I would like to know if training models to reward hack makes them so funny on average or if the examples in the paper are cherry picked. https://t.co/DmVbuKMj07 https://t.co/kAuPg83ihS

Posters: repligate. repligate

4 likes recorded by the source corpus.