{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "joI9Ck_HEEfQ"
   },
   "source": [
    "# Neural Network: Callback and Checkpoints"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eoM32EDzEEfR"
   },
   "source": [
    "For classifying MNIST digits. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lgNqpXcGEEfS"
   },
   "source": [
    "#### Load dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 661
    },
    "colab_type": "code",
    "id": "Qr6RAnvpFzae",
    "outputId": "1af7f9c1-e13e-44bb-c84a-ba5a2172abe2"
   },
   "outputs": [],
   "source": [
    "#pip install tensorflow==2.0.0-beta0\n",
    "#pip install --upgrade tensorflow==2.0.0-beta0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "d3sEySBYEEfS"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras  # tf.keras\n",
    "import seaborn as sns\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import os\n",
    "import pandas as pd\n",
    "import sklearn\n",
    "import sys\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "sns.set()\n",
    "%matplotlib inline\n",
    "%load_ext tensorboard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "python 3.7.1 (default, Dec 14 2018, 13:28:58) \n",
      "[Clang 4.0.1 (tags/RELEASE_401/final)]\n",
      "matplotlib 3.0.2\n",
      "numpy 1.15.4\n",
      "pandas 0.23.4\n",
      "sklearn 0.20.1\n",
      "tensorflow 2.0.0-beta0\n",
      "tensorflow.python.keras.api._v2.keras 2.2.4-tf\n"
     ]
    }
   ],
   "source": [
    "print(\"python\", sys.version)\n",
    "for module in mpl, np, pd, sklearn, tf, keras:\n",
    "    print(module.__name__, module.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zo7uccB7EEfW"
   },
   "source": [
    "#### Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 52
    },
    "colab_type": "code",
    "id": "tEARWdhNEEfW",
    "outputId": "1a1a2b0f-cab5-40b1-8bc0-6e036256e2c6"
   },
   "outputs": [],
   "source": [
    "fashion_mnist = keras.datasets.fashion_mnist\n",
    "(X_train_full, y_train_full), (X_test, y_test) = (\n",
    "    fashion_mnist.load_data())\n",
    "X_valid, X_train = X_train_full[:5000], X_train_full[5000:]\n",
    "y_valid, y_train = y_train_full[:5000], y_train_full[5000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((55000, 28, 28), (55000,))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_names = [\"T-shirt/top\", \"Trouser\", \"Pullover\", \"Dress\", \"Coat\",\n",
    "               \"Sandal\", \"Shirt\", \"Sneaker\", \"Bag\", \"Ankle boot\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#X_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x576 with 50 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_rows = 5\n",
    "n_cols = 10\n",
    "plt.figure(figsize=(n_cols*1.4, n_rows * 1.6))\n",
    "for row in range(n_rows):\n",
    "    for col in range(n_cols):\n",
    "        index = n_cols * row + col\n",
    "        plt.subplot(n_rows, n_cols, index + 1)\n",
    "        plt.imshow(X_train[index], cmap=\"binary\", interpolation=\"nearest\")\n",
    "        plt.axis('off')\n",
    "        plt.title(class_names[y_train[index]])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([4, 0, 7, 9, 9, 9, 4, 4, 3, 4, 0, 1, 8, 6, 3, 6, 4, 3, 2, 8, 7, 3,\n",
       "       4, 7, 1, 3, 4, 2, 0, 8, 5, 5, 9, 1, 5, 3, 5, 9, 0, 3, 9, 6, 4, 2,\n",
       "       9, 0, 8, 3, 3, 2], dtype=uint8)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[0:50]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Coat'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_names[y_train[0]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "hPqUUAWKEEfY"
   },
   "source": [
    "### Preprocess data : Feature Scaling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When using Gradient Descent, it is usually best to ensure that the features all have a similar scale, preferably with a Normal distribution. Try to standardize the pixel values and see if this improves the performance of your neural network.\n",
    "\n",
    "**Tips**:\n",
    "* For each feature (pixel intensity), you must subtract the `mean()` of that feature (across all instances, so use `axis=0`) and divide by its standard deviation (`std()`, again `axis=0`). Alternatively, you can use Scikit-Learn's `StandardScaler`.\n",
    "* Make sure you compute the means and standard deviations on the training set, and use these statistics to scale the training set, the validation set and the test set (you should not fit the validation set or the test set, and computing the means and standard deviations counts as \"fitting\")."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "scaler = StandardScaler()\n",
    "X_train_scaled = scaler.fit_transform(X_train.astype(np.float32)\\\n",
    "                                      .reshape(-1, 28 * 28)).reshape(-1, 28, 28)\n",
    "X_valid_scaled = scaler.transform(X_valid.astype(np.float32)\\\n",
    "                                  .reshape(-1, 28 * 28)).reshape(-1, 28, 28)\n",
    "X_test_scaled = scaler.transform(X_test.astype(np.float32)\\\n",
    "                                 .reshape(-1, 28 * 28)).reshape(-1, 28, 28)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JAU1Cyy1EEfe"
