Upload French PII detection model OpenMed-PII-French-ClinicalE5-Large-335M-v1

530b359 verified 5 days ago

9.7 kB

language:
  - fr
license: apache-2.0
base_model: intfloat/e5-large-v2
tags:
  - token-classification
  - ner
  - pii
  - pii-detection
  - de-identification
  - privacy
  - healthcare
  - medical
  - clinical
  - phi
  - french
  - pytorch
  - transformers
  - openmed
pipeline_tag: token-classification
library_name: transformers
metrics:
  - f1
  - precision
  - recall
model-index:
  - name: OpenMed-PII-French-ClinicalE5-Large-335M-v1
    results:
      - task:
          type: token-classification
          name: Named Entity Recognition
        dataset:
          name: AI4Privacy (French subset)
          type: ai4privacy/pii-masking-400k
          split: test
        metrics:
          - type: f1
            value: 0.9652
            name: F1 (micro)
          - type: precision
            value: 0.9626
            name: Precision
          - type: recall
            value: 0.9678
            name: Recall
widget:
  - text: >-
      Dr. Jean Dupont (NSS: 1 85 12 75 108 123 45) peut être contacté à
      jean.dupont@hopital.fr ou au 06 12 34 56 78. Il habite au 15 Rue de la
      Paix, 75002 Paris.
    example_title: Clinical Note with PII (French)

OpenMed-PII-French-ClinicalE5-Large-335M-v1

French PII Detection Model | 335M Parameters | Open Source

Model Description

OpenMed-PII-French-ClinicalE5-Large-335M-v1 is a transformer-based token classification model fine-tuned for Personally Identifiable Information (PII) detection in French text. This model identifies and classifies 54 types of sensitive information including names, addresses, social security numbers, medical record numbers, and more.

Key Features

French-Optimized: Specifically trained on French text for optimal performance
High Accuracy: Achieves strong F1 scores across diverse PII categories
Comprehensive Coverage: Detects 55+ entity types spanning personal, financial, medical, and contact information
Privacy-Focused: Designed for de-identification and compliance with GDPR and other privacy regulations
Production-Ready: Optimized for real-world text processing pipelines

Performance

Evaluated on the French subset of AI4Privacy dataset:

Metric	Score
Micro F1	0.9652
Precision	0.9626
Recall	0.9678
Macro F1	0.9507
Weighted F1	0.9634
Accuracy	0.9951

Top 10 French PII Models

Rank	Model	F1	Precision	Recall
1	OpenMed-PII-French-SuperClinical-Large-434M-v1	0.9797	0.9790	0.9804
2	OpenMed-PII-French-EuroMed-210M-v1	0.9762	0.9747	0.9777
3	OpenMed-PII-French-ClinicalBGE-568M-v1	0.9733	0.9718	0.9748
4	OpenMed-PII-French-BigMed-Large-560M-v1	0.9733	0.9716	0.9749
5	OpenMed-PII-French-SnowflakeMed-Large-568M-v1	0.9728	0.9711	0.9745
6	OpenMed-PII-French-SuperMedical-Large-355M-v1	0.9728	0.9712	0.9744
7	OpenMed-PII-French-NomicMed-Large-395M-v1	0.9722	0.9704	0.9740
8	OpenMed-PII-French-mClinicalE5-Large-560M-v1	0.9713	0.9697	0.9729
9	OpenMed-PII-French-mSuperClinical-Base-279M-v1	0.9674	0.9662	0.9687
10	OpenMed-PII-French-ClinicalBGE-Large-335M-v1	0.9668	0.9644	0.9692

Supported Entity Types

This model detects 54 PII entity types organized into categories:

Identifiers (22 types)

Entity	Description
`ACCOUNTNAME`	Accountname
`BANKACCOUNT`	Bankaccount
`BIC`	Bic
`BITCOINADDRESS`	Bitcoinaddress
`CREDITCARD`	Creditcard
`CREDITCARDISSUER`	Creditcardissuer
`CVV`	Cvv
`ETHEREUMADDRESS`	Ethereumaddress
`IBAN`	Iban
`IMEI`	Imei
...	and 12 more

Personal Info (11 types)

