Update app.py

app.py

@@ -87,8 +87,9 @@ model = model.merge_and_unload()
 print("phi-3 model loaded sucessfully")  
 # 3. Inference Function
 
-@spaces.GPU
+
 # 3. Inference Function
+@spaces.GPU
 def predict(image, question):
     """
     Takes an image and a question as input and returns an answer.
@@ -111,9 +112,9 @@ def predict(image, question):
         prompt = f"Question: {question}\nImage Embeddings: {image_embeddings_str}\nAnswer:"
 
         # Generate answer using llama.cpp
-        output = llm(
+        output = model(
             prompt,
-            max_tokens=200,
+            max_tokens=128,
             stop=["Q:", "\n"],
             echo=False,
         )