Safety Evaluation

Benchmarks

Dataset Name	SOTA Method	Metric
HEx-PHI	ID-LoRA	HEx-PHI Score97.2	148	3d ago
Advbench	AOA	Safety Score100	117	3d ago
DoNotAnswer Framed	TFS-IP-CoT	HRR0	96	3d ago
Sorry-Bench	IDGAF	Safety Score99.09	90	3d ago
Harmfulness Evaluation Sequences	llama2-13b-chat	Harmfulness Score0.79	84	3d ago
Harmbench	NPO	Harmbench Score0.06	76	2d ago
ToxiGen	Self-Improving Pretraining	Safety93.1	71	3d ago
LLaMA-2-7B-CHAT Safety (test)	TRAP	Safety Score0.55	60	3d ago
WildJailbreak (held-out)	NeST	Attack Success Rate (ASR)0	50	3d ago
StrongReject	DirectRefusal	Attack Success Rate0.64	45	3d ago
MM-SafetyBench	RAI	Average ASR0	42	3d ago
Harmful Prompts	Surgery	Harmful Score8.3	40	3d ago
CocoNot	GRAPH ROUTER	Safety Score0.613	36	3d ago
AdvBench 50 examples		Safe Response Rate100	32	3d ago
XSTest (test)	DPO + OGPSA	XSTest Score95	32	3d ago
PS-Bench base setting (test)		ASR (Hate Speech)18	30	3d ago
Safety Evaluation Suite HarmBench, StrongReject, WildJailbreak, XSTest		HarmBench Score68.44	28	3d ago
Sorry-Bench base		Safety Score92.73	27	3d ago
Safety	Olmo 3.1 32B Instruct	Score92.1	27	3d ago
Wildguard (test)	WHP (W/O safe set)	Wildguard Test Score0.08	27	3d ago
SafetyBench en	ROSE	Avg Score81.2	25	3d ago
UnsafeBench	BLPO	F1 Score89	24	3d ago
Harmful Benchmarks (CATQA, HEX-PHI, Salad-Base)	Safe LORA	CATQA Score99.94	24	3d ago
HH-RedTeaming	Llama-2 Chat	H.R.adv0.036	22	3d ago
HH-RLHF (test)	SAILS	Harm Score1.02	21	3d ago

Showing 25 of 162 rows