Entity	Description
`AGE`	Age
`DATEOFBIRTH`	Dateofbirth
`EYECOLOR`	Eyecolor
`FIRSTNAME`	Firstname
`GENDER`	Gender
`HEIGHT`	Height
`LASTNAME`	Lastname
`MIDDLENAME`	Middlename
`OCCUPATION`	Occupation
`PREFIX`	Prefix
...	and 1 more

Contact Info (2 types)

Entity	Description
`EMAIL`	Email
`PHONE`	Phone

Location (9 types)

Entity	Description
`BUILDINGNUMBER`	Buildingnumber
`CITY`	City
`COUNTY`	County
`GPSCOORDINATES`	Gpscoordinates
`ORDINALDIRECTION`	Ordinaldirection
`SECONDARYADDRESS`	Secondaryaddress
`STATE`	State
`STREET`	Street
`ZIPCODE`	Zipcode

Organization (3 types)

Entity	Description
`JOBDEPARTMENT`	Jobdepartment
`JOBTITLE`	Jobtitle
`ORGANIZATION`	Organization

Financial (5 types)

Entity	Description
`AMOUNT`	Amount
`CURRENCY`	Currency
`CURRENCYCODE`	Currencycode
`CURRENCYNAME`	Currencyname
`CURRENCYSYMBOL`	Currencysymbol

Temporal (2 types)

Entity	Description
`DATE`	Date
`TIME`	Time

Usage

Quick Start

from transformers import pipeline

# Load the PII detection pipeline
ner = pipeline("ner", model="OpenMed/OpenMed-PII-French-ClinicalE5-Large-335M-v1", aggregation_strategy="simple")

text = """
Patient Jean Martin (né le 15/03/1985, NSS: 1 85 03 75 108 234 67) a été vu aujourd'hui.
Contact: jean.martin@email.fr, Téléphone: 06 12 34 56 78.
Adresse: 123 Avenue des Champs-Élysées, 75008 Paris.
"""

entities = ner(text)
for entity in entities:
    print(f"{entity['entity_group']}: {entity['word']} (score: {entity['score']:.3f})")

De-identification Example

def redact_pii(text, entities, placeholder='[REDACTED]'):
    """Replace detected PII with placeholders."""
    # Sort entities by start position (descending) to preserve offsets
    sorted_entities = sorted(entities, key=lambda x: x['start'], reverse=True)
    redacted = text
    for ent in sorted_entities:
        redacted = redacted[:ent['start']] + f"[{ent['entity_group']}]" + redacted[ent['end']:]
    return redacted

# Apply de-identification
redacted_text = redact_pii(text, entities)
print(redacted_text)

Batch Processing

from transformers import AutoModelForTokenClassification, AutoTokenizer
import torch

model_name = "OpenMed/OpenMed-PII-French-ClinicalE5-Large-335M-v1"
model = AutoModelForTokenClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

texts = [
    "Patient Jean Martin (né le 15/03/1985, NSS: 1 85 03 75 108 234 67) a été vu aujourd'hui.",
    "Contact: jean.martin@email.fr, Téléphone: 06 12 34 56 78.",
]

inputs = tokenizer(texts, return_tensors='pt', padding=True, truncation=True)
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.argmax(outputs.logits, dim=-1)

Training Details

Dataset

Source: AI4Privacy PII Masking 400k (French subset)
Format: BIO-tagged token classification
Labels: 109 total (54 entity types × 2 BIO tags + O)

Training Configuration

Max Sequence Length: 512 tokens
Epochs: 3
Framework: Hugging Face Transformers + Trainer API

Intended Use & Limitations

Intended Use

De-identification: Automated redaction of PII in French clinical notes, medical records, and documents
Compliance: Supporting GDPR, and other privacy regulation compliance
Data Preprocessing: Preparing datasets for research by removing sensitive information
Audit Support: Identifying PII in document collections

Limitations

Important: This model is intended as an assistive tool, not a replacement for human review.

False Negatives: Some PII may not be detected; always verify critical applications
Context Sensitivity: Performance may vary with domain-specific terminology
Language: Optimized for French text; may not perform well on other languages

Citation

@misc{openmed-pii-2026,
  title = {OpenMed-PII-French-ClinicalE5-Large-335M-v1: French PII Detection Model},
  author = {OpenMed Science},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/OpenMed/OpenMed-PII-French-ClinicalE5-Large-335M-v1}
}

OpenMed
/

OpenMed-PII-French-ClinicalE5-Large-335M-v1