《动手学深度学习》TF2.0 实现

港控/mmm° 2023-06-21 04:57 36阅读 0赞

![cover.png][]

[本项目][Link 1]将[《动手学深度学习》][Link 2] 原书中MXNet代码实现改为TensorFlow2.0实现。经过我的导师咨询李沐老师，这个项目的实施已得到李沐老师的同意。原书作者：阿斯顿·张、李沐、扎卡里 C. 立顿、亚历山大 J. 斯莫拉以及其他社区贡献者，GitHub地址：[https://github.com/d2l-ai/d2l-zh][https_github.com_d2l-ai_d2l-zh]

此书的[中][Link 3]、[英][Link 4]版本存在一些不同，本项目主要针对此书的中文版进行TensorFlow2.0重构。另外，本项目也参考了对此书中文版进行PyTorch重构的项目[Dive-into-DL-PyTorch][]，在此表示感谢。

**现已更新到第五章，持续更新中。。。**

## 简介 ##

本仓库主要包含code和docs两个文件夹（外加一些数据存放在data中）。其中code文件夹就是每章相关jupyter notebook代码（基于TensorFlow2.0）；docs文件夹就是markdown格式的《动手学深度学习》书中的相关内容，然后利用[docsify][]将网页文档部署到GitHub Pages上，由于原书使用的是MXNet框架，所以docs内容可能与原书略有不同，但是整体内容是一样的。欢迎对本项目做出贡献或提出issue。

## 面向人群 ##

本项目面向对深度学习感兴趣，尤其是想使用TensorFlow进行深度学习的童鞋。本项目并不要求你有任何深度学习或者机器学习的背景知识，你只需了解基础的数学和编程，如基础的线性代数、微分和概率，以及基础的Python编程。

## 食用方法 ##

### 方法一 ###

本仓库包含一些latex公式，但github的markdown原生是不支持公式显示的，而docs文件夹已经利用[docsify][]被部署到了GitHub Pages上，所以查看文档最简便的方法就是直接访问[本项目网页版][Link 1]。当然如果你还想跑一下运行相关代码的话还是得把本项目clone下来，然后运行code文件夹下相关代码。

### 方法二 ###

你还可以在本地访问文档，先安装`docsify-cli`工具:

npm i docsify-cli -g

然后将本项目clone到本地:

git clone https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0
    cd Dive-into-DL-TensorFlow2.0

然后运行一个本地服务器，这样就可以很方便的在`http://localhost:3000`实时访问文档网页渲染效果。

docsify serve docs

## 目录 ##

*  [简介][Link 5]
 *  [阅读指南][Link 6]
 *  [1. 深度学习简介][1.]
 *  1.  预备知识
    
     *  [2.1 环境配置][2.1]
     *  [2.2 数据操作][2.2]
     *  [2.3 自动求梯度][2.3]
     *  [2.4 查阅文档][2.4]
 *  1.  深度学习基础
    
     *  [3.1 线性回归][3.1]
     *  [3.2 线性回归的从零开始实现][3.2]
     *  [3.3 线性回归的简洁实现][3.3]
     *  [3.4 softmax回归][3.4 softmax]
     *  [3.5 图像分类数据集（Fashion-MNIST）][3.5 _Fashion-MNIST]
     *  [3.6 softmax回归的从零开始实现][3.6 softmax]
     *  [3.7 softmax回归的简洁实现][3.7 softmax]
     *  [3.8 多层感知机][3.8]
     *  [3.9 多层感知机的从零开始实现][3.9]
     *  [3.10 多层感知机的简洁实现][3.10]
     *  [3.11 模型选择、欠拟合和过拟合][3.11]
     *  [3.12 权重衰减][3.12]
     *  [3.13 丢弃法][3.13]
     *  [3.14 正向传播、反向传播和计算图][3.14]
     *  [3.15 数值稳定性和模型初始化][3.15]
     *  [3.16 实战Kaggle比赛：房价预测][3.16 _Kaggle]
 *  1.  深度学习计算
    
     *  [4.1 模型构造][4.1]
     *  [4.2 模型参数的访问、初始化和共享][4.2]
     *  [4.3 模型参数的延后初始化][4.3]
     *  [4.4 自定义层][4.4]
     *  [4.5 读取和存储][4.5]
     *  [4.6 GPU计算][4.6 GPU]
 *  1.  卷积神经网络
    
