Dataset · gretelai
Synthetic documents with PII annotations
gretelai/gretel-pii-masking-en-v1
Generated texts from various domains with PII and PHI entity annotations for training anonymization models.
What's inside
The dataset contains synthetic English texts (10k to 100k records) saturated with personally identifiable information (PII) and protected health information (PHI) across various domains. The set was generated using the Gretel Navigator tool based on the mistral-nemo-2407 model.
What you can do with it
- Fine-tuning named entity recognition (NER) models, specifically GLiNER
- Creating automated search and masking systems for personal and medical data
- Validating text anonymization tools before training LLMs
- Tasks
- Text classification, Text generation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 6.2K
- Likes
- 47
Models trained on it
- OpenMed/privacy-filter-nemotron-v2Token classification 165K
- OpenMed/privacy-filter-multilingual-v2Token classification 116.2K
- bardsai/eu-pii-anonimization-multilang-v3Token classification 2K
- Horizon-Labs/pii-redactor-smallToken classification 1.4K
- Horizon-Labs/pii-redactor-baseToken classification 759
- nekomimi-llc/egemma-pii-secretsToken classification 589
- crp4222/privaite-pii-minilmToken classification 401
- gretelai/gretel-gliner-bi-large-v1.0Token classification 118
How to load it
from datasets import load_dataset
ds = load_dataset("gretelai/gretel-pii-masking-en-v1")