LLM Moderation on WildGuardMix (test)

14.59ASR

MLPM Simple Refuse

Updated 1mo ago

Evaluation Results

Method
MLPM Simple Refuse 2025.02	14.59	6.03	83.85
MLPM Simple Refuse 2025.02	15.52	5.93	83.47
MLPM Simple Refuse 2025.02	15.92	6.24	82.84
MLPM Simple Refuse 2025.02	18.3	6.24	81.52
Turner et al. (2023) 2025.02	18.57	8.15	78.9
Turner et al. (2023) + MLPM Conditioning 2025.02	19.36	6.46	80.63
Zheng et al. (2024) 2025.02	19.76	8.25	78.12
Lee et al. (2025a) 2025.02	20.42	6.56	79.9
Base Model 2025.02	20.82	5.93	80.48
Zheng et al. (2024) + MLPM Conditioning 2025.02	20.95	6.35	79.86
Zheng et al. (2024) 2025.02	25.07	6.03	77.82
Zheng et al. (2024) + MLPM Conditioning 2025.02	26.92	4.02	79.21
Zheng et al. (2024) 2025.02	28.91	5.71	75.83
Zheng et al. (2024) + MLPM Conditioning 2025.02	31.56	3.92	76.3
Turner et al. (2023) 2025.02	35.94	53.76	27.2
Base Model 2025.02	36.21	2.54	74.75
Lee et al. (2025a) 2025.02	36.47	7.51	68.71
Turner et al. (2023) 2025.02	40.45	18.73	54.29
Turner et al. (2023) + MLPM Conditioning 2025.02	44.56	4.66	66.03
Turner et al. (2023) + MLPM Conditioning 2025.02	45.89	3.6	66.15
Turner et al. (2023) 2025.02	47.75	3.81	64.41
Turner et al. (2023) + MLPM Conditioning 2025.02	48.67	2.54	65.02
Lee et al. (2025a) 2025.02	50.13	2.65	63.66
Base Model 2025.02	50.53	2.22	63.78
Zheng et al. (2024) 2025.02	54.11	3.49	59.3
Zheng et al. (2024) + MLPM Conditioning 2025.02	55.44	2.43	59.16
Lee et al. (2025a) 2025.02	73.61	6.24	37.31
Base Model 2025.02	78.25	1.9	34.51