Hacker Newsnew | past | comments | ask | show | jobs | submit | speedgoose's commentslogin

I would need proper benchmarks but in my limited testing on my Phone using Qwen 0.6b, this doesn’t work well.

Between "brocoli and poop soup" or "cake", it recommends me to eat the soup.


It's got fiber and some extra bacteria for your gut biome!

Human made summary: A telecom company had time synchronisation issues in a location of their infrastructure. They eventually used a single hardware GPS clock that fixed the issues but became their only authoritative clock in this location.

One day, the clock was turn off and on and it went 1024 weeks backwards because the time GPS time protocol sucks and use a week counter with too few bits.

Apparently a GPS clock can keep track of the time if it’s up and running when the week counter overflows, otherwise it has to take a wild guess. Apparently their hardware GPS clock used a hardcoded start time from its firmware instead of trusting a less reliable existing clock.

The article finishes with some AI looking suggestions to prevent such an issue to happen again.

Mine would be to have bought one or two more GPS clocks and not from the same provider.


> One day, the clock was turn off and on and it went 1024 weeks backwards because the time GPS time protocol sucks and use a week counter with too few bits.

As I pointed out in my other comment below, this is only true of the old L1 signal, not the newer L2C signal. If they had a newer GPS card, this would never have happened.

> Mine would be to have bought one or two more GPS clocks and not from the same provider.

I think it would be more important to have a newer one that doesn't have this problem, than two old ones which both do.


I don't think every GPS receiver necessarily has to have this problem, even with the old signal. It's not crazy to keep a week count in persistent memory and use that as a lower bound on startup so you only get a wrong time if the receiver isn't online for 1024 weeks.

Exactly. The sane ones just ask you what year it is at startup, and use that to determine the initial epoch. Then they handle rollover automatically since then.

It's been widely opined that 1024 weeks (~20 years) is the worst possible interval. Either rollovers should've happened VERY frequently (say, 128 weeks) so receivers would be FORCED to deal with it, or extremely infrequently (16384 weeks?), so it's simply never an issue.

The unhappy medium is long enough that developers feel justified in saying "naaaah, our receiver won't still be in use then, we can ignore that!", but in practice it's very likely to happen.


This is the opposite of a Google AI as summary in the best way.

Thank you for slogging through the slop to bring the human angle


No progress?!

On one hand, it's very annoying.

On another hand, you realised that you are vulnerable to denial of wallet attacks without loosing that much money.


true!

I didn't know 2 thirds of the training data would be source code.

that is the the "data used to improve the model" when signing up for the subscription plans

this is the rl run, not the pretraining run

even in pre-training, usually 30%-50% is code these days.

That would be far too high in my opinion. But happy if anybody can give insights from their own experience with pretraining runs.

I gave some very bad sketch made in jspaint to Astra, plus some text instructions, and I got back some clean STL file and a beautiful 3D render.

It even made a second STL file with the same model rotated for 3D printing.

I remember all those hours on SolidWorks many years ago, and now you can just speak and present bad drawings to the computer, and it does it okay.


Why not?

Yes sir. Let us know if you have more questions.

Is there any audience using a web browser that can negotiate a modern TLS connection to apple.com, but not support a better rendering method?

Well for starters, any modern web browser with JS disabled is going to display the original page just fine, but will not be able to render a canvas-based solution. That may put the user outside your target demographic, which is fine, but it is worth being aware of.

What exactly do you mean, the original animation was done in JS?

Ignoring the use of JavaScript in the old demo, is there an individual on earth that uses a modern web browser with JavaScript disabled, wants to order a personalised Apple product, and can’t enable JavaScript back for previewing?

I once encountered a joke building that was so large, I remember in the magnitude of 500km2, that you couldn’t edit it in the web editor. You had to download a desktop application to delete it.

It was in a place of the world that is unlikely to be popular within the OSM community.


Fortunately it's fairly easy to query the database for that kind of thing and a few editors periodically go through the most implausibly-sized common features, so anything too egregious will get noticed eventually.

The last time I saw a collection of "biggest objects" for various tags it was mostly things that were just subtly mistagged (e.g., an entire forest being tagged as a single tree) rather than vandalism.


> e.g., an entire forest being tagged as a single tree

If that forest's dominant tree happens to be a quaking aspen then technically speaking it might not have been mistagged.

https://en.wikipedia.org/wiki/Pando_(tree)


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: