layer8

The article makes the point that not all polynomial algorithms are created equal. If you combine ten O(n^2) algorithms, you can end up with an O(n^1024) algorithm, which suddenly isn’t practical anymore, despite being polynomial. That’s inconvenient.

It’s not the same size at all, much wider and heavier (both wider and heavier than the Pro Max). One-handed use is a very different experience on the Duo than on the mini, and it’s quite thick and heavy in one’s pocket.

It’s only fitting that the peak Windows was introduced by Patrick Stewart.

Memory fabs and logic chip fabs are quite different.

> it required specialized/miniaturized hardware

Looking at the earlier smaller iPhones and the budget SE line, at least the “miniaturized” part isn’t very convincing.

The logical screen size was the same as the iPhone 11 Pro (375x812), the iPhone mini didn’t add a new display size.

Hence the smiley.

Same for the iPhone mini. ;)

That Astra is just using the same methods as humans do (like abliteration) for modifying the behavior of a model, and that there is no particular connection to Astra itself being a model.

Schema support is the way to go, because applications will have a formal or informal schema no matter what, but then it’s better to drop most of the schemaless syntax rules in the first place. If someone wants `10.20` to mean the exact decimal value, or wants `30m` to mean “30 meters” instead of a duration, then that should be possible. Similarly, self-delimiting syntax like URLs shouldn’t require quotes.

He wants to give those posts exposure by posting them on his blog, so a second blog would counteract that. What he should do is put “Guest post: ” in front of each title.

> what you see is what you get: […] '10.20' and 10.20 are a string and a float, exactly as written.

This seems both underspecified and incorrect. Assuming a single-precision binary IEEE 754 floating-point number, 10.20 is actually 10.19999980926513671875. With double precision, it would be 10.199999999999999289457264239899814128875732421875. So what you see is not what you get, and it’s also unclear what exactly you’ll get.

> A value's type comes from its syntax, never from its content […] allowed-origins: [ 'https://example.com' 'https://example.org' ]

This looks like URLs, but are strings. What if you had:

> allowed-origins: [ 'httpș://example.com' 'ħttps://example.org' ]

Still strings and therefore fine for the SOML parser, but probably not a valid configuration in terms of the application-level types of the values. The syntax telling you the types is of limited utility as soon as the application schema has more specific ideas about the types than the configuration language.

[I wanted to put “ørg” instead of “org”, but the HN comment parser helpfully converts the former to a punycode URL.]

Like many other posts on Tao’s blog, this is a post by guest authors, not by Tao himself.

I’m more shocked by how many commenters in this thread seem to be misunderstanding what is going on here.

The identified factor is exercising one’s own bones, not eating them.

Read the article?

The one unit you buy is an insignificant drop in the bucket for them, and getting their actually good underlying hardware and display firmware is much more valuable for you in comparison.

The characters you see on screen are a mix of the writing and the acting. In long-running shows like TNG, the writers pick up on how an actor plays their character, and it informs their writing. TNG is generally considered to have come into its full form only after Roddenberry’s involvement leveled off. It might have become a very different show with an actor other than Patrick Steward as Jean-Luc Picard.

The bounty payouts must be public in order to serve as an incentive.

Enough that if you have to ask, you can’t afford it. ;)

No “here’s why it matters” and friends, thankfully.

This looks the same as the iPad Pro tandem OLED pixel layout. That I happen to write this comment on.

You could buy one from the price difference of the touch MacBook.

I think there’s a wide space between refusing any AI use, and refusing AI use requirements imposed by an employer. For example, one might find AI valuable for search, research, and getting a second opinion, but refuse it for primary generation of content.

I’d be surprised to see current LLMs write as well (and as non-annoying) as the writing in this book.

I spot-checked a few random chapters, and it looks clearly human-written to me. But I share your general worry.

The new or the classic context menu? (https://learn.microsoft.com/en-us/answers/questions/2287432/...)

Alexa had already been using other models like Claude depending on the question, and the French version for example uses Mistral.

Just as friendly feedback, monospace fonts are difficult to read for me for that kind of content. The site also doesn’t explain what it is about. It says “terminal-grade communities” (whatever that means), but Fluxer doesn’t seem to be bound to terminals?