autointent.modules.embedding.LogregAimedEmbedding#

class autointent.modules.embedding.LogregAimedEmbedding(embedder_config, cv=3)#

Bases: autointent.modules.base.BaseEmbedding

Module for configuring embeddings optimized for linear classification.

The main purpose of this module is to be used at embedding node for optimizing embedding configuration using its logreg classification quality as a sort of proxy metric.

Parameters:

embedder_config (autointent.configs.EmbedderConfig | str | dict[str, Any]) – Config of the embedder used for creating embeddings
cv (pydantic.PositiveInt) – Number of folds used in LogisticRegressionCV

Examples:#

from autointent.modules.embedding import LogregAimedEmbedding
utterances = ["bye", "how are you?", "good morning"]
labels = [0, 1, 1]
retrieval = LogregAimedEmbedding(
    embedder_config="sergeyzh/rubert-tiny-turbo",
    cv=2
)
retrieval.fit(utterances, labels)

name = 'logreg_embedding'#: Name of the module.

supports_multiclass = True#: Whether the module supports multiclass classification

supports_multilabel = True#: Whether the module supports multilabel classification

supports_oos = False#: Whether the module supports oos data

embedder_config#

cv = 3#

classmethod from_context(context, embedder_config, cv=3)#

Create a LogregAimedEmbedding instance using a Context object.

Parameters:

context (autointent.Context) – Context containing configurations and utilities
cv (pydantic.PositiveInt) – Number of folds used in LogisticRegressionCV
embedder_config (autointent.configs.EmbedderConfig | str) – Config of the embedder to use

Return type:

LogregAimedEmbedding

clear_cache()#

Clear embedder from memory.

Return type:: None

fit(utterances, labels)#

Train the logistic regression model using the provided utterances and labels.

Parameters:

utterances (list[str]) – List of text data to index
labels (autointent.custom_types.ListOfLabels) – List of corresponding labels for the utterances

Return type:

None

score_ho(context, metrics)#

Evaluate the embedding model using specified metric functions.

Parameters:

context (autointent.Context) – Context containing test data and labels
metrics (list[str]) – List of metric names to compute

Returns:

Dictionary of computed metric values for the test set

Return type:

dict[str, float]

score_cv(context, metrics)#

Evaluate the embedding model using specified metric functions.

Parameters:

context (autointent.Context) – Context containing test data and labels
metrics (list[str]) – List of metric names to compute

Returns:

Dictionary of computed metric values for the test set

Return type:

dict[str, float]

get_assets()#

Get the classifier artifacts for this module.

Returns:: EmbeddingArtifact object containing embedder information
Return type:: autointent.context.optimization_info.EmbeddingArtifact

predict(utterances)#

Predict probabilities for input utterances.

Parameters:: utterances (list[str]) – List of texts to predict probabilities for
Returns:: Array of predicted probabilities
Return type:: numpy.typing.NDArray[numpy.float64]