Create app.py

app.py

@@ -0,0 +1,26 @@
+import gradio as gr
+from transformers import pipeline
+
+# Replace this with your actual Hugging Face model repo (trained using classify.py)
+MODEL_NAME = "CIRCL/cwe-vulnerability-classification-codebert-base"
+
+# Load the text classification pipeline
+classifier = pipeline("text-classification", model=MODEL_NAME, return_all_scores=True)
+
+def classify_cwe(text):
+    results = classifier(text)[0]
+    # Sort by confidence score descending
+    sorted_results = sorted(results, key=lambda x: x["score"], reverse=True)
+    return {res["label"]: round(res["score"], 4) for res in sorted_results[:5]}
+
+# Create the Gradio interface
+interface = gr.Interface(
+    fn=classify_cwe,
+    inputs=gr.Textbox(lines=5, placeholder="Enter vulnerability description..."),
+    outputs=gr.Label(num_top_classes=5),
+    title="CWE Vulnerability Classifier",
+    description="Enter a vulnerability description to predict the most likely CWE types."
+)
+
+# Launch the Gradio app
+interface.launch()