Marten Lienen

@AmitMY These three steps would work for my use case :+1:

I have tried `:type nil` to get automatic byte-compilation etc. for local mini-packages that I have in my large dotfiles repository. So they are not in their own package repository...

Also works for me. Thanks a lot!

I have found a workaround. Let's say you have the following function ```python def f(x: TensorType["batch", "feature"]): return x.sum() ``` which you want to use in TorchScript. TorchScript does not...

You are right, I just saw the wrong reference targets in a paper of mine where I used hyperref's `backref`. If you think that it is not possible or too...

+1

Could you please post the exact error message? And also the change in the opening post is unacceptable. You cannot just delete the comment and change the behavior of a...

I think, the easiest solution is to install react via npm with `npm install --save react`. This will install it into the `node_modules` directory, from where you can require it.

`assets:precompile` should be run in `production` environment, so `RAILS_ENV=production rake assets:precompile` or simply `rake assets:precompile`, because this task uses the `production` environment by default, I think. Does this fix it?

Thanks for the contribution! Are you running the code on windows? Do your changes also work on other OS? Can you explain the rationale behind your changes to me?