Magnitude: A Fast, Efficient Universal Vector Embedding Utility Package

TL;DR

Magnitude is a fast, lightweight Python package that efficiently handles large-scale vector embeddings in NLP, offering significant speed improvements and robustness features.

Contribution

It introduces Magnitude, a novel open-source utility for efficient storage, manipulation, and robustness in vector embeddings, outperforming existing tools like Gensim.

Findings

Magnitudes performs operations up to 6,000 times faster than Gensim.

It features a compact storage format for large embedding collections.

Includes out-of-vocabulary lookup for improved robustness.

Abstract

Vector space embedding models like word2vec, GloVe, fastText, and ELMo are extremely popular representations in natural language processing (NLP) applications. We present Magnitude, a fast, lightweight tool for utilizing and processing embeddings. Magnitude is an open source Python package with a compact vector storage file format that allows for efficient manipulation of huge numbers of embeddings. Magnitude performs common operations up to 60 to 6,000 times faster than Gensim. Magnitude introduces several novel features for improved robustness like out-of-vocabulary lookups.