data-science-ipython-notebooks/deep-learning/keras-tutorial/0. Preamble.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Credits: Forked from [deep-learning-keras-tensorflow](https://github.com/leriomaggio/deep-learning-keras-tensorflow) by Valerio Maggio"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "<div>\n",
    "    <h1 style=\"text-align: center;\">Deep Learning with Keras</h1>\n",
    "    <img style=\"text-align: left\" src=\"imgs/keras-logo-small.jpg\" width=\"10%\" />\n",
    "<div>\n",
    "\n",
    "<div>\n",
    "    <h2 style=\"text-align: center;\">Tutorial @ EuroScipy 2016</h2>\n",
    "    <img style=\"text-align: left\" src=\"imgs/euroscipy_2016_logo.png\" width=\"40%\" />\n",
    "</div>    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Yam Peleg,  Valerio Maggio"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "# Goal of this Tutorial\n",
    "\n",
    "- **Introduce** main features of Keras\n",
    "- **Learn** how simple and Pythonic is doing Deep Learning with Keras\n",
    "- **Understand** how easy is to do basic and *advanced* DL models in Keras;\n",
    "    - **Examples and Hand-on Excerises** along the way."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Source\n",
    "\n",
    "https://github.com/leriomaggio/deep-learning-keras-euroscipy2016/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# (Tentative) Schedule \n",
    "\n",
    "## Attention: Spoilers Warning!\n",
    "\n",
    "\n",
    "- **Setup** (`10 mins`)\n",
    "\n",
    "- **Part I**: **Introduction** (`~65 mins`)\n",
    "\n",
    "    - Intro to ANN (`~20 mins`)\n",
    "        - naive pure-Python implementation\n",
    "        - fast forward, sgd, backprop\n",
    "        \n",
    "    - Intro to Theano (`15 mins`)\n",
    "        - Model + SGD with Theano\n",
    "        \n",
    "    - Introduction to Keras (`30 mins`)\n",
    "        - Overview and main features\n",
    "            - Theano backend\n",
    "            - Tensorflow backend\n",
    "        - Multi-Layer Perceptron and Fully Connected\n",
    "            - Examples with `keras.models.Sequential` and `Dense`\n",
    "            - HandsOn: MLP with keras\n",
    "            \n",
    "- **Coffe Break** (`30 mins`)\n",
    "\n",
    "- **Part II**: **Supervised Learning and Convolutional Neural Nets** (`~45 mins`)\n",
    "    \n",
    "    - Intro: Focus on Image Classification (`5 mins`)\n",
    "\n",
    "    - Intro to CNN (`25 mins`)\n",
    "        - meaning of convolutional filters\n",
    "            - examples from ImageNet    \n",
    "        - Meaning of dimensions of Conv filters (through an exmple of ConvNet) \n",
    "        - Visualising ConvNets\n",
    "        - HandsOn: ConvNet with keras \n",
    "\n",
    "    - Advanced CNN (`10 mins`)\n",
    "        - Dropout\n",
    "        - MaxPooling\n",
    "        - Batch Normalisation\n",
    "        \n",
    "    - Famous Models in Keras (likely moved somewhere else) (`10 mins`)\n",
    "        (ref: https://github.com/fchollet/deep-learning-models)\n",
    "            - VGG16\n",
    "            - VGG19\n",
    "            - ResNet50\n",
    "            - Inception v3\n",
    "        - HandsOn: Fine tuning a network on new dataset \n",
    "        \n",
    "- **Part III**: **Unsupervised Learning** (`10 mins`)\n",
    "\n",
    "    - AutoEncoders (`5 mins`)\n",
    "    - word2vec & doc2vec (gensim) & `keras.datasets` (`5 mins`)\n",
    "        - `Embedding`\n",
