Machine-Generated Text Detection on DetectRL Llama-2-70b (test)
90.2Detection Score (Llama-2-70b)Log-Rank-M
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Log-Rank-Mapplied_calibration=Markov-informed score calibration2026.02 | 90.2 | 75.32 | 84.34 | 83.29 | |
| Likelihood-Mapplied_calibration=Markov-informed score calibration2026.02 | 87.61 | 73.7 | 81.21 | 80.84 | |
| Log-Rankapplied_calibration=None2026.02 | 79.67 | 65.85 | 70.66 | 72.06 | |
| DetectGPT-Mapplied_calibration=Markov-informed score calibration2026.02 | 78.81 | 64.15 | 75.9 | 72.95 | |
| Likelihoodapplied_calibration=None2026.02 | 78.58 | 66.61 | 71.42 | 72.2 | |
| DNA-GPT-Mapplied_calibration=Markov-informed score calibration2026.02 | 74.39 | 64.36 | 70.5 | 69.75 | |
| DNA-GPTapplied_calibration=None2026.02 | 71.91 | 64.14 | 67.32 | 67.79 | |
| Entropy-Mapplied_calibration=Markov-informed score calibration2026.02 | 69.97 | 65.26 | 65.82 | 67.02 | |
| FastGPTapplied_calibration=None2026.02 | 67.72 | 58.5 | 56.68 | 60.97 | |
| FastGPT-Mapplied_calibration=Markov-informed score calibration2026.02 | 66.55 | 55.19 | 63.3 | 61.68 | |
| Entropyapplied_calibration=None2026.02 | 66.21 | 63.09 | 60.73 | 63.34 | |
| DetectGPTapplied_calibration=None2026.02 | 52.37 | 50.22 | 43.19 | 48.6 |