Thanks for sharing these findings! Do you think results would improve by using inference microservices and/or specialized LLMs? (Ex. @Get_Writer
’s Palmyra Creative or Fin models)

→ View original post on X — @martyswant