HN user

arilotter

493 karma
Posts0
Comments80
View on HN
No posts found.
Iroh 1.0 1 month ago

My company was using Iroh for a production distributed ML training system & we LOVED it. The team was incredibly responsive even before we hooked up with an enterprise support contract, they're incredibly knowledgeable and the library itself worked amazingly. ++ to this lib. would use again over libp2p anytime.

This title is moderately clickbait-y and comes with a subtle implication that Rust might be getting removed from the kernel. IMO it should be changed to "Rust in the kernel is no longer experimental"

big ++ for iroh - using it for a project at work, it's evolving really nicely, has reliable holepunching, and the team is super responsive and happy to dig thru gigs of logs to find a bug :)

you could absolutely use DisTrO for federated learning. The DeMo optimizer on its own doesn't solve the adverserial aspects of training on local-only data, nor does it solve tensor parallelism across devices, so you're still limited to only what fits on your local GPUs, but it does enable distributed data parallelism over the internet at a bandwidth orders of magnitude lower than before.

Are you planning to eventually do a SETI@home style thing where anyone can join?

That's one of the goals of the stuff we're working on right now, I personally hope we can make it work!

What was Durk Kingma's involvement in DisTrO?

As a co-author on the paper, we brought him in to bounce some ideas off of & have him validate DeMo's design and implementation to ensure we weren't hallucinating these results.

We're excited about the potential and want to find other folks also excited about it that are interested in working for/with us to build things on the foundations of DisTrO! Plus also it's so cool and mind boggling to us that we wanted to share the hype a little bit, it was hard not being able to tell anyone we were working on it

We're pretty confident it's not a fluke, and paper + code are the next step, within a couple months. It's not "synchronize every step", but it's "do something every step".

We double and triple and quadruple checked our results, to make sure that we are in fact getting results like this while only doing our thing every step, and it really keeps holding up.

Don't trust our word for it, though, you'll see when the paper comes out :)

It's a number in a database that's assigned an "owner". Some NFTs can be controlled by a central authority, kind of like your bank account, and some of them are totally decentralized and can only be modified/traded if you, the owner, wills it so.

Someone can associate some data with that unique number, if they want.

AFAIK they are using a reverse-engineered Apple binary that does the real iMessage - and since Apple doesn't update apps outside of iOS updates, Apple can't, in theory, patch it without also breaking iMessage on older iPhones.

Beeper Mini does not MITM messages - it's a reverse engineering of how iMessage works, and runs entirely on-device, without putting messages thru Beeper's servers. It talks directly to Apple and pretends to be an iPhone.

You're thinking of Beeper Cloud, which does iMessage thru a Mac Mini in the cloud.

oh my god I just spend the last 6 months building something exactly like this as an internal tool because i couldn't find ANYTHING open-source that was remotely powerful enouhh for what i needed. this fits basically every single use case i had, i would only have to write a custom storage format to make data git-trackable.

i'm gonna go cry now.

please don't tell me this has been out for a while and i missed it.

Yes, but only for the stress it causes me in thinking really really hard all day and then being exhausted at night, which i think is not unique to a computers career. I don't regret it in that it's an incredibly high paying job with very low social barriers to entry.

Built one of these with my dad when I was about 8, and this was one of the things that got me bitten by the "working with computers" bug.