Does anybody know of a nice explainer or visualization of how current LLM-chatbots are constructed? Especially concerning the interplay between the core LLM component and other parts (e.g. search, ranking, word filters, classifiers, "calculators", ...)?
@simonSimon Willison, maybe?