Final quiz · 6 questions From model to system Answer every question, then check. With at least 5 correct answers, you earn the course badge. 1. Where in a long context does a model make the least use of information? At the beginning In the middle At the end Equally everywhere 2. Why ask for the steps of a calculation? To go faster So that each step supports the next and stays verifiable To reduce the cost To switch models 3. What does retrieval-augmented generation do before answering? It retrains the model It retrieves relevant passages and places them in the context It translates the question It deletes the history 4. In retrieval-augmented generation, where does a bad answer often come from? The choice of font A wrong passage brought back by the retrieval The temperature The number of layers 5. In a mixture of experts, what goes down? The memory required The computation per token The total number of parameters The size of the vocabulary 6. What does an open-weight model make possible? No longer needing hardware Running it on your own machines Avoiding every error Learning from every conversation Check my answers