Fast Model Editing at Scale

Eric Mitchell, Charles Lin, Antoine Bosselut +2

While large pre-trained models have enabled impressive results on a variety of downstream tasks, the largest existing models still make errors, and even accurate predictions may be…