Hugging Face's logo

Accelerate documentation

Experiment Trackers

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v1.14.0).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

Experiment Trackers

GeneralTracker

class accelerate.tracking.GeneralTracker

< >

( _blank = False )

A base Tracker class to be used for all logging integration implementations.

Each function should take in **kwargs that will automatically be passed in from a base dictionary provided to Accelerator.

Should implement name, requires_logging_directory, and tracker properties such that:

name (str): String representation of the tracker class name, such as “TensorBoard” requires_logging_directory (bool): Whether the logger requires a directory to store their logs. tracker (object): Should return internal tracking mechanism used by a tracker class (such as the run for wandb)

Implementations can also include a main_process_only (bool) attribute to toggle if relevant logging, init, and other functions should occur on the main process or across all processes (by default will use True)

finish

< >

( )

Should run any finalizing functions within the tracking API. If the API should not have one, just don’t overwrite that method.

log

< >

( values: dictstep: typing.Optional[int] = None**kwargs )

Parameters

  • values (Dictionary str to str, float, or int) — Values to be logged as key-value pairs. The values need to have type str, float, or int.
  • step (int, optional) — The run step. If included, the log will be affiliated with this step.

Logs values to the current run. Base log implementations of a tracking API should go in here, along with special behavior for the `step parameter.

start

< >

( )

Lazy initialization of the tracker inside Accelerator to avoid initializing PartialState before InitProcessGroupKwargs.

store_init_configuration

< >

( values: dict )

Parameters

  • values (Dictionary str to bool, str, float or int) — Values to be stored as initial hyperparameters as key-value pairs. The values need to have type bool, str, float, int, or None.

Logs values as hyperparameters for the run. Implementations should use the experiment configuration functionality of a tracking API.

register_tracker_class

accelerate.tracking.register_tracker_class

< >

( tracker_class: type )

Parameters

  • tracker_class (subclass of GeneralTracker) — The tracker class to register. It must subclass GeneralTracker and define a non-empty name class attribute and a requires_logging_directory class attribute. The class is instantiated with the project_name passed to Accelerator.init_trackers() as the first positional argument. When requires_logging_directory is True, the logging directory is passed as the second positional argument (matching the built-in tracker convention). Tracker-specific keyword arguments from init_kwargs are forwarded as **kwargs.

Registers a custom GeneralTracker subclass so it can be referenced by its name in the log_with argument of Accelerator, the same way as the built-in trackers.

The tracker must be registered before instantiating the Accelerator that uses it.

Example:

from accelerate import Accelerator
from accelerate.tracking import GeneralTracker, register_tracker_class


class MyTracker(GeneralTracker):
    name = "my_tracker"
    requires_logging_directory = False
    # ... implement the rest of the `GeneralTracker` interface


register_tracker_class(MyTracker)
accelerator = Accelerator(log_with="my_tracker")

TensorBoardTracker

class accelerate.tracking.TensorBoardTracker

< >

( run_name: strlogging_dir: typing.Union[str, os.PathLike]**kwargs )

Parameters

  • run_name (str) — The name of the experiment run
  • logging_dir (str, os.PathLike) — Location for TensorBoard logs to be stored.
  • **kwargs (additional keyword arguments, optional) — Additional key word arguments passed along to the tensorboard.SummaryWriter.__init__ method.

A Tracker class that supports tensorboard. Should be initialized at the start of your script.

__init__

< >

( run_name: strlogging_dir: typing.Union[str, os.PathLike]**kwargs )

WandBTracker

class accelerate.tracking.WandBTracker

< >

( run_name: str**kwargs )

Parameters

  • run_name (str) — The name of the experiment run.
  • **kwargs (additional keyword arguments, optional) — Additional key word arguments passed along to the wandb.init method.

A Tracker class that supports wandb. Should be initialized at the start of your script.

__init__

< >

( run_name: str**kwargs )

Trackio

class accelerate.tracking.TrackioTracker

< >

( run_name: str**kwargs )

Parameters

  • run_name (str) — The name of the experiment run. Will be used as the project name when instantiating trackio.
  • **kwargs (additional keyword arguments, optional) — Additional key word arguments passed along to the trackio.init method. Refer to this init to see all supported key word arguments.

A Tracker class that supports trackio. Should be initialized at the start of your script.

