deel-ai / influenciae

👋 Influenciae is a Tensorflow Toolbox for Influence Functions
https://deel-ai.github.io/influenciae
Other
55 stars 3 forks source link

Thomas/wip Introduce base setup #1

Closed fel-thomas closed 3 years ago

fel-thomas commented 3 years ago

Initial Setup of Influenciae

This PR aims at setting up the basic structure for the influenciae project: the test environment, the ci, the pypi deployment and the documentation.


We start by c7524b888abe5823cf84610294e9d7f11e15920b that introduce the MIT License, same as the other DEEL libs (deel-lip / xplique). To ensure we don't push our envs file 574a6f8e55d07ec7f3e28946b1882c37b52c0531 introduce a basic python gitignore. To finish the preparation, we add base dependencies bb5027d2a59a2942c759770bc3f506276b014a64 : tensorflow and numpy. Is it enough for the moment or shoud we add other lib?

We can introduce our base structure 853f7a6e735cd799f4beb2986b1f1d91fb4516ef for influenciae, for now a common module for utils, a typing module if we need to define custom types and an influence module where we will have our methods. 5b2b27f6ad863e794bae9093d42680dd24dc1194 add the setup to build the lib.

We introduce the documentation c7353ca5abcbb1ff730a909f229c0e06eaf56a19 build on top of mkdocs and mkdocs-material , numkdoc for parsing. It means numpy like docstring, you validate ? For typing, a simple mypi config 6e02f858944073923dc53e61d37eedaaafe1f44c. For the versionning, a13cac4846ade423e547c4ad86cadd96f5f147b4 bump2version to help us.

Finaly, for the CI pipeline d6bb644ac643e95781ab7a83077ed3a023fd22f3 0c9765670ebc4d62834f1798ddbe2eb311cfe744 introduce all we need to have a proper ci pipeline for testing linting and publishing on pypi

lucashervier commented 3 years ago

LGTM :) I just want to draw attention on the TF version which might be a tedious subject!