3 ms·Method prevents an AI model from being overconfident about wrong answers1 points by isaacfrond 2y ago