# Generality of the learning regime as Sellarsian fit for LLM philosophy > Something like: the kind of system an LLM is may already make it apt for a Sellarsian conception of philosophy, before we start itemizing particular philosophical maneuvers. > > And next-token training over a massive mixed corpus may itself be a way of forcing a system to form broad, portable, cross-context representations. So the affinity may lie not just in the breadth of the corpus, but in the generalizing character of the learning regime itself.