---
license: apache-2.0
language:
- en
library_name: gliner
pipeline_tag: token-classification
tags:
- entity linking
- GLiNER
- GLiNKER
- ner
- bi-encoder
base_model:
- microsoft/deberta-v3-base
- microsoft/deberta-v3-large
---

# GLiNER-Linker: Entity Disambiguation Model

![image](/static-proxy?url=https%3A%2F%2Fcdn-uploads.huggingface.co%2Fproduction%2Fuploads%2F6405f62ba577649430be5124%2FzTJU_W307kOrJSwcQubxP.png)

Bi-encoder model for entity disambiguation - the neural component (Layer 3) of the [GLiNKER framework](https://github.com/Knowledgator/GLinker).

## Models

| Model | Base Encoder | Use Case |
|-------|--------------|----------|
| [gliner-linker-base-v1.0](https://huggingface.co/knowledgator/gliner-linker-base-v1.0) | deberta-v3-base | Balanced performance |
| [gliner-linker-large-v1.0](https://huggingface.co/knowledgator/gliner-linker-large-v1.0) | deberta-v3-large | Maximum accuracy |
| [gliner-linker-rerank-v1.0](https://huggingface.co/knowledgator/gliner-linker-rerank-v1.0) | ettin-encoder-68m | Reranking |

*Both text encoder and label encoder use the same DeBERTa model*

## Quick Start with GLiNKER
```bash
pip install git+https://github.com/Knowledgator/GLinker.git
```
```python
from glinker import ConfigBuilder, DAGExecutor

# Build pipeline
builder = ConfigBuilder(name="entity_linking")

# L1: Extract mentions
builder.l1.gliner(
    model="knowledgator/gliner-bi-base-v2.0",
    labels=["person", "organization", "location"]
)

# L2: Candidate retrieval
builder.l2.add("dict", priority=0)

# L3: Disambiguation with GLiNER-Linker
builder.l3.configure(
    model="knowledgator/gliner-linker-large-v1.0",
    use_precomputed_embeddings=True
)

# Execute
executor = DAGExecutor(builder.get_config())
executor.load_entities("entities.jsonl", target_layers=["dict"])

result = executor.execute({
    "texts": ["Apple announced new iPhone"]
})

# Get linked entities
l0_result = result.get("l0_result")
for entity in l0_result.entities:
    if entity.linked_entity:
        print(f"{entity.mention_text} → {entity.linked_entity.label}")
        print(f"  Score: {entity.linked_entity.score:.3f}")
```

## Precomputed Embeddings for Speed

Enable precomputed embeddings in L2 for 10-100× speedup:
```python
builder.l2.embeddings(
    enabled=True,
    model_name="knowledgator/gliner-linker-large-v1.0"
)

# Precompute embeddings when loading entities
executor.load_entities("entities.jsonl", target_layers=["dict"])
executor.precompute_embeddings(target_layers=["postgres"], batch_size=8)
```

## Set up Reranker
```python
builder = ConfigBuilder(name="reranked")
builder.l1.gliner(model="knowledgator/gliner-bi-base-v2.0", labels=["gene", "disease"])
builder.l3.configure(model="knowledgator/gliner-linker-base-v1.0")
builder.l4.configure(
    model="knowledgator/gliner-linker-rerank-v1.0",
    threshold=0.3,
    max_labels=5,
)
builder.save("config.yaml")  # Generates L1 → L2 → L3 → L4 → L0
```

## Entity Format (JSONL)
```json
{"entity_id": "Q312", "label": "Apple Inc.", "description": "American technology company", "entity_type": "organization"}
{"entity_id": "Q89", "label": "Apple", "description": "Edible fruit of apple tree", "entity_type": "food"}
```

## Resources

- **GLiNKER Framework**: [GitHub](https://github.com/Knowledgator/GLinker)
- **Documentation**: [GLiNKER Docs](https://github.com/Knowledgator/GLinker/blob/main/DOCUMENTATION.md)
- **Discord**: [Join Community](https://discord.gg/HbW9aNJ9)

## Citation
```bibtex
@misc{stepanov2024glinermultitask,
      title={GLiNER multi-task: Generalist Lightweight Model for Various Information Extraction Tasks}, 
      author={Ihor Stepanov and Mykhailo Shtopko},
      year={2024},
      eprint={2406.12925},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}
```

---

Developed by [Knowledgator](https://knowledgator.com)