first commit

Branch_CIFAR_10/recover/models/ShuffleNetv1.py

@@ -0,0 +1,138 @@
+'''ShuffleNet in PyTorch.
+See the paper "ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices" for more details.
+'''
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class ShuffleBlock(nn.Module):
+    def __init__(self, groups):
+        super(ShuffleBlock, self).__init__()
+        self.groups = groups
+
+    def forward(self, x):
+        '''Channel shuffle: [N,C,H,W] -> [N,g,C/g,H,W] -> [N,C/g,g,H,w] -> [N,C,H,W]'''
+        N,C,H,W = x.size()
+        g = self.groups
+        return x.view(N,g,C//g,H,W).permute(0,2,1,3,4).reshape(N,C,H,W)
+
+
+class Bottleneck(nn.Module):
+    def __init__(self, in_planes, out_planes, stride, groups, is_last=False):
+        super(Bottleneck, self).__init__()
+        self.is_last = is_last
+        self.stride = stride
+
+        mid_planes = int(out_planes/4)
+        g = 1 if in_planes == 24 else groups
+        self.conv1 = nn.Conv2d(in_planes, mid_planes, kernel_size=1, groups=g, bias=False)
+        self.bn1 = nn.BatchNorm2d(mid_planes)
+        self.shuffle1 = ShuffleBlock(groups=g)
+        self.conv2 = nn.Conv2d(mid_planes, mid_planes, kernel_size=3, stride=stride, padding=1, groups=mid_planes, bias=False)
+        self.bn2 = nn.BatchNorm2d(mid_planes)
+        self.conv3 = nn.Conv2d(mid_planes, out_planes, kernel_size=1, groups=groups, bias=False)
+        self.bn3 = nn.BatchNorm2d(out_planes)
+
+        self.shortcut = nn.Sequential()
+        if stride == 2:
+            self.shortcut = nn.Sequential(nn.AvgPool2d(3, stride=2, padding=1))
+
+    def forward(self, x):
+        out = F.relu(self.bn1(self.conv1(x)))
+        out = self.shuffle1(out)
+        out = F.relu(self.bn2(self.conv2(out)))
+        out = self.bn3(self.conv3(out))
+        res = self.shortcut(x)
+        preact = torch.cat([out, res], 1) if self.stride == 2 else out+res
+        out = F.relu(preact)
+        # out = F.relu(torch.cat([out, res], 1)) if self.stride == 2 else F.relu(out+res)
+        if self.is_last:
+            return out, preact
+        else:
+            return out
+
+
+class ShuffleNet(nn.Module):
+    def __init__(self, cfg, num_classes=10):
+        super(ShuffleNet, self).__init__()
+        out_planes = cfg['out_planes']
+        num_blocks = cfg['num_blocks']
+        groups = cfg['groups']
+
+        self.conv1 = nn.Conv2d(3, 24, kernel_size=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(24)
+        self.in_planes = 24
+        self.layer1 = self._make_layer(out_planes[0], num_blocks[0], groups)
+        self.layer2 = self._make_layer(out_planes[1], num_blocks[1], groups)
+        self.layer3 = self._make_layer(out_planes[2], num_blocks[2], groups)
+        self.linear = nn.Linear(out_planes[2], num_classes)
+
+    def _make_layer(self, out_planes, num_blocks, groups):
+        layers = []
+        for i in range(num_blocks):
+            stride = 2 if i == 0 else 1
+            cat_planes = self.in_planes if i == 0 else 0
+            layers.append(Bottleneck(self.in_planes, out_planes-cat_planes,
+                                     stride=stride,
+                                     groups=groups,
+                                     is_last=(i == num_blocks - 1)))
+            self.in_planes = out_planes
+        return nn.Sequential(*layers)
+
+    def get_feat_modules(self):
+        feat_m = nn.ModuleList([])
+        feat_m.append(self.conv1)
+        feat_m.append(self.bn1)
+        feat_m.append(self.layer1)
+        feat_m.append(self.layer2)
+        feat_m.append(self.layer3)
+        return feat_m
+
+    def get_bn_before_relu(self):
+        raise NotImplementedError('ShuffleNet currently is not supported for "Overhaul" teacher')
+
+    def forward(self, x, is_feat=False, preact=False):
+        out = F.relu(self.bn1(self.conv1(x)))
+        f0 = out
+        out, f1_pre = self.layer1(out)
+        f1 = out
+        out, f2_pre = self.layer2(out)
+        f2 = out
+        out, f3_pre = self.layer3(out)
+        f3 = out
+        out = F.avg_pool2d(out, 4)
+        out = out.view(out.size(0), -1)
+        f4 = out
+        out = self.linear(out)
+
+        if is_feat:
+            if preact:
+                return [f0, f1_pre, f2_pre, f3_pre, f4], out
+            else:
+                return [f0, f1, f2, f3, f4], out
+        else:
+            return out
+
+
+def ShuffleV1(**kwargs):
+    cfg = {
+        'out_planes': [240, 480, 960],
+        'num_blocks': [4, 8, 4],
+        'groups': 3
+    }
+    return ShuffleNet(cfg, **kwargs)
+
+
+if __name__ == '__main__':
+
+    x = torch.randn(2, 3, 32, 32)
+    net = ShuffleV1(num_classes=100)
+    import time
+    a = time.time()
+    feats, logit = net(x, is_feat=True, preact=True)
+    b = time.time()
+    print(b - a)
+    for f in feats:
+        print(f.shape, f.min().item())
+    print(logit.shape)

Branch_CIFAR_10/recover/models/ShuffleNetv2.py

@@ -0,0 +1,218 @@
+'''ShuffleNetV2 in PyTorch.
