Centroid-difference steering for sycophancy in Llama-3-8B reduces agreement with factually correct statements as well as sycophantic ones, and this non-specificity is invisible under standard single-stance evaluation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Dual-Stance Evaluation of Sycophancy: The Structure of Agreement and the Limits of Intervention
Centroid-difference steering for sycophancy in Llama-3-8B reduces agreement with factually correct statements as well as sycophantic ones, and this non-specificity is invisible under standard single-stance evaluation.