scvi.autotune.AutotuneExperiment

class scvi.autotune.AutotuneExperiment(model_cls, data, metrics, mode, search_space, num_samples, scheduler='asha', searcher='hyperopt', seed=None, resources=None, name=None, logging_dir=None, scheduler_kwargs=None, searcher_kwargs=None)

BETA Track hyperparameter tuning experiments.

Parameters:

• model_cls (BaseModelClass) – Model class on which to tune hyperparameters. Must implement a constructor and a train method.

• data (Union[AnnData, MuData, LightningDataModule]) – AnnData or MuData that has been setup with model_cls or a LightningDataModule (EXPERIMENTAL).

• metrics (str | list[str]) – Either a single metric or a list of metrics to track during the experiment. If a list is provided, the primary metric will be the first element in the list.

• mode (Literal['min', 'max']) – Optimization mode for the primary metric. One of "min" or "max".

• search_space (dict[str, dict[Literal['model_args', 'train_args'], dict[str, Any]]]) –

  Dictionary of hyperparameter names and their respective search spaces. See the API for available search specifications. Must only contain the following top-level keys:

  • "model_args": parameters to pass to the model constructor.

  • "train_args": parameters to pass to the model’s train method.

  Passed into Tuner as param_space.

• num_samples (int) – Total number of hyperparameter configurations to sample from the search space. Passed into TuneConfig.

• scheduler (Literal['asha', 'hyperband', 'median', 'fifo'] (default: 'asha')) –

  Ray Tune scheduler to use. One of the following:

  • "asha": AsyncHyperBandScheduler

  • "hyperband": HyperBandScheduler

  • "median": MedianStoppingRule

  • "fifo": FIFOScheduler

  Configured with reasonable defaults, which can be overridden with scheduler_kwargs.

• searcher (Literal['hyperopt', 'random'] (default: 'hyperopt')) –

  Ray Tune search algorithm to use. One of the following:

  • "hyperopt": HyperOptSearch

  • "random": BasicVariantGenerator

  Configured with reasonable defaults, which can be overridden with searcher_kwargs.

• seed (int | None (default: None)) – Random seed to use for the experiment. Propagated to seed and search algorithms. If not provided, defaults to seed.

• resources (dict[Literal['cpu', 'gpu', 'memory'], float] | None (default: None)) –

  Dictionary of resources to allocate per trial in the experiment. Available keys include:

  • "cpu": number of CPU cores

  • "gpu": number of GPUs

  • "memory": amount of memory

  Passed into with_resources().

• name (str | None (default: None)) – Name of the experiment, used for logging purposes. Defaults to a unique ID.

• logging_dir (str | None (default: None)) – Base directory to store experiment logs. Defaults to :attr:scvi.settings.logging_dir.

• scheduler_kwargs (dict | None (default: None)) – Additional keyword arguments to pass to the scheduler.

• searcher_kwargs (dict | None (default: None)) – Additional keyword arguments to pass to the search algorithm.

Notes

Lifecycle: beta

See also

run_autotune()

Attributes table

data	Data on which to tune hyperparameters.
id	Unique identifier for the experiment.
logging_dir	Base directory to store experiment logs.
metrics	Metrics to track during the experiment.
metrics_callback
mode	Optimization mode for the primary metric.
model_cls	Model class on which to tune hyperparameters.
name	Name of the experiment.
num_samples	Total number of hyperparameter configurations to sample.
resources	Resources to allocate per trial in the experiment.
result_grid	Result grid for the experiment.
scheduler	Ray Tune scheduler to use.
scheduler_kwargs	Additional keyword arguments to pass to the scheduler.
search_space	Search space for hyperparameters.
searcher	Ray Tune search algorithm to use.
searcher_kwargs	Additional keyword arguments to pass to the search algorithm.
seed	Random seed to use for the experiment.
setup_method_args	Keyword arguments for the setup method.
setup_method_name	Either "setup_anndata" or "setup_mudata".

Methods table

get_logger(trial_name)	Configure TensorBoard logger for a trial in this experiment.
get_tuner()	Configure a Tuner from this experiment.

Attributes

AutotuneExperiment.data

Data on which to tune hyperparameters.

AutotuneExperiment.id

Unique identifier for the experiment.

AutotuneExperiment.logging_dir

Base directory to store experiment logs.

AutotuneExperiment.metrics

Metrics to track during the experiment.

AutotuneExperiment.metrics_callback

AutotuneExperiment.mode

Optimization mode for the primary metric.

AutotuneExperiment.model_cls

Model class on which to tune hyperparameters.

AutotuneExperiment.name

Name of the experiment.

AutotuneExperiment.num_samples

Total number of hyperparameter configurations to sample.

AutotuneExperiment.resources

Resources to allocate per trial in the experiment.

AutotuneExperiment.result_grid

Result grid for the experiment.

AutotuneExperiment.scheduler

Ray Tune scheduler to use.

AutotuneExperiment.scheduler_kwargs

Additional keyword arguments to pass to the scheduler.

AutotuneExperiment.search_space

Search space for hyperparameters.

AutotuneExperiment.searcher

Ray Tune search algorithm to use.

AutotuneExperiment.searcher_kwargs

Additional keyword arguments to pass to the search algorithm.

AutotuneExperiment.seed

Random seed to use for the experiment.

AutotuneExperiment.setup_method_args

Keyword arguments for the setup method.

AutotuneExperiment.setup_method_name

Either "setup_anndata" or "setup_mudata".

Methods

AutotuneExperiment.get_logger(trial_name)

Configure TensorBoard logger for a trial in this experiment.

Return type:

TensorBoardLogger

AutotuneExperiment.get_tuner()

Configure a Tuner from this experiment.

Return type:

Tuner