Update added quanitzation config

app.py

@@ -1,12 +1,14 @@
 import torch
 import gradio as gr
-from transformers import AutoProcessor, Blip2ForConditionalGeneration
+from transformers import AutoProcessor, Blip2ForConditionalGeneration, BitsAndBytesConfig
 from peft import LoraConfig, get_peft_model, PeftModel
 
+
 # Load the processor
 processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
 
 # Load the base model from the original repository
+quantization_config = BitsAndBytesConfig(load_in_8bit=True)
 base_model = Blip2ForConditionalGeneration.from_pretrained(
     "ybelkada/blip2-opt-2.7b-fp16-sharded",
     device_map="auto",