Supported Python versions

Python 3.7.13 or a more recent version is required to setup and use relex.

Requirements

Handling of deep neural networks and of their learning processes in relex is done basically using pytorch as backend environment, along with the following amazing packages:

• flair, a Pytorch-based library for high-level (NLP).

• torchmetrics, a Python library for creating evaluation metrics.

• numpy, a Python library for scientific computing.

• pandas, a Python library for data manipulation and analysis.

Installation

Relex package can be easily installed with:

$ pip install relex

Sub-packages

To achieve its goal of being a simple and easy-to-use framework for building relation extraction systems, relex (the package) has been thought to include the following tree of sub-packages:

• relex.data
• relex.models
• relex.learners
• relex.predictors
• relex.utilities

More details about the motivations behind the choice of this package structure and the purpose of each sub-package can be found in this paragraph. Please refer to contributions guidelines for more information on how to contribute to relex.