Update app.py

app.py

@@ -16,6 +16,14 @@ import torch
 from emu3.mllm.processing_emu3 import Emu3Processor
 import spaces
 
+import subprocess
+# Install flash attention, skipping CUDA build if necessary
+subprocess.run(
+    "pip install flash-attn --no-build-isolation",
+    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
+    shell=True,
+)
+
 # Model paths
 EMU_GEN_HUB = "BAAI/Emu3-Gen"
 EMU_CHAT_HUB = "BAAI/Emu3-Chat"