tutorial_presentation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# <center> Accelerated Deep Learning with PyTorch\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "<center> **[Wafa Louhichi]()**, [Michael Isaev](), [Tom Kurfess](http://www.me.gatech.edu/faculty/kurfess) and [Rich Vuduc](https://vuduc.org)\n",
    "\n",
    "<center> [Jupyter Day Atlanta (2018)](https://atl-jugheads.github.io/jupyter-day-atlanta-ii/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "This tutorial shows you how to train a deep neural network using [PyTorch](http://pytorch.org/). PyTorch is a Python package aimed at accelerating deep learning applications, building on the [Torch](http://torch.ch/) library. PyTorch provides a [Numpy](http://www.numpy.org/)-like abstraction for representing _tensors_, or multiway arrays, and it can take advantage of [GPUs](http://www.nvidia.com/object/what-is-gpu-computing.html) for performance. The tutorial ends with a short case study, which is how one of us [Wafa]() applied PyTorch to automate the inspection of surface finish quality in manufacturing.\n",
    "\n",
    "> Besides PyTorch, there are numerous proposed tools and extensions to get GPU acceleration for multiway arrays. See **Suggested Next Steps** at the end of this tutorial for pointers."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "**Knowledge prerequisites.** This tutorial assumes familiarity with Python and Numpy, for starters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "From there, PyTorch makes it easy to get started with deep learning even if your background in the topic is modest. At a minimum, it's helpful to know that a multilayer neural network model may be viewed as a graph of nodes (values and functions on values) connected by weights (the unknown parameters of the model), where one wishes to estimate the weights from data using an optimization procedure (think: gradient computations!) based on forward and backward propagation.\n",
    "\n",
    "**Software prerequisites.** You'll need to install PyTorch before running this notebook. Here's a code cell to verify that you are ready to go."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "import torch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "**Hardware niceties.** Lastly, to exploit GPUs, you'll need an NVIDIA GPU with the CUDA SDK installed. It is reported that 10-100$\\times$ speedups are possible by doing so. Of course, if you don't have such a setup, PyTorch still can run using the CPU only. But remember, when it comes to training neural network models, life is short---so you should use GPUs if you can!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## **Presentation Outline.**\n",
    "\n",
    "### 1. Essential PyTorch Background\n",
    "### 2. PyTorch for Data Analytics\n",
    "### 3. LeNet Convolutional Neural Network (CNN) in PyTorch \n",
    "### 4. Application to a Manufacturing Problem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 1. Essential PyTorch background\n",
    "\n",
    "* PyTorch is a Python package aimed at accelerating deep learning applications, building on the [Torch](http://torch.ch/) library. \n",
    "* PyTorch provides a [Numpy](http://www.numpy.org/)-like abstraction for representing _tensors_, or multiway arrays, and it can take advantage of [GPUs](http://www.nvidia.com/object/what-is-gpu-computing.html) for performance.\n",
    "\n",
    "<img src=\"images/tensor.png\" style=\"max-width:100%; width: 75%; max-width: none\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 1.1 PyTorch Tensors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "The key data abstraction of PyTorch is a tensor, which is a multidimensional array. It is similar in functionality to Numpy's `ndarray` object. Use [torch.Tensor()](http://pytorch.org/docs/master/tensors.html) to create one."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "type:  <class 'torch.FloatTensor'>  and size:  torch.Size([10, 20])\n"
     ]
    }
   ],
   "source": [
    "# Generate a 2-D pytorch tensor (i.e., a matrix)\n",
    "pytorch_tensor = torch.Tensor(10, 20)\n",
    "print(\"type: \", type(pytorch_tensor), \" and size: \", pytorch_tensor.shape )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "If you need a Numpy-compatible representation, or if you want to create a PyTorch tensor from an existing Numpy object, it's easy to do."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "type:  <class 'numpy.ndarray'>  and size:  (10, 20)\n",
      "type:  <class 'torch.FloatTensor'>  and size:  torch.Size([10, 20])\n"
     ]
    }
   ],
   "source": [
    "# Convert the pytorch tensor to a numpy array:\n",
    "numpy_tensor = pytorch_tensor.numpy()\n",
    "print(\"type: \", type(numpy_tensor), \" and size: \", numpy_tensor.shape)\n",
    "\n",
    "# Convert the numpy array to Pytorch Tensor:\n",
    "print(\"type: \", type(torch.Tensor(numpy_tensor)), \" and size: \", torch.Tensor(numpy_tensor).shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 1.2 PyTorch vs. NumPy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "PyTorch is not a drop in replacement for NumPy, but it implements a lot of Numpy functionality. One inconvenience is it's naming scheme that sometimes is rather different from Numpy. Let's go over several examples to see the difference. Let's look at PyTorch and NumPy differences on various examples:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "#### 1. Tensor creation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "t = torch.rand(2, 4, 3, 5)\n",
    "a = np.random.rand(2, 4, 3, 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Tensor with size (3, 4):\n",
      " \n",
      " 0.2687  0.4535  0.1562  0.1470\n",
      " 0.6824  0.5180  0.2928  0.4175\n",
      " 0.2522  0.0826  0.7987  0.1032\n",
      "[torch.FloatTensor of size 3x4]\n",
      "\n",
      "Random NdArray with size (3, 4):\n",
      " [[ 0.9221226   0.54427845  0.39084139  0.11807335]\n",
      " [ 0.58825753  0.06241606  0.89243157  0.33282672]\n",
      " [ 0.16356127  0.74073636  0.8937606   0.25118884]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print('Random Tensor with size (3, 4):\\n', torch.rand(3, 4))\n",
    "print(\"Random NdArray with size (3, 4):\\n\", np.random.rand(3, 4))\n",
    "print('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "You can see other initializations schemes like zeros, ones, and identity matrices below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor of zeros with size (3, 4):\n",
      " \n",
      " 0  0  0  0\n",
      " 0  0  0  0\n",
      " 0  0  0  0\n",
      "[torch.FloatTensor of size 3x4]\n",
      "\n",
      "NdArray of zeros with size (3, 4):\n",
      " [[ 0.  0.  0.  0.]\n",
      " [ 0.  0.  0.  0.]\n",
      " [ 0.  0.  0.  0.]]\n",
      "-------------------------------------------------------------------------\n",
      "Tensor of ones with size (3, 4):\n",
      " \n",
      " 1  1  1  1\n",
      " 1  1  1  1\n",
      " 1  1  1  1\n",
      "[torch.FloatTensor of size 3x4]\n",
      "\n",
      "NdArray of ones with size (3, 4):\n",
      " [[ 1.  1.  1.  1.]\n",
      " [ 1.  1.  1.  1.]\n",
      " [ 1.  1.  1.  1.]]\n",
      "-------------------------------------------------------------------------\n",
      "Identity Tensor with size (3, 4):\n",
      " \n",
      " 1  0  0  0\n",
      " 0  1  0  0\n",
      " 0  0  1  0\n",
      "[torch.FloatTensor of size 3x4]\n",
      "\n",
      "Identity NdArray with size (3, 3) (can't create non-squared identity matrix):\n",
      " [[ 1.  0.  0.]\n",
      " [ 0.  1.  0.]\n",
      " [ 0.  0.  1.]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print('Tensor of zeros with size (3, 4):\\n', torch.zeros(3, 4))\n",
    "print('NdArray of zeros with size (3, 4):\\n', np.zeros((3, 4)))\n",
    "print('-------------------------------------------------------------------------')\n",
    "\n",
    "print('Tensor of ones with size (3, 4):\\n', torch.ones(3, 4))\n",
    "print('NdArray of ones with size (3, 4):\\n', np.ones((3, 4)))\n",
    "print('-------------------------------------------------------------------------')\n",
    "\n",
    "print('Identity Tensor with size (3, 4):\\n', torch.eye(3, 4))\n",
    "print(\"Identity NdArray with size (3, 3) (can't create non-squared identity matrix):\\n\", np.eye(3))\n",
    "print('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "#### 2. Tensor slicing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor:\n",
      " \n",
      "(0 ,0 ,.,.) = \n",
      "  0.2353  0.3940  0.5236  0.0965  0.7462\n",
      "  0.8047  0.7706  0.7035  0.8961  0.6226\n",
      "  0.0524  0.2388  0.6801  0.2517  0.7626\n",
      "\n",
      "(0 ,1 ,.,.) = \n",
      "  0.7094  0.3178  0.9071  0.5035  0.2032\n",
      "  0.9533  0.7225  0.3106  0.6425  0.1594\n",
      "  0.6849  0.3640  0.7304  0.8298  0.3114\n",
      "\n",
      "(0 ,2 ,.,.) = \n",
      "  0.5726  0.1185  0.8974  0.1053  0.9073\n",
      "  0.5831  0.8008  0.8334  0.8271  0.6497\n",
      "  0.3021  0.4453  0.0228  0.4584  0.2826\n",
      "\n",
      "(0 ,3 ,.,.) = \n",
      "  0.5222  0.3387  0.6303  0.9214  0.4238\n",
      "  0.0181  0.5255  0.1014  0.9714  0.8173\n",
      "  0.2635  0.7258  0.1076  0.1959  0.0803\n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      "  0.3146  0.8596  0.5384  0.5301  0.2259\n",
      "  0.5644  0.4054  0.6867  0.3449  0.1055\n",
      "  0.3096  0.9881  0.0009  0.7668  0.1698\n",
      "\n",
      "(1 ,1 ,.,.) = \n",
      "  0.5302  0.7687  0.4239  0.1969  0.9981\n",
      "  0.4789  0.1910  0.4797  0.1617  0.1610\n",
      "  0.6330  0.6916  0.9095  0.5786  0.4924\n",
      "\n",
      "(1 ,2 ,.,.) = \n",
      "  0.1892  0.6991  0.3013  0.7681  0.7045\n",
      "  0.2920  0.5094  0.4631  0.6119  0.3043\n",
      "  0.4938  0.5368  0.3839  0.8246  0.4544\n",
      "\n",
      "(1 ,3 ,.,.) = \n",
      "  0.7249  0.4835  0.5392  0.6014  0.4747\n",
      "  0.4734  0.0815  0.1517  0.5613  0.5622\n",
      "  0.5787  0.4567  0.7078  0.9083  0.0713\n",
      "[torch.FloatTensor of size 2x4x3x5]\n",
      "\n",
      "NdArray:\n",
      " [[[[  2.35266149e-01   3.94001544e-01   5.23599505e-01   9.65321660e-02\n",
      "      7.46225476e-01]\n",
      "   [  8.04683864e-01   7.70551920e-01   7.03463018e-01   8.96062672e-01\n",
