Analyzer

systematica.tuners.neptune_ai.analyzer

NeptuneOptunaAnalyzer

NeptuneOptunaAnalyzer(
    project: str,
    api_token: str = None,
)

Neptune runs and fetching experiment data analyzer.

Method generated by attrs for class NeptuneOptunaAnalyzer.

Ancestors

• systematica.tuners.base.BaseOptunaAnalyzer
• systematica.tuners.base.BaseAnalyzer
• abc.ABC

Methods

run_context

run_context(
    self,
    **kwargs,
) ‑> neptune.metadata_containers.run.Run

Context manager to initialize and stop a Neptune run.

Parameters:

Name	Type	Default	Description
kwargs	tp.Kwargs	--	Additional arguments to pass to neptune.init_run (e.g., with_id, mode).

Yields:

Type	Description
neptune.Run	The initialized Neptune run object.

fetch_trials

fetch_trials(
    self,
    metadata_tag: str,
    run_ids: Tuple[str],
    is_trials: bool,
    is_file: bool = False,
    mode: str = 'read-only',
) ‑> pandas.core.frame.DataFrame

Fetch metadata from Neptune runs and return as a DataFrame.

Parameters:

Name	Type	Default	Description
metadata_tag	str	--	The metadata field to fetch (e.g., ‘best/values’, ‘trials/trials’).
run_ids	tp.Tuple[str]	--	The ID(s) of the Neptune run(s) to fetch.
is_trials	bool	--	Whether the metadata represents trials (True) or other data (False).
is_file	bool	True	Flag to fetch compressed file trials if True. Use namespace fetch() otherwise. Defaults to True.
mode	str	read-only	Connection mode for Neptune. Options: ‘async’, ‘sync’, ‘offline’, ‘read-only’, ‘debug’. Defaults to ‘read-only’.

Returns:

Type	Description
pd.DataFrame	DataFrame containing the fetched metadata.

fetch_feature_config

fetch_feature_config(
    self,
    run,
    run_id: str,
) ‑> Dict[str, Any]

Fetch feature config from Neptune.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run.
run_id	str	--	The ID of the Neptune run to fetch.

Raises:

Type	Description
SystematicaError	No metadata found for run_id.

Returns:

Type	Description
tp.Kwargs	metadata params.

get_all_trials

get_all_trials(
    self,
    run_ids: str | List[str],
    is_file: bool = True,
) ‑> pandas.core.frame.DataFrame

Fetch all trials from Neptune runs.

Parameters:

Name	Type	Default	Description
run_ids	str	--	The ID(s) of the Neptune run(s) to fetch.
is_file	bool	True	Flag to fetch compressed file trials through ‘trials/compressed_trials’ if True. Fetch ‘trials/trials’ otherwise. Defaults to True.

Returns:

Type	Description
pd.DataFrame	All trial data, including ‘run_id’ and ‘number’.

fetch_optuna_study

fetch_optuna_study(
    self,
    run_id: str,
) ‑> optuna.study.study.Study

Loads Optuna study from an existing Neptune run.

Loading mechanics depend on the study storage type used during the run:

• If the study used InMemoryStorage, it will be loaded from the logged pickled study object.
• If the study used database storage, it will be loaded from the logged database URL.

To resume an existing run, you need the run ID. It is stored in the run’s sys/id field.

Parameters:

Name	Type	Default	Description
run_id	str	--	The ID of the Neptune run to fetch.

Returns:

Type	Description
optuna.Study	The Optuna study object.