Callbacks

class pyhopper.callbacks.Callback

load_state_dict(state_dict)

Restores the internal state of the callback. :param state_dict: A dict created by the self.state_dict method

on_duplicate_sampled(candidate, info)

Called if candidate was sampled twice

Parameters

• candidate (dict) – Parameter value of the sampled candidate

• info (ParamInfo) –

on_evaluate_end(candidate, f, info)

Called after candidate was successfully evaluated

Parameters

• candidate (dict) – Parameter value of the evaluated candidate

• f (float) – Value of the objective function corresponding to the candidate

• info (ParamInfo) –

on_evaluate_nan(candidate, info)

Called if candidate is evaluated to NaN and the ignore_nans argument of run was set to True

Parameters

• candidate (dict) – Parameter value that evaluate to NaN

• info (ParamInfo) –

on_evaluate_pruned(candidate, info)

Called if candidate was pruned (by an pyhopper.pruners.Pruner())

Parameters

• candidate (dict) – Parameter value of the pruned candidate

• info (ParamInfo) –

on_evaluate_start(candidate, info)

Called after candidate was sampled and scheduled for evaluation

Parameters

• candidate (dict) – Parameter value of the candidate to be evaluated

• info (ParamInfo) –

on_new_best(new_best, f, info)

Called when a new best parameter is found

Parameters

• new_best (dict) – Value of the new best parameter

• f (float) – Value of the objective function corresponding to the new best parameter

• info (ParamInfo) –

on_search_end()

Called at the end of the search process

on_search_start(search)

Called at the beginning of the search

Parameters

search (Search) – pyhopper.Search object handling the search

state_dict()

Called when a checkpoint of the hyperparameter search is created to backup the state :return A dict containing the internal (runtime) state of the callback. None if the callback has no state

class pyhopper.ParamInfo(candidate_type, sampled_at)

Holds auxiliary information about a parameter candidate

Attributes:

type Enum specifying how the parameter was sampled (valid values are pyhopper.CandidateType.INIT, pyhopper.CandidateType.MANUALLY_ADDED, pyhopper.CandidateType.RANDOM_SEEDING, pyhopper.CandidateType.LOCAL_SAMPLING).

sampled_at UNIX epoch timestamp when the parameter candidate was sampled.

finished_at UNIX epoch timestamp when the evaluation of the candidate was finished.

is_pruned Bool indicating if the candidate was pruned.

class pyhopper.callbacks.History(log_candidates=True)

Public API for the history of the search. Can be used by the user for plotting and analyzing the search space. Persistent over several consecutive calls of `run`

load_state_dict(state_dict)

Restores the internal state of the callback. :param state_dict: A dict created by the self.state_dict method

on_evaluate_end(candidate, f, info)

Called after candidate was successfully evaluated

Parameters

• candidate (dict) – Parameter value of the evaluated candidate

• f (float) – Value of the objective function corresponding to the candidate

• info (ParamInfo) –

on_evaluate_nan(candidate, info)

Called if candidate is evaluated to NaN and the ignore_nans argument of run was set to True

Parameters

• candidate (dict) – Parameter value that evaluate to NaN

• info (ParamInfo) –

on_evaluate_pruned(candidate, info)

Called if candidate was pruned (by an pyhopper.pruners.Pruner())

Parameters

• candidate (dict) – Parameter value of the pruned candidate

• info (ParamInfo) –

on_new_best(new_best, f, info)

Called when a new best parameter is found

Parameters

• new_best (dict) – Value of the new best parameter

• f (float) – Value of the objective function corresponding to the new best parameter

• info (ParamInfo) –

on_search_start(search)

Called at the beginning of the search

Parameters

search (Search) – pyhopper.Search object handling the search

state_dict()

Called when a checkpoint of the hyperparameter search is created to backup the state :return A dict containing the internal (runtime) state of the callback. None if the callback has no state