/RTRL

A RTRL implementation using CUDA

Primary LanguageMakefileApache License 2.0Apache-2.0

Fully Connected Neural Network

RTRL

RTRL is a CUDA based implementation of the real time recurrent learning algorithm as first described by Ronald J. Williams and David Zipser in A Learning Algorithm for Continually Running Fully Recurrent Neural Networks.

Quick Start

Dependencies

RTRL has the following dependencies:

Building the Executable

Once the dependencies are installed and the environment is set up to allow their access, one can compile RTRL by opening a terminal and entering the following commands:

localhost:~ kdavis$ cd <RTRL Root Directory>
localhost:RTRL kdavis$ make

This compiles the RTRL executable rtrl and places it into the RTRL root directory.