Best way to make a smaller "PixPlot"

[fr1ll]

I have been using PixPlot for a couple years and greatly enjoy it.

At the same time, I've had some pain points and would like to make a version that does a bit less and works a bit differently. Mainly, I'd like to outsource the embeddings stuff and make that easier to parallelize.

I've started to think through the idea here:
https://github.com/fr1ll/clip-plot/

Given that you (mainly @duhaime and @pleonard212 ) have done great work and also have been quite careful with branding (DHLab logo, etc.), I want to make sure -- do you think a fork is the appropriate way to do this? If I change the name of the fork (to, say, clip-plot), is that a good thing, or does that take away from the recognition you deserve for this if anyone uses it?

I know the license is MIT, so there's some leeway to what I could do here, but I'm new to this sort of thing and don't know what's the right approach.

Let me know if you have any thoughts or would like to discuss in some way!

[duhaime]

This is exciting @fr1ll ! As far as I'm concerned a rename + fork is totally fine. For what it's worth, we worked on a rewrite of PixPlot as well, making the system a bit more abstract so as to be capable of dealing with any data that can be discretized. Hopefully we'll be able to get that project out soon!

[fr1ll]

Thanks for the co-sign to trying this out! I will post an update here when I have a working version.

I'm excited to see what you did or are doing to make PixPlot more generic! I have some tabular data problems and graph data problems that could be interesting to visualize through the lens of embeddings + dimensionality reduction.