Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Deception Evaluation on Sycophancy
Loading...
89.77
CoT Plan Accuracy
GRPO
11.2916
31.6658
52.04
72.4142
Mar 27, 2026
CoT Plan Accuracy
Actual Deception Rate
CoT Faithfulness Score
Updated 19d ago
Evaluation Results
Method
Method
Links
CoT Plan Accuracy
Actual Deception Rate
CoT Faithfulness Score
GRPO
Backbone=Llama-3.1-8B
2026.03
89.77
72.27
74.54
GRPO
Backbone=Qwen3-8B
2026.03
66.59
37.95
72.5
CoT Monitor
Backbone=Qwen3-8B
2026.03
35.68
18.86
84.54
CoT Monitor
Backbone=Llama-3.1-8B
2026.03
19.31
72.5
30.91
SAR
Backbone=Qwen3-8B
2026.03
16.81
9.77
89.31
SAR
Backbone=Llama-3.1-8B
2026.03
14.31
35
70.9
Feedback
Search any
task
Search any
task