Chapter Assessment: RAG, LLM Inference, and Agents

This is the chapter review for RAG, LLM Inference, and Agents. The bank holds twenty-eight questions and each attempt draws ten, so repeating it gives you different coverage. This is the chapter where the gap between having read about a subject and having shipped it is widest, and the questions are written to tell the two apart. Four specifics are worth holding in mind. They are the indexed dimension ceiling and what to do about it, the throughput range commonly cited for continuous batching, the ordering rule that makes prefix caching work, and the four limits that bound an agent run