> STARFlow-V is trained on 96 H100 GPUs using approximately 20 million videos.

They don’t say for how long.

Quest 3 is $499 and Quest 3S is $299 in the US

But what if we just use milliseconds as our bigserial? And maybe add some hw-random number at the end to avoid conflicts? Wait

Somehow +1 on this comment just doesn't feel like enough.

Oh yeah, it would be an identifier but it would be unique. Across the universe of all devices, effectively. Should come up with a name for that

“client” here may refer to a backend app server. So you can have 10-100s of backend servers inserting into a same table without having a single authority coordinating IDs.

That table is still a single authority, isn't it? But I guess fewer steps is still faster.

Except if you're using a sharding or clustering database system, where the record itself may be stored to separate servers as well as the key generation itself.

In those cases yes. There's still a case for sequential there depending on the use pattern, but write-heavy benefits from not waiting on one server for IDs.

Apple is exempt from US-China tariffs.

https://www.npr.org/2025/04/12/nx-s1-5363025/apple-iphone-ta...

wait what. i had no idea. These tarrifs are a shitshow.

You download directly from microsoft, here is a useful guide: https://massgrave.dev/windows_ltsc_links

Obama authorized 54 strikes in Pakistan in his first year, resulting in estimated 100 civilians dead. He received Nobel peace prize that year.

https://en.wikipedia.org/wiki/Drone_strikes_in_Pakistan

2I/Borisov was not discovered by Zwicky Transient Facility .

The comet was discovered on 30 August 2019 by amateur astronomer Gennadiy Borisov at his personal observatory MARGO in Nauchnij, Crimea, using a 0.65 meter telescope he designed and built himself.

That’s interesting. If it was possible to do on a homemade telescope, then why were no interstellar objects discovered before 1A?

The hard part isn't having a telescope, but analyzing the images for objects that have moved between successive observations. Digital astrophotography and analysis software have been getting steadily cheaper and better, which leads to more amateur comet hunters each watching more sky, which has rapidly improved the odds of catching rare objects.

I'm not sure how the progress of institutional and amateur observations compare. Obviously the big guys benefit from the same technological advancement, but I don't know whether the fraction of new objects discovered by amateurs has been growing or not. I suspect the odds of the first interstellar object being found by an amateur were still pretty long.

A tamper-proof electronic collar with some C4.

I like the mobile version and pricing page.