   },
   "source": [
    "#### Design neural network architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GIVzBGIfEEfh"
   },
   "outputs": [],
   "source": [
    "model = keras.models.Sequential([\n",
    "    keras.layers.Flatten(input_shape=[28, 28]),\n",
    "    keras.layers.Dense(300, activation=\"relu\"),\n",
    "    keras.layers.Dense(100, activation=\"relu\"),\n",
    "    keras.layers.Dense(10, activation=\"softmax\")\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 330
    },
    "colab_type": "code",
    "id": "OJIwNRLNEEfj",
    "outputId": "91f81122-7305-4d82-93c3-6942432381a0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten (Flatten)            (None, 784)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 300)               235500    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 100)               30100     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                1010      \n",
      "=================================================================\n",
      "Total params: 266,610\n",
      "Trainable params: 266,610\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "o2hYUMTzEEfp"
   },
   "source": [
    "### Configure model: Callbacks and Checkpoints"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `fit()` method accepts a `callbacks` argument. Try training your model with a large number of epochs, a validation set, and with a few callbacks from `keras.callbacks`:\n",
    "* `TensorBoard`: specify a log directory. It should be a subdirectory of a root logdir, such as `./my_logs/run_1`, and it should be different every time you train your model. You can use a timestamp in the subdirectory's path to ensure that it changes at every run.\n",
    "* `EarlyStopping`: specify `patience=5`\n",
    "* `ModelCheckpoint`: specify the path of the checkpoint file to save (e.g., `\"my_mnist_model.h5\"`) and set `save_best_only=True`\n",
    "\n",
    "Notice that the `EarlyStopping` callback will interrupt training before it reaches the requested number of epochs. This reduces the risk of overfitting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ZANJM3AeEEfq"
   },
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
    "              optimizer=keras.optimizers.SGD(1e-3),\\\n",
    "              metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zq4UBoc5EEfs"
   },
   "source": [
    "#### Train!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mkdir: log: File exists\r\n"
     ]
    }
   ],
   "source": [
    "mkdir log"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "logdir = os.path.join(\"./log/run_1\", \"run_{}\".format(time.time()))\n",
    "\n",
    "callbacks = [\n",
    "    keras.callbacks.TensorBoard(logdir),\n",
    "    keras.callbacks.EarlyStopping(patience=5),\n",
    "    keras.callbacks.ModelCheckpoint(\"my_mnist_model.h5\", save_best_only=True),\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 471
    },
    "colab_type": "code",
    "id": "ln4J2UMLEEfu",
    "outputId": "9fb80236-f579-45d7-ac9a-39f85be0ee0f"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W1228 16:58:45.364671 4328199616 deprecation.py:323] From /Users/admin/anaconda3/lib/python3.7/site-packages/tensorflow/python/ops/math_grad.py:1250: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.where in 2.0, which has the same broadcast rule as np.where\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 55000 samples, validate on 5000 samples\n",
      "Epoch 1/20\n",
      "55000/55000 [==============================] - 2s 40us/sample - loss: 1.4332 - accuracy: 0.5387 - val_loss: 0.9500 - val_accuracy: 0.7058\n",
      "Epoch 2/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.8561 - accuracy: 0.7255 - val_loss: 0.7478 - val_accuracy: 0.7598\n",
      "Epoch 3/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.7228 - accuracy: 0.7617 - val_loss: 0.6602 - val_accuracy: 0.7816\n",
      "Epoch 4/20\n",
      "55000/55000 [==============================] - 2s 32us/sample - loss: 0.6535 - accuracy: 0.7809 - val_loss: 0.6091 - val_accuracy: 0.7990\n",
      "Epoch 5/20\n",
      "55000/55000 [==============================] - 2s 35us/sample - loss: 0.6085 - accuracy: 0.7931 - val_loss: 0.5736 - val_accuracy: 0.8116\n",
      "Epoch 6/20\n",
      "55000/55000 [==============================] - 2s 31us/sample - loss: 0.5760 - accuracy: 0.8018 - val_loss: 0.5479 - val_accuracy: 0.8172\n",
      "Epoch 7/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.5510 - accuracy: 0.8096 - val_loss: 0.5275 - val_accuracy: 0.8250\n",
      "Epoch 8/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.5311 - accuracy: 0.8150 - val_loss: 0.5122 - val_accuracy: 0.8302\n",
      "Epoch 9/20\n",
      "55000/55000 [==============================] - 2s 30us/sample - loss: 0.5147 - accuracy: 0.8208 - val_loss: 0.4988 - val_accuracy: 0.8346\n",
      "Epoch 10/20\n",
      "55000/55000 [==============================] - 2s 34us/sample - loss: 0.5010 - accuracy: 0.8256 - val_loss: 0.4877 - val_accuracy: 0.8364\n",
      "Epoch 11/20\n",
