, reward, done, info = self.eval_env.step(action)
rew = rew + reward
if done:
if self.eval_env.env.enemy.life <= 0:
wins.append(1)
else:
wins.append(0)
ls.append(s)
break
rs.append(rew)
self.lengths.append(np.mean(ls))
self.rewards.append(np.mean(rs))
with open(f'{self.raw_data_dir}/wins.csv', mode='a') as wins_file:
wins_writer = csv.writer(wins_file, delimiter=',', quotechar='\'', quoting=csv.QUOTE_NONNUMERIC)
with open(f'{self.raw_data_dir}/rewards.csv', mode='a') as rewards_file:
rewards_writer = csv.writer(rewards_file, delimiter=',', quotechar='\'', quoting=csv.QUOTE_NONNUMERIC)
wins_writer.writerow([self.n_calls, self.n_eval_episodes, ''] + wins)
rewards_writer.writerow([self.n_calls, self.n_eval_episodes, ''] + rs)
def _on_training_end(self) -> None:
self.evaluate()
with open(f'{self.lengths_path}/Evaluation_lengths.csv', mode='a') as eval_lengths_file:
l_writer = csv.writer(eval_lengths_file, delimiter=',', quotechar='\'', quoting=csv.QUOTE_NONNUMERIC)
with open(f'{self.rewards_path}/Evaluation_rewards.csv', mode='a') as eval_rewards_file:
r_writer = csv.writer(eval_rewards_file, delimiter=',', quotechar='\'', quoting=csv.QUOTE_NONNUMERIC)
l_writer.writerow(self.lengths_prepend+self.lengths)
r_writer.writerow(self.rewards_prepend+self.rewards)
for run in range(runs_start, runs_start+runs):
print(f'Starting run {run}!')
if randomini:
baseDir = f'FinalData/RandomIni/{algorithm}/run{run}'
else:
baseDir = f'FinalData/StaticIni/{algorithm}/run{run}'
if not os.path.exists(baseDir):
os.makedirs(baseDir)
for enemy_id, enemy_envs in environments:
enemyDir = f'{baseDir}/{id_to_name(enemy_id)}'
if not os.path.exists(enemyDir):
os.makedirs(enemyDir)
for env, eval_env in enemy_envs:
modelDir = f'{enemyDir}/models/{({env.env.weight_player_hitpoint}, {env.env.weight_enemy_hitpoint})}'
videoDir = f'{enemyDir}/videos/{({env.env.weight_player_hitpoint}, {env.env.weight_enemy_hitpoint})}'
rawDataDir = f'{enemyDir}/raw-data/{({env.env.weight_player_hitpoint}, {env.env.weight_enemy_hitpoint})}'
if not os.path.exists(modelDir):
os.makedirs(modelDir)
if not os.path.exists(videoDir):
os.makedirs(videoDir)
if not os.path.exists(rawDataDir):
os.makedirs(rawDataDir)
env.env.keep_frames = False
model = PPO('MlpPolicy', env)
l_prepend = [f'{id_to_name(enemy_id)}', ""]
r_prepend = [f'{id_to_name(enemy_id)} ({env.env.weight_player_hitpoint}, {env.env.weight_enemy_hitpoint})', str(env.env.win_value())]
model.learn(total_timesteps=int(2.5e6), callback=EvalEnvCallback(
eval_env=eval_env,
lengths_path=enemyDir,
rewards_path=enemyDir,
models_dir=modelDir,
video_dir=videoDir,
raw_data_dir=rawDataDir,
lengths_prepend=l_prepend,
rewards_prepend=r_prepend,
eval_freq=12500,
n_eval_episodes=25,
))
with open(f'{enemyDir}/Training_lengths.csv', mode='a') as train_lengths_file:
train_lengths_writer = csv.writer(train_lengths_file, delimiter=',', quotechar='\'',
quoting=csv.QUOTE_NONNUMERIC)
with open(f'{enemyDir}/Training_rewards.csv', mode='a') as train_rewards_file:
train_rewards_writer = csv.writer(train_rewards_file, delimiter=',', quotechar='\'',
quoting=csv.QUOTE_NONNUMERIC)
train_lengths_writer.writerow(l_prepend+env.get_episode_lengths())
train_rewards_writer.writerow(r_prepend+env.get_episode_rewards())
print(f'\nFinished {id_to_name(enemy_id)} ({env.env.weight_player_hitpoint}, {env.env.weight_enemy_hitpoint})')
print(f'\n\nFinished {id_to_name(enemy_id)} completely\n\n')