Spectral analysis of transformer hidden states reveals phase transitions distinguishing reasoning from recall, with the spectral exponent alpha predicting answer correctness at AUC up to 1.000 before output generation.
For token-level dynamics (Finding 5–6), we use a sliding window of w= 10 tokens, computing SVD on H(ℓ) window ∈ Rw×d at each generation step
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
-
The Spectral Geometry of Thought: Phase Transitions, Instruction Reversal, Token-Level Dynamics, and Perfect Correctness Prediction in How Transformers Reason
Spectral analysis of transformer hidden states reveals phase transitions distinguishing reasoning from recall, with the spectral exponent alpha predicting answer correctness at AUC up to 1.000 before output generation.