Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
numpad0
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
451.
▲
by
numpad0
5mo ago
...why would webcam head tracking need a new standalone vibecoded project? I thought there are plenty of those already.
452.
▲
by
numpad0
5mo ago
The bottom half of Surface Laptop(with the bending snake hinge) is a not-the-Thunderbolt dock. It has an optional GPU and fan in addition to what's obviously in it. So you're attaching battery at different SOC, and PCIe device, an
453.
▲
by
numpad0
5mo ago
Because everyone wants 100% PostScript-compatible printers and nothing else. The PostScript was created by ex-PARC people as they were founding a small startup called Adobe Systems, and it was chosen by Apple for its revolutionary 1985 Lase
454.
▲
by
numpad0
5mo ago
Apple is dogmatically anti games, they only spare bare minimum of effort to support videogames as an necessary evil. It's some Jobs' teaching thing.
455.
▲
by
numpad0
5mo ago
> We elected not to deploy these weapons for whatever reason. Russia/China insisted that Americans perfecting the nuke interception game is an existential threat to them, and threatened to do something horrible like retaliating mili
456.
▲
by
numpad0
5mo ago
If I were to hallucinate what it is and why it's worded that way: AI robot space is in need of a hyper-realistic game engine with better physics than Unity/Unreal style non-deformable rigid body mechanics, that's also way fas
457.
▲
by
numpad0
5mo ago
That's the wrong way around. Matte is better for performance but gloss attract consumers so Apple puts "Pro" and glossy front in the same product. Actual pro displays are always matte, a bit like how actual pro cameras always
458.
▲
by
numpad0
5mo ago
It's seeming to be going to be an another DGX Sparks that aren't so faster than maxed out Mac Studio, nor cheaper than 4x Blackwell on a workstation, nor cloud tokens. That's why.
459.
▲
by
numpad0
5mo ago
MediaTek used to supply a lot of lower-performance MIPS and ARM chips during 2010s. GP probably has that MediaTek == bad experience association. The actual chips weren't unreliable or horribly engineered, just slow, but people don'
460.
▲
by
numpad0
5mo ago
MS announced Surface Laptop Ultra with this SoC. 15", 128GB RAM, no pricing yet. 1: https://www.theverge.com/tech/940584/microsoft-surface-lapto...
461.
▲
by
numpad0
5mo ago
The actual canned response to this is [1] and humidity. Daily levels of humidity in East/Southeast Asia is whole another level from US/EU(or vice versa). 1: https://en.wikipedia.org/wiki/Glico_Morinaga_case
462.
▲
by
numpad0
5mo ago
This is really about Ubisoft's The Crew, a one-time-paid mostly-singleplayer car race game about infights and revenges in an illegal street racing group, that required Internet connection, which server got shut down. So yeah. The requi
463.
▲
by
numpad0
5mo ago
(update: looks like there was someone/some groups ragebaiting Korean and Japanese Twitter users with Korean transition into the Hangul phonetic script for Twitter impression incentives money. Those tweets had not reached me at the time
464.
▲
by
numpad0
5mo ago
But is an algorithmic AI detector really a thing? I get the idea: get 10k each samples of human data and AI data, train a simple classifier until it gets 99.9999% accuracy or <10k false negatives per day at your scale, ship it as a scree
465.
▲
by
numpad0
5mo ago
It could be. They seem to be getting enormous amount of politics or crypto related AI fake reels that real people fall for. They probably do need means to control spams.
466.
▲
by
numpad0
5mo ago
The problem is that it's not SOME false positives, AI detectors so far have been all so comically bad that they might be classified as pseudoscience. Or an artificial false positive generators even.
467.
▲
by
numpad0
5mo ago
This definitely needs substantiation. I've NEVER seen such usable tools EVER. AI flagging in general has always been very sketchy IME.
468.
▲
by
numpad0
5mo ago
Likely not the "base model" of AutoCAD. Those tools are used in ways that they're integral to processes. They have their equivalents of ticket systems that are linked to code repositories with LFSs and bunch of IDE type tools
469.
▲
by
numpad0
5mo ago
One guy retire and a college graduate goes in.
470.
▲
by
numpad0
5mo ago
> IVD is the best compromise we can have in this situation. Maybe, but no one is running an ivdfy-filter through every single Japanese documents and the issue keeps going. Maybe one way to make it happen is to make the Simplified forms s
471.
▲
by
numpad0
5mo ago
> “If a creator doesn’t specify whether or not they used AI, but our systems detect significant photorealistic AI use, we will now automatically apply a label,” YouTube said. > YouTube creators who believe their content was inc
472.
▲
by
numpad0
5mo ago
Chinese AI labs are reducing Japanese images and text out of AI models - they leave much smaller amount for text models that has to be literate in Japanese, and explicitly nuke it out of dataset for image models so that it only supports Sim
473.
▲
by
numpad0
5mo ago
Google/Apple needed it to fill the moat for Japanese phone market - for Google it was because Japanese carriers were stripping emoji from outgoing emails, and for Apple it was because iPhone as a real phone and not an Internet-connecte
474.
▲
by
numpad0
5mo ago
RPA was a big thing just towards the tail end of 2010s, though. MS had free official tools for Windows 10 to do it. I think at some point a Japanese bank had a robotic arm set up to flip a contract binder, scan the page and stamp it page by
475.
▲
by
numpad0
5mo ago
Tokenizer efficiency varying by languages, by as much as up to 15x, is very well known and established https://www.google.com/search?q=tokenizer+efficiency+by+language
476.
▲
by
numpad0
5mo ago
Next generation of. The Type 25 HVGP just entered service few months ago. That one is just a two-stage Scud, but the Block 3 or NG or Type 35 or whatever of that could have this tech.
477.
▲
by
numpad0
5mo ago
Air breathing engines don't need the oxidizer tank, so like the 2/3 of a rocket just goes away before even touching Tsiolkovsky math. That improves payload mass fraction massively. Also, this doesn't scale down to Mach 3-4 an
478.
▲
by
numpad0
5mo ago
Yeah. What are people even talking about? The rear looks a bit too R34, the bottom part of front bumper looks a bit 992, and the car overall looks a bit too comical looking, but other than that, this is just completely fine. I can almost se
479.
▲
by
numpad0
5mo ago
They always sound like an obnoxious American tourist talking through a translator, the chatbot training dataset is the same and foundation models are always built with >50% American English data for some reason.
480.
▲
by
numpad0
5mo ago
... why would cm precision be often not possible?
More ›