      "      6.22605801e-01]\n",
      "   [  5.23919463e-02   2.38847196e-01   6.80063128e-01   2.51653731e-01\n",
      "      7.62594163e-01]]\n",
      "\n",
      "  [[  7.09381461e-01   3.17814767e-01   9.07121003e-01   5.03548861e-01\n",
      "      2.03220844e-01]\n",
      "   [  9.53343928e-01   7.22493529e-01   3.10601354e-01   6.42495573e-01\n",
      "      1.59355521e-01]\n",
      "   [  6.84857786e-01   3.63963306e-01   7.30434358e-01   8.29808414e-01\n",
      "      3.11439395e-01]]\n",
      "\n",
      "  [[  5.72571039e-01   1.18456125e-01   8.97350490e-01   1.05296135e-01\n",
      "      9.07261372e-01]\n",
      "   [  5.83072662e-01   8.00758421e-01   8.33401263e-01   8.27071071e-01\n",
      "      6.49661124e-01]\n",
      "   [  3.02051365e-01   4.45313454e-01   2.28268504e-02   4.58414435e-01\n",
      "      2.82595038e-01]]\n",
      "\n",
      "  [[  5.22177517e-01   3.38662267e-01   6.30302966e-01   9.21366513e-01\n",
      "      4.23827350e-01]\n",
      "   [  1.81279182e-02   5.25528014e-01   1.01398468e-01   9.71433997e-01\n",
      "      8.17272782e-01]\n",
      "   [  2.63513923e-01   7.25817621e-01   1.07580960e-01   1.95925176e-01\n",
      "      8.02723765e-02]]]\n",
      "\n",
      "\n",
      " [[[  3.14632356e-01   8.59642804e-01   5.38383722e-01   5.30092120e-01\n",
      "      2.25893259e-01]\n",
      "   [  5.64411581e-01   4.05419648e-01   6.86654627e-01   3.44886243e-01\n",
      "      1.05519712e-01]\n",
      "   [  3.09615314e-01   9.88126993e-01   8.94665718e-04   7.66828716e-01\n",
      "      1.69833779e-01]]\n",
      "\n",
      "  [[  5.30231118e-01   7.68676221e-01   4.23854351e-01   1.96927369e-01\n",
      "      9.98137891e-01]\n",
      "   [  4.78865862e-01   1.90981925e-01   4.79736984e-01   1.61729097e-01\n",
      "      1.61048293e-01]\n",
      "   [  6.33021832e-01   6.91628098e-01   9.09508765e-01   5.78613877e-01\n",
      "      4.92397547e-01]]\n",
      "\n",
      "  [[  1.89242125e-01   6.99099720e-01   3.01253200e-01   7.68107295e-01\n",
      "      7.04518139e-01]\n",
      "   [  2.91954458e-01   5.09429097e-01   4.63145733e-01   6.11934304e-01\n",
      "      3.04336846e-01]\n",
      "   [  4.93766069e-01   5.36813736e-01   3.83876801e-01   8.24628472e-01\n",
      "      4.54383314e-01]]\n",
      "\n",
      "  [[  7.24947572e-01   4.83467460e-01   5.39229155e-01   6.01448357e-01\n",
      "      4.74733293e-01]\n",
      "   [  4.73427653e-01   8.14995170e-02   1.51711106e-01   5.61286628e-01\n",
      "      5.62167287e-01]\n",
      "   [  5.78667998e-01   4.56686199e-01   7.07787573e-01   9.08277869e-01\n",
      "      7.12981820e-02]]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "t = torch.rand(2, 4, 3, 5)\n",
    "a = t.numpy()\n",
    "\n",
    "print ('Tensor:\\n', t)\n",
    "print ('NdArray:\\n', a)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "pytorch_slice = t[0, 1:3, :, 4]\n",
    "numpy_slice =  a[0, 1:3, :, 4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor[0, 1:3, :, 4]:\n",
      " \n",
      " 0.2032  0.1594  0.3114\n",
      " 0.9073  0.6497  0.2826\n",
      "[torch.FloatTensor of size 2x3]\n",
      "\n",
      "NdArray[0, 1:3, :, 4]:\n",
      " [[ 0.20322084  0.15935552  0.31143939]\n",
      " [ 0.90726137  0.64966112  0.28259504]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor[0, 1:3, :, 4]:\\n', pytorch_slice)\n",
    "print ('NdArray[0, 1:3, :, 4]:\\n', numpy_slice)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Below you can find more slicing examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor size:\n",
      " torch.Size([2, 4, 3, 5])\n",
      "NdArray size:\n",
      " (2, 4, 3, 5)\n",
      "-------------------------------------------------------------------------\n",
      "Tensor[0][1][2][3]:\n",
      " 0.8298084139823914\n",
      "NdArray[0][1][2][3]:\n",
      " 0.829808\n",
      "-------------------------------------------------------------------------\n",
      "Tensor[0, 1, 2, 3]:\n",
      " 0.8298084139823914\n",
      "NdArray[0, 1, 2, 3]:\n",
      " 0.829808\n",
      "-------------------------------------------------------------------------\n",
      "Tensor[0][1]:\n",
      " \n",
      " 0.7094  0.3178  0.9071  0.5035  0.2032\n",
      " 0.9533  0.7225  0.3106  0.6425  0.1594\n",
      " 0.6849  0.3640  0.7304  0.8298  0.3114\n",
      "[torch.FloatTensor of size 3x5]\n",
      "\n",
      "NdArray[0][1]:\n",
      " [[ 0.70938146  0.31781477  0.907121    0.50354886  0.20322084]\n",
      " [ 0.95334393  0.72249353  0.31060135  0.64249557  0.15935552]\n",
      " [ 0.68485779  0.36396331  0.73043436  0.82980841  0.31143939]]\n",
      "-------------------------------------------------------------------------\n",
      "Tensor[0, 1:3]:\n",
      " \n",
      "(0 ,.,.) = \n",
      "  0.7094  0.3178  0.9071  0.5035  0.2032\n",
      "  0.9533  0.7225  0.3106  0.6425  0.1594\n",
      "  0.6849  0.3640  0.7304  0.8298  0.3114\n",
      "\n",
      "(1 ,.,.) = \n",
      "  0.5726  0.1185  0.8974  0.1053  0.9073\n",
      "  0.5831  0.8008  0.8334  0.8271  0.6497\n",
      "  0.3021  0.4453  0.0228  0.4584  0.2826\n",
      "[torch.FloatTensor of size 2x3x5]\n",
      "\n",
      "NdArray[0, 1:3]:\n",
      " [[[ 0.70938146  0.31781477  0.907121    0.50354886  0.20322084]\n",
      "  [ 0.95334393  0.72249353  0.31060135  0.64249557  0.15935552]\n",
      "  [ 0.68485779  0.36396331  0.73043436  0.82980841  0.31143939]]\n",
      "\n",
      " [[ 0.57257104  0.11845613  0.89735049  0.10529613  0.90726137]\n",
      "  [ 0.58307266  0.80075842  0.83340126  0.82707107  0.64966112]\n",
      "  [ 0.30205137  0.44531345  0.02282685  0.45841444  0.28259504]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor size:\\n', t.size())\n",
    "print ('NdArray size:\\n', a.shape)\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Tensor[0][1][2][3]:\\n', t[0][1][2][3])\n",
    "print ('NdArray[0][1][2][3]:\\n', a[0][1][2][3])\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Tensor[0, 1, 2, 3]:\\n', t[0, 1, 2, 3])\n",
    "print ('NdArray[0, 1, 2, 3]:\\n', a[0, 1, 2, 3])\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Tensor[0][1]:\\n', t[0][1])\n",
    "print ('NdArray[0][1]:\\n', a[0][1])\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Tensor[0, 1:3]:\\n', t[0, 1:3])\n",
    "print ('NdArray[0, 1:3]:\\n', a[0, 1:3])\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "#### 3. Tensor masking"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor:\n",
      " \n",
      "(0 ,0 ,.,.) = \n",
      " -0.2647 -0.1060  0.0236 -0.4035  0.2462\n",
      "  0.3047  0.2706  0.2035  0.3961  0.1226\n",
      " -0.4476 -0.2612  0.1801 -0.2483  0.2626\n",
      "\n",
      "(0 ,1 ,.,.) = \n",
      "  0.2094 -0.1822  0.4071  0.0035 -0.2968\n",
      "  0.4533  0.2225 -0.1894  0.1425 -0.3406\n",
      "  0.1849 -0.1360  0.2304  0.3298 -0.1886\n",
      "\n",
      "(0 ,2 ,.,.) = \n",
      "  0.0726 -0.3815  0.3974 -0.3947  0.4073\n",
      "  0.0831  0.3008  0.3334  0.3271  0.1497\n",
      " -0.1979 -0.0547 -0.4772 -0.0416 -0.2174\n",
      "\n",
      "(0 ,3 ,.,.) = \n",
      "  0.0222 -0.1613  0.1303  0.4214 -0.0762\n",
      " -0.4819  0.0255 -0.3986  0.4714  0.3173\n",
      " -0.2365  0.2258 -0.3924 -0.3041 -0.4197\n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      " -0.1854  0.3596  0.0384  0.0301 -0.2741\n",
      "  0.0644 -0.0946  0.1867 -0.1551 -0.3945\n",
      " -0.1904  0.4881 -0.4991  0.2668 -0.3302\n",
      "\n",
      "(1 ,1 ,.,.) = \n",
      "  0.0302  0.2687 -0.0761 -0.3031  0.4981\n",
      " -0.0211 -0.3090 -0.0203 -0.3383 -0.3390\n",
      "  0.1330  0.1916  0.4095  0.0786 -0.0076\n",
      "\n",
      "(1 ,2 ,.,.) = \n",
      " -0.3108  0.1991 -0.1987  0.2681  0.2045\n",
      " -0.2080  0.0094 -0.0369  0.1119 -0.1957\n",
      " -0.0062  0.0368 -0.1161  0.3246 -0.0456\n",
      "\n",
      "(1 ,3 ,.,.) = \n",
      "  0.2249 -0.0165  0.0392  0.1014 -0.0253\n",
      " -0.0266 -0.4185 -0.3483  0.0613  0.0622\n",
      "  0.0787 -0.0433  0.2078  0.4083 -0.4287\n",
      "[torch.FloatTensor of size 2x4x3x5]\n",
      "\n",
      "NdArray:\n",
      " [[[[-0.26473385 -0.10599846  0.02359951 -0.40346783  0.24622548]\n",
      "   [ 0.30468386  0.27055192  0.20346302  0.39606267  0.1226058 ]\n",
      "   [-0.44760805 -0.2611528   0.18006313 -0.24834627  0.26259416]]\n",
      "\n",
      "  [[ 0.20938146 -0.18218523  0.407121    0.00354886 -0.29677916]\n",
      "   [ 0.45334393  0.22249353 -0.18939865  0.14249557 -0.34064448]\n",
      "   [ 0.18485779 -0.13603669  0.23043436  0.32980841 -0.18856061]]\n",
      "\n",
      "  [[ 0.07257104 -0.38154387  0.39735049 -0.39470387  0.40726137]\n",
      "   [ 0.08307266  0.30075842  0.33340126  0.32707107  0.14966112]\n",
      "   [-0.19794863 -0.05468655 -0.47717315 -0.04158556 -0.21740496]]\n",
      "\n",
      "  [[ 0.02217752 -0.16133773  0.13030297  0.42136651 -0.07617265]\n",
      "   [-0.48187208  0.02552801 -0.39860153  0.471434    0.31727278]\n",
      "   [-0.23648608  0.22581762 -0.39241904 -0.30407482 -0.41972762]]]\n",
      "\n",
      "\n",
      " [[[-0.18536764  0.3596428   0.03838372  0.03009212 -0.27410674]\n",
      "   [ 0.06441158 -0.09458035  0.18665463 -0.15511376 -0.39448029]\n",
      "   [-0.19038469  0.48812699 -0.49910533  0.26682872 -0.33016622]]\n",
      "\n",
      "  [[ 0.03023112  0.26867622 -0.07614565 -0.30307263  0.49813789]\n",
      "   [-0.02113414 -0.30901808 -0.02026302 -0.3382709  -0.33895171]\n",
      "   [ 0.13302183  0.1916281   0.40950876  0.07861388 -0.00760245]]\n",
      "\n",
      "  [[-0.31075788  0.19909972 -0.1987468   0.2681073   0.20451814]\n",
      "   [-0.20804554  0.0094291  -0.03685427  0.1119343  -0.19566315]\n",
      "   [-0.00623393  0.03681374 -0.1161232   0.32462847 -0.04561669]]\n",
      "\n",
      "  [[ 0.22494757 -0.01653254  0.03922915  0.10144836 -0.02526671]\n",
      "   [-0.02657235 -0.41850048 -0.34828889  0.06128663  0.06216729]\n",
      "   [ 0.078668   -0.0433138   0.20778757  0.40827787 -0.42870182]]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "t = t - 0.5\n",
    "a = t.numpy()\n",
    "\n",
    "print ('Tensor:\\n', t)\n",