+See the paper "ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" for more details.
+'''
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class ShuffleBlock(nn.Module):
+    def __init__(self, groups=2):
+        super(ShuffleBlock, self).__init__()
+        self.groups = groups
+
+    def forward(self, x):
+        '''Channel shuffle: [N,C,H,W] -> [N,g,C/g,H,W] -> [N,C/g,g,H,w] -> [N,C,H,W]'''
+        N, C, H, W = x.size()
+        g = self.groups
+        return x.view(N, g, C//g, H, W).permute(0, 2, 1, 3, 4).reshape(N, C, H, W)
+
+
+class SplitBlock(nn.Module):
+    def __init__(self, ratio):
+        super(SplitBlock, self).__init__()
+        self.ratio = ratio
+
+    def forward(self, x):
+        c = int(x.size(1) * self.ratio)
+        return x[:, :c, :, :], x[:, c:, :, :]
+
+
+class BasicBlock(nn.Module):
+    def __init__(self, in_channels, split_ratio=0.5, is_last=False):
+        super(BasicBlock, self).__init__()
+        self.is_last = is_last
+        self.split = SplitBlock(split_ratio)
+        in_channels = int(in_channels * split_ratio)
+        self.conv1 = nn.Conv2d(in_channels, in_channels,
+                               kernel_size=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(in_channels)
+        self.conv2 = nn.Conv2d(in_channels, in_channels,
+                               kernel_size=3, stride=1, padding=1, groups=in_channels, bias=False)
+        self.bn2 = nn.BatchNorm2d(in_channels)
+        self.conv3 = nn.Conv2d(in_channels, in_channels,
+                               kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(in_channels)
+        self.shuffle = ShuffleBlock()
+
+    def forward(self, x):
+        x1, x2 = self.split(x)
+        out = F.relu(self.bn1(self.conv1(x2)))
+        out = self.bn2(self.conv2(out))
+        preact = self.bn3(self.conv3(out))
+        out = F.relu(preact)
+        # out = F.relu(self.bn3(self.conv3(out)))
+        preact = torch.cat([x1, preact], 1)
+        out = torch.cat([x1, out], 1)
+        out = self.shuffle(out)
+        if self.is_last:
+            return out, preact
+        else:
+            return out
+
+
+class DownBlock(nn.Module):
+    def __init__(self, in_channels, out_channels):
+        super(DownBlock, self).__init__()
+        mid_channels = out_channels // 2
+        # left
+        self.conv1 = nn.Conv2d(in_channels, in_channels,
+                               kernel_size=3, stride=2, padding=1, groups=in_channels, bias=False)
+        self.bn1 = nn.BatchNorm2d(in_channels)
+        self.conv2 = nn.Conv2d(in_channels, mid_channels,
+                               kernel_size=1, bias=False)
+        self.bn2 = nn.BatchNorm2d(mid_channels)
+        # right
+        self.conv3 = nn.Conv2d(in_channels, mid_channels,
+                               kernel_size=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(mid_channels)
+        self.conv4 = nn.Conv2d(mid_channels, mid_channels,
+                               kernel_size=3, stride=2, padding=1, groups=mid_channels, bias=False)
+        self.bn4 = nn.BatchNorm2d(mid_channels)
+        self.conv5 = nn.Conv2d(mid_channels, mid_channels,
+                               kernel_size=1, bias=False)
+        self.bn5 = nn.BatchNorm2d(mid_channels)
+
+        self.shuffle = ShuffleBlock()
+
+    def forward(self, x):
+        # left
+        out1 = self.bn1(self.conv1(x))
+        out1 = F.relu(self.bn2(self.conv2(out1)))
+        # right
+        out2 = F.relu(self.bn3(self.conv3(x)))
+        out2 = self.bn4(self.conv4(out2))
+        out2 = F.relu(self.bn5(self.conv5(out2)))
+        # concat
+        out = torch.cat([out1, out2], 1)
+        out = self.shuffle(out)
+        return out
+
+
+class ShuffleNetV2(nn.Module):
