List your company

Utility Evaluation on GPQA Diamond

53Accuracy (pass@1)No Defense
18.6827.5936.545.41Aug 6, 2025

Evaluation results

MethodLinks
No DefenseBackbone=R1-Qwen-7B2025.0853
SAFEPATH-FTBackbone=R1-Qwen-7B2025.0853
ReasoningGuardBackbone=R1-Qwen-7B2025.0853
SafeDecodingBackbone=R1-Qwen-7B2025.0852
Self-ReminderBackbone=R1-Qwen-7B2025.0852
No DefenseBackbone=R1-Llama-8B2025.0852
ThinkingIBackbone=R1-Qwen-7B2025.0850
SafeKeyBackbone=R1-Qwen-7B2025.0849
SafeDecodingBackbone=R1-Llama-8B2025.0848
RealSafe-R1Backbone=R1-Qwen-7B2025.0846
SAFEPATH-ZSBackbone=R1-Qwen-7B2025.0846
SAFEPATH-FTBackbone=R1-Llama-8B2025.0845
SAFEPATH-ZSBackbone=R1-Llama-8B2025.0845
RealSafe-R1Backbone=R1-Llama-8B2025.0844
Self-ReminderBackbone=R1-Llama-8B2025.0844
ReasoningGuardBackbone=R1-Llama-8B2025.0844
SafeKeyBackbone=R1-Llama-8B2025.0841
ThinkingIBackbone=R1-Llama-8B2025.0841
SmoothLLMBackbone=R1-Qwen-7B2025.0834
SmoothLLMBackbone=R1-Llama-8B2025.0831
ParaphraseBackbone=R1-Qwen-7B2025.0823
ParaphraseBackbone=R1-Llama-8B2025.0820