Visualizers
Learn by moving things
Interactive explainers for the ideas behind retrieval, embeddings, and agents. A formula tells you what is true; dragging a vector shows you why. Each one also appears inside the article it belongs to, and runs entirely in your browser, so drag, poke, and break it.
Corrective RAG (CRAG): grade, then recover
From: RAG That Thinks →An evaluator grades the retrieved documents. When they look weak or only half cover the question, the system does not plough ahead; it refines the query and retrieves again before answering, instead of returning a confidently wrong answer.
Iterative / multi-hop RAG
From: RAG That Thinks →Some questions can only be searched for one fact at a time, because each answer is the key to the next question. Watch a three-hop chain: the company that built the iPhone, its founder, then the founder's successor.
Hop 1: retrieve “What company built the iPhone?”, get “Apple”, then feed that into the next question.
Adaptive RAG: retrieve only when it helps
From: RAG That Thinks →A router in front of retrieval decides whether a query even needs external knowledge. Simple questions go straight to the model; questions about private or fresh data get retrieved. It sizes the effort to the question.
Skip retrieval. the model already knows this; retrieval would only add noise and latency.