Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.
Chain-of- thought reasoning in the wild is not always faithful
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Overthinking: Amplifying Reasoning Weights to Extract Learned Secrets
Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.