ejaasaari/lemur
LEMUR reduces multi-vector retrieval for late interaction models such as ColBERT into regular single-vector retrieval.
LEMUR helps developers who are building search and retrieval systems to make them much faster. It takes collections of document embeddings and query embeddings, and outputs a ranked list of relevant documents. This is useful for anyone working with large text datasets who needs to quickly find the most relevant documents for a given query.
Use this if you are a developer building a search system that uses late interaction models like ColBERT and need to significantly speed up your retrieval process.
Not ideal if you do not have an AVX-512 compatible CPU or are not comfortable working with Python development tools.
Stars
26
Forks
3
Language
Python
License
MIT
Category
Last pushed
Feb 23, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/embeddings/ejaasaari/lemur"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
FlagOpen/FlagEmbedding
Retrieval and Retrieval-augmented LLMs
qdrant/fastembed
Fast, Accurate, Lightweight Python library to make State of the Art Embedding
Blaizzy/mlx-embeddings
MLX-Embeddings is the best package for running Vision and Language Embedding models locally on...
Merck/Sapiens
Sapiens is a human antibody language model based on BERT.
amansrivastava17/embedding-as-service
One-Stop Solution to encode sentence to fixed length vectors from various embedding techniques