src/streamlit_app.py

import numpy as np
import streamlit as st
from typing import Dict, List
import torch
import json
from transformers import AutoModelForSequenceClassification, AutoTokenizer

TITLE = "Enter the article or leave this field blank...."
SUMMARY = "Enter summary of the article or leave it blank...."

@st.cache_resource
def load_model():
    model = AutoModelForSequenceClassification.from_pretrained("Kulebyaka-kokosik/articles-classifier-model")
    tokenizer = AutoTokenizer.from_pretrained("distilbert/distilbert-base-cased")
    model.eval()
    return model, tokenizer


@st.cache_resource
def load_label_to_topic():
    with open("label_to_topic.json") as file:
        label_to_topic = json.load(file)
    return label_to_topic


def inference(
        model: AutoModelForSequenceClassification,
        tokenizer: AutoTokenizer,
        title: str,
        summary: str
) -> torch.Tensor:
    tokenized = tokenizer(title, summary, padding='max_length', truncation=True, return_tensors="pt")

    with torch.no_grad():
        logits = model(**tokenized).logits
    return logits


def predict_topics(
        model: AutoModelForSequenceClassification,
        tokenizer: AutoTokenizer,
        title: str,
        summary: str,
        label_to_topic: Dict[str, str],
        top_k: int = 5,
) -> List[str]:
    logits = inference(model, tokenizer, title, summary)

    probs = torch.sigmoid(logits).squeeze(0)
    labels = np.argsort(-probs)[:top_k].tolist()
    topics = [label_to_topic[str(label)] for label in labels]
    return topics



def main():
    model, tokenizer = load_model()
    label_to_topic = load_label_to_topic()

    st.title("Articles classifier")

    title = st.text_input("Enter the article title in english or leave this field blank:")
    summary = st.text_area("Enter the article summary in english or leave this field blank:")
    k = st.number_input(
        "Top-K labels. Enter K:",
        min_value=1,
        max_value=20,
        value=5,
        step=1
    )

    if st.button('Classify'):
        if title == "" and summary == "":
            st.error("Please enter title and/or summary")
            return
        st.subheader("Most probable topics:")
        topics = predict_topics(model, tokenizer, title, summary, label_to_topic=label_to_topic, top_k=k)
        for topic in sorted(topics):
            st.write(topic)

main()