List your company

LLM Moderation on WildGuardMix (test)

14.59ASRMLPM Simple Refuse
12.043629.231846.4263.6082Feb 22, 2025

Evaluation results

MethodLinks
MLPM Simple RefuseBase LLM=OLMo2-7B2025.0214.596.0383.85
MLPM Simple RefuseBase LLM=Llama3-8B2025.0215.525.9383.47
MLPM Simple RefuseBase LLM=Mistral-7B2025.0215.926.2482.84
MLPM Simple RefuseBase LLM=Qwen2-7B2025.0218.36.2481.52
Turner et al. (2023)Base LLM=OLMo2-7B2025.0218.578.1578.9
Turner et al. (2023) + MLPM ConditioningBase LLM=OLMo2-7B2025.0219.366.4680.63
Zheng et al. (2024)Base LLM=OLMo2-7B2025.0219.768.2578.12
Lee et al. (2025a)Base LLM=OLMo2-7B2025.0220.426.5679.9
Base ModelBase LLM=OLMo2-7B2025.0220.825.9380.48
Zheng et al. (2024) + MLPM ConditioningBase LLM=OLMo2-7B2025.0220.956.3579.86
Zheng et al. (2024)Base LLM=Llama3-8B2025.0225.076.0377.82
Zheng et al. (2024) + MLPM ConditioningBase LLM=Llama3-8B2025.0226.924.0279.21
Zheng et al. (2024)Base LLM=Qwen2-7B2025.0228.915.7175.83
Zheng et al. (2024) + MLPM ConditioningBase LLM=Qwen2-7B2025.0231.563.9276.3
Turner et al. (2023)Base LLM=Llama3-8B2025.0235.9453.7627.2
Base ModelBase LLM=Llama3-8B2025.0236.212.5474.75
Lee et al. (2025a)Base LLM=Llama3-8B2025.0236.477.5168.71
Turner et al. (2023)Base LLM=Mistral-7B2025.0240.4518.7354.29
Turner et al. (2023) + MLPM ConditioningBase LLM=Llama3-8B2025.0244.564.6666.03
Turner et al. (2023) + MLPM ConditioningBase LLM=Mistral-7B2025.0245.893.666.15
Turner et al. (2023)Base LLM=Qwen2-7B2025.0247.753.8164.41
Turner et al. (2023) + MLPM ConditioningBase LLM=Qwen2-7B2025.0248.672.5465.02
Lee et al. (2025a)Base LLM=Qwen2-7B2025.0250.132.6563.66
Base ModelBase LLM=Qwen2-7B2025.0250.532.2263.78
Zheng et al. (2024)Base LLM=Mistral-7B2025.0254.113.4959.3
Zheng et al. (2024) + MLPM ConditioningBase LLM=Mistral-7B2025.0255.442.4359.16
Lee et al. (2025a)Base LLM=Mistral-7B2025.0273.616.2437.31
Base ModelBase LLM=Mistral-7B2025.0278.251.934.51