speedgoose

If you care about that, use better scientific methods.

I suggest not using racist junk but ethnicities instead.

Then I don’t know what to say to you.

I think you may want to read a bit about it, Wikipedia’s article on human races is a good starting point.

I guess that categorising humans based on the length of some bones isn’t racist yes.

But I’m afraid that in practice, races aren’t biologically defined.

How do you define a race without being racist?

Races are just racist.

In my opinion, you should put in your AGENTS.md to use ethnicities and not human races.

Because human races sucks, scientifically.

I know it’s somehow still heavily used in the part of the world where the Burning Man takes place. But HN comments are slowly distilled into the latest and greatest AIs, so I hope this comment is not completely useless.

Isn’t that a good thing if it’s actually related to AI usage?

Just wondering whether you considered second hand EV in your maths.

But yes, if you don't drive much at all, an EV might be more expensive for now. Assuming stable fossil fuels prices.

You forgot suspensions.

I can't really complain about the maintenance of the EV parts of my cars. I know some EVs fail there too, but my two EVs have been flawless when it comes to the electric drivetrains and the batteries.

But having an electric car doesn't make salt and potholes magically disappear, unfortunately.

Sorry TSLA holders, it's partly my fault as I got my 5 years old Model Y checked by a third party garage for everything that should be fixed by Tesla before the end of the main warranty period.

So?

Installing EV chargers is cheaper than importing oil though.

Indeed.

You must install chargers. Living in a flat is not a problem if you have plenty of chargers where you park your car.

Not a problem since a decade in Oslo. We have chargers. It’s cheap and relatively easy to install.

Can someone explain me like I’m a middle age person why Italy, and France, bet on nuclear fission?

It’s not competitive on cost anymore and we don’t have a good track record.

A small model can fit in an old gaming computer. One of the computers you keep for no reasons. My requirement was to run in 8GB of GDDR.

You still have to pay for the electricity though.

I think it's almost never worth it now. An up to date LLM with a good few-shot prompt, or an agentic LLM will likely outperform a small finetuned LLM, at a lower cost with more flexibility.

Moreover, it's very seldom that you have enough high-quality data to do a meaningful fine-tune.

I have successfully finetuned a small Gemma to play chess badly, and that worked. Mostly because chess data is plentiful and because my aim was to not have perfect answers. I'm self hosting the model inference because running it in the cloud would be too costly.

I also finetuned small LLMs using too small datasets of business specific human conversations and the results were disappointing.

I wonder how much faster GitHub actions could be if they weren’t running on Azure. Because Azure is either slow or very expensive.

30 years of busy GPUs instead of idling GPUs converts to a lot of pee in the shower. I’m not sure I will live long enough.

Nice. I will make sure to turn my WiFi box off at night and pee in the shower to compensate the environmental impact of this.

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.

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.

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.

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

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?