Tianshou: a Highly Modularized Deep Reinforcement Learning Library

TL;DR

Tianshou is a modular, research-friendly Python library for deep reinforcement learning that supports numerous algorithms and includes a benchmark suite for MuJoCo environments, facilitating research and comparison.

Contribution

It introduces Tianshou, a flexible, reliable DRL library with extensive algorithm support and a benchmark suite, enhancing research and development in the field.

Findings

Supports over 20 classic DRL algorithms

Provides a benchmark for MuJoCo environments

Achieves state-of-the-art performance on benchmark tasks

Abstract

In this paper, we present Tianshou, a highly modularized Python library for deep reinforcement learning (DRL) that uses PyTorch as its backend. Tianshou intends to be research-friendly by providing a flexible and reliable infrastructure of DRL algorithms. It supports online and offline training with more than 20 classic algorithms through a unified interface. To facilitate related research and prove Tianshou's reliability, we have released Tianshou's benchmark of MuJoCo environments, covering eight classic algorithms with state-of-the-art performance. We open-sourced Tianshou at https://github.com/thu-ml/tianshou/.