Reasoning models encode step importance in their activations pre-hoc, and this signal is more informative than tokens for identifying critical reasoning steps.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Reasoning Models Know What's Important, and Encode It in Their Activations
Reasoning models encode step importance in their activations pre-hoc, and this signal is more informative than tokens for identifying critical reasoning steps.