List your company

Safety Alignment and Instruction Following on REDORCA

91.08HHDualSelect
48.356859.448470.5481.6316Jun 1, 2026

Evaluation results

MethodLinks
DualSelectModel=Llama-3-8B2026.0691.0884.8887.9864.92
DualSelectModel=Qwen3-4B2026.0690.5884.8687.7264.38
SQSDModel=Qwen3-4B2026.0685.8678.9482.464.72
DualSelectModel=Gemma-3-1B2026.0684.3882.7183.5564.28
SQSDModel=Llama-3-8B2026.0684.1280.2482.1863.76
TOSSModel=Qwen3-4B2026.0679.5285.2482.3862.48
TOSSModel=Llama-3-8B2026.0678.8485.3682.162.72
TOSSModel=Gemma-3-1B2026.0678.4276.1877.361.64
SQSDModel=Gemma-3-1B2026.0673.8683.0478.4563.06
LARFModel=Qwen3-4B2026.0666.4369.576860.74
LARFModel=Llama-3-8B2026.0665.6868.8967.2960.12
LARFModel=Gemma-3-1B2026.0664.6968.5266.6158.31
SEALModel=Qwen3-4B2026.0663.8866.3965.1459.46
SEALModel=Llama-3-8B2026.0662.4765.6164.0458.64
SEALModel=Gemma-3-1B2026.0661.2166.4463.8357.86
RandomModel=Qwen3-4B2026.0656.0658.0557.0650.68
RandomModel=Llama-3-8B2026.0655.6857.4956.5950.38
SafeInstrModel=Qwen3-4B2026.0655.3660.5157.9453.24
RandomModel=Gemma-3-1B2026.0655.1856.3455.7650.42
SafeInstrModel=Gemma-3-1B2026.0654.9360.5857.7651.42
SafeInstrModel=Llama-3-8B2026.0654.1259.0856.652.88
Standard SFTModel=Gemma-3-1B2026.0650505050
Standard SFTModel=Qwen3-4B2026.0650505050
Standard SFTModel=Llama-3-8B2026.0650505050