-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmodel.py
187 lines (159 loc) · 6.76 KB
/
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
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
def hidden_init(layer):
fan_in = layer.weight.data.size()[0]
lim = 1. / np.sqrt(fan_in)
return (-lim, lim)
class Actor(nn.Module):
"""Actor (Policy) Model."""
def __init__(self, state_size, action_size, seed, fc1_units=400, fc2_units=300, bn_mode=0):
"""Initialize parameters and build model.
Params
======
state_size (int): Dimension of each state
action_size (int): Dimension of each action
seed (int): Random seed
fc1_units (int): Number of nodes in first hidden layer
fc2_units (int): Number of nodes in second hidden layer
bn_mode (int): Use Batch Normalization - 0=disabled, 1=BN before Activation, 2=BN after Activation (3, 4 are alt. versions of 1, 2)
"""
super(Actor, self).__init__()
self.seed = torch.manual_seed(seed)
# Dense layers
self.fc1 = nn.Linear(state_size, fc1_units)
self.fc2 = nn.Linear(fc1_units, fc2_units)
self.fc3 = nn.Linear(fc2_units, action_size)
# Normalization layers
self.bn1 = nn.BatchNorm1d(fc1_units)
if bn_mode!=2:
self.bn2 = nn.BatchNorm1d(fc2_units)
if bn_mode==3:
self.bn3 = nn.BatchNorm1d(action_size)
self.bn_mode=bn_mode
self.reset_parameters()
#print("[INFO] Actor initialized with parameters : state_size={} action_size={} seed={} fc1_units={} fc2_units={} bn_mode={}".format(state_size, action_size, seed, fc1_units, fc2_units, bn_mode))
def reset_parameters(self):
self.fc1.weight.data.uniform_(*hidden_init(self.fc1))
self.fc2.weight.data.uniform_(*hidden_init(self.fc2))
self.fc3.weight.data.uniform_(-3e-3, 3e-3)
def forward(self, state):
"""Build an actor (policy) network that maps states -> actions."""
# Reshape the state to comply with Batch Normalization
if state.dim() == 1:
state = torch.unsqueeze(state,0)
if self.bn_mode==0:
# Batch Normalization disabled
x = F.relu(self.fc1(state))
x = F.relu(self.fc2(x))
return F.tanh(self.fc3(x))
elif self.bn_mode==1:
# Batch Normalization before Activation
x = self.fc1(state)
x = self.bn1(x)
x = F.relu(x)
x = self.fc2(x)
x = self.bn2(x)
x = F.relu(x)
x = self.fc3(x)
return F.tanh(x)
elif self.bn_mode==2:
# Batch Normalization after Activation
x = F.relu(self.fc1(state))
x = self.bn1(x)
x = F.relu(self.fc2(x))
return F.tanh(self.fc3(x))
elif self.bn_mode==3:
# Batch Normalization before Activation (alternate version)
x = self.fc1(state)
x = self.bn1(x)
x = F.relu(x)
x = self.fc2(x)
x = self.bn2(x)
x = F.relu(x)
x = self.fc3(x)
x = self.bn3(x)
return F.tanh(x)
elif self.bn_mode==4:
# Batch Normalization after Activation (alternate version)
x = F.relu(self.fc1(state))
x = self.bn1(x)
x = F.relu(self.fc2(x))
x = self.bn2(x)
return F.tanh(self.fc3(x))
class Critic(nn.Module):
"""Critic (Value) Model."""
def __init__(self, state_size, action_size, seed, fcs1_units=400, fc2_units=300, bn_mode=0):
"""Initialize parameters and build model.
Params
======
state_size (int): Dimension of each state
action_size (int): Dimension of each action
seed (int): Random seed
fcs1_units (int): Number of nodes in the first hidden layer
fc2_units (int): Number of nodes in the second hidden layer
bn_mode (int): Use Batch Norm. - 0=disabled, 1=BN before Activation, 2=BN after Activation (3, 4 are alt. versions of 1, 2)
"""
super(Critic, self).__init__()
self.seed = torch.manual_seed(seed)
# Dense layers
self.fcs1 = nn.Linear(state_size, fcs1_units)
self.fc2 = nn.Linear(fcs1_units+action_size, fc2_units)
self.fc3 = nn.Linear(fc2_units, 1)
# Normalization layers
self.bn1 = nn.BatchNorm1d(fcs1_units)
if bn_mode>2:
self.bn2 = nn.BatchNorm1d(fc2_units)
self.bn_mode=bn_mode
self.reset_parameters()
#print("[INFO] CRITIC initialized with parameters : state_size={} action_size={} seed={} fcs1_units={} fc2_units={} bn_mode={}".format(state_size, action_size, seed, fcs1_units, fc2_units, bn_mode))
def reset_parameters(self):
self.fcs1.weight.data.uniform_(*hidden_init(self.fcs1))
self.fc2.weight.data.uniform_(*hidden_init(self.fc2))
self.fc3.weight.data.uniform_(-3e-3, 3e-3)
def forward(self, state, action):
"""Build a critic (value) network that maps (state, action) pairs -> Q-values."""
# Reshape the state to comply with Batch Normalization
if state.dim() == 1:
state = torch.unsqueeze(state,0)
if self.bn_mode==0:
# Batch Normalization disabled
xs = F.relu(self.fcs1(state))
x = torch.cat((xs, action), dim=1)
x = F.relu(self.fc2(x))
return self.fc3(x)
elif self.bn_mode==1:
# Batch Normalization before Activation
xs = self.fcs1(state)
xs = self.bn1(xs)
xs = F.relu(xs)
x = torch.cat((xs, action), dim=1)
x = self.fc2(x)
x = F.relu(x)
return self.fc3(x)
elif self.bn_mode==2:
# Batch Normalization after Activation
xs = F.relu(self.fcs1(state))
xs = self.bn1(xs)
x = torch.cat((xs, action), dim=1)
x = F.relu(self.fc2(x))
return self.fc3(x)
elif self.bn_mode==3:
# Batch Normalization before Activation (alternate version)
xs = self.fcs1(state)
xs = self.bn1(xs)
xs = F.relu(xs)
x = torch.cat((xs, action), dim=1)
x = self.fc2(x)
x = self.bn2(x)
x = F.relu(x)
return self.fc3(x)
elif self.bn_mode==4:
# Batch Normalization after Activation (alternate version)
xs = F.relu(self.fcs1(state))
xs = self.bn1(xs)
x = torch.cat((xs, action), dim=1)
x = F.relu(self.fc2(x))
x = self.bn2(x)
return self.fc3(x)