ValueError: Shape must be rank 4 but is rank 5 for '{{node conv2d_transpose}} = Conv2DBackpropInput[T=DT_FLOAT, data_format="NHWC", dilations=[1, 1, 1, 1], explicit_paddings=[], padding="SAME", strides=[1, 4, 4, 1], use_cudnn_on_gpu=true](conv2d_transpose/input_sizes, Generator_2/deconv_ir/deconv1/kernel/read, SOURCE_ir)' with input shapes: [4], [3,3,1,1], [1,64,64,4,1].