Update app.py

app.py

@@ -75,6 +75,7 @@ def highlight_text(text, text_list,color="black"):
 
 @spaces.GPU(duration=30)
 def warmup(model):
+    model.cuda()
     conv = get_conversation_template(args.model_type)
 
     if args.model_type == "llama-2-chat":
@@ -277,7 +278,7 @@ model = LlamaForCausalLM.from_pretrained(
     load_in_8bit=args.load_in_8bit,
     device_map="auto",
 )
-model.cuda()
+
 model.eval()
 tokenizer=AutoTokenizer.from_pretrained(args.base_model_path)
 warmup(model)