RU · PII
corrupt-hivetrace PII detection benchmark results
A frozen evaluation slice with 1,810 rows, 1,667 normalized gold annotations and source-level provenance.
1,810input rows
1,667gold annotations
164,420characters
56measured detectors
Dataset provenance
- Language group
- RU
- Task
- PII
- License
- apache-2.0
- Data categories
- ADDRESS, CONTACT, ID, PERSON, SECRET
- Project generated
- No
- Corrupted copy
- Yes
Rows without annotations are not verified clean. Reported extra masking is not a human-confirmed false-positive rate.
Detector outcomes
Sorted by fully hidden annotations. Training-source overlaps are excluded.
| Detector | Family | Fully hidden ↑ | Detected / any overlap ↑ | Extra masking ↓ | Char F1 ↑ |
|---|---|---|---|---|---|
| pplx+homoglyph | PPLX | 93.82% | 96.52% | 1.45% | 0.905 |
| pplx | PPLX | 93.70% | 96.52% | 1.51% | 0.904 |
| ru-pii-ner | RUPII | 83.32% | 88.84% | 0.71% | 0.854 |
| apararti | OPF | 80.38% | 90.16% | 4.87% | 0.862 |
| gliner2-fastino-ru | GLINER2 | 79.66% | 85.78% | 0.27% | 0.788 |
| traciora | OPF | 74.21% | 89.08% | 11.17% | 0.823 |
| gliner2-hivetrace-omni-ru | GLINER2 | 73.79% | 81.76% | 1.26% | 0.741 |
| opf-kz-ru | OPF | 73.25% | 87.04% | 2.52% | 0.833 |
| nym-base | HF | 71.51% | 97.18% | 0.49% | 0.931 |
| pii-shield-onnx | ONNX | 71.33% | 87.88% | 15.27% | 0.825 |
| nym-base+homoglyph | HF | 71.33% | 97.36% | 0.40% | 0.934 |
| gliner-nvidia-ru | GLINER | 71.15% | 84.22% | 1.88% | 0.811 |
| gliner-urchade-ru | GLINER | 69.17% | 72.89% | 0.53% | 0.733 |
| ru-legal-ner | HF | 68.87% | 88.54% | 22.36% | 0.747 |
| gliner-nvidia+homoglyph | GLINER | 68.75% | 79.30% | 6.22% | 0.733 |
| opf-ru-v2+homoglyph | OPF | 68.69% | 87.10% | 1.44% | 0.873 |
| openai-base | OPF | 68.63% | 77.50% | 2.48% | 0.822 |
| ru-legal-ner+homoglyph | HF | 68.63% | 88.30% | 18.65% | 0.756 |
| gliner-nvidia | GLINER | 68.39% | 78.64% | 5.96% | 0.728 |
| gliner2-vladlinv-ru | GLINER2 | 68.33% | 73.13% | 0.46% | 0.765 |
| opf-ru-v2 | OPF | 68.27% | 87.22% | 1.65% | 0.874 |
| nuner-zero | GLINER | 67.25% | 84.28% | 2.07% | 0.771 |
| gliner2-fastino | GLINER2 | 64.43% | 69.71% | 0.95% | 0.692 |
| opf-ru | OPF | 63.53% | 91.24% | 9.48% | 0.829 |
| gliner-pii-edge | GLINER | 63.23% | 74.63% | 7.20% | 0.650 |
| nym-small | ONNX | 62.87% | 95.08% | 1.68% | 0.901 |
| gliner-urchade | GLINER | 62.57% | 66.71% | 2.01% | 0.687 |
| gravitee-small | HF | 62.57% | 74.93% | 11.52% | 0.543 |
| gliner25-fastino-ru | GLINER2 | 62.15% | 68.99% | 0.43% | 0.717 |
| mmbert32k | HF | 61.79% | 95.02% | 4.08% | 0.853 |
| gliner2-vladlinv+homoglyph | GLINER2 | 60.77% | 65.51% | 0.44% | 0.723 |
| gliner2-vladlinv | GLINER2 | 60.59% | 65.27% | 0.46% | 0.721 |
| bardsai-eu | ONNX | 59.51% | 80.26% | 0.25% | 0.851 |
| gliner2-hivetrace-omni | GLINER2 | 57.71% | 63.23% | 1.13% | 0.684 |
| openmed-multilingual | HF | 53.75% | 84.04% | 3.69% | 0.783 |
| gliner2-hivetrace-uni | GLINER2 | 50.69% | 64.73% | 4.19% | 0.576 |
| gliner2-large | GLINER2 | 49.85% | 59.93% | 2.50% | 0.588 |
| gliner25-fastino | GLINER2 | 47.21% | 53.51% | 0.98% | 0.636 |
| fef2-secret-ru | HF | 45.53% | 59.87% | 0.89% | 0.683 |
| gliner-stream-pii | GLINER | 41.81% | 70.85% | 1.62% | 0.677 |
| rules-ru | RULES | 41.57% | 45.77% | 0.00% | 0.564 |
| kalyan-ettin | HF | 37.67% | 66.41% | 2.12% | 0.725 |
| openmed-nemotron | HF | 32.09% | 62.45% | 11.59% | 0.594 |
| gliner-multi-v21-ru | GLINER | 31.43% | 47.69% | 0.71% | 0.490 |
| gliner-pii-base | GLINER | 29.57% | 31.85% | 0.78% | 0.477 |
| gliner-multi-v21 | GLINER | 28.01% | 44.15% | 2.74% | 0.454 |
| gliner2-hivetrace-uni-ru | GLINER2 | 21.54% | 26.09% | 1.83% | 0.279 |
| stanza-ru | STANZA | 21.30% | 32.75% | 13.55% | 0.322 |
| ner-ru-yqelz | HF | 19.56% | 33.23% | 5.19% | 0.352 |
| davlan-xlmr | HF | 12.54% | 22.74% | 0.75% | 0.320 |
| spacy-ru-lg | SPACY | 12.42% | 21.12% | 8.16% | 0.235 |
| davlan-mbert | HF | 11.58% | 22.44% | 0.57% | 0.291 |
| natasha | NATASHA | 9.06% | 16.80% | 2.11% | 0.185 |
| spacy-alrosait | SPACY | 6.78% | 12.42% | 0.22% | 0.219 |
| ner-ru-gherman | HF | 1.44% | 27.65% | 0.38% | 0.376 |
| ner-ru-gherman+homoglyph | HF | 1.38% | 28.43% | 0.37% | 0.386 |