plasticityai/magnitude

Health v2 · maintenance only

Adoption not part of the score

1668 stars · 122 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Magnitude is a Python package and file format for fast, memory-efficient loading and querying of vector embeddings like word2vec, GloVe, and fastText. It serves as a faster, simpler alternative to Gensim and can act as a generic key-vector store, with features like out-of-vocabulary lookups and HTTP streaming of large models.

Use cases

• load word2vec or GloVe embeddings quickly in Python
• query word vectors with low memory usage
• handle out-of-vocabulary words and misspellings in embeddings
• stream large embedding models over HTTP without downloading
• feed embeddings into Keras or other ML models
• use a fast key-vector store outside of NLP

When to choose

• you need fast, memory-efficient access to word embeddings in Python
• Gensim is too slow or complex for your embedding lookups
• you need OOV handling or remote streaming of embedding models

When to avoid

• you need to train new embedding models from scratch
• you need actively maintained software with recent updates
• you need general-purpose vector databases with ANN search at scale

Facets

library · maturity maintenance

nlp machine-learning serialization search-engine machine-learning data-science python cross-platform word-embeddings word2vec glove fasttext vector-store gensim-alternative memory-efficient natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="plasticityai/magnitude")