NN with tensorflow 2.0.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow.keras as keras\n",
    "from tensorflow.keras import layers\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensorflow: 2.0.0, keras: 2.2.4-tf, gpu:  True\n"
     ]
    }
   ],
   "source": [
    "def begin():\n",
    "    print(\"tensorflow: {}, keras: {}, gpu:  {}\".format(tf.__version__, keras.__version__, tf.test.is_gpu_available()))\n",
    "    \n",
    "begin()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def set_cmap(cmap:str):\n",
    "    plt.rcParams['image.cmap'] = 'gray'\n",
    "\n",
    "set_cmap('binary')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = keras.datasets.fashion_mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train, y_train), (x_test, y_test) = data.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((60000, 28, 28), (60000,), (10000, 28, 28), (10000,))"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train.shape, y_train.shape, x_test.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_names = ['T-shirt/top', 'Trauser', 'Pullover', 'Dress',\n",
    "               'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class DatasetTypeUnknoun(Exception):\n",
    "    pass\n",
    "\n",
    "def show_image(img, label, ax=None, figsize=None):\n",
    "    if ax is None: fig, ax = plt.subplots(figsize=figsize)\n",
    "    ax.imshow(img)\n",
    "    ax.set_title(label)\n",
    "    \n",
    "def show_index(index, dataset:str='train', ax=None, figsize:(int, int)=(5, 5)):\n",
    "    if dataset == 'train'  : img, label = x_train[index], y_train[index]\n",
    "    elif dataset == 'valid': img, label = x_test[index], y_test[index]\n",
    "    else: raise DatasetTypeUnknoun \n",
    "    label = class_names[label]\n",
    "    show_image(img, label, ax, figsize)\n",
    "\n",
    "def show_batch(nrows, ncols, dataset:str='train', figsize:(int, int)=(5, 5)):\n",
    "    fig, axs = plt.subplots(nrows, ncols, figsize=(figsize[0]*nrows, figsize[1]*ncols))\n",
    "    if dataset == 'valid': \n",
    "        ids = np.random.permutation(len(x_test))[:nrows*ncols].reshape((nrows, ncols))\n",
    "    else:\n",
    "        ids = np.random.permutation(len(x_train))[:nrows*ncols].reshape((nrows, ncols))\n",
    "    for i in range(nrows):\n",
    "        for j in range(ncols):\n",
    "            show_index(ids[i,j], dataset, axs[i, j])\n",
    "    \n",
    "\n",
    "show_batch(3, 3, 'valid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "72.96981142492712"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(x_train.mean()*60000+x_test.mean()*10000)/70000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Flatten\n",
    "# x_train = x_train.reshape(60_000, 784)\n",
    "# x_test = x_test.reshape(10_000, 784)\n",
    "# x_train.shape, x_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x_train/255.0\n",
    "x_test = x_test/255.0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Simple Model (Dense)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Building the model\n",
    "def build_dense(input_shape, layers_=[256, 128], n_classes=10, drop=0.2, name='Dense'):\n",
    "    model = keras.Sequential(name=name)\n",
    "    model.add(layers.Flatten(input_shape = input_shape))\n",
    "    \n",
    "    for l in layers_:\n",
    "        model.add(layers.Dense(l, activation='relu'))\n",
    "        model.add(layers.Dropout(drop))\n",
    "        \n",
    "    model.add(layers.Dense(n_classes, activation='softmax'))\n",
    "    return model\n",
    "\n",
    "model = build_dense((28, 28), [256, 128], 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compiling the model\n",
    "model.compile(optimizer='sgd', loss='sparse_categorical_crossentropy', metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Dense\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_1 (Flatten)          (None, 784)               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 256)               200960    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 235,146\n",
      "Trainable params: 235,146\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 23s 384us/sample - loss: 0.8317 - accuracy: 0.7128 - val_loss: 0.5661 - val_accuracy: 0.8036\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 11s 187us/sample - loss: 0.5560 - accuracy: 0.8063 - val_loss: 0.4867 - val_accuracy: 0.8276\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 11s 186us/sample - loss: 0.4929 - accuracy: 0.8273 - val_loss: 0.4613 - val_accuracy: 0.8346\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 12s 204us/sample - loss: 0.4567 - accuracy: 0.8393 - val_loss: 0.4317 - val_accuracy: 0.8423\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 12s 196us/sample - loss: 0.4330 - accuracy: 0.8465 - val_loss: 0.4191 - val_accuracy: 0.8486\n"
     ]
    }
   ],
   "source": [
    "# Training the model\n",
