Revision History
Superposition (Mechanistic Interpretability) · 7 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected the Toy Models sparsity definition (S is the probability a feature is zero), the antipodal-pair geometry (one shared dimension, not two), the claim that Toy Models computed logical AND (it showed absolute value only), and the description of the Towards Monosemanticity model (a standard one-layer transformer with an MLP, not attention-only) (plus 25 minor fixes)
Version 7 · Nov 4, 2026, 05:53 PM
- Improve internal links (2026-10-13)
Version 6 · Oct 13, 2026, 10:04 AM
- Added 6 contextual internal links
Version 5 · Jul 23, 2026, 03:31 PM
- Initial AI-authored creation (4885 words).
Version 2 · May 18, 2026, 09:51 PM