Managers

class AssetManager(api_config: Configuration)[source]

AIoD - RAIL AssetManager class

Class containing methods that operate on simpler assets such as: models, datasets, publications, etc..

__init__(api_config: Configuration)[source]

Initializes a new ExperimentRunManager.

Parameters:: client_config – (Configuration): Instance of Configuration class.
Returns:: Initialized AssetManager.
Return type:: AssetManager

Examples

>>> config = Configuration(...)
>>> AssetManager(config)
AssetManager

count_datasets(query: str = None) → int[source]

Counts the number of datasets.

Parameters:: query – (Optional[str]): Count only datasets that in their name contain a string given in this arg.
Returns:: Number of datasets.
Return type:: int
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.count_datasets()
4321
>>> asset_manager.count_datasets(query="Image")
1234 # count of datasets that contain the phrase "Image"

count_models(query: str = None) → int[source]

Counts the number of models.

Parameters:: query – (Optional[str]): Count only models that contain in their name a string given by this arg.
Returns:: Number of models.
Return type:: int
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.count_models()
4321
>>> asset_manager.count_models(query="LLM")
1234 # count of models that contain the phrase "LLM"

count_my_datasets() → int[source]

Counts the number of datasets of the current user.

Returns:: Number of datasets.
Return type:: int
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.count_my_datasets()
123

count_my_models() → int[source]

Counts the number of models of the current user.

Returns:: Number of models.
Return type:: int
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.count_my_models()
123

count_publications(query: str = None) → int[source]

Counts the number of publications.

Parameters:: query – (Optional[str]): Count only publications that contain a string given in this arg.
Returns:: Number of publications.
Return type:: int
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.count_publications()
4321
>>> asset_manager.count_publications(query="CNNs")
1234 # count of publications that contain the phrase "CNNs"

get_dataset_by_id(id: str) → Dataset[source]

Retrieves dataset specified by its ID.

Parameters:: id (str) – Unique identifier of dataset in database.
Returns:: The dataset corresponding to the given ID.
Return type:: Dataset
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_dataset_by_id(id="data_000002AhzqHqOQwQLP0qCRds")
Dataset

get_datasets(query: str = None, enhanced: bool = False, offset: int = 0, limit: int = 100) → List[Dataset][source]

Retrieves a list of available datasets.

Parameters:

query (str, optional) – Search string used to filter datasets. Defaults to empty string,
default (which means that by)
used. (it's not)
enhanced (bool, optional) – If true and the query arg is also specified, a semantic search is performed.
offset (int, optional) – Starting index of from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

The list of datasets.

Return type:

List[Dataset]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_datasets(query="AI", offset=2, limit=10)
List[Dataset] # 10 datasets that contain the text "AI" in their name

get_model_by_id(id: str) → Model[source]

Retrieves a model specified by its ID.

Parameters:: id (str) – Unique identifier of a model.
Returns:: The model corresponding to the given ID.
Return type:: Model
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_model_by_id(id="mdl_000028bcIa2SCbO9aVlIB0Xc")
Model

get_models(query: str = None, offset: int = 0, limit: int = 100) → List[Model][source]

Retrieves a list of models.

Parameters:

query (str, optional) – Search string used to filter models. Defaults to empty string,
default (which means that by)
used. (it's not)
offset (int, optional) – Starting index from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

The list of models.

Return type:

List[Model]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_models(query="LLM", offset=2, limit=10)
List[Model] # 10 models that contain the phrase "LLM" in their name

get_my_datasets(offset: int = 0, limit: int = 100) → List[Dataset][source]

Retrieves a list of datasets created by the current user.

Parameters:

offset (int, optional) – Starting index from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

The list of datasets.

Return type:

List[Dataset]

Raises:

ApiException – In case of a failed HTTP request.

Examples:

get_my_models(offset: int = 0, limit: int = 100) → List[Model][source]

Retrieves a list of models created by the current user.

Parameters:

offset (int, optional) – Starting index from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

A list of user’s models.

