Large Language Models Do Gradient Descent at Runtimetil.nicklothian.com2 points·nl··0 commentsOpen articleSaveView on HN