pytorch

Module for dense and sparseindexing and searching using PyTorch.

The PyTorch Indexing and Searching unifies both dense and sparse retrieval pipelines into a single streamlined interface. It utilizes PyTorch tensors and GPU acceleration to efficiently encode, store, and query documents, offering dedicated architectures for both continuous vector embeddings and high-dimensional sparse representations.

This module provides the classes and configurations for dense and sparse indexing and searching using PyTorch in the Lightning IR framework.

Modules

dense_indexer
dense_searcher	Torch-based Dense Searcher for Lightning IR Framework
sparse_indexer	Torch-based Sparse Indexer for Lightning IR Framework
sparse_searcher	Torch-based Sparse Searcher for Lightning IR Framework