feat: Fix Gradio warnings and implement ZeroGPU startup hook for vision pipeline

app.py

@@ -8,19 +8,32 @@ try:
 except FileNotFoundError:
     custom_css = ""
 
-# Build the Gradio App with a modern theme and our custom CSS injected
-demo = gr.Blocks(
-    css=custom_css, 
-    theme=gr.themes.Default(
-        primary_hue="blue", 
-        neutral_hue="slate",
-        font=[gr.themes.GoogleFont("Inter"), "sans-serif"]
-    )
+# Build the Gradio App
+demo = gr.Blocks()
+my_theme = gr.themes.Default(
+    primary_hue="blue", 
+    neutral_hue="slate"
 )
 
 with demo:
     # Initialize the UI layout from the ui folder
-    create_ui()
+    image_input, audio_input, submit_btn, chatbot, msg_input = create_ui()
+    
+    # Bind the submit button to the vision function
+    def handle_analyze(image):
+        from tools.vision import process_receipt_image
+        if not image:
+            return [("System", "Please upload an image first!")]
+        
+        # Trigger the ZeroGPU
+        result = process_receipt_image(image)
+        return [("User", "Uploaded a bill..."), ("Agent", result)]
+        
+    submit_btn.click(
+        fn=handle_analyze,
+        inputs=[image_input],
+        outputs=[chatbot]
+    )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(theme=my_theme, css=custom_css)

tools/vision.py

@@ -0,0 +1,15 @@
+import spaces
+import torch
+
+@spaces.GPU
+def process_receipt_image(image_path):
+    """
+    Milestone 2: Vision Pipeline.
+    For now, this just verifies the ZeroGPU is working.
+    We will load the massive MiniCPM-V-2_6 model here next.
+    """
+    if torch.cuda.is_available():
+        gpu_name = torch.cuda.get_device_name(0)
+        return f"Success! ZeroGPU is awake. Allocated: {gpu_name}. Ready to load MiniCPM-V!"
+    else:
+        return "Warning: Running on CPU. ZeroGPU did not wake up."

ui/layout.py

@@ -24,4 +24,5 @@ def create_ui():
                 chatbot = gr.Chatbot(label="Agent", elem_classes=["chat-window"], height=500)
                 msg_input = gr.Textbox(label="Ask a question", placeholder="e.g. How much did I spend on junk food?", elem_classes=["chat-input"])
 
-    # TODO: In Milestone 2 & 3, we will bind functions to the submit_btn and msg_input.
+    # Return the components so app.py can bind events to them
+    return image_input, audio_input, submit_btn, chatbot, msg_input