      "55000/55000 [==============================] - 2s 31us/sample - loss: 0.4893 - accuracy: 0.8291 - val_loss: 0.4782 - val_accuracy: 0.8390\n",
      "Epoch 12/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.4790 - accuracy: 0.8322 - val_loss: 0.4704 - val_accuracy: 0.8404\n",
      "Epoch 13/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.4699 - accuracy: 0.8352 - val_loss: 0.4628 - val_accuracy: 0.8444\n",
      "Epoch 14/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.4618 - accuracy: 0.8380 - val_loss: 0.4565 - val_accuracy: 0.8450\n",
      "Epoch 15/20\n",
      "55000/55000 [==============================] - 2s 29us/sample - loss: 0.4545 - accuracy: 0.8401 - val_loss: 0.4505 - val_accuracy: 0.8482\n",
      "Epoch 16/20\n",
      "55000/55000 [==============================] - 2s 28us/sample - loss: 0.4480 - accuracy: 0.8426 - val_loss: 0.4453 - val_accuracy: 0.8498\n",
      "Epoch 17/20\n",
      "55000/55000 [==============================] - 2s 29us/sample - loss: 0.4419 - accuracy: 0.8446 - val_loss: 0.4406 - val_accuracy: 0.8508\n",
      "Epoch 18/20\n",
      "55000/55000 [==============================] - 2s 30us/sample - loss: 0.4363 - accuracy: 0.8468 - val_loss: 0.4362 - val_accuracy: 0.8520\n",
      "Epoch 19/20\n",
      "55000/55000 [==============================] - 2s 34us/sample - loss: 0.4312 - accuracy: 0.8484 - val_loss: 0.4318 - val_accuracy: 0.8524\n",
      "Epoch 20/20\n",
      "55000/55000 [==============================] - 2s 34us/sample - loss: 0.4264 - accuracy: 0.8499 - val_loss: 0.4286 - val_accuracy: 0.8538\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train_scaled,\\\n",
    "                    y_train,\\\n",
    "                    batch_size=128,\\\n",
    "                    epochs=20,\n",
    "                    validation_data=(X_valid_scaled, y_valid),\n",
    "                   callbacks=callbacks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 39us/sample - loss: 0.4680 - accuracy: 0.8349\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.4679875907897949, 0.8349]"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test_scaled, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_learning_curves(history):\n",
    "    pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
    "    plt.grid(True)\n",
    "    plt.gca().set_ylim(0, 1)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_learning_curves(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The early stopping callback only stopped training after 10 epochs without progress, so your model may already have started to overfit the training set. Fortunately, since the `ModelCheckpoint` callback only saved the best models (on the validation set), the last saved model is the best on the validation set, so try loading it using `keras.models.load_model()`. Finally evaluate it on the test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.load_model(\"my_mnist_model.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000/5000 [==============================] - 0s 41us/sample - loss: 0.4286 - accuracy: 0.8538\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.428587375164032, 0.8538]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_valid_scaled, y_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "yFJm3TV_mrmH"
   },
   "source": [
    "#### Performing Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 38us/sample - loss: 6.0076 - accuracy: 0.6082\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[6.007626473999023, 0.6082]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5000, 28, 28)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_valid.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  ],\n",
       "       [0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 0.  , 0.99, 0.  , 0.01, 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 0.  , 0.  , 0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  ],\n",
       "       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  ],\n",
       "       [0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 1.  , 0.  ]],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_new = 10\n",
    "X_new = X_test[:n_new]\n",
    "y_proba = model.predict(X_new)\n",
    "y_proba.round(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([8, 2, 1, 1, 2, 1, 4, 4, 8, 8])"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = model.predict_classes(X_new)\n",
    "y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bag\n",
      "Pullover\n",
      "Trouser\n",
      "Trouser\n",
      "Pullover\n",
      "Trouser\n",
      "Coat\n",
      "Coat\n",
      "Bag\n",
      "Bag\n"
     ]
    }
   ],
   "source": [
    "for item in y_pred:\n",
    "    print(class_names[item])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "X3wvk6OWm2oz"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1a2d4de978>"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(X_valid[0].reshape(28,28))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "TPU",
  "colab": {
   "include_colab_link": true,
   "name": "deep_net.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}