Does the `install-gpu.sh` install pip stuff at conda root? and why?

I’m looking through https://github.com/fastai/courses/blob/master/setup/install-gpu.sh

Why isn’t there an extra line at 26 say “conda create -n fastai” or something similar to make sure that everything is installed in a conda virtual env? What is the benefit of not doing this? I thought the point of conda was to “vm” everything.

I certainly wouldn’t say that’s the “point” of conda. For us, we use it for simplicity, and creating an additional environment is less simple. In general I prefer not to use environments or virtualenvs for data science prototyping or learning - you may find them helpful however for developing when there are multiple projects with different platform needs.

BTW, could you please change your username? Calling yourself ‘fastai’ may lead to confusion if other folks assume you are a site admin.

I see.

Sure. Do I need to make a new account?

Ok. this is my new account. I didn’t think of the confusion it might cause, i just couldn’t think of a username :slight_smile:

BTW, @jeremy thank you for running this class and answering my question(s)

Edit: why does this keep getting marked as spam?

Thanks for changing your username! It got marked as spam since there were 2 accounts created at similar times from the same IP address, but I’ve cleared you now.