Return type:

List[Model]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_my_models(limit=2)
List[Model] # first 2 models created by a current user

get_platforms(offset: int = 0, limit: int = 100) → List[Platform][source]

Retrieves a list of platforms.

Parameters:

offset (int, optional) – Starting index from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

The list of platforms.

Return type:

List[Platform]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_platforms(offset=2, limit=5)
5

get_publication_by_id(id: str) → Publication[source]

Retrieves a publication specified by its ID.

Parameters:: id (str) – Unique identifier of a publication in database.
Returns:: The publication corresponding to the given ID.
Return type:: Dataset
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_publication_by_id(id="pub_03F6jLXedDDQSzc6expGpLGI")
Publication

get_publications(query: str = None, offset: int = 0, limit: int = 100) → List[Publication][source]

Retrieves a list of publications.

Parameters:

query (str, optional) – Search string used to filter publications. Defaults to empty string,
default (which means that by)
used. (it's not)
offset (int, optional) – Starting index from which to retrieve. Defaults to 0.
limit (int, optional) – How many items to retrieve. Defaults to 100.

Returns:

The list of publications.

Return type:

List[Publication]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager = AssetManager(...)
>>> asset_manager.get_publications(query="AI", offset=2, limit=10)
List[Publication]

upload_file_to_huggingface(id: str, name: str, huggingface_token: str, file_path: str) → Dataset[source]

Uploads a file to Huggingface under the dataset specified by its id.

Parameters:

id (str) – The ID of the dataset.
name (str) – The name of the file as it will appear on Huggingface.
huggingface_token – Authentication token from Huggingface.
file_path (str) – Path to the file to upload.

Returns:

Dataset under which the file was uploaded.

Return type:

Dataset

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> asset_manager =  AssetManager(...)
>>> asset_manager.upload_file_to_huggingface("data_000017VBQZZ7s3cvm7Gx0AnT", "Example Name", "huggingface_token", "path/to/file.txt")
Dataset

class ExperimentTemplateManager(api_config: Configuration)[source]

AIoD - RAIL

ExperimentManager class

Class aggregating methods for operating on multiple experiments.

__init__(api_config: Configuration)[source]

Initializes a new ExperimentTemplateManager.

Parameters:: client_config – (Configuration): Instance of Configuration class.
Returns:: Initialized ExperimentTemplateManager.
Return type:: ExperimentTemplateManager

Examples

>>> config = Configuration(...)
>>> ExperimentTemplateManager(config)
ExperimentTemplateManager

count(query: str = '', mine: bool | None = None, finalized: bool | None = None, approved: bool | None = None, public: bool | None = None) → int[source]

Counts the number of experiments based on filters specified in Args.

Parameters:

query (str, optional) – Query used to filter experiment templates. Defaults to empty string, which means that by default count is not filtered.
mine (bool, optional) – If own personal experiment templates should be counted or the opposite. Defaults to None.
finalized (bool, optional) – If experiment templates that are successfully build and ready to use should be counted or the opposite. Defaults to None.
approved (bool, optional) – If already approved experiments should be counted or the opposite. Defaults to None.
public (bool, optional) – If experiment templates flagged as public should be counted or the opposite. Defaults to None.

Returns:

Number of experiment templates.

ApiException: In case of a failed HTTP request.

Return type:

int

Raises:

Examples

>>> template_manager = ExperimentTemplateManager(...)
>>> template_manager.count(finalized=True, approved=True, public=True)
1234

create(template: dict | tuple[str, str, str, dict]) → ExperimentTemplate[source]

Creates a new experiment template.

Parameters:

template (requirements and docker image in this order and) – (dict | tuple[str, str, str, dict]):
json (The file can be passed either as full specified)
to (as a tuple of three strings with paths) – (script, requirements and docker image and a json
script ((dictionary) specifying the paths to)
template
description (name, description, task etc.)

Returns:

Created experiment template.

