Hacker News new | ask | show | jobs
by amelius 1267 days ago
Transformer based models have been shown to reproduce training text verbatim.