問題文
In the typical architecture of an LLM-powered test infrastructure, one part accepts what the tester submitted, performs authentication and data retrieval, prepares the prompt, calls the model, and then post-processes the raw answer before it is shown. Which part carries out that work?
選択肢
- The vector database, because it stores the encoded embeddings and serves the closest matches within the wider test infrastructure.
- The back end, because it manages the incoming requests and refines the raw answers within the wider test infrastructure.
- The front end, because it collects the typed commands and shows the returned findings within the wider test infrastructure.
- The hosted model, because it reads the structured prompts and writes the fresh sentences within the wider test infrastructure.