From 4b4a17d1842655297cc80d8366e8c5065b19b18c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E7=AB=A5=E6=AF=93=E6=B3=BD?= <13204402429@stu.ecnu.edu.cn>
Date: Fri, 15 Jan 2021 20:12:40 +0800
Subject: [PATCH] =?UTF-8?q?=E5=88=A0=E9=99=A4=20'code'?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 code/.ipynb_checkpoints/mnist-checkpoint.ipynb | 606 -------------------------
 code/mnist.ipynb                               | 606 -------------------------
 2 files changed, 1212 deletions(-)
 delete mode 100644 code/.ipynb_checkpoints/mnist-checkpoint.ipynb
 delete mode 100644 code/mnist.ipynb

diff --git a/code/.ipynb_checkpoints/mnist-checkpoint.ipynb b/code/.ipynb_checkpoints/mnist-checkpoint.ipynb
deleted file mode 100644
index 5838583..0000000
--- a/code/.ipynb_checkpoints/mnist-checkpoint.ipynb
+++ /dev/null
@@ -1,606 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=6)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_6\")\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 5,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.4680 - accuracy: 0.8846\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.46795412821769716, 0.8846]"
-      ]
-     },
-     "execution_count": 5,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "metadata": {},
-   "outputs": [
-    {
-     "ename": "NameError",
-     "evalue": "name 'model' is not defined",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-1-e80b0bf1617b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mevaluate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_test\u001b[0m\u001b[0;34m,\u001b[0m  \u001b[0my_test\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
-      "\u001b[0;31mNameError\u001b[0m: name 'model' is not defined"
-     ]
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/conda/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
-      "  from ._conv import register_converters as _register_converters\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "<Figure size 640x480 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/6\n",
-      "60000/60000 [==============================] - 3s 58us/sample - loss: 1.6360 - accuracy: 0.5506\n",
-      "Epoch 2/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.9633 - accuracy: 0.7638\n",
-      "Epoch 3/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.7347 - accuracy: 0.8098\n",
-      "Epoch 4/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.6252 - accuracy: 0.8334\n",
-      "Epoch 5/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5630 - accuracy: 0.8467\n",
-      "Epoch 6/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5185 - accuracy: 0.8568\n",
-      "10000/10000 [==============================] - 0s 37us/sample - loss: 0.4304 - accuracy: 0.8923\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "<Figure size 640x480 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=6)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_6\")\n",
-    "\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 3,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.4304 - accuracy: 0.8923\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.43036172440052034, 0.8923]"
-      ]
-     },
-     "execution_count": 3,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 4,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/9\n",
-      "60000/60000 [==============================] - 3s 57us/sample - loss: 1.7841 - accuracy: 0.4954\n",
-      "Epoch 2/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 1.0707 - accuracy: 0.7433\n",
-      "Epoch 3/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.7987 - accuracy: 0.7936\n",
-      "Epoch 4/9\n",
-      "60000/60000 [==============================] - 3s 56us/sample - loss: 0.6696 - accuracy: 0.8208\n",
-      "Epoch 5/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5945 - accuracy: 0.8386\n",
-      "Epoch 6/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5450 - accuracy: 0.8495\n",
-      "Epoch 7/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5106 - accuracy: 0.8577\n",
-      "Epoch 8/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.4841 - accuracy: 0.8640\n",
-      "Epoch 9/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.4627 - accuracy: 0.8687\n",
-      "10000/10000 [==============================] - 0s 38us/sample - loss: 0.3818 - accuracy: 0.9001\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=9)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_9\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 5,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.3818 - accuracy: 0.9001\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.381765931224823, 0.9001]"
-      ]
-     },
-     "execution_count": 5,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 8,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/15\n",
-      "60000/60000 [==============================] - 4s 64us/sample - loss: 0.2953 - accuracy: 0.9138\n",
-      "Epoch 2/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.1440 - accuracy: 0.9567\n",
-      "Epoch 3/15\n",
-      "60000/60000 [==============================] - 4s 63us/sample - loss: 0.1086 - accuracy: 0.9677\n",
-      "Epoch 4/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0888 - accuracy: 0.9721\n",
-      "Epoch 5/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0763 - accuracy: 0.9758\n",
