Ma-Lab-Berkeley/deep-representation-learning-book

Learning Deep Representations of Data Distributions

What it solves

This repository provides the source code and build tools for the book "Learning Deep Representations of Data Distributions." It is designed for users who want to generate the book from scratch, reproduce the code used for its figures, or contribute content and translations to the project.

How it works

The project uses LaTeX for document typesetting and Python (managed via uv) for executing the code associated with the book's chapters. Users can build the full book or individual chapters using latexmk. To run the code, the repository uses uv to manage a dedicated virtual environment, allowing users to execute specific scripts found within chapter folders.

Who it’s for

This repository is intended for readers who want to engage with the book's source material, including developers, researchers, and students who wish to build the book locally, reproduce its technical figures, or contribute improvements and translations to the text.

Highlights

  • Build the full book or individual chapters using standard LaTeX tools.
  • Reproduce figures and technical examples using Python scripts managed by uv.
  • Generate a web-based version of the book using make4ht and MathJax.
  • Structured workflow for contributing content or translations via GitHub pull requests.
  • Includes clear instructions for reporting issues and system requirements for local development.

Related

  • Project
    fraunhoferportugal/tsfel

    A Python library for time series analysis that provides a centralized set of over 65 features across statistical, temporal, spectral, and fractal domains for machine learning preparation.

  • Project
    catboost/tutorials

    A collection of hands-on tutorials and code examples for the CatBoost gradient boosting library, covering everything from basic model training to advanced deployment and analysis.

  • Project
    AdaptiveMotorControlLab/CEBRA

    A PyTorch-based library for estimating consistent embeddings of high-dimensional time-series recordings using self-supervised learning, specifically designed for joint behavioral and neural analysis.

  • Project
    NovaSearch-Team/RAG-Retrieval

    A comprehensive toolkit for training, fine-tuning, and distilling RAG retrieval models, including embedding models, late interaction models, and rerankers.