Yet another dockerfile

In order to get rid of the blocking message:

The following packages have unmet dependencies:
nvidia-docker2 : Depends: docker-ce (= 18.03.1~ce-0~ubuntu) but 18.05.0~ce~3-0~ubuntu is to be installed or
docker-ee (= 18.03.1~ee-0~ubuntu) but it is not installable

I’m told by Nvidia-docker admin that edge releases are not supported and to disable the edge channel from your Docker repository entry in /etc/apt/sources.list or /etc/apt/sources.list.d/…
exchange is here:

I did replace edge by stable in that file but the error message remains the same, !?!
i would really like to move on this week-end with Fastai course and tests…

i finally managed to install without docker by following the fastai papersspace script on this page:

went flawlessly,

:nerd_face::tada::clap::clap::clap::confetti_ball: Congratulations!
You got it done. Excellent.

Congrats! I guess the docker version is more complicated, the reason I use it is, that I want to isolate other environments from the pytorch/ installation. Have fun with the course!

i would have liked to use Docker too for the same reason, but after 3 weeks i had to get going.

But i will revisit it, i think the solution to my problem is simple, according to Docker admin, i had to remove my Docker CE installation first (it was the edge version) and re-install the Stable version.

But not sure why i ended up with the edge version, indeed my Docker.list file contained edge instead of stable.

Issue resolution is here for these who bump into the same thing:

I have got everything working with this docker file, so it is loading cats! Hooray!

I quickly run into a problem as I start in with the Convlearner. It downloads the resnet34 file correctly into the container, but it keeps throwing the same error (even after rebuilding).

UnicodeDecodeError: ‘utf-8’ codec can’t decode byte 0xff in position 30: invalid start byte

Any hints? I am guessing there is data it is not seeing on the host or in the container.

Edit: I needed to run git pull as suggested by this thread.