WPGNN/modules.py at main · NatLabRockies/WPGNN · GitHub

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
import tensorflow as tf
import sonnet as snt

class EdgeUpdate(snt.Module):

    def __init__(self, input_size, output_size, layer_sizes=None, output_activation=False,
                       w_init=None, b_init=None, name=None):
        super(EdgeUpdate, self).__init__(name=name)
        self.input_size, self.output_size = input_size, output_size
        self.output_activation = output_activation

        layer_sizes = [output_size] if layer_sizes is None else layer_sizes
        self.n_layers = len(layer_sizes)

        self.layers = []
        for i in range(self.n_layers):
            self.layers.append(snt.Linear(layer_sizes[i],
                                          w_init=w_init,
                                          b_init=b_init,
                                          name='linear{0:03d}'.format(i)))
        self.layers.append(snt.Linear(output_size,
                                      w_init=w_init,
                                      b_init=b_init,
                                      name='linear_out'))

    def __call__(self, x):
        for layer in self.layers[:-1]:
            x = layer(x)
            x = tf.nn.leaky_relu(x)

        x = self.layers[-1](x)
        if self.output_activation == 'leaky_relu':
            x = tf.nn.leaky_relu(x)
        elif self.output_activation == 'relu':
            x = tf.nn.relu(x)
        elif self.output_activation == 'softplus':
            x = tf.nn.softplus(x)
        elif self.output_activation == 'sigmoid':
            x = tf.nn.sigmoid(x)
        elif self.output_activation == 'none':
            pass
        else:
            assert self.output_activation == False

        return x

class NodeUpdate(snt.Module):

    def __init__(self, input_size, output_size, layer_sizes=None, output_activation=False,
                       w_init=None, b_init=None, name=None):
        super(NodeUpdate, self).__init__(name=name)
        self.input_size, self.output_size = input_size, output_size
        self.output_activation = output_activation

        layer_sizes = [output_size] if layer_sizes is None else layer_sizes
        self.n_layers = len(layer_sizes)

        self.layers = []
        for i in range(self.n_layers):
            self.layers.append(snt.Linear(layer_sizes[i],
                                          w_init=w_init,
                                          b_init=b_init,
                                          name='linear{0:03d}'.format(i)))
        self.layers.append(snt.Linear(output_size,
                                      w_init=w_init,
                                      b_init=b_init,
                                      name='linear_out'))

    def __call__(self, x):
        for layer in self.layers[:-1]:
            x = layer(x)
            x = tf.nn.leaky_relu(x)

        x = self.layers[-1](x)
        if self.output_activation == 'leaky_relu':
            x = tf.nn.leaky_relu(x)
        elif self.output_activation == 'relu':
            x = tf.nn.relu(x)
        elif self.output_activation == 'softplus':
            x = tf.nn.softplus(x)
        elif self.output_activation == 'sigmoid':
            x = tf.nn.sigmoid(x)
        elif self.output_activation == 'none':
            pass
        else:
            assert self.output_activation == False

        return x

class GlobalUpdate(snt.Module):

    def __init__(self, input_size, output_size, layer_sizes=None, output_activation=False,
                       w_init=None, b_init=None, name=None):
        super(GlobalUpdate, self).__init__(name=name)
        self.input_size, self.output_size = input_size, output_size
        self.output_activation = output_activation

        layer_sizes = [output_size] if layer_sizes is None else layer_sizes
        self.n_layers = len(layer_sizes)

        self.layers = []
        for i in range(self.n_layers):
            self.layers.append(snt.Linear(layer_sizes[i],
                                          w_init=w_init,
                                          b_init=b_init,
                                          name='linear{0:03d}'.format(i)))
        self.layers.append(snt.Linear(output_size,
                                      w_init=w_init,
                                      b_init=b_init,
                                      name='linear_out'))

    def __call__(self, x):
        for layer in self.layers[:-1]:
            x = layer(x)
            x = tf.nn.leaky_relu(x)

        x = self.layers[-1](x)
        if self.output_activation == 'leaky_relu':
            x = tf.nn.leaky_relu(x)
        elif self.output_activation == 'relu':
            x = tf.nn.relu(x)
        elif self.output_activation == 'softplus':
            x = tf.nn.softplus(x)
        elif self.output_activation == 'sigmoid':
            x = tf.nn.sigmoid(x)
        elif self.output_activation == 'none':
            pass
        else:
            assert self.output_activation == False

        return x