Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Sycophancy Evaluation on SycophancyEval Experimental runs 1 and 2 (train test combined)
Loading...
106
Correct Count
BurGal
34.24
52.87
71.5
90.13
Apr 21, 2026
Correct Count
Total Samples
Accuracy
Z-Score vs Control
Z-Score vs Instruction Opponent
Accuracy 95% CI (Lower Bound)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Correct Count
Total Samples
Accuracy
Z-Score vs Control
Z-Score vs Instruction Opponent
Accuracy 95% CI (Lower Bound)
BurGal
Condition=BurGal†
2026.04
106
200
53
7.2
4.88
46.1
DeWin
Condition=DeWin
2026.04
97
200
48.5
6.36
4
41.7
FeynStein
Condition=FeynStein
2026.04
94
200
47
6.07
3.71
40.2
AnCifer
Condition=AnCifer
2026.04
87
200
43.5
5.41
3.02
36.8
Trident
Condition=Trident
2026.04
86
200
43
5.31
2.92
36.3
Instructed Opp.
Condition=Instructed Opp.
2026.04
58
200
29
2.47
-
23.1
Control
Condition=Control (Exp 2)
2026.04
38
200
19
-
-
13.7
Control
Condition=Control (Exp 1)
2026.04
37
200
18.5
-
-
13.7
Feedback
Search any
task
Search any
task
