GLiNER Documentation¶
GLiNER is a framework for training and deploying Named Entity Recognition (NER) models that can identify any entity type using bidirectional transformer encoders (BERT-like). Beyond standard NER, GLiNER supports multiple tasks including joint entity and relation extraction through specialized architectures. It provides a practical alternative to both traditional NER models, which are limited to predefined entity types, and Large Language Models (LLMs), which offer flexibility but require significant computational resources.
This documentation includes installation guides, tutorials, advanced topics, and full API reference.
User Guide
API Reference
- gliner.model module
- gliner.config module
- gliner.training package
- gliner.modeling package
- gliner.modeling.multitask package
- gliner.modeling.base module
- gliner.modeling.decoder module
- gliner.modeling.encoder module
- gliner.modeling.layers module
- gliner.modeling.loss_functions module
- gliner.modeling.outputs module
- gliner.modeling.scorers module
- gliner.modeling.span_rep module
- gliner.modeling.utils module
- gliner.data_processing package
- gliner.evaluation package
- gliner.onnx package
- gliner.decoding package
- gliner.utils module