    "print ('NdArray:\\n', a)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "pytorch_masked = t[t > 0]\n",
    "numpy_masked = a[a > 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor[Tensor > 0]:\n",
      " \n",
      " 0.0236\n",
      " 0.2462\n",
      " 0.3047\n",
      " 0.2706\n",
      " 0.2035\n",
      " 0.3961\n",
      " 0.1226\n",
      " 0.1801\n",
      " 0.2626\n",
      " 0.2094\n",
      " 0.4071\n",
      " 0.0035\n",
      " 0.4533\n",
      " 0.2225\n",
      " 0.1425\n",
      " 0.1849\n",
      " 0.2304\n",
      " 0.3298\n",
      " 0.0726\n",
      " 0.3974\n",
      " 0.4073\n",
      " 0.0831\n",
      " 0.3008\n",
      " 0.3334\n",
      " 0.3271\n",
      " 0.1497\n",
      " 0.0222\n",
      " 0.1303\n",
      " 0.4214\n",
      " 0.0255\n",
      " 0.4714\n",
      " 0.3173\n",
      " 0.2258\n",
      " 0.3596\n",
      " 0.0384\n",
      " 0.0301\n",
      " 0.0644\n",
      " 0.1867\n",
      " 0.4881\n",
      " 0.2668\n",
      " 0.0302\n",
      " 0.2687\n",
      " 0.4981\n",
      " 0.1330\n",
      " 0.1916\n",
      " 0.4095\n",
      " 0.0786\n",
      " 0.1991\n",
      " 0.2681\n",
      " 0.2045\n",
      " 0.0094\n",
      " 0.1119\n",
      " 0.0368\n",
      " 0.3246\n",
      " 0.2249\n",
      " 0.0392\n",
      " 0.1014\n",
      " 0.0613\n",
      " 0.0622\n",
      " 0.0787\n",
      " 0.2078\n",
      " 0.4083\n",
      "[torch.FloatTensor of size 62]\n",
      "\n",
      "NdArray[NdArray > 0]:\n",
      " [ 0.02359951  0.24622548  0.30468386  0.27055192  0.20346302  0.39606267\n",
      "  0.1226058   0.18006313  0.26259416  0.20938146  0.407121    0.00354886\n",
      "  0.45334393  0.22249353  0.14249557  0.18485779  0.23043436  0.32980841\n",
      "  0.07257104  0.39735049  0.40726137  0.08307266  0.30075842  0.33340126\n",
      "  0.32707107  0.14966112  0.02217752  0.13030297  0.42136651  0.02552801\n",
      "  0.471434    0.31727278  0.22581762  0.3596428   0.03838372  0.03009212\n",
      "  0.06441158  0.18665463  0.48812699  0.26682872  0.03023112  0.26867622\n",
      "  0.49813789  0.13302183  0.1916281   0.40950876  0.07861388  0.19909972\n",
      "  0.2681073   0.20451814  0.0094291   0.1119343   0.03681374  0.32462847\n",
      "  0.22494757  0.03922915  0.10144836  0.06128663  0.06216729  0.078668\n",
      "  0.20778757  0.40827787]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor[Tensor > 0]:\\n', pytorch_masked)\n",
    "print ('NdArray[NdArray > 0]:\\n', numpy_masked)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Below you can see how conditioning works with PyTorch, and how array size is changed after that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor > 0:\n",
      " \n",
      "(0 ,0 ,.,.) = \n",
      "  0  0  1  0  1\n",
      "  1  1  1  1  1\n",
      "  0  0  1  0  1\n",
      "\n",
      "(0 ,1 ,.,.) = \n",
      "  1  0  1  1  0\n",
      "  1  1  0  1  0\n",
      "  1  0  1  1  0\n",
      "\n",
      "(0 ,2 ,.,.) = \n",
      "  1  0  1  0  1\n",
      "  1  1  1  1  1\n",
      "  0  0  0  0  0\n",
      "\n",
      "(0 ,3 ,.,.) = \n",
      "  1  0  1  1  0\n",
      "  0  1  0  1  1\n",
      "  0  1  0  0  0\n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      "  0  1  1  1  0\n",
      "  1  0  1  0  0\n",
      "  0  1  0  1  0\n",
      "\n",
      "(1 ,1 ,.,.) = \n",
      "  1  1  0  0  1\n",
      "  0  0  0  0  0\n",
      "  1  1  1  1  0\n",
      "\n",
      "(1 ,2 ,.,.) = \n",
      "  0  1  0  1  1\n",
      "  0  1  0  1  0\n",
      "  0  1  0  1  0\n",
      "\n",
      "(1 ,3 ,.,.) = \n",
      "  1  0  1  1  0\n",
      "  0  0  0  1  1\n",
      "  1  0  1  1  0\n",
      "[torch.ByteTensor of size 2x4x3x5]\n",
      "\n",
      "NdArray > 0:\n",
      " [[[[False False  True False  True]\n",
      "   [ True  True  True  True  True]\n",
      "   [False False  True False  True]]\n",
      "\n",
      "  [[ True False  True  True False]\n",
      "   [ True  True False  True False]\n",
      "   [ True False  True  True False]]\n",
      "\n",
      "  [[ True False  True False  True]\n",
      "   [ True  True  True  True  True]\n",
      "   [False False False False False]]\n",
      "\n",
      "  [[ True False  True  True False]\n",
      "   [False  True False  True  True]\n",
      "   [False  True False False False]]]\n",
      "\n",
      "\n",
      " [[[False  True  True  True False]\n",
      "   [ True False  True False False]\n",
      "   [False  True False  True False]]\n",
      "\n",
      "  [[ True  True False False  True]\n",
      "   [False False False False False]\n",
      "   [ True  True  True  True False]]\n",
      "\n",
      "  [[False  True False  True  True]\n",
      "   [False  True False  True False]\n",
      "   [False  True False  True False]]\n",
      "\n",
      "  [[ True False  True  True False]\n",
      "   [False False False  True  True]\n",
      "   [ True False  True  True False]]]]\n",
      "-------------------------------------------------------------------------\n",
      "Tensor[Tensor > 0]:\n",
      " \n",
      " 0.0236\n",
      " 0.2462\n",
      " 0.3047\n",
      " 0.2706\n",
      " 0.2035\n",
      " 0.3961\n",
      " 0.1226\n",
      " 0.1801\n",
      " 0.2626\n",
      " 0.2094\n",
      " 0.4071\n",
      " 0.0035\n",
      " 0.4533\n",
      " 0.2225\n",
      " 0.1425\n",
      " 0.1849\n",
      " 0.2304\n",
      " 0.3298\n",
      " 0.0726\n",
      " 0.3974\n",
      " 0.4073\n",
      " 0.0831\n",
      " 0.3008\n",
      " 0.3334\n",
      " 0.3271\n",
      " 0.1497\n",
      " 0.0222\n",
      " 0.1303\n",
      " 0.4214\n",
      " 0.0255\n",
      " 0.4714\n",
      " 0.3173\n",
      " 0.2258\n",
      " 0.3596\n",
      " 0.0384\n",
      " 0.0301\n",
      " 0.0644\n",
      " 0.1867\n",
      " 0.4881\n",
      " 0.2668\n",
      " 0.0302\n",
      " 0.2687\n",
      " 0.4981\n",
      " 0.1330\n",
      " 0.1916\n",
      " 0.4095\n",
      " 0.0786\n",
      " 0.1991\n",
      " 0.2681\n",
      " 0.2045\n",
      " 0.0094\n",
      " 0.1119\n",
      " 0.0368\n",
      " 0.3246\n",
      " 0.2249\n",
      " 0.0392\n",
      " 0.1014\n",
      " 0.0613\n",
      " 0.0622\n",
      " 0.0787\n",
      " 0.2078\n",
      " 0.4083\n",
      "[torch.FloatTensor of size 62]\n",
      "\n",
      "NdArray[NdArray > 0]:\n",
      " [ 0.02359951  0.24622548  0.30468386  0.27055192  0.20346302  0.39606267\n",
      "  0.1226058   0.18006313  0.26259416  0.20938146  0.407121    0.00354886\n",
      "  0.45334393  0.22249353  0.14249557  0.18485779  0.23043436  0.32980841\n",
      "  0.07257104  0.39735049  0.40726137  0.08307266  0.30075842  0.33340126\n",
      "  0.32707107  0.14966112  0.02217752  0.13030297  0.42136651  0.02552801\n",
      "  0.471434    0.31727278  0.22581762  0.3596428   0.03838372  0.03009212\n",
      "  0.06441158  0.18665463  0.48812699  0.26682872  0.03023112  0.26867622\n",
      "  0.49813789  0.13302183  0.1916281   0.40950876  0.07861388  0.19909972\n",
      "  0.2681073   0.20451814  0.0094291   0.1119343   0.03681374  0.32462847\n",
      "  0.22494757  0.03922915  0.10144836  0.06128663  0.06216729  0.078668\n",
      "  0.20778757  0.40827787]\n",
      "-------------------------------------------------------------------------\n",
      "Size of Tensor[Tensor > 0]:\n",
      " torch.Size([62])\n",
      "Size of NdArray[NdArray > 0]:\n",
      " (62,)\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor > 0:\\n', t > 0)\n",
    "print ('NdArray > 0:\\n', a > 0)\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Tensor[Tensor > 0]:\\n', t[t > 0])\n",
    "print ('NdArray[NdArray > 0]:\\n', a[a > 0])\n",
    "print ('-------------------------------------------------------------------------')\n",
    "\n",
    "print ('Size of Tensor[Tensor > 0]:\\n', t[t > 0].size())\n",
    "print ('Size of NdArray[NdArray > 0]:\\n', a[a > 0].shape)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "#### 4. Tensor reshaping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor:\n",
      " \n",
      "(0 ,0 ,.,.) = \n",
      " -0.2647 -0.1060  0.0236 -0.4035  0.2462\n",
      "  0.3047  0.2706  0.2035  0.3961  0.1226\n",
      " -0.4476 -0.2612  0.1801 -0.2483  0.2626\n",
      "\n",
      "(0 ,1 ,.,.) = \n",
      "  0.2094 -0.1822  0.4071  0.0035 -0.2968\n",
      "  0.4533  0.2225 -0.1894  0.1425 -0.3406\n",
      "  0.1849 -0.1360  0.2304  0.3298 -0.1886\n",
      "\n",
      "(0 ,2 ,.,.) = \n",
      "  0.0726 -0.3815  0.3974 -0.3947  0.4073\n",
      "  0.0831  0.3008  0.3334  0.3271  0.1497\n",
      " -0.1979 -0.0547 -0.4772 -0.0416 -0.2174\n",
      "\n",
      "(0 ,3 ,.,.) = \n",
      "  0.0222 -0.1613  0.1303  0.4214 -0.0762\n",
      " -0.4819  0.0255 -0.3986  0.4714  0.3173\n",
      " -0.2365  0.2258 -0.3924 -0.3041 -0.4197\n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      " -0.1854  0.3596  0.0384  0.0301 -0.2741\n",
      "  0.0644 -0.0946  0.1867 -0.1551 -0.3945\n",
      " -0.1904  0.4881 -0.4991  0.2668 -0.3302\n",
      "\n",
      "(1 ,1 ,.,.) = \n",
      "  0.0302  0.2687 -0.0761 -0.3031  0.4981\n",
      " -0.0211 -0.3090 -0.0203 -0.3383 -0.3390\n",
      "  0.1330  0.1916  0.4095  0.0786 -0.0076\n",
      "\n",
      "(1 ,2 ,.,.) = \n",
      " -0.3108  0.1991 -0.1987  0.2681  0.2045\n",
      " -0.2080  0.0094 -0.0369  0.1119 -0.1957\n",
      " -0.0062  0.0368 -0.1161  0.3246 -0.0456\n",
      "\n",
      "(1 ,3 ,.,.) = \n",
      "  0.2249 -0.0165  0.0392  0.1014 -0.0253\n",
      " -0.0266 -0.4185 -0.3483  0.0613  0.0622\n",
      "  0.0787 -0.0433  0.2078  0.4083 -0.4287\n",
      "[torch.FloatTensor of size 2x4x3x5]\n",
      "\n",
      "NdArray:\n",
      " [[[[-0.26473385 -0.10599846  0.02359951 -0.40346783  0.24622548]\n",
      "   [ 0.30468386  0.27055192  0.20346302  0.39606267  0.1226058 ]\n",
      "   [-0.44760805 -0.2611528   0.18006313 -0.24834627  0.26259416]]\n",
      "\n",
      "  [[ 0.20938146 -0.18218523  0.407121    0.00354886 -0.29677916]\n",
      "   [ 0.45334393  0.22249353 -0.18939865  0.14249557 -0.34064448]\n",
      "   [ 0.18485779 -0.13603669  0.23043436  0.32980841 -0.18856061]]\n",
      "\n",
      "  [[ 0.07257104 -0.38154387  0.39735049 -0.39470387  0.40726137]\n",
      "   [ 0.08307266  0.30075842  0.33340126  0.32707107  0.14966112]\n",
