The End of Finetuning – With Jeremy Howard of Fast.ai (latent.space)

Fascinating how Jeremy openly and honestly says that a technique he developed and everyone uses is now not the best way to go.

I have become obsessed with smaller LLMs lately. Every time I use mistral-7B I wonder what exactly the developers did - anyone have links to good papers by the Mistral developers?

thanks for submitting this! unfortunately HN doesnt seem to like podcasts :/