app.py

from typing import Dict, Any

import gradio as gr
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification


def predict(text: str) -> Dict[str, Any]:
    """Classify text for PII detection."""
    if not text or text.strip() == "":
        return {"No input provided": 0.0}

    try:
        # Tokenize input
        inputs = tokenizer(
            text,
            return_tensors="pt",
            padding="max_length",
            max_length=512,
            truncation=True
        )

        # Run inference
        with torch.no_grad():
            outputs = model(**inputs)
            logits = outputs.logits
            probabilities = torch.sigmoid(logits)
            probs = probabilities.squeeze().tolist()

        # Create results dictionary
        results = {
            "Asking for PII": float(probs[0]),
            "Giving PII": float(probs[1])
        }

        return results

    except Exception as e:
        return {"Error": str(e)}


# Example test cases
examples = [
    ["Do you have the blue app?"],
    ["I live at 901 Roosevelt St, Redwood City"],
]


if __name__ == "__main__":
    # Model configuration
    model_id = "Roblox/PII-OSS-Private-Not-Public"

    # Load model and tokenizer
    # When deployed as a Hugging Face Space in the same organization,
    # authentication is handled automatically
    print(f"Loading model: {model_id}")
    try:
        model = AutoModelForSequenceClassification.from_pretrained(model_id)
        tokenizer = AutoTokenizer.from_pretrained(model_id)
        model.eval()
        print("Model loaded successfully!")
    except Exception as e:
        print(f"Failed to load model: {e}")
        print("If running locally, you may need to login with: huggingface-cli login")
        exit(1)

    # Create Gradio interface
    demo = gr.Interface(
        fn=predict,
        inputs=gr.Textbox(
            lines=3,
            placeholder="Enter text to analyze for PII content...",
            label="Input Text"
        ),
        outputs=gr.Label(
            num_top_classes=2,
            label="Classification Results"
        ),
        title="PII Detection Demo",
        description="This model detects whether text is asking for or giving personal information (PII).",
        examples=examples,
        flagging_mode="never",
    )

    demo.launch()