Boundary Detection on RoFT-chatgpt GPT-3.5-turbo generated (test) benchmark leaderboard · SOTA2 Research

Boundary Detection on RoFT-chatgpt GPT-3.5-turbo generated (test)

67.65AccuracyGigaCheck (DN-DAB-DETR)

Evaluation Results

Method	Links
GigaCheck (DN-DAB-DETR)2024.10		67.65	88.98	1.03
ROBERTa + SEP2023.11		54.61	79.03	3.06
RoBERTa + SEP2024.10		54.61	79.03	3.06
Phi-1.5 Perpl. + GB classifier2023.11		50.8	74.1	4.59
Phi-2 Perpl. + GB classifier2023.11		48.4	70.3	5.39
Phi-1.5 Perpl. + LR classifier2023.11		47.3	72.7	4.77
Phi-1.52024.10		47.3	72.7	4.77
Phi-2 Perpl. + LR classifier2023.11		44.4	71.5	5.42
LLaMA-2-7B Perpl. + GB classifier2023.11		41.5	65	6.32
ROBERTa2023.11		39.01	75.18	3.15
RoBERTa2024.10		39.01	75.18	3.15
Phi-1 Perpl. + GB classifier2023.11		36.5	56	9.13
GPT-2 Perpl. + GB classifier2023.11		34.94	59.8	7.46
GPT-2 Perpl. + LR classifier2023.11		33.5	57.56	9.25
Phi-1.5 Perpl. + GB regressor2023.11		32	71	3.07
Phi-1.52024.10		32	71	3.07
Phi-1 Perpl. + LR classifier2023.11		31	53.9	9.76
Phi-2 Perpl. + GB regressor2023.11		30.8	68.7	3.33
LLaMA-2-7B Perpl. + LR classifier2023.11		29.7	58.4	7.9
Length + GB2023.11		25.72	46.18	18.99
LLaMA-2-7B Perpl. + GB regressor2023.11		23.6	60.3	3.89
DetectGPT + GB classifier2023.11		21.69	43.52	6.87
DetectGPT2024.10		21.69	43.52	6.87
TLE + TS Binary2023.11		20.02	34.58	18.52
TLE + TS Binary2024.10		20.02	34.58	18.52
GPT-2 Perpl. + GB regressor2023.11		19.74	54.03	4.89
Phi-1 Perpl. + GB regressor2023.11		17.4	49.9	5.51
PHD + TS ML2023.11		17.29	35.81	14.45
PHD + TS ML2024.10		17.29	35.81	14.45
DetectGPT + LR classifier2023.11		15.35	41.43	7.22
DetectGPT2024.10		15.35	41.43	7.22
Majority class2023.11		13.83	24.42	26.46
gpt-42023.11		9.1	25	24.1
gpt-3.5-turbo2023.11		7.1	23.9	25.5