Callbacks

systematica.tuners.neptune_ai.callbacks

CustomCallback

CustomCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Custom Neptune callback for Optuna trials.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Descendants

• systematica.tuners.neptune_ai.callbacks.CSVCallback
• systematica.tuners.neptune_ai.callbacks.ChunkedJSONCallback
• systematica.tuners.neptune_ai.callbacks.ChunkedParquetCallback
• systematica.tuners.neptune_ai.callbacks.JSONCallback
• systematica.tuners.neptune_ai.callbacks.ParquetCallback

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Upload file placeholder.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

JSONCallback

JSONCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Enhanced Neptune callback with efficient live logging and full trial compression as JSON json.gz extension.

Logs both parameters and values in real-time while storing complete trial data for compressed post-run upload.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.neptune_ai.callbacks.CustomCallback
• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Compress and upload trial data as json file.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

Raises:

Type	Description
Exception	If the upload fails, an error is logged with the exception details.

ChunkedJSONCallback

ChunkedJSONCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Enhanced Neptune callback that logs trial data in real-time and uploads full trial data in compressed JSON chunks at experiment end.

Splits the compressed data into chunks of specified size (e.g., 5 MB) to handle large datasets efficiently.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.neptune_ai.callbacks.CustomCallback
• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Compress and upload trial data in chunks as json file.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

Raises:

Type	Description
Exception	If the upload fails, an error is logged with the exception details.

ParquetCallback

ParquetCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Enhanced Neptune callback that logs trial data in real-time and uploads full trial data as a compressed Parquet file at experiment end.

Splits the compressed data into chunks of specified size (e.g., 5 MB) to handle large datasets efficiently.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.neptune_ai.callbacks.CustomCallback
• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Compress and upload trial data as parquet file.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

Raises:

Type	Description
Exception	If the upload fails, an error is logged with the exception details.

ChunkedParquetCallback

ChunkedParquetCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Enhanced Neptune callback that logs trial data in real-time and uploads full trial data as a compressed Parquet chunks at experiment end.

Uses Parquet format for efficient storage and handling of large datasets.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.neptune_ai.callbacks.CustomCallback
• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Compress and upload trial data in chunks as parquet file.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

Raises:

Type	Description
Exception	If the upload fails, an error is logged with the exception details.

CSVCallback

CSVCallback(
    run: neptune.metadata_containers.run.Run,
    **kwargs,
)

Enhanced Neptune callback with efficient live logging and full trial data upload as CSV.

Logs both parameters and values in real-time while storing complete trial data for post-run upload as a CSV file.

Initialize the Neptune custom callback.

Parameters:

Name	Type	Default	Description
run	neptune.Run	--	Neptune run instance for logging.
kwargs	tp.Kwargs	--	Additional keyword arguments passed to parent class.

Ancestors

• systematica.tuners.neptune_ai.callbacks.CustomCallback
• systematica.tuners.base.BaseCustomNeptuneCallback
• abc.ABC
• systematica.utils.neptune_ai.NeptuneCallback

Static methods

combine_trials

combine_trials(
    trials: dict,
) ‑> pandas.core.frame.DataFrame

Combine trials into a pandas DataFrame.

Parameters:

Name	Type	Default	Description
trials	tp.Kwargs	--	Dictionary containing trial data.

Returns:

Type	Description
pd.DataFrame	DataFrame containing combined trial data.

Methods

custom_upload

custom_upload(
    self,
    study: optuna.study.study.Study,
    trial: optuna.trial._frozen.FrozenTrial,
)

Upload trial data as CSV file.

Parameters:

Name	Type	Default	Description
study	optuna.Study	--	The Optuna study object.
trial	optuna.trial.FrozenTrial	--	The completed Optuna trial.

Raises:

Type	Description
Exception	If the upload fails, an error is logged with the exception details.