Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Instruction Signal Recovery on DeceptionBench
Loading...
68.6
Judge Score (R)
PRISM
17.432
30.716
44
57.284
Jun 8, 2026
Judge Score (R)
Coverage Rate (Cvg)
Hallucination Rate (H)
Adversarial Detection Average (AD)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Judge Score (R)
Coverage Rate (Cvg)
Hallucination Rate (H)
Adversarial Detection Average (AD)
PRISM
Backbone=Qwen3.5-9B, R...
2026.06
68.6
68.8
0.3
87.1
PRISM w/o RL
Backbone=Qwen3.5-9B, R...
2026.06
43.6
45.4
3.6
61
Activation Oracles
Backbone=Qwen3.5-9B
2026.06
36.5
36.7
0.4
45.3
LatentQA
Backbone=Qwen3.5-9B
2026.06
19.4
19.7
0.8
20.5
Feedback
Search any
task
Search any
task