Re:LLMs can't self-correct in reasoning tasks, DeepMind stud...
by jaidhyani on Monday October 09, 2023 @04:21PM
(#37825011)
As the other commenter said, this is incorrect. The input was a sequence of legal moves (not even "real" moves - most of the training data was synthetically generated with "generate legal moves" as the only constraint).
Deducing board state from this is extremely non-trivial.
Deducing board state from this is extremely non-trivial.