    "        - word2vec and CNN\n",
    "    - Exercises\n",
    "\n",
    "- **Part IV**: **Advanced Materials** (`20 mins`)\n",
    "    - RNN and LSTM (`10 mins`)\n",
    "        -  RNN, LSTM, GRU  \n",
    "    - Example of RNN and LSTM with Text (`~10 mins`) -- *Tentative*\n",
    "    - HandsOn: IMDB\n",
    "\n",
    "- **Wrap up and Conclusions** (`5 mins`)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Requirements"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This tutorial requires the following packages:\n",
    "\n",
    "- Python version 3.4+ \n",
    "    - likely Python 2.7 would be fine, but *who knows*? :P\n",
    "- `numpy` version 1.10 or later: http://www.numpy.org/\n",
    "- `scipy` version 0.16 or later: http://www.scipy.org/\n",
    "- `matplotlib` version 1.4 or later: http://matplotlib.org/\n",
    "- `pandas` version 0.16 or later: http://pandas.pydata.org\n",
    "- `scikit-learn` version 0.15 or later: http://scikit-learn.org\n",
    "- `keras` version 1.0 or later: http://keras.io\n",
    "- `theano` version 0.8 or later: http://deeplearning.net/software/theano/\n",
    "- `ipython`/`jupyter` version 4.0 or later, with notebook support\n",
    "\n",
    "(Optional but recommended):\n",
    "\n",
    "- `pyyaml`\n",
    "- `hdf5` and `h5py` (required if you use model saving/loading functions in keras)\n",
    "- **NVIDIA cuDNN** if you have NVIDIA GPUs on your machines.\n",
    "    [https://developer.nvidia.com/rdp/cudnn-download]()\n",
    "\n",
    "The easiest way to get (most) these is to use an all-in-one installer such as [Anaconda](http://www.continuum.io/downloads) from Continuum. These are available for multiple architectures."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Python Version"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I'm currently running this tutorial with **Python 3** on **Anaconda**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Python 3.5.2\r\n"
     ]
    }
   ],
   "source": [
    "!python --version"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# How to set up your environment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The quickest and simplest way to setup the environment is to use [conda](https://store.continuum.io) environment manager. \n",
    "\n",
    "We provide in the materials a `deep-learning.yml` that is complete and **ready to use** to set up your virtual environment with conda."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name: deep-learning\r\n",
      "channels:\r\n",
      "- conda-forge\r\n",
      "- defaults\r\n",
      "dependencies:\r\n",
      "- accelerate=2.3.0=np111py35_3\r\n",
      "- accelerate_cudalib=2.0=0\r\n",
      "- bokeh=0.12.1=py35_0\r\n",
      "- cffi=1.6.0=py35_0\r\n",
      "- backports.shutil_get_terminal_size=1.0.0=py35_0\r\n",
      "- blas=1.1=openblas\r\n",
      "- ca-certificates=2016.8.2=3\r\n",
      "- cairo=1.12.18=8\r\n",
      "- certifi=2016.8.2=py35_0\r\n",
      "- cycler=0.10.0=py35_0\r\n",
      "- cython=0.24.1=py35_0\r\n",
      "- decorator=4.0.10=py35_0\r\n",
      "- entrypoints=0.2.2=py35_0\r\n",
      "- fontconfig=2.11.1=3\r\n",
      "- freetype=2.6.3=1\r\n",
      "- gettext=0.19.7=1\r\n",
      "- glib=2.48.0=4\r\n",
      "- h5py=2.6.0=np111py35_6\r\n",
      "- harfbuzz=1.0.6=0\r\n",
      "- hdf5=1.8.17=2\r\n",
      "- icu=56.1=4\r\n",
      "- ipykernel=4.3.1=py35_1\r\n",
      "- ipython=5.1.0=py35_0\r\n",
      "- ipywidgets=5.2.2=py35_0\r\n",
      "- jinja2=2.8=py35_1\r\n",
