3 ms·
IIUC the argument is that while unsolved there was much work done on them that shows up in the training data. The idea being that the LLM is limited to a small
by fc417fc802 28d ago
IIUC the argument is that while unsolved there was much work done on them that shows up in the training data. The idea being that the LLM is limited to a small amount of inference over externally supplied data.
- eieje1 28d ago[dead]
- vonneumannstan 26d agoAnd yet the best humans could not use that same available data to solve the problems.