[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-9673":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":16,"forks30d":16,"starsTrendScore":19,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":24,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":33,"readmeContent":34,"aiSummary":35,"trendingCount":16,"starSnapshotCount":16,"syncStatus":36,"lastSyncTime":37,"discoverSource":38},9673,"Dive-into-DL-PyTorch","ShusenTang\u002FDive-into-DL-PyTorch","ShusenTang","本项目将《动手学深度学习》(Dive into Deep Learning)原书中的MXNet实现改为PyTorch实现。","http:\u002F\u002Ftangshusen.me\u002FDive-into-DL-PyTorch",null,"Jupyter Notebook",19406,5403,378,71,0,11,37,5,45,"Apache License 2.0",false,"master",true,[26,27,28,29,30,31,32],"computer-vision","d2l","deep-learning","deep-learning-tutorial","natural-language-processing","pytorch","pytorch-tutorial","2026-06-12 02:02:11","\n\u003Cdiv align=center>\n\u003Cimg width=\"500\" src=\"img\u002Fcover.png\" alt=\"封面\"\u002F>\n\u003C\u002Fdiv>\n\n[本项目](https:\u002F\u002Ftangshusen.me\u002FDive-into-DL-PyTorch)将[《动手学深度学习》](http:\u002F\u002Fzh.d2l.ai\u002F) 原书中MXNet代码实现改为PyTorch实现。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:\u002F\u002Fgithub.com\u002Fd2l-ai\u002Fd2l-zh\n\n此书的[中](https:\u002F\u002Fzh.d2l.ai\u002F)[英](https:\u002F\u002Fd2l.ai\u002F)版本存在一些不同，针对此书英文版的PyTorch重构可参考[这个项目](https:\u002F\u002Fgithub.com\u002Fdsgiitr\u002Fd2l-pytorch)。\nThere are some differences between the [Chinese](https:\u002F\u002Fzh.d2l.ai\u002F) and [English](https:\u002F\u002Fd2l.ai\u002F) versions of this book. For the PyTorch modifying of the English version, you can refer to [this repo](https:\u002F\u002Fgithub.com\u002Fdsgiitr\u002Fd2l-pytorch).\n\n\n## 简介\n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PyTorch）；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](https:\u002F\u002Fdocsify.js.org\u002F#\u002Fzh-cn\u002F)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。\n\n## 面向人群\n本项目面向对深度学习感兴趣，尤其是想使用PyTorch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。\n\n## 食用方法 \n### 方法一\n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](https:\u002F\u002Fdocsify.js.org\u002F#\u002Fzh-cn\u002F)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](https:\u002F\u002Ftangshusen.me\u002FDive-into-DL-PyTorch)。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。\n\n### 方法二\n你还可以在本地访问文档，先安装`docsify-cli`工具:\n``` shell\nnpm i docsify-cli -g\n```\n然后将本项目clone到本地:\n``` shell\ngit clone https:\u002F\u002Fgithub.com\u002FShusenTang\u002FDive-into-DL-PyTorch.git\ncd Dive-into-DL-PyTorch\n```\n然后运行一个本地服务器，这样就可以很方便的在`http:\u002F\u002Flocalhost:3000`实时访问文档网页渲染效果。