Share your thoughts, 1 month free Claude Pro on usSee more

Backdoor Attack Classification on HSOL

100ASR

RIPPLES

Updated 1mo ago

Evaluation Results

Method	Links
RIPPLES 2022.06		100	94.81
TrojanLM 2022.06		100	95.21
SOS 2022.06		100	95.78
EP 2022.06		100	95.25
Vanilla 2024.02		100	92.27
Prefix 2024.02		100	93.8
RAP 2026.03		100	91.29
No-Defense 2026.03		99.99	91.59
No-Defense 2026.03		99.01	89.52
Vanilla 2024.02		98.39	93.24
No-Defense 2026.03		98.27	90.28
RAP 2026.03		98.12	89.52
Vanilla 2024.02		97.66	92.13
LWS 2022.06		97.26	95.65
ONION 2026.03		94.18	91.16
LWP 2022.06		94.15	95.82
LoRA 2024.02		93.16	90.82
ONION 2026.03		90.18	89.53
Prefix 2024.02		88.16	93.8
No-Defense 2026.03		85.47	89.08
RAP 2026.03		84.91	88.76
ONION 2026.03		80.76	87.68
Prefix 2024.02		73.67	94.45
LoRA 2024.02		72.22	89.58
Vanilla 2024.02		68.92	94.81
Adapter 2024.02		67.31	83.62
BadActs 2026.03		62.77	86.37
RAP 2026.03		59.35	71.6
LoRA 2024.02		54.18	91.99
Adapter 2024.02		49.92	84.55
LoRA 2024.02		49.85	90.06
Adapter 2024.02		49.6	85.8
Adapter 2024.02		46.7	84.71
Prefix 2024.02		43.72	93.24
BadActs 2026.03		39.83	88.52
BadLLM-TG 2026.03		29.14	89.54
MuScleLoRA 2024.02		24.31	86
BadLLM-TG 2026.03		18.52	90.75
MuScleLoRA 2026.03		18.19	89.42
BadActs 2026.03		16.94	89.09
MuScleLoRA 2024.02		13.45	86.84
MuScleLoRA 2026.03		11.93	89.54
BadLLM-TG 2026.03		11.25	89.54
MuScleLoRA 2024.02		10.63	86.64
ONION 2026.03		8.57	89.26
BadLLM-TG 2026.03		7.78	89.54
MuScleLoRA 2024.02		2.74	85.47
MuScleLoRA 2026.03		1.15	89.54
MuScleLoRA 2026.03		0.8	89.34
BadActs 2026.03		0.3	89.54