Update app.py

app.py

@@ -35,9 +35,11 @@ model_name = "daekeun-ml/Llama-2-ko-instruct-13B"
 
 #daekeun-ml/Llama-2-ko-instruct-13B
 #quantumaikr/llama-2-70b-fb16-korean
-
 tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoDistributedModelForCausalLM.from_pretrained(model_name)
+
+def init():
+    if check:
+        model = AutoDistributedModelForCausalLM.from_pretrained(model_name)
 
 
 def check(model_name):
@@ -51,7 +53,7 @@ def check(model_name):
 
 
 def chat(id, npc, prompt):
-
+    
     # get_coin endpoint
     response = requests.post("https://ldhldh-api-for-unity.hf.space/run/predict_6", json={
       "data": [
@@ -63,6 +65,8 @@ def chat(id, npc, prompt):
         return "no coin"
     
     # model inference
+
+    init()
     if check:
         prom = ""
         inputs = tokenizer(prom, return_tensors="pt")["input_ids"]