A platform for Applied Reinforcement Learning (Applied RL)
Clone or download
czxttkl and facebook-github-bot Support DQN with Multi-Step Learning
Summary:
We are adding support for multi-step RL. This diff updates the pytorch codes for several new features:
1. read multi-step RL config
2. read hive table with multi-step data
3. Q-learning with multi steps

Two previous diffs are related: D13049646 added a spark pipeline for reading multi-step data, and D13125518 tested the spark pipeline in dataswarm.

Reviewed By: MisterTea

Differential Revision: D13345496

fbshipit-source-id: dee1dfd8b4d0af00ba0997173b5268edc18d75aa
Latest commit 123f124 Dec 16, 2018

README.md

Banner

Applied Reinforcement Learning @ Facebook

Build Status

Overview

Horizon is an open source end-to-end platform for applied reinforcement learning (RL) developed and used at Facebook. Horizon is built in Python and uses PyTorch for modeling and training and Caffe2 for model serving. The platform contains workflows to train popular deep RL algorithms and includes data preprocessing, feature transformation, distributed training, counterfactual policy evaluation, and optimized serving. For more detailed information about Horizon see the white paper here.

Algorithms Supported

Installation

Horizon can be installed via. Docker or manually. Detailed instructions on how to install Horizon can be found here.

Usage

Detailed instructions on how to use Horizon can be found here.

License

Horizon is released under a BSD license. Find out more about it here.

Logo