using transformers to handle the model

app.py

@@ -8,6 +8,7 @@ from pathlib import Path
 from transformers import AutoProcessor, AutoModelForVision2Seq
 import subprocess
 import logging
+subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)