Update app.py

app.py

@@ -2,8 +2,13 @@
 import subprocess
 import sys
 
-# Install flash-attn for GPU only
+# CRITICAL: Import spaces FIRST before any CUDA initialization
+import spaces
+
+# Now we can import torch and other packages
 import torch
+
+# Install flash-attn for GPU only (after spaces import)
 if torch.cuda.is_available():
     print("CUDA detected - installing flash-attn for optimal GPU performance...")
     subprocess.run(
@@ -13,7 +18,6 @@ if torch.cuda.is_available():
     )
 
 import gradio as gr
-import spaces
 from PIL import Image
 from io import BytesIO
 import pypdfium2 as pdfium