snap-stanford/stark

(NeurIPS D&B 2024) STaRK: Benchmarking LLM Retrieval on Textual and Relational Knowledge Bases

49
/ 100
Emerging

Combines textual search with structured relational reasoning across three large-scale knowledge bases (Amazon products, academic papers, biomedical data) using hybrid retrieval approaches. Provides pre-computed embeddings from multiple models (text-embedding-ada-002, GritLM, Colbert) and supports custom embedding generation, enabling direct evaluation of LLM retrieval performance on semi-structured data. Available as a pip package with automatic HuggingFace dataset downloads and includes an interactive knowledge base explorer for query analysis.

330 stars.

No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 9 / 25
Community 20 / 25

How are scores calculated?

Stars

330

Forks

51

Language

Python

License

MIT

Last pushed

Feb 06, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/embeddings/snap-stanford/stark"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.