app.py

import streamlit as st
import torch
import time
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Streamlit page configuration
st.set_page_config(page_title="Review Keypoint Extractor (BART-Large-CNN)", page_icon="🔑")

# Define the model
model_name = "facebook/bart-large-cnn"

# Cache the model and tokenizer to avoid reloading
@st.cache_resource
def load_model_and_tokenizer():
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model = model.to(device)
    return tokenizer, model, device

# Keypoint generation function
def generate_keypoint(review, max_new_tokens=64):
    tokenizer, model, device = load_model_and_tokenizer()
    
    start_time = time.time()
    
    # BART-specific prompt (no additional prompt engineering)
    prompt = review
    
    # Inference
    inputs = tokenizer(prompt, return_tensors="pt", truncation=True, padding=True).to(device)
    with torch.no_grad():
        outputs = model.generate(**inputs, max_new_tokens=max_new_tokens)
    keypoint = tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
    
    # Post-process: Normalize "no key point" outputs
    if keypoint.lower() in ["none", "no keypoint", "no key point", "n/a", "na", "", "nothing"]:
        keypoint = "No key point"
    
    elapsed = time.time() - start_time
    return keypoint, elapsed

# Streamlit UI
st.title("🔑 Review Keypoint Extractor (BART-Large-CNN)")
st.write("Enter a product review below to extract its key points using the facebook/bart-large-cnn model.")

# Input field for review
review = st.text_area("Product Review", placeholder="e.g., The Jackery power station is lightweight and charges quickly, but the battery life could be longer.")

# Button to generate keypoint
if st.button("Extract Keypoint"):
    if review.strip():
        with st.spinner("Generating keypoint..."):
            keypoint, elapsed = generate_keypoint(review)
            st.success(f"✅ Keypoint generated in {elapsed:.2f} seconds!")
            st.subheader("Results")
            st.write(f"**Review:** {review}")
            st.write(f"**Keypoint:** {keypoint}")
    else:
        st.error("⚠️ Please enter a valid review.")

# Footer
st.markdown("---")
st.markdown("Powered by [Hugging Face Transformers](https://huggingface.co/) and [Streamlit](https://streamlit.io/)")