File size: 5,731 Bytes
d4ebf73
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
#!/usr/bin/env python3
# encoding: utf-8
# @Time    : 2018/9/28 下午12:13
# @Author  : yuchangqian
# @Contact : [email protected]
# @File    : init_func.py.py
import math
import warnings
import torch
import torch.nn as nn
from utils.seg_opr.conv_2_5d import Conv2_5D_depth, Conv2_5D_disp


def __init_weight(feature, conv_init, norm_layer, bn_eps, bn_momentum,
                  **kwargs):
    for name, m in feature.named_modules():
        if isinstance(m, (nn.Conv1d, nn.Conv2d, nn.Conv3d)):
            conv_init(m.weight, **kwargs)
        elif isinstance(m, Conv2_5D_depth):
            conv_init(m.weight_0, **kwargs)
            conv_init(m.weight_1, **kwargs)
            conv_init(m.weight_2, **kwargs)
        elif isinstance(m, Conv2_5D_disp):
            conv_init(m.weight_0, **kwargs)
            conv_init(m.weight_1, **kwargs)
            conv_init(m.weight_2, **kwargs)
        elif isinstance(m, norm_layer):
            m.eps = bn_eps
            m.momentum = bn_momentum
            nn.init.constant_(m.weight, 1)
            nn.init.constant_(m.bias, 0)


def init_weight(module_list, conv_init, norm_layer, bn_eps, bn_momentum,
                **kwargs):
    if isinstance(module_list, list):
        for feature in module_list:
            __init_weight(feature, conv_init, norm_layer, bn_eps, bn_momentum,
                          **kwargs)
    else:
        __init_weight(module_list, conv_init, norm_layer, bn_eps, bn_momentum,
                      **kwargs)


def group_weight(weight_group, module, norm_layer, lr):
    group_decay = []
    group_no_decay = []
    for m in module.modules():
        if isinstance(m, nn.Linear):
            group_decay.append(m.weight)
            if m.bias is not None:
                group_no_decay.append(m.bias)
        elif isinstance(m, (nn.Conv1d, nn.Conv2d, nn.Conv3d, nn.ConvTranspose2d, nn.ConvTranspose3d)):
            group_decay.append(m.weight)
            if m.bias is not None:
                group_no_decay.append(m.bias)
        elif isinstance(m, Conv2_5D_depth):
            group_decay.append(m.weight_0)
            group_decay.append(m.weight_1)
            group_decay.append(m.weight_2)
            if m.bias is not  None:
                group_no_decay.append(m.bias)
        elif isinstance(m, Conv2_5D_disp):
            group_decay.append(m.weight_0)
            group_decay.append(m.weight_1)
            group_decay.append(m.weight_2)
            if m.bias is not  None:
                group_no_decay.append(m.bias)
        elif isinstance(m, norm_layer) or isinstance(m, nn.BatchNorm1d) or isinstance(m, nn.BatchNorm2d) \
                or isinstance(m, nn.BatchNorm3d) or isinstance(m, nn.GroupNorm):
            if m.weight is not None:
                group_no_decay.append(m.weight)
            if m.bias is not None:
                group_no_decay.append(m.bias)
        elif isinstance(m, nn.Parameter):
            group_decay.append(m)
        elif isinstance(m, nn.Embedding):
            group_decay.append(m)
        # else:
        #     print(m, norm_layer)
    # print(module.modules)
    # print( len(list(module.parameters())) , 'HHHHHHHHHHHHHHHHH',  len(group_decay) + len(
    #    group_no_decay))
    assert len(list(module.parameters())) == len(group_decay) + len(
       group_no_decay)
    weight_group.append(dict(params=group_decay, lr=lr))
    weight_group.append(dict(params=group_no_decay, weight_decay=.0, lr=lr))
    return weight_group

def _no_grad_trunc_normal_(tensor, mean, std, a, b):
    # Cut & paste from PyTorch official master until it's in a few official releases - RW
    # Method based on https://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf
    def norm_cdf(x):
        # Computes standard normal cumulative distribution function
        return (1. + math.erf(x / math.sqrt(2.))) / 2.

    if (mean < a - 2 * std) or (mean > b + 2 * std):
        warnings.warn("mean is more than 2 std from [a, b] in nn.init.trunc_normal_. "
                      "The distribution of values may be incorrect.",
                      stacklevel=2)

    with torch.no_grad():
        # Values are generated by using a truncated uniform distribution and
        # then using the inverse CDF for the normal distribution.
        # Get upper and lower cdf values
        l = norm_cdf((a - mean) / std)
        u = norm_cdf((b - mean) / std)

        # Uniformly fill tensor with values from [l, u], then translate to
        # [2l-1, 2u-1].
        tensor.uniform_(2 * l - 1, 2 * u - 1)

        # Use inverse cdf transform for normal distribution to get truncated
        # standard normal
        tensor.erfinv_()

        # Transform to proper mean, std
        tensor.mul_(std * math.sqrt(2.))
        tensor.add_(mean)

        # Clamp to ensure it's in the proper range
        tensor.clamp_(min=a, max=b)
        return tensor


def trunc_normal_(tensor, mean=0., std=1., a=-2., b=2.):
    r"""Fills the input Tensor with values drawn from a truncated
    normal distribution. The values are effectively drawn from the
    normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)`
    with values outside :math:`[a, b]` redrawn until they are within
    the bounds. The method used for generating the random values works
    best when :math:`a \leq \text{mean} \leq b`.
    Args:
        tensor: an n-dimensional `torch.Tensor`
        mean: the mean of the normal distribution
        std: the standard deviation of the normal distribution
        a: the minimum cutoff value
        b: the maximum cutoff value
    Examples:
        >>> w = torch.empty(3, 5)
        >>> nn.init.trunc_normal_(w)
    """
    return _no_grad_trunc_normal_(tensor, mean, std, a, b)