Problem in model comparison

Thank you for your response. But, in the warning part, the answers are all true which mean that the computation of LOO/WAIC may not be reliable. Concerning the weights, how the best model doesn’t have the highest weight (probability) to be true?