Watermark Detection on HumanEval
85.9F1 ScoreEWD
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EWDScenario=Low-entropy, FPR threshold=Best2024.03 | 85.9 | — | — | |
| SWEETScenario=Low-entropy, FPR threshold=Best2024.03 | 84.2 | — | — | |
| EWDScenario=Low-entropy, FPR threshold=5%2024.03 | 79.7 | 69.2 | — | |
| SWEETScenario=Low-entropy, FPR threshold=5%2024.03 | 77.8 | 66.7 | — | |
| KGWScenario=Low-entropy, FPR threshold=Best2024.03 | 77.2 | — | — | |
| EWDScenario=Low-entropy, FPR threshold=1%2024.03 | 63.3 | 46.6 | — | |
| SWEETScenario=Low-entropy, FPR threshold=1%2024.03 | 62.2 | 45.5 | — | |
| KGWScenario=Low-entropy, FPR threshold=5%2024.03 | 56.7 | 41.4 | — | |
| KGWScenario=Low-entropy, FPR threshold=1%2024.03 | 49.4 | 33.1 | — | |
| EXP-editWatermarking Status=Watermarked2026.01 | — | — | 48.6 | |
| SWaRLWatermarking Status=Watermarked2026.01 | — | — | 72.5 | |
| SWEETWatermarking Status=Watermarked2026.01 | — | — | 79.8 | |
| WLLMWatermarking Status=Watermarked2026.01 | — | — | 77.1 |