Update resnet_in.py

[DiggyBee]

Corrected issue with resnet_in.py where initial_kernel_size argument was missing from the ResNet class. This caused num_classes argument to be overwritten by the intended initial_kernel_size value.

[DiggyBee]

Sorry if I have messed up the pull request. It is my first time doing it. I have corrected an issue with resnet_in.py which caused any model using resnet_in.py to use the inital_kernel_size argument, in the R50_ImgNet example (see below), to be used for the num_classes.

elif model=='R50_ImgNet':

        from models import resnet_in as resnet

        rnet_hr = resnet.ResNet(resnet.BasicBlock, [3,4,6,3], **7**, 1000)
        rnet_lr = resnet.ResNet(resnet.BasicBlock, [3,4,6,3], **7**, 1000)
        agent = resnet.ResNet(resnet.BasicBlock, [2,2,2,2], **3**, 16)

The third argument, either 3 or 7, in the example above is passed into the ResNet class below.

class ResNet(nn.Module):

    def __init__(self, block, layers, num_classes=1000, zero_init_residual=False,
                 groups=1, width_per_group=64, replace_stride_with_dilation=None,
                 norm_layer=None):
        super(ResNet, self).__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        self._norm_layer = norm_layer

...

The third argument of ResNet is num_classes. By passing a value of 3 or 7, as found in the utils.py file, we find that the intended inital_kernel_size argument ends up being used for the num_classes argument.