Just embarking on the fast ai Practical deep learning journey…Just getting started…
I noticed that there’s a workshop by Jeremy with a few required installation tutorials. My question is do I need to install the Ubuntu app to go through the course, I already have anaconda installed and a jupyter notebook environment.
For Windows users, you will need to install a terminal with the bash shell (Mac and Linux already have one). Note that this is only necessary if you don’t use one of the ready-to-run Jupyter options for the course."
Hi Sujatha. The Ubuntu “app” in Windows 10 is provided through the “WSL,” or Windows Subsystem for Linux. It sort of a “linux emulator” running inside of Windows, so it acts similarly but not identically to a real install of Ubuntu. Notably, it doesn’t seem to have access to the GPU and there’s some weirdness about file locations and performance of the file system. That will, however, provide a viable bash-compatible terminal that works mostly like a terminal on an Ubuntu system.
It seems like some people have been able to do the workbooks in Windows, but I found it to be an exercise in frustration. On my desktop system with an nvidia card, I bought a small SSD and installed Ubuntu on it (allowing me too boot into either Ubuntu or Windows). I have found that to be much better. If you have never used linux it can be a bit intimidating, but Ubuntu makes it about as easy as it’s going to get. If you plan on doing much ML stuff it’s probably worth learning. It would seem as though linux in general, Ubuntu in particular, are what most people use and are therefore most likely to just work without too much headache. Use Ubuntu Desktop 18.04 LTS, and you will have to turn on the support for nvidia binary drivers if you want to use an nvidia card.
Alternatively, you could use one of the many remote server options discussed in the documentation. This is almost certainly the way to go if you are using a laptop (which likely doesn’t have a fast nvidia gpu). Then you don’t have to install anything locally. That’s the recommended approach anyway.
Checkout https://course.fast.ai/ and look at the “Getting started” and “Server setup” sections on the left-hand navigation menu.