shwaj

RFK is still on the ballot in more than half of the states. He should be added to your site.

https://www.cbsnews.com/news/rfk-jr-map-on-the-ballot-states...

Editorializing is defined (by one dictionary) as: “making comments or expressing opinions rather than just reporting the facts”.

By this definition, changing the title to “Austrialian Federal government…” doesn’t run afoul of the policy because there is not opinion involved. It’s just clarifying which federal government is the subject.

What’s the alternative? You can’t just say “don’t say that”. There needs to be something you can say instead, 5 syllables at the most, which evokes the same feeling of confident wrongness, without falling into anthropomorphism. It’s a tall order.

This is the first I’ve heard of perovskite so I have no idea: are the cited toxicity issues fixable? Or are they inherent to the chemistry of this class of panels? How do we decide how much extra short term toxicity is acceptable? Do we do that blindly without theoretical/empirical results that demonstrate a better end state?

There are layers of abstraction on top of “the math”. The back propagation math for a transformer is no different than for a multi-layer perception, yet a transformer is vastly more capable than a MLP. More to the point, it took a series of non-trivial steps to arrive at the transformer architecture. In other words, understanding the lowest-level math is no guarantee that you understand the whole thing, otherwise the transformer architecture would have been obvious.

I’m not familiar with json-ld, other than a quick skim of a Google search that I just did. What is so revolutionary about it (and other technologies in the space, such as…?) that it represents a whole new revision of the web paradigm, comparable from the shift from static pages in web1 to interactive sites in web2?

Groq is claiming 284 tokens/second on Llama 3.1 70b, so they’re in the same ballpark.

https://groq.com/12-hours-later-groq-is-running-llama-3-inst...

He didn't do it a third time.

Which half of the court would be likely to rule which way?

A bit of a tangent, but I like Pomobabble. Even the footnotes have footnotes.

https://michiganlawreview.org/journal/pomobabble-postmodern-...

Why is that funny? Sometimes compression is lossy, like JPEG and H.265

I’m pretty sure Altman still cares about full AGI, if only because of the power inherent in achieving that goal.

It’s a core technology issue.

The AI isn’t the product, e.g. the ChatGPT interface is the main product that is layered above the core AI tech.

The issue is trustworthiness isn’t solvable by applying standard product management techniques on a predictable schedule. It requires scientific research.

That mirrors my experience from when I used to live in northern Canada

The Cue tests may be unreliable, but at least they’re expensive!

My company was providing Cue test cartridges for free. They always struck me as over engineered.

I like to think that there are subversives working in ad agencies and that such gaffes are intentional.

(I know it’s probably not true, it just makes the dystopia a bit more entertaining)

In this case the subtext would be something like: AI is coming and will provide all of the content for the rubes to consume on their shiny iPads, and damn the creatives who used to make a meaningful living creating it. But at least we can raise a middle finger via this ad, and mock the execs who okayed it!

I find merino wool tshirts can be worn for several days, whereas polyester tshirts marketed as “moisture wicking” will be stinky in a matter of hours.

If I wear one of the latter for a workout, if I don’t take a shower immediately, then I at least change into a cotton or wool shirt afterward.

Edit: typo

I just felt a jolt of anxiety reading this. I feel physically unwell when in a room lit by 6000K "daylight" bulbs, and even imagining being in such a room gives me a twinge.

I don't know how I respond to these "blue pump" LEDs; are they currently on the market? How would I know whether a given bulb uses this technology? Maybe I'm OK with them (fingers crossed).

The difference between LED bulbs cannot possibly be a major contributor to CO2 emmissions. If they're going to be draconian about this, look forward to "15 minute cities" where you're not allowed to drive outside of your zone, since that would save orders of magnitude more energy.

Perhaps this will be true, for the types of insights that also require tremendous intuition for people to see. One difference is that it will be possible to examine the full state of the AI which lead to the insight, and perhaps learn from it.

(edit: rewrote to flow a bit better)

Tether isn’t crypto though, in the sense of being decentralized, permissionless, etc

I was going to ask for a reference to support this (although please provide one if handy), but the search term “catastrophic forgetting” is a great entry into the literature. Thanks.

This seems intentionally obtuse. What you say is true, but it is very obvious that this is much more of a pain than if you had the source code. On the other hand, fine tuning is just as easy, regardless of whether you have the original training data.

The grandparent provided an argument why we might not want to use an algorithm, even if it provided a Pareto improvement.

I suggested that the same argument could be used to say that we should actively degrade performance of the algorithm, in the name of equity. This is absurd, and illustrates that the GP argument is maybe not as strong as it appears.

I don't think it's a given that we shouldn't keep using that codec. For example, maybe the improvement is due to an open source hacker working in their spare time to make the world a better place. Do we tell them their contribution isn't welcome until it meets the community's benchmark for equity?

Your same argument can also be used to degrade the performance for all other groups, so that group X isn't unfairly disadvantaged. Or, it can even be used to argue that the performance for other groups should be degraded to be even worse than group X, to compensate for other factors that disadvantage group X.

This is argumentum ad absurdum, but it goes to show that the issue isn't as black and white as you seem to think it is.

Can you precisely define what you mean by "generating" and "pretends", in such a way that this neural network does both these things, but a conventional modern audio codec doesn't?

"Pretends" is a problematic choice of words, because it anthropomorphizes the algorithm. It would be more accurate and less misleading to replace "pretends to be" with "approximates". But then it wouldn't serve your goal of (seeming to) establish a categorical difference between this approach and "regular algorithms", because that's what a regular algorithm does too.

I apologize, because the above might sound rude. It's not intended to be.

This might be true for people in general, but people who have been diagnosed with Celiac disease are definitely allergic to wheat.

The word sling also refers to a weapon used to thow projectiles. https://youtu.be/6irNkhLdApk

Yang Luchan was the creator of Yang style taijiquan.

Haha, you can sure pen a phrase. I’d be shocked if you’re a circa 2024 GPT.

Username checks out. Funny, I remember GR as being filthier, such parts don’t jump to mind regarding AtD. AtD is one of my all-time favorite books.