• Memory 1 -- How much do linear layers memorize?

  5 min read   ·   February 09, 2026

  2026   ·   Physics-of-AI   Memory     ·   AI  

• Transformers don't learn Newton's laws? They learn Kepler's laws!

  5 min read   ·   February 08, 2026

  2026   ·   Physics-of-AI   World-model     ·   AI  

• When I say "toy models", what do I mean?

  10 min read   ·   February 07, 2026

  2026   ·   Physics-of-AI   Methodology     ·   AI  

• On the physical interpretation of drifting generative models

  7 min read   ·   February 06, 2026

  2026   ·   Physics-of-AI   Diffusion     ·   AI  

• Physics 2 -- Transformers fail to maintain physical cosistency for circular motion

  4 min read   ·   February 05, 2026

  2026   ·   Physics-of-AI   AI-for-Physics   Physics   World-model     ·   AI