PyTorch is awesome. However, PyTorch functionalities only include widely used operations. As is mentioned in the contributing guide.
"Adding operators/algorithms from recently-released research is generally not accepted unless there is overwhelming evidence that this newly published work has ground-breaking results and will eventually become a standard in the field. If you are not sure where your method falls, open an issue first before implementing a PR."
However, we want new things! We can look for independent Github repos, but it is tiring, and not everyone is conformed to the PyTorch standard (some implementations might be wrong). Motivated by Tensorflow Addons, we implement an addon library where one can use a standard PyTorch interface to try out newly created methods! Activation, optimizers, layers, and networks. Whatever you want.
Note that not all code is contributed by the core authors. If it is borrowed from another repo, then make sure to mention them! Also, this is highly experimental, so no guarantee it will work! If it doesn't, either the new idea probably doesn't really work, or maybe it is a bug. Always feel free to raise an issue.
If you want something, also raise an issue. We will get to you as soon as possible.
The documentation will be generated by sphinx, similar to that of PyTorch.
Standard Github contribution routine (nothing new). However, pay attention that we want to use sphinx to generate documentation. Therefore pay attention to the docs!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent