PR

Privacy Filter

PrivacyGGUF by LocalAI-ioUpdated 2026-06-18
Download on Hugging Face
Parameters1.4B
Context128K
Layers8
Architectureopenai-privacy-filter
Smallest quant1.6 GB
Quick answer

Privacy Filter needs about 2.2 GB of memory at its main file (a 1.6 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 1.6 GB to 2.8 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 0.1 GB.

Bidirectional token‑classification model for PII detection

Privacy Filter is a 1.4B‑parameter model that labels tokens for personally identifiable information. It uses a 128,000‑token context window and a constrained Viterbi decoder for coherent span extraction. The model is open‑source under Apache‑2.0.

Good at

  • PII detection and masking
  • Long‑context token classification
  • Fast single‑pass inference

Limits

  • No vision support
  • Requires custom runtime (privacy‑filter.cpp)
  • Large context may need more RAM

Best for

  • On‑prem data sanitization
  • Batch PII redaction
  • Compliance workflows

Who it’s for. Teams that need a fast, on‑prem PII detector that can be fine‑tuned for specific data distributions. The model is suitable for developers and data engineers building privacy‑aware pipelines.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—1.6 GB 2.2 GB
Fits Get file →
F162.8 GB 3.4 GB
Fits Get file →

For 8 GB VRAM, the unknown quant file (1.64 GB) needs ~2.2 GB RAM at 8 k context, while the F16 quant (2.82 GB) needs ~3.4 GB. Both fit comfortably; choose the smaller file for lower memory usage.

What GPU do you need for Privacy Filter?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is F16 (2.8 GB file, about 3.4 GB in use).

Specifications

Architecture from the GGUF header

Architectureopenai-privacy-filter
Parameters1.4B (approximate, mixture of experts)
Layers8
Trained context131,072 tokens
KV cache (fp16)4 KB per token

Source

Quantized byLocalAI-io
FamilyPrivacy
Licenseapache-2.0
Downloads (HF)3,879
Last updated2026-06-18

Frequently asked questions about Privacy Filter

How much VRAM does Privacy Filter need?

About 2.2 GB at its main file with an 8K context window. That is the 1.6 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Privacy Filter on an 8 GB GPU?

Yes, with F16: the 2.8 GB file needs about 3.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Privacy Filter GGUF?

It depends on the quant: file is 1.6 GB, F16 is 2.8 GB.

What context length does Privacy Filter support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.0 GB at 8K and 0.1 GB at 32K.

Where can I download Privacy Filter GGUF?

The files are on Hugging Face in the LocalAI-io/privacy-filter-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Privacy Filter use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run this model on 8 GB VRAM?

Yes. The unknown quant requires ~2.2 GB RAM and the F16 quant ~3.4 GB, both below 8 GB.

Does this model support multiple languages?

The base model has no language list; multilingual support is available in a separate fine‑tuned variant.

Can I fine‑tune this model?

Yes, the model is designed for easy, data‑efficient fine‑tuning to adapt to specific data distributions.

Similar models

Browse all →

More from LocalAI-io

All by LocalAI-io →

More Privacy models

All Privacy →

Other tiny models

All tiny models →