► KerasTuner: Hyperparam Tuning / API documentation / Oracles / The base Oracle class

The base Oracle class

`Oracle` class

keras_tuner.Oracle(
    objective=None,
    max_trials=None,
    hyperparameters=None,
    allow_new_entries=True,
    tune_new_entries=True,
    seed=None,
    max_retries_per_trial=0,
    max_consecutive_failed_trials=3,
)

Implements a hyperparameter optimization algorithm.

In a parallel tuning setting, there is only one Oracle instance. The workers would communicate with the centralized Oracle instance with gPRC calls to the Oracle methods.

Trial objects are often used as the communication packet through the gPRC calls to pass information between the worker Tuner instances and the Oracle. For example, Oracle.create_trial() returns a Trial object, and Oracle.end_trial() accepts a Trial in its arguments.

New copies of the same Trial instance are reconstructed as it going through the gRPC calls. The changes to the Trial objects in the worker Tuners are synced to the original copy in the Oracle when they are passed back to the Oracle by calling Oracle.end_trial().

Arguments

objective: A string, keras_tuner.Objective instance, or a list of keras_tuner.Objectives and strings. If a string, the direction of the optimization (min or max) will be inferred. If a list of keras_tuner.Objective, we will minimize the sum of all the objectives to minimize subtracting the sum of all the objectives to maximize. The objective argument is optional when Tuner.run_trial() or HyperModel.fit() returns a single float as the objective to minimize.
max_trials: Integer, the total number of trials (model configurations) to test at most. Note that the oracle may interrupt the search before max_trial models have been tested if the search space has been exhausted.
hyperparameters: Optional HyperParameters instance. Can be used to override (or register in advance) hyperparameters in the search space.
tune_new_entries: Boolean, whether hyperparameter entries that are requested by the hypermodel but that were not specified in hyperparameters should be added to the search space, or not. If not, then the default value for these parameters will be used. Defaults to True.
allow_new_entries: Boolean, whether the hypermodel is allowed to request hyperparameter entries not listed in hyperparameters. Defaults to True.
seed: Int. Random seed.
max_retries_per_trial: Integer. Defaults to 0. The maximum number of times to retry a Trial if the trial crashed or the results are invalid.
max_consecutive_failed_trials: Integer. Defaults to 3. The maximum number of consecutive failed Trials. When this number is reached, the search will be stopped. A Trial is marked as failed when none of the retries succeeded.

[source]

`wrapped_func` function

keras_tuner.Oracle.create_trial()

[source]

`wrapped_func` function

keras_tuner.Oracle.end_trial()

[source]

`get_best_trials` method

Oracle.get_best_trials(num_trials=1)

Returns the best Trials.

[source]

`get_state` method

Oracle.get_state()

Returns the current state of this object.

This method is called during save.

Returns

A dictionary of serializable objects as the state.

[source]

`set_state` method

Oracle.set_state(state)

Sets the current state of this object.

This method is called during reload.

Arguments

state: A dictionary of serialized objects as the state to restore.

[source]

`score_trial` method

Oracle.score_trial(trial)

Score a completed Trial.

This method can be overridden in subclasses to provide a score for a set of hyperparameter values. This method is called from end_trial on completed Trials.

Arguments

trial: A completed Trial object.

[source]

`populate_space` method

Oracle.populate_space(trial_id)

Fill the hyperparameter space with values for a trial.

This method should be overridden in subclasses and called in create_trial in order to populate the hyperparameter space with values.

Arguments

trial_id: A string, the ID for this Trial.

Returns

A dictionary with keys "values" and "status", where "values" is a mapping of parameter names to suggested values, and "status" should be one of "RUNNING" (the trial can start normally), "IDLE" (the oracle is waiting on something and cannot create a trial), or "STOPPED" (the oracle has finished searching and no new trial should be created).

[source]