Update llm_handler.py

llm_handler.py

@@ -22,10 +22,39 @@ settings = provider.get_provider_default_settings()
 settings.max_tokens = 2000
 settings.stream = True
 
+from llama_cpp import Llama
+from llama_cpp_agent import LlamaCppAgent
+from llama_cpp_agent import MessagesFormatterType
+from llama_cpp_agent.providers import LlamaCppPythonProvider
+
+# Initialize the Llama model
+llama_model = Llama("Arcee-Spark-GGUF/Arcee-Spark-Q4_K_M.gguf", n_batch=1024, n_threads=10, n_gpu_layers=33, n_ctx=2048, verbose=False)
+
+# Create the provider
+provider = LlamaCppPythonProvider(llama_model)
+
+# Create the agent
+agent = LlamaCppAgent(
+    provider,
+    system_prompt="You are a helpful assistant.",
+    predefined_messages_formatter_type=MessagesFormatterType.CHATML,
+    debug_output=True
+)
+
+# Set provider settings
+settings = provider.get_provider_default_settings()
+settings.max_tokens = 2000
+settings.stream = True
+
 def send_to_llm(provider, msg_list):
     try:
-        response = agent.get_chat_response(msg_list, settings=settings)
+
+        provider.apply_settings(settings)
+        
+        # Call get_chat_response without the settings parameter
+        response = agent.get_chat_response(msg_list)
         return response.content, None  # We don't have usage info in this case
     except Exception as e:
         print(f"Error in send_to_llm: {str(e)}")
         return f"Error: {str(e)}", None
+