moving examples out of the package

[Borda]

Hello, nice peace of work. I was wondering if it would be easier to have examples out of the package (more intuitive to finds and keep the package simple) as well as all tests?
(e.g. pytorch-lightning/pytorch_lightning/testing_models/lm_test_module.py)

[williamFalcon]

yeah, the problem was that there were issues with the test models being outside of the package.

But if you want to make the change, I do think it would be cleaner. Feel free to submit a PR! It would be super helpful!

Tests will also need to be updated.

[williamFalcon]

Awesome PR, thanks for submitting! Interested in looking at any of the other issues we need help with?

@Borda

[Borda]

I was thinking about some workaround automatic testing, formation and coverage...
I will prepare another PR :)

[williamFalcon]

amazing! That would be great. Right now a big issue is having to manually run the GPU tests for free.

I'll create an issue to track these changes there

[gregunz]

I am new to python packages, modules and stuff but I think that now that the directory examples is in the root, the setup.py detects it as a package with find_packages().

This results in anyone installing your pytorch-lightning package to also have a examples package installed into their environment.

As a matter of fact, it is preventing me from loading one of my local module named examples.

[gregunz]

The above commit could solve this.
It simply excludes the directory from the package.

Note that it also excludes the tests which should be causing the same type of problems.
I will proceed with a pull request if it makes sense to you @williamFalcon @Borda.

[Borda]

I do agree with the change, or even explicitly name the 'pytorch-lightning'
Btw, why the package is imported pytorch... not just torch... such as PyTorch

[gregunz]

Whitelisting as you suggest seems a good idea BUT doing so, sub-packages will not be imported.

About the name, I also think torch-lightning makes a lot of sense.

[williamFalcon]

let’s keep it as is for now. i don’t want to introduce breaking changes often (pytorch-lightning)