load model change

app.py

@@ -11,29 +11,33 @@ login(access_token)
 model_id = "google/gemma-2-9b-it"
 tokenizer = None
 model = None
+model_loaded = False  # Flag to check if the model is loaded
 
 @spaces.GPU
 def load_model():
-    global tokenizer, model
-    print("Model loading started")
-    tokenizer = AutoTokenizer.from_pretrained(model_id)
-    model = AutoModelForCausalLM.from_pretrained(
-        model_id,
-        device_map="auto",
-        torch_dtype=torch.bfloat16,
-    )
-    print("Model loading completed. Device of the model:", model.device)
-
-load_model()
+    global tokenizer, model, model_loaded
+    if not model_loaded:  # Load model only if it's not already loaded
+        print("Model loading started")
+        tokenizer = AutoTokenizer.from_pretrained(model_id)
+        model = AutoModelForCausalLM.from_pretrained(
+            model_id,
+            device_map="auto",
+            torch_dtype=torch.bfloat16,
+        )
+        model_loaded = True
+        print("Model loading completed. Device of the model:", model.device)
+        return model, tokenizer
+    else:
+        print("Model is already loaded")
+        return model, tokenizer
 
 @spaces.GPU
 def ask(prompt):
-
-    global tokenizer, model
-
     if not prompt:
         return {"error": "Prompt is missing"}
 
+    if not model_loaded:
+        model, tokenizer = load_model()  # Ensure the model is loaded before processing
     print("Device of the model:", model.device)
     messages = [
         {"role": "user", "content": f"{prompt}"},