I love my kindle for this exact reason. I never shy away from a book that's "too big" anymore because I have no idea how big it is!

For those as confused as me, I'm pretty sure this isn't "new". Numberphile talked about doing even better than this several years ago: https://www.youtube.com/watch?v=5q32heFz1bs

According to the wikipedia article linked at https://news.ycombinator.com/item?id=49559416 this discovery is from around the same time as that video came out. This submission should be tagged (2023).

ETA: Oh I just looked at the linked article again and it was just published. Weird.

Mine gets stuck at loading 90%.

The traffic right now is insane, 200 people trying the demo at the same time :\

Isn't it just static files? Use a cdn like cloudflare or AWS CloudFront or depending on size - if you can break it into chunks of like 50mb each, GitHub pages (free)

200 people or 100k people wouldn't matter then, it'll be cached.

It will also make the download based almost entirely on client download speed.

is 200 a lot of concurrent people? What is required from the server in order to render the demo, can't most of the work be pushed to the edge devices?

yeah I probably shouldnt have included a 100mb gaussian splat file to load each time.. I had no idea so many people would try it :)

Any video demo available - love to see it in action

Why submit an issue rather than just making the fix and adding the tests in PR? Seems like they're just making work for the maintainers.

OP here: A bug report just needs a proof of existence for the condition while a bug fix needs a proof of correctness. Sometimes is the best to let the developers who are day to day in the codebase to choose the best fix and if they what to fix it.

Just wanted an open place to explore interesting mathematical concepts :)

Isn’t the number of turns most important? Some agents take repeated input, while others can mostly one-shot what I’m looking for.

I think he means that if you ever accidentally tie the knot wrong, then it can end up difficult to untie.

I don't think I tied it wrong. I think it eventually just tightened up with usage and got really hard to work with.

Fun game! There should be a check mate though, right?

I thought the goal was minimal instruction to let Claude determine the best way to solve the problem. Not adding this to my workflow anytime soon.

It is not for the end user, it is more for things like wrappers and automation scripts.

Nobody expects the end user to prompt the AI using a structured language like xml

Am I the only one who thinks taking a huge model trained on the entire internet and fine tuning it is a complete waste? How is your small bit of data going to affect it in the least?