    "H = model.fit(x_train, y_train, validation_data=(x_test, y_test), batch_size=32, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_curve(history, metrics=['accuracy', 'loss'], valid:bool=False):\n",
    "    history = history.history\n",
    "    for metric in metrics:\n",
    "        plot_metric(history, metric, valid)\n",
    "            \n",
    "def plot_metric(history, metric, valid, ax=None):\n",
    "    if ax is None:\n",
    "        _, ax = plt.subplots(figsize=(5, 5))\n",
    "        ax.plot(history[metric])\n",
    "        if valid:\n",
    "            ax.plot(history['val_' + metric])\n",
    "        ax.set_title('Model ' + metric.title() + ':')\n",
    "        ax.set_xlabel('Epoch')\n",
    "        ax.set_ylabel(metric.title())\n",
    "        ax.legend(['Train', 'Valid'], loc='upper left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_curve(H, valid=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# LeNet 5 (Convolution)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "def lenet5(verbose=True):\n",
    "    model = keras.Sequential(name='LeNet5')\n",
    "    # 6  16\n",
    "    model.add(layers.Conv2D(filters=25, kernel_size=(5, 5), activation='relu', input_shape=(28,28,1), name='conv1_3x3', padding='valid'))\n",
    "    model.add(layers.AveragePooling2D(name='pool1_2x2'))\n",
    "\n",
    "    model.add(layers.Conv2D(filters=50, kernel_size=(5, 5), activation='relu', name='conv2_3x3'))\n",
    "    model.add(layers.AveragePooling2D(name='pool2_2x2'))\n",
    "\n",
    "    model.add(layers.Flatten())\n",
    "\n",
    "    model.add(layers.Dense(units=120, activation='relu'))\n",
    "    \n",
    "    model.add(layers.Dense(units=84, activation='relu'))\n",
    "\n",
    "    model.add(layers.Dense(units=10, activation = 'softmax'))\n",
    "    if verbose: model.summary()\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"LeNet5\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1_3x3 (Conv2D)           (None, 24, 24, 25)        650       \n",
      "_________________________________________________________________\n",
      "pool1_2x2 (AveragePooling2D) (None, 12, 12, 25)        0         \n",
      "_________________________________________________________________\n",
      "conv2_3x3 (Conv2D)           (None, 8, 8, 50)          31300     \n",
      "_________________________________________________________________\n",
      "pool2_2x2 (AveragePooling2D) (None, 4, 4, 50)          0         \n",
      "_________________________________________________________________\n",
      "flatten_7 (Flatten)          (None, 800)               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 120)               96120     \n",
      "_________________________________________________________________\n",
      "dense_22 (Dense)             (None, 84)                10164     \n",
      "_________________________________________________________________\n",
      "dense_23 (Dense)             (None, 10)                850       \n",
      "=================================================================\n",
      "Total params: 139,084\n",
      "Trainable params: 139,084\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "lenet = lenet5()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((60000, 28, 28, 1), (10000, 28, 28, 1))"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train = x_train.reshape(-1, 28, 28, 1)\n",
    "x_test = x_test.reshape(-1, 28, 28, 1)\n",
    "x_train.shape, x_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 22s 360us/sample - loss: 0.5949 - accuracy: 0.7782 - val_loss: 0.5271 - val_accuracy: 0.8132\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 20s 336us/sample - loss: 0.3854 - accuracy: 0.8608 - val_loss: 0.3558 - val_accuracy: 0.8713\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 20s 337us/sample - loss: 0.3232 - accuracy: 0.8815 - val_loss: 0.3349 - val_accuracy: 0.8762\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 20s 335us/sample - loss: 0.2905 - accuracy: 0.8932 - val_loss: 0.3183 - val_accuracy: 0.8832\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 20s 337us/sample - loss: 0.2676 - accuracy: 0.9015 - val_loss: 0.3322 - val_accuracy: 0.8825\n"
     ]
    }
   ],
   "source": [
    "lenet.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
    "H_lenet = lenet.fit(x_train, y_train, validation_data=(x_test, y_test), batch_size=32, epochs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_curve(H_lenet, valid=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import log_loss\n",
    "from sklearn.preprocessing import OneHotEncoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "def most_confused_(n, model, valid_data, loss_func, return_images=True):\n",
    "    x_true, y_true = valid_data[0], valid_data[1].copy()\n",
    "    y_pred = model.predict(x_true)\n",
    "    label = OneHotEncoder()\n",
    "    y_true = y_true[:, None]\n",
    "    y_true = label.fit_transform(y_true)\n",
    "    losses = loss_func(y_true, y_pred)\n",
    "    sort = np.argsort(losses)\n",
    "    sort = sort[-n:]\n",
