GLINER2 detector
gliner2-vladlinv PII detection benchmark
Measured on 41 of 41 retained datasets under the same frozen scoring and masking protocol.
46.09%fully hidden annotations
49.61%detected / overlapped
0.91%extra masking
690CPU characters / second
Strengths
Highest complete-masking rates by normalized data type.
Phone numbers & email82.75%18,190 / 21,981
People's names78.10%51,340 / 65,736
Dates & times71.28%3,922 / 5,502
Watch list
Lowest observed complete-masking rates; inspect before deployment.
Network identifiers1.33%341 / 25,631
Organizations1.61%310 / 19,274
Other sensitive attributes3.74%120 / 3,207
Language coverage
The overall leader can differ from the best choice for one language slice.
| Language | Datasets | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| English | 17 | 51.57% | 53.95% | 0.93% |
| Russian | 19 | 43.18% | 47.52% | 0.89% |
| Multilingual | 5 | 49.64% | 51.89% | 0.18% |
Configuration, source and version
- Family
- GLINER2
- Revision
- d4c05667b10d8e8db0f0626934135731d0530983
- Eligible datasets
- 41 / 41
- Character F1
- 0.572
- Untouched diagnostic
- 50.39% · 114,622 / 227,466
- Benchmark version
- 1.0.3 · 2026-09-09
- CPU seconds / 10k chars
- 14.493
- GPU seconds / 10k chars
- 0.173
Results by dataset
Complete masking is primary; detection shows whether any character overlapped.
| Dataset | Gold annotations | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| corrupt-secrets-issues | 288 | 4.17% | 9.72% | 0.06% |
| secrets-issues | 288 | 6.60% | 9.72% | 0.04% |
| rubai-ru | 3,458 | 6.82% | 80.19% | — |
| leak-museum | 101 | 8.91% | 10.89% | 0.00% |
| leaky-repo | 95 | 17.89% | 18.95% | 0.00% |
| tab-echr | 3,830 | 19.14% | 31.02% | — |
| synth-secrets-en | 581 | 25.13% | 27.02% | 3.28% |
| alexen2 | 1,261 | 26.33% | 100.00% | 0.00% |
| synth-ru-tickets | 32,099 | 26.50% | 28.23% | — |
| synth-secrets-ru | 581 | 28.06% | 30.64% | 3.43% |
| synth-env-configs | 392 | 30.36% | 32.14% | 0.45% |
| multiconer-ru | 1,208 | 32.04% | 39.40% | 3.65% |
| privy | 1,899 | 33.33% | 33.81% | 1.63% |
| factrueval | 7,966 | 33.42% | 33.54% | — |
| nerel | 24,369 | 35.96% | 37.38% | — |
| ameau01 | 2,242 | 39.12% | 39.16% | 0.05% |
| synth-jira-comments | 2,705 | 39.33% | 39.93% | — |
| secrets-rules | 746 | 41.02% | 44.24% | 10.30% |
| corrupt-redmadrobot | 5,531 | 41.31% | 42.27% | 0.10% |
| gretel-multi | 4,484 | 45.00% | 49.42% | 0.18% |
| scanpatch | 8,708 | 46.41% | 48.11% | 0.57% |
| dialogpii-multi | 10,345 | 47.13% | 52.69% | — |
| redact-multi | 22,864 | 49.27% | 50.86% | — |
| redmadrobot | 5,516 | 49.31% | 49.91% | 0.10% |
| redact-ru | 9,230 | 49.64% | 51.65% | 3.76% |
| kiji-multi | 7,565 | 52.62% | 52.99% | — |
| nym-multi | 8,169 | 53.63% | 54.09% | 0.00% |
| kiji-en | 7,623 | 54.87% | 55.21% | — |
| dialogpii-en | 3,087 | 55.88% | 61.81% | 3.80% |
| creddata | 776 | 56.31% | 59.41% | 1.37% |
| nym-en | 3,584 | 60.18% | 61.19% | 1.13% |
| corrupt-hivetrace | 1,667 | 60.59% | 65.27% | 0.46% |
| nemotron-pii | 9,391 | 63.49% | 63.91% | 0.36% |
| jayguard | 1,195 | 64.52% | 74.81% | 0.71% |
| nym-ru | 9,497 | 64.75% | 65.52% | — |
| hivetrace | 1,667 | 68.87% | 70.79% | 0.17% |
| synth-wiki-tables | 13,094 | 69.10% | 73.45% | — |
| russian-pii-66k | 4,805 | 69.80% | 69.89% | — |
| tonicai | 2,417 | 74.27% | 76.42% | 0.42% |
| alrosait | 1,862 | 90.71% | 97.15% | 1.59% |
| arthur-passwords | 280 | 96.07% | 96.07% | 6.93% |