-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathRL_SAC_model.py
168 lines (134 loc) · 5.82 KB
/
RL_SAC_model.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
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.distributions import Normal
LOG_SIG_MAX = 2
LOG_SIG_MIN = -20
epsilon = 1e-6
# Initialize Policy weights
def weights_init_(m):
if isinstance(m, nn.Linear):
torch.nn.init.xavier_uniform_(m.weight, gain=1)
torch.nn.init.constant_(m.bias, 0)
class ValueNetwork(nn.Module):
def __init__(self, num_inputs, hidden_dim):
super(ValueNetwork, self).__init__()
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim, hidden_dim)
self.linear3 = nn.Linear(hidden_dim, 1)
self.apply(weights_init_)
def forward(self, state):
x = F.relu(self.linear1(state))
x = F.relu(self.linear2(x))
x = self.linear3(x)
return x
class QNetwork(nn.Module):
def __init__(self, num_inputs, num_actions, hidden_dim):
super(QNetwork, self).__init__()
# Q1 architecture
self.linear1 = nn.Linear(num_inputs + num_actions, hidden_dim)
self.linear2 = nn.Linear(hidden_dim, hidden_dim)
self.linear3 = nn.Linear(hidden_dim, 1)
# Q2 architecture
self.linear4 = nn.Linear(num_inputs + num_actions, hidden_dim)
self.linear5 = nn.Linear(hidden_dim, hidden_dim)
self.linear6 = nn.Linear(hidden_dim, 1)
self.apply(weights_init_)
def forward(self, state, action):
xu = torch.cat([state, action], 1)
x1 = F.relu(self.linear1(xu))
x1 = F.relu(self.linear2(x1))
x1 = self.linear3(x1)
x2 = F.relu(self.linear4(xu))
x2 = F.relu(self.linear5(x2))
x2 = self.linear6(x2)
return x1, x2
class DeterministicPolicy(nn.Module):
def __init__(self, num_inputs, num_actions, hidden_dim, action_space=None):
super(DeterministicPolicy, self).__init__()
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim, hidden_dim)
self.mean = nn.Linear(hidden_dim, num_actions)
self.mean_bhp = nn.Linear(hidden_dim, 5)
self.mean_rate = nn.Linear(hidden_dim, 4)
self.noise = torch.Tensor(num_actions).to(self.device)
self.apply(weights_init_)
self.scale_bhp = (2500-2200)/(4069.2-2200)
self.bias_bhp = (2200-2200)/(4069.2-2200)
self.scale_rate = (1.0e6-1.0e5)/(1.2e6-0)
self.bias_rate = (1.0e5-0)/(1.2e6-0)
# action rescaling
if action_space is None:
self.action_scale = 1.
self.action_bias = 0.
else:
self.action_scale = torch.FloatTensor(
(action_space.high - action_space.low) / 2.)
self.action_bias = torch.FloatTensor(
(action_space.high + action_space.low) / 2.)
def forward(self, state):
x = F.relu(self.linear1(state))
x = F.relu(self.linear2(x))
mean_bhp = torch.sigmoid(self.mean_bhp(x))*self.scale_bhp+self.bias_bhp
mean_rate = torch.sigmoid(self.mean_rate(x))*self.scale_rate+self.bias_rate
# mean_bhp = torch.sigmoid(self.mean_bhp(x))
# mean_rate = torch.sigmoid(self.mean_rate(x))
mean = torch.cat((mean_bhp, mean_rate),dim=-1)
return mean
def sample(self, state):
mean = self.forward(state)
# noise = self.noise.normal_(0., std=0.05)
# noise = noise.clamp(-0.10, 0.10)
# action = mean + noise
action = mean
return action, torch.tensor(0.), mean
# def to(self, device):
# self.action_scale = self.action_scale.to(device)
# self.action_bias = self.action_bias.to(device)
# self.noise = self.noise.to(device)
# return super(DeterministicPolicy, self).to(device)
class GaussianPolicy(nn.Module):
def __init__(self, num_inputs, num_actions, hidden_dim, action_space=None):
super(GaussianPolicy, self).__init__()
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim, hidden_dim)
self.mean_linear = nn.Linear(hidden_dim, num_actions)
self.log_std_linear = nn.Linear(hidden_dim, num_actions)
self.apply(weights_init_)
# action rescaling
if action_space is None:
self.action_scale = torch.tensor(1.)
self.action_bias = torch.tensor(0.)
else:
self.action_scale = torch.FloatTensor(
(action_space.high - action_space.low) / 2.)
self.action_bias = torch.FloatTensor(
(action_space.high + action_space.low) / 2.)
def forward(self, state):
x = F.relu(self.linear1(state))
x = F.relu(self.linear2(x))
mean = self.mean_linear(x)
log_std = self.log_std_linear(x)
log_std = torch.clamp(log_std, min=LOG_SIG_MIN, max=LOG_SIG_MAX)
return mean, log_std
def sample(self, state):
mean, log_std = self.forward(state)
std = log_std.exp()
normal = Normal(mean, std)
x_t = normal.rsample() # for reparameterization trick (mean + std * N(0,1))
# y_t = torch.tanh(x_t)
# action = y_t * self.action_scale + self.action_bias
y_t = torch.sigmoid(x_t)
action = y_t * self.action_scale + self.action_bias
log_prob = normal.log_prob(x_t)
# Enforcing Action Bound
log_prob -= torch.log(self.action_scale * (1 - y_t.pow(2)) + epsilon)
log_prob = log_prob.sum(1, keepdim=True)
# mean = torch.tanh(mean) * self.action_scale + self.action_bias
mean = torch.sigmoid(mean)
return action, log_prob, mean
# def to(self, device):
# self.action_scale = self.action_scale.to(device)
# self.action_bias = self.action_bias.to(device)
# return super(GaussianPolicy, self).to(device)