v1.7.0

• add sample weights support via ‘weight’ parameter of GPModel()
• option to initialize regression coefficients and aux_pars from iid model for linear regression models via ‘init_coef_aux_pars_from_iid_model’ parameter (true by default)
• enforce lower bound on error (nugget) variance for gaussian likelihoods to avoid numerical issues
• support calculation of standard errors for covariance and auxiliary parameters for non-Gaussian likelihoods
• fix bug for GPBoost algorithm with offset
• add ‘gaussian_heteroscedastic’ likelihood with fixed effects on variance, log(error variance) = fixed_effects_2 and mean = random_effects + fixed_effects_1