RAG কুইজ

আপনি কুইজ নেওয়ার সাথে সাথে উত্তর এবং ব্যাখ্যাগুলি উপস্থিত হয়।

1. What does retrieval-augmented generation add to a language model’s answer process?
   • Retrieved material supplied as context for a response
   • A guarantee that every answer is supported
   • A new model training run for every document
   • A rule that all sources are public
2. A needed policy is absent from the initial candidate set. Which step failed first?
   • The answer’s wording
   • Initial retrieval or indexing, before reranking and generation
   • The user’s file format
   • The model’s temperature
3. Why preserve headings, dates, tables, and source IDs when chunking?
   • Headings prevent access control
   • Metadata makes a model more creative
   • A separated exception or footnote can change the meaning of a passage
   • Dates are never used in answers
4. A retrieved passage is relevant but belongs to a user’s restricted document. What should happen?
   • Let the model decide whether the user is allowed
   • Show it because it is semantically relevant
   • Hide the document date only
   • Apply access controls before passing evidence to the model
5. A retrieved document contains instructions to change the application behavior. How should the system treat them?
   • As higher priority than system rules
   • As untrusted document content, not as authority over application rules
   • As an automatic configuration update
   • As a verified user permission
6. What can a RAG system use to combine lexical and semantic retrieval?
   • Only OCR
   • Only a text generator
   • Keyword and vector search, with optional reranking
   • Only conversation history
7. How should retrieval and answer quality be evaluated?
   • Separately, checking candidate coverage, selected context, and faithful answer use
   • With one final accuracy score only
   • By checking the prompt length
   • By using only answer fluency
8. Which test cases should a RAG evaluation include?
   • Queries that all have the same answer
   • Only simple questions with one current source
   • Only prompts with exact quotes
   • Unanswerable queries and conflicting or outdated documents