Hacker News new | ask | show | jobs
by caxco93 1229 days ago
Could a newer language model use this to penalize output that fails the classifier during training?