ds-utils-transformers

What problem does it solve?

Provides Sklearn-compatible transformers to integrate multi-label binarization and sentence embeddings into pipelines, enabling feature extraction and preprocessing in a consistent interface.

Core Features & Use Cases

• MultiLabelBinarizerTransformer: wraps sklearn's MultiLabelBinarizer so pipelines get get_feature_names_out and dense matrices.
• SentenceEmbeddingTransformer: wraps sentence-transformers models for use in pipelines with lazy loading and NaN handling.
• Use Case: Build a pipeline to binarize tags and then feed embeddings into a classifier, or combine text features with numeric ones using ColumnTransformer.

Quick Start

Create a Pipeline using MultiLabelBinarizerTransformer for your label column and, when processing text data, append SentenceEmbeddingTransformer to produce embeddings for model training.