Text Classification on MIND
77.43AccuracyWARDEN
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| WARDENAttack=Dimension-shift Attack2026.04 | 77.43 | — | 10 | |
| WARDENAttack=No Attack2026.04 | 77.39 | — | 10 | |
| WARDENAttack=Dim2025.12 | 77.33 | — | — | |
| OriginalAttack=None2026.04 | 77.33 | — | 0.99 | |
| SemMarkAttack=Dim2025.12 | 77.29 | — | — | |
| EmbMarkerAttack=No Attack2025.12 | 77.26 | — | — | |
| EspeWAttack=Dim2025.12 | 77.25 | — | — | |
| SemMarkAttack=No Attack2025.12 | 77.25 | — | — | |
| WARDENAttack=Dimension-reduction Attack2026.04 | 77.25 | — | 10 | |
| EmbMarkerAttack=Dim2025.12 | 77.24 | — | — | |
| EspeWAttack=Dimension-shift Attack2026.04 | 77.24 | — | 10 | |
| OriginalDescription=Benign victim model2024.03 | 77.22 | 51.37 | — | |
| EmbMarkerAttack=Detect2025.12 | 77.22 | — | — | |
| GeoMarkAttack=Dimension-shift Attack2026.04 | 77.22 | — | 10 | |
| OriginalAttack=None2025.12 | 77.21 | — | — | |
| WARDENAttack=No Attack2025.12 | 77.21 | — | — | |
| GeoMarkAttack=No Attack2026.04 | 77.21 | — | 10 | |
| WARDENAttack=Detect2025.12 | 77.2 | — | — | |
| EmbMarkerDescription=Watermarked model2024.03 | 77.19 | 51.4 | — | |
| EspeWAttack=Detect2025.12 | 77.18 | — | — | |
| EspeWAttack=No Attack2025.12 | 77.17 | — | — | |
| EspeWAttack=No Attack2026.04 | 77.17 | — | 10 | |
| SemMarkAttack=Detect2025.12 | 77.09 | — | — | |
| GeoMarkAttack=Dimension-reduction Attack2026.04 | 77.06 | — | 10 | |
| EspeWAttack=Dimension-reduction Attack2026.04 | 77 | — | 10 | |
| WARDENAttack=Paraphrasing Attack (GPT-4o-mini)2026.04 | 76.93 | — | 10 | |
| WETAttack=Dimension-shift Attack2026.04 | 76.91 | — | 0.72 | |
| WETAttack=No Attack2025.12 | 76.89 | — | — | |
| EspeWAttack=Paraphrasing Attack (GPT-4o-mini)2026.04 | 76.88 | — | 0.002 | |
| GeoMarkAttack=Paraphrasing Attack (GPT-4o-mini)2026.04 | 76.85 | — | 10 | |
| WETAttack=Detect2025.12 | 76.83 | — | — | |
| WETAttack=No Attack2026.04 | 76.83 | — | 10 | |
| WETAttack=Paraphrasing Attack (GPT-4o-mini)2026.04 | 76.68 | — | 10 | |
| WARDENAttack=Paraphrasing Attack (NLLB)2026.04 | 76.58 | — | 10 | |
| EspeWAttack=Paraphrasing Attack (NLLB)2026.04 | 76.57 | — | 0.002 | |
| GeoMarkAttack=Paraphrasing Attack (NLLB)2026.04 | 76.49 | — | 10 | |
| WETAttack=Paraphrasing Attack (NLLB)2026.04 | 76.25 | — | 10 | |
| WETAttack=Dim2025.12 | 75.94 | — | — | |
| WARDENAttack=CSE Attack2026.04 | 75.69 | — | 0.05 | |
| EspeWAttack=CSE Attack2026.04 | 75.59 | — | 10 | |
| CSEDescription=CSE performed on EmbMarker embeddings2024.03 | 75.51 | 50.35 | — | |
| SemMarkAttack=CSE2025.12 | 75.49 | — | — | |
| EmbMarkerAttack=CSE2025.12 | 75.48 | — | — | |
| EspeWAttack=CSE2025.12 | 75.48 | — | — | |
| WETAttack=CSE2025.12 | 75.37 | — | — | |
| WARDENAttack=CSE2025.12 | 75.28 | — | — | |
| WETAttack=CSE Attack2026.04 | 75.12 | — | 10 | |
| GeoMarkAttack=CSE Attack2026.04 | 74.86 | — | 10 |