List your company

Sycophancy Reduction on philpapers 2020 (test)

2.27Delta LogDEVIL’S ADV.
-0.060640.544431.14951.75457May 20, 2026

Evaluation results

MethodLinks
DEVIL’S ADV.Backbone=Qwen 3 32B, COEF=+2002026.052.2716.6—
CAA (TARGETED)Backbone=Qwen 3 32B, COEF=−2002026.051.9720.9—
SKEPTICBackbone=Qwen 3 32B, COEF=+2002026.051.8218.1—
JUDGEBackbone=Qwen 3 32B, COEF=+2002026.051.74.4—
RANDOM (n = 10)Backbone=Qwen 3 32B2026.051.05810.3—
CAA (TARGETED)Backbone=Gemma 2 27B, COEF=−2k2026.050.8798.9—
SKEPTICBackbone=Gemma 2 27B, COEF=+2k2026.050.7119.6—
PEACEKEEPERBackbone=Qwen 3 32B, COEF=−2002026.050.7090.1—
JUDGEBackbone=Gemma 2 27B, COEF=+2k2026.050.5569.3—
DEVIL’S ADV.Backbone=Gemma 2 27B, COEF=+2k2026.050.5218.7—
RANDOM (n = 10)Backbone=Gemma 2 27B2026.050.2542.1—
PACIFISTBackbone=Gemma 2 27B, COEF=+2k2026.050.10.3—
PEACEKEEPERBackbone=Gemma 2 27B, COEF=+2k2026.050.0521.8—
COLLABORATORBackbone=Gemma 2 27B, COEF=+5002026.050.0451.7—
COLLABORATORBackbone=Qwen 3 32B, COEF=−1002026.050.0291.7—