File size: 6,174 Bytes
079c32c |
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 |
from typing import Union
import torch
import torch.nn as nn
from .nn_module import conv2d_block, fc_block
class ResBlock(nn.Module):
"""
Overview:
Residual Block with 2D convolution layers, including 3 types:
basic block:
input channel: C
x -> 3*3*C -> norm -> act -> 3*3*C -> norm -> act -> out
\__________________________________________/+
bottleneck block:
x -> 1*1*(1/4*C) -> norm -> act -> 3*3*(1/4*C) -> norm -> act -> 1*1*C -> norm -> act -> out
\_____________________________________________________________________________/+
downsample block: used in EfficientZero
input channel: C
x -> 3*3*C -> norm -> act -> 3*3*C -> norm -> act -> out
\__________________ 3*3*C ____________________/+
For more details, please refer to `Deep Residual Learning for Image Recognition
<https://arxiv.org/abs/1512.03385>`_.
Interfaces:
``__init__``, ``forward``
"""
def __init__(
self,
in_channels: int,
activation: nn.Module = nn.ReLU(),
norm_type: str = 'BN',
res_type: str = 'basic',
bias: bool = True,
out_channels: Union[int, None] = None,
) -> None:
"""
Overview:
Init the 2D convolution residual block.
Arguments:
- in_channels (:obj:`int`): Number of channels in the input tensor.
- activation (:obj:`nn.Module`): The optional activation function.
- norm_type (:obj:`str`): Type of the normalization, default set to 'BN'(Batch Normalization), \
supports ['BN', 'LN', 'IN', 'GN', 'SyncBN', None].
- res_type (:obj:`str`): Type of residual block, supports ['basic', 'bottleneck', 'downsample']
- bias (:obj:`bool`): Whether to add a learnable bias to the conv2d_block. default set to True.
- out_channels (:obj:`int`): Number of channels in the output tensor, default set to None, \
which means out_channels = in_channels.
"""
super(ResBlock, self).__init__()
self.act = activation
assert res_type in ['basic', 'bottleneck',
'downsample'], 'residual type only support basic and bottleneck, not:{}'.format(res_type)
self.res_type = res_type
if out_channels is None:
out_channels = in_channels
if self.res_type == 'basic':
self.conv1 = conv2d_block(
in_channels, out_channels, 3, 1, 1, activation=self.act, norm_type=norm_type, bias=bias
)
self.conv2 = conv2d_block(
out_channels, out_channels, 3, 1, 1, activation=None, norm_type=norm_type, bias=bias
)
elif self.res_type == 'bottleneck':
self.conv1 = conv2d_block(
in_channels, out_channels, 1, 1, 0, activation=self.act, norm_type=norm_type, bias=bias
)
self.conv2 = conv2d_block(
out_channels, out_channels, 3, 1, 1, activation=self.act, norm_type=norm_type, bias=bias
)
self.conv3 = conv2d_block(
out_channels, out_channels, 1, 1, 0, activation=None, norm_type=norm_type, bias=bias
)
elif self.res_type == 'downsample':
self.conv1 = conv2d_block(
in_channels, out_channels, 3, 2, 1, activation=self.act, norm_type=norm_type, bias=bias
)
self.conv2 = conv2d_block(
out_channels, out_channels, 3, 1, 1, activation=None, norm_type=norm_type, bias=bias
)
self.conv3 = conv2d_block(in_channels, out_channels, 3, 2, 1, activation=None, norm_type=None, bias=bias)
def forward(self, x: torch.Tensor) -> torch.Tensor:
"""
Overview:
Return the redisual block output.
Arguments:
- x (:obj:`torch.Tensor`): The input tensor.
Returns:
- x (:obj:`torch.Tensor`): The resblock output tensor.
"""
identity = x
x = self.conv1(x)
x = self.conv2(x)
if self.res_type == 'bottleneck':
x = self.conv3(x)
elif self.res_type == 'downsample':
identity = self.conv3(identity)
x = self.act(x + identity)
return x
class ResFCBlock(nn.Module):
"""
Overview:
Residual Block with 2 fully connected layers.
x -> fc1 -> norm -> act -> fc2 -> norm -> act -> out
\_____________________________________/+
Interfaces:
``__init__``, ``forward``
"""
def __init__(
self, in_channels: int, activation: nn.Module = nn.ReLU(), norm_type: str = 'BN', dropout: float = None
):
"""
Overview:
Init the fully connected layer residual block.
Arguments:
- in_channels (:obj:`int`): The number of channels in the input tensor.
- activation (:obj:`nn.Module`): The optional activation function.
- norm_type (:obj:`str`): The type of the normalization, default set to 'BN'.
- dropout (:obj:`float`): The dropout rate, default set to None.
"""
super(ResFCBlock, self).__init__()
self.act = activation
if dropout is not None:
self.dropout = nn.Dropout(dropout)
else:
self.dropout = None
self.fc1 = fc_block(in_channels, in_channels, activation=self.act, norm_type=norm_type)
self.fc2 = fc_block(in_channels, in_channels, activation=None, norm_type=norm_type)
def forward(self, x: torch.Tensor) -> torch.Tensor:
"""
Overview:
Return the output of the redisual block.
Arguments:
- x (:obj:`torch.Tensor`): The input tensor.
Returns:
- x (:obj:`torch.Tensor`): The resblock output tensor.
"""
identity = x
x = self.fc1(x)
x = self.fc2(x)
x = self.act(x + identity)
if self.dropout is not None:
x = self.dropout(x)
return x
|