Comment by tgv

Because there's such a large corpus of 5000 year old cuneiforms waiting to be added to the training set? Ok, that sounds a bit snarky, but how would you expect to improve certainty when there is so little data?