Update app.py

app.py

@@ -70,7 +70,10 @@ def initialize_settings(model_path):
     Settings.llm = LlamaCPP(
         model_path=model_path,
         temperature=0.7,
-        n_gpu_layers=20  # 🔥 Mempercepat inferensi dengan GPU jika tersedia
+        context_window=4096,
+        max_new_tokens=512,
+        # n_gpu_layers=20,  # ❌ Hapus jika error
+        model_kwargs={"n_ctx": 4096}
     )
 
 # ===================================