Archive - Manatee of manatees

Inference-time scaling makes interpretability harder

Some quick, probably-non-original thoughts

Mar 8 •

February 2025

Homogeneity puzzles

Why results about homogeneous neural networks can apply to some non-homogeneous ones

Feb 13 •

September 2024

How neural networks learn: margin maximisation

A partial answer regarding how neural networks generalise

Sep 24, 2024 •

August 2024

The start of Vǫluspá, translated

Here’s my translation of the first five stanzas of Vǫluspá, in which a seeress seeks permisison to recount (and then begins recounting) a creation myth.

Aug 28, 2024 •

June 2024

May 2024

Why deep learning is mysterious

Deep neural networks trained with gradient descent generalise well. This is surprising.

May 30, 2024 •

Symmetry arguments

Symmetry arguments are a type of argument that shows up in several fields.

May 15, 2024 •

How do languages change?

Some historical linguistics

May 14, 2024 •

February 2023

This is Manatee of manatees.

Feb 26, 2023 •

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts