L2 - Basic RAG Pipeline Chunking Strategy

In the Basic RAG implementation the text from the pdf pages is merged into 1 document. Is that chunked before being fed to the vector index or is it embedded as a whole? If it is chunked what is the logic?