I'm a staff ML engineer trying to internalize modern GenAI concepts, and wrote this to explain LLM buzzwords (RAG, RLHF, MoE, etc.) through a fantasy metaphor.
It’s a follow-up to two other stories on how LLMs and RAG systems work. Would love feedback from the HN crowd.
It's way too long and too obviously generated for my tastes. The idea is ok but the metaphors should generally be enhancing understanding. And the part about scribes being "modified" lost the fantasy plot altogether.
And nonsense like this:
"It’s like equipping a Knight with a special glove instead of retraining them entirely."
does not signal to the reader that you produced this with a lot of care or thought. I'm not saying you didn't, just that this is a risk of saving time with LLMs; missing the occasional nonsense hallucination is enough to change people's mind about the piece, maybe even you as a writer.
reply