return to table of content

Training language models to be warm and empathetic makes them less reliable

373 comments