williamtrask

6 ms·

1.
Double Blind Evals: Resolving the Dual Confidentiality Dilemma in AI Safety...(storage.googleapis.com)
2 points by williamtrask 14d ago|0 comments
2.
Piloting the first double-blind AI evaluations(deepmind.google)
1 points by williamtrask 15d ago|0 comments
3.
frontier AI labs lost the "deep research" frontier(andrewtrask.substack.com)
2 points by williamtrask 2mo ago|0 comments
4.
Today's Frontier AI companies will never exceed the AI capability frontier again(andrewtrask.substack.com)
26 points by williamtrask 3mo ago|9 comments
5.
Frontier AI companies will never exceed the capability frontier again(andrewtrask.substack.com)
9 points by williamtrask 3mo ago|0 comments
6.
Show HN: AbcGPT(twitter.com)
2 points by williamtrask 4mo ago|0 comments
7.
by williamtrask 4mo ago
The scandals/year page has a little more umph to it than the main page https://ycombinator.fyi/timeline
8.
by williamtrask 4mo ago
Yeah... probably right. I do hold out hope that this is mostly a timeframe thing. Like, the library, printing press, etc. all had their moments of centralization. But eventually they federated.
9.
by williamtrask 4mo ago
fwiw - i think the design looks good.
10.
by williamtrask 4mo ago
I wonder if a popularization moment for local AI will ultimately be the pin-prick that pops the AI bubble. Like the deepseek or openclaw moments but bigger/next.
11.
by williamtrask 4mo ago
A non-profit to deconcentrate power over AI through better infrastructure for external auditing/oversight, and better infrastructure for local/federated inference/training https://openmined.org/ Also, we&#x27
12.
by williamtrask 5mo ago
i like your comment better than mine. more please.
13.
by williamtrask 5mo ago
i'm not sure it's productive to think this way. senators could be making more money on prediction markets. they took a nice step which will lead them to make no money on prediction markets (less money overall). it also sets a prec
14.
by williamtrask 5mo ago
Sometimes I find joy in noticing the importance of comma placement: Everything the incumbents ship, in an open codebase your firm owns. vs Everything the incumbents ship in an open codebase, your firm owns.
15.
Decentralized AI from Scratch (Python Tutorial) [video](youtube.com)
3 points by williamtrask 5mo ago|0 comments
16.
by williamtrask 5mo ago
<3 - have ambitious plans to release a new video/blogpost each week, but that's pretty ambitious we'll see
17.
Decentralized AI in 50 Lines of Python(iamtrask.github.io)
5 points by williamtrask 5mo ago|2 comments
18.
Decentralized AI from Scratch(github.com)
3 points by williamtrask 5mo ago|1 comments
19.
Zero-Setup Federated Learning: Train Models Across Private Datasets with GColab(openmined.org)
1 points by williamtrask 8mo ago|0 comments
20.
by williamtrask 9mo ago
...with a price :)
21.
Leadership Lab: The Craft of Writing Effectively [video](youtube.com)
2 points by williamtrask 9mo ago|0 comments
22.
by williamtrask 9mo ago
"Conclusion Our monorepo isn't about following a trend. It's about removing friction between things that naturally belong together, something that is critical when related context is everything. When a feature touches the bac
23.
by williamtrask 11mo ago
tried searching for "noodlesUK" and didn't find anything meaningful
24.
by williamtrask 11mo ago
Nit: regarding (2), Phil Blunsom did (same Blunsom from the article, and who was leading language modeling at DeepMind for about 7-8 years). He would often opine at Oxford (where he taught) that solving next word prediction is a viable meta
25.
The Bitter Lesson's Bitter Lesson(twitter.com)
5 points by williamtrask 1y ago|1 comments
26.
by williamtrask 1y ago
"This is not the reason, the reason is that this data is private. LLMs do not just learn from data, they can often reproduce it verbatim, you cannot give medical records or bank records of real people, that will put them at a very real
27.
by williamtrask 1y ago
With you on this one. I do think ABC is a step in the right direction to improve things. <3
28.
by williamtrask 1y ago
"The claim that humans need petabytes of data to develop their mind seems completely indefensible to me." And yet every human you know is using petabytes of data to develop their mind. :)
29.
by williamtrask 1y ago
I'm relatively close to publishing my PhD thesis which is broadly a survey paper of what you're describing. Will share ( almost done with revisions).
30.
by williamtrask 1y ago
I think this is the right question to ask. I think it depends on the task. For example, if you want to predict whether someone has cancer, then access to avast amounts of medical information would be important.

More ›