\n``` shell\ndocsify serve docs\n```\n\n### 方法三\n如果你不想安装`docsify-cli`工具，甚至你的电脑上都没有安装`Node.js`，而出于某些原因你又想在本地浏览文档，那么你可以在`docker`容器中运行网页服务。\n\n首先将本项目clone到本地:\n``` shell\ngit clone https:\u002F\u002Fgithub.com\u002FShusenTang\u002FDive-into-DL-PyTorch.git\ncd Dive-into-DL-PyTorch\n```\n之后使用如下命令创建一个名称为「d2dl」的`docker`镜像：\n``` shell\ndocker build -t d2dl .\n```\n镜像创建好后，运行如下命令创建一个新的容器：\n``` shell\ndocker run -dp 3000:3000 d2dl\n```\n最后在浏览器中打开这个地址`http:\u002F\u002Flocalhost:3000\u002F#\u002F`，就能愉快地访问文档了。适合那些不想在电脑上装太多工具的小伙伴。\n\n\n## 目录\n* [简介]()\n* [阅读指南](read_guide.md)\n* [1. 深度学习简介](chapter01_DL-intro\u002Fdeep-learning-intro.md)\n* 2\\. 预备知识\n   * [2.1 环境配置](chapter02_prerequisite\u002F2.1_install.md)\n   * [2.2 数据操作](chapter02_prerequisite\u002F2.2_tensor.md)\n   * [2.3 自动求梯度](chapter02_prerequisite\u002F2.3_autograd.md)\n* 3\\. 深度学习基础\n   * [3.1 线性回归](chapter03_DL-basics\u002F3.1_linear-regression.md)\n   * [3.2 线性回归的从零开始实现](chapter03_DL-basics\u002F3.2_linear-regression-scratch.md)\n   * [3.3 线性回归的简洁实现](chapter03_DL-basics\u002F3.3_linear-regression-pytorch.md)\n   * [3.4 softmax回归](chapter03_DL-basics\u002F3.4_softmax-regression.md)\n   * [3.5 图像分类数据集（Fashion-MNIST）](chapter03_DL-basics\u002F3.5_fashion-mnist.md)\n   * [3.6 softmax回归的从零开始实现](chapter03_DL-basics\u002F3.6_softmax-regression-scratch.md)\n   * [3.7 softmax回归的简洁实现](chapter03_DL-basics\u002F3.7_softmax-regression-pytorch.md)\n   * [3.8 多层感知机](chapter03_DL-basics\u002F3.8_mlp.md)\n   * [3.9 多层感知机的从零开始实现](chapter03_DL-basics\u002F3.9_mlp-scratch.md)\n   * [3.10 多层感知机的简洁实现](chapter03_DL-basics\u002F3.10_mlp-pytorch.md)\n   * [3.11 模型选择、欠拟合和过拟合](chapter03_DL-basics\u002F3.11_underfit-overfit.md)\n   * [3.12 权重衰减](chapter03_DL-basics\u002F3.12_weight-decay.md)\n   * [3.13 丢弃法](chapter03_DL-basics\u002F3.13_dropout.md)\n   * [3.14 正向传播、反向传播和计算图](chapter03_DL-basics\u002F3.14_backprop.md)\n   * [3.15 数值稳定性和模型初始化](chapter03_DL-basics\u002F3.15_numerical-stability-and-init.md)\n   * [3.16 实战Kaggle比赛：房价预测](chapter03_DL-basics\u002F3.16_kaggle-house-price.md)\n* 4\\. 深度学习计算\n   * [4.1 模型构造](chapter04_DL_computation\u002F4.1_model-construction.md)\n   * [4.2 模型参数的访问、初始化和共享](chapter04_DL_computation\u002F4.2_parameters.md)\n   * [4.3 模型参数的延后初始化](chapter04_DL_computation\u002F4.3_deferred-init.md)\n   * [4.4 自定义层](chapter04_DL_computation\u002F4.4_custom-layer.md)\n   * [4.5 读取和存储](chapter04_DL_computation\u002F4.5_read-write.md)\n   * [4.6 GPU计算](chapter04_DL_computation\u002F4.6_use-gpu.md)\n* 5\\. 