List your company
ResearchFormerly WizWand

LLM-generated text detection on Xsum, WritingPrompts, and SQuAD generated by GPT-4.1-mini (test)

80.25AUROCSurpMark
37.058848.271959.48570.6981Oct 8, 2025
Updated 5mo ago

Evaluation results

MethodLinks
SurpMarkBlack-box setting=true, k (number of discretization states)=6, B (reference-per-test budget)=22025.1080.25
SurpMarkBlack-box setting=true, k (number of discretization states)=7, B (reference-per-test budget)=22025.1078.48
R-DetectBlack-box setting=true2025.1071.64
BinocularsBlack-box setting=true2025.1071.12
DetectNPRBlack-box setting=true, B (reference-per-test budget)=1002025.1070.83
DetectGPTBlack-box setting=true, B (reference-per-test budget)=1002025.1070.08
Fast-DetectGPTBlack-box setting=true2025.1068.32
Lastde++Black-box setting=true, B (reference-per-test budget)=1002025.1068.23
LogRankBlack-box setting=true2025.1066.8
LikelihoodBlack-box setting=true2025.1066.77
DetectLRRBlack-box setting=true2025.1063.29
FourierGPTBlack-box setting=true2025.1063.05
LastdeBlack-box setting=true2025.1057.28
DNA-GPTBlack-box setting=true, B (reference-per-test budget)=102025.1056.71
EntropyBlack-box setting=true2025.1038.72