AttributeError: 'ResNet50' object has no attribute 'conv1'

vision
1

Basically I am trying to implement different learning rate for different layers in my modified ResNet50. My code below.

class ResNet50(nn.Module):
    def __init__(self, pretrained):
        super(ResNet50, self).__init__()
        if pretrained is True:
            self.model = models.resnet50(pretrained="imagenet")
        else:
            self.model = models.resnet50(pretrained=None)

        self.l0 = nn.Linear(2048, 2) 
        self.l1 = nn.Linear(2048, 2)  
        self.l2 = nn.Linear(2048, 2)  
  

    def forward(self, x):
        bs, _, _, _ = x.shape
        x = self.model.features(x)
        x = F.adaptive_avg_pool2d(x, 1).reshape(bs, -1)
        l0  = self.l0(x)
        l1  = self.l1(x)
        l2  = self.l2(x)

        return l0, l1, l2

model = ResNet50(pretrained=True)

for name, param in model.named_parameters():
    if param.requires_grad:
        print(name)

The output looks like:

model.conv1.weight
model.bn1.weight
model.bn1.bias
model.layer1.0.conv1.weight
model.layer1.0.bn1.weight
model.layer1.0.bn1.bias
model.layer1.0.conv2.weight
model.layer1.0.bn2.weight
model.layer1.0.bn2.bias
model.layer1.0.conv3.weight
model.layer1.0.bn3.weight
model.layer1.0.bn3.bias
model.layer1.0.downsample.0.weight
model.layer1.0.downsample.1.weight
model.layer1.0.downsample.1.bias
model.layer1.1.conv1.weight
model.layer1.1.bn1.weight
model.layer1.1.bn1.bias
model.layer1.1.conv2.weight
model.layer1.1.bn2.weight
model.layer1.1.bn2.bias
model.layer1.1.conv3.weight
model.layer1.1.bn3.weight
model.layer1.1.bn3.bias
model.layer1.2.conv1.weight
model.layer1.2.bn1.weight
model.layer1.2.bn1.bias
model.layer1.2.conv2.weight
model.layer1.2.bn2.weight
model.layer1.2.bn2.bias
model.layer1.2.conv3.weight
model.layer1.2.bn3.weight
model.layer1.2.bn3.bias
model.layer2.0.conv1.weight
model.layer2.0.bn1.weight
model.layer2.0.bn1.bias
model.layer2.0.conv2.weight
model.layer2.0.bn2.weight
model.layer2.0.bn2.bias
model.layer2.0.conv3.weight
model.layer2.0.bn3.weight
model.layer2.0.bn3.bias
model.layer2.0.downsample.0.weight
model.layer2.0.downsample.1.weight
model.layer2.0.downsample.1.bias
model.layer2.1.conv1.weight
model.layer2.1.bn1.weight
model.layer2.1.bn1.bias
model.layer2.1.conv2.weight
model.layer2.1.bn2.weight
model.layer2.1.bn2.bias
model.layer2.1.conv3.weight
model.layer2.1.bn3.weight
model.layer2.1.bn3.bias
model.layer2.2.conv1.weight
model.layer2.2.bn1.weight
model.layer2.2.bn1.bias
model.layer2.2.conv2.weight
model.layer2.2.bn2.weight
model.layer2.2.bn2.bias
model.layer2.2.conv3.weight
model.layer2.2.bn3.weight
model.layer2.2.bn3.bias
model.layer2.3.conv1.weight
model.layer2.3.bn1.weight
model.layer2.3.bn1.bias
model.layer2.3.conv2.weight
model.layer2.3.bn2.weight
model.layer2.3.bn2.bias
model.layer2.3.conv3.weight
model.layer2.3.bn3.weight
model.layer2.3.bn3.bias
model.layer3.0.conv1.weight
model.layer3.0.bn1.weight
model.layer3.0.bn1.bias
model.layer3.0.conv2.weight
model.layer3.0.bn2.weight
model.layer3.0.bn2.bias
model.layer3.0.conv3.weight
model.layer3.0.bn3.weight
model.layer3.0.bn3.bias
model.layer3.0.downsample.0.weight
model.layer3.0.downsample.1.weight
model.layer3.0.downsample.1.bias
model.layer3.1.conv1.weight
model.layer3.1.bn1.weight
model.layer3.1.bn1.bias
model.layer3.1.conv2.weight
model.layer3.1.bn2.weight
model.layer3.1.bn2.bias
model.layer3.1.conv3.weight
model.layer3.1.bn3.weight
model.layer3.1.bn3.bias
model.layer3.2.conv1.weight
model.layer3.2.bn1.weight
model.layer3.2.bn1.bias
model.layer3.2.conv2.weight
model.layer3.2.bn2.weight
model.layer3.2.bn2.bias
model.layer3.2.conv3.weight
model.layer3.2.bn3.weight
model.layer3.2.bn3.bias
model.layer3.3.conv1.weight
model.layer3.3.bn1.weight
model.layer3.3.bn1.bias
model.layer3.3.conv2.weight
model.layer3.3.bn2.weight
model.layer3.3.bn2.bias
model.layer3.3.conv3.weight
model.layer3.3.bn3.weight
model.layer3.3.bn3.bias
model.layer3.4.conv1.weight
model.layer3.4.bn1.weight
model.layer3.4.bn1.bias
model.layer3.4.conv2.weight
model.layer3.4.bn2.weight
model.layer3.4.bn2.bias
model.layer3.4.conv3.weight
model.layer3.4.bn3.weight
model.layer3.4.bn3.bias
model.layer3.5.conv1.weight
model.layer3.5.bn1.weight
model.layer3.5.bn1.bias
model.layer3.5.conv2.weight
model.layer3.5.bn2.weight
model.layer3.5.bn2.bias
model.layer3.5.conv3.weight
model.layer3.5.bn3.weight
model.layer3.5.bn3.bias
model.layer4.0.conv1.weight
model.layer4.0.bn1.weight
model.layer4.0.bn1.bias
model.layer4.0.conv2.weight
model.layer4.0.bn2.weight
model.layer4.0.bn2.bias
model.layer4.0.conv3.weight
model.layer4.0.bn3.weight
model.layer4.0.bn3.bias
model.layer4.0.downsample.0.weight
model.layer4.0.downsample.1.weight
model.layer4.0.downsample.1.bias
model.layer4.1.conv1.weight
model.layer4.1.bn1.weight
model.layer4.1.bn1.bias
model.layer4.1.conv2.weight
model.layer4.1.bn2.weight
model.layer4.1.bn2.bias
model.layer4.1.conv3.weight
model.layer4.1.bn3.weight
model.layer4.1.bn3.bias
model.layer4.2.conv1.weight
model.layer4.2.bn1.weight
model.layer4.2.bn1.bias
model.layer4.2.conv2.weight
model.layer4.2.bn2.weight
model.layer4.2.bn2.bias
model.layer4.2.conv3.weight
model.layer4.2.bn3.weight
model.layer4.2.bn3.bias
model.fc.weight
model.fc.bias
l0.weight
l0.bias
l1.weight
l1.bias
l2.weight
l2.bias

