Skip to main content

Documentation Index

Fetch the complete documentation index at: https://www.getmaxim.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

View module source on GitHub

​
current_generation

def current_generation() -> Optional[Generation]
Get the current generation from the context variable. Returns:
NameDescription
Optional[[Generation](/sdk/python/references/logger/components/generation)]The current generation instance if one exists,
otherwise None.

​
generation

def generation(logger: Optional[Logger] = None,
               id: Optional[Union[Callable, str]] = None,
               name: Optional[str] = None,
               maxim_prompt_id: Optional[str] = None,
               tags: Optional[Dict[str, str]] = None,
               evaluators: Optional[List[str]] = None,
               evaluator_variables: Optional[Dict[str, str]] = None)
Decorator for tracking AI model generations with Maxim logging. This decorator wraps functions to automatically create and manage Generation objects for tracking AI model calls, including inputs, outputs, and metadata. The decorated function must be called within a @trace or @span decorated context. Arguments:
NameTypeDescription
loggerOptional[Logger]Maxim logger instance. If None, uses the current
logger from context.
  • id Optional[Union[Callable, str]] - Generation ID. Can be a string or a callable that returns a string. If None, generates a UUID.
  • name Optional[str] - Human-readable name for the generation.
  • maxim_prompt_id Optional[str] - ID of the Maxim prompt template used.
  • tags Optional[Dict[str, str]] - Key-value pairs for tagging the generation.
  • evaluators Optional[List[str]] - List of evaluator names to run on this generation.
  • evaluator_variables Optional[Dict[str, str]] - Variables to pass to evaluators.
Returns:
NameDescription
CallableThe decorator function that wraps the target function.
Raises:
  • ValueError - If no logger is found or if called outside of a trace/span context when raise_exceptions is True.
Example: 
import maxim

logger = maxim.Logger(api_key="your-api-key")

@logger.trace()
def my_ai_workflow():
    result = generate_text("Hello world")
    return result

@generation(
    name="text_generation",
    tags={"model": "gpt-4", "temperature": "0.7"},
    evaluators=["coherence", "relevance"]
)
def generate_text(prompt: str) -> str:
    # Your AI generation logic here
    return "Generated response"