Panther: Faster and Cheaper Computations with Randomized Numerical Linear Algebra

TL;DR

Panther is a PyTorch-compatible library that integrates randomized numerical linear algebra algorithms to reduce memory usage and accelerate computations in deep learning models without sacrificing performance.

Contribution

Panther provides a unified, high-performance, production-grade library implementing RandNLA techniques for deep learning, enabling easy adoption and significant memory savings.

Findings

Up to 75% memory reduction on BERT models

Maintains comparable loss with standard models

Easy integration with minimal code changes

Abstract

Training modern deep learning models is increasingly constrained by GPU memory and compute limits. While Randomized Numerical Linear Algebra (RandNLA) offers proven techniques to compress these models, the lack of a unified, production-grade library prevents widely adopting these methods. We present Panther, a PyTorch-compatible library that consolidates established RandNLA algorithms into a single high-performance framework. Panther engineers efficient, drop-in replacements for standard components including sketched linear layers, 2D convolution, multi-head attention, and randomized matrix decompositions (such as pivoted CholeskyQR). By implementing a custom C++/CUDA backend (pawX), Panther provides an optimized implementation that can run on both CPUs and GPUs. We demonstrate the effectiveness of RandNLA techniques and Panther's ease of adoption. By replacing standard PyTorch linear layers with Panther layers (requiring only a few lines of code) we achieve significant memory savings (up to 75%) on BERT while maintaining comparable loss. Source code is available (MIT License) at https://github.com/FahdSeddik/panther, along with demonstration video at https://youtu.be/7M3RQb4KWxs.