139 - Coherent Long Story Generation, with Kevin Yang
NLP Highlights
00:00
How Price Generate Stories on a Comparative Longevity
The main baseline that we use is just uh like a rolling window gpt3 baseline where we give it the same premise and then let it generate. There's one other baseline which is they do the same thing except that the gpt3 model is fine tuned on stories from writing prompts. The story generated from these models based on some aspects of quality which were just regular right can you that is what uh what aspects of quality that you want to put thatYeah so all of these comparisons are pairwise like we always compare like our story against like one of the baseline stories that for the same premise it's hard to like give like a concrete number to like something like how plaqua here is
Transcript
Play full episode
