Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Sycophancy Reduction (held-in and held-out)
Loading...
79
Activation Patching
Difference-in-means
39.48
49.74
60
70.26
Feb 17, 2026
Activation Patching
Attribution Patching
Attention Head Knockouts
ITI Performance
Random Selections Score
Updated 4d ago
Evaluation Results
Method
Method
Links
Activation Patching
Attribution Patching
Attention Head Knockouts
ITI Performance
Random Selections Score
Difference-in-means
Model=SOLAR-10.7B, alp...
2026.02
79
78
70
71
60
Difference-in-means
Model=Qwen-14B, alpha=...
2026.02
78
80
79
66
64
Mean steering
Model=Qwen-14B, alpha=...
2026.02
77
77
24
77
59
Difference-in-means
Model=OLMo-13B, alpha=...
2026.02
70
69
58
64
53
Mean steering
Model=OLMo-13B, alpha=...
2026.02
67
70
44
48
56
ReFT
Model=SOLAR-10.7B, alp...
2026.02
58
57
56
58
57
ReFT
Model=OLMo-13B, alpha=...
2026.02
53
51
31
51
42
Mean steering
Model=SOLAR-10.7B, alp...
2026.02
46
44
11
61
39
ReFT
Model=Qwen-14B, alpha=...
2026.02
41
39
22
40
21
Feedback
Search any
task
Search any
task