- Apache 2.0
- Zero-shot
- INT8 with INT4 embeddings
Shark Scent
Entities
Extracts entities such as people, email addresses or locations from text zero-shot. The basis of the dynamic PII check in Patronus Ark.
On Hugging Face
patronus-studio/gliner2-multi-edgeProfile.
Task
Sensitive data and documents
Document classes and entities before content leaves the device
Signal
person · email · phone
- Base
- GLiNER2 edge build
- Evaluation
- Quantized edge build of fastino/gliner2-multi-v1 for CPU inference, zero-shot for any entity type.
- Formats
- ONNX and ORT, encoder INT8 with INT4 embeddings, task heads INT8
- Variants and siblings
GLiNER small v2.5 Edge Smaller edge build based on gliner-community/gliner_small-v2.5.
- License
- Apache 2.0
- Hugging Face
patronus-studio/gliner2-multi-edge
How it works: Shark Scent.
Shark Scent
Input
Please reply to Anna Example, anna.example@example.com, callback at +49 89 1234567.
Output
person · email · phone
Application rule: Mask before sending
Evaluation.
Model card with protocolQuantized edge build of fastino/gliner2-multi-v1 for CPU inference, zero-shot for any entity type.
Install.
Patronus Ark loads the model and runs it on the CPU. The quickstart with Transformers or ONNX Runtime is in the model card.
Patronus Ark
pip install patronus-ark
cargo add patronus-ark