Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
phire
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
121.
▲
by
phire
4mo ago
The design of the exposed mechanism is explicitly about booting unsigned versions of MacOS. There is zero support for booting anything else, but no enforcement that it must be MacOS. However, apple's justification for exposing this mec
122.
▲
by
phire
4mo ago
And if Apple were going to change their mind and try to block linux, they would intentionally modify the bootloader to remove that functionality, not break the boot picker.
123.
▲
by
phire
4mo ago
Speedrunning communities generally hate it when having more money leads to an advantage. If you ban deliberately smudging/scratching the disc, then some runner with a lot of money will just buy a lot of copies of the disc, find the one
124.
▲
by
phire
4mo ago
Yes, they loved the compactness and convenience (well, I’m not sure anyone ever loved the rewinding/fastfowarding experience) But the quality/color was always a noticeable downgrade from broadcast quality video (and that was a not
125.
▲
by
phire
4mo ago
True, it would be better at some tasks. My thinking is that for most tasks, a byte-orientated LLM still needs something like the wide "single activation per word" formatting that the tokeniser mostly provides. And it will likely w
126.
▲
by
phire
4mo ago
I wouldn't use the word necessary. IMO, we are probably talking about a 6x slow down (for typical english). You would need to be absolutely stupid not to implement some kind of optimisation along these lines. Slower and maybe a little
127.
▲
by
phire
4mo ago
The tokeniser is not a dictionary. It doesn't provide definitions, or give the LLM any kind of mapping at all. At best, it's a wordlist. It gives the LLM some idea of what humans consider to be common words. But it doesn't te
128.
▲
by
phire
4mo ago
IMO, the way the error bars combine is very intuitive. You are really just rounding to 6 or 12 sig-figs after every operation. People just seem to get really hung up on the point that error bars exist in the first place, and combine. I su
129.
▲
by
phire
5mo ago
That was only an issue because they were fired pretty much straight up; They only went 500km down range. You can also reduce peek deceleration forces by using aerodynamic lift to stretch out the reentry over a longer period.
130.
▲
by
phire
5mo ago
The individual aircraft could be operationally viable on certain routes, but the whole program was not commercially viable.
131.
▲
by
phire
5mo ago
Is a Mach-5 passenger aircraft actually the goal of this project? Seems more likely that Japan is designing this engine for a hypersonic cruise missile program, and the passenger aircraft concept is somewhat of a cover. IMO, there is no poi
132.
▲
by
phire
5mo ago
No contest really: RISCV is a much better ISA, VexRISC is a hyper-optimised implementation of it (for FPGAs), and it's not hindered by trying to be microcode compatible. The roughly equivalent VexRISC configuration (full with MMU) is o
133.
▲
by
phire
5mo ago
I wouldn’t say it didn’t have any microcode. It actually had a small PLA for sequencing the multi-cycle instructions. [0] I don’t think anyone would actually label it as microcode (not when the entire point of RISC was to avoid microcod
134.
▲
by
phire
5mo ago
It's a bit harder to avoid windows than it is to avoid Bun. More importantly, it's not the same thing at all. All the code in windows (at least until recently) was written by humans, understood by humans and reviewed by humans. An
135.
▲
by
phire
5mo ago
It's a pretty strict ban, with an exception. That exception is experimental and somewhat limited; Only allows "well tested, high-quality" PRs on parts of the codebase that have a low probability of causing soundness issues, a
136.
▲
by
phire
5mo ago
Yeah, I'm a little suspicious about that claim. Bluetooth tethering is a thing, actually predates wifi tethering. Though it's not enabled unless you enable Personal Hotspot in your phone settings (and Android requires it to be ena
137.
▲
by
phire
5mo ago
This is barking up the wrong tree. Using IFUNC to patch sshd was kind of elegant, it achieved rootkit like behaviour with a pre-existing mechanism. And sure, it might be possible for a secure daemon like sshd to drop enough privileges that
138.
▲
by
phire
5mo ago
I think this question is easier to answer if you look at the inverse: "Could a model maliciously smuggle intentions through a roundtrip of compressed representation without them being human readable" And skimming through the paper
139.
▲
by
phire
5mo ago
The NLA also hallucinates, so it's still not revealing the models actual "thoughts" of the model; The paper also points out that since the NLA is a full LLM, it can make inferences that aren't actually in the activation
140.
▲
by
phire
5mo ago
Because they are both well-known and really well documented in a way that's easy for beginners.
141.
▲
by
phire
5mo ago
Yeah, it might be about isolating the APU as much as possible from potential sources of noise. Not that they ever put optical isolators on the data lines between the APU and CPU, but just keeping them out of phase probably helped a lot. Ano
142.
▲
by
phire
5mo ago
I don't understand why Nintendo decided to give the APU a dedicated clock crystal. Was it that important to have an exactly 32000 Hz sample rate? They could have used a divide-by-21 to get within 0.2%, which would have been more acc
143.
▲
by
phire
6mo ago
Also, you probably spend much more energy moving the bits around the chip and out to RAM than you do on the actual calculation.
144.
▲
by
phire
6mo ago
Thanks, I suspected there might be something from the minicomputer era. I've only really looked at a single AM2900 implementation (and it was far from optimal). Guess I need to dig deeper at some point. > The ONES operation forces
145.
▲
by
phire
6mo ago
I'm not actually aware of any CPUs that preform a XOR faster than a SUB. And more importantly, they have identical timings on the 8086, which is where this pattern comes from.
146.
▲
by
phire
6mo ago
I agree that if you put a high-speed line between Auckland and Wellington, and get the travel time down to 3-4 hours, people would use it. It would actually be faster than going to the airport for central Auckland, or anything further north
147.
▲
by
phire
6mo ago
The railway services NZ are trying to bring back are regional commuter services. Auckland to Hamilton (now in operation); Auckland to Tauranga; Wellington to Palmerston North (Capital Connection, has been in operation for 35 years, about to
148.
▲
by
phire
6mo ago
Population density is one thing. Another issue is timing. New Zealand was a really young country when railway technology came along, and didn't really have enough time or money to invest in a good railway network before other technolog
149.
▲
by
phire
6mo ago
I don't think it counts as a signature, because it can't be verified without revealing the same secret used to create it.
150.
▲
by
phire
6mo ago
I'm not sure it's a good idea. But it's trivial to make a secret checksum. Just take the key, concatenate it with a secret 256-bit key that only the servers know and hash it with sha256. External users might know the length o
More ›