- class ray.tune.TuneConfig(mode: str | None = None, metric: str | None = None, search_alg: Searcher | SearchAlgorithm | None = None, scheduler: TrialScheduler | None = None, num_samples: int = 1, max_concurrent_trials: int | None = None, time_budget_s: int | float | timedelta | None = None, reuse_actors: bool | None = None, trial_name_creator: Callable[[Trial], str] | None = None, trial_dirname_creator: Callable[[Trial], str] | None = None, chdir_to_trial_dir: bool = 'DEPRECATED')#
Tune specific configs.
metric – Metric to optimize. This metric should be reported with
tune.report(). If set, will be passed to the search algorithm and scheduler.
mode – Must be one of [min, max]. Determines whether objective is minimizing or maximizing the metric attribute. If set, will be passed to the search algorithm and scheduler.
search_alg – Search algorithm for optimization. Default to random search.
scheduler – Scheduler for executing the experiment. Choose among FIFO (default), MedianStopping, AsyncHyperBand, HyperBand and PopulationBasedTraining. Refer to ray.tune.schedulers for more options.
num_samples – Number of times to sample from the hyperparameter space. Defaults to 1. If
grid_searchis provided as an argument, the grid will be repeated
num_samplesof times. If this is -1, (virtually) infinite samples are generated until a stopping condition is met.
max_concurrent_trials – Maximum number of trials to run concurrently. Must be non-negative. If None or 0, no limit will be applied. This is achieved by wrapping the
ConcurrencyLimiter, and thus setting this argument will raise an exception if the
search_algis already a
ConcurrencyLimiter. Defaults to None.
time_budget_s – Global time budget in seconds after which all trials are stopped. Can also be a
reuse_actors – Whether to reuse actors between different trials when possible. This can drastically speed up experiments that start and stop actors often (e.g., PBT in time-multiplexing mode). This requires trials to have the same resource requirements. Defaults to
Truefor function trainables (including most Ray Train Trainers) and
Falsefor class and registered trainables (e.g. RLlib).
trial_name_creator – Optional function that takes in a Trial and returns its name (i.e. its string representation). Be sure to include some unique identifier (such as
Trial.trial_id) in each trial’s name. NOTE: This API is in alpha and subject to change.
trial_dirname_creator – Optional function that takes in a trial and generates its trial directory name as a string. Be sure to include some unique identifier (such as
Trial.trial_id) is used in each trial’s directory name. Otherwise, trials could overwrite artifacts and checkpoints of other trials. The return value cannot be a path. NOTE: This API is in alpha and subject to change.
chdir_to_trial_dir – Deprecated. Use the
RAY_CHDIR_TO_TRIAL_DIR=0environment variable instead. Whether to change the working directory of each worker to its corresponding trial directory. Defaults to
Trueto prevent contention between workers saving trial-level outputs. If set to
False, files are accessible with paths relative to the original working directory. However, all workers on the same node now share the same working directory, so be sure to use
ray.train.get_context().get_trial_dir()as the path to save any outputs.
PublicAPI (beta): This API is in beta and may change before becoming stable.