A Generative AI Engineer is loading 150 million embeddings into a vector database that takes a maximum of 100 million. Which TWO actions can they take to reduce the record count?

Increasing document chunk size means each chunk covers more text, so fewer total chunks are needed to represent the same corpus, directly reducing the embedding record count.

Decreasing overlap between chunks reduces the number of duplicate or near-duplicate token windows that generate separate embeddings, further lowering the total record count.

Decreasing chunk size produces more, smaller chunks from the same documents, which increases the total number of embeddings beyond 150 million.

Increasing overlap generates more overlapping windows between adjacent chunks, creating additional redundant embeddings and increasing the record count.

Using a smaller embedding model changes the dimensionality of each vector but does not reduce the number of chunks or records being stored.