GLINER detector
gliner-urchade-ru PII detection benchmark
Measured on 19 of 41 retained datasets under the same frozen scoring and masking protocol.
45.35%fully hidden annotations
54.49%detected / overlapped
1.30%extra masking
1,608CPU characters / second
Strengths
Highest complete-masking rates by normalized data type.
Phone numbers & email74.11%10,655 / 14,377
Documents & identifiers71.48%6,218 / 8,699
Organizations66.90%7,564 / 11,306
Watch list
Lowest observed complete-masking rates; inspect before deployment.
People's names8.53%2,894 / 33,934
Network identifiers35.30%8,151 / 23,092
Other sensitive attributes38.19%425 / 1,113
Language coverage
The overall leader can differ from the best choice for one language slice.
| Language | Datasets | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| Russian | 19 | 45.35% | 54.49% | 1.30% |
Configuration, source and version
- Family
- GLINER
- Revision
- 1fcf13e85f4eef5394e1fcd406cf2ca9ea82351d
- Eligible datasets
- 19 / 41
- Character F1
- 0.637
- Untouched diagnostic
- 45.51% · 62,081 / 136,419
- Benchmark version
- 1.0.3 · 2026-09-09
- CPU seconds / 10k chars
- 6.219
- GPU seconds / 10k chars
- Not measured
Russian labels
Results by dataset
Complete masking is primary; detection shows whether any character overlapped.
| Dataset | Gold annotations | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| rubai-ru | 3,458 | 3.09% | 67.84% | — |
| alexen2 | 1,261 | 9.04% | 72.01% | 2.00% |
| synth-secrets-ru | 581 | 20.48% | 21.34% | 0.91% |
| jayguard | 1,195 | 33.56% | 34.39% | 0.43% |
| synth-jira-comments | 2,705 | 33.79% | 34.57% | — |
| factrueval | 7,966 | 37.13% | 37.56% | — |
| redact-ru | 9,230 | 37.45% | 43.49% | 10.69% |
| synth-ru-tickets | 32,099 | 43.32% | 63.72% | — |
| nerel | 24,369 | 43.42% | 46.51% | — |
| alrosait | 1,862 | 46.72% | 47.69% | 2.99% |
| scanpatch | 8,708 | 47.00% | 52.18% | 5.52% |
| multiconer-ru | 1,208 | 48.76% | 54.80% | 6.72% |
| nym-ru | 9,497 | 51.69% | 52.88% | — |
| synth-wiki-tables | 13,094 | 52.40% | 56.74% | — |
| corrupt-redmadrobot | 5,531 | 54.22% | 56.10% | 0.76% |
| redmadrobot | 5,516 | 59.97% | 61.02% | 1.09% |
| russian-pii-66k | 4,805 | 67.99% | 68.84% | — |
| corrupt-hivetrace | 1,667 | 69.17% | 72.89% | 0.53% |
| hivetrace | 1,667 | 76.06% | 77.08% | 0.58% |