-      "Epoch 6/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0658 - accuracy: 0.9789\n",
-      "Epoch 7/15\n",
-      "60000/60000 [==============================] - 4s 60us/sample - loss: 0.0572 - accuracy: 0.9813\n",
-      "Epoch 8/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0519 - accuracy: 0.9832\n",
-      "Epoch 9/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0503 - accuracy: 0.9830\n",
-      "Epoch 10/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0430 - accuracy: 0.9857\n",
-      "Epoch 11/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0414 - accuracy: 0.9861\n",
-      "Epoch 12/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0383 - accuracy: 0.9877\n",
-      "Epoch 13/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0362 - accuracy: 0.9876\n",
-      "Epoch 14/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0342 - accuracy: 0.9881\n",
-      "Epoch 15/15\n",
-      "60000/60000 [==============================] - ETA: 0s - loss: 0.0336 - accuracy: 0.98 - 4s 61us/sample - loss: 0.0336 - accuracy: 0.9884\n",
-      "10000/10000 [==============================] - 0s 37us/sample - loss: 0.0869 - accuracy: 0.9791\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='Adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=15)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_15\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 11,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.0869 - accuracy: 0.9791\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.08690820527771356, 0.9791]"
-      ]
-     },
-     "execution_count": 11,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "tensorflow-2.0",
-   "language": "python",
-   "name": "tensorflow-2.0"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.6.5"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}
diff --git a/code/mnist.ipynb b/code/mnist.ipynb
deleted file mode 100644
index 5838583..0000000
--- a/code/mnist.ipynb
+++ /dev/null
@@ -1,606 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=6)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_6\")\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 5,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.4680 - accuracy: 0.8846\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.46795412821769716, 0.8846]"
-      ]
-     },
-     "execution_count": 5,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "metadata": {},
-   "outputs": [
-    {
-     "ename": "NameError",
-     "evalue": "name 'model' is not defined",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
-      "\u001b[0;32m<ipython-input-1-e80b0bf1617b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mevaluate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_test\u001b[0m\u001b[0;34m,\u001b[0m  \u001b[0my_test\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
-      "\u001b[0;31mNameError\u001b[0m: name 'model' is not defined"
-     ]
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/conda/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
-      "  from ._conv import register_converters as _register_converters\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "<Figure size 640x480 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/6\n",
-      "60000/60000 [==============================] - 3s 58us/sample - loss: 1.6360 - accuracy: 0.5506\n",
-      "Epoch 2/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.9633 - accuracy: 0.7638\n",
-      "Epoch 3/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.7347 - accuracy: 0.8098\n",
-      "Epoch 4/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.6252 - accuracy: 0.8334\n",
-      "Epoch 5/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5630 - accuracy: 0.8467\n",
-      "Epoch 6/6\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5185 - accuracy: 0.8568\n",
-      "10000/10000 [==============================] - 0s 37us/sample - loss: 0.4304 - accuracy: 0.8923\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "<Figure size 640x480 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=6)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_6\")\n",
-    "\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 3,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.4304 - accuracy: 0.8923\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.43036172440052034, 0.8923]"
-      ]
-     },
-     "execution_count": 3,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 4,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/9\n",
-      "60000/60000 [==============================] - 3s 57us/sample - loss: 1.7841 - accuracy: 0.4954\n",
-      "Epoch 2/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 1.0707 - accuracy: 0.7433\n",
-      "Epoch 3/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.7987 - accuracy: 0.7936\n",
-      "Epoch 4/9\n",
-      "60000/60000 [==============================] - 3s 56us/sample - loss: 0.6696 - accuracy: 0.8208\n",
-      "Epoch 5/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5945 - accuracy: 0.8386\n",
-      "Epoch 6/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5450 - accuracy: 0.8495\n",
-      "Epoch 7/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.5106 - accuracy: 0.8577\n",
-      "Epoch 8/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.4841 - accuracy: 0.8640\n",
