rocker VS tensorflow

I am, strictly speaking, not a big R user, so take my opinion with a grain of salt, but if I were using R extensively, I would absolutely use the Rocker project containers to manage different R versions and different sets of dependencies for different projects: https://rocker-project.org/

Perhaps the main difference is r package are not available as binaries from rstudio (posit) repo but their is a cran2copr repo that works really well or you can still install from source in your home. For more info on cran2copr see: https://cran.rstudio.com/bin/linux/fedora/ . Personally I am slowly moving to container based workflow with podman (and not toolbox as you end up having your r package install directly in home but that can be worked out by specifying the ribs path). I use docker image from the rocker project: https://rocker-project.org/

If you have such specific requirements it’s often easier to use a container like the one from rocker (runs in der docker) instead. Btw wouldn’t be surprised if you’d get the latest version running in there as well.

Check out the Rocker Project, comprising of Docker containers for R, and can be used with RStudio. Also, virtual environments e.g., renv package can also help solve the package versioning issue, aside from containerization, and is transferable to a new machine via the renv::restore() function.

https://hub.docker.com/u/rocker has a lot of R-related images and they look pretty legit (look at "Tags" to find different versions). Don't use weird looking images. There's a lot of malware out there. Here's a guide on nice docker files: https://docs.docker.com/develop/develop-images/dockerfile\_best-practices/

Other people have mentioned renv and packrat already (hasn't renv basically superseded packrat at this point?), but what is also nearly ready-made to deal with this is rocker's R images. They have a bunch of images preconfigured for typical TidyVerse stuff, Shiny, etc.

The Apple M1 is ARM. As far as my knowledge, ARM isn't supported. Looks like the rocker project is aware of it. Considering how popular these chips are, i'm confident a lot of smart people are working on it. :)

But, with docker, there are many prebuilt images provided by the RStudio team directly, and other great repositories from rocker. These are basically images for your full SDLC with R, from development to deployments.

To get up to speed with TensorFlow, check their quickstart Support TensorFlow on GitHub ⭐

Explore here.

Part of the major userbase pull in GitHub revolves around hosting a considerable number of popular projects including Angular, React, Kubernetes, cpython, Ruby, tensorflow, and well even the software that powers this site Forem.

Right but that's not an inherent GPU determinism issue. It's a software issue.

https://github.com/tensorflow/tensorflow/issues/3103#issueco... is correct that it's not necessary, it's a choice.

Your line of reasoning appears to be "GPUs are inherently non-deterministic don't be quick to judge someone's code" which as far as I can tell is dead wrong.

Admittedly there are some cases and instructions that may result in non-determinism but they are inherently necessary. The author should thinking carefully before introducing non-determinism. There are many scenarios where it is irrelevant, but ultimately the issue we are discussing here isn't the GPU's fault.

and things like y = layers.ELU()(y) work as expected. I wanted to see a list of the available layers so I went to the Tensorflow GitHub repository and to the keras directory. There's a warning in that directory that says:

Exactly what language do you think TensorFlow is written in? :)

You can use Tensorflow's deep learning API for this.

Since you've specifically mentioned ML, here's Tenserflow's GitHub. I'm sure a quick glance through that will change your mind.