jasonmorton
Joined 17 years ago
Karma
116
Submissions
41
Member since
Oct 15
Recent submissions
Comment · 1 year ago
This lets you verify the signature on the model. It won’t help you tell that a decision came from that model. If you want to verify the inference that a model makes, check out https://github.com/zkonduit/ezkl (our project).
Comment · 2 years ago
You can ZK an XGBoost or neural net with https://github.com/zkonduit/ezkl
Comment · 3 years ago
Here is the twitter thread providing an introduction and some background: https://twitter.com/ezklxyz/status/1712940546182774936
Comment · 3 years ago
> Checking the work takes the same resources as doing the work.
> As much as this would be wonderful in a universe where it's possible, it's simply not possible.
We do live in that universe, under some currently believed assumptions. An NP-complete problem is an example of something where checking a solution is (thought to be) easier than finding one.
Zero-knowledge proofs make it such that checking that a computation (such as inference) has been done correctly is easier than doing the computation (even keeping some parts private). A great reference is here: https://people.cs.georgetown.edu/jthaler/ProofsArgsAndZK.pdf
Comment · 3 years ago
One can cryptographically prove the correct inference of a small AI model now with https://github.com/zkonduit/ezkl (our open-source package).
Comment · 3 years ago
Our project proves AI model execution with cryptography, but without any trusted hardware (using zero-knowledge proofs): https://github.com/zkonduit/ezkl
Comment · 4 years ago
There are definitely certain scenarios, especially around verifiable compute, in which you can think of ZK as a software alternative to a secure enclave.
Interestingly, if you ask people which they would trust for a rollup, most say ZK is the more trustworthy technology. In part just because it doesn't depend on one vendor (the enclave manufacturer).
Comment · 4 years ago
Here are two good places to start:
https://www.youtube.com/playlist?list=PLj80z0cJm8QErn3akRcqv...
Comment · 4 years ago
Examples include the various zkEVM projects, risc0, and slightly more limited stuff like our effort to compile PyTorch/ONNX to a zk circuit (https://github.com/zkonduit/ezkl).
Speed and memory usage is improving quickly.
Comment · 5 years ago
Not OP, but for me adding or deleting a character in the middle of a paragraph in a latex doc hangs for maybe .5s or more, often. This is a fairly new phenomenon, never happened 2+ years ago.
Comment · 5 years ago
I've been using emacs for many years now, and I'm sad to say that it has never been slower, despite the hardware getting better. Now it lags like crazy when editing a latex document, no matter the recent version. Something went off the rails in the last year or so, and even Atom is more responsive.
Comment · 6 years ago
Many ads are not clicked on by a real person, but a bot, or someone clicking ads as fast as possible because it is their job.
Many comments would be considered "fraudulent" in the same sense, in that they were not authored by a real person, but a bot, or someone writing comments as fast as possible because it is their job.
Comment · 6 years ago
We've been working with Modulo as well; the experience is excellent and is constantly adapting to make it work for our kids, rather than imposing a formula. They enjoy having some autonomy in their education. It's a very nice community as well, and they have the chance to make friends all over the country.
Comment · 7 years ago
Thanks, looks like this is: https://github.com/jitsi/jitsi-meet