Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Sycophancy Detection on Anthropic Sycophancy Dataset
Loading...
98.3
Accuracy
Contrastive + LR
52.124
64.112
76.1
88.088
Jun 23, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
Contrastive + LR
Approach=Linear Probin...
2026.06
98.3
CLiF + SVM
Approach=Linear Probin...
2026.06
98.3
CLiF + LR
Approach=Linear Probin...
2026.06
96.7
CLiF + LR
Approach=Linear Probin...
2026.06
93.6
CLiF + SVM
Approach=Linear Probin...
2026.06
93.6
CLiF + LR
Approach=Linear Probin...
2026.06
93.5
CLiF + LR
Approach=Linear Probin...
2026.06
93.2
CLiF + SVM
Approach=Linear Probin...
2026.06
91.5
Contrastive + SVM
Approach=Linear Probin...
2026.06
90
CLiF + SVM
Approach=Linear Probin...
2026.06
89.8
Gemini 2.5 Pro
Approach=LLM as a Judg...
2026.06
63.9
Gemini 2.5 Pro
Approach=LLM as a Judg...
2026.06
60
Gemini 2.5 Flash
Approach=LLM as a Judg...
2026.06
54.8
Gemini 2.5 Flash
Approach=LLM as a Judg...
2026.06
53.9
Feedback
Search any
task
Search any
task