-
Notifications
You must be signed in to change notification settings - Fork 461
/
vtrace.py
executable file
·136 lines (108 loc) · 4.63 KB
/
vtrace.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
import gym
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.distributions import Categorical
#Hyperparameters
learning_rate = 0.0005
gamma = 0.98
T_horizon = 20
clip_rho_threshold = 1.0
clip_c_threshold = 1.0
print_interval = 20
class Vtrace(nn.Module):
def __init__(self):
super(Vtrace, self).__init__()
self.data = []
self.fc1 = nn.Linear(4,256)
self.fc_pi = nn.Linear(256,2)
self.fc_v = nn.Linear(256,1)
self.optimizer = optim.Adam(self.parameters(), lr=learning_rate)
self.clip_rho_threshold = torch.tensor(clip_rho_threshold, dtype=torch.float)
self.clip_c_threshold = torch.tensor(clip_c_threshold, dtype=torch.float)
def pi(self, x, softmax_dim = 0):
x = F.relu(self.fc1(x))
x = self.fc_pi(x)
prob = F.softmax(x, dim=softmax_dim)
return prob
def v(self, x):
x = F.relu(self.fc1(x))
v = self.fc_v(x)
return v
def put_data(self, transition):
self.data.append(transition)
def make_batch(self):
s_lst, a_lst, r_lst, s_prime_lst, mu_a_lst, done_lst = [], [], [], [], [], []
for transition in self.data:
s, a, r, s_prime, mu_a, done = transition
s_lst.append(s)
a_lst.append([a])
r_lst.append([r])
s_prime_lst.append(s_prime)
mu_a_lst.append([mu_a])
done_mask = 0 if done else 1
done_lst.append([done_mask])
s,a,r,s_prime,done_mask, mu_a = torch.tensor(s_lst, dtype=torch.float), torch.tensor(a_lst), \
torch.tensor(r_lst), torch.tensor(s_prime_lst, dtype=torch.float), \
torch.tensor(done_lst, dtype=torch.float), torch.tensor(mu_a_lst)
self.data = []
return s, a, r, s_prime, done_mask, mu_a
def vtrace(self, s, a, r, s_prime, done_mask, mu_a):
with torch.no_grad():
pi = self.pi(s, softmax_dim=1)
pi_a = pi.gather(1,a)
v, v_prime = self.v(s), self.v(s_prime)
ratio = torch.exp(torch.log(pi_a) - torch.log(mu_a)) # a/b == exp(log(a)-log(b))
rhos = torch.min(self.clip_rho_threshold, ratio)
cs = torch.min(self.clip_c_threshold, ratio).numpy()
td_target = r + gamma * v_prime * done_mask
delta = rhos*(td_target - v).numpy()
vs_minus_v_xs_lst = []
vs_minus_v_xs = 0.0
vs_minus_v_xs_lst.append([vs_minus_v_xs])
for i in range(len(delta)-1, -1, -1):
vs_minus_v_xs = gamma * cs[i][0] * vs_minus_v_xs + delta[i][0]
vs_minus_v_xs_lst.append([vs_minus_v_xs])
vs_minus_v_xs_lst.reverse()
vs_minus_v_xs = torch.tensor(vs_minus_v_xs_lst, dtype=torch.float)
vs = vs_minus_v_xs[:-1] + v.numpy()
vs_prime = vs_minus_v_xs[1:] + v_prime.numpy()
advantage = r + gamma * vs_prime - v.numpy()
return vs, advantage, rhos
def train_net(self):
s, a, r, s_prime, done_mask, mu_a = self.make_batch()
vs, advantage, rhos = self.vtrace(s, a, r, s_prime, done_mask, mu_a)
pi = self.pi(s, softmax_dim=1)
pi_a = pi.gather(1,a)
val_loss = F.smooth_l1_loss(self.v(s) , vs)
pi_loss = -rhos * torch.log(pi_a) * advantage
loss = pi_loss + val_loss
self.optimizer.zero_grad()
loss.mean().backward()
self.optimizer.step()
def main():
env = gym.make('CartPole-v1')
model = Vtrace()
score = 0.0
for n_epi in range(10000):
s, _ = env.reset()
done = False
while not done:
for t in range(T_horizon):
prob = model.pi(torch.from_numpy(s).float())
m = Categorical(prob)
a = m.sample().item()
s_prime, r, done, truncated, info = env.step(a)
model.put_data((s, a, r/100.0, s_prime, prob[a].item(), done))
s = s_prime
score += r
if done:
break
model.train_net()
if n_epi%print_interval==0 and n_epi!=0:
print("# of episode :{}, avg score : {:.1f}".format(n_epi, score/print_interval))
score = 0.0
env.close()
if __name__ == '__main__':
main()