EN · PII
dialogpii-en PII detection benchmark results
A frozen evaluation slice with 147 rows, 3,087 normalized gold annotations and source-level provenance.
147input rows
3,087gold annotations
569,190characters
55measured detectors
Dataset provenance
- Language group
- EN
- Task
- PII
- License
- cc-by-4.0
- Data categories
- ADDRESS, CONTACT, ID, NET, ORG, PERSON
- 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 ↑ |
|---|---|---|---|---|---|
| gliner2-fastino | GLINER2 | 10.14% | 81.92% | 10.71% | 0.602 |
| gliner-nvidia+sent300 | GLINER | 13.28% | 77.52% | 11.20% | 0.587 |
| gliner2-large | GLINER2 | 14.16% | 78.00% | 9.92% | 0.571 |
| nuner-zero | GLINER | 14.64% | 80.24% | 12.29% | 0.535 |
| gliner2-hivetrace-omni | GLINER2 | 16.78% | 75.25% | 6.73% | 0.691 |
| stanza-ru | STANZA | 17.01% | 78.81% | 40.89% | 0.187 |
| bardsai-eu | ONNX | 17.07% | 71.66% | 5.86% | 0.799 |
| gliner-nvidia+ov100 | GLINER | 17.20% | 73.31% | 8.64% | 0.635 |
| gliner-nvidia | GLINER | 17.43% | 73.28% | 8.38% | 0.608 |
| gliner-multi-v21 | GLINER | 17.62% | 75.06% | 12.48% | 0.531 |
| gliner-urchade | GLINER | 18.46% | 75.38% | 13.79% | 0.507 |
| gliner-pii-base | GLINER | 19.99% | 71.59% | 5.15% | 0.643 |
| gliner-pii-edge | GLINER | 20.21% | 71.17% | 2.52% | 0.723 |
| openmed-multilingual | HF | 21.38% | 56.24% | 6.43% | 0.603 |
| nym-small | ONNX | 22.03% | 58.28% | 3.83% | 0.757 |
| nym-base+sent300 | HF | 22.51% | 58.31% | 3.38% | 0.754 |
| gliner25-fastino+ov100 | GLINER2 | 22.74% | 70.10% | 6.65% | 0.674 |
| davlan-xlmr | HF | 23.26% | 66.41% | 2.33% | 0.780 |
| nym-base | HF | 23.29% | 57.60% | 3.38% | 0.750 |
| gliner25-fastino+sent300 | GLINER2 | 23.71% | 68.90% | 9.36% | 0.611 |
| gliner25-fastino | GLINER2 | 24.04% | 69.00% | 8.12% | 0.655 |
| pplx | PPLX | 24.10% | 71.01% | 5.64% | 0.761 |
| pplx+sent300 | PPLX | 24.17% | 69.97% | 5.37% | 0.762 |
| pplx+ov100 | PPLX | 24.36% | 71.36% | 6.13% | 0.756 |
| nym-base+ov100 | HF | 24.43% | 56.59% | 3.31% | 0.738 |
| gravitee-small | HF | 25.79% | 65.95% | 4.06% | 0.707 |
| gliner2-hivetrace-uni | GLINER2 | 26.47% | 64.20% | 5.04% | 0.731 |
| davlan-mbert | HF | 27.08% | 62.97% | 2.33% | 0.759 |
| openmed-nemotron | HF | 29.35% | 51.15% | 8.76% | 0.523 |
| kalyan-ettin | HF | 31.29% | 50.31% | 5.45% | 0.616 |
| gliner25-fastino+nochunk | GLINER2 | 32.26% | 60.87% | 3.19% | 0.758 |
| ner-ru-yqelz | HF | 33.24% | 60.12% | 4.06% | 0.665 |
| ner-ru-gherman | HF | 34.11% | 46.10% | 2.07% | 0.634 |
| mmbert32k | HF | 35.47% | 46.65% | 4.85% | 0.543 |
| gliner2-vladlinv | GLINER2 | 38.19% | 55.88% | 3.80% | 0.633 |
| gliner-stream-pii | GLINER | 39.94% | 53.35% | 1.99% | 0.685 |
| ru-pii-ner | RUPII | 42.05% | 53.74% | 8.08% | 0.476 |
| openai-base | OPF | 42.40% | 54.13% | 12.89% | 0.415 |
| apararti | OPF | 43.21% | 51.96% | 9.13% | 0.455 |
| opf-ru | OPF | 43.70% | 45.19% | 10.26% | 0.392 |
| opf-kz-ru | OPF | 51.64% | 43.34% | 6.39% | 0.443 |
| traciora | OPF | 53.61% | 40.30% | 4.43% | 0.580 |
| mmbert32k+nochunk | HF | 58.47% | 26.50% | 3.53% | 0.487 |
| opf-ru-v2+ov100 | OPF | 61.68% | 33.56% | 5.41% | 0.480 |
| opf-ru-v2+sent300 | OPF | 62.36% | 32.85% | 4.36% | 0.483 |
| opf-ru-v2 | OPF | 62.71% | 32.30% | 5.04% | 0.483 |
| pii-shield-onnx | ONNX | 72.63% | 17.65% | 7.48% | 0.321 |
| ru-legal-ner+sent300 | HF | 72.69% | 16.65% | 2.33% | 0.344 |
| ru-legal-ner+ov100 | HF | 83.67% | 7.52% | 0.75% | 0.267 |
| ru-legal-ner | HF | 85.26% | 6.51% | 1.47% | 0.244 |
| spacy-ru-lg | SPACY | 85.65% | 13.74% | 1.58% | 0.180 |
| natasha | NATASHA | 89.99% | 9.59% | 0.60% | 0.124 |
| rules-ru | RULES | 96.44% | 0.32% | 0.00% | 0.096 |
| fef2-secret-ru | HF | 98.22% | 1.17% | 0.00% | 0.038 |
| spacy-alrosait | SPACY | 99.87% | 0.13% | 0.00% | 0.002 |