The math is often behind me, I just try to visualize what I can. So I don't know how interesting this would be to you, but these folks are pretty good at math and trying to figure out what's going on in the latent spaces: https://transformer-circuits.pub/

Author of the project here. Definitely appreciating the supportive comments. I'd be happy to answer questions folks have and am very interested in what kind of data folks end up visualizing with it!

Did you look at using PHATE for dimension reduction?

I hadn't seen it, looking at the API it seems like it could be pretty straightforward to drop it in and see how the projections look.

I'd be really interested to see if that works out. There's some interesting comparisons here; https://www.nature.com/articles/s42003-022-03628-x

some tutorials i've collected recently * https://mode.com/sql-tutorial/sql-joins/ * https://github.com/NUKnightLab/sql-mysteries * https://selectstarsql.com/ * https://sqlbolt.com/ * https://www.craft.do/s/VzvaPiNX6jvxX5

one thing I love about DuckDB is that it supports Parquet files, which means you can get great compression on the data. Here's an examples getting a 1 million row CSV under 50mb and interactive querying in the browser: https://observablehq.com/@observablehq/bandcamp-sales-data?c...

the other big thing is better native data types, especially dates. With SQLite if you want to work with timeseries you need to do your own date/time casting.

Yes, it is always difficult to use dates in SQLite... DuckDB makes dates easier - like they should be!

We are hosting a d3 meetup online Thursday 8/27 (tomorrow) more info here: http://meetu.ps/e/Jh8GY/1kwFr/d

Mike Bostock will be doing an AMA at the end if you have burning d3v6 questions (though you can expect some more documentation on upgrading to come very soon!)

I'll also plug our other two speakers, Mike Freeman and Amelia Wattenberger who are amazing d3 educators & authors.

It's exciting that this is the first big online d3 meetup since the pandemic put a stop to in-person events, and as a consequence of being online anyone can participate!

oh man, I'd love to attend but that is such an awkward time for me.

Will the talk be recorded and hosted anywhere?

Yes, the recording will be posted to the Observable to YouTube channel after the event. https://www.youtube.com/channel/UCCD2tAKN32ya7V639gkbWhg

yup, our meetup recordings end up on YouTube at https://www.youtube.com/observablehq

clearly it's the sophons.

(reference to the Three Body Problem, an awesome book)

it is alacart and baked in at the same time! This article breaks down the API in graleater detail. https://medium.com/@Elijah_Meeks/d3-is-not-a-data-visualizat...

All of the functionality in d3 is split up into smaller modules that can be required independently.

I think this article does a great job breaking down the d3 API and answers your question in detail “D3 is not a Data Visualization Library” @Elijah_Meeks https://medium.com/@Elijah_Meeks/d3-is-not-a-data-visualizat...

I think this is why it is necessary to fund "searching for problems" research. We don't need to solve the 1970s vision of computing, we need to solve the 2020s vision of computing, whatever that may be.

Don't forget that programming languages get more abstract and hence more powerful with fewer lines of code. RealTalk has functions you can call to get the physical location of a piece of paper and being based on Lua makes it much more expressive than C++. You can keep digging down levels of abstraction but the interesting parts of this system are extremely accessible.