3 ms·
> Learning a transformer language model on such data simply learns a distribution over sequences of tokens. If statistical distributions can model higher leve
by echelon 3y ago
> Learning a transformer language model on such data simply learns a distribution over sequences of tokens.
If statistical distributions can model higher level polypeptide structure, then it could be useful.