diff --git a/letta/llm_api/google_vertex_client.py b/letta/llm_api/google_vertex_client.py index 028871db..49941696 100644 --- a/letta/llm_api/google_vertex_client.py +++ b/letta/llm_api/google_vertex_client.py @@ -239,7 +239,7 @@ class GoogleVertexClient(LLMClientBase): request_data["config"]["response_mime_type"] = "application/json" request_data["config"]["response_schema"] = self.get_function_call_response_schema(tools[0]) del request_data["config"]["tools"] - else: + elif tools: tool_config = ToolConfig( function_calling_config=FunctionCallingConfig( # ANY mode forces the model to predict only function calls diff --git a/letta/services/summarizer/summarizer.py b/letta/services/summarizer/summarizer.py index 699998e6..e9c65d0c 100644 --- a/letta/services/summarizer/summarizer.py +++ b/letta/services/summarizer/summarizer.py @@ -348,12 +348,9 @@ async def simple_summary(messages: List[Message], llm_config: LLMConfig, actor: {"role": "system", "content": system_prompt}, {"role": "user", "content": summary_transcript}, ] - print("messages going to summarizer:", input_messages) input_messages_obj = [simple_message_wrapper(msg) for msg in input_messages] - print("messages going to summarizer (objs):", input_messages_obj) - request_data = llm_client.build_request_data(input_messages_obj, llm_config, tools=[]) - print("request data:", request_data) + # NOTE: we should disable the inner_thoughts_in_kwargs here, because we don't use it # I'm leaving it commented it out for now for safety but is fine assuming the var here is a copy not a reference # llm_config.put_inner_thoughts_in_kwargs = False