alt Hacker NewsI effectively distill the frontier models by building whole sets of skills, personas, and other artifacts that I can then run on smaller models and get 10% even 20% improvements on models like haiku or local models.
There's a lot of room for improving the smaller models at many levels of the stack.
This is a good point. It didn't really work on older small models but the latest crop are quite good at following instructions and paying attention to detail, they just lack a lot of the sophistication and nuance that the frontier models have these days. So they are often capable of doing very complex tasks, they just need more detailed and foolproof instructions than the larger models would.