 | Been building with LLMs for a while and kept hitting terms where the standard definition was useless for making engineering decisions. So I kept a personal doc, eventually it hit 30+ terms across inference, retrieval, agents, training, and prompting. Each entry has the plain-English definition plus the production implication, the thing that actually affects your architecture or debugging. Cleaned it up, built a small interactive UI with search and category filtering, and put it on GitHub. Not trying to compete with papers or courses, it's more of a field reference for when you're mid-build and need the practical version of a term fast. Would genuinely appreciate corrections or additions. The bar I set for new terms: does the definition help someone make a better engineering decision? [link] [comments] |