Update handler.py

handler.py

@@ -1,10 +1,13 @@
 import torch
-from transformers import AutoModelForVision2Seq, AutoProcessor, BitsAndBytesConfig
 import sys
 from subprocess import run
 from PIL import Image
 
 run("pip install flash-attn --no-build-isolation", shell=True, check=True)
+run("pip install bitsandbytes", shell=True, check=True)
+
+from transformers import AutoModelForVision2Seq, AutoProcessor, BitsAndBytesConfig
+
 model_id = "ibm-granite/granite-vision-3.2-2b"
 
 bnb_config = BitsAndBytesConfig(
@@ -16,7 +19,6 @@ bnb_config = BitsAndBytesConfig(
         llm_int8_enable_fp32_cpu_offload=True
 )
 
-
 try:
     import flash_attn
     print("FlashAttention is installed")