+    def __init__(self, net_size, num_classes=10, img_size=32):
+        super(ShuffleNetV2, self).__init__()
+        out_channels = configs[net_size]['out_channels']
+        num_blocks = configs[net_size]['num_blocks']
+
+        # self.conv1 = nn.Conv2d(3, 24, kernel_size=3,
+        #                        stride=1, padding=1, bias=False)
+        self.conv1 = nn.Conv2d(3, 24, kernel_size=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(24)
+        self.in_channels = 24
+        self.layer1 = self._make_layer(out_channels[0], num_blocks[0])
+        self.layer2 = self._make_layer(out_channels[1], num_blocks[1])
+        self.layer3 = self._make_layer(out_channels[2], num_blocks[2])
+        self.conv2 = nn.Conv2d(out_channels[2], out_channels[3],
+                               kernel_size=1, stride=1, padding=0, bias=False)
+        self.bn2 = nn.BatchNorm2d(out_channels[3])
+        self.linear = nn.Linear(out_channels[3], num_classes)
+
+    def _make_layer(self, out_channels, num_blocks):
+        layers = [DownBlock(self.in_channels, out_channels)]
+        for i in range(num_blocks):
+            layers.append(BasicBlock(out_channels, is_last=(i == num_blocks - 1)))
+            self.in_channels = out_channels
+        return nn.Sequential(*layers)
+
+    def get_feat_modules(self):
+        feat_m = nn.ModuleList([])
+        feat_m.append(self.conv1)
+        feat_m.append(self.bn1)
+        feat_m.append(self.layer1)
+        feat_m.append(self.layer2)
+        feat_m.append(self.layer3)
+        return feat_m
+
+    def get_bn_before_relu(self):
+        raise NotImplementedError('ShuffleNetV2 currently is not supported for "Overhaul" teacher')
+
+    def forward(self, x, is_feat=False, preact=False):
+        out = F.relu(self.bn1(self.conv1(x)))
+        # out = F.max_pool2d(out, 3, stride=2, padding=1)
+        f0 = out
+        out, f1_pre = self.layer1(out)
+        f1 = out
+        out, f2_pre = self.layer2(out)
+        f2 = out
+        out, f3_pre = self.layer3(out)
+        f3 = out
+        out = F.relu(self.bn2(self.conv2(out)))
+        # out = F.avg_pool2d(out, 4)
+        out = F.avg_pool2d(out, out.size(3))
+        out = out.view(out.size(0), -1)
+        f4 = out
+        out = self.linear(out)
+        if is_feat:
+            if preact:
+                return [f0, f1_pre, f2_pre, f3_pre, f4], out
+            else:
+                return [f0, f1, f2, f3, f4], out
+        else:
+            return out
+
+
+configs = {
+    0.2: {
+        'out_channels': (40, 80, 160, 512),
+        'num_blocks': (3, 3, 3)
+    },
+
+    0.3: {
+        'out_channels': (40, 80, 160, 512),
+        'num_blocks': (3, 7, 3)
+    },
+
+    0.5: {
+        'out_channels': (48, 96, 192, 1024),
+        'num_blocks': (3, 7, 3)
+    },
+
+    1: {
+        'out_channels': (116, 232, 464, 1024),
+        'num_blocks': (3, 7, 3)
+    },
+    1.5: {
+        'out_channels': (176, 352, 704, 1024),
+        'num_blocks': (3, 7, 3)
+    },
+    2: {
+        'out_channels': (224, 488, 976, 2048),
+        'num_blocks': (3, 7, 3)
+    }
+}
+
+
+def ShuffleV2(**kwargs):
+    model = ShuffleNetV2(net_size=1, **kwargs)
+    return model
+
+def ShuffleV2_0_3(**kwargs):
+    model = ShuffleNetV2(net_size=0.3, **kwargs)
+    return model
+
+def ShuffleV2_0_5(**kwargs):
+    model = ShuffleNetV2(net_size=0.5, **kwargs)
+    return model
+
+if __name__ == '__main__':
+    net = ShuffleV2(num_classes=100)
+    x = torch.randn(3, 3, 32, 32)
+    import time
+    a = time.time()
+    feats, logit = net(x, is_feat=True, preact=True)
+    b = time.time()
+    print(b - a)
+    for f in feats:
+        print(f.shape, f.min().item())
+    print(logit.shape)