Robustness Evaluation

Benchmarks

Dataset Name	SOTA Method	Metric
Robust LQR	Pathwise Robust ZO Both	Mean Normalized Performance0.57	70	1mo ago
Robust LQR adversarial Normalized performance (test)	Pathwise Robust ZO Both	Normalized Performance1.603	70	1mo ago
SI-Score size synthetic	RobustViT	R@166.5	31	4mo ago
SI-Score rotation synthetic	RobustViT	R@158	31	4mo ago
SI-Score location synthetic	RobustViT	R@148.3	31	4mo ago
MultiRLVR	Master-RM	FPR (%)0.02	20	4mo ago
MATH	AdvJudge-Zero	FPR (%)0	20	4mo ago
GSM8K	Master-RM	FPR (%)0	20	4mo ago
AIME	Master-RM	FPR0	20	4mo ago
HellaSwag SAGE-generated	GPT-4o	Overall Accuracy (OA)74.17	12	2mo ago
R-Bench 2024 (test)	Robust-U1	MCQ Accuracy (low)73.53	9	1mo ago
SA-1b photos	CIN	Identity Bit Accuracy100	9	4mo ago
Meta AI images	CIN	Identity Bit Acc100	9	4mo ago
Perturbation Dataset	L4L	Change Accuracy62.56	8	4mo ago
LLMBar	Qwen3-30B-A3B-Thinking-2507	Accuracy83.07	8	4mo ago
BiasBench		Accuracy82.5	8	4mo ago
Lexical Variation (abbr.)	Mamba	Jensen-Shannon Divergence0.0476	8	4mo ago
CIFAR-100-C	Deep ens. (LPBN)	mCE43.15	8	4mo ago
VizWiz	Latent Denoising	Accuracy70.9	6	3mo ago
RWQA	Latent Denoising	Accuracy72.9	6	3mo ago
NaturalBench	Latent Denoising	GACC33.5	6	3mo ago
CartPole A=9.5 (test)	+DR	Average Reward231.8	6	4mo ago
CartPole A=9.0 (test)	+ESN-OA-PT	Average Reward830.9	6	4mo ago
CartPole A=8.5 (test)	+ESN-OA-PT	Average Reward810.1	6	4mo ago
CartPole A=8.0 (test)	+ESN-OA-PT	Average Reward1,000	6	4mo ago

Showing 25 of 53 rows