evoc-rs

EVoC clustering for high-dimensional embeddings. The Rust crate does the work; this is a thin scikit-learn shaped layer over it.

uv pip install evoc-rs

import numpy as np
import evoc_rs

rng = np.random.default_rng(0)
X = np.vstack([rng.normal(c * 20, 1, (500, 32)) for c in range(4)]).astype(np.float32)

model = evoc_rs.EVoC(n_neighbours=15).fit(X)
model.labels_

What EVoC does differently

Most density-based clustering runs on your data as it stands. EVoC embeds the kNN graph first, with a UMAP-like optimiser, then builds an MST over that and pulls clusters out by persistence. Two consequences follow.

It is fast on wide data. The expensive stages see a 4-to-16 dimensional embedding, not your 768-dimensional one, and the only stage that touches the original space is the kNN search.

And you get a hierarchy, not a labelling. cluster_layers_ holds one labelling per granularity, persistence_scores_ says how stable each is, and labels_ picks the most stable for you. That last part is a convenience, not the answer: the layers either side are often the interesting ones. See the cluster hierarchy.

Where to go next

Quickstart is the five-minute version. kNN backends covers the seven graph builders and when the default is wrong. GPU covers EVoCGpu and when it pays off.

Credit

Port of evoc by Leland McInnes. Where behaviour diverges, the Python original is the source of truth.