PII Arena
HF detector

ner-ru-gherman PII detection benchmark

Measured on 41 of 41 retained datasets under the same frozen scoring and masking protocol.

27.63%fully hidden annotations
48.78%detected / overlapped
0.28%extra masking
4,026CPU characters / second

Strengths

Highest complete-masking rates by normalized data type.

People's names58.61%38,527 / 65,736
Addresses & locations52.98%20,980 / 39,602
Logins & usernames20.55%2,339 / 11,382

Watch list

Lowest observed complete-masking rates; inspect before deployment.

Network identifiers0.00%0 / 25,631
Dates & times0.02%1 / 5,502
Phone numbers & email0.03%7 / 21,981

Language coverage

The overall leader can differ from the best choice for one language slice.

LanguageDatasetsFully hidden ↑Detected ↑Extra masking ↓
English1731.65%43.80%0.18%
Russian1923.06%47.84%0.51%
Multilingual536.44%54.70%0.08%

Configuration, source and version

Family
HF
Revision
fc6b2c5a2c5d7c82da7416f7fd9c055159bbb984
Eligible datasets
41 / 41
Character F1
0.453
Untouched diagnostic
51.22% · 116,497 / 227,466
Benchmark version
1.0.3 · 2026-09-09
CPU seconds / 10k chars
2.484
GPU seconds / 10k chars
0.150

Results by dataset

Complete masking is primary; detection shows whether any character overlapped.

DatasetGold annotationsFully hidden ↑Detected ↑Extra masking ↓
corrupt-secrets-issues2880.00%0.00%0.10%
creddata7760.00%0.00%0.06%
leak-museum1010.00%0.00%0.09%
leaky-repo950.00%0.00%0.03%
secrets-issues2880.00%0.00%0.11%
synth-secrets-en5810.00%0.00%0.01%
synth-secrets-ru5810.00%0.00%0.00%
secrets-rules7460.13%0.13%0.13%
synth-env-configs3920.26%0.26%0.13%
hivetrace1,6670.90%30.29%0.11%
arthur-passwords2801.07%1.07%4.13%
corrupt-hivetrace1,6671.44%27.65%0.38%
rubai-ru3,4581.76%55.32%
alrosait1,8621.77%96.19%2.51%
synth-jira-comments2,7052.26%27.02%
synth-wiki-tables13,0943.89%39.12%
synth-ru-tickets32,0994.66%27.19%
alexen21,2614.76%52.18%9.53%
multiconer-ru1,20810.18%30.79%1.52%
ameau012,24210.48%29.79%0.07%
gretel-multi4,48413.20%52.45%0.08%
tab-echr3,83015.20%56.06%
privy1,89920.59%34.60%0.10%
nym-ru9,49725.38%37.85%
nym-multi8,16927.14%33.14%0.00%
russian-pii-66k4,80528.95%42.12%
nym-en3,58429.24%37.08%0.14%
corrupt-redmadrobot5,53130.65%39.63%0.27%
nemotron-pii9,39131.91%39.83%0.00%
redact-ru9,23032.86%60.46%3.79%
redmadrobot5,51634.97%44.29%0.19%
scanpatch8,70836.15%58.41%0.51%
redact-multi22,86436.48%60.13%
jayguard1,19537.15%79.00%0.88%
tonicai2,41738.85%49.86%0.08%
nerel24,36939.95%72.24%
dialogpii-multi10,34540.02%56.25%
dialogpii-en3,08746.10%65.89%2.07%
kiji-multi7,56555.27%60.81%
kiji-en7,62356.25%61.56%
factrueval7,96666.43%69.23%