Create app.py

app.py

@@ -0,0 +1,30 @@
+import torch
+from transformers import BertTokenizer, BertForSequenceClassification
+import gradio as gr
+
+# Load fine-tuned model
+model = BertForSequenceClassification.from_pretrained("bert-expense-classifier")
+tokenizer = BertTokenizer.from_pretrained("bert-expense-classifier")
+
+model.eval()
+
+label_map = {0: "statement", 1: "question"}
+
+def classify_sentence(text):
+    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True, max_length=128)
+    with torch.no_grad():
+        outputs = model(**inputs)
+        predicted_class = torch.argmax(outputs.logits, dim=1).item()
+    return label_map[predicted_class]
+
+# Gradio Interface
+interface = gr.Interface(
+    fn=classify_sentence,
+    inputs=gr.Textbox(lines=2, placeholder="Enter a sentence..."),
+    outputs=gr.Textbox(label="Prediction"),
+    title="Expense Sentence Classifier",
+    description="Classifies whether a sentence is a user question or a statement for an expense tracker."
+)
+
+if __name__ == "__main__":
+    interface.launch()