mattalex

5 ms·

1.
by mattalex 16d ago
The reason the German gdpr watchdog is not complaining about Heise is because what they are doing is completely legal: you can still disable the tracking with the "more options" below. Current legal doctrine is that _forcing_ paym
2.
by mattalex 21d ago
You assume that openai's inference is profitable and that they aren't just trying to bolster revenue before their IPO. The only indication that openai is profitable comes from openai (whom I wouldn't trust with any statement,
3.
by mattalex 2mo ago
You can already do that: Keyword you want to be looking for is NTN (non terrestrial networks). The complexity is how you actually deploy this at scale and what you want the actual satellites to do: For example, there is the option of having
4.
by mattalex 2mo ago
SAP has largely switched over to their own HANA database for new deployments. Even before that their "AnyDB" database worked with pretty much any backend and is still supported. The reason you see oracle has less to do with what S
5.
by mattalex 2mo ago
I wouldn't even say that VW is a failing company: They sold the largest number of EVs in the EU in 2025 ( https://autovista24.autovistagroup.com/news/which-brand-domi... ) and that is before you consider that e.g. A
6.
by mattalex 2mo ago
Why do you assume that the entire EU has brain damage and did not think of this _most obvious_ loophole? Sure the companies could commit crimes but if that is enough of a reason not to pass laws then we better start striking murder of the b
7.
by mattalex 2mo ago
150M at 5B revenue is not great: that's 3% margin! The bigger issue is that console manufacturer revenue is highly cyclical. This is hard to see in e.g. Xbox and Sony since they are both part of a larger conglomerate, but really obviou
8.
by mattalex 2mo ago
You can get rid of 99.9% of those attacks by simply dispatching the data consumption to a different instance of the LLM, see, for instance, some of the later patterns in https://arxiv.org/abs/2506.08837
9.
by mattalex 3mo ago
The issue with having a "no answer" option is that you implicitly add a decision problem into your test that depends on the "cost" of answering wrong. Specifically, your model now has two "correct" classes p(cl
10.
by mattalex 3mo ago
Because the size of the attention matrix depends on the number of tokens (this is what makes attention N^2). If you don't care about having a flexible number of input tokens (e.g. in image processing) you can learn a fixed routing matr
11.
by mattalex 4mo ago
Nothing is stopping them, it's just not worth it: Have a look at e.g. vast.ai's pricing ( https://vast.ai/pricing ). The V100 (2017 -> 9 years old) can be rented from $0.02 to $0.37/h (right now I can find a
12.
by mattalex 4mo ago
Regarding specifically depth anything: You're not running this on a microcontroller. In general, CNNs still reign supreme on microcontrollers since you have a way lower peak memory demand which is what usually kills you. Here in this c
13.
by mattalex 4mo ago
Then allow resale through the platform at the original purchase price. You still have no scalping, but you recover the ability to back out due to unforeseen events
14.
by mattalex 5mo ago
It works the same way: softmax is essentially just applying the normalization to the vector exp(x). From an "engineering" POV this effectively ensures that the vector you normalize has strictly positive entries, so the result ends
15.
by mattalex 5mo ago
I'm what way is that different? You return early and the call Cascades up the call chain until you handle it (otherwise it's always an "either" results) In practice you use something like an exception monad, which makes
16.
by mattalex 5mo ago
Of course you can: you just have to define it in your type. The output set becomes a union type of the normal output and whatever you want as an exception. If you write this as a monad, your get very similar syntax to procedural code.
17.
by mattalex 5mo ago
Effectively none. The US has a huge trade deficit with Germany/Europe so there is practically never a case where the US receives gold from Germany: It's always more then offset by the deficit. The equivalent for the US would be th
18.
by mattalex 5mo ago
Germany already repatriated about half of its gold reserves between 2013 and 2017 from paris and new york to frankfurt. There has been a recent (as in "18th of march" recent) petition to the Bundestag to repatriate the gold. The r
19.
by mattalex 5mo ago
Whether the US is capable of hiding their maleficence or not should not be an indicator of whether it is safe to deal with them. If your indicator for the US being a good partner in _anything_ is that "well we did corrupt things in the
20.
by mattalex 6mo ago
There were plenty of models the size of gpt3 in industry. The core insight necessary for chatgpt was not scaling (that was already widely accepted): the insight was that instead of finetuning for each individual task, you can finetune once
21.
by mattalex 7mo ago
Assuming this is real: Why do you think anthropic was put on what is essentially an "enemy of the state" list and openai didn't? The two things anthropic refused to do is mass surveillance and autonomous weapons, so why do _y
22.
by mattalex 1y ago
It might be that they pay less for anthropic depending how many tokens are generated by each model: total cost is token cost times number of tokens. I haven't checked gpt5, but it is not impossible that price wise they might be very co
23.
by mattalex 2y ago
This is essentially the principle behind algebraic effects (which, in practice, do get implemented as delimited continuations): When you have an impure effect (e.g. check a database, generate a random number, write to a file, nondeterminist
24.
by mattalex 2y ago
Once you have strong normalization you can just check local confluence and use Newman's lemma to get strong confluence. That should be pretty easy: just build all n^2 pairs and run them to termination (which you have proven before). If
25.
by mattalex 2y ago
That entirely depends on what AMD device you look at: gaming GPUs are not well supported, but their instinct line of accelerators works just as well as cuda. keep in mind that, in contrast to Nvidia, AMD uses different architectures for com
26.
by mattalex 2y ago
To expand on that: there's also the issue that these games have to be (somewhat) competitive multiplayer games: multiplayer because otherwise there's no way to create enough content, and competitive since otherwise there's le
27.
by mattalex 2y ago
There are alternatives to iron that have higher efficiency and lower prices. For instance https://hydrogenious.net/ does exactly that but with benzene like structures. The advantage of this is that you can reuse existing in
28.
by mattalex 2y ago
The paper I have mentioned can be found here https://arxiv.org/pdf/2206.09787 There are so many things that have only been invented in the last couple of years like RINS, MCF cuts, conflict analysis, symmetry detection
29.
by mattalex 2y ago
2008 is ancient for optimization! People have tested old year 2000 lp and milp solvers against recent ones while correcting for hardware. Hardware improvements made up ~20x improvement, while lp solvers in general sped up 180x. MILP solvers
30.
by mattalex 2y ago
You can solve L1 regression using linear programming at fantastically large scales. In fact in many applications you do the opposite: go from squared to absolute because the latter fits into in lp

More ›