PII Arena
RUPII detector

ru-pii-ner PII detection benchmark

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

53.11%fully hidden annotations
57.30%detected / overlapped
2.73%extra masking
7,053CPU characters / second

Strengths

Highest complete-masking rates by normalized data type.

Dates & times84.22%4,634 / 5,502
Documents & identifiers81.82%12,301 / 15,034
Phone numbers & email80.97%17,797 / 21,981

Watch list

Lowest observed complete-masking rates; inspect before deployment.

Organizations2.95%568 / 19,274
Other sensitive attributes8.73%280 / 3,207
Network identifiers9.12%2,337 / 25,631

Language coverage

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

LanguageDatasetsFully hidden ↑Detected ↑Extra masking ↓
English1753.30%56.47%2.43%
Russian1953.34%58.32%3.50%
Multilingual552.39%55.30%0.66%

Configuration, source and version

Family
RUPII
Revision
3b509ebbcec894e522adef32560644017832deea
Eligible datasets
41 / 41
Character F1
0.599
Untouched diagnostic
42.70% · 97,121 / 227,466
Benchmark version
1.0.3 · 2026-09-09
CPU seconds / 10k chars
1.418
GPU seconds / 10k chars
Not measured

Results by dataset

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

DatasetGold annotationsFully hidden ↑Detected ↑Extra masking ↓
leaky-repo955.26%11.58%0.09%
secrets-issues2887.29%12.15%1.26%
rubai-ru3,4588.01%94.88%
secrets-rules7469.92%14.34%11.96%
synth-env-configs39210.20%12.24%4.32%
synth-secrets-en58110.67%14.63%4.62%
leak-museum10111.88%16.83%0.76%
synth-secrets-ru58112.91%17.90%5.05%
tab-echr3,83015.74%30.34%
corrupt-secrets-issues28819.44%26.39%1.74%
multiconer-ru1,20825.50%34.69%3.66%
alexen21,26126.33%100.00%6.89%
factrueval7,96629.79%29.84%
nerel24,36932.84%33.49%
creddata77634.54%38.92%2.17%
dialogpii-multi10,34535.68%39.55%
synth-ru-tickets32,09939.14%41.29%
synth-jira-comments2,70543.81%44.10%
ameau012,24244.47%44.74%1.09%
tonicai2,41748.90%51.26%0.88%
privy1,89949.29%54.29%3.28%
dialogpii-en3,08753.74%57.95%8.08%
kiji-multi7,56553.88%54.78%
redact-multi22,86455.93%59.06%
gretel-multi4,48455.93%60.53%0.61%
arthur-passwords28056.79%56.79%10.85%
nym-multi8,16960.29%62.32%3.56%
nemotron-pii9,39165.20%65.99%1.03%
scanpatch8,70865.94%67.88%1.33%
synth-wiki-tables13,09466.67%71.25%
kiji-en7,62366.84%67.60%
jayguard1,19567.78%79.08%1.45%
corrupt-redmadrobot5,53172.93%78.50%5.89%
redact-ru9,23075.20%77.70%8.39%
redmadrobot5,51676.63%79.88%2.63%
nym-en3,58477.04%78.93%1.84%
corrupt-hivetrace1,66783.32%88.84%0.71%
nym-ru9,49783.39%84.29%
hivetrace1,66788.42%90.04%0.41%
alrosait1,86294.20%95.97%2.23%
russian-pii-66k4,80596.23%96.36%