Update app.py

app.py

@@ -1,9 +1,12 @@
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
+import os
+
+os.system('mkdir buff')
 
 # Load model and tokenizer
 model_name = "abhifdsdf/good_disease_8bit"  # Replace with your Hugging Face model repo name or local path
-model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", low_cpu_mem_usage=True)
+model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", low_cpu_mem_usage=True, offload_folder="buff")
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 # Define inference function