检索增强生成

题集来源:AMIT SHEKHAR(amitshekhariitbhu)收录项目:ai-engineering-interview-questions查看原题集 反馈(在 GitHub 新标签页打开)

题目目录

42 道原题
  1. What is re-ranking, and how does it improve RAG retrieval quality?

    原题
  2. What is ColBERT, and how does late interaction retrieval work?

    原题
  3. Compare reranker architectures: cross-encoder, ColBERT, and LLM-based rerankers.

    原题
  4. How do you handle multi-document and multi-hop questions in RAG?

    原题
  5. What is the "lost in the middle" problem in RAG systems?

    原题
  6. How do you evaluate a RAG system? Explain faithfulness, relevance, and context precision/recall.

    原题
  7. Explain Self-RAG. How does the model decide when to retrieve?

    原题
  8. What is GraphRAG, and when would you use it over traditional RAG?

    原题
  9. Vectorless RAG

    原题
  10. How do you handle structured data (tables, SQL databases) in a RAG pipeline?

    原题