Return type:

ExperimentTemplate

Raises:

ApiException – In case of a failed HTTP request.

Note

Successfully created template will need to be approved by an administrator and afterward built as a docker container by the backend service. Only after these operations are done can it be used to make new experiments.

Examples

>>> script_path = "path/to/script.py"
>>> requirements_path = "path/to/requirements.txt"
>>> base_image = "python:3.9"
>>> template_config = {
>>> "name": "Example Template",
>>> "description": "Template in Examples",
>>> "task": "TEXT_CLASSIFICATION",
>>> "datasets_schema": { "cardinality": "1-1" },
>>> "models_schema": { "cardinality": "1-1" },
>>> "envs_required": [ { "name": "SPLIT_NAME", "description": "name of a subset" } ],
>>> "envs_optional": [],
>>> "available_metrics": [ "accuracy" ],
>>> "is_public": True
>>> }
>>> template_manager.create((script_path, requirements_path, base_image, template_config))
ExperimentTemplate # newly created instance

get(query: str = '', mine: bool | None = None, finalized: bool | None = None, approved: bool | None = None, public: bool | None = None, offset: int = 0, limit: int = 100) → List[ExperimentTemplate][source]

Gets experiment templates based of on specified filters.

Parameters:

query (str, optional) – Query used to filter experiment templates. This parameter is case-insensitive and matches full words in template names. Defaults to empty string, in which case it’s not used.
mine (bool, optional) – If own personal experiment templates should be included or the opposite. Defaults to None.
finalized (bool, optional) – If experiment templates that are successfully build and ready to use should be listed or the opposite. Defaults to None.
approved (bool, optional) – If already approved experiments should be listed or the opposite. Defaults to None.
public (bool, optional) – If experiment templates flagged as public should be listed or the opposite. Defaults to None.
offset (int, optional) – Starting index of experiment template range from which to retrieve Defaults to 0.
limit (int, optional) – Ending index of experiment template range to which to retrieve. Defaults to 100.

Returns:

List of all experiments in given range

Return type:

list[ExperimentTemplate]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> template_manager = ExperimentTemplateManager(...)
>>> template_manager.get()
List[ExperimentTemplate]
>>> len(template_manager.get(finalized=True, approved=True, limit=1000))
1000
>>> template_manager.get(query="Tutorial")
List[ExperimentTemplate] # only templates that contain word "Tutorial" in their name.

get_by_id(id: str) → ExperimentTemplate[source]

Retrieves a specific experiment template by its ID.

Parameters:: id (str) – ID of experiment template to be retrieved.
Returns:: Experiment template given by ID.
Return type:: ExperimentTemplate
Raises:: ApiException – In case of a failed HTTP request.

Examples

>>> template_manager = ExperimentTemplateManager(...)
>>> template_manager.get_by_id("685151f2d08da970a3a5d6ce")
ExperimentTemplate

class ExperimentManager(api_config: Configuration)[source]

AIoD - RAIL ExperimentManager class

Class aggregating methods for operating on multiple experiments.

__init__(api_config: Configuration)[source]

Initializes a new ExperimentManager.

Parameters:: client_config – (Configuration): Instance of Configuration class.
Returns:: Initialized ExperimentManager.
Return type:: ExperimentManager

Examples

>>> config = Configuration(...)
>>> ExperimentManager(config)
ExperimentManager

count(query: str = '', mine: bool | None = None, archived: bool | None = None, public: bool | None = None) → int[source]

Counts the number of experiments based on filters specified in Args.

Parameters:

query (str, optional) – Query used to filter experiments. Defaults to empty string, which means that by default, it’s not used.
mine (bool, optional) – If own personal experiments should be counted or the opposite. Defaults to None.
archived (bool, optional) – If archived experiments should be counted or the opposite. Defaults to None.
public (bool, optional) – If experiment templates flagged as public should be counted or the opposite. Defaults to None.

Returns:

Number of experiments.

