I have a dataset that had 20 features and 65 samples. The performance of the models inside was extremely poor, so I did the interpolation with the help of scipy.rbf and added 300 samples to the dataset. The performance of the models was excellent. Almost all models reached R2 of 99% in the training and test sets. After taking cross validation, the result was the same at about 99%.

Is there a way for me to understand that my new data set is meaningful and false data has not reached 99% of R2?