nucleation

LIT-1543: Zoom In: An Introduction to Circuits

1 min read

Properties5
StatusActive — read 2026-10-26 (NOTE-1182) in full, from the Distill article; worth reading as the programme statement of circuits research: features as the unit, circuits as weighted subgraphs between them that can be read off the weights, and universality across models, each offered by the authors as a deliberately speculative claim. Read it knowing that the examples are from one vision model, InceptionV1, that the detailed evidence was promised to later articles of the thread, and that the universality claim rests, in the authors' word, on anecdote (THEORY-1013, THEORY-986).
Version2
Tagsrepresentation-learning, philosophy-of-science, anthology-candidate
Cited as “paper” byNOTE-1182 — Olah, Cammarata, Schubert, Goh, Petrov and Carter — Zoom in, an introduction to circuits
Cited as “source” byTHEORY-1013 — In trained vision networks, units or other directions in a layer's activation space are typically meaningful features, and the weights between understood features implement algorithms that can be read off them, so that features and their circuits are the units in which such a network can be understood, as Olah, Cammarata, Schubert, Goh, Petrov and Carter propose, THEORY-986 — Analogous features and circuits form across neural networks trained on different architectures, data and tasks, so that a feature or circuit found in one model can be expected in others, as Olah, Cammarata, Schubert, Goh, Petrov and Carter propose under the name universality

Graph View

Olah, C., Cammarata, N., Schubert, L., Goh, G., Petrov, M. and Carter, S. (2020). Zoom in: An introduction to circuits. Distill, 10 March 2020. doi:10.23915/distill.00024.001.

Read in full on 2026-10-26 (NOTE-1182). Two accounts were drawn from it, both Proposed: THEORY-1013 and THEORY-986.

Why this is here

Registered and read at the owner’s request of 2026-10-26, after an answer to “how’s the record on neural circuit theory?” found that the record held almost none of the core circuits literature. This is the work that names the programme. The record’s circuits line had its later results, Toy Models of Superposition (LIT-323), the grokking circuit (LIT-345) and the induction-head papers, without the statement of what a feature and a circuit are and why they should be the units. It also names the universality hypothesis, which CLAIM-263’s test for whether two networks compute the same way presupposes can fail, and it first describes as superposition the case that LIT-323 later models.

Backlinks