change to version 2

Author: L1aoXingyu

README.md

@@ -4,17 +4,88 @@ Learn Deep Learning with PyTorch
 
 非常感谢您能够购买此书，这个github repository包含有[深度学习入门之PyTorch](https://item.jd.com/17915495606.html)的实例代码。由于本人水平有限，在写此书的时候参考了一些网上的资料，在这里对他们表示敬意。由于深度学习的技术在飞速的发展，同时PyTorch也在不断更新，且本人在完成此书的时候也有诸多领域没有涉及，所以这个repository会不断更新作为购买次书的一个后续服务，希望我能够在您深度学习的入门道路上提供绵薄之力。
 
+**注意：由于PyTorch版本更迭，书中的代码可能会出现bug，所以一切代码以该github中的为主。**
+
 ![image.png](http://upload-images.jianshu.io/upload_images/3623720-7cc3a383f486d157.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
 
 ## 配置环境
 
 书中已经详细给出了如何基于Anaconda配置python环境，以及PyTorch的安装，如果你使用自己的电脑，并且有Nvidia的显卡，那么你可以愉快地进入深度学习的世界了，如果你没有Nvidia的显卡，那么我们需要一个云计算的平台来帮助我们学习深度学习之旅。[如何配置aws计算平台](https://github.com/SherlockLiao/code-of-learn-deep-learning-with-pytorch/blob/master/aws.md)
 
 
-
+**以下的课程目录和书中目录有出入，因为内容正在不断更新，所有的内容更新完成会更迭到书的第二版中！**
 ## 课程目录
-
-
+### part1: 深度学习基础
+- Chapter 2: PyTorch基础
+    - PyTorch和NumPy
+    - 自动求导机制
+    - 数据的读取
+    - Module和Sequential
+    - 自定义参数的初始化
+    - 模型保存和读取
+
+- Chapter 3: PyTorch高级
+    - tensorboard可视化
+    - 优化器
+    - 自定义loss和非标准层
+    - 数据并行和多GPU
+    - PyTorch的分布式应用
+    - 使用ONNX转化为Caffe2模型
+
+- Chapter 4: 多层感知器
+    - 线性模型
+    - Logistic 回归
+    - 多层神经网络
+
+- Chapter 5: 卷积神经网络
+    - 从0开始手动搭建卷积网络
+    - 批标准化
+    - 使用重复元素的深度网络，VGG
+    - 更加丰富化结构的网络，GoogLeNet
+    - 深度残差网络，ResNet
+    - 稠密连接的卷积网络，DenseNet
+
+- Chapter 6: 循环神经网络
+    - LSTM和GRU
+    - 使用RNN进行时间序列分析
+    - 使用RNN进行图像分类
+    - Word Embedding和N-Gram模型
+    - Seq-LSTM做词性预测
+
+### part2: 深度学习的应用
+- Chapter 7: 计算机视觉
+    - 图像增强的方法
+    - Fine-tuning: 通过微调进行迁移学习
+    - 语义分割: 通过卷积实现像素级别的分类
+    - 使用卷积网络进行目标检测
+    - 使用triplet loss进行人脸识别
+    - Neural Transfer: 通过卷积网络实现风格迁移
+    - Deep Dream: 探索卷积网络眼中的世界
+
+- Chapter 8: 自然语言处理
+    - char rnn实现文本生成
+    - 联合卷积网络实现图片字幕
+    - 使用rnn进行情感分析
+    - seq2seq实现机器翻译
+    - cnn+rnn+attention实现文本识别
+    - Tree-lstm实现语义相关性分析
+
+### part3: 高级内容
+- Chapter 9: 生成对抗网络
+    - 自动编码器
+    - 变分自动编码器
+    - 生成对抗网络的介绍
+    - 深度卷积对抗网络(DCGANs)
+    - Wasserstein-GANs
+    - 条件生成对抗网络(Conditional
+     GANs)
+    - Pix2Pix
+
+- Chapter 10: 深度增强学习
+    - 深度增强学习的介绍
+    - Policy gradient
+    - Actor-critic gradient
+    - Deep Q-networks
 
 ## 一些别的资源
 
@@ -24,7 +95,7 @@ Learn Deep Learning with PyTorch
 
 关于PyTorch的资源
 
-我的github repository [pytorch-beginner](https://github.com/SherlockLiao/pytorch-beginner)
+我的github repo [pytorch-beginner](https://github.com/SherlockLiao/pytorch-beginner)
 
 [pytorch-tutorial](https://github.com/yunjey/pytorch-tutorial)
 

aws.md

@@ -19,12 +19,11 @@
 ![2.png](http://upload-images.jianshu.io/upload_images/3623720-0fca7afcf3c0508e.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
 
 
-这个界面只需要注意三个地方，一个是右上角的地区，需要选择一个离你比较近的地区，整个亚太地区可以选择韩国，日本，新加坡和孟买。然后是左边的一个方框"限制"，如果你申请CPU的计算实例，那么不用管，如果你要申请GPU计算实例，就需要点击"限制"进行申请，因为GPU实例会产生费用，亚马逊需要和你确认这个事情，一般需要两到三个工作日。
+这个界面只需要注意三个地方，一个是右上角的地区，需要选择一个离你比较近的地区，整个亚太地区可以选择韩国，日本，新加坡和孟买，需要注意的是不同的地区实例价格是不同的，如果你有vpn，那么推荐选择俄勒冈，因为这个地区最便宜，比亚太地区便宜了4到5倍。然后是左边的一个方框"限制"，如果你申请CPU的计算实例，那么不用管，如果你要申请GPU计算实例，就需要点击"限制"进行申请，因为GPU实例会产生费用，亚马逊需要和你确认这个事情，一般需要两到三个工作日。
 
