tusKANNy

tusKANNy is a research team building state-of-the-art algorithms and open-source tools for ANN search across dense, sparse, and multi-vector neural embeddings.

[tusKANNy @tuskannysearch]

Together they let you express a wide variety of combinations of quantizers, indexes and optimizations, which is what makes it quick to try a new idea.

📄 https://t.co/UOEk8U17iR

@SilvioMartinico, @fmnardini, @cosimorulli1, @rventurini_

[tusKANNy @tuskannysearch]

Our paper on the core architecture behind tusKANNy is out.

Two libraries, one idea: separate the data from the index. 🧵

#VectorSearch #NeuralIR #RustLang

[tusKANNy @tuskannysearch]

kANNolo holds the index primitives: HNSW, IVF, RerankIndex (two-stage).

It can be used for retrieval on its own, or to power other systems. For instance, its HNSW is what TACHIOM runs on. 🧵