RepoRadar
Leave a Tip
Item detail

localai-org/privacy-filter.cpp

privacy-filter.cpp is a minimal MIT-licensed C++/GGML inference engine for OpenAI's privacy-filter NER family, shipping byte-precise PII spans with exact UTF-8 offsets, pre-converted GGUFs, and 7.7x faster CPU redacting than Hugging Face Transformers on the same hardware. It is the first serious local engine for OpenAI's privacy model and the only one that runs flat to 131k tokens where the refere

Open source →Back to radar
Score8.2
Popularity84.0
Risknone
TierGold
Score breakdown
Usefulness8.0
Novelty9.0
Momentum7.0
Maturity8.2
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease6.4

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

Useful for privacy engineers and local-AI builders who need a fast, dependency-free PII redactor that runs entirely on a single CPU or GPU: clone privacy-filter.cpp, point it at a Hugging Face GGUF, and integrate the byte offsets into the redaction layer of their product.

Who should use it

privacy engineers building redaction pipelines for LLM input/outputlocal-AI builders who need a fast, dependency-free PII redactorteams processing long documents that HF Transformers OOMs ondevelopers integrating byte-precise PII offsets into existing regex/token systems

Who should skip it

Skip if the source link, docs, or setup requirements do not match your workflow.

Risk explanation

No inherent user-impacting risk is flagged from the captured evidence.

Evidence links

Closest alternatives / related signals

nerpiiprivacyggmlcpplocalaiinferenceredaction