Question 24
Domain 2: Fundamentals of Generative AIWhat is the purpose of vector embeddings in a large language model (LLM)?
Correct answer: C
Explanation
Vector embeddings turn text into numeric vectors so the model can compare meaning mathematically. AWS describes embeddings as enabling semantic search and retrieval, where similar texts are close in vector space, which is why they support “mathematically compare texts.”
Why each option is right or wrong
A. Splitting text into manageable pieces of data
Chunking splits long documents into smaller pieces for retrieval and parsing.
B. Grouping a set of characters to be treated as a single unit
Tokenization groups characters or subwords into tokens for model input.
C. Providing the ability to mathematically compare texts
AWS Bedrock’s Titan Embeddings documentation describes embeddings as converting text into numerical vectors that can be compared in vector space, which is the basis for semantic similarity and retrieval. In practice, this lets an LLM system measure closeness between two passages mathematically rather than relying on exact keyword matches, enabling use cases like search and knowledge-base retrieval.[^201004][^201005]
D. Providing the count of every word in the input
Token counting tracks how many tokens appear in input, not word frequencies.