1wheel

10 ms·

1.
by 1wheel 6mo ago
2015 was about the last year you could get away with publishing an interactive graphic with a fixed width — this made it harder do really creative/original work.
2.
by 1wheel 6mo ago
https://visxai.io/ has a bunch more too — see the Hall of Fame section at the bottom for some of the highlights. I also made a dozen of these a couple years ago, my two favorites: - https://pair.withgoogle.com&#x
3.
When Models Manipulate Manifolds: The Geometry of a Counting Task(transformer-circuits.pub)
4 points by 1wheel 11mo ago|0 comments
4.
by 1wheel 1y ago
It can be! Here's a circuit showing how the model processes "PCB tracing stands for" to output "printed": https://www.neuronpedia.org/gemma-2-2b/graph?slug=pcb-tracin...
5.
by 1wheel 1y ago
Browsers won't use http2 unless https is on — chrome only allows six concurrent requests to the same domain if you're not using https!
6.
by 1wheel 2y ago
https://www.lkk-store.com/lkk-map/
7.
Advent of Code Solve Times(roadtolarissa.com)
2 points by 1wheel 2y ago|0 comments
8.
Anthropic raises an additional $4B from Amazon(techcrunch.com)
4 points by 1wheel 2y ago|0 comments
9.
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet(transformer-circuits.pub)
168 points by 1wheel 2y ago|124 comments
10.
The Claude 3 Model Family: Opus, Sonnet, Haiku [pdf](www-cdn.anthropic.com)
33 points by 1wheel 3y ago|3 comments
11.
Circuits Updates – January 2024(transformer-circuits.pub)
1 points by 1wheel 3y ago|0 comments
12.
Patchscopes: A Unifying Framework for Inspecting Hidden Representations of LMs(pair-code.github.io)
2 points by 1wheel 3y ago|0 comments
13.
by 1wheel 3y ago
There's also https://lmql.ai/
14.
by 1wheel 3y ago
Basically just a bunch of d3 — could be cleaned up significantly, but that's hard to do while iterating and polishing the charts. I also have a couple of little libraries for things like annotations, interleaving svg/canvas and ma
15.
Do Machine Learning Models Memorize or Generalize?(pair.withgoogle.com)
454 points by 1wheel 3y ago|210 comments
16.
An interactive introduction to grokking and mechanistic interpretability(pair.withgoogle.com)
1 points by 1wheel 3y ago|0 comments
17.
From Confidently Incorrect Models to Humble Ensembles(pair.withgoogle.com)
1 points by 1wheel 3y ago|1 comments
18.
by 1wheel 3y ago
https://pair.withgoogle.com/explorables/private-and-fair/
19.
Speeding up the JS ecosystem - npm scripts(marvinh.dev)
9 points by 1wheel 4y ago|3 comments
20.
Searching for Unintended Biases with Saliency(pair.withgoogle.com)
2 points by 1wheel 4y ago|1 comments
21.
by 1wheel 4y ago
Here's an example of that with a smaller BERT model: https://pair.withgoogle.com/explorables/fill-in-the-blank/
22.
Interactive Visualizations of Federated Learning(pair.withgoogle.com)
1 points by 1wheel 4y ago|0 comments
23.
How Federated Learning Protects Privacy(pair.withgoogle.com)
2 points by 1wheel 4y ago|0 comments
24.
by 1wheel 4y ago
https://www.westsiderag.com/2022/05/24/rumors-of-the-phone-b...
25.
It’s Not Spider-Man’s Fault: Why Best Picture Winners Aren’t Hits Anymore(roadtolarissa.com)
2 points by 1wheel 4y ago|0 comments
26.
Why Some Machine Learning Models Leak Data(pair.withgoogle.com)
3 points by 1wheel 5y ago|0 comments
27.
What Have Language Models Learned?(pair.withgoogle.com)
1 points by 1wheel 5y ago|0 comments
28.
Static RSS Reader(roadtolarissa.com)
1 points by 1wheel 5y ago|0 comments
29.
Election Forecast Correlations(roadtolarissa.com)
3 points by 1wheel 6y ago|0 comments
30.
by 1wheel 6y ago
Mine is 60 lines of js; the markdown library does most of the work. https://roadtolarissa.com/literate-blogging/

More ›