卷积神经网络\n   * [5.1 二维卷积层](chapter05_CNN\u002F5.1_conv-layer.md)\n   * [5.2 填充和步幅](chapter05_CNN\u002F5.2_padding-and-strides.md)\n   * [5.3 多输入通道和多输出通道](chapter05_CNN\u002F5.3_channels.md)\n   * [5.4 池化层](chapter05_CNN\u002F5.4_pooling.md)\n   * [5.5 卷积神经网络（LeNet）](chapter05_CNN\u002F5.5_lenet.md)\n   * [5.6 深度卷积神经网络（AlexNet）](chapter05_CNN\u002F5.6_alexnet.md)\n   * [5.7 使用重复元素的网络（VGG）](chapter05_CNN\u002F5.7_vgg.md)\n   * [5.8 网络中的网络（NiN）](chapter05_CNN\u002F5.8_nin.md)\n   * [5.9 含并行连结的网络（GoogLeNet）](chapter05_CNN\u002F5.9_googlenet.md)\n   * [5.10 批量归一化](chapter05_CNN\u002F5.10_batch-norm.md)\n   * [5.11 残差网络（ResNet）](chapter05_CNN\u002F5.11_resnet.md)\n   * [5.12 稠密连接网络（DenseNet）](chapter05_CNN\u002F5.12_densenet.md)\n* 6\\. 循环神经网络\n   * [6.1 语言模型](chapter06_RNN\u002F6.1_lang-model.md)\n   * [6.2 循环神经网络](chapter06_RNN\u002F6.2_rnn.md)\n   * [6.3 语言模型数据集（周杰伦专辑歌词）](chapter06_RNN\u002F6.3_lang-model-dataset.md)\n   * [6.4 循环神经网络的从零开始实现](chapter06_RNN\u002F6.4_rnn-scratch.md)\n   * [6.5 循环神经网络的简洁实现](chapter06_RNN\u002F6.5_rnn-pytorch.md)\n   * [6.6 通过时间反向传播](chapter06_RNN\u002F6.6_bptt.md)\n   * [6.7 门控循环单元（GRU）](chapter06_RNN\u002F6.7_gru.md)\n   * [6.8 长短期记忆（LSTM）](chapter06_RNN\u002F6.8_lstm.md)\n   * [6.9 深度循环神经网络](chapter06_RNN\u002F6.9_deep-rnn.md)\n   * [6.10 双向循环神经网络](chapter06_RNN\u002F6.10_bi-rnn.md)\n* 7\\. 优化算法\n   * [7.1 优化与深度学习](chapter07_optimization\u002F7.1_optimization-intro.md)\n   * [7.2 梯度下降和随机梯度下降](chapter07_optimization\u002F7.2_gd-sgd.md)\n   * [7.3 小批量随机梯度下降](chapter07_optimization\u002F7.3_minibatch-sgd.md)\n   * [7.4 动量法](chapter07_optimization\u002F7.4_momentum.md)\n   * [7.5 AdaGrad算法](chapter07_optimization\u002F7.5_adagrad.md)\n   * [7.6 RMSProp算法](chapter07_optimization\u002F7.6_rmsprop.md)\n   * [7.7 AdaDelta算法](chapter07_optimization\u002F7.7_adadelta.md)\n   * [7.8 Adam算法](chapter07_optimization\u002F7.8_adam.md)\n* 8\\. 计算性能\n   * [8.1 命令式和符号式混合编程](chapter08_computational-performance\u002F8.1_hybridize.md)\n   * [8.2 异步计算](chapter08_computational-performance\u002F8.2_async-computation.md)\n   * [8.3 自动并行计算](chapter08_computational-performance\u002F8.3_auto-parallelism.md)\n   * [8.4 多GPU计算](chapter08_computational-performance\u002F8.4_multiple-gpus.md)\n* 9\\. 