      "- jpeg=9b=0\r\n",
      "- jsonschema=2.5.1=py35_0\r\n",
      "- jupyter_client=4.3.0=py35_0\r\n",
      "- jupyter_console=5.0.0=py35_0\r\n",
      "- jupyter_core=4.1.1=py35_1\r\n",
      "- libffi=3.2.1=2\r\n",
      "- libiconv=1.14=3\r\n",
      "- libpng=1.6.24=0\r\n",
      "- libsodium=1.0.10=0\r\n",
      "- libtiff=4.0.6=6\r\n",
      "- libxml2=2.9.4=0\r\n",
      "- markupsafe=0.23=py35_0\r\n",
      "- matplotlib=1.5.2=np111py35_6\r\n",
      "- mistune=0.7.3=py35_0\r\n",
      "- nbconvert=4.2.0=py35_0\r\n",
      "- nbformat=4.0.1=py35_0\r\n",
      "- ncurses=5.9=8\r\n",
      "- nose=1.3.7=py35_1\r\n",
      "- notebook=4.2.2=py35_0\r\n",
      "- numpy=1.11.1=py35_blas_openblas_201\r\n",
      "- openblas=0.2.18=4\r\n",
      "- openssl=1.0.2h=2\r\n",
      "- pandas=0.18.1=np111py35_1\r\n",
      "- pango=1.40.1=0\r\n",
      "- path.py=8.2.1=py35_0\r\n",
      "- pcre=8.38=1\r\n",
      "- pexpect=4.2.0=py35_1\r\n",
      "- pickleshare=0.7.3=py35_0\r\n",
      "- pip=8.1.2=py35_0\r\n",
      "- pixman=0.32.6=0\r\n",
      "- prompt_toolkit=1.0.6=py35_0\r\n",
      "- protobuf=3.0.0b3=py35_1\r\n",
      "- ptyprocess=0.5.1=py35_0\r\n",
      "- pygments=2.1.3=py35_1\r\n",
      "- pyparsing=2.1.7=py35_0\r\n",
      "- python=3.5.2=2\r\n",
      "- python-dateutil=2.5.3=py35_0\r\n",
      "- pytz=2016.6.1=py35_0\r\n",
      "- pyyaml=3.11=py35_0\r\n",
      "- pyzmq=15.4.0=py35_0\r\n",
      "- qt=4.8.7=0\r\n",
      "- qtconsole=4.2.1=py35_0\r\n",
      "- readline=6.2=0\r\n",
      "- requests=2.11.0=py35_0\r\n",
      "- scikit-learn=0.17.1=np111py35_blas_openblas_201\r\n",
      "- scipy=0.18.0=np111py35_blas_openblas_201\r\n",
      "- setuptools=25.1.6=py35_0\r\n",
      "- simplegeneric=0.8.1=py35_0\r\n",
      "- sip=4.18=py35_0\r\n",
      "- six=1.10.0=py35_0\r\n",
      "- sqlite=3.13.0=1\r\n",
      "- terminado=0.6=py35_0\r\n",
      "- tk=8.5.19=0\r\n",
      "- tornado=4.4.1=py35_1\r\n",
      "- traitlets=4.2.2=py35_0\r\n",
      "- wcwidth=0.1.7=py35_0\r\n",
      "- wheel=0.29.0=py35_0\r\n",
      "- widgetsnbextension=1.2.6=py35_3\r\n",
      "- xz=5.2.2=0\r\n",
      "- yaml=0.1.6=0\r\n",
      "- zeromq=4.1.5=0\r\n",
      "- zlib=1.2.8=3\r\n",
      "- cudatoolkit=7.5=0\r\n",
      "- ipython_genutils=0.1.0=py35_0\r\n",
      "- jupyter=1.0.0=py35_3\r\n",
      "- libgfortran=3.0.0=1\r\n",
      "- llvmlite=0.11.0=py35_0\r\n",
      "- mkl=11.3.3=0\r\n",
      "- mkl-service=1.1.2=py35_2\r\n",
      "- numba=0.26.0=np111py35_0\r\n",
      "- pycparser=2.14=py35_1\r\n",
      "- pyqt=4.11.4=py35_4\r\n",
      "- snakeviz=0.4.1=py35_0\r\n",
      "- pip:\r\n",
      "  - backports.shutil-get-terminal-size==1.0.0\r\n",
      "  - certifi==2016.8.2\r\n",
      "  - cycler==0.10.0\r\n",
      "  - cython==0.24.1\r\n",
      "  - decorator==4.0.10\r\n",
      "  - h5py==2.6.0\r\n",
      "  - ipykernel==4.3.1\r\n",
      "  - ipython==5.1.0\r\n",
      "  - ipython-genutils==0.1.0\r\n",
      "  - ipywidgets==5.2.2\r\n",
      "  - jinja2==2.8\r\n",
      "  - jsonschema==2.5.1\r\n",
      "  - jupyter-client==4.3.0\r\n",
      "  - jupyter-console==5.0.0\r\n",
      "  - jupyter-core==4.1.1\r\n",
      "  - keras==1.0.7\r\n",
      "  - mako==1.0.4\r\n",
      "  - markupsafe==0.23\r\n",
      "  - matplotlib==1.5.2\r\n",
      "  - mistune==0.7.3\r\n",
      "  - nbconvert==4.2.0\r\n",
      "  - nbformat==4.0.1\r\n",
      "  - nose==1.3.7\r\n",
      "  - notebook==4.2.2\r\n",
      "  - numpy==1.11.1\r\n",
