Tag

mechanistic-interpretability

Pieces tagged mechanistic-interpretability, across every department of the publication. Tags trace threads — vocabulary, regulatory regime, technical method — through pieces that may sit in different departments.

1 piece.

• 2026-05-10

  Cornerstone

  Anthropic's Mechanistic Interpretability Work at Year Five — What Shipped

  By Annika Vogel

  A working census of what the lab's mechanistic-interpretability program has actually published since the transformer-circuits thread opened in 2021. Sparse autoencoders, feature dictionaries, attri…

Browse every department and tag, or open the full archive.