still a pain to run these models locally. when you do run them locally they max out your vram. RAG use cases are super interesting though. I've always wanted to build a local RAG personal assistant that can search my calendar, emails and nostr notes to find things. just gotta find time to play with it.

nostr:note1dmlyg5lg8jmysch3de8cgfwk8q6cqvl08px372kp0rw3g57fz02qjsnqv9