Method backward

intermediate_source/custom_function_conv_bn_tutorial.py:70–72

(ctx, grad_out)

    @staticmethod
    @once_differentiable
    def backward(ctx, grad_out):
        X, weight = ctx.saved_tensors
        return convolution_backward(grad_out, X, weight)

######################################################################
# When testing with ``gradcheck``, it is important to use double precision