extheat

All sound is vibration, so everything is affected by sound at some point as the vibrations propagate through and physically move around materials. Humans can hear because the audible frequency range moves receptors in human inner ears, and as you can imagine this isn't local to the ear, the whole brain will be affected (but biology has accounted for this). If you knock on wood that sound is audible in many frequencies--below human audible, audible and also above human audible (what we call 'ultrasound'). What is more dangerous generally is sustained noise at any frequency; lower frequency + high amplitude is one of the most harmful as that can physically push you.

The reconnect thing is actually quite helpful for mobile use cases. Say the user switches the tab, closes their browser or loses network and then they return. Since SSE is stateless from the client's perspective, the client can just reconnect and continue receiving messages. Whereas with WS there's handshakes to worry about--and also other annoyances like what to do with pending requests before connection was lose.

They reopened with formal understanding that there will be an executive order tomorrow to suspend the enforcement of the ban. That is a big deal and it's something that they can point to to defend themselves in court should that happen. When President Biden signed the bill, it gave him the ability to extend the deadline by an amount which he declined to do (beyond saying "I'll let Trump admin deal with it"); and soon-to-be President Trump is saying he will do it tomorrow.

The Biden administration signed the thing into law. Of course they need to comply. And people are acting as if somehow TikTok decided to self-ban and have now un-banned. No, it's only those with the app already installed that are able to continue to use it. It's still blocked on the app stores, and will presumably stay that way until tomorrow.

If I want to run what someone else has determined as "malware" on my computer, as far as I'm concerned, I should have the absolute right to do it. Same for spyware. Why? Because I don't want the government to make the determination for what is right or wrong for me on my own property. If the US government wants to block apps on their property, then they can go ahead and do that. But the moment it extends to my own property, it's quite ridiculous to think people are going to bend over backwards and comply with what's good for you. Especially in the context of some vague national security threat, why am I supposed to be subversive to the CIA?

How can you complain about the CCP banning foreign social media and censoring when you have your own government willing to do the same thing -- in the name of Protecting the Democracy?

It's not about privacy or data or whatever the facade is. The crime that we are committing is none other than allowing ourselves to be fed information that could threaten the United States. So, therefore, even according to the SCOTUS, if Congress plasters the magical words "national security" in their laws, then the Constitution takes a backseat and we too can be like China/Russia/Iran. Will we start banning VPNs next--which circumvent our new found love for censorship? I'd not be surprised.

made them vote together across party lines

Ha, is that uniparty vote supposed to be something meaningful? If the government had true concerns, they could 1) be aired to the public and 2) other senators like Thomas Massie and Rand Paul would not be speaking against the ban.

People can change their views and minds. It's only a problem when you lie and pretend you didn't. Pres Biden signed the law and could suspend it now if he wanted, but he chose not to do it as it'd be contradictory to his own signing. And of course soon-to-be President Trump will get the credit for reverting it. Nobody cares about the details beyond those invested into politik.

Not talked about beyond heat is maintenance. When things fail, who are you going to send up to fix things? Is it never going to fail, or are they planning to throw away big components or are there going to need to be servicing missions?

Bluesky isn’t a centralized service, the developers cannot themselves comply with judicial orders (any of them).

They can get rid of apps and block websites that they want, which is basically what they’re doing right now.

Yes and appeal to whom? Himself, who’s clearly shown himself to be a partisan? Why even need an executive when your judiciary can basically unilaterally function as executive be a legislator in one? Obviously they’re is not the US, but that’s not an excuse to a ridiculous system.

There is much more story that you’re either uninformed about or willfully ignoring. The correct move was to remove people from harms way for decisions they have no control over, hence their staff exit from the country.

This is actually really cool. I can see the usages in textarea and inputs but also even in plain HTML documents for large code formatting. No need for exploding the DOM with tons of spans and other HTML tags, if you can have the hardware accelerated font rendering software take care of everything. I’d imagine that would save lots of memory and CPU.

A simple equation to approximate it is `memory_in_gb = parameters_in_billions * (bits/8)`

So at 32 bit full precision, 70 * (32 / 8) ~= 280GB

fp16, 70 * (16 / 8) ~= 140GB

8 bit, 70 * (8 / 8) ~= 70GB

4 bit, 70 * (4 / 8) ~= 35GB

However in things like llama.cpp quants sometimes it's mixed so some of the weights are Q5, some Q4, etc, so you usually want to take the higher number.

Obviously because the world revolves around US politics. What else could there be to talk about beyond politics.

Had the same first thought (:

The production simplicity of having a standardized OS and being able to drop in a .exe and have it run everywhere without worrying about building for 1000 system combinations cannot be beat.

There's nothing special about Windows beyond the fact that you can run arbitrary executable files. The problem could just as easily have happened for Linux or iOS/Mac and in fact it has. ChromeOS kind of works if you want to run a web application that's hosted on some web server... but it's not appropriate for running programs where a dumb browser doesn't suffice.

People simply have no idea what they're talking about. It's just jumping on to the latest hype train. My first impression here was per the name that it was actually some sort of compiler in it of itself--ie programming language in and pure machine code or some other IR out. It's got bits and pieces of that here and there but that's not what it really is at all. It's more of a predictive engine for an optimizer and not a very generalized one for that.