     *  [5.1 二维卷积层][5.1]
     *  [5.2 填充和步幅][5.2]
     *  [5.3 多输入通道和多输出通道][5.3]
     *  [5.4 池化层][5.4]
     *  [5.5 卷积神经网络（LeNet）][5.5 _LeNet]
     *  [5.6 深度卷积神经网络（AlexNet）][5.6 _AlexNet]
     *  [5.7 使用重复元素的网络（VGG）][5.7 _VGG]
     *  [5.8 网络中的网络（NiN）][5.8 _NiN]
     *  [5.9 含并行连结的网络（GoogLeNet）][5.9 _GoogLeNet]
     *  [5.10 批量归一化][5.10]
     *  [5.11 残差网络（ResNet）][5.11 _ResNet]
     *  [5.12 稠密连接网络（DenseNet）][5.12 _DenseNet]
 *  1.  循环神经网络
    
     *  [6.1 语言模型][6.1]
     *  [6.2 循环神经网络][6.2]
     *  [6.3 语言模型数据集（周杰伦专辑歌词）][6.3]
     *  [6.4 循环神经网络的从零开始实现][6.4]
     *  [6.5 循环神经网络的简洁实现][6.5]
     *  [6.6 通过时间反向传播][6.6]
     *  [6.7 门控循环单元（GRU）][6.7 _GRU]
     *  [6.8 长短期记忆（LSTM）][6.8 _LSTM]
     *  [6.9 深度循环神经网络][6.9]
     *  [6.10 双向循环神经网络][6.10]
 *  1.  优化算法
    
     *  [7.1 优化与深度学习][7.1]
     *  [7.2 梯度下降和随机梯度下降][7.2]
     *  [7.3 小批量随机梯度下降][7.3]
     *  [7.4 动量法][7.4]
     *  [7.5 AdaGrad算法][7.5 AdaGrad]
     *  [7.6 RMSProp算法][7.6 RMSProp]
     *  [7.7 AdaDelta算法][7.7 AdaDelta]
     *  [7.8 Adam算法][7.8 Adam]
 *  1.  计算性能
    
     *  [8.1 命令式和符号式混合编程][8.1]
     *  [8.2 异步计算][8.2]
     *  [8.3 自动并行计算][8.3]
     *  [8.4 多GPU计算][8.4 _GPU]
 *  1.  计算机视觉
    
     *  [9.1 图像增广][9.1]
     *  [9.2 微调][9.2]
     *  [9.3 目标检测和边界框][9.3]
     *  [9.4 锚框][9.4]
     *  [9.5 多尺度目标检测][9.5]
     *  [9.6 目标检测数据集（皮卡丘）][9.6]
     *  待更新…
 *  1.  自然语言处理
    
     *  [10.1 词嵌入（word2vec）][10.1 _word2vec]
     *  [10.2 近似训练][10.2]
     *  [10.3 word2vec的实现][10.3 word2vec]
     *  [10.4 子词嵌入（fastText）][10.4 _fastText]
     *  [10.5 全局向量的词嵌入（GloVe）][10.5 _GloVe]
     *  [10.6 求近义词和类比词][10.6]
     *  [10.7 文本情感分类：使用循环神经网络][10.7]
     *  [10.8 文本情感分类：使用卷积神经网络（textCNN）][10.8 _textCNN]
     *  [10.9 编码器—解码器（seq2seq）][10.9 _seq2seq]
     *  [10.10 束搜索][10.10]
     *  [10.11 注意力机制][10.11]
     *  [10.12 机器翻译][10.12]

持续更新中…

## 原书地址 ##

中文版：[动手学深度学习][Link 3] | [Github仓库][https_github.com_d2l-ai_d2l-zh]  
English Version: [Dive into Deep Learning][Link 4] | [Github Repo][]

## 引用 ##

如果您在研究中使用了这个项目请引用原书:

@book{zhang2019dive,
        title={Dive into Deep Learning},
        author={Aston Zhang and Zachary C. Lipton and Mu Li and Alexander J. Smola},
        note={\url{http://www.d2l.ai}},
        year={2019}
    }

