1

I want to install the package mxnet on Rstudio Cloud, which is the cloud version of Rstudio. I tried it first with R 3.6, then R 3.5.3, neither worked. I tried to install from with this piece of code:

cran <- getOption("repos")
cran["dmlc"] <- "https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/R/CRAN/"
cran["dmlc"] <- "https://s3-us-west-2.amazonaws.com/apache-mxnet/R/CRAN/"
options(repos = cran)
install.packages("mxnet")

I got a 404 error file not found. Then i tried downloading the .tar.gz installation file associated with LInux/R/CPU, and it also gave an error:

ERROR: cannot extract package from ‘/cloud/project/apache-mxnet-src-1.5.0-incubating.tar.gz’

Here is my session info:

R version 3.5.3 (2019-03-11)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.6 LTS

Thanks for your help. I need to get this package installed and running on Rstudio Cloud.

agenis
  • 8,069
  • 5
  • 53
  • 102

1 Answers1

2

RStudio Cloud is linux based 

Platform: x86_64-pc-linux-gnu (64-bit)

The snippet you use is instruction for R on Windows CPU platform. R on Linux instruction there says to build it from source which wouldn't work on the cloud. Seems like no CRAN repo currently has binaries of the linux version of the mxnet package.

RStudio Cloud gives you access to the terminal and you can even download the sources for mxnet, but to build it you need sudo which the terminal doesn't provide so building from source stops on ./install_mxnet_ubuntu_r.sh command here:

git clone --recursive https://github.com/apache/incubator-mxnet.git mxnet
cd mxnet/docs/install
./install_mxnet_ubuntu_r.sh
isp-zax
  • 3,833
  • 13
  • 21
  • ok, so it's hopeless for now..? would you recommend another neural network package similar to mxnet, that i could install up there? – agenis Sep 08 '19 at 05:59
  • I was just thinking that since the terminal is available (even without sudo), you can try building the package binaries on a 64-bit linux machine, uploading binaries somewhere (like your public github repository), downloading them from the terminal using CURL and installing it from local directory. As for alternative, I have just tried and neuralnet package installs fine from CRAN – isp-zax Sep 08 '19 at 06:37
  • yes, indeed, neuralnet seems OK, as well as Keras and RSNNS. I will go with those ones (in fact I need a convolutionnal network). – agenis Sep 09 '19 at 07:11