app.py

import os
import streamlit as st
from datasets import load_dataset
from huggingface_hub import InferenceClient

api_key = os.getenv("HF_API_KEY")
client = InferenceClient(api_key=api_key)
dataset = load_dataset("andreska/adregadocs", split="test")

def read_dataset(dataset): 
    text = [] 
    for item in dataset: 
        text.append(item['text'])
    return "\n".join(text)

def chunk_text(text, max_length): 
    words = text.split() 
    for i in range(0, len(words), max_length): 
        yield " ".join(words[i:i+ max_length])

context = read_dataset(dataset)

max_chunk_length = 30000;
st.title("Adrega AI Help")
user_input = st.text_input('Ask me a question') 

if st.button("Submit"): 
    if user_input: 
        responses = []

        for chunk in chunk_text(context, max_chunk_length):
            messages = [
                {"role": "system", "content": f"Context: {context}"},
                {"role": "user","content": user_input}
            ]

            completion = client.chat.completions.create(
                model= "Qwen/Qwen2.5-72B-Instruct",
                #model="Qwen/Qwen2.5-Coder-32B-Instruct", 
                #model="HuggingFaceTB/SmolLM2-1.7B-Instruct",
            	messages=messages, 
            	max_tokens=500
            )
        
            answer = completion.choices[0].message['content']
            responses.append(answer)
        
        final_response = " ".join(responses)
        st.write(f"Adrega AI: {final_response}") 
    else: 
        st.write("Please enter a question.")