Metadata-Version: 2.1
Name: hierarchy_transformers
Version: 0.0.1
Summary: Language models as hierarchy encoders.
Author: Yuan He (KRR-Oxford)
Author-email: yuan.he@cs.ox.ac.uk
License: Apache License 2.0
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: datasets
Requires-Dist: deeponto>=0.8.8
Requires-Dist: geoopt==0.5.0
Requires-Dist: scikit_learn
Requires-Dist: seaborn
Requires-Dist: sentence_transformers
Requires-Dist: torch
Requires-Dist: tqdm
Requires-Dist: transformers[torch]

# Hierarchy Transformers (HiTs)

Code repository for the paper: "Language Models as Hierarchy Encoders".


## Main Dependencies

This repository follows a similar layout as the `sentence_transformers` library. It mainly depends on the following libraries:


- [Sentence Transformers](https://www.sbert.net/index.html) for language models.

- [DeepOnto](https://krr-oxford.github.io/DeepOnto/) for processing hierarchies and constructing datasets from hierarchies.

- [Geoopt](https://geoopt.readthedocs.io/en/latest/index.html) for arithmetic in hyperbolic space.


## Models on Huggingface Hub

Our HiT models will be released on the [Huggingface Hub](https://huggingface.co/Hierarchy-Transformers).

## Datasets

Datasets for HiTs are available at [Zenodo](https://zenodo.org/doi/10.5281/zenodo.10511042), including those constructed from:

- WordNet  
- SNOMED CT
- Schema.org 
- FoodOn
- DOID