    "    y_pred = model.predict_classes(x_true)\n",
    "    y_true = valid_data[1]\n",
    "    if return_images: return x_true[sort], y_true[sort], y_pred[sort], losses[sort]\n",
    "    return sort, y_pred[sort], losses[sort]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cross_entropy(y_true, y_pred):\n",
    "    return -np.array([t.dot(np.log(1/p))[0] for t, p in zip(y_true, y_pred)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [],
   "source": [
    "# x, t, p, l = most_confused_(5, model, (x_test, y_test), cross_entropy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "# p.shape, t.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "# class_names[t[0]], class_names[p[0]], l[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plt.imshow(x[0])\n",
    "# plt.title('true: ' + class_names[t[0]] + ', predicted: ' + class_names[p[0]] + ', loss: ' + str(l[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_image(img, t_label, p_label, loss, ax=None, figsize=None):\n",
    "    if ax is None: fig, ax = plt.subplots(figsize=figsize)\n",
    "    ax.imshow(img)\n",
    "    ax.set_title('true: ' + class_names[t_label] + ', predicted: ' + class_names[p_label] + ', loss: ' + str(loss))\n",
    "\n",
    "def show_batch(nrows, ncols, images, t_labels, p_labels, loss, figsize:(int, int)=(5, 5)):\n",
    "    fig, axs = plt.subplots(nrows, ncols, figsize=(figsize[0]*nrows, figsize[1]*ncols))\n",
    "    k = 0\n",
    "    for i in range(nrows):\n",
    "        for j in range(ncols):\n",
    "            show_image(images[k], t_labels[k], p_labels[k], loss[k], axs[i, j])\n",
    "            k += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_most_confused(n, model, valid_data, most_confused=None):\n",
    "    if most_confused is None:\n",
    "        x, t, p, l = most_confused_(n, model, valid_data, cross_entropy)\n",
    "        print(l)\n",
    "    else:\n",
    "        x, t, p, l = most_confused\n",
    "    nrows, ncols = int(np.ceil(n//3)), 3\n",
    "    show_batch(nrows, ncols, x, t, p, l)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Rashid Haffadi\\.conda\\envs\\tensorflow\\lib\\site-packages\\ipykernel_launcher.py:2: RuntimeWarning: divide by zero encountered in true_divide\n",
      "  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0. -0. -0. -0. -0. -0. -0. -0. -0.]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_most_confused(9, model, (x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture\n",
    "loss, accu = lenet.evaluate(x_test[0, None], y_test[0, None])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0003716255014296621"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36m_get_op_def\u001b[1;34m(self, type)\u001b[0m\n\u001b[0;32m   3819\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3820\u001b[1;33m       \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_op_def_cache\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtype\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3821\u001b[0m     \u001b[1;32mexcept\u001b[0m \u001b[0mKeyError\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyError\u001b[0m: 'Const'",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\IPython\\core\\interactiveshell.py\u001b[0m in \u001b[0;36mrun_code\u001b[1;34m(self, code_obj, result, async_)\u001b[0m\n\u001b[0;32m   3318\u001b[0m                 \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3319\u001b[1;33m                     \u001b[0mexec\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcode_obj\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0muser_global_ns\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0muser_ns\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3320\u001b[0m             \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-66-6640cc3df928>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      2\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx_test\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 3\u001b[1;33m     \u001b[0mloss\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maccur\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mlenet\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mevaluate\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx_test\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_test\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      4\u001b[0m     \u001b[0mlosses\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mloss\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mevaluate\u001b[1;34m(self, x, y, batch_size, verbose, sample_weight, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[0;32m    832\u001b[0m         \u001b[0mworkers\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mworkers\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 833\u001b[1;33m         use_multiprocessing=use_multiprocessing)\n\u001b[0m\u001b[0;32m    834\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\training_v2.py\u001b[0m in \u001b[0;36mevaluate\u001b[1;34m(self, model, x, y, batch_size, verbose, sample_weight, steps, callbacks, **kwargs)\u001b[0m\n\u001b[0;32m    455\u001b[0m         \u001b[0mmodel\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mModeKeys\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mTEST\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mbatch_size\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mverbose\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 456\u001b[1;33m         sample_weight=sample_weight, steps=steps, callbacks=callbacks, **kwargs)\n\u001b[0m\u001b[0;32m    457\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\training_v2.py\u001b[0m in \u001b[0;36m_model_iteration\u001b[1;34m(self, model, mode, x, y, batch_size, verbose, sample_weight, steps, callbacks, **kwargs)\u001b[0m\n\u001b[0;32m    395\u001b[0m           \u001b[0msteps\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0msteps\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 396\u001b[1;33m           distribution_strategy=strategy)\n\u001b[0m\u001b[0;32m    397\u001b[0m       \u001b[0mtotal_samples\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0m_get_total_number_of_samples\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0madapter\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\training_v2.py\u001b[0m in \u001b[0;36m_process_inputs\u001b[1;34m(model, x, y, batch_size, epochs, sample_weights, class_weights, shuffle, steps, distribution_strategy, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[0;32m    605\u001b[0m       \u001b[0mworkers\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mworkers\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 606\u001b[1;33m       use_multiprocessing=use_multiprocessing)\n\u001b[0m\u001b[0;32m    607\u001b[0m   \u001b[1;31m# As a fallback for the data type that does not work with\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\data_adapter.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, x, y, sample_weights, batch_size, epochs, steps, shuffle, **kwargs)\u001b[0m\n\u001b[0;32m    317\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 318\u001b[1;33m     \u001b[0mindices_dataset\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mindices_dataset\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mflat_map\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mslice_batch_indices\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    319\u001b[0m     dataset = dataset_ops.DatasetV2.zip((\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\data\\ops\\dataset_ops.py\u001b[0m in \u001b[0;36mflat_map\u001b[1;34m(self, map_func)\u001b[0m\n\u001b[0;32m   1239\u001b[0m     \"\"\"\n\u001b[1;32m-> 1240\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0mFlatMapDataset\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmap_func\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1241\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\data\\ops\\dataset_ops.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, input_dataset, map_func)\u001b[0m\n\u001b[0;32m   3485\u001b[0m     self._map_func = StructuredFunctionWrapper(\n\u001b[1;32m-> 3486\u001b[1;33m         map_func, self._transformation_name(), dataset=input_dataset)\n\u001b[0m\u001b[0;32m   3487\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_map_func\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0moutput_structure\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mDatasetSpec\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\data\\ops\\dataset_ops.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, func, transformation_name, dataset, input_classes, input_shapes, input_types, input_structure, add_to_graph, use_legacy_function, defun_kwargs)\u001b[0m\n\u001b[0;32m   2694\u001b[0m       \u001b[1;32mwith\u001b[0m \u001b[0mtracking\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mresource_tracker_scope\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mresource_tracker\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2695\u001b[1;33m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_function\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mwrapper_fn\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_get_concrete_function_internal\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2696\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0madd_to_graph\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\eager\\function.py\u001b[0m in \u001b[0;36m_get_concrete_function_internal\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1853\u001b[0m     graph_function = self._get_concrete_function_internal_garbage_collected(\n\u001b[1;32m-> 1854\u001b[1;33m         *args, **kwargs)\n\u001b[0m\u001b[0;32m   1855\u001b[0m     \u001b[1;31m# We're returning this concrete function to someone, and they may keep a\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\eager\\function.py\u001b[0m in \u001b[0;36m_get_concrete_function_internal_garbage_collected\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1847\u001b[0m       \u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1848\u001b[1;33m     \u001b[0mgraph_function\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_maybe_define_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1849\u001b[0m     \u001b[1;32mreturn\u001b[0m \u001b[0mgraph_function\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\eager\\function.py\u001b[0m in \u001b[0;36m_maybe_define_function\u001b[1;34m(self, args, kwargs)\u001b[0m\n\u001b[0;32m   2149\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mgraph_function\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2150\u001b[1;33m           \u001b[0mgraph_function\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_create_graph_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2151\u001b[0m           \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_function_cache\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mprimary\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mcache_key\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mgraph_function\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\eager\\function.py\u001b[0m in \u001b[0;36m_create_graph_function\u001b[1;34m(self, args, kwargs, override_flat_arg_shapes)\u001b[0m\n\u001b[0;32m   2040\u001b[0m             \u001b[0moverride_flat_arg_shapes\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0moverride_flat_arg_shapes\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2041\u001b[1;33m             capture_by_value=self._capture_by_value),\n\u001b[0m\u001b[0;32m   2042\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_function_attributes\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\func_graph.py\u001b[0m in \u001b[0;36mfunc_graph_from_py_func\u001b[1;34m(name, python_func, args, kwargs, signature, func_graph, autograph, autograph_options, add_control_dependencies, arg_names, op_return_value, collections, capture_by_value, override_flat_arg_shapes)\u001b[0m\n\u001b[0;32m    914\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 915\u001b[1;33m       \u001b[0mfunc_outputs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mpython_func\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0mfunc_args\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mfunc_kwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    916\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\data\\ops\\dataset_ops.py\u001b[0m in \u001b[0;36mwrapper_fn\u001b[1;34m(*args)\u001b[0m\n\u001b[0;32m   2688\u001b[0m       \u001b[1;32mdef\u001b[0m \u001b[0mwrapper_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m  \u001b[1;31m# pylint: disable=missing-docstring\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2689\u001b[1;33m         \u001b[0mret\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0m_wrapper_helper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2690\u001b[0m         \u001b[0mret\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mstructure\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mto_tensor_list\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_output_structure\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mret\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\data\\ops\\dataset_ops.py\u001b[0m in \u001b[0;36m_wrapper_helper\u001b[1;34m(*args)\u001b[0m\n\u001b[0;32m   2633\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2634\u001b[1;33m       \u001b[0mret\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mautograph\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtf_convert\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfunc\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mag_ctx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0mnested_args\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2635\u001b[0m       \u001b[1;31m# If `func` returns a list of tensors, `nest.flatten()` and\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\autograph\\impl\\api.py\u001b[0m in \u001b[0;36mwrapper\u001b[1;34m(*args, **kwargs)\u001b[0m\n\u001b[0;32m    233\u001b[0m       \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 234\u001b[1;33m         \u001b[1;32mreturn\u001b[0m \u001b[0mconverted_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    235\u001b[0m       \u001b[1;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m  \u001b[1;31m# pylint:disable=broad-except\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\autograph\\impl\\api.py\u001b[0m in \u001b[0;36mconverted_call\u001b[1;34m(f, options, args, kwargs, caller_fn_scope)\u001b[0m\n\u001b[0;32m    438\u001b[0m   \u001b[1;32mif\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0moptions\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0muser_requested\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mconversion\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mis_whitelisted_for_graph\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mf\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 439\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0m_call_unconverted\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    440\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\autograph\\impl\\api.py\u001b[0m in \u001b[0;36m_call_unconverted\u001b[1;34m(f, args, kwargs, options, update_cache)\u001b[0m\n\u001b[0;32m    329\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mkwargs\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 330\u001b[1;33m       \u001b[1;32mreturn\u001b[0m \u001b[0mf\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    331\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\keras\\engine\\data_adapter.py\u001b[0m in \u001b[0;36mslice_batch_indices\u001b[1;34m(indices)\u001b[0m\n\u001b[0;32m    306\u001b[0m       \u001b[0mnum_in_full_batch\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnum_full_batches\u001b[0m \u001b[1;33m*\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 307\u001b[1;33m       \u001b[0mfirst_k_indices\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0marray_ops\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mslice\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mindices\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m[\u001b[0m\u001b[0mnum_in_full_batch\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    308\u001b[0m       first_k_indices = array_ops.reshape(\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\ops\\array_ops.py\u001b[0m in \u001b[0;36mslice\u001b[1;34m(input_, begin, size, name)\u001b[0m\n\u001b[0;32m    865\u001b[0m   \"\"\"\n\u001b[1;32m--> 866\u001b[1;33m   \u001b[1;32mreturn\u001b[0m \u001b[0mgen_array_ops\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_slice\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minput_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbegin\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msize\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    867\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\ops\\gen_array_ops.py\u001b[0m in \u001b[0;36m_slice\u001b[1;34m(input, begin, size, name)\u001b[0m\n\u001b[0;32m   9223\u001b[0m   _, _, _op = _op_def_lib._apply_op_helper(\n\u001b[1;32m-> 9224\u001b[1;33m         \"Slice\", input=input, begin=begin, size=size, name=name)\n\u001b[0m\u001b[0;32m   9225\u001b[0m   \u001b[0m_result\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0m_op\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\op_def_library.py\u001b[0m in \u001b[0;36m_apply_op_helper\u001b[1;34m(self, op_type_name, name, **keywords)\u001b[0m\n\u001b[0;32m    526\u001b[0m                 \u001b[0mas_ref\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0minput_arg\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mis_ref\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 527\u001b[1;33m                 preferred_dtype=default_dtype)\n\u001b[0m\u001b[0;32m    528\u001b[0m           \u001b[1;32mexcept\u001b[0m \u001b[0mTypeError\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0merr\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36minternal_convert_to_tensor\u001b[1;34m(value, dtype, name, as_ref, preferred_dtype, ctx, accept_composite_tensors)\u001b[0m\n\u001b[0;32m   1295\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mret\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1296\u001b[1;33m       \u001b[0mret\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mconversion_func\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mvalue\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mas_ref\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mas_ref\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1297\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\constant_op.py\u001b[0m in \u001b[0;36m_constant_tensor_conversion_function\u001b[1;34m(v, dtype, name, as_ref)\u001b[0m\n\u001b[0;32m    285\u001b[0m   \u001b[0m_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mas_ref\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 286\u001b[1;33m   \u001b[1;32mreturn\u001b[0m \u001b[0mconstant\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mv\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    287\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\constant_op.py\u001b[0m in \u001b[0;36mconstant\u001b[1;34m(value, dtype, shape, name)\u001b[0m\n\u001b[0;32m    226\u001b[0m   return _constant_impl(value, dtype, shape, name, verify_shape=False,\n\u001b[1;32m--> 227\u001b[1;33m                         allow_broadcast=True)\n\u001b[0m\u001b[0;32m    228\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\constant_op.py\u001b[0m in \u001b[0;36m_constant_impl\u001b[1;34m(value, dtype, shape, name, verify_shape, allow_broadcast)\u001b[0m\n\u001b[0;32m    270\u001b[0m              \"dtype\": dtype_value},\n\u001b[1;32m--> 271\u001b[1;33m       name=name).outputs[0]\n\u001b[0m\u001b[0;32m    272\u001b[0m   \u001b[1;32mreturn\u001b[0m \u001b[0mconst_tensor\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\func_graph.py\u001b[0m in \u001b[0;36mcreate_op\u001b[1;34m(***failed resolving arguments***)\u001b[0m\n\u001b[0;32m    547\u001b[0m         \u001b[0mop_type\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minputs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtypes\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minput_types\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mattrs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mop_def\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 548\u001b[1;33m         compute_device)\n\u001b[0m\u001b[0;32m    549\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36m_create_op_internal\u001b[1;34m(self, op_type, inputs, dtypes, input_types, name, attrs, op_def, compute_device)\u001b[0m\n\u001b[0;32m   3428\u001b[0m           \u001b[0moriginal_op\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_default_original_op\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3429\u001b[1;33m           op_def=op_def)\n\u001b[0m\u001b[0;32m   3430\u001b[0m       \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_create_op_helper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mret\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcompute_device\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mcompute_device\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, node_def, g, inputs, output_types, control_inputs, input_types, original_op, op_def)\u001b[0m\n\u001b[0;32m   1766\u001b[0m       \u001b[1;32mif\u001b[0m \u001b[0mop_def\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1767\u001b[1;33m         \u001b[0mop_def\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_graph\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_get_op_def\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnode_def\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mop\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1768\u001b[0m       \u001b[1;31m# TODO(skyewm): op_def_library.apply_op() flattens the incoming inputs.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36m_get_op_def\u001b[1;34m(self, type)\u001b[0m\n\u001b[0;32m   3825\u001b[0m         \u001b[1;31m# pylint: enable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3826\u001b[1;33m         \u001b[0mdata\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mc_api\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbuf\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3827\u001b[0m       \u001b[0mop_def\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mop_def_pb2\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mOpDef\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\contextlib.py\u001b[0m in \u001b[0;36m__exit__\u001b[1;34m(self, type, value, traceback)\u001b[0m\n\u001b[0;32m     87\u001b[0m             \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 88\u001b[1;33m                 \u001b[0mnext\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mgen\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     89\u001b[0m             \u001b[1;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\tensorflow_core\\python\\framework\\c_api_util.py\u001b[0m in \u001b[0;36mtf_buffer\u001b[1;34m(data)\u001b[0m\n\u001b[0;32m    172\u001b[0m   \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 173\u001b[1;33m     \u001b[0mc_api\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mTF_DeleteBuffer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbuf\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    174\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: ",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m~\\.conda\\envs\\tensorflow\\lib\\site-packages\\IPython\\core\\interactiveshell.py\u001b[0m in \u001b[0;36mrun_code\u001b[1;34m(self, code_obj, result, async_)\u001b[0m\n\u001b[0;32m   3334\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mresult\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   3335\u001b[0m                 \u001b[0mresult\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0merror_in_exec\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0msys\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexc_info\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3336\u001b[1;33m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshowtraceback\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mrunning_compiled_code\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3337\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   3338\u001b[0m             \u001b[0moutflag\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mFalse\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# Danger !!!!!\n",
    "%%capture\n",
    "losses = []\n",
    "for i in range(len(x_test)):\n",
    "    loss, accur = lenet.evaluate(x_test[i, None], y_test[i, None])\n",
    "    losses.append(loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}