Comment by viccis
21 hours ago
I've said this in another comment but an example would be to train an LLM on a corpus with ALL mathematic content removed. Nothing at all. Then ask it what the shortest distance between two points is. That's an example of synthetic a priori knowledge.
I think it would be nearly impossible to prove that your corpus had no mathematical content. In fact it would be extremely contentious as to what was considered mathematical content. Do you remove all reference to numbers and counting? How about the words and, or and not.
I think the only criteria of no math that would satisfy some people would by definition fail because the model would have no concept of points or distance because some would certainly count those as mathematical content.