Binary Classification (Human vs Assistive) on CNN/DailyMail
0.99AUCRoBERTa
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RoBERTaGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.99 | 96 | |
| Role-Aware AIGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.99 | 94 | |
| Role-Aware AIGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.96 | 95 | |
| LogLikelihoodGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.85 | 77 | |
| LogRankGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.84 | 76 | |
| EntropyGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.75 | 67 | |
| RoBERTaGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.71 | 71 | |
| LogRankGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.67 | 61 | |
| CurvatureGenerator Model=LLaMA-3-Instruct, Generator Parameters=3 billion2026.05 | 0.63 | 62 | |
| EntropyGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.41 | 51 | |
| CurvatureGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.39 | 50 | |
| LogLikelihoodGenerator Model=GPT-2, Generator Parameters=124 million2026.05 | 0.14 | 51 |