Skip to content
Language models

ritual-pii-redactor

fine-tuned model for detecting and hiding personal data and secrets

2 min readupdated

The ritual-pii-redactor model is built on praveen-ritual/pii-redactor and specializes in detecting personally identifiable information (PII) in texts, logs, and HTTP request bodies. This new fine-tune trained the model to reliably find cryptocurrency mnemonics, private keys, authorization tokens, and credentials while leaving public addresses and blockchain hashes unchanged. The model weights are quantized to INT8 and exported to the ONNX format for quick integration into data processing pipelines. This model will be useful for developers and security engineers looking to automatically filter confidential information before sending it to LLMs.

Key features

  • Specialization in crypto secrets: precise identification of seed phrases, private keys, RPC links, and tokens.
  • Accuracy boost: secret search recall increased from a baseline of 1.4% to 73.9%, and the overall micro F1 metric increased to 84.9%.
  • Preservation of public data: wallet addresses and transaction hashes are not masked and pass through unchanged.
  • Ready-to-use INT8 ONNX: weights are optimized and quantized to INT8 grid for efficient inference via ONNX Runtime.

What you can do

  • Filtering LLM requests: integrate the model into a proxy server to clean text of personal data and secrets before calling language models.
  • Log protection: set up automatic scanning of backend logs and HTTP bodies for accidentally leaked private keys.
  • Dataset analysis: use the model for batch markup and masking of confidential information in text corpora.

Example

Example commands for preparing, training, ONNX exporting, and testing the model via the console:

unset HF_TOKEN HUGGING_FACE_HUB_TOKEN
uv run prepare-ritual --config configs/ritual_b.toml
uv run train --config configs/ritual_b.toml
uv run export-onnx --model runs/ritual_b/model --output exports/ritual-pii-redactor --group-size 2
uv run gate --parent praveen-ritual/pii-redactor --candidate exports/ritual-pii-redactor --ritual-eval data/prepared/ritual/ritual_b/eval.jsonl
Read moreDetails
Metrics: secret recall 0.7397 (baseline 0.0145), personal data recall 0.9963, micro F1 0.8497.
Training: LR 5e-06, 1 epoch, 30,000 replay documents and 4,000 specific documents.
Format: ONNX opset 17 weight-only INT8 with input_ids and attention_mask inputs.
Limitations: additional documents were generated using public templates rather than real live traffic.
Discuss Telegram 0 14

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 0.4 GB · ready files1.1 GB
  • BF16weights 0.2 GB0.8 GB
  • FP8 / INT8recommendedweights 0.1 GB0.7 GB
  • 4-bitweights 0.1 GB0.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F32 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Ritual-Net/ritual-pii-redactor — A30, FP8 / INT8
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv venv ~/vllm --python 3.12 && source ~/vllm/bin/activate
    uv pip install vllm
    vllm serve Ritual-Net/ritual-pii-redactor --max-model-len 8192 --quantization fp8
    Add more LoRAs to ritual-pii-redactor
  3. 3. Use it

    The server listens on port 8000 and speaks the OpenAI API: http://localhost:8000/v1 on the machine. From your computer: ssh -L 8000:localhost:8000 user@machine-IP.

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

  1. Multilingual-MiniLM-L12-H384family root
  2. pii-redactor
  3. ritual-pii-redactor

New posts about versions of Multilingual-MiniLM-L12-H384 (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 2All on Hugging Face

About the model

Ritual-Net/ritual-pii-redactor
Author
Ritual-Net
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

140

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments

Language models

Censer.8

Censer.8
Language models

Asmaa 2B

Asmaa 2B