4 ms·
If it were actually reasoning generally, though, it wouldn't need to be trained on each game. It could be told the rules and figure things out from there.
by bccdee 2y ago
If it were actually reasoning generally, though, it wouldn't need to be trained on each game. It could be told the rules and figure things out from there.
- yencabulator 2y agoEven worse, the LLM is supposed to already "know" Sudoku rules. Either that, or it doesn't "know" anything that was scraped from the web...