src/streamlit_app.py

import altair as alt
import numpy as np
import pandas as pd
import streamlit as st
import os

os.environ["TRANSFORMERS_CACHE"] = "/app/cache"
os.environ["HF_HOME"] = "/app/cache"

from huggingface_hub import login
hf_token = os.getenv("hf_token")
login(token=hf_token)

from transformers import AutoModelForSequenceClassification, AutoTokenizer, pipeline

@st.cache_resource
def load_classifier():
    model_name = "mahsharyahan/EMBEDDIA_crosloengual_bert_Sl"
    model = AutoModelForSequenceClassification.from_pretrained(model_name, token=hf_token)
    tokenizer = AutoTokenizer.from_pretrained(model_name, token=hf_token)
    return pipeline("text-classification", model=model, tokenizer=tokenizer)

# Define sample texts
sample_texts = [
    "Slovenija je čudovita država z bogato kulturo.",
    "Vreme danes ni najboljše, pričakuje se dež.",
    "Ta film mi je bil zelo všeč.",
    "Ne maram zamud pri javnem prevozu.",
    "To je bil odličen športni dogodek."
]

st.title("AI Text Detection(Prototype)")

# Sample selector
selected_sample = st.selectbox(
    "Or select a sample text to detect:",
    ["(Choose a sample)"] + sample_texts
)

# Text area for custom input, pre-filled if a sample is chosen
if selected_sample != "(Choose a sample)":
    user_input = st.text_area("Enter text to dectet:", value=selected_sample)
else:
    user_input = st.text_area("Enter text to dected AI:")

if st.button("Detect"):
    if user_input.strip():
        classifier = load_classifier()
        result = classifier(user_input)
        label = result[0]['label']
        score = result[0]['score']
        st.write(f"**Label:** {label}")
        st.write(f"**Confidence:** {score:.2f}")
    else:
        st.warning("Please enter some text.")