But when I try this

param_groups = [
    [model.conv1, model.bn1, model.layer1, model.layer2],
    [model.layer3, model.layer4],
    [model.fc.weight, model.fc.bias]
]

lrs = np.array([lr / 10, lr / 3, lr])

I get following error

AttributeError                            Traceback (most recent call last)
<ipython-input-110-28b2b69ee3ca> in <module>
      1 param_groups = [
----> 2     [model.conv1, model.bn1, model.layer1, model.layer2],
      3     [model.layer3, model.layer4],
      4     [model.fc.weight, model.fc.bias]
      5 ]

~/.local/lib/python3.6/site-packages/torch/nn/modules/module.py in __getattr__(self, name)
    533                 return modules[name]
    534         raise AttributeError("'{}' object has no attribute '{}'".format(
--> 535             type(self).__name__, name))
    536 
    537     def __setattr__(self, name, value):

AttributeError: 'ResNet50' object has no attribute 'conv1'
2

It’s because your class does not have those attributes but self.model. So you have to use model.model.conv1 and with others attributes as well

1 Like
3

Thanks, that worked.

Is there any way I can avoid writing model.model?

Edit: In order to avoid writing model.model, while creating the instance I used
model = ResNet50(pretrained=True).model

4

I code like it, but it said mobileNetV3 has no attribute ‘model’
And my code below

class MobileNetV3(MyNetwork):

def __init__(self, first_conv, blocks, final_expand_layer, feature_mix_layer, classifier):
    super(MobileNetV3, self).__init__()
    
    self.first_conv = first_conv
    self.blocks = nn.ModuleList(blocks)
    self.final_expand_layer = final_expand_layer
    self.feature_mix_layer = feature_mix_layer
    self.classifier = classifier
    self.quant = torch.quantization.QuantStub()
    self.dequant = torch.quantization.DeQuantStub()

def forward(self, x):
    x.contiguous(memory_format=torch.channels_last)
    x = self.quant(x)
    x = self.first_conv(x)
    for block in self.blocks:
        x = block(x)
    x = self.final_expand_layer(x)
    x = x.mean(3, keepdim=True).mean(2, keepdim=True)  # global average pooling
    x = self.feature_mix_layer(x)
    x = torch.squeeze(x)
    x = self.classifier(x)
    x = self.dequant(x)
    return x

@property
def module_str(self):
    _str = self.first_conv.module_str + '\n'
    for block in self.blocks:
        _str += block.module_str + '\n'
    _str += self.final_expand_layer.module_str + '\n'
    _str += self.feature_mix_layer.module_str + '\n'
    _str += self.classifier.module_str
    return _str

@property
def config(self):
    return {
        'name': MobileNetV3.__name__,
        'bn': self.get_bn_param(),
        'first_conv': self.first_conv.config,
        'blocks': [
            block.config for block in self.blocks
        ],
        'final_expand_layer': self.final_expand_layer.config,
        'feature_mix_layer': self.feature_mix_layer.config,
        'classifier': self.classifier.config,
    }

@staticmethod
def build_from_config(config):
    first_conv = set_layer_from_config(config['first_conv'])
    final_expand_layer = set_layer_from_config(config['final_expand_layer'])
    feature_mix_layer = set_layer_from_config(config['feature_mix_layer'])
    classifier = set_layer_from_config(config['classifier'])

    blocks = []
    for block_config in config['blocks']:
        blocks.append(MobileInvertedResidualBlock.build_from_config(block_config))

    net = MobileNetV3(first_conv, blocks, final_expand_layer, feature_mix_layer, classifier)
    if 'bn' in config:
        net.set_bn_param(**config['bn'])
    else:
        net.set_bn_param(momentum=0.1, eps=1e-3)

    return net

def zero_last_gamma(self):
    for m in self.modules():
        if isinstance(m, MobileInvertedResidualBlock):
            if isinstance(m.mobile_inverted_conv, MBInvertedConvLayer) and isinstance(m.shortcut, IdentityLayer):
                m.mobile_inverted_conv.point_linear.bn.weight.data.zero_()

But when I try to use

torch.quantization.fuse_modules(model, [[‘conv’, ‘bn’, ‘relu’]], inplace=True)
The error:
‘MobileNetV3’ object has no attribute ‘conv’

5

And when I try to add ‘.model’. It said mobileNetv3 has no attribute ‘model’.