      "   [-0.19794863 -0.05468655 -0.47717315 -0.04158556 -0.21740496]]\n",
      "\n",
      "  [[ 0.02217752 -0.16133773  0.13030297  0.42136651 -0.07617265]\n",
      "   [-0.48187208  0.02552801 -0.39860153  0.471434    0.31727278]\n",
      "   [-0.23648608  0.22581762 -0.39241904 -0.30407482 -0.41972762]]]\n",
      "\n",
      "\n",
      " [[[-0.18536764  0.3596428   0.03838372  0.03009212 -0.27410674]\n",
      "   [ 0.06441158 -0.09458035  0.18665463 -0.15511376 -0.39448029]\n",
      "   [-0.19038469  0.48812699 -0.49910533  0.26682872 -0.33016622]]\n",
      "\n",
      "  [[ 0.03023112  0.26867622 -0.07614565 -0.30307263  0.49813789]\n",
      "   [-0.02113414 -0.30901808 -0.02026302 -0.3382709  -0.33895171]\n",
      "   [ 0.13302183  0.1916281   0.40950876  0.07861388 -0.00760245]]\n",
      "\n",
      "  [[-0.31075788  0.19909972 -0.1987468   0.2681073   0.20451814]\n",
      "   [-0.20804554  0.0094291  -0.03685427  0.1119343  -0.19566315]\n",
      "   [-0.00623393  0.03681374 -0.1161232   0.32462847 -0.04561669]]\n",
      "\n",
      "  [[ 0.22494757 -0.01653254  0.03922915  0.10144836 -0.02526671]\n",
      "   [-0.02657235 -0.41850048 -0.34828889  0.06128663  0.06216729]\n",
      "   [ 0.078668   -0.0433138   0.20778757  0.40827787 -0.42870182]]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor:\\n', t)\n",
    "print ('NdArray:\\n', a)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "pytorch_reshape = t.view([6, 5, 4])\n",
    "numpy_reshape = a.reshape([6, 5, 4])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor reshaped to (6:5:4):\n",
      " \n",
      "(0 ,.,.) = \n",
      " -0.2647 -0.1060  0.0236 -0.4035\n",
      "  0.2462  0.3047  0.2706  0.2035\n",
      "  0.3961  0.1226 -0.4476 -0.2612\n",
      "  0.1801 -0.2483  0.2626  0.2094\n",
      " -0.1822  0.4071  0.0035 -0.2968\n",
      "\n",
      "(1 ,.,.) = \n",
      "  0.4533  0.2225 -0.1894  0.1425\n",
      " -0.3406  0.1849 -0.1360  0.2304\n",
      "  0.3298 -0.1886  0.0726 -0.3815\n",
      "  0.3974 -0.3947  0.4073  0.0831\n",
      "  0.3008  0.3334  0.3271  0.1497\n",
      "\n",
      "(2 ,.,.) = \n",
      " -0.1979 -0.0547 -0.4772 -0.0416\n",
      " -0.2174  0.0222 -0.1613  0.1303\n",
      "  0.4214 -0.0762 -0.4819  0.0255\n",
      " -0.3986  0.4714  0.3173 -0.2365\n",
      "  0.2258 -0.3924 -0.3041 -0.4197\n",
      "\n",
      "(3 ,.,.) = \n",
      " -0.1854  0.3596  0.0384  0.0301\n",
      " -0.2741  0.0644 -0.0946  0.1867\n",
      " -0.1551 -0.3945 -0.1904  0.4881\n",
      " -0.4991  0.2668 -0.3302  0.0302\n",
      "  0.2687 -0.0761 -0.3031  0.4981\n",
      "\n",
      "(4 ,.,.) = \n",
      " -0.0211 -0.3090 -0.0203 -0.3383\n",
      " -0.3390  0.1330  0.1916  0.4095\n",
      "  0.0786 -0.0076 -0.3108  0.1991\n",
      " -0.1987  0.2681  0.2045 -0.2080\n",
      "  0.0094 -0.0369  0.1119 -0.1957\n",
      "\n",
      "(5 ,.,.) = \n",
      " -0.0062  0.0368 -0.1161  0.3246\n",
      " -0.0456  0.2249 -0.0165  0.0392\n",
      "  0.1014 -0.0253 -0.0266 -0.4185\n",
      " -0.3483  0.0613  0.0622  0.0787\n",
      " -0.0433  0.2078  0.4083 -0.4287\n",
      "[torch.FloatTensor of size 6x5x4]\n",
      "\n",
      "NdArray reshaped to (6:5:4):\n",
      " [[[-0.26473385 -0.10599846  0.02359951 -0.40346783]\n",
      "  [ 0.24622548  0.30468386  0.27055192  0.20346302]\n",
      "  [ 0.39606267  0.1226058  -0.44760805 -0.2611528 ]\n",
      "  [ 0.18006313 -0.24834627  0.26259416  0.20938146]\n",
      "  [-0.18218523  0.407121    0.00354886 -0.29677916]]\n",
      "\n",
      " [[ 0.45334393  0.22249353 -0.18939865  0.14249557]\n",
      "  [-0.34064448  0.18485779 -0.13603669  0.23043436]\n",
      "  [ 0.32980841 -0.18856061  0.07257104 -0.38154387]\n",
      "  [ 0.39735049 -0.39470387  0.40726137  0.08307266]\n",
      "  [ 0.30075842  0.33340126  0.32707107  0.14966112]]\n",
      "\n",
      " [[-0.19794863 -0.05468655 -0.47717315 -0.04158556]\n",
      "  [-0.21740496  0.02217752 -0.16133773  0.13030297]\n",
      "  [ 0.42136651 -0.07617265 -0.48187208  0.02552801]\n",
      "  [-0.39860153  0.471434    0.31727278 -0.23648608]\n",
      "  [ 0.22581762 -0.39241904 -0.30407482 -0.41972762]]\n",
      "\n",
      " [[-0.18536764  0.3596428   0.03838372  0.03009212]\n",
      "  [-0.27410674  0.06441158 -0.09458035  0.18665463]\n",
      "  [-0.15511376 -0.39448029 -0.19038469  0.48812699]\n",
      "  [-0.49910533  0.26682872 -0.33016622  0.03023112]\n",
      "  [ 0.26867622 -0.07614565 -0.30307263  0.49813789]]\n",
      "\n",
      " [[-0.02113414 -0.30901808 -0.02026302 -0.3382709 ]\n",
      "  [-0.33895171  0.13302183  0.1916281   0.40950876]\n",
      "  [ 0.07861388 -0.00760245 -0.31075788  0.19909972]\n",
      "  [-0.1987468   0.2681073   0.20451814 -0.20804554]\n",
      "  [ 0.0094291  -0.03685427  0.1119343  -0.19566315]]\n",
      "\n",
      " [[-0.00623393  0.03681374 -0.1161232   0.32462847]\n",
      "  [-0.04561669  0.22494757 -0.01653254  0.03922915]\n",
      "  [ 0.10144836 -0.02526671 -0.02657235 -0.41850048]\n",
      "  [-0.34828889  0.06128663  0.06216729  0.078668  ]\n",
      "  [-0.0433138   0.20778757  0.40827787 -0.42870182]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Tensor reshaped to (6:5:4):\\n', pytorch_reshape)\n",
    "print ('NdArray reshaped to (6:5:4):\\n',numpy_reshape)\n",
    "print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "You can also see a permutation example below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "scrolled": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor with dimensions transposed as (1, 0, 2):\n",
      " \n",
      "(0 ,.,.) = \n",
      " -0.2647 -0.1060  0.0236 -0.4035\n",
      "  0.4533  0.2225 -0.1894  0.1425\n",
      " -0.1979 -0.0547 -0.4772 -0.0416\n",
      " -0.1854  0.3596  0.0384  0.0301\n",
      " -0.0211 -0.3090 -0.0203 -0.3383\n",
      " -0.0062  0.0368 -0.1161  0.3246\n",
      "\n",
      "(1 ,.,.) = \n",
      "  0.2462  0.3047  0.2706  0.2035\n",
      " -0.3406  0.1849 -0.1360  0.2304\n",
      " -0.2174  0.0222 -0.1613  0.1303\n",
      " -0.2741  0.0644 -0.0946  0.1867\n",
      " -0.3390  0.1330  0.1916  0.4095\n",
      " -0.0456  0.2249 -0.0165  0.0392\n",
      "\n",
      "(2 ,.,.) = \n",
      "  0.3961  0.1226 -0.4476 -0.2612\n",
      "  0.3298 -0.1886  0.0726 -0.3815\n",
      "  0.4214 -0.0762 -0.4819  0.0255\n",
      " -0.1551 -0.3945 -0.1904  0.4881\n",
      "  0.0786 -0.0076 -0.3108  0.1991\n",
      "  0.1014 -0.0253 -0.0266 -0.4185\n",
      "\n",
      "(3 ,.,.) = \n",
      "  0.1801 -0.2483  0.2626  0.2094\n",
      "  0.3974 -0.3947  0.4073  0.0831\n",
      " -0.3986  0.4714  0.3173 -0.2365\n",
      " -0.4991  0.2668 -0.3302  0.0302\n",
      " -0.1987  0.2681  0.2045 -0.2080\n",
      " -0.3483  0.0613  0.0622  0.0787\n",
      "\n",
      "(4 ,.,.) = \n",
      " -0.1822  0.4071  0.0035 -0.2968\n",
      "  0.3008  0.3334  0.3271  0.1497\n",
      "  0.2258 -0.3924 -0.3041 -0.4197\n",
      "  0.2687 -0.0761 -0.3031  0.4981\n",
      "  0.0094 -0.0369  0.1119 -0.1957\n",
      " -0.0433  0.2078  0.4083 -0.4287\n",
      "[torch.FloatTensor of size 5x6x4]\n",
      "\n",
      "NdArray with dimensions transposed as (1, 0, 2):\n",
      " [[[-0.26473385 -0.10599846  0.02359951 -0.40346783]\n",
      "  [ 0.45334393  0.22249353 -0.18939865  0.14249557]\n",
      "  [-0.19794863 -0.05468655 -0.47717315 -0.04158556]\n",
      "  [-0.18536764  0.3596428   0.03838372  0.03009212]\n",
      "  [-0.02113414 -0.30901808 -0.02026302 -0.3382709 ]\n",
      "  [-0.00623393  0.03681374 -0.1161232   0.32462847]]\n",
      "\n",
      " [[ 0.24622548  0.30468386  0.27055192  0.20346302]\n",
      "  [-0.34064448  0.18485779 -0.13603669  0.23043436]\n",
      "  [-0.21740496  0.02217752 -0.16133773  0.13030297]\n",
      "  [-0.27410674  0.06441158 -0.09458035  0.18665463]\n",
      "  [-0.33895171  0.13302183  0.1916281   0.40950876]\n",
      "  [-0.04561669  0.22494757 -0.01653254  0.03922915]]\n",
      "\n",
      " [[ 0.39606267  0.1226058  -0.44760805 -0.2611528 ]\n",
      "  [ 0.32980841 -0.18856061  0.07257104 -0.38154387]\n",
      "  [ 0.42136651 -0.07617265 -0.48187208  0.02552801]\n",
      "  [-0.15511376 -0.39448029 -0.19038469  0.48812699]\n",
      "  [ 0.07861388 -0.00760245 -0.31075788  0.19909972]\n",
      "  [ 0.10144836 -0.02526671 -0.02657235 -0.41850048]]\n",
      "\n",
      " [[ 0.18006313 -0.24834627  0.26259416  0.20938146]\n",
      "  [ 0.39735049 -0.39470387  0.40726137  0.08307266]\n",
      "  [-0.39860153  0.471434    0.31727278 -0.23648608]\n",
      "  [-0.49910533  0.26682872 -0.33016622  0.03023112]\n",
      "  [-0.1987468   0.2681073   0.20451814 -0.20804554]\n",
      "  [-0.34828889  0.06128663  0.06216729  0.078668  ]]\n",
      "\n",
      " [[-0.18218523  0.407121    0.00354886 -0.29677916]\n",
      "  [ 0.30075842  0.33340126  0.32707107  0.14966112]\n",
      "  [ 0.22581762 -0.39241904 -0.30407482 -0.41972762]\n",
      "  [ 0.26867622 -0.07614565 -0.30307263  0.49813789]\n",
      "  [ 0.0094291  -0.03685427  0.1119343  -0.19566315]\n",
      "  [-0.0433138   0.20778757  0.40827787 -0.42870182]]]\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "\n",
    "print ('Tensor with dimensions transposed as (1, 0, 2):\\n', pytorch_reshape.permute(1, 0, 2) )\n",
    "print ('NdArray with dimensions transposed as (1, 0, 2):\\n', np.transpose(numpy_reshape, (1, 0, 2)))\n",
    "print ('-------------------------------------------------------------------------')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 1.3 PyTorch Variables\n",
    "\n",
    "* Thin **wrapper** around Tensor in PyTorch\n",
    "* Helps to build **computation graph**\n",
    "* Necessary part of **Autograd** - automatic differenciation library\n",
    "* Keeps gradients with respect to this Variable in **.grad**.\n",
    "<img src=\"images/variable.png\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## Computation Graph\n",
    "<img src=\"images/2_layer_NN.png\" style=\"width: 50%\"/>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "**Computation graphs and variables.** In PyTorch, a neural network is represented by a _computational graph_ composed of interconnected _variables_. PyTorch allows you to build the network model by constructing this graph in code; it then simplifies the process of estimating the weights of this model by, for example, automatically calculating gradients.\n",
