RecurrentLayers

LuxRecurrentLayers.jl extends Lux.jl recurrent layers offering by providing implementations of additional recurrent layers not available in base deep learning libraries.

Features

The package offers multiple layers for Lux.jl. Currently there are 25+ cells implemented, together with multiple higher level implementations:

Short namePublication venueOfficial implementation
AntisymmetricRNN/GatedAntisymmetricRNNICLR 2019
ATREMNLP 2018bzhangGo/ATR
BR/BRCPLOS ONE 2021nvecoven/BRC
CFNICLR 2017
coRNNICLR 2021tk-rusch/coRNN
FastRNN/FastGRNNNeurIPS 2018Microsoft/EdgeML
IndRNNCVPR 2018Sunnydreamrain/IndRNNTheanoLasagne
JANETarXiv 2018JosvanderWesthuizen/janet
LEMICLR 2022tk-rusch/LEM
LiGRUIEEE Transactions on Emerging Topics in Computing 2018mravanelli/theano-kaldi-rnn
LightRUMDPI Electronics 2023
MinimalRNNNeurIPS 2017
MultiplicativeLSTMWorkshop ICLR 2017benkrause/mLSTM
MGUInternational Journal of Automation and Computing 2016
MUT1/MUT2/MUT3ICML 2015
NASarXiv 2016
OriginalLSTMNeural Computation 1997-
PeepholeLSTMJMLR 2002
RANarXiv 2017kentonl/ran
RHNICML 2017jzilly/RecurrentHighwayNetworks
SCRNICLR 2015facebookarchive/SCRNNs
SGRNIET 2018
STARIEEE Transactions on Pattern Analysis and Machine Intelligence 20220zgur0/STAckable-Recurrent-network
Typed RNN / GRU / LSTMICML 2016
UGRNNICLR 2017-
UnICORNNICML 2021tk-rusch/unicornn
WMCLSTMNeural Networks 2021

Installation

You can install LuxRecurrentLayers using either of:

using Pkg
Pkg.add("LuxRecurrentLayers")
julia> ]
pkg> add LuxRecurrentLayers