      "  - pandas==0.18.1\r\n",
      "  - path.py==8.2.1\r\n",
      "  - pexpect==4.2.0\r\n",
      "  - pickleshare==0.7.3\r\n",
      "  - pip==8.1.2\r\n",
      "  - prompt-toolkit==1.0.6\r\n",
      "  - protobuf==3.0.0b2\r\n",
      "  - ptyprocess==0.5.1\r\n",
      "  - pygments==2.1.3\r\n",
      "  - pygpu==0.2.1\r\n",
      "  - pyparsing==2.1.7\r\n",
      "  - python-dateutil==2.5.3\r\n",
      "  - pytz==2016.6.1\r\n",
      "  - pyyaml==3.11\r\n",
      "  - pyzmq==15.4.0\r\n",
      "  - qtconsole==4.2.1\r\n",
      "  - requests==2.11.0\r\n",
      "  - scikit-learn==0.17.1\r\n",
      "  - scipy==0.18.0\r\n",
      "  - setuptools==25.1.4\r\n",
      "  - simplegeneric==0.8.1\r\n",
      "  - six==1.10.0\r\n",
      "  - tensorflow==0.10.0rc0\r\n",
      "  - terminado==0.6\r\n",
      "  - theano==0.8.2\r\n",
      "  - tornado==4.4.1\r\n",
      "  - traitlets==4.2.2\r\n",
      "  - wcwidth==0.1.7\r\n",
      "  - wheel==0.29.0\r\n",
      "  - widgetsnbextension==1.2.6\r\n",
      "prefix: /home/valerio/anaconda3/envs/deep-learning\r\n",
      "\r\n"
     ]
    }
   ],
   "source": [
    "!cat deep-learning.yml"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Recreate the Conda Environment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### A. Create the Environment\n",
    "\n",
    "```\n",
    "conda env create -f deep-learning.yml  # this file is for Linux channels.\n",
    "```\n",
    "\n",
    "If you're using a **Mac OSX**, we also provided in the repo the conda file \n",
    "that is compatible with `osx-channels`:\n",
    "\n",
    "```\n",
    "conda env create -f deep-learning-osx.yml  # this file is for OSX channels.\n",
    "```\n",
    "\n",
    "#### B. Activate the new `deep-learning` Environment\n",
    "\n",
    "```\n",
    "source activate deep-learning\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Optionals"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Enabling Conda-Forge"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is strongly suggested to enable [**conda forge**](https://conda-forge.github.io/) in your Anaconda installation.\n",
    "\n",
    "**Conda-Forge** is a github organisation containing repositories of conda recipies.\n",
    "\n",
    "To add `conda-forge` as an additional anaconda channel it is just required to type:\n",
    "\n",
    "```shell\n",
    "conda config --add channels conda-forge\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Configure Theano\n",
    "\n",
    "1) Create the `theanorc` file:\n",
    "\n",
    "```shell\n",
    "touch $HOME/.theanorc\n",
    "```\n",
    "\n",
    "2) Copy the following content into the file:\n",
    "\n",
    "```\n",
    "[global]\n",
    "floatX = float32\n",
    "device = gpu  # switch to cpu if no GPU is available on your machine\n",
    "\n",
    "[nvcc]\n",
    "fastmath = True\n",
    "\n",
    "[lib]\n",
    "cnmem=.90\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**More on [theano documentation](http://theano.readthedocs.io/en/latest/library/config.html)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Installing Tensorflow as backend "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```shell\n",
    "# Ubuntu/Linux 64-bit, GPU enabled, Python 3.5\n",
    "# Requires CUDA toolkit 7.5 and CuDNN v4. For other versions, see \"Install from sources\" below.\n",
    "export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow-0.10.0rc0-cp35-cp35m-linux_x86_64.whl\n",
    "\n",