[cover.png]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/raw/master/docs/img/cover.png
[Link 1]: https://trickygo.github.io/Dive-into-DL-TensorFlow2.0
[Link 2]: http://zh.d2l.ai/
[https_github.com_d2l-ai_d2l-zh]: https://github.com/d2l-ai/d2l-zh
[Link 3]: https://zh.d2l.ai/
[Link 4]: https://d2l.ai/
[Dive-into-DL-PyTorch]: https://github.com/ShusenTang/Dive-into-DL-PyTorch
[docsify]: https://docsify.js.org/#/zh-cn/
[Link 5]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs
[Link 6]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/read_guide.md
[1.]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter01_DL-intro/deep-learning-intro.md
[2.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter02_prerequisite/2.1_install.md
[2.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter02_prerequisite/2.2_tensor.md
[2.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter02_prerequisite/2.3_autograd.md
[2.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter02_prerequisite/2.4_document.md
[3.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.1_linear-regression.md
[3.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.2_linear-regression-scratch.md
[3.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.3_linear-regression-tensorflow2.0.md
[3.4 softmax]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.4_softmax-regression.md
[3.5 _Fashion-MNIST]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.5_fashion-mnist.md
[3.6 softmax]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.6_softmax-regression-scratch.md
[3.7 softmax]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.7_softmax-regression-tensorflow2.0.md
[3.8]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.8_mlp.md
[3.9]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.9_mlp-scratch.md
[3.10]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.10_mlp-tensorflow2.0.md
[3.11]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.11_underfit-overfit.md
[3.12]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.12_weight-decay.md
[3.13]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.13_dropout.md
[3.14]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.14_backprop.md
[3.15]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.15_numerical-stability-and-init.md
[3.16 _Kaggle]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter03_DL-basics/3.16_kaggle-house-price.md
[4.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.1_model-construction.md
[4.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.2_parameters.md
[4.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.3_deferred-init.md
[4.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.4_custom-layer.md
[4.5]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.5_read-write.md
[4.6 GPU]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter04_DL-computation/4.6_use-gpu.md
[5.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.1_conv-layer.md
[5.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.2_padding-and-strides.md
[5.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.3_channels.md
[5.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.4_pooling.md
[5.5 _LeNet]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.5_lenet.md
[5.6 _AlexNet]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.6_alexnet.md
[5.7 _VGG]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.7_vgg.md
[5.8 _NiN]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.8_nin.md
[5.9 _GoogLeNet]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.9_googlenet.md
[5.10]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.10_batch-norm.md
[5.11 _ResNet]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.11_resnet.md
[5.12 _DenseNet]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter05_CNN/5.12_densenet.md
[6.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.1_lang-model.md
[6.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.2_rnn.md
[6.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.3_lang-model-dataset.md
[6.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.4_rnn-scratch.md
[6.5]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.5_rnn-pytorch.md
[6.6]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.6_bptt.md
[6.7 _GRU]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.7_gru.md
[6.8 _LSTM]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.8_lstm.md
[6.9]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.9_deep-rnn.md
[6.10]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter06_RNN/6.10_bi-rnn.md
[7.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.1_optimization-intro.md
[7.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.2_gd-sgd.md
[7.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.3_minibatch-sgd.md
[7.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.4_momentum.md
[7.5 AdaGrad]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.5_adagrad.md
[7.6 RMSProp]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.6_rmsprop.md
[7.7 AdaDelta]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.7_adadelta.md
[7.8 Adam]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter07_optimization/7.8_adam.md
[8.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter08_computational-performance/8.1_hybridize.md
[8.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter08_computational-performance/8.2_async-computation.md
[8.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter08_computational-performance/8.3_auto-parallelism.md
[8.4 _GPU]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter08_computational-performance/8.4_multiple-gpus.md
[9.1]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.1_image-augmentation.md
[9.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.2_fine-tuning.md
[9.3]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.3_bounding-box.md
[9.4]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.4_anchor.md
[9.5]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.5_multiscale-object-detection.md
[9.6]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter09_computer-vision/9.6_object-detection-dataset.md
[10.1 _word2vec]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.1_word2vec.md
[10.2]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.2_approx-training.md
[10.3 word2vec]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.3_word2vec-pytorch.md
[10.4 _fastText]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.4_fasttext.md
[10.5 _GloVe]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.5_glove.md
[10.6]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.6_similarity-analogy.md
[10.7]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.7_sentiment-analysis-rnn.md
[10.8 _textCNN]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.8_sentiment-analysis-cnn.md
[10.9 _seq2seq]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.9_seq2seq.md
[10.10]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.10_beam-search.md
[10.11]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.11_attention.md
[10.12]: https://github.com/TrickyGo/Dive-into-DL-TensorFlow2.0/blob/master/docs/chapter10_natural-language-processing/10.12_machine-translation.md
[Github Repo]: https://github.com/d2l-ai/d2l-en