Gp.predict gives mismatched dimensions error

Questions v5
1

Hi, I’m trying to apply kriging

however I get the error
ValueError: Input dimension mis-match. (input[0].shape[1] = 364, input[1].shape[1] = 2406)
when using gp.predict() on the test data locations

Any ideas where I’m going wrong?

    with pm.Model() as model:
        c = pm.Normal(name='c', mu=0, sigma=10,shape=1)
        ls = pm.Normal(name='ls', mu=[0.1, 0.1, 0.01], sigma=5,shape=3)
        k = pm.Exponential('k',0.01)
        cov_func = pm.gp.cov.ExpQuad(input_dim=3,ls=ls)
        mean_func = pm.gp.mean.Constant(c) #pm.gp.mean.Constant(c=c)
        noise = 5
        gp = pm.gp.Marginal(mean_func, k*cov_func)#,sigma=noise,observed={'X':X, 'Y':y})
        f = gp.marginal_likelihood("f", X, y, noise)
     
        
    with model:
        map_estimate = pm.find_MAP(model=model)
        y = y.flatten()
        y_star = gp.predict(X_star,point=map_estimate)