Apply a Chunking Strategy for a Given Document Structure and Model Constraints GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A team has implemented a chunking strategy for a set of medical research papers. They are evaluating the quality of the chunks by checking if each chunk is self-contained and coherent. They find that some chunks contain only a table and its caption, while others contain only the text that references the table. What is the most likely problem with their current strategy?

   • AThe chunking strategy is not respecting the document's structure, separating the table from the text that discusses it
   • BThe embedding model is not capable of understanding tables
   • CThe chunk size is too large, causing the table and its reference to be in the same chunk
   • DThe overlap is too high, causing duplicate content in the chunks
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   A generative AI engineer is selecting a chunk size for a document to be processed by a model with a 4,096-token context window. What is the primary constraint that determines the maximum chunk size?

   • AThe storage capacity of the vector database
   • BThe number of pages in the document
   • CThe model's maximum token limit
   • DThe speed of the embedding model
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   When analyzing a document's structure to inform chunking, which element is most likely to represent a natural semantic boundary?

   • AThe first word of every page
   • BA heading that introduces a new major topic
   • CThe exact middle of a sentence
   • DA random point in the middle of a paragraph
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A company is building a chatbot to answer questions from its internal policy documents. The documents are a mix of short paragraphs and long, dense tables. The target LLM has a context window of 4096 tokens. The team wants to ensure that the most relevant information is retrieved for each query. What is the most important consideration when chunking the tables?

   • ASplit the table into smaller chunks based on rows or groups of rows, ensuring each chunk fits within the model's context window
   • BKeep each table as a single chunk, even if it exceeds the model's token limit
   • CUse a fixed-size chunking strategy on the raw table text with a chunk size of 1024 tokens
   • DConvert the table to a text description and then chunk the text using a standard recursive strategy
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A team is building a RAG system for a large corpus of legal documents. The documents are very long, often exceeding 10,000 tokens. The target LLM has a context window of 4096 tokens. They are using a recursive chunking strategy with a chunk size of 3000 tokens and an overlap of 200 tokens. What is the most important thing to verify about their chunks?

   • AThat the chunks are all exactly 3000 tokens long
   • BThat the chunks are all semantically related to the document's main topic
   • CThat the overlap is exactly 200 tokens for every chunk
   • DThat no chunk exceeds the model's context window of 4096 tokens
   Reveal Solution Discussion Select an answer first