inspect_keras_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n"
     ]
    }
   ],
   "source": [
    "from ds_utils.imports import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((150, 4), (150,), (150, 3))"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "iris = sklearn.datasets.load_iris()\n",
    "X = iris.data\n",
    "y = iris.target\n",
    "y_cat = keras.utils.to_categorical(y)\n",
    "\n",
    "X.shape, y.shape, y_cat.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = keras.models.Sequential()\n",
    "model.add(keras.layers.Dense(3, activation='relu', input_dim=X.shape[1]))\n",
    "model.add(keras.layers.Dense(5, activation='relu'))\n",
    "model.add(keras.layers.Dense(y_cat.shape[1], activation='softmax'))\n",
    "model.compile(optimizer='rmsprop',\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f18fd918748>"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X, y_cat, epochs=10, batch_size=1, verbose=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_16 (Dense)             (None, 3)                 15        \n",
      "_________________________________________________________________\n",
      "dense_17 (Dense)             (None, 5)                 20        \n",
      "_________________________________________________________________\n",
      "dense_18 (Dense)             (None, 3)                 18        \n",
      "=================================================================\n",
      "Total params: 53.0\n",
      "Trainable params: 53\n",
      "Non-trainable params: 0.0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "layers = model.layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "layer = layers[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'activation': 'softmax',\n",
       " 'activity_regularizer': None,\n",
       " 'bias_constraint': None,\n",
       " 'bias_initializer': {'class_name': 'VarianceScaling',\n",
       "  'config': {'distribution': 'uniform',\n",
       "   'mode': 'fan_avg',\n",
       "   'scale': 1.0,\n",
       "   'seed': None}},\n",
       " 'bias_regularizer': None,\n",
       " 'kernel_constraint': None,\n",
       " 'kernel_initializer': {'class_name': 'VarianceScaling',\n",
       "  'config': {'distribution': 'uniform',\n",
       "   'mode': 'fan_avg',\n",
       "   'scale': 1.0,\n",
       "   'seed': None}},\n",
       " 'kernel_regularizer': None,\n",
       " 'name': 'dense_18',\n",
       " 'trainable': True,\n",
       " 'units': 3,\n",
       " 'use_bias': True}"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer.get_config()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([[ 0.42792454, -0.04814534,  0.12542209],\n",
       "        [-1.84589791, -0.35244051, -0.14400582],\n",
       "        [ 0.80219966, -0.26786217, -1.55027366],\n",
       "        [-0.21194243, -1.04536021,  0.23438835],\n",
       "        [ 0.92551601, -1.07620263, -0.87460256]], dtype=float32),\n",
       " array([ 0.54254174,  0.04203929, -0.41321945], dtype=float32)]"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "layer.get_weights()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "layer."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'class_name': 'Dense',\n",
       "  'config': {'activation': 'relu',\n",
       "   'activity_regularizer': None,\n",
       "   'batch_input_shape': (None, 4),\n",
       "   'bias_constraint': None,\n",
       "   'bias_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'bias_regularizer': None,\n",
       "   'dtype': 'float32',\n",
       "   'kernel_constraint': None,\n",
       "   'kernel_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'kernel_regularizer': None,\n",
       "   'name': 'dense_16',\n",
       "   'trainable': True,\n",
       "   'units': 3,\n",
       "   'use_bias': True}},\n",
       " {'class_name': 'Dense',\n",
       "  'config': {'activation': 'relu',\n",
       "   'activity_regularizer': None,\n",
       "   'bias_constraint': None,\n",
       "   'bias_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'bias_regularizer': None,\n",
       "   'kernel_constraint': None,\n",
       "   'kernel_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'kernel_regularizer': None,\n",
       "   'name': 'dense_17',\n",
       "   'trainable': True,\n",
       "   'units': 5,\n",
       "   'use_bias': True}},\n",
       " {'class_name': 'Dense',\n",
       "  'config': {'activation': 'softmax',\n",
       "   'activity_regularizer': None,\n",
       "   'bias_constraint': None,\n",
       "   'bias_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'bias_regularizer': None,\n",
       "   'kernel_constraint': None,\n",
       "   'kernel_initializer': {'class_name': 'VarianceScaling',\n",
       "    'config': {'distribution': 'uniform',\n",
       "     'mode': 'fan_avg',\n",
       "     'scale': 1.0,\n",
       "     'seed': None}},\n",
       "   'kernel_regularizer': None,\n",
       "   'name': 'dense_18',\n",
       "   'trainable': True,\n",
       "   'units': 3,\n",
       "   'use_bias': True}}]"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "config = model.get_config()\n",
    "config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>distribution</th>\n",
       "      <th>mode</th>\n",
       "      <th>scale</th>\n",
       "      <th>seed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  distribution     mode  scale  seed\n",
       "0      uniform  fan_avg    1.0  None\n",
       "1      uniform  fan_avg    1.0  None\n",
       "2      uniform  fan_avg    1.0  None"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([layer['config']['kernel_initializer']['config'] for layer in config])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>distribution</th>\n",
       "      <th>mode</th>\n",
       "      <th>scale</th>\n",
       "      <th>seed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>uniform</td>\n",
       "      <td>fan_avg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  distribution     mode  scale  seed\n",
       "0      uniform  fan_avg    1.0  None\n",
       "1      uniform  fan_avg    1.0  None\n",
       "2      uniform  fan_avg    1.0  None"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([layer['config']['bias_initializer']['config'] for layer in config])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>activation</th>\n",
       "      <th>activity_regularizer</th>\n",
       "      <th>batch_input_shape</th>\n",