__init__

< >

( run_name: str**kwargs )

CometMLTracker

class accelerate.tracking.CometMLTracker

< >

( run_name: str**kwargs )

Parameters

A Tracker class that supports comet_ml. Should be initialized at the start of your script.

API keys must be stored in a Comet config file.

Note: For comet_ml versions < 3.41.0, additional keyword arguments are passed to comet_ml.Experiment instead: https://www.comet.com/docs/v2/api-and-sdk/python-sdk/reference/Experiment/#comet_ml.Experiment.__init__

__init__

< >

( run_name: str**kwargs )

AimTracker

class accelerate.tracking.AimTracker

< >

( run_name: strlogging_dir: typing.Union[str, os.PathLike, NoneType] = '.'**kwargs )

Parameters

  • run_name (str) — The name of the experiment run.
  • **kwargs (additional keyword arguments, optional) — Additional key word arguments passed along to the Run.__init__ method.

A Tracker class that supports aim. Should be initialized at the start of your script.

__init__

< >

( run_name: strlogging_dir: typing.Union[str, os.PathLike, NoneType] = '.'**kwargs )

MLflowTracker

class accelerate.tracking.MLflowTracker

< >

( experiment_name: typing.Optional[str] = Nonelogging_dir: typing.Union[str, os.PathLike, NoneType] = Nonerun_id: typing.Optional[str] = Nonetags: typing.Union[dict[str, typing.Any], str, NoneType] = Nonenested_run: typing.Optional[bool] = Falserun_name: typing.Optional[str] = Nonedescription: typing.Optional[str] = None )

Parameters

  • experiment_name (str, optional) — Name of the experiment. Environment variable MLFLOW_EXPERIMENT_NAME has priority over this argument.
  • logging_dir (str or os.PathLike, defaults to ".") — Location for mlflow logs to be stored.
  • run_id (str, optional) — If specified, get the run with the specified UUID and log parameters and metrics under that run. The run’s end time is unset and its status is set to running, but the run’s other attributes (source_version, source_type, etc.) are not changed. Environment variable MLFLOW_RUN_ID has priority over this argument.
  • tags (Dict[str, str], optional) — An optional dict of str keys and values, or a str dump from a dict, to set as tags on the run. If a run is being resumed, these tags are set on the resumed run. If a new run is being created, these tags are set on the new run. Environment variable MLFLOW_TAGS has priority over this argument.
  • nested_run (bool, optional, defaults to False) — Controls whether run is nested in parent run. True creates a nested run. Environment variable MLFLOW_NESTED_RUN has priority over this argument.
  • run_name (str, optional) — Name of new run (stored as a mlflow.runName tag). Used only when run_id is unspecified.
  • description (str, optional) — An optional string that populates the description box of the run. If a run is being resumed, the description is set on the resumed run. If a new run is being created, the description is set on the new run.

A Tracker class that supports mlflow. Should be initialized at the start of your script.

__init__

< >

( experiment_name: typing.Optional[str] = Nonelogging_dir: typing.Union[str, os.PathLike, NoneType] = Nonerun_id: typing.Optional[str] = Nonetags: typing.Union[dict[str, typing.Any], str, NoneType] = Nonenested_run: typing.Optional[bool] = Falserun_name: typing.Optional[str] = Nonedescription: typing.Optional[str] = None )

ClearMLTracker

class accelerate.tracking.ClearMLTracker

< >

( run_name: typing.Optional[str] = None**kwargs )

Parameters

  • run_name (str, optional) — Name of the experiment. Environment variables CLEARML_PROJECT and CLEARML_TASK have priority over this argument.
  • **kwargs (additional keyword arguments, optional) — Kwargs passed along to the Task.__init__ method.

A Tracker class that supports clearml. Should be initialized at the start of your script.

__init__

< >

( run_name: typing.Optional[str] = None**kwargs )

SwanLabTracker

class accelerate.tracking.SwanLabTracker

< >

( run_name: str**kwargs )

Parameters

  • run_name (str) — The name of the experiment run.
  • **kwargs (additional keyword arguments, optional) — Additional key word arguments passed along to the swanlab.init method.

A Tracker class that supports swanlab. Should be initialized at the start of your script.

__init__

< >

( run_name: str**kwargs )

Update on GitHub

←DataLoaders, Optimizers, Schedulers Launchers→