It's just awkward, I've held the knife with my dominant hand all my life.

The idea would be to use autonomous drones, so they wouldn't need to communicate, the problem would be that the GPS signal is jammed.

If they're not communicating, how are they sending back lots of data?

I presume these are surveillance drones and are programmed to loop back to origin

Surveillance gathered by an completely autonomous drone with no outside data, stationed far enough away to require refueling, close enough to enemy operations to be useful, that then needs to make its way back to origin, intact, through hostile territory, quickly enough for the gathered information to be useful, seems like a preeeetty big lift. Something a startup would promise to tackle with a star team of technologists over the course of like 10 years? Sure. Something they’d have designed within the past, like, year while getting shot at? I’d have to see that believe it.

I don't know, but you can always use <() process substitution to create a temp file.

This new SAM model actually competes against SOTA models.

https://www.reddit.com/r/LocalLLaMA/comments/1pp9w31/ama_wit...

Their answer:

> If you are interested in how well we do compared to demucs in particular, we can use the MUSDB18 dataset since that is the domain that demucs is trained to work well on. There our net win rate against demucs is ~17%, meaning we do perform better on the MUSDB18 test set. There are actually stronger competitors on both this domain and our "in-the-wild" instrument stem separation domain that we built for SAM Audio Bench, but we either match or beat all of the ones we tested (AudioShake, LalalAI, MoisesAI, etc.)

So ~20% better than demucs, better than the ones they tested, but the acknowledge there are better models out there even today. So not sure "competes against SOTA models" is right, but "getting close to compete against SOTA models" might be more accurate.

Have you seen YouTube's front-page? It's pretty much 20-40min videos full of fluff.

I don't know if you wrote it as a form of satire, but obviously thete is no such thing as "YouTube's front-page". Everyone gets recommended different videos, based on various signals, even when not authenticated.

What privacy? Whoever is watching your traffic can see you accessed their website with HTTPS, they can guess with high accuracy which article you are reading based on the response size.

Any hops along the paths and whatever they split off to by whoever. And of course they can, even with HTTPS the Client Hello is unencrypted.

Unencrypted data transmission just isn't a thing I'm interested in with it being 2025.

You can enable it in the settings.

I'm guessing you're not storing the CLIP for every single frame, instead of every second or so? Also, are you using the cosine similarity? How are you finding the nearest vector?

I split per scene using pyscenedetect and sampled from each. Distance is via cosine similarity- I fed it into qdrant

BitTorrent v2 uses SHA-256, but in any case SHA-1 is still second-preimage resistant. And the BitTorrent piece hashes are included in the .torrent file, so you would need to find a double collision.