Extracting info from Files_in_Knowledgebase

Hallo Everyone, I am using just the prompts and knowledgebase to build my custom gpt with gpt builder, I just wanted to know how the information retrieval from the knowledgebase takes place, is it converted in to vectors or is the text extracted and included in the query text from the user ?, and as i have read in one of the posts, does it depend on size of the files

And does using a markdown format make it easier for the model to retrieve the info in the file than pdf or html formats that have a lot of presentation information in them?
Does the name of the files also have an impact ?
Thanks in advance