Neural networks that learn non-linearity without activation functions [pdf](tahabouhsine.com)17 points by mlnomadpy 1 year ago | 4 comments
A Mathematical Framework for Transformer Circuits (2021)(transformer-circuits.pub)95 points by Bluestein 1 month ago | 17 comments
Towards Monosemanticity: Decomposing Language Models with Dictionary Learning(transformer-circuits.pub)56 points by goodmachine 3 years ago | 3 comments
Why are neural networks and cryptographic ciphers so similar? (2025)(reiner.org)126 points by jxmorris12 5 months ago | 42 comments
Show HN: Microgpt is a GPT you can visualize in the browser(microgpt.boratto.ca)261 points by b44 8 months ago | 23 comments
![Screenshot of tahabouhsine.com: Neural networks that learn non-linearity without activation functions [pdf]](https://hn.urlbox.com/shots/renders/2025/8/3/6867fd91-fd75-4e6c-8067-1db64b2042f1.jpg)
![Screenshot of cdn.openai.com: Weight-sparse transformers have interpretable circuits [pdf]](https://hn.urlbox.com/shots/renders/2025/11/22/e1993c9b-4654-4516-a5a3-69700ad54a3c.jpg)
