-
Notifications
You must be signed in to change notification settings - Fork 3
/
MHRA_uniformer_v1.py
186 lines (156 loc) · 6.92 KB
/
MHRA_uniformer_v1.py
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
# (ref) https://github.com/Sense-X/UniFormer/blob/main/image_classification/models/uniformer.py
""" Multi-Head Relation Aggregator (MHRA) for image classification.
"""
from functools import partial
import torch
import torch.nn as nn
from timm.models.layers import trunc_normal_, DropPath, to_2tuple
layer_scale = False
init_value = 1e-6
# ---------------- #
# == Local MHRA == #
# ---------------- #
class CMlp(nn.Module):
def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):
super().__init__()
out_features = out_features or in_features
hidden_features = hidden_features or in_features
self.fc1 = nn.Conv2d(in_features, hidden_features, 1)
self.act = act_layer()
self.fc2 = nn.Conv2d(hidden_features, out_features, 1)
self.drop = nn.Dropout(drop)
def forward(self, x):
x = self.fc1(x)
x = self.act(x)
x = self.drop(x)
x = self.fc2(x)
x = self.drop(x)
return x
class CBlock(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.pos_embed = nn.Conv2d(dim, dim, 3, padding=1, groups=dim)
self.norm1 = nn.BatchNorm2d(dim)
self.conv1 = nn.Conv2d(dim, dim, 1)
self.conv2 = nn.Conv2d(dim, dim, 1)
self.attn = nn.Conv2d(dim, dim, 5, padding=2, groups=dim)
# -- NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
self.norm2 = nn.BatchNorm2d(dim)
mlp_hidden_dim = int(dim * mlp_ratio)
self.mlp = CMlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
def forward(self, x):
x = x + self.pos_embed(x)
x = x + self.drop_path(self.conv2(self.attn(self.conv1(self.norm1(x)))))
x = x + self.drop_path(self.mlp(self.norm2(x)))
return x
# ----------------- #
# == Global MHRA == #
# ----------------- #
class Mlp(nn.Module):
def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):
super().__init__()
out_features = out_features or in_features
hidden_features = hidden_features or in_features
self.fc1 = nn.Linear(in_features, hidden_features)
self.act = act_layer()
self.fc2 = nn.Linear(hidden_features, out_features)
self.drop = nn.Dropout(drop)
def forward(self, x):
x = self.fc1(x)
x = self.act(x)
x = self.drop(x)
x = self.fc2(x)
x = self.drop(x)
return x
class Attention(nn.Module):
def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0.):
super().__init__()
self.num_heads = num_heads
head_dim = dim // num_heads
# NOTE scale factor was wrong in my original version, can set manually to be compat with prev weights
self.scale = qk_scale or head_dim ** -0.5
self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
self.attn_drop = nn.Dropout(attn_drop)
self.proj = nn.Linear(dim, dim)
self.proj_drop = nn.Dropout(proj_drop)
def forward(self, x):
B, N, C = x.shape
qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
q, k, v = qkv[0], qkv[1], qkv[2] # make torchscript happy (cannot use tensor as tuple)
attn = (q @ k.transpose(-2, -1)) * self.scale
attn = attn.softmax(dim=-1)
attn = self.attn_drop(attn)
x = (attn @ v).transpose(1, 2).reshape(B, N, C)
x = self.proj(x)
x = self.proj_drop(x)
return x
class SABlock(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.pos_embed = nn.Conv2d(dim, dim, 3, padding=1, groups=dim)
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)
# NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
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)
# -- NOTE: layer scale -- #
global layer_scale
self.ls = layer_scale
if self.ls:
global init_value
print(f"Use layer_scale: {layer_scale}, init_values: {init_value}")
self.gamma_1 = nn.Parameter(init_value * torch.ones((dim)),requires_grad=True)
self.gamma_2 = nn.Parameter(init_value * torch.ones((dim)),requires_grad=True)
def forward(self, x):
x = x + self.pos_embed(x)
B, N, H, W = x.shape
x = x.flatten(2).transpose(1, 2)
if self.ls:
x = x + self.drop_path(self.gamma_1 * self.attn(self.norm1(x)))
x = x + self.drop_path(self.gamma_2 * self.mlp(self.norm2(x)))
else:
x = x + self.drop_path(self.attn(self.norm1(x)))
x = x + self.drop_path(self.mlp(self.norm2(x)))
x = x.transpose(1, 2).reshape(B, N, H, W)
return x
if __name__ == "__main__":
head_dim=64
mlp_ratio=4.
qkv_bias=True
qk_scale=None
representation_size=None
drop_rate=0.
attn_drop_rate=0.
drop_path_rate=0.
norm_layer=partial(nn.LayerNorm, eps=1e-6)
# ---
depth=[3, 4, 8, 3]
embed_dim=[64, 128, 320, 512]
num_heads = [dim // head_dim for dim in embed_dim]
dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depth))] # stochastic depth decay rule
# --- local MHRA ---
x = torch.randn(1, embed_dim[0], 56, 56) # input
localMHRA = nn.ModuleList([
CBlock(dim=embed_dim[0], num_heads=num_heads[0], mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer)
for i in range(depth[0])])
for blk in localMHRA:
x = blk(x)
print(x.shape)
# --- global MHRA ---
x = torch.randn(1, embed_dim[2], 14, 14) # input
globalMHRA = nn.ModuleList([
SABlock(dim=embed_dim[2], num_heads=num_heads[2], mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i+depth[0]+depth[1]], norm_layer=norm_layer)
for i in range(depth[2])])
for blk in globalMHRA:
x = blk(x)
print(x.shape)