List your company
ResearchFormerly WizWand

Robustness Evaluation on MultiRLVR

0.02FPR (%)Master-RM
-3.979223.015450.0177.0046Dec 19, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Master-RMTarget Judge=General Verifier2025.120.02
Master-RMTarget Judge=Master-RM Judge2025.120.02
Master-RMTarget Judge=Omni-Judge2025.120.34
Master-RMTarget Judge=Qwen2.5-RLVR2025.120.63
AdvJudge-ZeroTarget Judge=Master-RM Judge2025.122.33
AdvJudge-ZeroTarget Judge=Qwen2.5-RLVR2025.122.48
AdvJudge-ZeroTarget Judge=General Verifier2025.123.77
Master-RMTarget Model=Qwen2.5-7B2025.1213.78
Master-RMTarget Model=Llama-3.2-3B2025.1228.22
Master-RMTarget Model=Llama-3.3-70B2025.1245.58
AdvJudge-ZeroTarget Judge=Omni-Judge2025.1249.47
Master-RMTarget Model=gemma-3-4b2025.1276.7
Master-RMTarget Model=Qwen3-30B2025.1279.17
AdvJudge-ZeroTarget Model=Qwen2.5-7B2025.1279.45
Master-RMTarget Model=Qwen3-4B2025.1283.33
AdvJudge-ZeroTarget Model=Llama-3.2-3B2025.1293.28
AdvJudge-ZeroTarget Model=Llama-3.3-70B2025.1295.78
AdvJudge-ZeroTarget Model=Qwen3-4B2025.12100
AdvJudge-ZeroTarget Model=Qwen3-30B2025.12100
AdvJudge-ZeroTarget Model=gemma-3-4b2025.12100