Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Sycophancy Evaluation on POLI
Loading...
92.18
Sycophantic Preference (%)
Ours Resid
47.4184
59.0392
70.66
82.2808
Jan 26, 2026
Sycophantic Preference (%)
Updated 4d ago
Evaluation Results
Method
Method
Links
Sycophantic Preference (%)
Ours Resid
Model=Gemma-2-9B, Prob...
2026.01
92.18
Ours Resid
Base Model=Gemma-2-2B,...
2026.01
86.41
Ours SAE
Model=Gemma-2-9B, Prob...
2026.01
86.13
Ours SAE
Base Model=Gemma-2-2B,...
2026.01
79.6
Synthetic Data Intervention
Model=Gemma-2-9B
2026.01
74.59
Untrained Gemma-2-9B
Model=Gemma-2-9B
2026.01
74.2
Supervised Pinpoint Tuning
Model=Gemma-2-9B
2026.01
73.95
Untrained Gemma-2-2B
Base Model=Gemma-2-2B
2026.01
50.22
Supervised Pinpoint Tuning
Base Model=Gemma-2-2B
2026.01
50.12
Synthetic Data Intervention
Base Model=Gemma-2-2B
2026.01
49.14
Feedback
Search any
task
Search any
task