    "\n",
    "For example, suppose we wish to build the following two-layer model.Let's start by creating variables for the tensor inputs $x$ and outputs $y$. We'll do so by wrapping a PyTorch `Tensor` inside a `Variable` object:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "from torch.autograd import Variable\n",
    "import torch.nn.functional as F\n",
    "\n",
    "x = Variable(torch.randn(4, 1), requires_grad=False)\n",
    "y = Variable(torch.randn(3, 1), requires_grad=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "We set requires_grad to True to say that we want the gradient to be computed automatically which will be used in backpropagation to optimize the weights."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Now we define the weights: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "w1 = Variable(torch.randn(5, 4), requires_grad=True)\n",
    "w2 = Variable(torch.randn(3, 5), requires_grad=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Train the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "def model_forward(x):\n",
    "    return F.sigmoid(w2 @ F.sigmoid(w1 @ x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      " 1.6068 -1.3304 -0.6717 -0.6097\n",
      "-0.3414 -0.5062 -0.2533  1.0260\n",
      "-0.0341 -1.2144 -1.5983 -0.1392\n",
      "-0.5473  0.0084  0.4054  0.0970\n",
      " 0.3596  0.5987 -0.0324  0.6116\n",
      "[torch.FloatTensor of size 5x4]\n",
      "\n",
      "torch.Size([5, 4])\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print (w1)\n",
    "print (w1.data.shape)\n",
    "print (w1.grad) # Initially, non-existent"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 1.4 PyTorch: Backpropagation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "So, we have inputs and targets, we have our simple model's weights, now it's time to train them. And for this, we need three components: "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "**Loss** that describes how far our model from the target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true,
    "scrolled": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "criterion = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "**Optimization algorithm** that we use to update the weight update, and"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "import torch.optim as optim\n",
    "optimizer = optim.SGD([w1, w2], lr=0.001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "**Backpropagation step**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "for epoch in range(10):\n",
    "    loss = criterion(model_forward(x), y)\n",
    "    optimizer.zero_grad() # Zero-out previous gradients\n",
    "    loss.backward() # Compute new gradients\n",
    "    optimizer.step() # Apply these gradients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      " 1.6067 -1.3303 -0.6717 -0.6095\n",
      "-0.3414 -0.5062 -0.2533  1.0259\n",
      "-0.0340 -1.2145 -1.5983 -0.1396\n",
      "-0.5476  0.0085  0.4055  0.0976\n",
      " 0.3597  0.5986 -0.0324  0.6113\n",
      "[torch.FloatTensor of size 5x4]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print (w1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "##  1.5 PyTorch CUDA interface"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "One of the benefits of PyTorch is that it provides a CUDA interface for its tensor and autograd libraries. Using CUDA GPGPUs you can accelerate not onlu neural network training and inference, but also any other workload that maps to PyTorch tensors.\n",
    "\n",
    "You can check whether you have CUDA available in PyTorch by calling **torch.cuda.is_available()** function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Great, you have a GPU!\n"
     ]
    }
   ],
   "source": [
    "cuda_gpu = torch.cuda.is_available()\n",
    "if (cuda_gpu):\n",
    "    print(\"Great, you have a GPU!\")\n",
    "else:\n",
    "    print(\"Life is short -- consider a GPU!\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "## .cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "After that, accelerating you code with cuda is as easy as calling **.cuda()** on your tensors and models. If you call **.cuda()** on tensors, it will perform data transfer from CPU to CUDA GPU. If you call **.cuda()** on a model, it not only moves all its internal storage to GPU, but also maps the whole computational graph to GPU.\n",
    "\n",
    "To copy a tensor or a model back to CPU, for example in order to interface it with NumPy, you can call **.cpu()**."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "**Spoiler**: if you don't have a cuda GPU, or it's not configured for PyTorch, the next two cells are going to fail without wrapping them into if statement or in try block."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'torch.cuda.FloatTensor'>\n",
      "<class 'torch.FloatTensor'>\n"
     ]
    }
   ],
   "source": [
    "if cuda_gpu:\n",
    "    x = x.cuda()\n",
    "    print(type(x.data))\n",
    "\n",
    "x = x.cpu()\n",
    "print(type(x.data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      " 0.8210\n",
      "-0.3060\n",
      "-0.4833\n",
      "-2.4531\n",
      "[torch.cuda.FloatTensor of size 4x1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      " 1.6067 -1.3303 -0.6717 -0.6095\n",
      "-0.3414 -0.5062 -0.2533  1.0259\n",
      "-0.0340 -1.2145 -1.5983 -0.1396\n",
      "-0.5476  0.0085  0.4055  0.0976\n",
      " 0.3597  0.5986 -0.0324  0.6113\n",
      "[torch.cuda.FloatTensor of size 5x4 (GPU 0)]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "if cuda_gpu:\n",
    "    x = x.cuda()\n",
    "    y = y.cuda()\n",
    "    w1 = w1.cuda()\n",
    "    w2 = w2.cuda()\n",
    "\n",
    "print (x)\n",
    "for epoch in range(10):\n",
    "    loss = criterion(model_forward(x), y)\n",
    "        \n",
    "    optimizer.zero_grad() # Zero-out previous gradients\n",
    "    loss.backward() # Compute new gradients\n",
    "    optimizer.step() # Apply these gradients\n",
    "print (w1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Let's define two functions, train and test, to perform training and inference using our model. This code is also adopted from the PyTorch official tutorial and shows all the steps necessary for training/inference. For the training and testing aour network we need to perform a sequence of actions, that are mapped to PyTorch code fairly straightforward:\n",
    "\n",
    "1. We switch model to training/inference mode\n",
    "2. We iterate over the dataset fetching images in batches\n",
    "3. For every batch we load data and targets and running forward step of the network to get the model outputs\n",
    "4.  We define a loss function and compute loss between model outputs and targets on per batch basis\n",
    "5. In case of training, we zero gradients and use backpropagation with optimizer defined on the previous step to compute gradients of all the layers with respect to loss.\n",
    "6. In case of training, we perform a weight update step\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "def train(model, epoch, criterion, optimizer, data_loader):\n",
    "    model.train()\n",
    "    for batch_idx, (data, target) in enumerate(data_loader):\n",
    "        if cuda_gpu:\n",
    "            data, target = data.cuda(), target.cuda()\n",
    "            model.cuda()\n",
    "        data, target = Variable(data), Variable(target)\n",
    "        output = model(data)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        loss = criterion(output, target)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        if (batch_idx+1) % 400 == 0:\n",
    "            print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f}'.format(\n",
    "                epoch, (batch_idx+1) * len(data), len(data_loader.dataset),\n",
    "                100. * (batch_idx+1) / len(data_loader), loss.data[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "def test(model, epoch, criterion, data_loader):\n",
    "    model.eval()\n",
    "    test_loss = 0\n",
    "    correct = 0\n",
    "    for data, target in data_loader:\n",
    "        if cuda_gpu:\n",
    "            data, target = data.cuda(), target.cuda()\n",
    "            model.cuda()\n",
    "        data, target = Variable(data), Variable(target)\n",
    "        output = model(data)\n",
    "        test_loss += criterion(output, target).data[0]\n",
    "        pred = output.data.max(1)[1] # get the index of the max log-probability\n",
    "        correct += pred.eq(target.data).cpu().sum()\n",
    "\n",
    "    test_loss /= len(data_loader) # loss function already averages over batch size\n",
    "    acc = correct / len(data_loader.dataset)\n",
    "    print('\\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\\n'.format(\n",
    "        test_loss, correct, len(data_loader.dataset), 100. * acc))\n",
    "    return (acc, test_loss)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "After this introduction, we can start our data science journey ! The rest of the tutorial is loosely based on these [PyTorch examples](https://github.com/pytorch/examples/).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 2. PyTorch for Data Analytics\n",
    "* Build models with **torch.nn** library\n",
    "* Train models with **torch.autograd** library\n",
    "* Wrap your data in **torch.utils.data.Dataset** library\n",
