AI-generated text detection on Long-form QA 46K ShareGPT-augmented corpus
100Detection Accuracy (1% FPR)SP (Retrieval over 46K)
Evaluation Results
| Method | Links | |
|---|---|---|
| SP (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=Original2023.03 | 100 | |
| SP (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=Original2023.03 | 100 | |
| SP (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=Original2023.03 | 100 | |
| BM25 (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=Original2023.03 | 100 | |
| BM25 (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=Original2023.03 | 100 | |
| BM25 (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=Original2023.03 | 100 | |
| BM25 (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=+ 60L2023.03 | 99 | |
| BM25 (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=+ 60L2023.03 | 98.9 | |
| BM25 (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=+ 60L2023.03 | 98.4 | |
| BM25 (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=+ 60 L,O2023.03 | 97.5 | |
| BM25 (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=+ 60 L,O2023.03 | 97.3 | |
| BM25 (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=+ 60 L,O2023.03 | 95.5 | |
| SP (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=+ 60L2023.03 | 94.2 | |
| SP (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=+ 60L2023.03 | 94.1 | |
| SP (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=+ 60L2023.03 | 94 | |
| SP (Retrieval over 46K)Generator Model=GPT-3.5 (davinci-003), Query Transformation=+ 60 L,O2023.03 | 84.9 | |
| SP (Retrieval over 46K)Generator Model=GPT2-XL, Query Transformation=+ 60 L,O2023.03 | 84.8 | |
| SP (Retrieval over 46K)Generator Model=OPT-13B, Query Transformation=+ 60 L,O2023.03 | 84.7 |