2 comments

[ 2.6 ms ] story [ 18.2 ms ] thread
[flagged]
It's just yet another inconsistency, which will be trained out as soon as people know to train on it. It's not some deep property of reality that the models are incapable of dealing with.