Looks like another Deepseek distil like the new Ministrals. For every other use case that would be an insult, but for coding that's a great approach given how much lead in coding performance Qwen and Deepseek have on Mistral's internal datasets. The Small 24B seems to have a decent edge on 30BA3B, though it'll be comparatively extremely slow to run.

Gemini manages to seamlessly weave a desperate cry for assisted suicide into anything it does haha.

I never read what it writes, I added an observability feature because my boss asked for that and yeah I could see it trying to order MAID over telehealth

The real joke is that we'll ever get another Llama iteration.

Your engineers were so preoccupied with whether or not they could, they didn't stop to think if they should.

Those predictions sound good enough to get you CIA funding.

Are there any pretrained models with this architecture yet or is it all still completely theoretical beyond Google's unverifiable claims? They published the original Titans paper last year and nobody seems to have built on the idea.

https://github.com/lucidrains/titans-pytorch - is the only public iteration.

But no one appears to have taken the risk/time to properly validate it.

The fundamental ideas in the paper aren't particularly novel. They will probably work as advertised.

This is over 200 convoluted rules with multiple subpoints on 140 pages, who the hell checks compliance with this without it taking over a decade?

You should see some of the manuals I go through for our some of IoT devices. As far as how Lockheed Martin does it, this quote from the article linked by bri3d answers it:

> MISRA-C was used as the basis for the C applications and a coding standards was developed with the assistance of Bjarne Stroustrup, original author of the C++ language. For both C and C++ Static Code Analysis (SCA) tools are used to ensure that restricted features are not utilized. Arguments about the lack of reliability in either C or C++ are addressed by programming standards restrictions and SCA checks. In truth, this approach is probably more consistent and robust than the manual checks used for previous development efforts including Ada.

Alright that makes far more sense than doing it the "bureaucratic" way. Non-compliant doesn't compile. Must be really annoying working on this codebase hah.

I would argue that it would be very annoying to work on a code base which wasn't like this. Having to spend so much time figuring out where your bottlenecks are and why they are there. Though you would make a fair point in regards to how much code-fascism you need when you try to make C/C++ safer than ADA. I'm sure they didn't come to this conclusion lightly, but the first thing to pop into my head is that it seems odd to do this mainly because of hiring challenges.

But C/C++ certainly did well enough for Lockheed Martin considering it's now one of their principal languages, if not the principal.

> A big chunk of our code was doing things that are very hard to do in other languages. The resulting software did things our competitors’ software couldn’t do.

I've never seen a general purpose programming language that couldn't do everything the underlying hardware is capable of. It could only be unperformant enough that you could call it unfeasible at worst. What's so hard to do in languages other than Lisp? Spam parentheses?

Well it would've been better if they kept the video on the actual camera that ate it instead of switching back and forth constantly. It's not Liam Neeson climbing a fence, guys.

GGUF when? /s

Despite what the average multinational will have you believe, terms and conditions usually don't hold up in court. If they write some illegal bullshit into it, it's just that, bullshit.

That may be true but doesn't help if not accepting the terms prevents you from using the device.

On a practical level you then at best have a battle to get a third party (the retailer) to give you a refund and most people faced with the option of removing and returning a huge expensive device like a fridge with no guarantee of a refund are going to just leave it.

It does need some stubborn and tenacious people to make a stand and set a president - perhaps backed by a consumer rights group but it's an uphill battle.

> huge expensive device like a fridge with no guarantee of a refund are going to just leave it.

oh I'll fix it with a hammer, or glue a piece of cardboard on it.

I paid extra for devices without WiFi when I moved house this year.

Sure, but that depends on the thing actually being illegal first. Genuine question - how often in practice are terms and conditions successfully challenged? My thought is that companies like that would be able to drain plaintiffs out before it getting that far very often

And how often in practice are terms and conditions attempted to be enforced in the first place? No need to challenge them if you can ignore them

If ignoring them is your only option, and challenging them would fail, we would expect to see a lack of challenging them. Which we do.

Unless there's a solid track record of people consistently challenging them and winning, we can assume, based on bayesian priors, that most people cannot.

Which makes sense: court costs money.