Nadav pfp

Nadav

@nadav

22 Following
981 Followers
𝚐𝔪𝟾𝚡𝚡𝟾 pfp
𝚐𝔪𝟾𝚡𝚡𝟾
@gm8xx8
INTELLECT-1, the first 10B model trained via a decentralized framework, is now officially released. The release includes the base model, checkpoints, post-trained models, datasets, technical report, and training framework. Another w for open source! This demonstrates the potential of decentralized deep learning to collaboratively build large-scale models. https://www.primeintellect.ai/blog/intellect-1-release
2 replies
8 recasts
44 reactions
Nadav pfp
Nadav
@nadav
determinism without provability is probably under researched, though not sure if that’s remotely feasible given floats are non deterministic
1 reply
0 recast
1 reaction
Nadav pfp
Nadav
@nadav
@remco from Worldcoin had a cool talk on this in Devcon from a few years ago and prob best source for practical knowledge on zkML / zk neural nets. AFAIK for any model reasonably large or useful by modern standards though TEEs are the closest you can get today — cool stuff from Nous research here
1 reply
0 recast
1 reaction
Nadav pfp
Nadav
@nadav
what is the state of art for how wallets are determining ERC20 token lists nowadays?
0 reply
0 recast
15 reactions
Nadav pfp
Nadav
@nadav
In other words, I always thought “decentralized AI” was maybe a little bit of a solution looking for a problem because of OSS models like Llama. But that might no longer be the case
0 reply
0 recast
6 reactions
Nadav pfp
Nadav
@nadav
The emergence of a test-time compute overhang in scaling reasoning models implies a surprising importance to decentralized inference. If the most useful and insightful things AIs can do require enormous amounts of inference compute, it means the highest leverage work output in the world is censorable (or worse, abusable) by an incredibly small number of people.
1 reply
31 recasts
135 reactions
Nadav pfp
Nadav
@nadav
Sick
0 reply
0 recast
1 reaction
Nadav pfp
Nadav
@nadav
anyone know if they're on FC?
0 reply
0 recast
1 reaction
Nadav pfp
Nadav
@nadav
love Nous. should we get those guys in here?
2 replies
0 recast
4 reactions
Nadav pfp
Nadav
@nadav
https://huggingface.co/papers/2411.04905
0 reply
59 recasts
189 reactions
Nadav pfp
Nadav
@nadav
https://huggingface.co/papers/2411.10440
0 reply
5 recasts
10 reactions
Nadav pfp
Nadav
@nadav
interesting result from @danrobinson https://www.paradigm.xyz/2024/11/pm-amm
0 reply
3 recasts
24 reactions
Nadav pfp
Nadav
@nadav
first~!
1 reply
2 recasts
17 reactions
Nadav pfp
Nadav
@nadav
cc: @v @dwr.eth
1 reply
0 recast
3 reactions
Nadav pfp
Nadav
@nadav
Request for FC channel — high signal research papers across AI, Crypto, Biology, etc. Setting a goal to read 2-3 papers a week and need a backlog
1 reply
9 recasts
75 reactions
Nadav pfp
Nadav
@nadav
@vitalik.eth what is the steelman case for "DeSci"? I have a kneejerk reaction of skepticism to what feels like buzzword soup
0 reply
3 recasts
8 reactions
Nadav pfp
Nadav
@nadav
Can someone please summarize this truth terminal spiel on CT? I’ve been heads down and out of the loop
4 replies
2 recasts
12 reactions
Nadav pfp
Nadav
@nadav
Have seen a lot of disparate posts around preconfs, execution tickets, etc but most of that is half a year+ old — not sure what’s the latest state of the art on the topic. Thanks!
0 reply
0 recast
0 reaction
Nadav pfp
Nadav
@nadav
Cc: @justindrake
1 reply
0 recast
0 reaction
Nadav pfp
Nadav
@nadav
Are there any more fleshed out / precise articulations of how based rollups would work in practice than the OG post from Justin Drake? Trying to catch up on the various threads of research but a lot still seems underspecified or implicit
1 reply
2 recasts
8 reactions