Roadmap

There's a few things that would be nice to have. Feel free to start a discussion on these topics in github.

Curated Word Lists

If people want to investigate bias in word embeddings then it would help if they had word-lists at the ready. But you can also imagine that it would be nice to have a word list of words that have multiple meanings. It'd be especially nice if we can support these word lists in many languages.

Multiple Plot Metrics

At the moment we only project onto axes to get x/y coordinates. It might make sense to show the cosine distance to these axes instead.

And if we're allowing cosine distance ... we might allow for flexible distance metrics in general.

Table Summaries

We've got a focus on charts now, but one imagines that calculating tables with summary statistics is also relevant.

Languages

It might be nice if we could help the user automatically download embedding files for supported backends.

Testing

  • it would be nice to have a good way of testing the charts
  • it would be nice to be able to test multiple models without having to download gigabytes into github actions