-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathDynamicResNetBasicBlock.patch
77 lines (73 loc) · 4.2 KB
/
DynamicResNetBasicBlock.patch
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
--- /home/rick/nas_rram/ofa/once-for-all/ofa/imagenet_classification/elastic_nn/modules/dynamic_layers.py
+++ /home/rick/nas_rram/ofa/once-for-all/ofa/imagenet_classification/elastic_nn/modules/dynamic_layers.py
@@ -18,34 +18,28 @@
round(max(self.out_channel_list) * max(self.expand_ratio_list)), MyNetwork.CHANNEL_DIVISIBLE)
self.conv1 = nn.Sequential(OrderedDict([
- # ('conv', DynamicFConv2d(max(self.in_channel_list), max_middle_channel, kernel_size)),
- ('conv', DynamicFConv2d(max(self.in_channel_list), max_middle_channel, stride = stride)),
- # ('conv', DynamicFConv2d(max(self.in_channel_list), max(self.out_channel_list), stride = stride)),
+ ('conv', DynamicFConv2d(max(self.in_channel_list), max_middle_channel, kernel_size)),
+ # ('conv', DynamicFConv2d(max(self.in_channel_list), max_middle_channel, kernel_size, stride)),
('bn', DynamicBatchNorm2d(max_middle_channel)),
('act', build_activation(self.act_func, inplace=True)),
]))
# pad = get_same_padding(self.kernel_size)
self.conv2 = nn.Sequential(OrderedDict([
# ('conv', DynamicConv2d(max_middle_channel, max_middle_channel, kernel_size, stride)),
- # ('conv', DynamicFConv2d(max_middle_channel, max(self.out_channel_list), kernel_size, stride)),
- ('conv', DynamicFConv2d(max_middle_channel, max(self.out_channel_list))),
- # ('conv', DynamicFConv2d( max(self.out_channel_list), max(self.out_channel_list))),
+ ('conv', DynamicFConv2d(max_middle_channel, max(self.out_channel_list), kernel_size, stride)),
('bn', DynamicBatchNorm2d(max(self.out_channel_list))),
]))
if self.stride == 1 and self.in_channel_list == self.out_channel_list:
- # print("IdentityLayer")
self.downsample = IdentityLayer(max(self.in_channel_list), max(self.out_channel_list))
elif self.downsample_mode == 'conv':
- # print("conv")
self.downsample = nn.Sequential(OrderedDict([
# ('conv', DynamicConv2d(max(self.in_channel_list), max(self.out_channel_list), stride=stride)),
('conv', DynamicFConv2d(max(self.in_channel_list), max(self.out_channel_list), stride=stride)),
('bn', DynamicBatchNorm2d(max(self.out_channel_list))),
]))
elif self.downsample_mode == 'avgpool_conv':
- # print("avgpool_conv")
self.downsample = nn.Sequential(OrderedDict([
('avg_pool', nn.AvgPool2d(kernel_size=stride, stride=stride, padding=0, ceil_mode=True)),
# ('avg_pool', nn.AvgPool2d(kernel_size=stride, stride=stride, padding=0, ceil_mode=True)),
@@ -53,7 +47,6 @@
('conv', DynamicFConv2d(max(self.in_channel_list), max(self.out_channel_list))),
('bn', DynamicBatchNorm2d(max(self.out_channel_list))),
]))
-
else:
raise NotImplementedError
@@ -71,19 +64,12 @@
if not isinstance(self.downsample, IdentityLayer):
self.downsample.conv.active_out_channel = self.active_out_channel
# print('x_shape', x.shape)
- # import pdb
- # pdb.set_trace()
residual = self.downsample(x)
- # if (residual.shape == torch.Size([64, 64, 8, 8])):
- # import pdb
- # pdb.set_trace()
# print('residual', residual.shape)
x = self.conv1(x)
# print('x_conv1.shape', x.shape)
x = self.conv2(x)
# print('x_conv2.shape', x.shape)
- # print('max(self.in_channel_list)', max(self.in_channel_list))
- # print('max(self.out_channel_list)', max(self.out_channel_list))
x = x + residual
x = self.final_act(x)
return x
@@ -91,7 +77,7 @@
@property
def module_str(self):
return '(%s, %s)' % (
- '%dx%d_BaseConv_in->%d->%d_S%d' % (
+ '%dx%d_BottleneckConv_in->%d->%d_S%d' % (
self.kernel_size, self.kernel_size, self.active_middle_channels, self.active_out_channel, self.stride
),
'Identity' if isinstance(self.downsample, IdentityLayer) else self.downsample_mode,