-
Notifications
You must be signed in to change notification settings - Fork 1
/
main.py
187 lines (140 loc) · 6.44 KB
/
main.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
# Reference Code
# https://github.com/chagmgang/gail/blob/master/gail_move2beacon/policy_net.py
# 차금강님의 코드를 참조하였습니다.
# https://github.com/uidilr/gail_ppo_tf/blob/master/run_gail.py
# Thank you.
import os
import argparse
import gym
import numpy as np
import tensorflow as tf
from PPO import PPOalgorithm
from policy_network import PolicyNet
from discriminator_network import discriminator
def argparser():
parser = argparse.ArgumentParser()
parser.add_argument('--logdir', help='log directory', default='./log/train/')
parser.add_argument('--savedir', help='save directory', default='./save_model/')
parser.add_argument('--gamma', default=0.95)
parser.add_argument('--iteration', default=int(1000000))
return parser.parse_args()
def main(args):
env = gym.make('CartPole-v0')
env.seed(0)
state_dim = env.observation_space
action_dim = env.action_space.n
'''
PolicyNet(state_dim, action_dim, name, action_type)
'''
curPolicy = PolicyNet(state_dim, action_dim, 'currentPolicy', 'stochastic')
oldPolicy = PolicyNet(state_dim, action_dim, 'oldPolicy', 'stochastic')
PPO = PPOalgorithm( oldPolicy, curPolicy, gamma=args.gamma)
Discriminator = discriminator(state_dim, action_dim, 'Discriminator', 'stochastic')
expert_observations = np.genfromtxt('trajectory/observations.csv')
expert_actions = np.genfromtxt('trajectory/actions.csv', dtype=np.int32)
saver = tf.train.Saver()
checkpoint_path = os.path.join(args.savedir, "model")
ckpt = tf.train.get_checkpoint_state(args.savedir)
with tf.Session() as sess:
writer = tf.summary.FileWriter(args.logdir, sess.graph)
if ckpt and ckpt.model_checkpoint_path:
print("[Restore Model]")
saver.restore(sess, ckpt.model_checkpoint_path)
else:
print("[Initialzie Model]")
sess.run(tf.global_variables_initializer())
obs = env.reset()
reward = 0
success_num = 0
for iteration in range(args.iteration):
observations = []
actions = []
rewards = []
v_preds = []
run_policy_steps = 0
while True:
env.render()
run_policy_steps += 1
obs = np.stack([obs]).astype(dtype=np.float32) # prepare to feed placeholder Policy.obs
'''
현재 정책에 따라서 observation을 보고 action과 V(s)를 예측합니다.
'''
act, v_pred = curPolicy.estimate(obs=obs)
act = np.asscalar(act)
v_pred = np.asscalar(v_pred)
observations.append(obs)
actions.append(act)
rewards.append(reward)
v_preds.append(v_pred)
next_obs, reward, done, info = env.step(act)
if done:
v_preds_next = v_preds[1:] + [0] # next state of terminate state has 0 state value
obs = env.reset()
reward = -1
break
else:
obs = next_obs
writer.add_summary(tf.Summary(value=[tf.Summary.Value(tag='episode_length', simple_value=run_policy_steps)])
, iteration)
writer.add_summary(tf.Summary(value=[tf.Summary.Value(tag='episode_reward', simple_value=sum(rewards))])
, iteration)
print(iteration, sum(rewards), success_num)
if sum(rewards) >= 190:
success_num += 1
if success_num >= 5:
saver.save(sess, args.savedir + '/model.ckpt')
print('Clear!! Model saved.')
break
else:
success_num = 0
'''
tf.placholder에 데이터를 넣기 위해서 list를 numpy array로 변환 합니다.
'''
observations = np.reshape(observations, newshape=[-1] + list(state_dim.shape))
actions = np.array(actions).astype(dtype=np.int32)
'''
Discriminator를 업데이트 합니다.
'''
for i in range(2):
sample_indices = (np.random.randint(expert_observations.shape[0], size=observations.shape[0]))
inp = [expert_observations, expert_actions]
sampled_inp = [np.take(a=a, indices=sample_indices, axis=0) for a in inp] # sample training data
Discriminator.update(sampled_inp[0], sampled_inp[1], observations, actions)
'''
GAIL 의 논문을 참조하세요.
GAIL 논문의 reward 는 Discriminator의 (s,a) pair 에 대한 learner의 policy에서 생성된 것인지를 판단하는
확률 입니다.
'''
d_rewards = Discriminator.get_reward(learner_s=observations, learner_a=actions)
d_rewards = np.reshape(d_rewards, newshape=[-1]).astype(dtype=np.float32)
gaes = PPO.get_gaes(rewards=d_rewards, v_preds=v_preds, v_next_preds=v_preds_next)
gaes = np.array(gaes).astype(dtype=np.float32)
# gaes = (gaes - gaes.mean()) / gaes.std()
v_preds_next = np.array(v_preds_next).astype(dtype=np.float32)
'''
Policy network들을 update합니다.
'''
inp = [observations, actions, gaes, d_rewards, v_preds_next]
PPO.update()
for epoch in range(15):
'''
MiniBatch
'''
sample_indices = np.random.randint(low=0, high=observations.shape[0],
size=32)
sampled_inp = [np.take(a=a, indices=sample_indices, axis=0) for a in inp] # sample training data
PPO.train(obs=sampled_inp[0],
actions=sampled_inp[1],
GAEs=sampled_inp[2],
rewards=sampled_inp[3],
estimated_v=sampled_inp[4])
summary = PPO.get_summary(obs=inp[0],
actions=inp[1],
GAEs=inp[2],
rewards=inp[3],
estimated_v=inp[4])
writer.add_summary(summary, iteration)
writer.close()
if __name__=="__main__":
args = argparser()
main(args)