Spaces:
Running
on
Zero
Running
on
Zero
File size: 6,440 Bytes
11e6f7b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 |
# https://github.com/sxyu/pixel-nerf/blob/master/src/model/resnetfc.py
from torch import nn
import torch
from vit.vision_transformer import Mlp, DropPath
# Resnet Blocks
class ResnetBlockFC(nn.Module):
"""
Fully connected ResNet Block class.
Taken from DVR code.
:param size_in (int): input dimension
:param size_out (int): output dimension
:param size_h (int): hidden dimension
"""
def __init__(self, size_in, size_out=None, size_h=None, beta=0.0, init_as_zero=False):
super().__init__()
# Attributes
if size_out is None:
size_out = size_in
if size_h is None:
size_h = min(size_in, size_out)
self.size_in = size_in
self.size_h = size_h
self.size_out = size_out
# Submodules
self.fc_0 = nn.Linear(size_in, size_h)
self.fc_1 = nn.Linear(size_h, size_out)
# Init
nn.init.constant_(self.fc_0.bias, 0.0)
if init_as_zero:
nn.init.zeros_(self.fc_0.weight)
else:
nn.init.kaiming_normal_(self.fc_0.weight, a=0, mode="fan_in")
nn.init.constant_(self.fc_1.bias, 0.0)
nn.init.zeros_(self.fc_1.weight)
if beta > 0:
self.activation = nn.Softplus(beta=beta)
else:
self.activation = nn.ReLU()
if size_in == size_out:
self.shortcut = None
else:
self.shortcut = nn.Linear(size_in, size_out, bias=False)
# nn.init.constant_(self.shortcut.bias, 0.0)
nn.init.kaiming_normal_(self.shortcut.weight, a=0, mode="fan_in")
def forward(self, x):
# with profiler.record_function("resblock"):
net = self.fc_0(self.activation(x))
dx = self.fc_1(self.activation(net))
if self.shortcut is not None:
x_s = self.shortcut(x)
else:
x_s = x
return x_s + dx
# Resnet Blocks
class ResnetBlockFCViT(nn.Module):
"""
Fully connected ResNet Block class.
Taken from DVR code.
:param size_in (int): input dimension
:param size_out (int): output dimension
:param size_h (int): hidden dimension
"""
def __init__(self, size_in, size_out=None, size_h=None, beta=0.0, init_as_zero=False):
super().__init__()
# Attributes
if size_out is None:
size_out = size_in
if size_h is None:
size_h = min(size_in, size_out)
self.size_in = size_in
self.size_h = size_h
self.size_out = size_out
# Submodules
self.fc_0 = nn.Linear(size_in, size_h)
self.fc_1 = nn.Linear(size_h, size_out)
# Init
nn.init.constant_(self.fc_0.bias, 0.0)
if init_as_zero:
nn.init.zeros_(self.fc_0.weight)
else:
nn.init.kaiming_normal_(self.fc_0.weight, a=0, mode="fan_in")
nn.init.constant_(self.fc_1.bias, 0.0)
nn.init.zeros_(self.fc_1.weight)
if beta > 0:
self.activation = nn.Softplus(beta=beta)
else:
self.activation = nn.ReLU()
if size_in == size_out:
self.shortcut = None
else:
self.shortcut = nn.Linear(size_in, size_out, bias=False)
# nn.init.constant_(self.shortcut.bias, 0.0)
nn.init.kaiming_normal_(self.shortcut.weight, a=0, mode="fan_in")
def forward(self, x):
# with profiler.record_function("resblock"):
net = self.fc_0(self.activation(x))
dx = self.fc_1(self.activation(net))
if self.shortcut is not None:
x_s = self.shortcut(x)
else:
x_s = x
return x_s + dx
# class Block(nn.Module):
# def __init__(self,
# dim,
# num_heads,
# mlp_ratio=4.,
# qkv_bias=False,
# qk_scale=None,
# drop=0.,
# attn_drop=0.,
# drop_path=0.,
# act_layer=nn.GELU,
# norm_layer=nn.LayerNorm):
# super().__init__()
# self.norm1 = norm_layer(dim)
# self.attn = Attention(dim,
# num_heads=num_heads,
# qkv_bias=qkv_bias,
# qk_scale=qk_scale,
# attn_drop=attn_drop,
# proj_drop=drop)
# self.drop_path = DropPath(
# drop_path) if drop_path > 0. else nn.Identity()
# self.norm2 = norm_layer(dim)
# mlp_hidden_dim = int(dim * mlp_ratio)
# self.mlp = Mlp(in_features=dim,
# hidden_features=mlp_hidden_dim,
# act_layer=act_layer,
# drop=drop)
# def forward(self, x, return_attention=False):
# y, attn = self.attn(self.norm1(x))
# if return_attention:
# return attn
# x = x + self.drop_path(y)
# x = x + self.drop_path(self.mlp(self.norm2(x)))
# return x
class ResMlp(nn.Module):
def __init__(self,
size_in,
size_out=None,
size_h=None,
drop=0.,
drop_path=0.,
act_layer=nn.GELU,
norm_layer=nn.LayerNorm,
):
super().__init__()
# Attributes
if size_out is None:
size_out = size_in
if size_h is None:
size_h = min(size_in, size_out)
self.size_in = size_in
self.size_h = size_h
self.size_out = size_out
# Submodules
self.norm1 = norm_layer(size_in) # ? how to use
self.mlp = Mlp(in_features=size_in,
out_features=size_out,
act_layer=act_layer,
drop=drop)
# Residual shortcuts
if size_in == size_out:
self.shortcut = None
else:
self.shortcut = nn.Linear(size_in, size_out, bias=False)
self.norm2 = norm_layer(size_in)
self.drop_path = DropPath(
drop_path) if drop_path > 0. else nn.Identity()
def forward(self, x):
dx = self.mlp(self.norm1(x))
if self.shortcut is not None:
x_s = self.shortcut(self.norm2(x))
else:
x_s = x
return x_s + self.drop_path(dx) |