Spaces:
Sleeping
Sleeping
import torch | |
torch.device('cpu') | |
import chainlit as cl | |
from faissdenseretrieval import initialize_documents, initialize_faiss_document_store, initialize_rag_pipeline | |
import os | |
from dotenv import load_dotenv | |
# Load environment variables (if any) | |
load_dotenv("../.env") | |
load_dotenv() | |
serp = os.getenv("SERP_API_KEY") | |
openai_key = os.getenv("OPENAI_API_KEY") | |
# Initialize documents | |
documents = initialize_documents(serp_key=serp, nl_query="IMDB movie reviews for the Barbie movie (2023)") | |
# Initialize document store and retriever | |
document_store, retriever = initialize_faiss_document_store(documents=documents) | |
# Initialize pipeline | |
query_pipeline = initialize_rag_pipeline(retriever=retriever, openai_key=openai_key) | |
async def main(message: str): | |
# Use the pipeline to get a response | |
output = query_pipeline.run(query=message) | |
# Create a Chainlit message with the response | |
response = output['answers'][0].answer | |
msg = cl.Message(content=response) | |
# Send the message to the user | |
await msg.send() | |