Documentation request: sentence similarity with {tidybert}

[MattCowgill]

Hi,
Thank you for such a great package, bringing the power of BERT to R!
I've worked through the examples in the documentation, thank you for writing them.
This is a request for additional documentation demonstrating how to use {tidybert} to measure the cosine similarity of sentences/texts.
Thank you!

[jonthegeek]

We actually have a vignette related to that over in {torchtransformers}.

The problem with putting a vignette like that into {tidybert} is that off-the-shelf BERT isn't that great at creating meaningful sentence embeddings. I need to think a little about what we'd want to show (and potentially implement as a new function or feature) here.

[MattCowgill]

Thanks @jonthegeek