Update app.py

app.py

@@ -6,6 +6,9 @@ import os
 import numpy as np
 import spaces
 
+import subprocess
+subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
+
 model = AutoModelForCausalLM.from_pretrained('microsoft/Florence-2-large', trust_remote_code=True).to("cuda").eval()
 processor = AutoProcessor.from_pretrained('microsoft/Florence-2-large', trust_remote_code=True)