#42 · Primary category： Classical Machine Learning Frameworks
bolt
10x faster matrix and vector operations
Project last updated：10/12/22
GitHub Stars
2.5K
Forks
174
Contributors
5
License
MPL-2.0
Why we included this project
Bolt tackles a cost that shows up everywhere in data-heavy ML work: storing and comparing large vectors of real numbers. It compresses mostly-dense vectors, then runs dot products, squared distances, and nearest-neighbor search directly on the compressed form, with published error bounds explaining what you give up in accuracy. For teams doing kNN or embedding search on big CPU-only datasets, a 10-200x cut in space or time is often worth that loss. The repo also carries MADDNESS, a newer approximate matrix-multiplication algorithm aimed at faster neural-net inference, though it has no Python wrapper yet and appears in the source as "mithral". Treat this as a research codebase rather than a drop-in dependency; the Bolt Python wrapper has been known to stop building for people.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
scikit-learn
scikit-learn: machine learning in Python
pyod
A Python library for anomaly detection across tabular, time series, graph, text, image, and audio data. 60+ detectors, benchmark-backed ADEngine orchestration, and an agentic workflow for AI agents.
dowhy
DoWhy is a Python library for causal inference that supports explicit modeling and testing of causal assumptions. DoWhy is based on a unified language for causal inference, combining causal graphical models and potential outcomes frameworks.
umap
Uniform Manifold Approximation and Projection
imbalanced-learn
A Python Package to Tackle the Curse of Imbalanced Datasets in Machine Learning