LEAKS detector
credsweeper-noml PII detection benchmark
Measured on 4 of 41 retained datasets under the same frozen scoring and masking protocol.
30.36%fully hidden annotations
32.90%detected / overlapped
2.92%extra masking
—CPU characters / second
Strengths
Highest complete-masking rates by normalized data type.
Passwords, keys & tokens69.36%849 / 1,224
Bank accounts & cards0.00%0 / 169
Documents & identifiers0.00%0 / 609
Watch list
Lowest observed complete-masking rates; inspect before deployment.
Addresses & locations0.00%0 / 176
Phone numbers & email0.00%0 / 390
People's names0.00%0 / 228
Language coverage
The overall leader can differ from the best choice for one language slice.
| Language | Datasets | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| English | 3 | 74.22% | 80.51% | 3.06% |
| Russian | 1 | 0.66% | 0.66% | 0.00% |
Configuration, source and version
- Family
- LEAKS
- Revision
- 1.18.2
- Eligible datasets
- 4 / 41
- Character F1
- 0.541
- Untouched diagnostic
- 67.10% · 1,876 / 2,796
- Benchmark version
- 1.0.3 · 2026-09-09
- CPU seconds / 10k chars
- Not measured
- GPU seconds / 10k chars
- Not measured
source overlap: creddata
Results by dataset
Complete masking is primary; detection shows whether any character overlapped.
| Dataset | Gold annotations | Fully hidden ↑ | Detected ↑ | Extra masking ↓ |
|---|---|---|---|---|
| hivetrace | 1,667 | 0.66% | 0.66% | 0.00% |
| leaky-repo | 95 | 54.74% | 55.79% | 0.00% |
| secrets-issues | 288 | 60.07% | 81.94% | 1.82% |
| secrets-rules | 746 | 82.17% | 83.11% | 17.26% |