{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 参数传递\n",
    "\n",
    "`Ascend` `GPU` `CPU` `模型开发`\n",
    "\n",
    "[![在线运行](https://gitee.com/mindspore/docs/raw/r1.6/resource/_static/logo_modelarts.png)](https://authoring-modelarts-cnnorth4.huaweicloud.com/console/lab?share-url-b64=aHR0cHM6Ly9taW5kc3BvcmUtd2Vic2l0ZS5vYnMuY24tbm9ydGgtNC5teWh1YXdlaWNsb3VkLmNvbS9ub3RlYm9vay9tYXN0ZXIvcHJvZ3JhbW1pbmdfZ3VpZGUvemhfY24vbWluZHNwb3JlX2luZGVmaW5pdGVfcGFyYW1ldGVyLmlweW5i&imageid=65f636a0-56cf-49df-b941-7d2a07ba8c8c) [![下载Notebook](https://gitee.com/mindspore/docs/raw/r1.6/resource/_static/logo_notebook.png)](https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/r1.6/programming_guide/zh_cn/mindspore_indefinite_parameter.ipynb) [![下载样例代码](https://gitee.com/mindspore/docs/raw/r1.6/resource/_static/logo_download_code.png)](https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/r1.6/programming_guide/zh_cn/mindspore_indefinite_parameter.py) [![查看源文件](https://gitee.com/mindspore/docs/raw/r1.6/resource/_static/logo_source.png)](https://gitee.com/mindspore/docs/blob/r1.6/docs/mindspore/programming_guide/source_zh_cn/indefinite_parameter.ipynb)\n",
    "\n",
    "## 概述\n",
    "\n",
    "本文介绍不定参数在网络构建中的使用，指的是在构建网络时，可以使用不定个数的参数来构造，有两种构造方式，一种是直接传入一个tuple类型的参数，另一种是传入Python的可变参数(\\*参数)来构造。下面以两个例子来说明这两种构造方式的用法。\n",
    "\n",
    "## 传入tuple类型的参数\n",
    "\n",
    "构造一个单Add算子网络，该网络需要有两个输入，可以传入一个tuple类型的参数来代替两个输入。网络构造如下："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-04T06:27:38.930014Z",
     "start_time": "2022-01-04T06:27:37.268097Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import mindspore\n",
    "from mindspore import Tensor\n",
    "from mindspore.nn import Cell\n",
    "import mindspore.ops as op\n",
    "\n",
    "\n",
    "class AddModel(Cell):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.add = op.Add()\n",
    "\n",
    "    def construct(self, inputs):\n",
    "        return self.add(inputs[0], inputs[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`AddModel`网络的定义中，inputs表示的是一个tuple类型的参数，包含两个元素。调用方法如下："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-04T06:27:38.966557Z",
     "start_time": "2022-01-04T06:27:38.932034Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[2. 2. 2.]\n",
      " [2. 2. 2.]]\n"
     ]
    }
   ],
   "source": [
    "input_x = Tensor(np.ones((2, 3)), mindspore.float32)\n",
    "input_y = Tensor(np.ones((2, 3)), mindspore.float32)\n",
    "net = AddModel()\n",
    "\n",
    "y = net((input_x, input_y))\n",
    "print(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 传入Python的可变参数\n",
    "\n",
    "第二种用法是使用Python的可变参数(\\*参数)，网络构造如下："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-04T06:27:38.973200Z",
     "start_time": "2022-01-04T06:27:38.968135Z"
    }
   },
   "outputs": [],
   "source": [
    "class AddModel(Cell):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.add = op.Add()\n",
    "\n",
    "    def construct(self, *inputs):\n",
    "        return self.add(inputs[0], inputs[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "第二种用法，网络定义中，\\*inputs表示的是Python中的可变参数，可以在函数定义时收集位置参数组成一个tuple对象，也可以在函数调用时解包tuple对象中的每个参数，调用方法有两种，如下："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-04T06:27:39.005207Z",
     "start_time": "2022-01-04T06:27:38.974205Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[2. 2. 2.]\n",
      " [2. 2. 2.]]\n"
     ]
    }
   ],
   "source": [
    "input_x = Tensor(np.ones((2, 3)), mindspore.float32)\n",
    "input_y = Tensor(np.ones((2, 3)), mindspore.float32)\n",
    "net = AddModel()\n",
    "\n",
    "#1) The first call method\n",
    "y = net(input_x, input_y)\n",
    "\n",
    "#2) The second call method\n",
    "inputs = (input_x, input_y)\n",
    "y = net(*inputs)\n",
    "\n",
    "print(y)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "MindSpore",
   "language": "python",
   "name": "mindspore"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}