app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForSequenceClassification
from huggingface_hub import hf_hub_download
import torch
import json

def predict(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)

    with torch.no_grad():
        outputs = model(**inputs)

    probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
    predicted_class = torch.argmax(probs, dim=-1).item()
    return id2label[predicted_class], probs[0][predicted_class].item()

if __name__ == '__main__':
    model_path = "Dunateo/roberta-cwe-classifier-kelemia-v0.2"

    # init the model
    tokenizer = AutoTokenizer.from_pretrained(model_path)
    model = AutoModelForSequenceClassification.from_pretrained(model_path)

    # get the dict file
    label_dict_file = hf_hub_download(repo_id=model_path, filename="label_dict.json")


    with open(label_dict_file, "r") as f:
        content = f.read()
    label_dict = json.loads(content)

    global id2label
    id2label = {v: k for k, v in label_dict.items()}

    # gradio specific to create an IHM
    iface = gr.Interface(
        fn=predict,
        inputs=gr.Textbox(lines=5, label="Enter vulnerability description"),
        outputs=[gr.Label(label="Predicted CWE"), gr.Number(label="Confidence")],
        title="Vulnerability CWE Classification",
        description="Enter a vulnerability description to classify it into a CWE category."
    )

    iface.launch()