{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "%matplotlib inline" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "\n", "PyTorch是什么?\n", "================\n", "\n", "基于Python的科学计算包，服务于以下两种场景:\n", "\n", "- 作为NumPy的替代品，可以使用GPU的强大计算能力\n", "- 提供最大的灵活性和高速的深度学习研究平台\n", " \n", "\n", "开始\n", "---------------\n", "\n", "Tensors（张量）\n", "\n", "Tensors与Numpy中的 ndarrays类似，但是在PyTorch中\n", "Tensors 可以使用GPU进行计算.\n", "\n" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from __future__ import print_function\n", "import torch" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "创建一个 5x3 矩阵, 但是未初始化:\n", "\n" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000]])\n" ] } ], "source": [ "x = torch.empty(5, 3)\n", "print(x)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "创建一个随机初始化的矩阵:\n", "\n" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[0.6972, 0.0231, 0.3087],\n", " [0.2083, 0.6141, 0.6896],\n", " [0.7228, 0.9715, 0.5304],\n", " [0.7727, 0.1621, 0.9777],\n", " [0.6526, 0.6170, 0.2605]])\n" ] } ], "source": [ "x = torch.rand(5, 3)\n", "print(x)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "创建一个0填充的矩阵，数据类型为long:\n", "\n" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[0, 0, 0],\n", " [0, 0, 0],\n", " [0, 0, 0],\n", " [0, 0, 0],\n", " [0, 0, 0]])\n" ] } ], "source": [ "x = torch.zeros(5, 3, dtype=torch.long)\n", "print(x)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "创建tensor并使用现有数据初始化:\n", "\n" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([5.5000, 3.0000])\n" ] } ], "source": [ "x = torch.tensor([5.5, 3])\n", "print(x)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "根据现有的张量创建张量。 这些方法将重用输入张量的属性，例如， dtype，除非设置新的值进行覆盖" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[1., 1., 1.],\n", " [1., 1., 1.],\n", " [1., 1., 1.],\n", " [1., 1., 1.],\n", " [1., 1., 1.]], dtype=torch.float64)\n", "tensor([[ 0.5691, -2.0126, -0.4064],\n", " [-0.0863, 0.4692, -1.1209],\n", " [-1.1177, -0.5764, -0.5363],\n", " [-0.4390, 0.6688, 0.0889],\n", " [ 1.3334, -1.1600, 1.8457]])\n" ] } ], "source": [ "x = x.new_ones(5, 3, dtype=torch.double) # new_* 方法来创建对象\n", "print(x)\n", "\n", "x = torch.randn_like(x, dtype=torch.float) # 覆盖 dtype!\n", "print(x) # 对象的size 是相同的，只是值和类型发生了变化" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "获取 size\n", "\n", "***译者注：使用size方法与Numpy的shape属性返回的相同，张量也支持shape属性，后面会详细介绍***\n" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "torch.Size([5, 3])\n" ] } ], "source": [ "print(x.size())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "

Note

``torch.Size`` 返回值是 tuple类型, 所以它支持tuple类型的所有操作.

\n", "\n", "操作\n", "\n", "操作有多种语法。 \n", "\n", "我们将看一下加法运算。\n", "\n", "加法1:" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[ 0.7808, -1.4388, 0.3151],\n", " [-0.0076, 1.0716, -0.8465],\n", " [-0.8175, 0.3625, -0.2005],\n", " [ 0.2435, 0.8512, 0.7142],\n", " [ 1.4737, -0.8545, 2.4833]])\n" ] } ], "source": [ "y = torch.rand(5, 3)\n", "print(x + y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "加法2\n" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[ 0.7808, -1.4388, 0.3151],\n", " [-0.0076, 1.0716, -0.8465],\n", " [-0.8175, 0.3625, -0.2005],\n", " [ 0.2435, 0.8512, 0.7142],\n", " [ 1.4737, -0.8545, 2.4833]])\n" ] } ], "source": [ "print(torch.add(x, y))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "提供输出tensor作为参数\n", "\n" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[ 0.7808, -1.4388, 0.3151],\n", " [-0.0076, 1.0716, -0.8465],\n", " [-0.8175, 0.3625, -0.2005],\n", " [ 0.2435, 0.8512, 0.7142],\n", " [ 1.4737, -0.8545, 2.4833]])\n" ] } ], "source": [ "result = torch.empty(5, 3)\n", "torch.add(x, y, out=result)\n", "print(result)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "替换\n", "\n" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([[ 0.7808, -1.4388, 0.3151],\n", " [-0.0076, 1.0716, -0.8465],\n", " [-0.8175, 0.3625, -0.2005],\n", " [ 0.2435, 0.8512, 0.7142],\n", " [ 1.4737, -0.8545, 2.4833]])\n" ] } ], "source": [ "# adds x to y\n", "y.add_(x)\n", "print(y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "

Note

任何 以``_`` 结尾的操作都会用结果替换原变量.\n", " 例如: ``x.copy_(y)``, ``x.t_()``, 都会改变 ``x``.

\n", "\n", "你可以使用与NumPy索引方式相同的操作来进行对张量的操作\n", "\n" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([-2.0126, 0.4692, -0.5764, 0.6688, -1.1600])\n" ] } ], "source": [ "print(x[:, 1])" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "``torch.view``: 可以改变张量的维度和大小\n", "\n", "***译者注：torch.view 与Numpy的reshape类似***\n", "\n" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "torch.Size([4, 4]) torch.Size([16]) torch.Size([2, 8])\n" ] } ], "source": [ "x = torch.randn(4, 4)\n", "y = x.view(16)\n", "z = x.view(-1, 8) # size -1 从其他维度推断\n", "print(x.size(), y.size(), z.size())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "如果你有只有一个元素的张量，使用``.item()``来得到Python数据类型的数值\n" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([-0.2368])\n", "-0.23680149018764496\n" ] } ], "source": [ "x = torch.randn(1)\n", "print(x)\n", "print(x.item())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Read later:**\n", "\n", "\n", " 100+ Tensor operations, including transposing, indexing, slicing,\n", " mathematical operations, linear algebra, random numbers, etc.,\n", " are described\n", " `here `_.\n", "\n", "NumPy 转换\n", "------------\n", "\n", "将一个Torch Tensor转换为NumPy数组是一件轻松的事，反之亦然。\n", "\n", "Torch Tensor与NumPy数组共享底层内存地址，修改一个会导致另一个的变化。\n", "\n", "将一个Torch Tensor转换为NumPy数组\n", "\n" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([1., 1., 1., 1., 1.])\n" ] } ], "source": [ "a = torch.ones(5)\n", "print(a)" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[1. 1. 1. 1. 1.]\n" ] } ], "source": [ "b = a.numpy()\n", "print(b)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "观察numpy数组的值是如何改变的。\n", "\n" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([2., 2., 2., 2., 2.])\n", "[2. 2. 2. 2. 2.]\n" ] } ], "source": [ "a.add_(1)\n", "print(a)\n", "print(b)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ " NumPy Array 转化成 Torch Tensor\n", "\n", "使用from_numpy自动转化\n", "\n" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[2. 2. 2. 2. 2.]\n", "tensor([2., 2., 2., 2., 2.], dtype=torch.float64)\n" ] } ], "source": [ "import numpy as np\n", "a = np.ones(5)\n", "b = torch.from_numpy(a)\n", "np.add(a, 1, out=a)\n", "print(a)\n", "print(b)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "\n", "所有的 Tensor 类型默认都是基于CPU， CharTensor 类型不支持到\n", "NumPy 的转换.\n", "CUDA 张量\n", "------------\n", "\n", "使用``.to`` 方法 可以将Tensor移动到任何设备中\n", "\n" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "tensor([0.7632], device='cuda:0')\n", "tensor([0.7632], dtype=torch.float64)\n" ] } ], "source": [ "# is_available 函数判断是否有cuda可以使用\n", "# ``torch.device``将张量移动到指定的设备中\n", "if torch.cuda.is_available():\n", " device = torch.device(\"cuda\") # a CUDA 设备对象\n", " y = torch.ones_like(x, device=device) # 直接从GPU创建张量\n", " x = x.to(device) # 或者直接使用``.to(\"cuda\")``将张量移动到cuda中\n", " z = x + y\n", " print(z)\n", " print(z.to(\"cpu\", torch.double)) # ``.to`` 也会对变量的类型做更改" ] } ], "metadata": { "kernelspec": { "display_name": "Pytorch for Deeplearning", "language": "python", "name": "pytorch" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.7" } }, "nbformat": 4, "nbformat_minor": 1 }