rag_chain.py

# rag_chain.py
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.chains import RetrievalQA
from langchain.llms import HuggingFaceHub

def setup_rag_chain(docs):
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
    vectorstore = Chroma.from_documents(docs, embedding=embeddings)

    retriever = vectorstore.as_retriever()
    
    # Replace this with your own hosted LLaMA 3.1 if needed
    llm = HuggingFaceHub(
        repo_id="meta-llama/Meta-Llama-3-8B-Instruct",
        model_kwargs={"temperature": 0.3, "max_tokens": 512}
    )

    qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)
    return qa_chain