Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
rfoo
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
65 ms
·
571.
▲
by
rfoo
3y ago
Note that the team claiming a zero resistance replicate at 110K is the same team who said failed to replicate Meissner a few days ago. In the latest video they still don't think they have observed Meissner.
572.
▲
by
rfoo
3y ago
For reference, Bilibili started as a direct clone of Acfun (which itself is a clone of Niconico).
573.
▲
by
rfoo
3y ago
Bard is a 4.5B or so model.
574.
▲
by
rfoo
3y ago
> Easy: GL.iNet has tiny travel routers with OpenWRT supported out of the box. It supports modified OpenWRT with proprietary drivers, which are closed source. Still better than completely black-box travel routers but /shrug. From h
575.
▲
by
rfoo
3y ago
> but those are specs as well Well, sure, for weight. The other three I just can't tell until I try the thing for a week, it isn't spec if you can't tell anything after reading the spec. > I'm sure it can SSH. &quo
576.
▲
by
rfoo
3y ago
Beelink also offers barebone version if you live close enough to them. I'd prefer minisforum though, according to STH and other user reviews it's way quieter and cooler, and doesn't have the troublesome magnetically attached
577.
▲
by
rfoo
3y ago
> I'm doing CAD work/GPU work. Then you have a good reason to go for a gaming laptop, because your GPU isn't one-ssh-away. Gaming laptop also don't mean "high qualify", they just mean "high spec".
578.
▲
by
rfoo
3y ago
Yeah, I agree promoting this usage is as bad as promoting `curl | sh` in README.md. Similar to how you can inspect the content of a `curl | sh` script and then run it, the model is also in a clonable repo, you may just: git clone https
579.
▲
by
rfoo
3y ago
I like this pip metaphor. If we had required `--trust-remote-code` for every `npm install` we could have avoided left-pad and most of the software supply chain drama in the past years.
580.
▲
by
rfoo
3y ago
That's because the model architecture hasn't been added to huggingface/transformers yet, because it literally was just published today. >>> from transformers import AutoTokenizer, AutoModel >>> mod
581.
▲
by
rfoo
3y ago
Your comment is equally hilarious from the point of view of a native who lives in China now. idk if i'm smarter than the GFW but every time I rolled my own censorship-circumvention tool it worked well, even the most lazy way worked. I&
582.
▲
by
rfoo
3y ago
Yes but they are unfortunately targeted more than other censorship circumvention tools. Since everyone knows Tor/Obfs4/Snowflake it's easier to get your research published if you work on detecting that.
583.
▲
by
rfoo
3y ago
How is Microsoft's business in China crippled?
584.
▲
by
rfoo
3y ago
So rare earths are the DRAM of manufacturing industry?
585.
▲
by
rfoo
3y ago
Spectre as a general vulnerability class is NOT fixed. And you can write perfectly vulnerable-to-Spectre code in Rust, or anything you like. As long as there are "virtual" privilege separations under the same hardware privilege le
586.
▲
by
rfoo
3y ago
Spectre marks the beginning of a new bug class. It's more like "to write secure program on processors with speculative execution you must not do A, B and C". New processors fixed known bugs in the implementation of the silico
587.
▲
by
rfoo
3y ago
> I imagine that anyone who knows where to look can easily find dozens of interested buyers willing to pay a lot more than that, from intelligence agencies and their contractors to crime syndicates. This is unfortunately false. The runni
588.
▲
by
rfoo
3y ago
> integration into the existing Linux security stack It's less about integration into the security stack. It's just that the implementation of io_uring is hard to get right, and at 2023 there seems to be an endless stream of ra
589.
▲
by
rfoo
3y ago
> it's just that other manufacturers do not do that. Yeah. So it's just that people don't complain about other manufacturers.
590.
▲
by
rfoo
3y ago
No. I like TPMs. I hate DRMs. TPMs help me have a secure (as in preventing evil maid AND being forensics proof) computer to use. The current TPM2.0 specification had some problems (external TPMs are easy to MITM), so I can't trust it t
591.
▲
by
rfoo
3y ago
A good community-driven hardened Android distro needs someone with a strong personality. With strcat leaving I'm afraid that I can't trust GrapheneOS anymore and have to switch back to stock Android shipped by Google :( Edit: To b
592.
▲
by
rfoo
3y ago
TPM spec is fully open. Stop FUD.
593.
▲
by
rfoo
3y ago
It's not per parameter, you also need to hold activations for back prop to work.
594.
▲
by
rfoo
3y ago
It is more expensive (in engineering cost) to port all the world's research (and your own one) to the TPU of your choice, than just paying NVIDIA.
595.
▲
by
rfoo
3y ago
What about lists? Something like: --allowed-downtime-0-start=20230523170000Z --allowed-downtime-0-end=20230523170500Z --allowed-downtime-1-start=... Or --allowed-downtime-start='20230523170000Z,...' I can live wi
596.
▲
by
rfoo
3y ago
The thing is, this is compared to 36 months reserved instance, which also assumes 100% usage. For true on-demand you pay much more.
597.
▲
by
rfoo
3y ago
--flagfile gets me really far but to be fair it sucks when you need nested kv
598.
▲
by
rfoo
3y ago
Post it here and I'm sure some random guy are going to make you pay $100 next month.
599.
▲
by
rfoo
3y ago
Apple already ruined their laptops with the heavy-ish metal chassis. Even the new MacBook Air is ~1.24kg. Windows laptop companies, please DO NOT follow suit! Though some of them already did so, sigh. Please Apple, please just replace your
600.
▲
by
rfoo
3y ago
The unlocked bootloader warning screen isn't very different from what you described [1]. It says: > The bootloader is unlocked and software integrity cannot be guaranteed. Any data stored on the device may be available to attackers.
More ›