facebookresearch/StarSpace

Health v2 · maintenance only

Flags: no_releases archived

Adoption not part of the score

3952 stars · 524 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

StarSpace is a general-purpose neural model from Facebook Research that learns entity embeddings for classification, retrieval, ranking, and similarity tasks. It maps objects of different types into a common vector space so they can be compared and ranked against each other.

Use cases

• learn word, sentence, or document embeddings
• rank documents or entities for information retrieval
• classify text or other labeled data
• learn sentence or document similarity metrics
• build content-based or collaborative filtering recommendations
• embed multi-relational knowledge graphs like Freebase
• classify or retrieve images using ResNet features

When to choose

• you need a fast, general-purpose embedding model for ranking or classification across mixed entity types
• you want a lightweight C++ alternative to deep learning frameworks for embedding learning
• you need to compare entities of different types in a shared vector space

When to avoid

• you need state-of-the-art transformer-based embeddings for NLP tasks
• you want an actively developed project with frequent updates
• you need GPU-accelerated training or deep integration with modern ML ecosystems

Facets

library · maturity maintenance

machine-learning nlp search-engine machine-learning windows cpp python embeddings entity-embeddings ranking recommendation text-classification similarity-learning facebook-research natural-language-processing search linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/StarSpace")