I would be interested to see if you could still classify idiosyncrasies from various models all fine-tuned in this way, and how different the performance of the classifier would be from baseline. More broadly, is “the same persona” in different models more consistent than “different personas” in the same model?
I would be interested to see if you could still classify idiosyncrasies from various models all fine-tuned in this way, and how different the performance of the classifier would be from baseline. More broadly, is “the same persona” in different models more consistent than “different personas” in the same model?
That’s a good question. I think I will add it to my queue if no one else picks it up.
Have you looked at this yet or know of anyone that has? If not, I’d love to work on it if you wouldn’t mind me scooping the idea.
Me and David are doing some followup work on EM, but mostly didn’t follow this branch.