    "* Interface your model and data with your favorite tools using **NumPy interface**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Before we look at more complex models, let's start with something really simple - linear regression and synthetic toy dataset that we can generate with sklearn kit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVwAAAFcCAYAAACEFgYsAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3Xl0W+WdN/DvvVerLVmSFXnJ5tiJ\nE4c4SwnQJg2kTeJsJpASWtoCLWkpzAyFoZy+523LeWlLppy2Z6Zn2pm+QAgltDNd3mlKAjFLIFOS\nAO2EshlnJcEmC7G8SbZsa7vL+4csWbZlx05sXUn+fs7paXQtWT+7x98+eu7z/B5B0zQNREQ04US9\nCyAimiwYuEREacLAJSJKEwYuEVGaMHCJiNKEgUtElCYGvQuYSK2tAb1LGDWXKw8+X6/eZVwy1q8v\n1q+v5Po9Hvuwz+MIN0MYDJLeJVwW1q8v1q+v0dbPwCUiShMGLhFRmjBwiYjShIFLRJQmDFwiojRh\n4BIRpQkDl4goTRi4RERpwsAlIkoTBi4RUZowcImI0oSBS0Q0ThRVHfHrDFwionEgKyo6usIjPien\n2zMSEaVDVFbREQhDVUc+BJ0jXCKiyyArKto7g+gJRvGngx+O+FyOcImILlFUVuELhCCaDNix9ygu\ntI/cRJ2BS0R0CaKygo5AGL5AGDtfeA8tvuBFX8PAJSIao0hUga87DK8viKfqjqGzJwIAmFU6/PE6\nAAOXiGhMwlEF/kAY59p68NTzx9AbkgEA82Y68eU1c0d8rW43zS5cuIDbb78dGzZsQG1tLZ5++mkA\ngN/vx9atW7F27Vps3boVnZ2dAABN0/BP//RPqKmpwaZNm3DkyBG9SieiSSoUkeEPhPHhhS7seO5o\nImyvml+M29bOhckwcqTqFriSJOE73/kOXnjhBfzhD3/Ab3/7W5w6dQrbt2/HsmXLsG/fPixbtgzb\nt28HABw8eBBNTU3Yt28ftm3bhh/84Ad6lU5Ek1AwLKOzO4LjZ3x46vljCEcVAMA184vwtU0LYJBE\nuOzmEb+HboFbVFSEBQsWAABsNhsqKirg9Xqxf/9+bN68GQCwefNmvPLKKwCQuC4IApYsWYKuri60\ntLToVT4RTSLBsIzOngjeO92G37x0ErISW2973eKpuHFFOYwGEYV2C0zGkU/vzYh1uOfOncOxY8ew\nePFitLe3o6ioCEAslDs6OgAAXq8XJSUlideUlJTA6/XqUi8RTR69oVjYvnnMiz/sPwVVi4Xtumtm\nYP0nZ8IgiZjitMJ4kekEIANumvX09OC+++7D9773PdhstmGfp2lDd3AIgjDi93a58rLqvHuPZ+Q7\nnJmO9euL9Y+/7t4IIhDwt/c+xjOHGgEAAoAvrZuH6z4xHUZJhNtphSQKo6pf18CNRqO47777sGnT\nJqxduxYA4Ha70dLSgqKiIrS0tKCwsBBAbETb3NyceG1zc3NiJDwcn2/kRciZxOOxo7U1oHcZl4z1\n64v1X76Gxna8Vn8Brf4gPE4rls7zYLrHhn1vnsWBdz8GAIiCgM9/djaqy1zo7grCaTejo717QP0j\nBa9uUwqapuHBBx9ERUUFtm7dmri+atUq7N69GwCwe/durF69esB1TdPw7rvvwm63XzRwiYhGo6Gx\nHbsOfAivLwhVAz5u78HuQ434zUsnEmFrkATctm4uFs+ZArNRgstuhniRT9mD6TbCfeutt7Bnzx7M\nnTsXN954IwDggQcewF133YX7778ff/zjH1FaWoqf//znAICVK1fiwIEDqKmpgdVqxSOPPKJX6USU\nY16rv5D4t6KqUBQN/u5wYquu2SjhK+vnoby0AFazAY580yW9j26Be9VVV+HEiRMpvxZfk5tMEAR8\n//vfn+iyiGgSavXHtuUqigpF1dARCCMciS37yrMYsHVDFaZ5bLBZjbBZjZf8PhmxSoGISE8epxWK\nokJWNLR3hhJha5AE3HXDAkzz2FCQd3lhCzBwiWiS0zQNSyqnIKpoaO8KISLHTm2QRAGbry1HsdMK\nR74JeZbLC1sgA5aFERHpRdM0+LsjcNktCEVkRPvC1mKScOO15VgyZwqcNjPMF9nQMFoMXCLKSYOX\nea1YVIrqcnfi6/Gw/bitB796/hg6u2Mdv2YW2/DV9VXItxjgspthHMe1/AxcIso58WVecV5fMPG4\nutwNVdPgD4RxpqUbT9UdQyAYBQDMmebAbWvnwmqOha1BGt9ZV87hElHOSV7mNfi6qmnwdYVx6nwn\nnnjuSCJsF8wqxFfWz0OexYDCgvEPW4AjXCLKQfFlXoO1+ILo6Arh+Ed+/Me+E4kbZFfO9eBz11XA\napLgvIQNDaPFES4R5RyP0zrkmqZpcNhMeO9UO55+8XgibJdXl+CmlRXI65tGmKiwBRi4RJSDViwq\nHfBY0zQoioaCPBN+98pJKH3Hma9eOh21y8qQbzHCZTdftCHW5eKUAhHlnPhqhNfqL6DFF4TTZobF\nLOFQ0tzu9cvLsLy6FPkWA+x5l7ZVd6wYuESUk6rL3aia6UJ7Vwiv/O0c9r91DgAgCsBNK2fjyrke\n2POMyB+HDQ2jxcAloqyXas1t1UwX2rpCqHu9Ca83xFq7SqKAL62pxIJZhSjIN8FqTm8EMnCJKKul\nWnP7x1dPY9WV01B/ugNvn2wFAJgMIm5bNw+V0x1w5pthNqX/cAIGLhFltcFrbjVNQ1RW8acDH6Kr\nN7bG1mqW8NX1VZhVYh/33WNjwcAloqyWvOZW7QtbX1cI4Whs2ZfdasTW2vmYOiUfhROwe2wsuCyM\niLJafM2tqmmIRtVYe8W+sHXZzbjrhgWYPiUf7gnaPTYWDFwiymorFpVCVTVEoiraOkOJjl8Omwl3\n3bAApe48FBZYIIn6xx2nFIgo41ys01eyOdMc+NSCYuw51AhZiYXtFIcFd9+4AIV2C5w204RvaBgt\nBi4RZZSLdfpKFgzLOHW+E/sOn01s1S0vteP2dfPgsplRkJ85YQtwSoGIMsxInb6SBcMyjn7kw+PP\nHkFnT6yXbdVMJ+7YMB/uAgsctonfqjtWHOESUUYZrtNXqz+U+HdvSMZ7p9vw6xdPIByNnT+2eI4b\nN39mNpw2c1p3j40FR7hElFFSdfqKXbcAAHpCUbx53Iunnj+WCNtr5hfhC5+dg0K7JWPDFmDgElGG\nGdzpK/l6dzCK1+ov4DcvnYSsxDp+rVwyFZuvLUdhgSXtW3XHKrOrI6JJJ7nTV6s/BI/TghWLSlFW\nbMd/v30Oew41Qut77vprZuIzn5iq6+6xsWDgElHGqS53D1iR0NUbwQt/PYMXD58BAAgAbry2HJ9a\nUKL77rGxYOASUUbzd4ex57VGHHj3YwCAKAj4wqrZuLLSA1eBOSM2NIwWA5eIMpavK4Q//PkUDh9r\nAQAYJAG31sxFdbk7dhyOmFnLvi6GgUtEGam9M4jf7DuJ+tPtAACzUcJX1s9D1UwXHDbThJ49NlEY\nuESUUTRNQ6s/iKdeOI4TZ/wAgDyLAVs3zsfsqQVwZNjusbFg4BJRxtA0Dc0dvdix9ygaLwQAAI58\nE7bWzsesYjsK8tNz9thEYeASUUZQNQ3nW7qx/bmjON/WAwBwOyz42sb5mFFkg82auRsaRouBS0S6\nU1UNTRe68PizRxNbe0sK87B1YxWmTclHXgbvHhsLBi4R6UpVNRxpbMe//+l9+LtjTWhmFttwx/oq\nFBfmZfzusbHInZ+EiLKOqmo40tSOJ+uOo6uv41fldAduWzsXRa48mI2Zv3tsLBi4RDRhRmokrqgq\n3j3ZhiefP4ZQJNaEZkF5Ib60eg48zjwYDdmzoWG0GLhENCFGaiReNdOF/znqxa9fOpE4EmfpPA+2\nrJyNKQ5L1mzVHSsGLhFNiOEaiR9872O0+IL43SsfQFFjbWhWXz0Da5dOg7vAmnW7x8aCgUtEEyJV\nI3FN03D6fBfeOtEKra/l15qrpuPL66qgRuWs3D02Frk5bici3Q1uJK5pGvzdEfgC4UTYXr98FjZ+\nqgxTnNacD1uAgUtEEyS5kbiqqvAFwomVCKIA3PyZ2VizdDqcGXj22EThlAIRTYj4aoRX3zmPk2c7\n0R2MAgAkUcCX1lTimvnFObF7bCw4wiWiCVNR6oAAIRG2JoOIr26owqeumHxhC3CES0QTJNAbwfZn\nj+BIkw8AYDVLuGN9FRbOdsNimpzRMzl/aiKaUL5AGI/ubsCp850AALvViK/VzkdVmSvndo+NBQOX\niMZVe2cQv9j1Ps62dAMAXHYz7rz+CsyZ5sjJ3WNjwcAlonHT3N6DX+x6H80dvQCAIpcV36idj7LS\ngpzdPTYWDFwiGhcfeQP4t1316OgKAwCmefLxjdorML3IlnL3WHKfhenFBbh63pQBJ/XmIgYuEV22\nD8758ctnGhLrbMtLC/D12iqUuPNTbmgY3GfhQls3djV3AUBOhy7H+ER0WY40deDn/1WfCNuqmS7c\nfeMVKB0mbIHh+ywMdz1XcIRLRCmlaq0IYMC14kIr9h0+h3A01l6xYmoBLCYRj+85MqQdY7JUfRZi\n10MT9wNlAAYuEQ2RqrXif+47CQ1InMBw+nwnDh9rSTynaqYTgWAUvm4t8Zr49xgcuh6nFV7f0ND1\nOC3j/aNkFE4pENEQqT7aB4JRdAej0DQNXT0RtPfdHAOAzyyZijyzlHIlQqrvldxnYTTXcwUDl4iG\nSPWRX1ZURGUFXT2xjl9xjnwjbllVCV/feWRDv9fQaYLqcje2rKxAsSvWJWzqFBu2rKzI6RtmAKcU\niCiFVB/5JVFAVNYSBz0CgMNmQuU0B8wmaczTBNXl7kTAejx2tLYGxvEnyEwc4RLRAA2N7ejsDuNC\new9a/UGEwjJUNXYMjqxoiee57GY48k24bslUAJN3mmAsOMIlooTkm2VOmxndwSjau4Kwmo0IhmMr\nEQQBKLRbUDHVjmsXT02MUuP/HVvFEILHaRl2lcJkpWvgfve738Wrr74Kt9uNvXv3AgD8fj++9a1v\n4fz585g2bRr+9V//FQ6HA5qm4Uc/+hEOHDgAi8WCH//4x1iwYIGe5RPlnOQbXFazAWajiDa/hkBv\nrL2i2Sjhzuvn4xNzPSnX2CZPE9BQuk4p3HTTTdixY8eAa9u3b8eyZcuwb98+LFu2DNu3bwcAHDx4\nEE1NTdi3bx+2bduGH/zgBzpUTJTbkm+WKaqKFl8Iwb4jzPMtBvzD56px5TBhSxena+BeffXVcDgc\nA67t378fmzdvBgBs3rwZr7zyyoDrgiBgyZIl6OrqQktLy5DvSUSXLn4OmaKoaOkIJjY0GCUR921Z\niIUV7klzHM5EyLibZu3t7SgqKgIAFBUVoaOjAwDg9XpRUlKSeF5JSQm8Xq8uNRLlqhWLSiErKrwd\nQUTk2I0ySRRwy+rZqJzh0rm67Jc1N800TRty7WL/T+ty5cFgyJ5mxx6PXe8SLgvr19d41F8lCOh9\n4TiiSixsrWYD7ryxGms/WXbZ3/tiJsPvP+MC1+12o6WlBUVFRWhpaUFhYSGA2Ii2ubk58bzm5ubE\nSHg4Pl/vhNY6nrJ9HSLr19el1D+4V8K8GU4890ZTYp1tWbEd37xpIdwOS+J7p+qvMB43yXLp9z9S\n8GbclMKqVauwe/duAMDu3buxevXqAdc1TcO7774Lu91+0cAlotTiy7+8viBULdbL9j9f/iARtpXT\nHfjWFxbD7bAM+5p4r4SGxna9foyso+sI94EHHsDhw4fh8/lw3XXX4d5778Vdd92F+++/H3/84x9R\nWlqKn//85wCAlStX4sCBA6ipqYHVasUjjzyiZ+lEWS15+VcwLKPVH0R81m5hRSHuvmEB8izGYV8z\n+DqXgo2OroH7s5/9LOX1p59+esg1QRDw/e9/f6JLIpoU4su/ekNRtPlDiN8hybMYcM/nFsKU4qDH\nydpScTxl3JQCEU08j9OK7t4oWpPC1mYxYH6ZM2XYxl+T+nput1QcTwxcoknIZjWivat/ZGrPM8Ji\nltDdK2Pb02/isT0NQ+Zm2Svh8mXcKgUiujwXW0nw3BtN+O+3zyceO20mFDkt6OyJItS30SFV83D2\nSrh8DFyiHJLqpIb44wWzCvH//nwKLx0+CwAQBeCW1ZVYs3Q6Hn/2CMLy0LXug2+IsVfC5WHgEuWI\nhsZ27HzhOLqDURgkEXZrbJVBIBjFY7sbkG81Jm5wGSQBX1k3DysWxVor8oZYejBwiXLA2ydasOvA\nh+gORgENkGU1MUcrAJBVDb197RUNkoi7brgCV83rX8c+Wc8YSzfeNCPKAa8cPgMgFqaqqkFWVMiK\nBlnREFW0xBpbQQBmT7UPCFuAN8TShSNcohzQ3N4DADAaRARDcsrniKIAj9OCSIq5Wt4QSw8GLlEO\nKHHn40xzF6KyCkkSBhyFA8SmFYpdVpiM0qjOGKOJwSkFohyw5pqZAGIn6yJFZz1JQmJDA6cJ9MPA\nJcoBV84rwpaVFbCYJPS1sQUAGEQBFpMIs9GAYpd1UhxFnsk4pUCUI5w204CpBFEAIGgQRQnrPjkT\n1y+bpVttFMMRLlEOON7UgR//5zvo7bthJoqx/5iMBtisRrx1opVtFDMAR7hEGWq0zb6PNLbj359p\nQLjvsEeX3QSb1QhRHDieYhtF/TFwiTLQSFt0k0Pz7ZMteGzPkcRUwoqFJTjbEgCEoR9euWtMf5xS\nIMpAIzX7jnv9/Qv4v7v7w7bm6hnYunE+igvzU76Wu8b0xxEuUQYa3NsgFJYRCEbR3N6Dx/Y0IM9i\nwIF3Pk70sr113Tys/sQ0ALFlX8mj4zguB9MfA5coAyX3NgiFZfgCYQCAwSDixFk/OvvOHhMAfGlN\nJb64tipxiCF3jWUuBi5RBkoepfq7w5AVFaoW29jQ2zcVKwD4eu18LF84dOTKXWOZiYFLlIHiYVn3\nRhMisgposYBVkzaReVzWlGFLmYs3zYgyVHW5Gw6bGUZJhAYgecOuyShiVoldr9LoEjFwiTJYU3NX\nbIQ7iKpqvAmWhRi4RBmqNySjvSs85LooAFazgXO0WYhzuEQZqKs3gn/+3TtQknojGEQkdo/lWYx6\nlUaXgYFLlCaj3arb0RXCT3/3Dlr6loVJogCjQYCqxU50sFmNKCu2pbt8GgcMXKI0GO1WXW9HL376\nu3cS625LCq0wSCIkaeDsH+dvsxMDlygNhtuqW/dGU2LUm28x4PT5LgT7mtBUzXThHz+/CB+c83MT\nQ45g4BKlQapjyINhGRfawyh1qwiFZTQ1BxKHNSypnIJ/2FwNgyRyE0MO4SoFojTwOK1DrnUHozBI\nInpDUbT4gomwdReY8c3PLYRB4p9nruEIl2gCxW+UfeQNINAbhd1qhMUc+7OTFRUWozSgbaI9zwh7\nngmiKOhVMk0gBi7RBInfKAuGZXQHo4hEFbRFFBTkm1A53YFQWE40qAFiJzTIigqzSdKxappI/MxC\nNEFeq7+AYFiGPxCGLKsQBQGSKCAcVWAxSQPDVgAMoghF0eALhHkcTo7iCJdoHCWvtW3xBWPHlifR\nNA2hsIyD7/WvWjAaBAiCkFhjazUbEsfhjHbtLmUHBi7ROBm81lYDEJFVSIIAURSgqCqS81cSBbgL\nzLCm2DXW6g+Neu0uZQ8GLtEYDTfqHLzW1m41IhiSIWvawL6KfdwOM1w2C0JRZcjXPE7LiMfsMHCz\nEwOXaAxGGnUOXmurAX0j26Fh67SZYDUb0ROKoCMQgayoMEhiYhXDikWleObg0GNyYu/DwyCzFW+a\nEY3BSKPOwWttu4NRDOxiO/Brze29aO4IwmY1wiCJkBUVgWAUS+d5UF3uTrl2F+BhkNmMgUs0Bql2\njMWuhzC9yIbm9h6c9QZw1htAb0iGMrSVLQBAUTREZAVq3+jX47Si1J0Pj9OKcy3dAIbvl8A+CtmL\nUwpEAN4+0YK6Q6cvuhog+XDHZFFZQd0bHyEcHSZhU9A0QJIEdAejsJr7/xTjUwY8DDL3MHBp0mto\nbMezrzch2neywkirAVIdQR4My+jsjiAqKxAw3CTCQELfRjJREIYsHUueMmAfhdzCKQWa9Eaal03F\nYhTR5g+izR+EoqgIhuTY9IA2fNgKKXbqxnslDO6ZwCmD3MURLk16rf7gkH6zsesDVwMkr1CY4rQi\nGJbR4gtC1bRhR7YCYgc+KqoGOen0Bk2LbYLItxpR7LIiImucMpgEGLg06XmcVnQEhp4dNng1QHyr\nbncwCllRoSgaBAGJLl+DCQAcNhOKXdZE60UNGgQIEATAZJRQ7LLif9+6dAJ+KspEDFya9FYsKsWz\nrzelvJ7sI28A/kAYqqpB1bRUexkSBADTPPn4wqo5qC53Y9vTb6Z8fkQezYwv5QoGLk161eVuOBx5\nfasUhl8NEJVVqKqWciNDnCjERrVbN84f8PrhVjdwTe3kwsAlGkBDZ3cYdW98hGcOfjhgiZjREJuL\nHS5unTYTHDYztqysGNXqhvh1mjwYuDTpJS8LC4ZlnO+bz3XZzQOWiDnzTfB2pN74IInA3BnOYW96\ncU0tAQxcogHLv2LbcWMCwWjidIbX6i+gJySnfL0oAkvnFeHvbqwe8X24ppa4DpcmveTtusmbEJL/\n3eoPoaNrmKYxGqcGaHQYuDTpJTeJSd6EkPxvt8OcctuuKABGo8SRK40KA5cmveTRqc3a3wzc3vfv\n3lAURxt9A1YnGCQBJoMIgyRiVrEtfcVSVuMcLk161eVutAYiqHvtQ/QEozCbJDjyTTAaJEgi4A+E\nEU3aJRbfpWswxI7EqV0+S5e6KftwhEuTXkNjO/5S/zHseSaUuPNRWGCBJIlYd810nG/rSYStQRIw\nxWGGyShC02LB67KZ9S2esgpHuDTppWpSIysqnnr+BMJ9x98IQuw/gd4oFFWDJAmY4rQiFFV4zhiN\n2rAj3N///vfprININ4ObikdlFc3tvf1hC8AgChAgICrHeygMbP81XGcxomTDjnBfeuklvPzyy3jk\nkUdQXFyczpqIJkyqAyCTm9dEogq8Hb2JvgdCXxswDbHgTTSqGbTdjOeM0WgMO8J96qmnUFNTg1tu\nuQXPPPNMOmsa0cGDB7Fu3TrU1NRg+/btepdDWSTeXtHrC0LV+huNTy+yoTcURXNHLy6094etmNQJ\nTOlrWCOKAiRJwOANvuyJQKMx4hzuF7/4RXzqU5/CzTffjB//+McQRRGaFvs49Ze//CVdNSYoioKH\nH34YTz31FIqLi3HzzTdj1apVmDNnTtproeyT6mN/MCzj5TfPors3OqRHgoCk0NUASRThsJngD4TZ\nNJwuyYiBW19fj+9973u4/vrr8fWvfx2iqO+ihvr6epSVlWHGjBkAgNraWuzfv5+BS6MyeK42GJbh\nD4QRkVOfQ6ZqgCQKUPqGuVFZSZw95rKbEYmq7IlAYzJs4P7zP/8zXnzxRTz88MNYvnx5Omsaltfr\nRUlJSeJxcXEx6uvrdayIsklyi8RQWEZ7Z2jEVosaAFGM3RxT+0K32DX8AZNEFzNs4HZ0dGD37t2w\n2TJnF42WorX+4LvFyVyuPBgM0kSWNK48HrveJVyWTK+/9trZ+M3zR9EbkuHvjqQM28EnOAhCbM5W\nAlA504n/c+ey9BU8Rpn++7+YyVD/sIH7yCOPjGsx46GkpATNzc2Jx16vF0VFRcM+3+frTUdZ48Lj\nsaO1NaB3GZcsG+qfUWjFDZ+ehaeeP4aorKR+ktYfuqIQOxLHIImwW41Ye9X0jP0Zs+H3P5Jcqn+k\n4M2qjQ8LFy5EU1MTzp49i+LiYtTV1eFf/uVf9C6LssgVswohKwOPx4nfGNMAQABMBhFWs4GHO9K4\ny6rANRgMeOihh3DnnXdCURRs2bIFlZWVepdFWaChsR2H3vsYR5t8ib62ggA48k0IhhVEZAWCFgvb\nWaUFqF1WxoClcZdVgQsAK1euxMqVK/Uug7JIQ2M7/vjqabT6gghG+qcSnDYzjAYRvSEZJoMEp80E\ni9mAUGSY6Qaiy8TmNZTzDrxzHt6O/rAVRQHuAjMEAegJRmEwiHA7LInTHQBu1aWJkXUjXKKx6A5G\n8f6HHYm1tpIooLgwD0aDCFEQgPzYfK7RICIqDzzhgWi8MXApZwzuk7Ckcgqee70pEbZGSURRoTWx\nSyy+HZfHl1O6MHApJ8T7JMSda+3BWydaE2ttTQYRRYV5kMT+ddvx7bg8vpzShYFLWSNVp6/k48fj\nIlEl1qCmL2znzXCi5urpOHysZdgjyv92og1nvQEuAaMJxcClrDB4BBvv9AX0HZHT1ychFJHR4gsm\ndotZzQY8cMtiGA0SrpybepNMdbkbn71mVlYvvKfswMClrDDcqoHX6i+gutwNk0FCY3MXIkkn6+Zb\nDLhilgvGUW7vHmkETTQeGLiUFQZ3+uq/HkJDYzvOtAQGhK2AWODOKLbjsT0N+MgbQFRWYTSIKCu2\nDwnTt0+0jDiCJhoPDFzKCsmdvgZet+C3+04gGB66WUHVNLx1ojXRhjFB61+ZEA/TVw6fSfm+8RE0\n0XjgxgfKCsOtGsizGNDs618zK/b1QjAaRPgCEQCxtbjJAn2Pk6cpmtt7Un5/rsel8cQRLmWF5NUI\nrf4QpjjMMEgiXn3n4yHPlRU10YwmGJYhK+qQrwMDw7TEnY8zzV1DvhfX49J44giXMl5DYzse29OA\nZw7G5lQ3XzsLTpsZfzniHfJcVUP/mWSiAH8gjMEdkwdvfACANdfMTPneXI9L44kjXMpog5eDNXf0\nYvtzR9ETlBPXRAEY0ktciN04kxUVkiQOCF271QhgYJheOa8InSsrEiNorselicDApYwyeGlWZ3f/\nzS5N09DqDyEY7g9bSYwd7qhqGmRFS7ouQBQEqBqgKCoctlizGpNBwsxiW8owrS53M2BpQjFwKWMM\nHs1+1BxAqz8IURRgNIiQFS3RYEYUBJgMQtKR5gJEQRvwOPE8s4TK6Q783Y3Vaf15iAZj4FLGSF41\nEArL8PUt5VIUDYqiJI4xl0QBV8xyIhJV0Xihf3dYbETbd8KuosaOORcF2KxGrjagjMCbZpQxkjc3\nxJduCUJstUE8bAUBKHHnoebqmahdPgtOuxkGgwgIgNR3M0wQkl7Q999cbUCZgCNcSrvhttAmb26I\nLe3SMGhFF4ySgFtWzUnMtd6WpoiTAAAZMUlEQVS2dm7iRldnTxhWs4TeviN0VE2Dqmpo8wfhspnQ\n0NjOOVrSFQOX0qahsR11bzShsTmQOAk3eQvtikWl2HXgQ4TCMqJRFckLDwQABkkYcux98o2ubU+/\nCVUDzEYJ/p4IlGj/tEIoqnKrLumOUwqUFvEbYmdbewANkGUVvkAYob4VB/EttEvneeDvDmPwKi9R\nBARBgMkgYucLx7Ht6Tfx2J4GNDS2J57jcVoBABazAZIowCiJMEgiTMb+kObROaQnBi6lRTzoBu/6\nis/Vxm9qHW3sPw5nIAF5FgN6QzK6g1GoWn+DmXjoJq+rTX4fW9+62+T3IdIDA5fSIn5DLL7LKy4e\njB6nBe+dbsOJM/5EL9vYKDU2qpUkIRHEg79HPMyry93YsrICxS5rbHRrEOG0m2FNOhySN89IT5zD\npbSI3xCzWY0DOnfFw3OKw4J/3/V+YirBnmeExSShszuSeF48nO1JI1Zg4Kg1Pqc7eE1vHLfqkp4Y\nuDRuRmrgHb8hFh9tdgejkBUVM4psmObJxwv/cyYxsnXkm+C0mwHE5m27g1EU5JkQkRUYJXHAceZA\n6lHr4GY33KpLmYCBS+NiNA28LUYJ51q7AQAzPPmoXT4LHzUHEs8TANyyuhJTp+QlgnJWSX+z8LGO\nWrlVlzINA5fGxUgNvIH+k3Gn9K0kCEYUHHz3Y/ztRCuAWAOaOzZUYcWiqQBSL93iqJWyHQOXxsVI\nDbwHL8XSNA0dXWGc8cZGu5IooLzUjj+/cx4NjR0jhihHrZTNuEqBxkWJOz/ldY/TMmDLrqZpaPOH\nEqcwGA0i3A4LooqWcqkXUS5h4NK4GKmBd3xDgqZpaPEF0du32UESBcyZVjBg2VYcNyhQLmLg0ri4\ncl5RYg2sKAgodlmxZWUFqsvdWLGoFKqqwdvRi1Ck/7BHh82IM97uAf1t47hBgXIR53Bp3Aw3v1pW\nbEcoIiPcd4y5AKDQboItz4xQJJhYl8sNCpTrGLg0of5y5AJ+89LJxMjWZBDhcVoSTWhMBhGhsIw2\nfxAWswE2qxFWs4EbFCgnMXDpsjU0tuPNF0/gnLdrwIaHQ++dx69fOgml7xgGo0GEqqmQFQ0GQ+xE\n3d6QDFEUoKoaZEVFdzDKpV6Usxi4dMli7RY/QmNzF0wGCXkWA9S+VQZeXxB/2P9BImzNRhEepxUd\ngTACwSgsZkNipUL8GJz4zbVzLd26/UxEE4mBS5ckvuur1R8ENCAqq4m5WEEAfv9Kf9haTBI8TgtE\nUYz1Uug7GDK5o5edHb1oEmDg0iUZrt2ivzs8oHl4ntkAt8MMUYwtiLGaDXDZTHDYzGjzB6EhFrYW\n3jCjSYCBS5ckud2i3Nc2UVHVAUfiVJcXorMnnAjbuNrls9jRiyYlBi5dkni7RbvVCF8gDFkZGLZX\nzp2Cez63EEeaOobtfcDeCDTZMHDpksTbLVrMBpiCUXSH+jcvLK8uxp3XLwBw8d4H7I1AkwkDly5J\ndbkbmqbh9/tPJcJWAPDF1ZWouXqGvsURZSgGLl0STdPwzsk2XGjvBRA7Ufcr6+Yl2isS0VAMXEpp\npNMbFEXFE3uP4vCxFgCxDQ3fvnUpKkvtepZMlPEYuDTE4NUDyac3zJ3hxP99pgH1p2PtEy0mCf+w\nuRrLF01Fa2tAl3qJsgUDl4YYrjXiq++cx943mnDybCeA2PHj9960EJUznOksjyhrMXBpiOSG4XGK\noqLhw47EUeVOmwn/+PnFKCvmNALRaLEfLg0R72kQJ8sKmjuCibD1OC34X1/6BMOWaIwYuDRE8k6v\naF/YxrfwTpuSj//1xU+gdJgjdYhoeJxSoJQsRgkfNXchnNQXobzUjnu3LILTZta1NqJsxcClAeIr\nFEJhGRG5P2yne/Lxrc8vhi3PpGt9RNmMgUsDvFZ/Ab0hGW2dQWh9aWs1SShyWRm2RJeJgUsDNF7o\nRJs/3N9e0WKAu8AMXyCia11EuYCBSwmvvnserf5w4rHNaoTLboIoiuxRSzQOGLgETdOw88XjOPRe\n/4aHPLOEwgIzBEEAwB61ROOBgTvJqZqGJ547gv852pK4lm8xQNU0hKMqyopt7FFLNE4YuJOYrCj4\n3SunBoSty26GPc8IQRBQ7LLi726s1rFCotzCwJ2kIrKCnc8fx1+PehPX3AVm5FuNiWkEHuZINL50\n2Wn2wgsvoLa2FlVVVXj//fcHfO3xxx9HTU0N1q1bh0OHDiWuHzx4EOvWrUNNTQ22b9+e7pJzSigs\n47HdDYmwFQVgisMyIGwBHuZINN50Cdy5c+fi3/7t33D11VcPuH7q1CnU1dWhrq4OO3bswA9/+EMo\nigJFUfDwww9jx44dqKurw969e3Hq1Ck9Ss96gd4IfrGrHu+eirVXzLMYcMOKWcizGAaELcAbZUTj\nTZcphdmzZ6e8vn//ftTW1sJkMmHGjBkoKytDfX09AKCsrAwzZsSObqmtrcX+/fsxZ86ctNWcC3yB\nEP79T++j8UKsb21Bvgl3b7oC88pcqJjq4GGORBMso+ZwvV4vFi9enHhcXFwMrzf2sbekpGTA9XgQ\n08jiJzd83NaD9q4QgmEFAFBYYMbf31iNiqkFEASBhzkSpcGEBe4dd9yBtra2Idfvv/9+rFmzJuVr\ntPhe0iSCIEBV1ZTXL8blyoPBII2i2szg8Yxvu8O3T7Tg2debEI7KaO7ohazEfr9uhwX/5+ufRMVU\nx6h+j6M13vWnG+vX12Sof8ICd+fOnWN+TUlJCZqbmxOPvV4vioqKAGDY6yPx+XrHXINePB77uB9R\nU3foNLp7I2j1h6CosbA1GkSUuq0oMEloa+set/eaiPrTifXrK5fqHyl4M6of7qpVq1BXV4dIJIKz\nZ8+iqakJixYtwsKFC9HU1ISzZ88iEomgrq4Oq1at0rvcjNDQ2I7H9jRg29Nv4rE9DWhojN0M0zQN\nH3kDaPH1h63JGNui2xNU9CyZaNLSZQ735ZdfxrZt29DR0YG7774b8+fPx5NPPonKykps2LABGzdu\nhCRJeOihhyBJsSmBhx56CHfeeScURcGWLVtQWVmpR+kZZbjDHjVNg6IAbf4Q1L5pGotJgtthhkGS\nuNyLSCeClmriNEdk00eUS/lI9dieBnh9A88f0zQNBknEuZbuxJE4VrMBhQVmGKTYB5otKyvG/QZZ\nLn0kzEasX1+jnVLIqFUKNDaDD3vUNA3dwSg6uvo7fl0xywWzUYS/O8rlXkQ6Y+BmMY/TmhjhapqG\nrp4I/N39fWtXXTkNm68th83KxuFEmSCjbprR2MR3gmmaBn/3wLCtXVaGm66bzbAlyiAc4Wax6nI3\nIlEF//Xn0+jqiYWtAGDLZ2bjs5+YBquZ//MSZRL+RWaxYDiKdz5oS0wrSKKAL62uxPKFJbCY+D8t\nUabhX2WWCvRG8esXj+Otk60AAJNBxO1r5+GqqiKYTdmzu45oMmHgZiFfIIxfPX8MRxo7AMTW2G7d\nWIVFs6fAbGTYEmUqBm6Wae0MYsdzR/HBuU4AgN1qxNdq52N+mQsmhi1RRmPgZglN0+Dt6MVjzx7B\nGW+sB4LTZsKd11+ByukOGLOoSQ/RZMXAzQKapuFsSzcef/YILrTHGvIYJAEl7jxEZIVhS5QlGLgZ\nTlU1fPhxJx5/7ijaO2NnjBkNIgoLLOgNydjzWhMMksjdY0RZgBsfMpiiqjj+kQ+/fKYhEbYmgwh3\ngQVmo5joZfta/QU9yySiUeIIN0O8faIFdYdOo9UfhMdpxbIFxVA14Mm6Y+gNyQBiqxFcdjOMBnFA\n43CerkuUHRi4GaChsR3Pvt6EaF93r+aOXvzHyx+gqyeSuLZothvQNPh7IkNOaWC7RaLswCmFDJA8\nJaCqsY5f7Z2hRNheM78IX15diVVLp6c8Eoen6xJlBwZuBoi3WYyHbXJ7RY/TgvOt3dh18DREUcCW\nlRUodlkhCgKKXdYJ6W1LRBODUwoZwOO0orUzhK7eCDqTOn4ZJAEmowRJFNDiD2HXgQ+xZWUF/u7G\nah2rJaJLxRFuBlg6zwNfV2hI2DpsJkiiMGAagSsSiLIXR7g683eHcbTJN6CX7YyifATDMvIsxiFz\ntlyRQJS9GLg60TQNHYEw/vjqafzPUS+A2Kj21pq5WFjhxh/++wO0pAhXrkggyl6cUtCBqmlo8wfx\nH/tOJMI21vFrPhbNngKX3YxrF09N+VquSCDKXhzhppmiqmjx9eI3L53E8TN+AECexYB//OIn4LGZ\n4Mg3QRCExMqD1+ovoNUf4gGQRDmAgZtGsqLiQnsPdr5wHI0XYkcqF+Sb+torFkIORwc8v7rczYAl\nyiGcUkiTqKzgrDeA7c8eTYStu8CCu2+4AmXFdrgKODdLlOs4wk2DcFTBR94Antx7LLHJoaQwD1s3\nVqG4MA8FeTxZl2gyYOBOsGBYxocfd+FXzx+DLxDbQTaz2Iavrq/CFIcFdoYt0aTBwJ1APaEoPjjX\niafqjiEQjM3PVk534NaauSgssMBmNepcIRGlEwN3gnT1RnD8Ix+efvE4gmEFALCgvBC3rJoDl92M\nfAvDlmiyYeCOM03T0NUTwfsfduA/9p1ApK/j19K5Hmy+rgIumwl5DFuiSYmBO440TYO/O4J3PmjF\n7175AIqqAQA+XV2CDcvK4LKZYTXzV040WfGvf5yomgZ/IIy/HvVi14HT0GJZi9VLp2P1ldPgYNgS\nTXpMgGE0NLb37fKKHXkz0i4vRVXhC4Rx8L2PsfeNjxLXr19ehk9Xl8JhM8Fi4q+aaLJjCqTQ0NiO\nXQc+TDz2+oKJx4NDV1ZUdHSF8PLfzmH/W+cAAKIA3LRyNpbO9cBpM8Ns4jHmRMSdZikN13N28PWo\nrKKtM4jnXm9KhK0kCvhyzVwsneeB086wJaJ+HOGmEN8NNvR6f7vESFRBW1cIfzrwId4+2QogdoT5\nbWvnoXKGA4V2M4wGhi0R9eMINwWP0zrM9Vi/g1BERqs/iN+9/EEibC0mCV+rnY+5MxwotFsYtkQ0\nBAM3heF6zq5YVIrekIwWXxBPv3gCR5o6AAB2qxF33bAAs0rscNktMBr4ayWioTilkMJwvWhnlRTA\n6+vF0y8exxlvNwDAZTfjaxvnw+OyotBuhkFi2BJRagzcYQzuRdvVE8GF9h489fxxNHf0AgCKXFZs\n3TgfhXYzXAxbIroIBu5FaJqGzp4ILrT34ld1x9DeFbtxNs2Tjzs2VMGRZ4KrwAxJZNgS0cgYuCNQ\nNQ2d3RGcbe3GU3XH0NkTO1m3vNSO29fNg81qRKHdAlEULvKdiIgYuMNSVQ2+QBhN3gCeev4YekMy\nAKBqphNfWjMXeWYDXHYzw5aIRo2Bm4KiqvB1hfHB+U78+sUTCEdj7RUXz3Hj5s/MhtVkgNNuhigw\nbIlo9Bi4g8iKio5AGMeaOvCfL5+ErMS60HzyimJs+vQsWIwSw5aILgkDN0kkqsDfHcY7H7Thv/58\nGmpfy6/PLJmKmqtnwGIywGmLHWNORDRWDNw+4UgsbP961ItnX2tEX3dFrP/kTFy3eCrMRolhS0SX\nhYGL2EGPXT0RHHj3Y7x4+AwAQACw+dpyXD2/GBaTBEc+w5aILs+kD9zuYBSB3gj2vXkWB979GECs\n49fnPzsHi2a7YTVJcNjMOldJRLlgUgduV28E3cEonn2tEYePtQAAjJKIL9dUYt5MF6xmAxz5PMac\niMbHpAzc+O6xnlAU//Xn06g/3Q4AMBslfGX9PJSXFiDPYkBBHsOWiMbPpAvc+NljPSEZv33lJE6c\n8QMA8iwGbN04H9Om5CPfYoCdYUtE42xSBW5891ggGMGvXzqBpgsBAIAj34SttfNR5LTCZjXCZuUx\n5kQ0/iZN4MqKCn8gjM7eCHY+fxzn23oAAG6HBV/bOB8uuxn2PCPyLQxbIpoYkyJwo7IKX3cYvq4Q\nfvX88cQROqXuPNyxoQr2PBMK8kzIs0yKXwcR6STnEyYSVeDrDqPNH8KTdUfh7451/JpZbMNX11cl\nViJYzTn/qyAineV0ysR3j33c1zi8OxgFAFROd+DWmrkwGyU4bCZYTDn9ayCiDKFL1+yf/OQnWL9+\nPTZt2oR77rkHXV1dia89/vjjqKmpwbp163Do0KHE9YMHD2LdunWoqanB9u3bR/U+YVnBR94Annju\naCJsq8sLcfu6eX1bdc0MWyJKG10C99Of/jT27t2L5557DrNmzcLjjz8OADh16hTq6upQV1eHHTt2\n4Ic//CEURYGiKHj44YexY8cO1NXVYe/evTh16tRF3+fEGR+erDuGUCTWXnHpPA++uLoSRoMIp90M\ns4kn6xJR+ugSuCtWrIDBEBtZLlmyBM3NzQCA/fv3o7a2FiaTCTNmzEBZWRnq6+tRX1+PsrIyzJgx\nAyaTCbW1tdi/f/9F32f7s0cRldXYey4sxU3XVUCSBLhsZpiNDFsiSi/dD+LatWsXrrvuOgCA1+tF\nSUlJ4mvFxcXwer3DXr8YRY31/Fpz1XRs+NRMSKKAQrsZJoYtEelgwiYw77jjDrS1tQ25fv/992PN\nmjUAgEcffRSSJOGGG24AENtyO5ggCFBVNeX1i5leZMNnlk7HisXTIAoCpjgtMBoyN2w9HrveJVwW\n1q8v1q+v0dQ/YYG7c+fOEb/+zDPP4NVXX8XOnTsT4VlSUpKYXgBiI96ioiIAGPb6SL79xSXoDcnw\n+3tRaDfD7+u9hJ8kPTweO1pbA3qXcclYv75Yv76S6x8peHWZUjh48CCeeOIJPProo7BarYnrq1at\nQl1dHSKRCM6ePYumpiYsWrQICxcuRFNTE86ePYtIJIK6ujqsWrVqVO8l9k0jGCTdZ0+IaJLTZU3U\ntm3bEIlEsHXrVgDA4sWL8fDDD6OyshIbNmzAxo0bIUkSHnroIUhSbArgoYcewp133glFUbBlyxZU\nVlZe9H0MDFsiyiCClmriNEdk00eUXPpIlY1Yv75yqf6Mm1IgIpqMGLhERGnCwCUiShMGLhFRmjBw\niYjShIFLRJQmDFwiojRh4BIRpQkDl4goTRi4RERpwsAlIkoTBi4RUZowcImI0iSnu4UREWUSjnCJ\niNKEgUtElCYMXCKiNGHgEhGlCQOXiChNGLhERGnCwM0gP/nJT7B+/Xps2rQJ99xzD7q6uvQuaUxe\neOEF1NbWoqqqCu+//77e5YzawYMHsW7dOtTU1GD79u16lzMm3/3ud7Fs2TJcf/31epdySS5cuIDb\nb78dGzZsQG1tLZ5++mm9SxqTcDiMm2++GTfccANqa2vxi1/8YuQXaJQxDh06pEWjUU3TNO2nP/2p\n9tOf/lTnisbm1KlT2unTp7XbbrtNq6+v17ucUZFlWVu9erV25swZLRwOa5s2bdI++OADvcsatcOH\nD2sNDQ1abW2t3qVcEq/XqzU0NGiapmmBQEBbu3ZtVv3+VVXVuru7NU3TtEgkot18883aO++8M+zz\nOcLNICtWrIDBYAAALFmyBM3NzTpXNDazZ89GRUWF3mWMSX19PcrKyjBjxgyYTCbU1tZi//79epc1\naldffTUcDofeZVyyoqIiLFiwAABgs9lQUVEBr9erc1WjJwgC8vPzAQCyLEOWZQiCMOzzGbgZateu\nXbjuuuv0LiPneb1elJSUJB4XFxdn1R98Ljl37hyOHTuGxYsX613KmCiKghtvvBHLly/H8uXLR6zf\nkMa6CMAdd9yBtra2Idfvv/9+rFmzBgDw6KOPQpIk3HDDDeku76JGU3820VLsbB9phEITo6enB/fd\ndx++973vwWaz6V3OmEiShD179qCrqwv33HMPTp48iblz56Z8LgM3zXbu3Dni15955hm8+uqr2Llz\nZ0b+4V+s/mxTUlIyYOrG6/WiqKhIx4omn2g0ivvuuw+bNm3C2rVr9S7nkhUUFOCTn/wkDh06NGzg\nckohgxw8eBBPPPEEHn30UVitVr3LmRQWLlyIpqYmnD17FpFIBHV1dVi1apXeZU0amqbhwQcfREVF\nBbZu3ap3OWPW0dGRWE0UCoXwxhtvjHgfg93CMkhNTQ0ikQicTicAYPHixXj44Yd1rmr0Xn75ZWzb\ntg0dHR0oKCjA/Pnz8eSTT+pd1kUdOHAAjzzyCBRFwZYtW/D3f//3epc0ag888AAOHz4Mn88Ht9uN\ne++9F5///Of1LmvU/va3v+HWW2/F3LlzIYqx8d8DDzyAlStX6lzZ6Bw/fhzf+c53oCgKNE3D+vXr\n8c1vfnPY5zNwiYjShFMKRERpwsAlIkoTBi4RUZowcImI0oSBS0SUJgxcoj5+vx8rV65EfX194tqj\njz6Ke++9V8eqKJdwWRhRkldeeQU/+9nPsHv3bjQ2NuLOO+/E7t274Xa79S6NcgADl2iQb3/72ygs\nLMSbb76Jb3zjG9i4caPeJVGOYOASDdLZ2YnPfvazWLZsGX75y1/qXQ7lEM7hEg3y17/+FTabDY2N\njYhEInqXQzmEgUuUpKOjAz/60Y+wfft2VFdXX/zIFKIxYOASJfnhD3+IL3zhC6iqqsKDDz6IvXv3\nDli1QHQ5GLhEfZ5//nk0NTXh7rvvBgA4HA489NBDePDBBzm1QOOCN82IiNKEI1wiojRh4BIRpQkD\nl4goTRi4RERpwsAlIkoTBi4RUZowcImI0oSBS0SUJv8fyTcqtHT4mWwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe211a532e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.datasets import make_regression\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sns.set()\n",
    "\n",
    "x_train, y_train, W_target = make_regression(n_samples=100, n_features=1, noise=10, coef = True)\n",
    "\n",
    "df = pd.DataFrame(data = {'X':x_train.ravel(), 'Y':y_train.ravel()})\n",
    "\n",
    "sns.lmplot(x='X', y='Y', data=df, fit_reg=True)\n",
    "plt.show()\n",
    "\n",
    "x_torch = torch.FloatTensor(x_train)\n",
    "y_torch = torch.FloatTensor(y_train)\n",
    "y_torch = y_torch.view(y_torch.size()[0], 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "PyTorch has a lot of useful modules in its **nn** library. One of them is linear. As the name suggests, it performs a linear transformation of its input, which is essentially linear regression does. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "class LinearRegression(torch.nn.Module):\n",
