RU · PII
russian-pii-66k PII detection benchmark results
A frozen evaluation slice with 1,500 rows, 4,805 normalized gold annotations and source-level provenance.
1,500input rows
4,805gold annotations
195,840characters
50measured detectors
Dataset provenance
- Language group
- RU
- Task
- PII
- License
- not specified
- Data categories
- ACCOUNT, ADDRESS, CONTACT, DATE, ID, PERSON, SECRET
- Project generated
- No
- Corrupted copy
- No
Rows without annotations are not verified clean. Reported extra masking is not a human-confirmed false-positive rate.
Detector outcomes
Sorted by untouched annotations. Training-source overlaps are excluded.
| Detector | Family | Untouched ↓ | Fully hidden ↑ | Extra masking ↓ | Char F1 ↑ |
|---|---|---|---|---|---|
| nym-base | HF | 0.06% | 99.44% | — | 0.993 |
| pplx | PPLX | 0.23% | 97.94% | — | 0.955 |
| nym-small | ONNX | 0.29% | 99.42% | — | 0.991 |
| mmbert32k | HF | 2.64% | 75.96% | — | 0.942 |
| bardsai-eu | ONNX | 3.43% | 81.58% | — | 0.930 |
| gliner-nvidia-ru | GLINER | 3.50% | 91.95% | — | 0.961 |
| ru-pii-ner | RUPII | 3.64% | 96.23% | — | 0.958 |
| gliner-nvidia | GLINER | 3.93% | 91.09% | — | 0.936 |
| gliner2-fastino | GLINER2 | 5.06% | 92.26% | — | 0.965 |
| gliner2-fastino-ru | GLINER2 | 5.16% | 92.07% | — | 0.960 |
| ru-legal-ner | HF | 5.97% | 86.72% | — | 0.939 |
| gliner2-hivetrace-omni-ru | GLINER2 | 6.12% | 87.49% | — | 0.950 |
| gliner2-hivetrace-omni | GLINER2 | 6.60% | 86.85% | — | 0.951 |
| openai-base | OPF | 7.80% | 89.82% | — | 0.901 |
| apararti | OPF | 8.95% | 88.53% | — | 0.902 |
| traciora | OPF | 10.72% | 86.24% | — | 0.906 |
| openmed-multilingual | HF | 11.95% | 80.17% | — | 0.934 |
| opf-ru-v2 | OPF | 12.78% | 85.49% | — | 0.910 |
| nuner-zero | GLINER | 13.42% | 82.14% | — | 0.839 |
| opf-kz-ru | OPF | 13.44% | 79.65% | — | 0.874 |
| pii-shield-onnx | ONNX | 14.36% | 78.88% | — | 0.922 |
| fef2-secret-ru | HF | 18.58% | 67.35% | — | 0.867 |
| openmed-nemotron | HF | 20.23% | 66.53% | — | 0.790 |
| gliner2-vladlinv-ru | GLINER2 | 20.94% | 78.96% | — | 0.878 |
| gliner-stream-pii | GLINER | 21.12% | 64.60% | — | 0.816 |
| gliner-pii-edge | GLINER | 21.21% | 73.92% | — | 0.844 |
| gliner25-fastino-ru | GLINER2 | 22.21% | 75.48% | — | 0.858 |
| gliner2-large | GLINER2 | 22.50% | 76.15% | — | 0.892 |
| kalyan-ettin | HF | 23.16% | 61.79% | — | 0.820 |
| gliner-multi-v21-ru | GLINER | 28.41% | 60.69% | — | 0.704 |
| gliner-urchade | GLINER | 28.51% | 70.36% | — | 0.876 |
| gliner2-vladlinv | GLINER2 | 30.11% | 69.80% | — | 0.824 |
| gravitee-small | HF | 30.41% | 62.96% | — | 0.651 |
| gliner-urchade-ru | GLINER | 31.16% | 67.99% | — | 0.858 |
| gliner2-hivetrace-uni | GLINER2 | 32.17% | 62.60% | — | 0.863 |
| gliner25-fastino | GLINER2 | 33.76% | 63.89% | — | 0.787 |
| gliner-multi-v21 | GLINER | 34.80% | 54.80% | — | 0.653 |
| stanza-ru | STANZA | 50.43% | 39.13% | — | 0.551 |
| gliner-pii-base | GLINER | 57.42% | 39.19% | — | 0.627 |
| ner-ru-gherman | HF | 57.88% | 28.95% | — | 0.409 |
| ner-ru-yqelz | HF | 59.40% | 32.47% | — | 0.432 |
| davlan-xlmr | HF | 61.02% | 35.46% | — | 0.403 |
| spacy-ru-lg | SPACY | 62.16% | 29.30% | — | 0.436 |
| spacy-alrosait | SPACY | 62.98% | 36.13% | — | 0.373 |
| davlan-mbert | HF | 63.73% | 31.26% | — | 0.383 |
| gliner2-hivetrace-uni-ru | GLINER2 | 64.54% | 33.88% | — | 0.415 |
| rules-ru | RULES | 66.14% | 33.63% | — | 0.623 |
| natasha | NATASHA | 69.89% | 23.56% | — | 0.298 |
| betterleaks | LEAKS | 100.00% | 0.00% | — | 0.000 |
| gitleaks | LEAKS | 100.00% | 0.00% | — | 0.000 |