Y
Hacker News
new
|
ask
|
show
|
jobs
by
astrange
42 days ago
Because cheating is easier than actually doing work, if you use this to train future models, it's likely you'll end up with cheating instead of actual generalization.