app.py

from langchain import PromptTemplate
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS
from langchain.llms import CTransformers
from langchain.chains import RetrievalQA
import chainlit as cl

DB_FAISS_PATH = "vectorstores/db_faiss"

custom_prompt_template = """Use the following pieces of information to answer the user's question.
If you don't know the answer, just say that you don't know, don't try to make up an answer.

Context: {context}
Question: {question}

only return the helpful answer below and nothing else.
Helpful answer:
"""

def set_custom_prompt():
    """
    Prompt template for QA retrieval for each vectorstore
    """
    prompt = PromptTemplate(template=custom_prompt_template,
                            input_variables=['context', 'question'])
    return prompt


def load_llm():
    """
    Load the language model
    """
    llm = CTransformers(
        model="C:/Users/sanath/Downloads/llama-2-7b-chat.ggmlv3.q8_0.bin",
        model_type = "llama",
        max_new_tokens = 512,
        temperature = 0.5
    )
    return llm

def retrieval_QA_chain(llm,prompt,db):
  qachain = RetrievalQA.from_chain_type(
      llm=llm,
      chain_type="stuff",
      retriever=db.as_retriever(search_kwargs={'k':2}),
      return_source_documents=True,
      chain_type_kwargs={'prompt':prompt}

  )
  return qachain


def qa_bot():
  embeddings = HuggingFaceEmbeddings(model_name='sentence-transformers/all-miniLM-L6-V2', model_kwargs={'device': 'cpu'})
  db = FAISS.load_local(DB_FAISS_PATH,embeddings)
  llm = load_llm()
  qa_prompt = set_custom_prompt()
  qa = retrieval_QA_chain(llm,qa_prompt,db)
  return qa

def final_result(query):
  qa_result = qa_bot()
  response = qa_result({'query':query})
  return response

@cl.on_chat_start
async def start():
  chain = qa_bot()
  msg = cl.Message(content = "Starting the bot...")
  await msg.send()
  msg.content = "Hi, Welcome to the Medical bot. What is your query?"
  await msg.update()

  cl.user_session.set("chain",chain)

@cl.on_message
async def main(message):
  chain = cl.user_session.get("chain")
  cb = cl.AsyncLangchainCallbackHandler(
      stream_final_answer = True, answer_prefix_tokens = ["FINAL","ANSWER"]
  )
  cb.answer_reached = True
  res = await chain.acall(message, callbacks=[cb])
  answer = res["result"]
  sources = res["source_documents"]
  if sources:
    answer += f"\nSources:" + str(sources)
  else:
    answer += "\nNo sources found"
  
  await cl.Message(content=answer).send()