OpenTensor: Reproducing Faster Matrix Multiplication Discovering Algorithms

TL;DR

OpenTensor is a framework that reproduces and improves upon AlphaTensor's matrix multiplication algorithms using Deep Reinforcement Learning, making the process more transparent and accessible.

Contribution

We provide a cleaned-up, clarified, and improved implementation of AlphaTensor, enabling easier reproduction and further development of efficient matrix multiplication algorithms.

Findings

OpenTensor successfully reproduces AlphaTensor's algorithms.

It discovers new efficient matrix multiplication algorithms.

The framework improves reproducibility and understanding of the original methods.

Abstract

OpenTensor is a reproduction of AlphaTensor, which discovered a new algorithm that outperforms the state-of-the-art methods for matrix multiplication by Deep Reinforcement Learning (DRL). While AlphaTensor provides a promising framework for solving scientific problems, it is really hard to reproduce due to the massive tricks and lack of source codes. In this paper, we clean up the algorithm pipeline, clarify the technical details, and make some improvements to the training process. Computational results show that OpenTensor can successfully find efficient matrix multiplication algorithms.