langchain.Tracer

Tracing and instrumentation utilities for Langchain integration.

View module source on GitHub

MaximLangchainTracer

class MaximLangchainTracer(BaseCallbackHandler)

A callback handler that logs langchain outputs to Maxim logger

Arguments:

Name	Description
logger	[Logger](/sdk/python/references/logger/logger): Maxim Logger instance to log outputs

__init__

def __init__(logger: Logger,
             metadata: Optional[Dict[str, Any]] = None,
             eval_config: Optional[Dict[str, List[str]]] = None) -> None

Initializes the Langchain Tracer

Arguments:

Name	Description
logger	[Logger](/sdk/python/references/logger/logger): Maxim Logger instance to log outputs

on_llm_start

def on_llm_start(serialized: dict[str, Any],
                 prompts: list[str],
                 *,
                 run_id: UUID,
                 parent_run_id: Optional[UUID] = None,
                 tags: Optional[list[str]] = None,
                 metadata: Optional[dict[str, Any]] = None,
                 **kwargs: Any) -> Any

Runs when LLM starts

on_chat_model_start

def on_chat_model_start(serialized: Dict[str, Any],
                        messages: List[List[BaseMessage]],
                        metadata: Optional[Dict[str, Any]] = None,
                        **kwargs: Any) -> Any

Runs when a chat model call starts

on_llm_new_token

def on_llm_new_token(token: str, **kwargs: Any) -> Any

Run on new LLM token. Only available when streaming is enabled.

on_llm_end

def on_llm_end(response: LLMResult, **kwargs: Any) -> Any

Run when LLM ends running.

on_llm_error

def on_llm_error(error: Union[Exception, BaseException, KeyboardInterrupt],
                 **kwargs: Any) -> Any

Run when LLM errors.

on_retriever_start

def on_retriever_start(serialized: dict[str, Any],
                       query: str,
                       *,
                       run_id: UUID,
                       parent_run_id: Optional[UUID] = None,
                       tags: Optional[list[str]] = None,
                       metadata: Optional[dict[str, Any]] = None,
                       **kwargs: Any) -> Any

Run when Retriever starts running.