gpm

6 ms·

61.
by gpm 2mo ago
It's reported in bytes - thus it's all digital. It's a "cloud storage vendor" - thus it's all digital. Nine PBS might well have non-digital data, but it's clearly not the data this article is about.
62.
by gpm 2mo ago
50TiB is like $4k consumer SSDs in a shoebox at this point (not actually the recommended solution, but...). And that's massively inflated thanks to the AI boom.
63.
by gpm 2mo ago
Well, yes and no. By non-deterministic I think people really mean "chaotic" in the chaos theory sense. Small perturbations in the input lead to wild and unpredictable changes in the output. Even with temperature parameters a fixed
64.
by gpm 2mo ago
Yeah, I think there's a lot of fans of "pair programming", including the zed people...
65.
by gpm 2mo ago
By a strict definition of memory safety it isn't - you can tear two-pointer-wide values using data races and cause arbitrary memory issues if you try to using only normal code. It's close enough for most purposes... but it isn&#x2
66.
by gpm 2mo ago
5% within a week is a very high set of odds for such rare events. Or to look at it from the other more relevant direction 70% of magnitude 7 or larger quakes have detected foreshocks... https://en.wikipedia.org/wiki/For
67.
by gpm 2mo ago
Really I suspect that's the only hard part. If we had accurate 3d models of the structure (well 4d, 3d models over time) it would probably be reasonably easy to estimate how much strain it can take, and how much strain there is... Mayb
68.
by gpm 2mo ago
In practice... people don't use hoods (or maybe they don't work, but almost certainly the former). See for example this study https://www.jaci-inpractice.org/article/S2213-2198(26)00537-... It was done on peo
69.
by gpm 2mo ago
Along the same lines - replacing gas stoves with electric ones is effective for treating asthma: Article: https://theconversation.com/replacing-gas-stoves-with-electr... Underling Study: https://www.jaci-inpracti
70.
by gpm 2mo ago
Yeah, so my (electric) stove has knobs, but my oven and microwave are both touch buttons and this has literally never been an issue. Admittedly I have other complaints about the microwave controls (why is it so damn many button presses to v
71.
by gpm 2mo ago
I'm thinking super long context length or something to that effect. I can imagine schemes for instance where context is compressed into chunks and then chunks that are ranked highly relevant for the token are decompressed. Which would
72.
by gpm 2mo ago
Trade agricultural futures based on it maybe?
73.
by gpm 2mo ago
This is like saying that all the countries in the world that speak English are England, that's not how it works. Countries are not defined by the language they speak, and the language they speak most definitely varies over time. Moreov
74.
by gpm 2mo ago
> the Holy Roman Empire inherited the name Eh, inheritance implies the predecessor willed it to you upon death. In the case of the "Holy Roman Empire" the predecessor died 334 years earlier with no successor. This isn't a
75.
by gpm 2mo ago
Collaborative (as in simultaneous multiplayer) editing I think, which has been Zed's mission since before LLMs were a thing... See https://zed.dev/blog/introducing-deltadb but also https://zed.dev/
76.
by gpm 2mo ago
More destructive weather events are good for expensive pieces of infrastructure? Higher temperatures are good for things that struggle to cool themselves? I don't see it.
77.
by gpm 2mo ago
The value here is pretty simple, the frequency at which windows are rendered useless due to weather. Yes, the change we are undergoing and that the current administration seems to be doing their best to accelerate includes an increase in he
78.
by gpm 2mo ago
> Would they take off if medevac flight - there was probably some urgency.
79.
by gpm 2mo ago
If I know stopping a particular butterfly from flapping its wings prevents thousands of people from dying and devastating effects on infrastructure (the typical result of a hurricane) I don't particularly care if they "cause"
80.
by gpm 2mo ago
Not useless if you can identify and stop the butterflies that cause hurricanes.
81.
by gpm 2mo ago
The weather already does that on a regular basis... though they are doing their best to make the weather, on average (i.e. what you might call a climate), worse.
82.
by gpm 2mo ago
Yeah, sure, you're right, and good point about non-towered. Are you suggesting however that "The flight crew acknowledged the clearance and responded that they would cancel IFR “in just a couple of minutes.”" meant the crew w
83.
by gpm 2mo ago
Yes... though I'd include the subsequent quote "and responded that they would cancel IFR “in just a couple of minutes.”" Also someone is correctly pointing out to me downthread that a visual approach is technically an IFR pro
84.
by gpm 2mo ago
No, you didn't, because they apparently had instrument clearance from the tower at the time they crashed into the mountain. The only correct conclusion here is more investigation is necessary than what has been publicly published.
85.
by gpm 2mo ago
Note that after the pilots said that they could go visual the preliminary report documents the following > The ARTCC controller subsequently cleared the flight to SRR, cleared the flight for a visual approach, and informed them they co
86.
by gpm 2mo ago
Well for one I would expect the final report to actually come to any sort of conclusion and not just transcribe the conversations between air traffic control and the plane, and the data recorded by the plane. The preliinary report doesn&#x2
87.
by gpm 2mo ago
So Irregular used the model to hack some real foo.com with the excuse that they meant to run in an environment where they were serving a fake foo.com!? This seems like a very mild reaction on OpenAIs part.
88.
by gpm 2mo ago
Because it's a tool in search of a use case (or many use cases) and mathematics is the most natural use case for it. Mathematics is by definition the art of putting words on a page in a rigorously defined "correct manner" (
89.
by gpm 2mo ago
Between a lean proof, and a peer reviewed paper, the former is a lot less likely to be mistaken... Nothing is perfect.
90.
by gpm 2mo ago
I mean, they're verified in the sense that the lean proof checks out... and presumably OpenAI read them.

More ›