rag-chatbot-3

1
from langchain_community.embeddings import HuggingFaceEmbeddings
2
from langchain_community.vectorstores import FAISS
3
from langchain_community.document_loaders import PyPDFLoader, DirectoryLoader
4
from langchain.text_splitter import RecursiveCharacterTextSplitter 
5

6
DATA_PATH = 'data/'
7
DB_FAISS_PATH = 'vectorstore/db_faiss'
8

9
# Create vector database
10
def create_vector_db():
11
    loader = DirectoryLoader(DATA_PATH, glob='*.pdf', loader_cls=PyPDFLoader)
12

13
    documents = loader.load()
14
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
15
    texts = text_splitter.split_documents(documents)
16

17
    embeddings = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2', model_kwargs={'device': 'cpu'})
18

19
    db = FAISS.from_documents(texts, embeddings)
20
    db.save_local(DB_FAISS_PATH)
21

22
if __name__ == "__main__":
23
    create_vector_db()
24

25