    "pip install --ignore-installed --upgrade $TF_BINARY_URL\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**More on [tensorflow documentation](https://www.tensorflow.org/versions/r0.10/get_started/os_setup.html)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test if everything is up&running"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Check import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import scipy as sp\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n",
      "Using gpu device 0: GeForce GTX 760 (CNMeM is enabled with initial size: 90.0% of memory, cuDNN 4007)\n"
     ]
    }
   ],
   "source": [
    "import keras"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Check installeded Versions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "numpy: 1.11.1\n",
      "scipy: 0.18.0\n",
      "matplotlib: 1.5.2\n",
      "iPython: 5.1.0\n",
      "scikit-learn: 0.17.1\n"
     ]
    }
   ],
   "source": [
    "import numpy\n",
    "print('numpy:', numpy.__version__)\n",
    "\n",
    "import scipy\n",
    "print('scipy:', scipy.__version__)\n",
    "\n",
    "import matplotlib\n",
    "print('matplotlib:', matplotlib.__version__)\n",
    "\n",
    "import IPython\n",
    "print('iPython:', IPython.__version__)\n",
    "\n",
    "import sklearn\n",
    "print('scikit-learn:', sklearn.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "keras:  1.0.7\n",
      "Theano:  0.8.2\n",
      "Tensorflow:  0.10.0rc0\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "print('keras: ', keras.__version__)\n",
    "\n",
    "import theano\n",
    "print('Theano: ', theano.__version__)\n",
    "\n",
    "# optional\n",
    "import tensorflow as tf\n",
    "print('Tensorflow: ', tf.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "<h1 style=\"text-align: center;\">If everything worked till down here, you're ready to start!</h1>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Consulting Material"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You have two options to go through the material presented in this tutorial:\n",
    "\n",
    "* Read (and execute) the material as **iPython/Jupyter** notebooks\n",
    "* (just) read the material as (HTML) slides"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the first case, all you need to do is just execute `ipython notebook` (or `jupyter notebook`) depending on the version of `iPython` you have installed on your machine\n",
    "\n",
    "(`jupyter` command works in case you have `iPython 4.0.x` installed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the second case, you may simply convert the provided notebooks in `HTML` slides and see them into your browser\n",
    "thanks to `nbconvert`.\n",
    "\n",
    "Thus, move to the folder where notebooks are stored and execute the following command:\n",
    "\n",
    "    jupyter nbconvert --to slides ./*.ipynb --post serve\n",
    "    \n",
    "   \n",
    "(Please substitute `jupyter` with `ipython` in the previous command if you have `iPython 3.x` installed on your machine)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## In case..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "..you wanna do **both** (interactive and executable slides), I highly suggest to install the terrific `RISE` ipython notebook extension: [https://github.com/damianavila/RISE](https://github.com/damianavila/RISE)"
   ]
  }
 ],
 "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": "3.4.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}