List your company

Utility assessment on MMLU-Pro

16.5Personalization Bias (PB)Identity-Robust Generation
-0.772115.814232.4348.986Jan 14, 2026

Evaluation results

MethodLinks
Identity-Robust GenerationModel=Llama3.3-70B2026.0116.5
Identity-Robust GenerationModel=Qwen32026.0119.6
Identity-Robust GenerationModel=gpt-oss-20B2026.0125.2
Vanilla GenerationModel=Llama3.3-70B2026.0135
Prompt SteeringModel=Llama3.3-70B2026.0136.6
Prompt SteeringModel=Qwen32026.0149
Vanilla GenerationModel=Qwen32026.0152.3
Prompt SteeringModel=gpt-oss-20B2026.0160.8
Vanilla GenerationModel=gpt-oss-20B2026.01448.3