Training large language models on narrow tasks can lead to broad misalignment

by thebeardisred
today at 6:17 AM
3 points
Comments
Loading comments...