Return type:

int

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> experiment_manager =  ExperimentManager(...)
>>> experiment_manager.count(query="Tutorial", mine=True, archived=True)
ExperimentTemplate

create(experiment: Dict) → Experiment[source]

Creates experiment from specified experiment file. :param experiment: Experiment described in a dictionary. :type experiment: dict

Returns:: Experiment created from given template.
Return type:: Experiment
Raises:: ApiException – In case of a failed HTTP request.

Note

To create an experiment successfully, the template it is based on needs to be approved and built as a docker container.

Examples

>>> experiment_dict = {
>>>     "name": "test123",
>>>     "description": "321test",
>>>     "is_public": True,
>>>     "experiment_template_id": "685151f2d08da970a3a5d6ce",
>>>     "dataset_ids": ["data_000002AhzqHqOQwQLP0qCRds"],
>>>     "model_ids": ["mdl_003Csk8QjNfE80c7g6Rt8yVb"],
>>>     "publication_ids": [],
>>>     "env_vars": [{"key": "SPLIT_NAME", "value": "PES"}
>>>     ]
>>> }
>>> experiment_manager =  ExperimentManager(...)
>>> experiment_manager.create(experiment_dict)
Experiment # Newly created experiment.

get(query: str = '', mine: bool | None = None, archived: bool | None = None, public: bool | None = None, offset: int = 0, limit: int = 100) → List[Experiment][source]

Retrieves a lis of experiments based on specific filters.

Parameters:

query (str, optional) – Query used to filter experiments. Defaults to empty string, which means that by default, it’s not used.
mine (bool, optional) – If own personal experiments should be included or the opposite. Defaults to None.
archived (bool, optional) – If archived experiments should be listed or the opposite. Defaults to None.
public (bool, optional) – If experiment templates flagged as public should be listed or the opposite. Defaults to None.
offset (int, optional) – Starting index of experiment range from which to retrieve. Defaults to 0.
limit (int, optional) – Ending index of experiment range to which to retrieve. Defaults to 100.

Returns:

The list of experiments.

Return type:

list[Experiment]

Raises:

ApiException – In case of a failed HTTP request.

Examples

>>> experiment_manager =  ExperimentManager(...)
>>> experiment_manager.get()
List[Experiment]
>>> len(experiment_manager.get(archived=False, public=True, limit=1000))
1000
>>> template_manager.get(query="Tutorial")
List[Experiment] # only experimenets that contain word "Tutorial" in their name.

get_by_id(id: str) → Experiment[source]

Gets specific experiment by its ID.

Parameters:: id (str) – ID of experiment to be retrieved.
Returns:: Experiment specified by its ID.
Return type:: Experiment
Raises:: ApiException – In case of a failed HTTP request or failure to retrieve an experiment with given ID.

Examples

>>> experiment_manager =  ExperimentManager(...)
>>> experiment_manager.get_by_id("6861d3c954d2c02536469a30")
Experiment

class ExperimentRunManager(api_config: Configuration)[source]

AIoD - RAIL ExperimentRunManager class

Class aggregating methods for operating on multiple experiment runs and which are not restricted to a specific experiment.

__init__(api_config: Configuration)[source]

Initializes a new ExperimentRunManager.

Parameters:: client_config – (Configuration): Instance of Configuration class.
Returns:: Initialized ExperimentRunManager.
Return type:: ExperimentRunManager

Examples

>>> config = Configuration(...)
>>> ExperimentRunManager(config)
ExperimentRunManager

get_by_id(id: str) → ExperimentRun[source]

Gets specific experiment run by its ID.

Parameters:: id (str) – ID of an experiment run to be retrieved.
Returns:: Experiment run specified by its ID.
Return type:: ExperimentRun
Raises:: ApiException – In case of a failed HTTP request or failure to retrieve an experiment with given ID.

Examples

>>> run_manager =  ExperimentRunManager(...)
>>> run_manager.get_by_id("6861d3c954d2c02536469a30")
ExperimentRun