Plot GP prior/ get GP prior samples

I’ve got a working GP on some simple example data (i.e. sin(x)). Now I want to examine the GP prior before I start an expensive sampling process. How can this be done?

 with pm.Model() as model:
    ls = pm.HalfCauchy('ls', beta=5)
    # zero mean
    mean =
    # input_dim = the total columns of X
    cov =, ls=ls)
    # The GP
    gp = , cov)
    # Observed data
    f = gp.marginal_likelihood('y_obs', X, Y.flatten(), noise=0)
    # Conditional GP
    f_star = gp.conditional('f*', x) 
    step = pm.NUTS()

    trace = pm.sample(chains=1, step=step, tune=50, draws=100)