Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces mlx-vis, a Python library built on Apple's MLX framework, enabling GPU-accelerated dimensionality reduction and visualization directly on Apple Silicon. It implements UMAP, t-SNE, PaCMAP, TriMap, DREAMS, CNE, and NNDescent, along with a GPU-accelerated circle-splatting renderer. Benchmarks on Fashion-MNIST demonstrate fast embedding (2.1-3.8s) and animation rendering (1.4s) on an M3 Ultra, offering a complete pipeline from data to video in under 6 seconds.
Ditch matplotlib for blazing-fast, GPU-powered dimensionality reduction and visualization on your Mac with mlx-vis.
mlx-vis is a Python library that implements six dimensionality reduction methods and a k-nearest neighbor graph algorithm entirely in MLX, Apple's array framework for Apple Silicon. The library provides UMAP, t-SNE, PaCMAP, TriMap, DREAMS, CNE, and NNDescent, all executing on Metal GPU through a unified fit_transform interface. Beyond embedding computation, mlx-vis includes a GPU-accelerated circle-splatting renderer that produces scatter plots and smooth animations without matplotlib, composing frames via scatter-add alpha blending on GPU and piping them to hardware H.264 encoding. On Fashion-MNIST with 70,000 points, all methods complete embedding in 2.1-3.8 seconds and render 800-frame animations in 1.4 seconds on an M3 Ultra, with the full pipeline from raw data to rendered video finishing in 3.6-5.2 seconds. The library depends only on MLX and NumPy, is released under the Apache 2.0 license, and is available at https://github.com/hanxiao/mlx-vis.