What would be more interesting is training a large model on pure (code, assembly) pairs like a normal translation task. Presumably a very generalized model would be good at even doing the inverse: given some assembly, write code that will produce the given assembly. Unlike human language there is a finite set of possible correct answers here and you have the convenience of being able to generate synthetic data for cheap. I think optimizations would arise as a natural side effect this way: if there's multiple trees of possible generations (like choosing between logits in an LLM) you could try different branches to see what's smaller in terms of byte code or faster in terms of execution.

What's the point of it then? If humans are not consuming it then there is no point to doing anything at all out there is it?

I'm not sure if it's intentional or not, but this seems to be barely discussed, and the page reads more like an slick advertisement. There is a paragraph expert about it and it's mentioned at start of the video but I don't see anywhere showing how the full device is setup or how it looks. From the images/video b-roll alone it seems almost as if the wrist bad is what's controlling it and not your toe. Not to nitpick of course, just makes things seem a bit rather misleading.

The "National security" needs of the day trumps the 1st amendment. Only U.S. (and a certain other place which won't be named) propaganda is allowed.

I am thankful we have LLMs so we don't have to deal with SO. Ideally, as little as possible. SO can be a pretty toxic place filled with elitism and care for procedure over actually helping people, which is not totally unreasonable from their standpoint but it's definitely not what people are visiting the site for. Quite ironically, one of the major complaints I get is that LLMs output wrong answers here and there, ignoring that many of the answers on SO are also completely wrong or irrelevant to the core question being asked. And mind you, also outdated (I regularly have to click through the sorting to make sure answers are actually still relevant).

If we could merge the two to get the best of both worlds, and have LLMs that know how to write well and are validated by humans on the site, that would be great. Maybe not great for the folks looking to accrue internet points but absolutely great for users.

Cool stuff. It's kind of interesting because the time zones are several hours off, so I wonder how well it works at night. From the renders I see the quality doesn't seem that great (for better or worse), still quite fascinating to look through I'm sure.

It's what enables you to even be here to hear OP. Somewhere along the line the magic got lost, in the name of higher level human problems, but it is quite literally one of humanity's most important inventions.

I'm not sure how much people actually read the link, but I'm talking about this "problem point" specifically relating to network congestion:

However, there’s a huge problem: the proposed rules make it possible for mobile ISPs to start picking applications and putting them in a fast lane - where they’ll perform better generally and much better if the network gets congested.

T-Mobile, AT&T and Verizon are all testing ways to create these 5G fast lanes for apps such as video conferencing, games, and video where the ISP chooses and controls what gets boosted. They use a technical feature in 5G called network slicing, where part of their radio spectrum gets used as a special lane for the chosen app or apps, separated from the usual internet traffic.

The FCC’s draft order opens the door to these fast lanes, so long as the app provider isn’t charged for them.

My views on this have changed over time. More and more it seems like policy people encroaching on technical decisions without fully understanding the rationale behind them. Is there a good reason to segment different types of content, such as for optimal network performance? Ultimately all wireless communications have to work within the bounds of physics no matter what the made up human laws say. If there's a limited amount of bandwidth you have to work with, does it make sense to waste all the bandwidth on streaming HD videos and block off all the other traffic in the name of "neutrality"?

I would understand if there's some profit motive behind this, but I'm not seeing it.

I'm personally looking forward to it. One of the biggest advantages with running locally is cost. Yes, the upfront cost to getting your own hardware is much higher, but I'd much rather pay for something once then worry about paying a monthly bill for using someone else's infra that they may at any time decide to pull out of service or hike the price for.

The "no data sent to servers" point I agree is super important for work things. I pay for and use Github Copilot for personal things but I'm not allowed to use it at work for obvious reasons. If we could remove the privacy concerns and run these models on local hardware (that people actually have) then it'd be game changing.

Reading some of the other anti-AI comments is a bit odd. I guess it's going to depend on what you do for work, but as someone who writes software not using Copilot is like comparing myself from a 0.25x developer to a 2x developer. I mean that when I say it, but I get that you do need to be able to identify wrong code faster than you can run it.

The industry (OpenAI, DeepMind, et al) are not seriously working on generate AI. Thats a phase. They are working on AGI. AGI definitionally includes all human cognitive capabilities and that includes vehicular operation and real world interaction (in form of embodied robots). It is likely possible that even self driving cars can be done safely without needing AGI with sufficient scale, and I don’t think it’s that far off from happening.

Even better on Node, you have EventEmitter. A much simpler interface.

There are so many real things to be worrying about over whatever this image generator is doing. Obviously it was wrong what they did, and I don't think it was an "accident" in the sense that they definitely knowingly inserted some diversity-related prompt into the model's built-in instructions. However, who did this actually harm? Why should I or the general public care? Why would you want to use the crappy low-quality Gemini image generator over a superior tool to Midjourney (or even DallE) in the first place? There is no doubt that Google is very left leaning, an artifact of where they are incorporated, but going as so far as using this as some gatcha to Google and how it shows the rot in the company, etc.... that's giving way too much credit to whoever wrote a few sentences of a system prompt.