Update app.py

app.py

@@ -28,20 +28,29 @@ def respond(
     top_p,
 ):
     from llama_cpp import Llama
+    from llama_cpp_agent import LlamaCppAgent
+    from llama_cpp_agent import MessagesFormatterType
+    from llama_cpp_agent.providers import LlamaCppPythonProvider
+    
     llm = Llama(
         model_path="models/mistral-7b-instruct-v0.2.Q6_K.gguf",
         n_gpu_layers=33,
     )
-    stream = llm.create_chat_completion(
-        messages = [
-            {"role": "system", "content": f"{system_message}"},
-            {
-                "role": "user",
-                "content": f"{message}"
-            }
-        ],
-        stream=True,
+    provider = LlamaCppPythonProvider(llm)
+
+    agent = LlamaCppAgent(
+        provider,
+        system_prompt="You are a helpful assistant.",
+        predefined_messages_formatter_type=MessagesFormatterType.MISTRAL,
+        debug_output=True
     )
+    
+    settings = provider.get_provider_default_settings()
+    settings.max_tokens = 2000
+    settings.stream = True
+    
+    stream = agent.get_chat_response(message, llm_sampling_settings=settings, returns_streaming_generator=True)
+    
     outputs = ""
     for output in stream:
         print(output)