autointent.modules.base.BaseModule

class autointent.modules.base.BaseModule

Bases: abc.ABC

Base module for all intent classification modules.

supports_oos: bool

Whether the module supports oos data

supports_multilabel: bool

Whether the module supports multilabel classification

supports_multiclass: bool

Whether the module supports multiclass classification

name: str

Name of the module.

abstract fit(*args, **kwargs)

Fit the model.

Parameters:

• *args (tuple[Any]) – Args to fit

• **kwargs (dict[str, Any]) – Kwargs to fit

Return type:

None

score(context, metrics)

Calculate metric on test set and return metric value.

Parameters:

• context (autointent.context.Context) – Context to score

• metrics (list[str]) – Metrics to score

Raises:

ValueError – If unknown scheme is provided

Return type:

dict[str, float]

abstract score_cv(context, metrics)

Parameters:

• context (autointent.context.Context)

• metrics (list[str])

Return type:

dict[str, float]

abstract score_ho(context, metrics)

Parameters:

• context (autointent.context.Context)

• metrics (list[str])

Return type:

dict[str, float]

abstract get_assets()

Return useful assets that represent intermediate data into context.

Returns:

Artifact containing intermediate data

Return type:

autointent.context.optimization_info.Artifact

abstract clear_cache()

Clear cache.

Return type:

None

dump(path)

Dump all data needed for inference.

Parameters:

path (str) – Path to dump

Return type:

None

load(path, embedder_config=None, cross_encoder_config=None)

Load data from file system.

Parameters:

• path (str) – Path to load

• embedder_config (autointent.configs.EmbedderConfig | None) – one can override presaved settings

• cross_encoder_config (autointent.configs.CrossEncoderConfig | None) – one can override presaved settings

Return type:

None

abstract predict(*args, **kwargs)

Predict on the input.

Parameters:

• *args (list[str] | numpy.typing.NDArray[Any]) – args to predict

• **kwargs (dict[str, Any]) – kwargs to predict

Return type:

autointent.custom_types.ListOfGenericLabels | numpy.typing.NDArray[Any]

predict_with_metadata(*args, **kwargs)

Predict on the input with metadata.

Parameters:

• *args (list[str] | numpy.typing.NDArray[Any]) – args to predict

• **kwargs (dict[str, Any]) – kwargs to predict

Returns:

Tuple of predictions and metadata

Return type:

tuple[autointent.custom_types.ListOfGenericLabels | numpy.typing.NDArray[Any], list[dict[str, Any]] | None]

classmethod from_context(context, **kwargs)

Abstractmethod:

Parameters:

• context (autointent.context.Context)

• kwargs (dict[str, Any])

Return type:

BaseModule

Initialize self from context.

Parameters:

• context (autointent.context.Context) – Context to init from

• **kwargs (dict[str, Any]) – Additional kwargs

Returns:

Initialized module

Return type:

BaseModule

get_embedder_config()

Get the config of the embedder.

Return type:

dict[str, Any] | None

static score_metrics_ho(params, metrics_dict)

Score metrics on the test set.

Parameters:

• params (tuple[Any, Any]) – Params to score

• metrics_dict (dict[str, Any]) – Dictionary of metrics to compute

Return type:

dict[str, float]

score_metrics_cv(metrics_dict, cv_iterator, **fit_kwargs)

Score metrics using cross-validation.

Parameters:

• metrics_dict (dict[str, Any]) – Dictionary of metrics to compute

• cv_iterator (collections.abc.Iterable[tuple[list[str], autointent.custom_types.ListOfLabels, list[str], autointent.custom_types.ListOfLabels]]) – Cross-validation iterator

• **fit_kwargs – Additional arguments for fit method

Return type:

tuple[dict[str, float], list[autointent.custom_types.ListOfGenericLabels] | list[numpy.typing.NDArray[Any]]]