gravitee-io/gliner4j-gliner2-privacy-filter-PII-multi

GLiNER4j ONNX — GLiNER2 PII (42 labels)

ONNX export of fastino/gliner2-privacy-filter-PII-multi for Java inference via ONNX Runtime.

Part of the GLiNER4j project.

Supported Tasks

Task	Description
Named Entity Recognition	Extract typed PII entity spans from text with confidence scores

Supports entity descriptions for improved accuracy and per-call overrides without model reloading.

Repository Structure

├── gliner4j_config.json        # Shared model configuration
├── tokenizer.json              # Shared HuggingFace tokenizer
├── tokenizer_config.json
├── onnx/                       # Base FP32 (~1.1 GB)
│   ├── ner_full.onnx
│   └── classifier_full.onnx
├── onnx_fp16/                  # FP16 (~588 MB, ~50% smaller)
│   ├── ner_full.onnx
│   └── classifier_full.onnx
└── onnx_quantized/             # INT8 dynamic quantization (~350 MB, ~70% smaller)
    ├── ner_full.onnx
    └── classifier_full.onnx

An onnx_optimized_cpu/ folder with the same two files may also be present (ONNX Runtime graph-optimized for CPU).

Model Architecture

Each variant ships two merged, self-contained ONNX graphs — one per task:

Graph	Description
ner_full.onnx	Transformer encoder + span representation + count-aware scoring head (NER)
classifier_full.onnx	Transformer encoder + classifier head MLP (Classification)

The graphs are fused at export time from the encoder and task heads; the intermediate split modules are not published.

Variants

Variant	Folder	Precision	Size	Use case
Base	onnx/	FP32	~1.1 GB	Maximum accuracy
FP16	onnx_fp16/	FP16	~588 MB, ~50% smaller	Good accuracy/size trade-off
Quantized	onnx_quantized/	INT8 (QUInt8, per-channel)	~350 MB, ~70% smaller	Smallest footprint, fastest on CPU

To download a specific variant only:

huggingface-cli download <repo> --include "onnx_fp16/*" "*.json"

Configuration

Parameter	Value
Hidden size	768
Max span width	8
Max count	20
Span mode	SpanMarkerV0
Token pooling	first
ONNX opset	17

Usage

Use with GLiNER4j, a Java library for GLiNER2 inference via ONNX Runtime.

Named Entity Recognition (PII)

var entities = List.of(
    new EntityDefinition("email", "Email address"),
    new EntityDefinition("phone_number", "Phone or mobile number"),
    new EntityDefinition("card_number", "Credit / debit card number"),
    new EntityDefinition("iban", "IBAN"),
    new EntityDefinition("api_key", "API key")
);
var gliner = GLiNER4jNER.load(modelDir, entities);
Map<String, List<EntitySpan>> results = gliner.extract(
    "Charge card 4111-1111-1111-1111 to john.smith@example.com."
);

The full label set (42 PII types) is documented in the upstream model card on Hugging Face. See gliner4j-demo (run task demo:pii) for an interactive example.

Model Variants

// FP16 variant
var gliner = GLiNER4jNER.load(modelDir, entities, "onnx_fp16");

// Quantized variant
var gliner = GLiNER4jNER.load(modelDir, entities, "onnx_quantized");

License

Apache License 2.0