Update app.py

app.py

@@ -11,7 +11,7 @@ model = AutoModelForCausalLM.from_pretrained(
     return_dict=True,
     load_in_8bit=False,
     device_map=torch.device("cuda") if torch.cuda.is_available() else "cpu",
-    load_in_8bit_fp32_cpu_offload=True
+    #load_in_8bit_fp32_cpu_offload=True
 )
 tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)