Self-contained reproducible environment

[daniel-j-h]

Hey folks, we ported the r(2+1)d model and weights to PyTorch over in

https://github.com/moabitcoin/ig65m-pytorch

and had to run this project to e.g. get out activations to compare and verify the ported model.

The problems we ran into when trying to run this project were

• No dependency management: Python packages and system packages are listen in the installation document only and are missing a few key packages
• The project requiring Python2 and e.g. its cPickle (btw Python2 will EOL in a few months)
• We have to compile Caffe2 from source and make sure the video ops are in

Here are a few improvements which will give us self-contained and more importantly a reproducible environment for users, engineers, researchers, and scientists to run out their experiments.

Changes

• Adding Dockerfiles for both CPU as well as GPU (based on nvidia-docker)
• Tracking Python dependencies and locking down the dependency tree
• Providing a convenient Makefile to build and run the environment

This should resolve the problems I noted above.

Let me know what you think.

[facebook-github-bot]

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file. In order for us to review and merge your code, please sign up at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need the corporate CLA signed.

If you have received this in error or have any questions, please contact us at [email protected]. Thanks!

[facebook-github-bot]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Facebook open source project. Thanks!