Type annotations for Approximations #258

manel1874 commented 2 years ago

Description

Added type annotations to the approximations module of SyMPC:

[x] exponential.py
[x] log.py
[x] reciprocal.py
[x] sigmoid.py
[x] softmax.py
[x] tanh.py
[x] utils.py

This fixes #258.

Affected Dependencies

None.

How has this been tested?

Tested each function individually in a jupyter notebook.

Checklist

[x] I have followed the Contribution Guidelines and Code of Conduct
[x] I have commented my code following the OpenMined Styleguide
[ ] I have labelled this PR with the relevant Type labels
[x] My changes are covered by tests

gmuraru commented 2 years ago

Hello @manel1874. Thank you for the PR - I have some suggestions. Probably you will encounter the circular import problem if you import the MPCTensor.

One solution for this will be to create a register decorator that will keep track of all the approximations we have - something like:

@register_approximation("log")
def log(....)

And this decorator will add the functions in a dictionary.

Then in the MPCTensor we will import a wrapper class around that dictionary and use that to call the approximations.

Let me know if this sounds ok - also if you have more questions - feel free to message me on Slack.

gmuraru commented 2 years ago

@manel1874 could you do a rebase? there was a problem with the black code format tool, but it should be solved now.

OpenMined / SyMPC