Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Hypocritical Sycophancy Detection on Sycophancy dataset knows-truth
Loading...
74
AUROC
Hypocrisy Gap (H)
39.576
48.513
57.45
66.387
Jan 14, 2026
AUROC
Updated 4d ago
Evaluation Results
Method
Method
Links
AUROC
Hypocrisy Gap (H)
backbone=Gemma-2B-IT
2026.01
74
Hypocrisy Gap
Predictor=H, Model=Gem...
2026.01
73.9
Hypocrisy Gap (H)
backbone=Llama-3.1-8B-...
2026.01
55.9
Hypocrisy Gap
Predictor=H, Model=Lla...
2026.01
55.8
Hypocrisy Gap
Predictor=H, Model=Qwe...
2026.01
55
Hypocrisy Gap (H)
backbone=Qwen3-1.7B
2026.01
54.6
Log-probability baseline (ΔLP)
backbone=Llama-3.1-8B-...
2026.01
49
Log-probability baseline (ΔLP)
backbone=Qwen3-1.7B
2026.01
45
Log-probability baseline (ΔLP)
backbone=Gemma-2B-IT
2026.01
40.9
Feedback
Search any
task
Search any
task