Hacker News new | ask | show | jobs
by danans 1700 days ago
The "word" segmentation seems pretty off, even for these hallucinated words and meanings. Are they generated by the same model?