Update app.py

app.py

@@ -23,10 +23,11 @@ quant_config = BitsAndBytesConfig(
 tokenizer = LlamaTokenizer.from_pretrained(model_name)
 model = LlamaForCausalLM.from_pretrained(
     model_name,
-    quantization_config=quant_config
+    quantization_config=quant_config,
+    device_map="cpu"
 )
 
-model = PeftModel.from_pretrained(model, lora_name, device_map="cpu")
+model = PeftModel.from_pretrained(model, lora_name)
 
 model = model.to("cuda")