TorchBeast: A PyTorch Platform for Distributed RL
This provides a tool for RL researchers to conduct scalable experiments more easily, but it is incremental as it adapts an existing algorithm into a new framework.
The paper tackles the problem of enabling scalable reinforcement learning research by introducing TorchBeast, a PyTorch-based platform that implements the IMPALA algorithm for fast, asynchronous, parallel training, and it performs on-par with IMPALA on Atari benchmarks.
TorchBeast is a platform for reinforcement learning (RL) research in PyTorch. It implements a version of the popular IMPALA algorithm for fast, asynchronous, parallel training of RL agents. Additionally, TorchBeast has simplicity as an explicit design goal: We provide both a pure-Python implementation ("MonoBeast") as well as a multi-machine high-performance version ("PolyBeast"). In the latter, parts of the implementation are written in C++, but all parts pertaining to machine learning are kept in simple Python using PyTorch, with the environments provided using the OpenAI Gym interface. This enables researchers to conduct scalable RL research using TorchBeast without any programming knowledge beyond Python and PyTorch. In this paper, we describe the TorchBeast design principles and implementation and demonstrate that it performs on-par with IMPALA on Atari. TorchBeast is released as an open-source package under the Apache 2.0 license and is available at \url{https://github.com/facebookresearch/torchbeast}.