Skip to content

Releases: d2l-ai/d2l-en

Release v0.14.0

08 Jul 17:53
@mli mli
282c685
Compare
Choose a tag to compare

Highlights

We have added both PyTorch and TensorFlow implementations up to Chapter 7 (Modern CNNs).

Improvements

  • We updated the text to be framework neutral, such as now we call ndarray as tensor.
  • Readers can click the tab in the HTML version to switch between frameworks, both colab button and discussion thread will change properly.
  • We changed the release process, d2l.ai will host the latest release (i.e. the release branch), instead of the contents from the master branch. We unified the version number of both text and the d2l package. That's why we jumped from v0.8 to v0.14.0
  • The notebook zip contains three folders, mxnet, pytorch and tensorflow (though we only build the PDF for mxnet yet).

Release v0.8.0

30 May 06:51
41dad43
Compare
Choose a tag to compare

Highlights

D2L is now runnable on Amazon SageMaker and Google Colab.

New Contents

The following chapters are re-organized:

  • Natural Language Processing: Pretraining
  • Natural Language Processing: Applications

The following sections are added:

  • Subword Embedding (Byte-pair encoding)
  • Bidirectional Encoder Representations from Transformers (BERT)
  • The Dataset for Pretraining BERT
  • Pretraining BERT
  • Natural Language Inference and the Dataset
  • Natural Language Inference: Using Attention
  • Fine-Tuning BERT for Sequence-Level and Token-Level Applications
  • Natural Language Inference: Fine-Tuning BERT

Improvements

There have been many light revisions and improvements throughout the book.

Release v0.7.0

18 Dec 00:46
Compare
Choose a tag to compare

Highlights

  • D2L is now based on the NumPy interface. All the code samples are rewritten.

New Contents

  • Recommender Systems

    • Overview of Recommender Systems
    • The MovieLens Dataset
    • Matrix Factorization
    • AutoRec: Rating Prediction with Autoencoders
    • Personalized Ranking for Recommender Systems
    • Neural Collaborative Filtering for Personalized Ranking
    • Sequence-Aware Recommender Systems
    • Feature-Rich Recommender Systems
    • Factorization Machines
    • Deep Factorization Machines
  • Appendix: Mathematics for Deep Learning

    • Geometry and Linear Algebraic Operations
    • Eigendecompositions
    • Single Variable Calculus
    • Multivariable Calculus
    • Integral Calculus
    • Random Variables
    • Maximum Likelihood
    • Distributions
    • Naive Bayes
    • Statistics
    • Information Theory
  • Attention Mechanisms

    • Attention Mechanism
    • Sequence to Sequence with Attention Mechanism
    • Transformer
  • Generative Adversarial Networks

    • Generative Adversarial Networks
    • Deep Convolutional Generative Adversarial Networks
  • Preliminaries

    • Data Preprocessing
    • Calculus

Improvements

  • The Preliminaries chapter is improved.
  • More theoretical analysis is added to the Optimization chapter.

Preview Version

Hard copies of a D2L preview version based on this release (excluding chapters of Recommender Systems and Generative Adversarial Networks) are distributed at AWS re:Invent 2019 and NeurIPS 2019.

Release v0.6.0

11 Apr 18:00
f66891a
Compare
Choose a tag to compare

Change of Contents

We heavily revised the following chapters, especially during teaching STAT 157 at Berkeley.

  • Preface
  • Installation
  • Introduction
  • The Preliminaries: A Crashcourse
  • Linear Neural Networks
  • Multilayer Perceptrons
  • Recurrent Neural Networks

The Community Are Translating D2L into Korean and Japanese

d2l-ko in Korean (website: ko.d2l.ai) joins d2l.ai! Thank Muhyun Kim, Kyoungsu Lee, Ji hye Seo, Jiyang Kang and many other contributors!

d2l-ja in Japanese (website: ja.d2l.ai) joins d2l.ai! Thank Masaki Samejima!

Thanks to Our Contributors

@alxnorden, @avinashingit, @bowen0701, @brettkoonce, Chaitanya Prakash Bapat, @cryptonaut, Davide Fiocco, @edgarroman, @gkutiel, John Mitro, Liang Pu, Rahul Agarwal, @mohamed-ali, @mstewart141, Mike Müller, @NRauschmayr, @prakhar Srivastav, @sad-, @sfermigier, Sheng Zha, @sundeepteki, @topecongiro, @tpdi, @vermicelli, Vishaal Kapoor, @vishwesh5, @YaYaB, Yuhong Chen, Evgeniy Smirnov, @lgov, Simon Corston-Oliver, @IgorDzreyev, @trungha-ngx, @pmuens, @alukovenko, @senorcinco, @vfdev-5, @dsweet, Mohammad Mahdi Rahimi, Abhishek Gupta, @uwsd, @domkm, Lisa Oakley, @vfdev-5, @bowen0701, @arush15june, @prasanth5reddy.

Release v0.5.0

25 Jan 07:45
Compare
Choose a tag to compare

Contents

  • Translated contents from https://github.com/d2l-ai/d2l-zh, including the following chapters

    • Introduction
    • A Taste of Deep Learning
    • Deep Learning Basics
    • Deep Learning Computation
    • Convolutional Neural Networks
    • Recurrent Neural Networks
    • Optimization Algorithms
    • Computational Performance
    • Computer Vision
    • Natural Language Processing
    • Appendix
  • Added new contents in the following chapters

    • Introduction
    • A Taste of Deep Learning
    • Deep Learning Basics
    • Deep Learning Computation
    • Convolutional Neural Networks

Style

  • Improved HTML styles
  • Improved PDF styles

Chinese Version

v1.0.0-rc0 is released: https://github.com/d2l-ai/d2l-zh/releases/tag/v1.0.0-rc0
The physical book will be published soon.

Thanks to Our Contributors

alxnorden, avinashingit, bowen0701, brettkoonce, Chaitanya Prakash Bapat, cryptonaut, Davide Fiocco, edgarroman, gkutiel, John Mitro, Liang Pu, Rahul Agarwal, mohamed-ali, mstewart141, Mike Müller, NRauschmayr, Prakhar Srivastav, sad-, sfermigier, Sheng Zha, sundeepteki, topecongiro, tpdi, vermicelli, Vishaal Kapoor, vishwesh5, YaYaB