Update app.py

app.py

@@ -1,13 +1,19 @@
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
+import tempfile
+import os
+
+# Create the offload folder
+offload_dir = './offload'
+os.makedirs(offload_dir, exist_ok=True)
 
 model = AutoModelForCausalLM.from_pretrained(
     "tiiuae/falcon-7b-instruct",
     torch_dtype=torch.bfloat16,
     device_map="auto",
     low_cpu_mem_usage=True,
-    offload_folder="/offload_folder",
+    offload_folder=offload_dir
 )
 tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-7b-instruct")