Im trying to create a RAG, i start by breaking down the document into chunks, send it to a localy hosted embedding model, get the vectors back, and then i get stuck with the FAISS part.
my problem is that all of what i find seems to want to connect to hugging face or something, and the langchain stuff seems to be geared towards doing that so i will get my embeddings from an external source or have to spend days re-running code until i get a model to download fully without telling me the hash doesnt match because it only tryed downloading 98.4% of it. why do that when i have a perfectly good server.
so please help me get on track, because ive been trying to find an answer for this for hours.
hears my proof of concept code