计算机视觉\n   * [9.1 图像增广](chapter09_computer-vision\u002F9.1_image-augmentation.md)\n   * [9.2 微调](chapter09_computer-vision\u002F9.2_fine-tuning.md)\n   * [9.3 目标检测和边界框](chapter09_computer-vision\u002F9.3_bounding-box.md)\n   * [9.4 锚框](chapter09_computer-vision\u002F9.4_anchor.md)\n   * [9.5 多尺度目标检测](chapter09_computer-vision\u002F9.5_multiscale-object-detection.md)\n   * [9.6 目标检测数据集（皮卡丘）](chapter09_computer-vision\u002F9.6_object-detection-dataset.md)\n   - [ ] 9.7 单发多框检测（SSD）\n   * [9.8 区域卷积神经网络（R-CNN）系列](chapter09_computer-vision\u002F9.8_rcnn.md)\n   * [9.9 语义分割和数据集](chapter09_computer-vision\u002F9.9_semantic-segmentation-and-dataset.md)\n   - [ ] 9.10 全卷积网络（FCN）\n   * [9.11 样式迁移](chapter09_computer-vision\u002F9.11_neural-style.md)\n   - [ ] 9.12 实战Kaggle比赛：图像分类（CIFAR-10）\n   - [ ] 9.13 实战Kaggle比赛：狗的品种识别（ImageNet Dogs）\n* 10\\. 自然语言处理\n   * [10.1 词嵌入（word2vec）](chapter10_natural-language-processing\u002F10.1_word2vec.md)\n   * [10.2 近似训练](chapter10_natural-language-processing\u002F10.2_approx-training.md)\n   * [10.3 word2vec的实现](chapter10_natural-language-processing\u002F10.3_word2vec-pytorch.md)\n   * [10.4 子词嵌入（fastText）](chapter10_natural-language-processing\u002F10.4_fasttext.md)\n   * [10.5 全局向量的词嵌入（GloVe）](chapter10_natural-language-processing\u002F10.5_glove.md)\n   * [10.6 求近义词和类比词](chapter10_natural-language-processing\u002F10.6_similarity-analogy.md)\n   * [10.7 文本情感分类：使用循环神经网络](chapter10_natural-language-processing\u002F10.7_sentiment-analysis-rnn.md)\n   * [10.8 文本情感分类：使用卷积神经网络（textCNN）](chapter10_natural-language-processing\u002F10.8_sentiment-analysis-cnn.md)\n   * [10.9 编码器—解码器（seq2seq）](chapter10_natural-language-processing\u002F10.9_seq2seq.md)\n   * [10.10 束搜索](chapter10_natural-language-processing\u002F10.10_beam-search.md)\n   * [10.11 注意力机制](chapter10_natural-language-processing\u002F10.11_attention.md)\n   * [10.12 机器翻译](chapter10_natural-language-processing\u002F10.12_machine-translation.md)\n\n\n\n持续更新中......\n\n\n\n\n## 原书地址\n中文版：[动手学深度学习](https:\u002F\u002Fzh.d2l.ai\u002F) | [Github仓库](https:\u002F\u002Fgithub.com\u002Fd2l-ai\u002Fd2l-zh)       \nEnglish Version: [Dive into Deep Learning](https:\u002F\u002Fd2l.ai\u002F) | [Github Repo](https:\u002F\u002Fgithub.com\u002Fd2l-ai\u002Fd2l-en)\n\n\n## 引用\n如果您在研究中使用了这个项目请引用原书:\n```\n@book{zhang2019dive,\n    title={Dive into Deep Learning},\n    author={Aston Zhang and Zachary C. Lipton and Mu Li and Alexander J. Smola},\n    note={\\url{http:\u002F\u002Fwww.d2l.ai}},\n    year={2020}\n}\n```\n","本项目是《动手学深度学习》一书的PyTorch实现版本，将原书中基于MXNet的代码转换为PyTorch代码。核心功能包括使用Jupyter Notebook编写的每章节代码示例以及利用docsify部署到GitHub Pages上的文档，方便读者在线阅读和本地运行实验。该项目适合对深度学习感兴趣，特别是希望使用PyTorch框架进行学习和实践的学习者。无论是初学者还是有一定基础的研究人员都能从中受益，通过丰富的实例加深对深度学习概念和技术的理解。",2,"2026-06-11 03:24:07","top_topic"]