🐫 CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org
Python agentai-societiesartificial-intelligencecommunicative-aicooperative-aideep-learninglarge-language-modelsmulti-agent-systemsnatural-language-processing
fix: prevent NameError in _arun method when agent_session_id exists (#3630)
Lei Zhangd3ebb4c9e66d
diff --git a/camel/models/openai_model.py b/camel/models/openai_model.pyindex f5a890ae1..316dee400 100644--- a/camel/models/openai_model.py+++ b/camel/models/openai_model.py@@ -270,6 +270,24 @@ class OpenAIModel(BaseModelBackend):self._token_counter = OpenAITokenCounter(self.model_type)return self._token_counter+ def _log_and_trace(self) -> None:+ r"""Update Langfuse trace with session metadata and log it."""+ agent_session_id = get_current_agent_session_id() or "no-session-id"+ model_type_str = str(self.model_type)+ metadata = {+ "source": "camel",+ "agent_id": agent_session_id,+ "agent_type": "camel_chat_agent",+ "model_type": model_type_str,+ }+ metadata = {k: str(v) for k, v in metadata.items()}+ update_langfuse_trace(+ session_id=agent_session_id,+ metadata=metadata,+ tags=["CAMEL-AI", model_type_str],+ )+ logger.info(f"metadata: {metadata}")+@observe()def _run(self,@@ -299,26 +317,7 @@ class OpenAIModel(BaseModelBackend):or `ChatCompletionStreamManager[BaseModel]` forstructured output streaming."""-- # Update Langfuse trace with current agent session and metadata- agent_session_id = get_current_agent_session_id()- model_type_str = str(self.model_type)- if not agent_session_id:- agent_session_id = "no-session-id"- metadata = {- "source": "camel",- "agent_id": agent_session_id,- "agent_type": "camel_chat_agent",- "model_type": model_type_str,- }- metadata = {k: str(v) for k, v in metadata.items()}- if agent_session_id:- update_langfuse_trace(- session_id=agent_session_id,- metadata=metadata,- tags=["CAMEL-AI", model_type_str],- )- logger.info(f"metadata: {metadata}")+ self._log_and_trace()messages = self._adapt_messages_for_o1_models(messages)response_format = response_format or self.model_config_dict.get(@@ -371,25 +370,7 @@ class OpenAIModel(BaseModelBackend):`AsyncChatCompletionStreamManager[BaseModel]` forstructured output streaming."""-- # Update Langfuse trace with current agent session and metadata- agent_session_id = get_current_agent_session_id()- model_type_str = str(self.model_type)- if not agent_session_id:- agent_session_id = "no-session-id"- metadata = {- "source": "camel",- "agent_id": agent_session_id,- "agent_type": "camel_chat_agent",- "model_type": model_type_str,- }- metadata = {k: str(v) for k, v in metadata.items()}- update_langfuse_trace(- session_id=agent_session_id,- metadata=metadata,- tags=["CAMEL-AI", model_type_str],- )- logger.info(f"metadata: {metadata}")+ self._log_and_trace()messages = self._adapt_messages_for_o1_models(messages)response_format = response_format or self.model_config_dict.get(