bibkey: “elhage2022toy” authors: “Nelson Elhage; Tristan Hume; Catherine Olsson; Nicholas Schiefer; Tom Henighan; Shauna Kravec; Zac Hatfield-Dodds; Robert Lasenby; Dawn Drain; Carol Chen; Roger Grosse; Sam McCandlish; Jared Kaplan; Dario Amodei; Martin Wattenberg; Christopher Olah” year: 2022 title: “Toy Models of Superposition” doi: null claim: “Small ReLU reconstruction models store more sparse features than dimensions by placing them in superposition, with geometric phase changes as sparsity and importance vary.” strata_touched: [] license: “citation-only” triage: “anchor” url: “https://arxiv.org/abs/2209.10652”

Toy Models of Superposition

Verified locator

Transformer Circuits Thread, 2022; arXiv:2209.10652. Cited as background for superposition and sparse dictionaries; no theorem from it is used in the fiber-law volume.

Declared identifiers: https://arxiv.org/abs/2209.10652.