-
Notifications
You must be signed in to change notification settings - Fork 19
/
test.py
64 lines (51 loc) · 1.69 KB
/
test.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
"""
### NOTICE ###
You DO NOT need to upload this file
"""
import argparse
import numpy as np
from environment import Environment
seed = 11037
def parse():
parser = argparse.ArgumentParser(description="MLDS 2018 HW4")
parser.add_argument('--test_pg', action='store_true', help='whether test policy gradient')
parser.add_argument('--test_dqn', action='store_true', help='whether test DQN')
try:
from argument import add_arguments
parser = add_arguments(parser)
except:
pass
args = parser.parse_args()
return args
def test(agent, env, total_episodes=30):
rewards = []
env.seed(seed)
for i in range(total_episodes):
state = env.reset()
agent.init_game_setting()
done = False
episode_reward = 0.0
#playing one game
while(not done):
#env.env.render()
action = agent.make_action(state, test=True)
state, reward, done, info = env.step(action)
episode_reward += reward
rewards.append(episode_reward)
print('[ episode ', i, '] upclipped reward :', episode_reward)
print('Run %d episodes'%(total_episodes))
print('Mean:', np.mean(rewards))
def run(args):
if args.test_pg:
env = Environment('Pong-v0', args, test=True)
from agent_dir.agent_pg import Agent_PG
agent = Agent_PG(env, args)
test(agent, env)
if args.test_dqn:
env = Environment('BreakoutNoFrameskip-v4', args, atari_wrapper=True, test=True)
from agent_dir.agent_dqn import Agent_DQN
agent = Agent_DQN(env, args)
test(agent, env, total_episodes=100)
if __name__ == '__main__':
args = parse()
run(args)