2 ms·
It would also be interesting to train these models on the entire math literature. One would want to use this to verify the math literature, translating the pro
by pfdietz 3y ago
It would also be interesting to train these models on the entire math literature. One would want to use this to verify the math literature, translating the proofs there to checkable forms. Eventually, we would get a model that knows every theorem and every correct proof in the math literature.