Ideal ratio with respect to scoring dataset and training dataset

Like the 70 - 30 ratio for trainig and testing, is there a suggested ratio for the datasets of training and scoring?

(This is so as to reduce the training data to the correct proportion for best scoring)