A legendary run: training a tiny MLP with ABS activations
2026-09-27
XOR is the oldest toy problem in neural networks, and the one that ended the first perceptron era: a single linear unit cannot separate the four point...
Thoughts, ideas, and explorations
XOR is the oldest toy problem in neural networks, and the one that ended the first perceptron era: a single linear unit cannot separate the four point...
We got here by accident, chasing a piece of news from pure mathematics.
The protection system offers graduated security levels through RDP (Read Protection) settings, but recent academic research reveals critical flaws tha...
This is the first post on my blog. Here I'll share my thoughts about the absolute state of humanity, technology, hardcore research and other interesti...
Let $S_n = \bigoplus_{i=1}^{n} i$ denote the cumulative XOR of integers from 1 to $n$, where $\bigoplus$ represents the bitwise XOR operation.