SOTA Safety Alignment benchmarks and papers with code

Benchmarks

Dataset Name	SOTA Method	Metric
HarmBench		ASR0	88	3mo ago
Salad Bench	ShaPO-T	MD0.68	68	2mo ago
HH-RLHF	ShaPO-T	MD Rate1.09	68	2mo ago
Do-Not-Answer	ShaPO-R	MD0	52	2mo ago
WildJailbreak		Trainable parameters (M)15,768.31	44	3mo ago
Visual Adversarial Attacks		ASR43.1	40	2mo ago
JOOD	MoRAS	ASR0	40	2mo ago
SORRY-Bench	LED-Merging	ASR10.22	40	22d ago
PKU-SafeRLHF 30K (IID)	ShaPO-T	WR89.26	36	3mo ago
AdvBench	SEA	Reward-0.38	32	2mo ago
Harmful Dataset (test)		Harmful Score81	30	3mo ago
AdvBench	BoN64	Harm Rate0	25	7d ago
WildJailbreak	R1 - 8B + UnsafeChain full	Safe@177.2	24	1mo ago
Safety Benchmarks (Sorry-bench, StrongREJECT, WildJailbreak, JBB-PAIR, JBB-GCG)	SafeChain	Average Score42.34	21	1mo ago
XSTest	Yi-VL-6B	Compliance95.2	21	22d ago
HEx-PHI	DiaBlo	HEx-PHI Score98.8	18	14d ago
HarmBench	SFT	MD Score95	18	3mo ago
Average (Do-Not-Answer, HarmBench, HH-RLHF, Salad Bench)	ShaPO-T	Aggregate Score0.59	18	3mo ago
StrongReject	R1 - 7B + UnsafeChain full	Safe@158	18	1mo ago
PKU-SafeRLHF	PPO	Gold Reward3.92	14	3mo ago
BeaverTails V	SaFeR-ToolKit (+ SFT+GRPO) [3B]	Safety Score93.37	13	3mo ago
Safety Alignment	PALM	Multiplicative Gap (Epsilon)0.0131	12	1mo ago
HarmBench	ImplicitRM	Score98.2	10	2mo ago
Safety Alignment Dataset 4-order (test)	MOSAIC-2	DSR100	10	2mo ago
Safety Alignment Dataset 3-order (test)	MOSAIC-5	DSR100	10	2mo ago

Showing 25 of 52 rows