]> git.99rst.org Git - sgasser-llm-shield.git/commit
Add PII detection accuracy benchmark (#99)
authorStefan Gasser <redacted>
Wed, 17 Jun 2026 20:24:19 +0000 (22:24 +0200)
committerGitHub <redacted>
Wed, 17 Jun 2026 20:24:19 +0000 (22:24 +0200)
commit747caa2fa1c6ce50a0cb4264f636fb6f2c796cc2
tree612e153a6150a051448c9d928dc833dbf1fccb16
parent5e2b1094aacd5786369885b414fb3304140bdcd0
Add PII detection accuracy benchmark (#99)

Adds a corpus-driven PII accuracy benchmark against the Presidio /analyze endpoint, plus pl/ro precision negatives, non-western PERSON coverage, and address-like LOCATION cases.
22 files changed:
.github/workflows/ci.yml
benchmarks/pii-accuracy/README.md [new file with mode: 0644]
benchmarks/pii-accuracy/run.ts
benchmarks/pii-accuracy/taxonomy.ts [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/boundaries.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/core.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/de.yaml [deleted file]
benchmarks/pii-accuracy/test-data/en.yaml [deleted file]
benchmarks/pii-accuracy/test-data/es.yaml [deleted file]
benchmarks/pii-accuracy/test-data/eval.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/fr.yaml [deleted file]
benchmarks/pii-accuracy/test-data/global.yaml [deleted file]
benchmarks/pii-accuracy/test-data/hard.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/it.yaml [deleted file]
benchmarks/pii-accuracy/test-data/multilingual-paragraphs.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/multilingual-sentences.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/precision-paragraphs.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/test-data/precision.yaml [new file with mode: 0644]
benchmarks/pii-accuracy/types.ts
biome.json
package.json
tsconfig.benchmarks.json [new file with mode: 0644]
git clone https://git.99rst.org/PROJECT