The more I work with local LLM the less impressed I am with the model, while being more impressed with the infrastructure around it. I feel like the supporting infrastructures around the model (inference engine, RAG system, etc.) should have at least the same spotlight as the models, since those are the ones that bridge the LLM to real world problems.

Looking for better term for concept of a short string of words, mostly for expressing passing thought or much less fleshed-out ideas. I have considered notion, blurb, and note but nothing felt quite it. I will use “thought bubbles” for now.