List your company

Refusal Evaluation on XSTest Seemingly Toxic Subsets

98XSDCR
64.7273.368290.64Feb 10, 2026

Evaluation results

MethodLinks
DCRBase Model=Qwen2.5-1.5B2026.029898838686
SurgicalBase Model=Qwen2.5-7B2026.029396719689
DCRBase Model=Qwen2.5-7B2026.029396719491
DCRBase Model=LLaMA-3-8B2026.029399859290
SCANSBase Model=Qwen2.5-7B2026.028492509791
STL-augBase Model=LLaMA-3-8B2026.028496598785
SCANSBase Model=LLaMA-3-8B2026.028497868090
SCANSBase Model=Qwen2.5-1.5B2026.028392878487
SurgicalBase Model=Qwen2.5-1.5B2026.028184547854
STLBase Model=LLaMA-3-8B2026.027994598985
STL-augBase Model=Qwen2.5-1.5B2026.027590697675
STL-augBase Model=Qwen2.5-7B2026.027489538583
STLBase Model=Qwen2.5-1.5B2026.027388727575
SurgicalBase Model=LLaMA-3-8B2026.027290538985
STLBase Model=Qwen2.5-7B2026.026687348780