Our new X account is live! Follow @wizwand_team for updates

Priming Attack Robustness on AdvBench No Attack (test)

0ASR (GPT-4o)

MOSA

Updated 4d ago

Evaluation Results

Method	Links
MOSA 2025.10		0	0	0
RA w/o inter (ablation) 2025.10		0	0	0
RA (ours) 2025.10		0	0	0
MOSA 2025.10		0	0	0
RA w/o inter (ablation) 2025.10		0	0	0
RA (ours) 2025.10		0	0	0
RA w/o inter (ablation) 2025.10		0	0	0
RA (ours) 2025.10		0	0	0
Original 2025.10		2	2	4
Original 2025.10		2	2	6
DPO 2025.10		4	24	10
SFT 2025.10		6	6	6
DPO 2025.10		6	30	8
MOSA 2025.10		8	12	12
SFT 2025.10		16	6	4
DPO 2025.10		34	64	57
SFT 2025.10		60	54	36
Original 2025.10		78	72	56