Update app.py

app.py

@@ -1,27 +1,35 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+from fastapi import FastAPI, Request
+from gradio.routes import mount_gradio_app
 
-# Create inference client for your model (no need for token if model is public)
 client = InferenceClient("dlucidone/kumaoni-mbart-lora")
 
-def translate(text):
+def translate(text: str):
     if not text.strip():
         return "Please enter text."
     try:
-        # Use the correct API method for text translation
-        result = client.text_generation(text, max_new_tokens=256)
-        return result.strip()
+        response = client.post(json={"inputs": text})
+        if isinstance(response, list) and len(response) and "generated_text" in response[0]:
+            return response[0]["generated_text"]
+        return str(response)
     except Exception as e:
         return f"Error: {e}"
 
-# Gradio interface for UI + API
-iface = gr.Interface(
-    fn=translate,
-    inputs=gr.Textbox(label="text", placeholder="Enter text to translate"),
-    outputs=gr.Textbox(label="output"),
-    title="Kumaoni Translator API"
-)
+# Create a FastAPI app
+api = FastAPI()
 
-# Enable API access for frontend fetch() calls
-iface.queue(api_open=True)
-iface.launch()
+# Add /predict endpoint
+@api.post("/predict")
+async def predict(request: Request):
+    body = await request.json()
+    text = body.get("text", "")
+    return {"translation": translate(text)}
+
+# Mount Gradio interface for human testing
+gr_app = gr.Interface(fn=translate, inputs="text", outputs="text", title="Kumaoni Translator")
+app = mount_gradio_app(api, gr_app, path="/")
+
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)