nlpflow

nlpflow is a Python library designed to simplify the process of building and deploying natural language processing (NLP) applications. It leverages well-established tools like spaCy, scikit-learn, and optuna for text preprocessing and model development. nlpflow provides a powerful toolkit for automating NLP workflows, making it easier for data scientists and developers to efficiently build models.

Features

Although nlpflow is in early-stage development, it already offers the following:

• A full text preprocessing pipeline with optional preprocessing steps.
• Development of Naive Bayes models with hyperparameter tuning using Bayesian optimization.

Planned features include:

• Tuning based on metrics beyond simple accuracy.
• Sentiment analysis using pretrained models.
• Named Entity Recognition (NER) using spaCy, with added functionality.
• Support for additional model types with hyperparameter tuning.
• Integration with popular tools such as MLFlow, GitHub Actions, and Docker.

Installation

You can install nlpflow using pip:

pip install nlpflow

Documentation

Documentation is still under development. For example usage of the library, check the examples.ipynb file.

Contributing

For contributions, please contact me directly on LinkedIn here.

License

This project is licensed under the Apache License 2.0.

Acknowledgements

Thanks to the developers and contributors of the libraries this tool depends on! Let's continue improving the open-source community together!