I am using llama2 model to create a RAG based LLM with huggingface sentence-transformers/all-MiniLM-L6-v2 embedding and pinecone to store information as Vector database. I am feeding a pdf and am able to get results using similarity search on query from vectors in pinecone. I am facing issue in providing the best-k results as context in RetrievalQA function, specifically its retrieval argument:

qa=RetrievalQA.from_chain_type(
    llm=llm, 
    chain_type="stuff", 
    retriever=??
    return_source_documents=True, 
    chain_type_kwargs=chain_type_kwargs)

Here is the complete code:

import os
from pinecone import Pinecone
from pinecone import ServerlessSpec

# initialize connection to pinecone (get API key at app.pinecone.io)
api_key = PINECONE_API_KEY
# configure client
pc = Pinecone(api_key=api_key)
cloud = os.environ.get('PINECONE_CLOUD') or 'aws'
region = os.environ.get('PINECONE_REGION') or 'us-east-1'
spec = ServerlessSpec(cloud=cloud, region=region)
index_name = 'semantic-search-fast'
import time

existing_indexes = [
    index_info["name"] for index_info in pc.list_indexes()
]

# check if index already exists (it shouldn't if this is first time)
if index_name not in existing_indexes:
    # if does not exist, create index
    pc.create_index(
        index_name,
        dimension=384,  # dimensionality of minilm
        metric='dotproduct',
        spec=spec
    )
    # wait for index to be initialized
    while not pc.describe_index(index_name).status['ready']:
        time.sleep(1)

# connect to index
index = pc.Index(index_name)
time.sleep(1)
# populating vectors
for i, t in zip(range(len(text_chunks)), text_chunks):
   query_result = embeddings.embed_query(t.page_content)
   index.upsert(
   vectors=[
        {
            "id": str(i),  # Convert i to a string
            "values": query_result, 
            "metadata": {"text":str(text_chunks[i].page_content)} # meta data as dic
        }
    ],
    namespace="real" 
)
question_embedding = embeddings.embed_query("What is acne?")
query_result = index.query(namespace="real",
    vector=question_embedding,
    top_k=3,
    include_metadata=True
)
relevant_docs = [hit["metadata"]["text"] for hit in query_result["matches"]]

I tried different approaches like creating vectorstore using

vectorstore = Pinecone.from_existing_index(index_name=index_name, embedding=HuggingFaceEmbeddings(), namespace='real') 

and then thought of using retriever=vectorstore.as_retriever() in the argument for RetrievalQA.
But initializing vectorstore throws an error stating :

__init__() missing 1 required positional argument: 'host'

and doesn’t work. I want to know if I could somehow get the above code to work the way it is, without the vectorstore