ch06 model overfitting comment

[nicolaleo]

in the notebook of the chapter there is this line of comment "However, based on the slightly lower test set performance, we can see that the model overfits the training data to a very small degree".
But the training and validation curve are quite similar without any gap on the final accuracy while is the test set accuracy that shows a little gap. Wouldn't it be correct to say that the problem is the overfitting on the dev set and therefore a larger dev set would be needed?

[rasbt]

Thanks for the comment. You raise a good point. I am not sure if a larger dev set could potentially solve it (it might, but that's a good hypothesis). I'd say the overfitting to the training set (and to the validation/dev set) could also come from tweaking hyperparameters slightly to get a good validation set performance.

I modified the note as follows:

• However, based on the slightly lower test set performance, we can see that the model overfits the training data to a very small degree, as well as the validation data that has been used for tweaking some of the hyperparameters, such as the learning rate

(However, these are only short notes in the Jupyter notebook; I will say more in the chapter itself)