GLINER detector
gliner-nvidia-ru PII detection benchmark
Measured on 19 of 41 retained datasets under the same frozen scoring and masking protocol.
50.97%fully hidden annotations
60.95%detected / overlapped
1.80%extra masking
500CPU characters / second
Strengths
Highest complete-masking rates by normalized data type.
Phone numbers & email73.34%10,544 / 14,377
Documents & identifiers72.97%6,348 / 8,699
Dates & times67.92%1,791 / 2,637
Watch list
Lowest observed complete-masking rates; inspect before deployment.
Network identifiers28.88%6,669 / 23,092
Other sensitive attributes32.35%360 / 1,113
Customer & employee IDs33.03%1,333 / 4,036
Language coverage
The overall leader can differ from the best choice for one language slice.
| Language | Datasets | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| Russian | 19 | 50.97% | 60.95% | 1.80% |
Configuration, source and version
- Family
- GLINER
- Revision
- bd23e8ef4425fd04e34c5204ab49ffaa706eae79
- Eligible datasets
- 19 / 41
- Character F1
- 0.641
- Untouched diagnostic
- 39.05% · 53,269 / 136,419
- Benchmark version
- 1.0.3 · 2026-09-09
- CPU seconds / 10k chars
- 20.000
- GPU seconds / 10k chars
- Not measured
Russian labels; source overlap: nemotron-pii
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 | 6.04% | 76.08% | — |
| alexen2 | 1,261 | 23.47% | 98.02% | 8.63% |
| multiconer-ru | 1,208 | 34.60% | 48.68% | 6.46% |
| synth-ru-tickets | 32,099 | 35.18% | 38.94% | — |
| synth-wiki-tables | 13,094 | 41.10% | 47.92% | — |
| synth-secrets-ru | 581 | 42.34% | 43.03% | 3.52% |
| scanpatch | 8,708 | 49.48% | 66.59% | 1.64% |
| redact-ru | 9,230 | 50.07% | 72.15% | 9.12% |
| nerel | 24,369 | 51.71% | 61.55% | — |
| factrueval | 7,966 | 52.54% | 53.24% | — |
| synth-jira-comments | 2,705 | 55.45% | 61.55% | — |
| jayguard | 1,195 | 56.15% | 71.13% | 0.99% |
| alrosait | 1,862 | 60.58% | 82.98% | 3.84% |
| corrupt-redmadrobot | 5,531 | 65.32% | 69.70% | 1.25% |
| corrupt-hivetrace | 1,667 | 71.15% | 84.22% | 1.88% |
| redmadrobot | 5,516 | 77.61% | 80.71% | 0.83% |
| nym-ru | 9,497 | 80.73% | 84.32% | — |
| hivetrace | 1,667 | 90.64% | 93.52% | 0.75% |
| russian-pii-66k | 4,805 | 91.95% | 96.50% | — |