Update app.py

app.py

@@ -44,7 +44,8 @@ PROMPT_DICT = {
         "Instruction:\n{instruction}\n\nResponse:"
     ),
 }
-
+model = None
+tokenizer = None
 def generate_prompt(instruction, input=None):
     if input:
         return PROMPT_DICT["prompt_input"].format(instruction=instruction,input=input)
@@ -68,7 +69,6 @@ def generator(input_ids, generation_config, max_new_tokens):
     return generation_output
     
 def loadModel():
-    global model, tokenizer
     if model is None:
         from llama_rope_scaled_monkey_patch import replace_llama_rope_with_scaled_rope
         replace_llama_rope_with_scaled_rope()