app.py

!pip install qdrant_client
import os
import streamlit as st
import requests

import qdrant_client



client = qdrant_client.QdrantClient("http://localhost:6333", prefer_grpc=True)
client.get_collections()


url = "https://api-ares.traversaal.ai/live/predict"
headers = {
  "x-api-key": "ares_5e61d51f3abc8feb37710d8784fa49e11426ee25d7ec5236b80362832f306ed2",
  "content-type": "application/json"
}

st.title('#@ck-RAG')
def inference(query):
    payload = { "query": [query] }
    response = requests.post(url, json=payload, headers=headers)
    # st.error(response)
    # st.error(response.text)
    response_text=response.json().get('data').get('response_text')
    urls=response.json().get('data').get('web_url')
    return response_text, urls

prompt = st.text_input('Enter a query', value='')

if prompt:
    

    results = client.query(
    collection_name="knowledge-base",
    query_text=prompt,
    limit=10,
    )
    #results

    context = "Hotel Name: " + "\n".join(r.document for r in results ) 
    #context 
    metaprompt = f"""
    Based on the context provided, provide information about the Question. You can give multiple points based on the question asked or context.

    Question: {prompt.strip()}

    Context: 
    {context.strip()}

    Answer:
    """
    response_text,urls = inference(metaprompt)

    # Look at the full metaprompt
    # print(metaprompt)
    st.write('Query Results:')
    st.write(response_text)
    st.write('Sources:')
    st.write(urls)