When looking to train a model, does it make sense to have a 60-20-20 train val test split, first hyper parameter tuning over the training dataset, using the validation set, picking the best model. Then training over train+val and the final test occurring on the test set?
I would say that this depends heavily on the type of data that you have and the task at hand. If the available dataset is sufficiently large, you can add a larger validation and test set. If you only have limited data available, you might consider decreasing the size of the validation and test set in order to improve the model by providing it with more data for training.
But generally speaking, without having any further information about your case, the approach is fine.