       "      <th>bias_constraint</th>\n",
       "      <th>bias_initializer</th>\n",
       "      <th>bias_regularizer</th>\n",
       "      <th>dtype</th>\n",
       "      <th>kernel_constraint</th>\n",
       "      <th>kernel_initializer</th>\n",
       "      <th>kernel_regularizer</th>\n",
       "      <th>name</th>\n",
       "      <th>trainable</th>\n",
       "      <th>units</th>\n",
       "      <th>use_bias</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>relu</td>\n",
       "      <td>None</td>\n",
       "      <td>(None, 4)</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>float32</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>dense_1</td>\n",
       "      <td>True</td>\n",
       "      <td>32</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>relu</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>dense_2</td>\n",
       "      <td>True</td>\n",
       "      <td>16</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>softmax</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "      <td>None</td>\n",
       "      <td>{'class_name': 'VarianceScaling', 'config': {'...</td>\n",
       "      <td>None</td>\n",
       "      <td>dense_3</td>\n",
       "      <td>True</td>\n",
       "      <td>3</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  activation activity_regularizer batch_input_shape bias_constraint  \\\n",
       "0       relu                 None         (None, 4)            None   \n",
       "1       relu                 None               NaN            None   \n",
       "2    softmax                 None               NaN            None   \n",
       "\n",
       "                                    bias_initializer bias_regularizer  \\\n",
       "0  {'class_name': 'VarianceScaling', 'config': {'...             None   \n",
       "1  {'class_name': 'VarianceScaling', 'config': {'...             None   \n",
       "2  {'class_name': 'VarianceScaling', 'config': {'...             None   \n",
       "\n",
       "     dtype kernel_constraint  \\\n",
       "0  float32              None   \n",
       "1      NaN              None   \n",
       "2      NaN              None   \n",
       "\n",
       "                                  kernel_initializer kernel_regularizer  \\\n",
       "0  {'class_name': 'VarianceScaling', 'config': {'...               None   \n",
       "1  {'class_name': 'VarianceScaling', 'config': {'...               None   \n",
       "2  {'class_name': 'VarianceScaling', 'config': {'...               None   \n",
       "\n",
       "      name trainable  units use_bias  \n",
       "0  dense_1      True     32     True  \n",
       "1  dense_2      True     16     True  \n",
       "2  dense_3      True      3     True  "
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([layer['config'] for layer in config])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(4, 3), (3,), (3, 5), (5,), (5, 3), (3,)]"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[w.shape for w in weights]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'name': 'dense_16', 'trainable': True, 'batch_input_shape': (None, 4), 'dtype': 'float32', 'units': 3, 'activation': 'relu', 'use_bias': True, 'kernel_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'bias_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'kernel_regularizer': None, 'bias_regularizer': None, 'activity_regularizer': None, 'kernel_constraint': None, 'bias_constraint': None}\n",
      "[array([[-0.13858747, -0.14601076, -0.21684699],\n",
      "       [-0.78277177, -0.50497305, -0.34725636],\n",
      "       [ 0.20980954,  0.8297739 , -0.58691067],\n",
      "       [-0.07665556,  0.93628305, -0.76365799]], dtype=float32), array([ 0.        ,  0.03256533,  0.        ], dtype=float32)]\n",
      "{'name': 'dense_17', 'trainable': True, 'units': 5, 'activation': 'relu', 'use_bias': True, 'kernel_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'bias_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'kernel_regularizer': None, 'bias_regularizer': None, 'activity_regularizer': None, 'kernel_constraint': None, 'bias_constraint': None}\n",
      "[array([[-0.19894423, -0.36462492,  0.10011253, -0.79210269,  0.10489256],\n",
      "       [-0.00242784,  1.08594978, -0.51033384,  0.12644957, -0.85477775],\n",
      "       [ 0.55898291,  0.06260493, -0.29105553, -0.64320433, -0.51722598]], dtype=float32), array([-0.00316228, -0.01347832,  0.81724524, -0.24956909,  0.81557822], dtype=float32)]\n",
      "{'name': 'dense_18', 'trainable': True, 'units': 3, 'activation': 'softmax', 'use_bias': True, 'kernel_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'bias_initializer': {'class_name': 'VarianceScaling', 'config': {'scale': 1.0, 'mode': 'fan_avg', 'distribution': 'uniform', 'seed': None}}, 'kernel_regularizer': None, 'bias_regularizer': None, 'activity_regularizer': None, 'kernel_constraint': None, 'bias_constraint': None}\n",
      "[array([[ 0.42792454, -0.04814534,  0.12542209],\n",
      "       [-1.84589791, -0.35244051, -0.14400582],\n",
      "       [ 0.80219966, -0.26786217, -1.55027366],\n",
      "       [-0.21194243, -1.04536021,  0.23438835],\n",
      "       [ 0.92551601, -1.07620263, -0.87460256]], dtype=float32), array([ 0.54254174,  0.04203929, -0.41321945], dtype=float32)]\n"
     ]
    }
   ],
   "source": [
    "for layer in model.layers:\n",
    "    print(layer.get_config())\n",
    "    print(layer.get_weights())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.13858747, -0.14601076, -0.21684699],\n",
       "       [-0.78277177, -0.50497305, -0.34725636],\n",
       "       [ 0.20980954,  0.8297739 , -0.58691067],\n",
       "       [-0.07665556,  0.93628305, -0.76365799]], dtype=float32)"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-3.1680950567126276"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.dot(X[0], weights[0][:,0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-3.16809506, -1.16312047, -3.29572343])"
      ]
     },
     "execution_count": 129,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.dot(X[0], weights[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "assert np.dot(X[0], weights[0][:,0]) == (np.dot(X[0], weights[0]))[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.0"
  },
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "86px",
    "width": "252px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": true,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}