Update app.py

app.py

@@ -15,6 +15,7 @@ from transformers import (
     AutoProcessor,
     TextIteratorStreamer,
 )
+
 from gradio.themes import Soft
 from gradio.themes.utils import colors, fonts, sizes
 
@@ -27,6 +28,7 @@ except subprocess.CalledProcessError as e:
     print(f"Error installing flash-attn: {e}")
     print("Continuing without flash-attn.")
 
+import flash_attn_2_cuda as flash_attn_gpu
 
 colors.steel_blue = colors.Color(
     name="steel_blue",