Update app.py

app.py

@@ -5,14 +5,12 @@ import re
 import time
 import torch
 import spaces
-import subprocess
-subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
+
 
 from io import BytesIO
 
 processor = AutoProcessor.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct")
 model = AutoModelForImageTextToText.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct", 
-                                               _attn_implementation="flash_attention_2",
                                                torch_dtype=torch.bfloat16).to("cuda:0")