-      "Epoch 9/9\n",
-      "60000/60000 [==============================] - 3s 55us/sample - loss: 0.4627 - accuracy: 0.8687\n",
-      "10000/10000 [==============================] - 0s 38us/sample - loss: 0.3818 - accuracy: 0.9001\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=9)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_9\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 5,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.3818 - accuracy: 0.9001\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.381765931224823, 0.9001]"
-      ]
-     },
-     "execution_count": 5,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 8,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(60000, 28, 28)\n",
-      "(60000,)\n",
-      "(10000, 28, 28)\n",
-      "(10000,)\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[5 0 4 1 9 2 1 3 1]\n",
-      "Epoch 1/15\n",
-      "60000/60000 [==============================] - 4s 64us/sample - loss: 0.2953 - accuracy: 0.9138\n",
-      "Epoch 2/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.1440 - accuracy: 0.9567\n",
-      "Epoch 3/15\n",
-      "60000/60000 [==============================] - 4s 63us/sample - loss: 0.1086 - accuracy: 0.9677\n",
-      "Epoch 4/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0888 - accuracy: 0.9721\n",
-      "Epoch 5/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0763 - accuracy: 0.9758\n",
-      "Epoch 6/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0658 - accuracy: 0.9789\n",
-      "Epoch 7/15\n",
-      "60000/60000 [==============================] - 4s 60us/sample - loss: 0.0572 - accuracy: 0.9813\n",
-      "Epoch 8/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0519 - accuracy: 0.9832\n",
-      "Epoch 9/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0503 - accuracy: 0.9830\n",
-      "Epoch 10/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0430 - accuracy: 0.9857\n",
-      "Epoch 11/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0414 - accuracy: 0.9861\n",
-      "Epoch 12/15\n",
-      "60000/60000 [==============================] - 4s 62us/sample - loss: 0.0383 - accuracy: 0.9877\n",
-      "Epoch 13/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0362 - accuracy: 0.9876\n",
-      "Epoch 14/15\n",
-      "60000/60000 [==============================] - 4s 61us/sample - loss: 0.0342 - accuracy: 0.9881\n",
-      "Epoch 15/15\n",
-      "60000/60000 [==============================] - ETA: 0s - loss: 0.0336 - accuracy: 0.98 - 4s 61us/sample - loss: 0.0336 - accuracy: 0.9884\n",
-      "10000/10000 [==============================] - 0s 37us/sample - loss: 0.0869 - accuracy: 0.9791\n"
-     ]
-    },
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 432x288 with 9 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import tensorflow as tf\n",
-    "import matplotlib.pyplot as plt\n",
-    "import numpy as np\n",
-    "\n",
-    "mnist = tf.keras.datasets.mnist\n",
-    "(x_train, y_train),(x_test, y_test) = mnist.load_data(path=\"/data/data/mnist.npz\")   #加载mnist数据集\n",
-    "\n",
-    "#验证mnist数据集大小。x为数据，y为标签。mnist每张图的像素为28*28\n",
-    "print(x_train.shape)\n",
-    "print(y_train.shape)\n",
-    "print(x_test.shape)\n",
-    "print(y_test.shape)\n",
-    "\n",
-    "#打印训练集中前9张，看看是什么数字\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_train[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印相应的标签\n",
-    "print(y_train[:9])\n",
-    "\n",
-    "#基操：将像素标准化一下\n",
-    "x_train, x_test = x_train / 255.0, x_test / 255.0\n",
-    "\n",
-    "#搭建一个两层神经网络\n",
-    "model = tf.keras.models.Sequential([\n",
-    "  tf.keras.layers.Flatten(input_shape=(28, 28)),    #拉伸图像成一维向量\n",
-    "  tf.keras.layers.Dense(128, activation='relu'),    #第一层全连接+ReLU激活\n",
-    "  tf.keras.layers.Dropout(0.2),                     #dropout层\n",
-    "  tf.keras.layers.Dense(10, activation='softmax')   #第二层全连接+softmax激活，输出预测标签\n",
-    "])\n",
-    "\n",
-    "#设置训练超参，优化器为sgd，损失函数为交叉熵，训练衡量指标为accuracy\n",
-    "model.compile(optimizer='Adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
-    "\n",
-    "#开始训练，训练5个epoch，一个epoch代表所有图像计算一遍。每一个epoch能观察到训练精度的提升\n",
-    "model.fit(x_train, y_train, epochs=15)\n",
-    "\n",
-    "#计算训练了5个epoch的模型在测试集上的表现\n",
-    "model.evaluate(x_test,  y_test)\n",
-    "\n",
-    "\n",
-    "#直观看一下模型预测结果，打印测试集中的前9张图像\n",
-    "for i in range(9):  \n",
-    "    plt.subplot(3,3,1+i)\n",
-    "    plt.imshow(x_test[i], cmap='gray')\n",
-    "plt.show()\n",
-    "\n",
-    "#打印模型识别的数字，是否正确？\n",
-    "np.argmax(model(x_test[:9]).numpy(), axis=1)\n",
-    "\n",
-    "#保存训练好的模型\n",
-    "model.save(\"/data/output/model_epoch_15\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 11,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "10000/10000 [==============================] - 0s 30us/sample - loss: 0.0869 - accuracy: 0.9791\n"
-     ]
-    },
-    {
-     "data": {
-      "text/plain": [
-       "[0.08690820527771356, 0.9791]"
-      ]
-     },
-     "execution_count": 11,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "model.evaluate(x_test,  y_test)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "tensorflow-2.0",
-   "language": "python",
-   "name": "tensorflow-2.0"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.6.5"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}