Small transformers on HMM prediction tasks exhibit correlated scaling between performance and linear encoding of belief distributions in residual activations.
Riechers, Thomas J
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
-
Structure and Scale in Simplicial Sequence Modelling
Small transformers on HMM prediction tasks exhibit correlated scaling between performance and linear encoding of belief distributions in residual activations.