 接下面就可以开始启动实例了，点击中间的红框即可开始。
 
 
-
 ### 申请实例并启动
 
 

chapter2_PyTorch-Basics/PyTorch-and-NumPy.ipynb

@@ -0,0 +1,192 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# PyTorch Basic\n",
+    "In this notebook, we will introduce much pytorch basic knowledge which you need to get into deep learning world."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Use PyTorch as NumPy\n",
+    "PyTorch is a Tensors and Dynamic neural networks in Python with strong GPU acceleration Library. So we can use it as numpy but it is faster than numpy."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {
+    "collapsed": false
+   },
+   "outputs": [],
+   "source": [
+    "# create a numpy ndarray\n",
+    "numpy_tensor = np.random.randn(10, 20)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "we can convert numpy ndarray to pytorch Tensor in two ways"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "from torch import nn\n",
+    "import numpy as np\n",
+    "from torch.autograd import Variable\n",
+    "from torch.utils.data import Dataset, DataLoader\n",
+    "from torchvision.datasets import ImageFolder\n",
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "# =============================Tensor================================\n",
+    "# Define 3x2 matrix with given values\n",
+    "a = torch.Tensor([[2, 3], [4, 8], [7, 9]])\n",
+    "print('a is: {}'.format(a))\n",
+    "print('a size is {}'.format(a.size()))  # a.size() = 3, 2\n",
+    "\n",
+    "b = torch.LongTensor([[2, 3], [4, 8], [7, 9]])\n",
+    "print('b is : {}'.format(b))\n",
+    "\n",
+    "c = torch.zeros((3, 2))\n",
+    "print('zero tensor: {}'.format(c))\n",
+    "\n",
+    "d = torch.randn((3, 2))\n",
+    "print('normal randon is : {}'.format(d))\n",
+    "\n",
+    "a[0, 1] = 100\n",
+    "print('changed a is: {}'.format(a))\n",
+    "\n",
+    "numpy_b = b.numpy()\n",
+    "print('conver to numpy is \\n {}'.format(numpy_b))\n",
+    "\n",
+    "e = np.array([[2, 3], [4, 5]])\n",
+    "torch_e = torch.from_numpy(e)\n",
+    "print('from numpy to torch.Tensor is {}'.format(torch_e))\n",
+    "f_torch_e = torch_e.float()\n",
+    "print('change data type to float tensor: {}'.format(f_torch_e))\n",
+    "\n",
+    "if torch.cuda.is_available():\n",
+    "    a_cuda = a.cuda()\n",
+    "    print(a_cuda)\n",
+    "\n",
+    "# =============================Variable===================================\n",
+    "\n",
+    "# Create Variable\n",
+    "x = Variable(torch.Tensor([1]), requires_grad=True)\n",
+    "w = Variable(torch.Tensor([2]), requires_grad=True)\n",
+    "b = Variable(torch.Tensor([3]), requires_grad=True)\n",
+    "\n",
+    "# Build a computational graph.\n",
+    "y = w * x + b    # y = 2 * x + 3\n",
+    "\n",
+    "# Compute gradients\n",
+    "y.backward()  # same as y.backward(torch.FloatTensor([1]))\n",
+    "# Print out the gradients.\n",
+    "print(x.grad)    # x.grad = 2\n",
+    "print(w.grad)    # w.grad = 1\n",
+    "print(b.grad)    # b.grad = 1\n",
+    "\n",
+    "x = torch.randn(3)\n",
+    "x = Variable(x, requires_grad=True)\n",
+    "\n",
+    "y = x * 2\n",
+    "print(y)\n",
+    "\n",
+    "y.backward(torch.FloatTensor([1, 0.1, 0.01]))\n",
+    "print(x.grad)\n",
+    "\n",
+    "\n",
+    "# ==============================nn.Module=================================\n",
+    "class net_name(nn.Module):\n",
+    "    def __init__(self, other_arguments):\n",
+    "        super(net_name, self).__init__()\n",
+    "        self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size)\n",
+    "        # other network layer\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        x = self.conv1(x)\n",
+    "        return x\n",
+    "\n",
+    "# ============================Dataset====================================\n",
+    "\n",
+    "\n",
+    "class myDataset(Dataset):\n",
+    "    def __init__(self, csv_file, txt_file, root_dir, other_file):\n",
+    "        self.csv_data = pd.read_csv(csv_file)\n",
+    "        with open(txt_file, 'r') as f:\n",
+    "            data_list = f.readlines()\n",
+    "        self.txt_data = data_list\n",
+    "        self.root_dir = root_dir\n",
+    "\n",
+    "    def __len__(self):\n",
+    "        return len(self.csv_data)\n",
+    "\n",
+    "    def __getitem__(self, idx):\n",
+    "        data = (self.csv_data[idx], self.txt_data[idx])\n",
+    "        return data\n",
+    "\n",
+    "\n",
+    "dataiter = DataLoader(myDataset, batch_size=32, shuffle=True,\n",
+    "                      collate_fn=default_collate)\n",
+    "\n",
+    "dset = ImageFolder(root='root_path', transform=None,\n",
+    "                   loader=default_loader)\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "mx",
+   "language": "python",
+   "name": "mx"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}