MULTI · PII
gretel-multi PII detection benchmark results
A frozen evaluation slice with 697 rows, 4,484 normalized gold annotations and source-level provenance.
697input rows
4,484gold annotations
916,566characters
38measured detectors
Dataset provenance
- Language group
- MULTI
- Task
- PII
- License
- apache-2.0
- Data categories
- ACCOUNT, ADDRESS, CONTACT, DATE, ID, NET, ORG, 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 ↑ |
|---|---|---|---|---|---|
| gravitee-small | HF | 16.24% | 79.66% | 2.73% | 0.716 |
| gliner-nvidia | GLINER | 16.93% | 50.87% | 2.52% | 0.707 |
| gliner2-fastino | GLINER2 | 17.33% | 74.49% | 0.75% | 0.765 |
| nym-small | ONNX | 17.33% | 46.83% | 1.68% | 0.709 |
| nym-base | HF | 17.75% | 45.63% | 2.02% | 0.728 |
| pplx | PPLX | 18.24% | 79.55% | 1.45% | 0.720 |
| bardsai-eu | ONNX | 18.49% | 66.57% | 2.09% | 0.752 |
| kalyan-ettin | HF | 20.29% | 43.09% | 2.32% | 0.706 |
| openmed-nemotron | HF | 21.14% | 40.63% | 2.67% | 0.641 |
| gliner2-large | GLINER2 | 22.32% | 65.77% | 3.06% | 0.678 |
| mmbert32k | HF | 22.70% | 56.31% | 2.93% | 0.645 |
| gliner25-fastino | GLINER2 | 23.39% | 70.56% | 0.76% | 0.784 |
| gliner-urchade | GLINER | 23.91% | 73.75% | 2.79% | 0.694 |
| opf-ru | OPF | 25.89% | 51.90% | 0.55% | 0.699 |
| stanza-ru | STANZA | 26.00% | 56.67% | 20.56% | 0.262 |
| nuner-zero | GLINER | 26.43% | 69.11% | 2.32% | 0.720 |
| apararti | OPF | 26.81% | 65.66% | 1.54% | 0.705 |
| gliner-pii-edge | GLINER | 27.83% | 55.71% | 1.29% | 0.646 |
| openai-base | OPF | 28.81% | 64.74% | 1.44% | 0.716 |
| gliner-stream-pii | GLINER | 29.19% | 41.86% | 0.45% | 0.723 |
| opf-kz-ru | OPF | 29.75% | 61.71% | 1.54% | 0.689 |
| traciora | OPF | 35.53% | 54.35% | 0.10% | 0.757 |
| gliner-multi-v21 | GLINER | 36.69% | 53.86% | 1.93% | 0.607 |
| opf-ru-v2 | OPF | 37.42% | 52.65% | 0.12% | 0.747 |
| davlan-xlmr | HF | 37.56% | 44.96% | 0.41% | 0.658 |
| davlan-mbert | HF | 39.34% | 42.28% | 0.29% | 0.642 |
| ru-pii-ner | RUPII | 39.47% | 55.93% | 0.61% | 0.652 |
| ru-legal-ner | HF | 43.78% | 27.99% | 16.63% | 0.239 |
| pii-shield-onnx | ONNX | 46.12% | 34.72% | 15.13% | 0.286 |
| ner-ru-gherman | HF | 47.55% | 13.20% | 0.08% | 0.536 |
| gliner2-vladlinv | GLINER2 | 50.58% | 45.00% | 0.18% | 0.590 |
| ner-ru-yqelz | HF | 52.74% | 29.88% | 4.21% | 0.447 |
| gliner-pii-base | GLINER | 54.19% | 39.12% | 0.69% | 0.574 |
| rules-ru | RULES | 85.28% | 14.41% | 0.04% | 0.306 |
| spacy-ru-lg | SPACY | 86.20% | 7.83% | 1.56% | 0.117 |
| fef2-secret-ru | HF | 87.07% | 9.63% | 0.07% | 0.253 |
| natasha | NATASHA | 89.36% | 4.77% | 0.27% | 0.125 |
| spacy-alrosait | SPACY | 100.00% | 0.00% | 0.00% | 0.000 |