Lancelot provides Ruby-native bindings to Lance, which is a modern columnar data format tailored for machine learning workloads. It supports efficient storage and searching of embeddings, text, and structured data in one platform. The latest version, v0.3.1, includes a Hybrid search feature utilizing Reciprocal Rank Fusion, allowing users to combine semantic vector and keyword text searches in one query. Key functionalities include vector search with index support, full-text search with BM25 scoring, and an API that follows Ruby conventions, making it suitable for various applications such as semantic search and recommendation systems.
Lancelot provides Ruby-native bindings to Lance, a modern columnar data format designed for ML workloads. It's ideal for storing and searching embeddings, text, and structured data.
Version 0.3.1 introduces Hybrid search with Reciprocal Rank Fusion (RRF), merging semantic vector search with keyword text search in a single query for improved results.
Key features include vector search (ANN) with index support, full-text search with BM25 scoring, and a Ruby-idiomatic API with Enumerable support.
This release is aimed at RAG applications, semantic search, recommendation systems, and ML feature stores, eliminating the need for PostgreSQL extensions or external vector databases.
Collection
[
|
...
]