yorwba

28 ms·

661.
by yorwba 11mo ago
My thesis is that marketing doesn't brainwash people. You can use marketing to increase awareness of your product, which in turn increases sales when people would e.g. otherwise have bought from a competitor, but you can't magical
662.
by yorwba 11mo ago
Alternatively, since brainwashing is a fiction trope that doesn't work in the real world, they can brainwash the same (0) number of people for less money. Or, more realistically, companies selling social media influence operations as a
663.
by yorwba 11mo ago
Noto Sans comes in SC, TC, HK, CJK JP and CJK KR variants. There's no Noto Serif for CJK characters, nor does Deja Vu support them.
664.
by yorwba 11mo ago
The CCP controlling the government doesn't mean they micromanage everything. Some Chinese AI companies release the weights of even their best models (DeepSeek, Moonshot AI), others release weights for small models, but not the largest
665.
by yorwba 11mo ago
How do you know how well OpenAI's unreleased experimental model does on biology or history questions?
666.
by yorwba 11mo ago
DeepSeekMath-V2 is also an LLM doing math and not a specific formal math system. What interpretation of "general purpose" were you using where one of them is "general purpose" and the other isn't?
667.
by yorwba 11mo ago
That's a different model: https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Speciale
668.
by yorwba 11mo ago
Both OpenAI and Google used models made specifically for the task, not their general-purpose products. OpenAI: https://xcancel.com/alexwei_/status/1946477756738629827#m "we are releasing GPT-5 soon, and we’re
669.
by yorwba 11mo ago
Previous discussion: https://news.ycombinator.com/item?id=46072786 218 points 3 days ago, 48 comments
670.
by yorwba 11mo ago
Yes, "new math" is neither magical nor unrelated to existing math, but that doesn't mean any new theorem or proof is automatically "new math." I think the term is usually reserved for the definition of a new kind of
671.
by yorwba 11mo ago
I think they must've messed up validation somehow. The performance drops relative to the base model are sometimes quite dramatic, which should've been caught by corresponding deterioration in validation performance. They write &qu
672.
by yorwba 11mo ago
"Join the Waitlist"
673.
by yorwba 11mo ago
Fortunately it's not true. GrapheneOS seem https://xcancel.com/GrapheneOS/status/1993061892324311480#m to be reacting to news coverage https://archive.ph/UrlvK saying that although legitimate
674.
by yorwba 11mo ago
I think La Quadrature du Net don't consider it relevant whether the accused are guilty or not; they don't want encrypted communication to ever be included in the evidence against someone. I agree that that's a bit too much bi
675.
by yorwba 11mo ago
English version https://www.laquadrature.net/en/2023/06/05/criminalization-o...
676.
by yorwba 11mo ago
European companies have been plenty successful at monetizing their innovations. If you look at OECD statistics https://goingdigital.oecd.org/datakitchen/#/explorer/1/toolk... in the industry of "Com
677.
by yorwba 11mo ago
If the text is written interactively on the canvas (as opposed to extraction from pixels) this task is known as "online handwriting recognition" ("online" because you can watch the text being formed incrementally, which
678.
by yorwba 11mo ago
> I worked on it a ton a couple years ago. Now I'm too busy. The current codebase is stuck in development hell on some branch of trying to make the big preprocessing sweep be more efficient (with multiprocessing). That's plausi
679.
by yorwba 11mo ago
Yes, mixture of experts is basically structured activation sparsity. You could imagine concatenating the expert matrices into a huge block matrix and multiplying by an input vector where only the coefficients corresponding to activated expe
680.
by yorwba 1y ago
"The accessible data items used in the study are the same that are public for anyone who knows a user's phone number and consist of: phone number, public keys, timestamps, and, if set to public, about text and profile picture.&quo
681.
by yorwba 1y ago
In theory, multiplying a matrix and a highly sparse vector should be much faster than the dense equivalent, because you only need to read the columns of the matrix that correspond to nonzero elements of the vector. But in this paper, the ve
682.
by yorwba 1y ago
The approach in the paper specifically addresses the case where an LLM can usually solve a task when it requires few steps, but fails for the same kind of task with more steps because it randomly gets a step in the middle wrong and then der
683.
by yorwba 1y ago
About 5,300 paid subscribers in January: https://www.astralcodexten.com/p/subscrive-drive-25-free-unl...
684.
by yorwba 1y ago
The comparison is against general models which are explicitly fine-tuned. Specifically, they pre-train their models on unlabeled in-domain images and take DINO models pre-trained on internet-scale general images, then fine-tune both of th
685.
by yorwba 1y ago
https://www.nand2tetris.org/
686.
by yorwba 1y ago
I recommend writing a blog post explaining the constraints in the FPGA example in detail and pointing out the constraint conflict your method identified. That would make for a better sales pitch than merely asserting the existence of this e
687.
by yorwba 1y ago
Deterministic output is a property of the inference framework, not the model. E.g. https://docs.sglang.ai/advanced_features/deterministic_infer... https://github.com/ggml-org/llama.cpp/pull&#x
688.
by yorwba 1y ago
It cannot. But it can detect whether the terminal emulator is shifting the cursor by the correct number of cells when displaying a character. Glyph support is typically left to the system font stack, but positioning and hence cursor movemen
689.
by yorwba 1y ago
This is to protect against shoulder-surfing. Having the information present on the client and accessible to the user while appearing redacted is very much the point.
690.
by yorwba 1y ago
A jpeg of a copyrighted image can be copyright infringement, but isn't necessarily. A trained model can be copyright infringement, but isn't necessarily. A human reciting poetry can be copyright infringement, but isn't

More ›