Scikit-learn: How to specify multiple input kernels in gaussian process regression?

Question

This is what I am trying:

kernel = C(np.ones(11), (1e-3, 1e3)) * RBF(length_scale=np.ones(11), length_scale_bounds=(1e-5, 1e3)) + WhiteKernel(
    noise_level=np.array(11*[1e-4]), noise_level_bounds=(1e-5, 1)) 
gp = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=9, normalize_y=True )

and this is the error I am getting:

    286     # unbounded variables must use None, not +-inf, for the optimizer to work properly
ValueError: length of x0 != length of bounds

But when I try the following, it works:

kernel = C(1.0, (1e-3, 1e3)) * RBF(length_scale=np.ones(11), length_scale_bounds=(1e-5, 1e3)) + WhiteKernel(
noise_level=1, noise_level_bounds=(1e-5, 1)) 

However, I would like to try the first kernel as I have 11 inputs and 8 outputs and I want better predictions from my gp.