princeton-nlp/PURE
PyTorch-based system that uses pre-trained language models to perform entity and relation extraction from text.
Not currently ranked — collecting fresh signals.
star history
PURE implements a two-stage pipeline for entity and relation extraction: an entity model predicts all entities in a text passage simultaneously, followed by a relation model that considers every entity pair independently with typed entity markers to predict relation types. The system leverages pre-trained language models (PyTorch) and includes pre-trained models for ACE05 and SciERC datasets.
Frequently asked
- What is princeton-nlp/PURE?
- PyTorch-based system that uses pre-trained language models to perform entity and relation extraction from text.
- Is PURE open source?
- Yes — princeton-nlp/PURE is open source, released under the MIT license.
- What language is PURE written in?
- princeton-nlp/PURE is primarily written in Python.
- How popular is PURE?
- princeton-nlp/PURE has 813 stars on GitHub.
- Where can I find PURE?
- princeton-nlp/PURE is on GitHub at https://github.com/princeton-nlp/PURE.