Hacker Newsnew | past | comments | ask | show | jobs | submit | sorenjan's commentslogin

Related:

FCC seek comments on NextNav petition for rulemaking on lower 900MHz ISM band - https://news.ycombinator.com/item?id=41226802

NextNav's Callous Land-Grab to Privatize 900 MHz - https://news.ycombinator.com/item?id=41535994


That's how EU's digital wallet is supposed to work:

> The selective disclosure of attributes will allow you to only share the specific information requested by a service provider, without revealing extra information.

> For example, with the selective disclosure of attributes you could choose to share your date of birth, but without revealing any other identifying details that could be used for profiling.

https://ec.europa.eu/digital-building-blocks/sites/spaces/EU...


You want to know what the global impression of the US is right now? Here's a translated quote from a newspaper today, from a source in our military:

> – The US has the most qualified intelligence organizations in the world at its disposal. Both the CIA and the FBI have been politicized under the current regime. I find it difficult to see how we will be able to maintain the trusting cooperation we have had with the US in the past after this.

The actions of the current administration speaks far louder than any font ever could, and it's tearing down decades of good will and trust.


> Both the CIA and the FBI have been politicized under the current regime.

The CIA and FBI were politicised well before the current regime. If you live in the US you will be aware of the Russiagate hoax.


> AN OFFICIAL WEBSITE OF THE UNITED STATES GOVERNMENT

> What's the biggest brand in the world? If you said Trump, you're not wrong.

This is beyond satire by now, it reminds me of Idi Amin and his official title:

His full self-bestowed title ultimately became: "His Excellency, President for Life, Field Marshal Al Hadji Doctor Idi Amin Dada, VC, DSO, MC, CBE, Lord of All the Beasts of the Earth and Fishes of the Seas and Conqueror of the British Empire in Africa in General and Uganda in Particular"


5G does not mean shorter waves/higher frequencies, that's just a common deployment. In Sweden we have 5G on the 700 MHz band, 5900 MHz, and several others in between.

Instagib with ASMD shock rifles.

Man after my own heart <3

    ++foo; // Increment value of foo by one

    foo += 1 # Increment value of foo by one

Something more complicated maybe ? Like say parse arguments from a command line ?

Is it the Cuda cores that run the vertex/fragment/etc shaders in normal GPUs? Where does the ray tracing units fit in? How much of a modern Nvidia GPU is general purpose vs specialized to graphics pipelines?


A datacenter GPU has next to nothing left related to graphics. You can't use them to render graphics. It's a pure computational kernel machine.


Qualcomm has confirmed it's coming to Snapdragon phones soon[0], which maybe hints that it's dependent on the SoC drivers? Samsung uses a mix of Snapdragon and their own Exynos, but I can't see them not releasing it to their Snapdragon phones when others do, and then they pretty much have to release it to their Exynos phones too.

[0] https://www.notebookcheck.net/Qualcomm-confirms-Quick-Share-...


An implementation of AWDL on Linux requires a Wi-Fi card that supports "active monitor mode with frame injection". [1] I looked into using it with an Intel Wi-Fi card I had and it appeared mine wasn't supported. I'm guessing the situation is similar on Android in terms of SoC support.

[1]: https://github.com/seemoo-lab/owl?tab=readme-ov-file#require...


SAM3 is probably a great model to distill from when training smaller segmentation models, but isn't their DINOv2 a better example of a large foundation model to distill from for various computer vision tasks? I've seen it used for as starting point for models doing segmentation and depth estimation. Maybe there's a v3 coming soon?

https://dinov2.metademolab.com/


DINOv3 was released earlier this year: https://ai.meta.com/dinov3/

I'm not sure if the work they did with DINOv3 went into SAM3. I don't see any mention of it in the paper, though I just skimmed it.


We used DINOv2 as the backbone of our RF-DETR model, which is SOTA on realtime object detection and segmentation: https://github.com/roboflow/rf-detr

It makes a great target to distill SAM3 to.


> It makes a great target to distill SAM3 to.

Could you expand on that? Do you mean you're starting with the pretrained DINO model and then using SAM3 to generate training data to make DINO into a segmentation model? Do you freeze the DINO weights and add a small adapter at the end to turn its output into segmentations?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: