De-identification test UI development only · synthetic data please (stored names are not encrypted yet)

Run c65a8950-2bcc-40bc-862a-ee545eb1ed42

Status
COMPLETED
Output
RELEASED · download anonymized file
Document
text/plain, 4 bytes
Chunks / labels placed
1 / –

Legend: PERSONLOCATIONORGANIZATIONHEALTHCAREDATEAGEPHONEEMAILURLNATIONAL_IDOTHER_IDPROFESSIONPASSAGE

Anonymized

Test

Restore the original

Allowed: admin. Every attempt is audited.

Restore labels in edited text

Chunks and confidentiality decisions

#SpanRiskDecisionReasonsSLM reviewDisposition
00–40 NONE– – KEEP

Labels in this document

LabelTypeOccurrences

Detections

StateTypeSourceCount

KEPT = used for a label. SUPERSEDED = overlapped by a stronger detection. PROPAGATION = found by searching known names again.

Run details

chunkersegment-aware(max=4000,overlap=200)
cipherAES256-GCM:a7c34a0b
deid0.1.0
encoderdeduce+crf
encoder_versiondeduce=3.0.6;crf=deidentify-0.7.3;model=model_crf_ons_tuned-v0.2.0@0a78b6e811e9;refine=tighten-v2(crf:Other),snap-v1,non-identifying-v3(2026-10-07.1:5e51ad55a66e)
keyerHMAC-SHA256:3f86e44c
normalizerdefault-v3
resolverexact-normalized-v1
risk_calculatordistinct-rule-sum-v1
risk_policythreshold-v1(threshold=10)
slmllama.cpp:fietje-2-instruct-q4_k_m
structurespreadsheet-structure-v1(2026-10-07.1:f92c0f97d935)
analyze_ms9
chunks1
identities0
label_collisions0
mentions_kept0
mentions_propagated0
mentions_superseded0
parse_ms0
peak_rss_mb1234
render_ms4
replacements0
resolve_ms9
slm_failures0
slm_reviews0
structure_mentions0
total_ms64
warnings[]