Apply a Chunking Strategy for a Given Document Structure and Model Constraints GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   Which chunking strategy is best suited for a document with clear hierarchical sections, such as a legal contract with numbered clauses?

   • ARandom chunking at arbitrary character positions
   • BFixed-size chunking with a token count of 200
   • CStructure-based chunking that respects the numbered clauses
   • DSemantic chunking using sentence embeddings
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A team is building a RAG system for a collection of user manuals. The manuals contain step-by-step instructions. Each step is a separate paragraph. The target model has a context window of 2048 tokens. The team is using a fixed-size chunking strategy with a chunk size of 1000 tokens and an overlap of 100 tokens. During evaluation, they find that the system sometimes fails to answer questions about a sequence of steps that spans a chunk boundary. What is the most effective way to fix this?

   • AIncrease the overlap to 500 tokens
   • BUse a semantic chunking strategy that groups steps by embedding similarity
   • CChange the chunking strategy to split on paragraph boundaries, ensuring that each chunk contains a complete set of steps
   • DDecrease the chunk size to 500 tokens and increase the overlap to 200 tokens
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   Why is it important to include overlap between consecutive chunks in a chunking strategy?

   • ATo eliminate the need for embeddings
   • BTo maintain context continuity across chunk boundaries
   • CTo reduce the total number of chunks
   • DTo ensure each chunk has the exact same token count
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A team has implemented a chunking strategy for a corpus of legal contracts. They are evaluating the quality of the chunks by running a set of representative questions through their RAG pipeline. They notice that some answers are incomplete because the relevant information is split across two chunks. What is the best way to evaluate and improve the chunking strategy?

   • ASwitch to a different embedding model with a larger context window
   • BReview the chunks that correspond to the failing questions to see if a concept is split, then adjust the chunk boundaries or add overlap
   • CIncrease the number of chunks retrieved for each query from 3 to 5
   • DMeasure the average token count of the chunks and ensure it is below the model's limit
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which metric is most directly used to evaluate the retrieval performance of a chunking strategy?

   • ARecall@k
   • BNumber of chunks generated
   • CToken count per chunk
   • DTime to embed all chunks
   Reveal Solution Discussion Select an answer first