the following is resnet50 fragment code.
def conv_block(input_tensor, kernel_size, filters, stage, block, strides=(2, 2)):
nb_filter1, nb_filter2, nb_filte
