Feature composition in SAEs collapses asymptotically when the Gaussian mean width of the signal cone is exceeded, with ReLU inducing a ratchet-like accumulation of interference from correlations.
Scaling Monosemanticity: Extracting Interpretable Features from
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Structural Instability of Feature Composition
Feature composition in SAEs collapses asymptotically when the Gaussian mean width of the signal cone is exceeded, with ReLU inducing a ratchet-like accumulation of interference from correlations.