> they produce incoherent sentences with no relation to the text they are supposed to extrapolate I think we’re agreeing here? Language models of the size of

> they produce incoherent sentences with no relation to the text they are supposed to extrapolate I think we’re agreeing here? Language models of the size of SD produce literally useless output. SD can produce kind of fun output that people can make use of sometimes?