I included comments from the refactor pull request that were not addressed:
[ ] Repository is too big. We will need to remove all files that were added here by mistake.
[ ] Add LICENSE, CODE_OF_CONDUCT.md, and CONTRIBUTING.md files to be the real open-source software:)
[ ] If possible add tests and set up some Continuous Integration (CI)
[ ] Create proper installation instructions, with required dependencies, constraints, etc. Make Dockerfile with it (at least for running prediction, not training).
[x] Make sure we are not bound to Python 3.6 as it will be deprecated soon
I included comments from the refactor pull request that were not addressed: