The voice actors usually try to match the timing of the mouth flaps in the animation. It's hard to do that while also conveying the meaning of what is said correctly (enough) -- and practically impossible to do that while sounding natural in English too.

That doesn't help for sure, but the intonation is ... odd.

I do something similar with Redirector, but a bit more involved (to also redirect mobile links to the regular pages + a few other cases):

    {
        "description": "Wikipedia Mobile -> Regular",
        "exampleUrl": "https://en.m.wikipedia.org/wiki/Hypernucleus",
        "exampleResult": "https://en.wikipedia.org/wiki/Hypernucleus?useskin=vector",
        "error": null,
        "includePattern": "https://([^.]+).m.wikipedia.org/wiki/([^#?]+)(#(.+))?",
        "excludePattern": "",
        "patternDesc": "Remove .m and add ?useskin=vector to end",
        "redirectUrl": "https://$1.wikipedia.org/wiki/$2?useskin=vector$3",
        "patternType": "R",
        "processMatches": "noProcessing",
        "disabled": false,
        "grouped": false,
        "appliesTo": [
            "main_frame"
        ]
    },
    {
        "description": "Wikipedia (Use Vector instead of Vector2022 skin)",
        "exampleUrl": "https://en.wikipedia.org/wiki/Potato",
        "exampleResult": "https://en.wikipedia.org/wiki/Potato?useskin=vector",
        "error": null,
        "includePattern": "https://([^.]+).wikipedia.org/wiki/([^#?]+)(#(.+))?",
        "excludePattern": "(https://[^.]+.wikipedia.org/wiki/Special:.+)|(https://([^.]+).wikipedia.org/wiki/[^?]+[?].*useskin=.+)",
        "patternDesc": "Add ?useskin=vector to end",
        "redirectUrl": "https://$1.wikipedia.org/wiki/$2?useskin=vector$3",
        "patternType": "R",
        "processMatches": "noProcessing",
        "disabled": false,
        "grouped": false,
        "appliesTo": [
            "main_frame"
        ]
    },
    {
        "description": "Wikipedia (Use Vector instead of Vector2022 skin) -- Query String Redirect",
        "exampleUrl": "https://en.wikipedia.org/w/index.php?title=Petabyte&redirect=no",
        "exampleResult": "https://en.wikipedia.org/w/index.php?title=Petabyte&redirect=no&useskin=vector",
        "error": null,
        "includePattern": "https://([^.]+).wikipedia.org/((?:w/index.php)|(?:wiki/([^#?]+)))[?]([^#]+)(#(.+))?",
        "excludePattern": "(https://[^.]+.wikipedia.org/wiki/Special:.+)|(https://([^.]+).wikipedia.org/wiki/[^?]+[?].*useskin=.+)",
        "patternDesc": "https://([^.]+).wikipedia.org/wiki/([^#?]+)[?].+useskin=.+",
        "redirectUrl": "https://$1.wikipedia.org/$2$3?$4&useskin=vector$5",
        "patternType": "R",
        "processMatches": "noProcessing",
        "disabled": false,
        "grouped": false,
        "appliesTo": [
            "main_frame"
        ]
    }

> it pulls around 400W

Try dropping the power cap on your GPU if it supports it; you can often get much lower energy usage with minimal loss of tok/s (particularly during generation) than whatever the GPU defaults to. There's a sweet spot around 200W on the GPU I'm currently testing that gives me about ~75% of the max pp and 97% of tg while using 100W less than the default/max 300W power cap -- and the card runs much quieter as a result.

> AI for product development and management would be far more impactful than automating rote coding tasks [...]

Yeah, if this stuff actually worked that well already, OpenAI et al. would just run AI CEOs and engineers. Why get some other company to pay you at all when you can automate every other company out of existence and take all the money they make?

The fact of the matter is that while the tech has some uses, it sure as hell isn't a full scale replacement and you almost always actually have to massage the input into LLMs to get anything decent back out in practice. Some CEOs and managers can learn to do this, of course, and some already are... but that quickly turns into a second full time job. A "programmer" is still needed. The job might change from mostly hand-writing C++/JS/Python to prompt engineering + some manual coding to fix all the stupid fuck-ups that the bots can't solve themselves, but you still need someone to actually prompt the bot.

When that changes, it won't just be engineers losing work; there will be no reason to even have a human CEO any more.

> When that changes, it won't just be engineers losing work; there will be no reason to even have a human CEO any more.

The human race isn’t ready for that world IMHO. The only reason there is a middle class is because people have leverage in the form of their labor. When that becomes worthless … the people who own stuff and make their living from doing so won’t hesitate to get rid of everyone else - whom are now worthless to them.

Humans will revolt and mutiny on the ai ceo so fast its not even funny

Imagine being deluded enough that you, a CEO, can continue to go buy groceries and drink at a cafe relaxing in that future

27B is the dense one. Try the Qwen3.6-35B-A3B variants for the MoE release. That's what I'm running on a Framework Desktop and I get ~50 tok/s plus or minus a few. The dense one is similarly slow for me -- not sure what to expect on your hardware from the MoE but it should probably be much faster.

Thanks!

I appreciate the sentiment, but this law looks counter-productive if I'm reading it right. This is OBVIOUSLY going to push companies to make games into a subscription service to bypass the law even if they wouldn't have normally:

> (b) This section does not apply to any of the following:

> (1) Any subscription-based service that advertises or offers for sale access to any digital game solely for the duration of the subscription.

This exemption NEEDS to be removed. If a game's official servers are taken down, the community needs to be given the ability to keep running it themselves. Full stop. No exceptions.

Games that the the company is allowed to kill whenever they want already are a subscription. Forcing companies to be upfront about that when they are not willing to provide the claimed deal is not counter-productive.

I don't think GitHub has made a single UI change since ~2023 (when it went JS heavy) that I've liked. (Admittedly though, I've moved away from it for everything I have a choice about at this point, so it's possible they snuck in some good stuff when I wasn't looking.)

Also: having trouble getting this specific link to load -- just getting the unicorn error over and over.

> The notation ExMm denotes a format with x exponent bits and y mantissa bits.

Shouldn't that be m mantissa bits (not y) -- i.e. typo here -- or am I misunderstanding something?

You're correct yeah, 'ExMy'.

> it's better anyways and surely close to a decade after coming out, we'd expect devices to support it well enough.

A lot of people, myself included, are still using quite old hardware. The GPU in my daily driver is ~10 years old at this point. Between crypto, COVID, and this AI craze raising GPU costs by insane amounts, it hasn't made sense to replace it with something newer. I know I'm not alone on that...

For legacy devices, VP8/VP9 is a good option. Intel Added VP8 hardware decoding to Broadwell which was 12 years ago. Nvidia had hardware VP9 decoding 10 years ago on the Geforce 10 series. AMD had hardware VP9 decode support 9 years ago on the Radeon 400 series.

If my 10 year old card can't encode in hardware, it's a nonstarter.

I still happily use my 2012 27" iMac for all my work. So I'm with the parent comment.

I work in AI and I'm surrounded by RTX-4090 and H100 servers but for much of the day to day AI training I use my RTX-970 in the desktop on my desk for convenience and it works just fine for most cases.

Literally 2 meters from my desk is a 2x RTX-4090 server and many times I just use my 8 year old GPU anyway so you don't need it.

For a long time I thought my RTX-2060 was just not capable and the other day I did a ffmpeg GPU transcode and was surprised by how well it did. So now I am thinking about putting on some of Google's new Gemma edge models (probably the smallest will work with my 6GB VRAM + 2 GB) setup. I am not a 100% sure what that 2GB is but I think it is borrowing from the system in some manner.

Video encoding uses dedicated silicon, it's not using the card's compute.