Deep Learning with PyTorch

DenseBlock

Let's look at the code for DenseBlock and then walk through it:

class _DenseBlock(nn.Sequential):    def __init__(self, num_layers, num_input_features, bn_size, growth_rate, drop_rate):        super(_DenseBlock, self).__init__()        for i in range(num_layers):            layer = _DenseLayer(num_input_features + i * growth_rate, growth_rate, bn_size, drop_rate)            self.add_module('denselayer%d' % (i + 1), layer)

DenseBlock is a sequential module where we add layers in a sequential order. Based on the number of layers (num_layers) in the block, we add that number of _Denselayer objects along with a name to it. All the magic is happening inside the DenseLayer. Let's look at what goes on inside the DenseLayer.