Create app.py

app.py

@@ -0,0 +1,55 @@
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import gradio as gr
+import torch
+
+# Load the model and tokenizer from Hugging Face Hub
+model_name = "julian-schelb/xlm-roberta-base-latin-intertextuality"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
+model = model.to(device)
+
+
+def predict_intertextuality(sentence1, sentence2):
+    """
+    Predict intertextuality using the specified model.
+    """
+    # Prepare input for the model
+    inputs = tokenizer(
+        sentence1,
+        sentence2,
+        return_tensors="pt",
+        truncation=True,
+        padding="max_length",
+        max_length=512  # Adjust based on model's configuration
+    ).to(device)
+
+    # Perform inference
+    model.eval()
+    with torch.no_grad():
+        outputs = model(**inputs)
+        logits = outputs.logits
+        probs = torch.softmax(logits, dim=1).squeeze().cpu().numpy()
+
+    # Map probabilities to labels
+    return {"Yes": probs[1], "No": probs[0]}
+
+
+# Define the Gradio interface
+inputs = [
+    gr.Textbox(label="Latin Sentence 1"),
+    gr.Textbox(label="Latin Sentence 2")
+]
+outputs = gr.Label(label="Intertextuality Probabilities", num_top_classes=2)
+
+gradio_app = gr.Interface(
+    fn=predict_intertextuality,
+    inputs=inputs,
+    outputs=outputs,
+    title="Latin Intertextuality Checker",
+    description="Enter two Latin sentences to get the probabilities for 'Yes' (intertextual) or 'No' (not intertextual).",
+    # flagging="never"  # Disable the flag button
+)
+
+if __name__ == "__main__":
+    gradio_app.launch()