AIGT detection on HC3 PWWS attack, AI to Human (in-domain) benchmark leaderboard · SOTA2 Research

AIGT detection on HC3 PWWS attack, AI to Human (in-domain)

100Overall AccuracyBERT

Evaluation Results

Method	Links
BERT2024.06		100	1.5	98.5	1,070.74
RoBERTa2024.06		100	38.5	61.5	1,332.6
DeBERTa2024.06		100	3	97	1,170.89
Flooding2024.06		100	23	77	1,422.6
RanMASK2024.06		100	50	50	1,562.84
RMLM2024.06		100	73.5	26.5	1,561.35
SCRN2024.06		100	94.5	5.5	1,665.53
RoBERTa2024.06		100	69	31	1,278.18
DeBERTa2024.06		100	51.25	48.75	1,197.21
Flooding2024.06		100	61.5	38.5	1,323.81
RanMASK2024.06		100	75	25	1,404.4
RMLM2024.06		100	86	14	1,388.94
SCRN2024.06		100	97.25	2.75	1,445.28
SeqXGPT2024.06		99.75	55.5	44.36	1,296.47
BERT2024.06		99.75	50	49.87	1,140.96
RDrop2024.06		99.75	83.75	16.04	1,405.08
SeqXGPT2024.06		99.5	11	88.94	1,368.07
RDrop2024.06		99.5	67.5	32.16	1,585.15
ChatGPT-Detector2024.06		99	49.75	49.75	1,149.79
ChatGPT-Detector2024.06		98	0	100	1,074.98
Log-Likelihood2024.06		98	49.5	49.49	1,090.48
Log-Rank2024.06		97.75	49.25	49.62	1,103.9
GLTR2024.06		97	41.75	56.96	1,061.31
Log-Rank2024.06		96.5	0	100	974.2
Log-Likelihood2024.06		96	0	100	957.42
GLTR2024.06		95	0	100	986.1
Entropy2024.06		90.5	39.5	56.35	1,037.86
Entropy2024.06		86	0	100	962.97