File size: 3,847 Bytes
5d2263b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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 torch
import torch.nn as nn


def count_params(model):
    total_params = sum(p.numel() for p in model.parameters())
    return total_params


class ActNorm(nn.Module):
    def __init__(self, num_features, logdet=False, affine=True,
                 allow_reverse_init=False):
        assert affine
        super().__init__()
        self.logdet = logdet
        self.loc = nn.Parameter(torch.zeros(1, num_features, 1, 1))
        self.scale = nn.Parameter(torch.ones(1, num_features, 1, 1))
        self.allow_reverse_init = allow_reverse_init

        self.register_buffer('initialized', torch.tensor(0, dtype=torch.uint8))

    def initialize(self, input):
        with torch.no_grad():
            flatten = input.permute(1, 0, 2, 3).contiguous().view(input.shape[1], -1)
            mean = (
                flatten.mean(1)
                .unsqueeze(1)
                .unsqueeze(2)
                .unsqueeze(3)
                .permute(1, 0, 2, 3)
            )
            std = (
                flatten.std(1)
                .unsqueeze(1)
                .unsqueeze(2)
                .unsqueeze(3)
                .permute(1, 0, 2, 3)
            )

            self.loc.data.copy_(-mean)
            self.scale.data.copy_(1 / (std + 1e-6))

    def forward(self, input, reverse=False):
        if reverse:
            return self.reverse(input)
        if len(input.shape) == 2:
            input = input[:,:,None,None]
            squeeze = True
        else:
            squeeze = False

        _, _, height, width = input.shape

        if self.training and self.initialized.item() == 0:
            self.initialize(input)
            self.initialized.fill_(1)

        h = self.scale * (input + self.loc)

        if squeeze:
            h = h.squeeze(-1).squeeze(-1)

        if self.logdet:
            log_abs = torch.log(torch.abs(self.scale))
            logdet = height*width*torch.sum(log_abs)
            logdet = logdet * torch.ones(input.shape[0]).to(input)
            return h, logdet

        return h

    def reverse(self, output):
        if self.training and self.initialized.item() == 0:
            if not self.allow_reverse_init:
                raise RuntimeError(
                    "Initializing ActNorm in reverse direction is "
                    "disabled by default. Use allow_reverse_init=True to enable."
                )
            else:
                self.initialize(output)
                self.initialized.fill_(1)

        if len(output.shape) == 2:
            output = output[:,:,None,None]
            squeeze = True
        else:
            squeeze = False

        h = output / self.scale - self.loc

        if squeeze:
            h = h.squeeze(-1).squeeze(-1)
        return h


class AbstractEncoder(nn.Module):
    def __init__(self):
        super().__init__()

    def encode(self, *args, **kwargs):
        raise NotImplementedError


class Labelator(AbstractEncoder):
    """Net2Net Interface for Class-Conditional Model"""
    def __init__(self, n_classes, quantize_interface=True):
        super().__init__()
        self.n_classes = n_classes
        self.quantize_interface = quantize_interface

    def encode(self, c):
        c = c[:,None]
        if self.quantize_interface:
            return c, None, [None, None, c.long()]
        return c


class SOSProvider(AbstractEncoder):
    # for unconditional training
    def __init__(self, sos_token, quantize_interface=True):
        super().__init__()
        self.sos_token = sos_token
        self.quantize_interface = quantize_interface

    def encode(self, x):
        # get batch size from data and replicate sos_token
        c = torch.ones(x.shape[0], 1)*self.sos_token
        c = c.long().to(x.device)
        if self.quantize_interface:
            return c, None, [None, None, c]
        return c