teravor

9 ms·

151.
by teravor 4mo ago
what I meant was dictating the data structures for the code (transformations) the LLM is going to write. "Show me your flowchart and conceal your tables, and I shall continue to be mystified. Show me your tables, and I won't
152.
by teravor 4mo ago
with AI, documentation driven development is an understatement, if you take the time not just to document but to also provide lots of examples and potentially even data structures for the implementation (including intermediary data structur
153.
by teravor 4mo ago
the rise of the TUIs is an indictment of existing GUIs, you can technically create a GUI that has identical functionality to a TUI and then start adding features that a TUI can't match.
154.
by teravor 4mo ago
forcing citizens to buy goods which are adversarial to them with their own money is rarely a good idea. paradoxically, citizens expect the government to be adversarial towards them in big things (see America) but it's the 'little&
155.
by teravor 4mo ago
i found the exact opposite, the pytorch version on the cpu barely does over 2 times realtime while i can get the onnx int8 version to reach 5x.
156.
by teravor 4mo ago
kokoro is decent but pocket-tts is much better especially when you rip a good voice. https://github.com/kyutai-labs/pocket-tts the onnx version of pocket-tts does perform better. https://huggingface.co/
157.
by teravor 4mo ago
certainly a concern, and a good reason to use multiple cryptosystems together. unfortunately there are probably similar papers for Kyber which are NSA property and will never see the light of day. they do employ a lot of mathematicians. for
158.
by teravor 4mo ago
the McEliece cryptosystem goes back to the 70s, doesn't mean it's as well studied as RSA. obviously people study popular cryptographic primitives more. having said that, I would trust McEliece more than Kyber.
159.
by teravor 4mo ago
it's worth noting that AMD's software is universally weak and not worth any degree of reliance. and it's not just ROCm, every few months they merge a serious regression into amdgpu and sometimes even backport it into stable.
160.
by teravor 4mo ago
> Unfortunately, we can't yet say that about lattice cryptography, despite it being approximately as well-studied as ECC. this is an absurd claim, lattices may be as well studied as elliptic curves, but not the cryptograph
161.
by teravor 4mo ago
tool use is another factor, every time the agent uses a tool the entire context is priced at cache rate on top. the same happens when it asks you for input.
162.
by teravor 4mo ago
a website can serve you a keylogger client and no one will ever know most likely, harder to do with binaries/sources
163.
by teravor 4mo ago
apparently those scrubbers are less efficient the lower the co2 is (also need to move much more air which ruins stealth). iirc submarines are routinely exposed to multiple thousands.
164.
by teravor 4mo ago
there is some ambiguity here. submariners are exposed to really high co2 levels and are doing fine. it's possible that regular high co2 is just a proxy for bad air, and that you can acclimate to purely high co2 pretty well. i have felt
165.
by teravor 4mo ago
in that case my mistake. i always assumed that the `NSA ploy` was strategic bullshit, the sort of thing you say to get support from NSA haters. it wouldn't even occur to me that someone would take time addressing it without being one o
166.
by teravor 4mo ago
> That is possible, but given the recent 2029 timelines from large Internet providers, I think it's prudent to prepare for Q-Day even if it never arrives. no one argues we shouldn't. you made the argument that we shou
167.
by teravor 4mo ago
> Hybrid PQ+ECDH is a hedged bet against an algorithm break before Q-Day, but is utterly fucking useless over Pure PQ once Q-Day occurs. there is also the likelihood that Q-Day never arrives, either because something we don&#x
168.
by teravor 4mo ago
> It's impossible for someone to doubt their own sentience. The literal act of doubting is enough to dissipate all doubt. i never found this convincing. just because you can loop does not mean you are sentient/conscio
169.
by teravor 4mo ago
with mmap you also don't have to worry about committing too much system memory, if another application needs it it will start evicting your cache.
170.
by teravor 4mo ago
> Blocking access to attestation or DRM will prevent using the functionality of the app depending on it or the whole app unless it's implemented incorrectly. which is what the user would want or be fine with in this case..
171.
by teravor 4mo ago
giving the option to completely block attestation and DRM API would be a good start. > hardware attestation, which is not fingerprinting this is false, the attestation middleman Google server can fingerprint your unique device s
172.
by teravor 4mo ago
> We're legally allowed to provide compatibility with Google Play via our sandboxed Google Play compatibility layer. and they are legally allowed to fingerprint grapheneos and block Play functionality. maybe once that happ
173.
by teravor 4mo ago
> where does it get a key to decrypt memory on resume? you enter it...
174.
by teravor 4mo ago
on the subject of encryption keys and memory there is something you can do: - if your CPU supports it, enable memory encryption. - if your TPM module supports this look for MemoryOverwriteRequestControl & MemoryOverwriteRequestControlLo
175.
by teravor 4mo ago
i believe it's more complicated than that. i know that nvidia offers TEE for their overpriced offerings. i would assume they make use of that so the weights are encrypted. this doesn't mean it cannot leak but it would be a major u
176.
by teravor 4mo ago
it's an electron app, it highlights wrong spelling but doesn't suggest corrections. how does someone exhibit so much incompetence?
177.
by teravor 4mo ago
the Chinese they are trying to catch must be amateurs, first thing you should do is construct a sandbox which looks indistinguishable from a common user. second thing is to put it behind a residential proxy.
178.
by teravor 4mo ago
mullvad's mistake was to respond. it's all universally performative, the only reason it elicits reactions (1000+ comments here) when the view of a megacorp don't is because people anticipate the ability to elicit a response f
179.
by teravor 4mo ago
there is a small problem though, everything in those power-plants is monitored. so there is no radiation increase anywhere to be found. if it's the pollutants as the Nature paper claims without evidence, then any other industrial plant
180.
by teravor 4mo ago
in a world where there is no safe low radiation dose, it would be quite easy to generate the data to demonstrate this. so either low doses cause no harm or cause such minimal harm as to be safely disregarded. luckily the government is movin

More ›