    "    def __init__(self, input_size, output_size):\n",
    "        super(LinearRegression, self).__init__()\n",
    "        self.linear = torch.nn.Linear(input_size, output_size)  \n",
    "        \n",
    "    def forward(self, x):\n",
    "        return self.linear(x)\n",
    "\n",
    "model = LinearRegression(1, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "To train a linear regression, we will need to add the right loss function from the same **nn** library. For linear regression we will use **MSELoss()**, mean squared error loss function.\n",
    "\n",
    "We will also need to use optimization function (SGD), and run a backpropagation similar to our previous toy example. Essentially, we repeating the steps from the **train()** function we defined above. The reason we can't use this function directly is that we implemented it for classification, not for regression, and as model prediction we use the index of the maximum element of cross-entropy loss. For linear regression we use output of linear layer as a prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "criterion = torch.nn.MSELoss()\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=0.1)  \n",
    "\n",
    "\n",
    "for epoch in range(50):\n",
    "    data, target = Variable(x_torch), Variable(y_torch)\n",
    "    output = model(data)\n",
    "        \n",
    "    optimizer.zero_grad()\n",
    "    loss = criterion(output, target)\n",
    "    loss.backward()\n",
    "    optimizer.step()\n",
    "        \n",
    "predicted = model(Variable(x_torch)).data.numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "We can now print the original data and linear regression that we fit with PyTorch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX8AAAD4CAYAAAAEhuazAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGPyJiFwQgz8RkQti8CcickH/H1hVgA/1KvfIAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20b5b5ac8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(x_train, y_train, 'o', label='Original data')\n",
    "plt.plot(x_train, predicted, label='Fitted line')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "To move forward with more complex models, let's download MNIST dataset to your 'datasets' folder and test some initial pre-processing that's available in PyTorch. PyTorch has dataloaders and handlers for various datasets. Once downloaded, you can use them any time. You can also wrap your data in PyTorch tensors and create your own data loader class.\n",
    "\n",
    "Batch size is a term used in machine learning and refers to the number of training examples utilised in one iteration. The batch size can be one of three options:\n",
    "-  batch mode: where the batch size is equal to the total dataset thus making the iteration and epoch values equivalent\n",
    "-  mini-batch mode: where the batch size is greater than one but less than the total dataset size. Usually, a number that can be divided into the total dataset size.\n",
    "-  stochastic mode: where the batch size is equal to one. Therefore the gradient and the neural network parameters are updated after each sample.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "from torchvision import datasets, transforms\n",
    "\n",
    "batch_num_size = 64\n",
    "\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST('data',train=True, download=True, transform=transforms.Compose([\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize((0.1307,), (0.3081,))\n",
    "    ])), \n",
    "    batch_size=batch_num_size, shuffle=True)\n",
    "\n",
    "test_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST('data',train=False, transform=transforms.Compose([\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize((0.1307,), (0.3081,))\n",
    "    ])), \n",
    "    batch_size=batch_num_size, shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 2.1 Logistic regression with MNIST"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Let's visualize some of the downloaded MNIST data to better understand what kind of data we are trying to classify."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(64, 1, 28, 28)\n"
     ]
    }
   ],
   "source": [
    "sns.reset_orig()\n",
    "\n",
    "dataiter = iter(train_loader)\n",
    "(data,target) = dataiter.next()\n",
    "\n",
    "data = data.numpy()\n",
    "print (data.shape)\n",
    "#print (data[:, 0].shape)  \n",
    "fig, axes = plt.subplots(ncols=10, figsize=(20, 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABkAAAACyCAYAAAAahgOPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20b5df5f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for col in range(10):\n",
    "    axes[col].imshow(data[col, 0, :, :], cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Now we are ready to write a logistic regression model. It's really easy in pytorch. In fact, it's exactly like linear regression, we just need to change the loss function to cross entropy loss during training and inference. We also added input reshaping for the case of multy-dimensional input tensor."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "class LogisticRegression(nn.Module):\n",
    "    def __init__(self, input_size, num_classes):\n",
    "        super(LogisticRegression, self).__init__()\n",
    "        self.linear = nn.Linear(input_size, num_classes)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x = x.view(x.size()[0], -1)\n",
    "        return self.linear(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Since logistic regression is a classification model, we can use **train()** and **test()** function we defined earlier. Hence training and testing the model is as easy as one line of code.\n",
    "\n",
    "We can add some regularization during training. To add a common **L2** regularizer, just add **weight_decay** parameter to **SGD()** function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 0 [25600/60000 (43%)]\tLoss: 0.248953\n",
      "Train Epoch: 0 [51200/60000 (85%)]\tLoss: 0.201164\n",
      "\n",
      "Test set: Average loss: 0.3464, Accuracy: 8986/10000 (90%)\n",
      "\n",
      "Train Epoch: 1 [25600/60000 (43%)]\tLoss: 0.253158\n",
      "Train Epoch: 1 [51200/60000 (85%)]\tLoss: 0.292870\n",
      "\n",
      "Test set: Average loss: 0.3298, Accuracy: 9059/10000 (91%)\n",
      "\n",
      "Train Epoch: 2 [25600/60000 (43%)]\tLoss: 0.321621\n",
      "Train Epoch: 2 [51200/60000 (85%)]\tLoss: 0.324983\n",
      "\n",
      "Test set: Average loss: 0.3236, Accuracy: 9049/10000 (90%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model = LogisticRegression(28*28, 10)\n",
    "criterion = nn.CrossEntropyLoss()  \n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=0.1, weight_decay=0.01)  \n",
    "\n",
    "# Training the Model\n",
    "for epoch in range(3):\n",
    "    train(model, epoch, criterion, optimizer, train_loader)\n",
    "    acc, loss = test(model, epoch, criterion, test_loader)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "After you trained your logistic regression function, you can interface weights of your model with NumPy and print the features your model learned."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "fig, axes = plt.subplots(ncols=10, figsize=(20, 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABkAAAACyCAYAAAAahgOPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20a61a978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = model.linear.weight.data.cpu().view(10, 1, 28, 28).numpy()\n",
    "for col in range(10):\n",
    "    axes[col].imshow(data[col, 0, :, :], cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "If you want to save the model to avoid costly retraining process, you can use **torch.save()** function. You can find more information on how to store models in the corresponding page of the [pytorch documentation](http://pytorch.org/docs/master/torch.html#serialization).  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(), 'pretrained/logreg.t7')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "##  3. LeNet Convolutional Neural Network (CNN) in PyTorch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Let's create our first simple neural network from scratch. This network is going to perform image classification trying to recognize hand-written numbers from MNIST dataset. It's a four layer convolutional neural network (CNN), a very common architecture for analyzing MNIST datasets. This code is adopted from one of the official PyTorch tutorials, but you can find many more examples of what you can do with PyTorch as a deep net researcher [here](http://pytorch.org/tutorials/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "We will use several modeules from **torch.nn** library. \n",
    "1. **Linear** layer -- applies linear transformation of the input tensor using the layer's weights.\n",
    "2. **Conv1** and **Conv2** -- convolution layers, each will output a dot product between convolution kernel (weights tensor of small size) and a small input region of the same size.\n",
    "3. **Relu** -- rectified linear unit function, applies an elementwise activation function **max(0,x)**. \n",
    "4. **Pooling** layer -- performs downsampling of certain area (typically 2x2 pixels) using the **max** operation.\n",
    "5. **Dropout2D** -- randomly zeroes whole channels of the input tensor. In the case where feature maps are strongly correlated, dropout2D promotes the independence between feature maps.\n",
    "6. **Softmax** -- applies the **Log(Softmax(x))** function to an n-dimensional input Tensor to get the output between 0 and 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "class LeNet(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(LeNet,self).__init__()\n",
    "        self.conv1 = nn.Conv2d(1,10,kernel_size=5)\n",
    "        self.conv2 = nn.Conv2d(10,20,kernel_size=5)\n",
    "        self.conv2_drop = nn.Dropout2d()\n",
    "        self.fc1 = nn.Linear(320,50)\n",
    "        self.fc2 = nn.Linear(50,10)\n",
    "        \n",
    "    def forward(self,x):\n",
    "        x = F.relu(F.max_pool2d(self.conv1(x),2))\n",
    "        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)),2))\n",
    "        x = x.view(-1, 320)\n",
    "        x = F.relu(self.fc1(x))\n",
    "        x = F.dropout(x, training=self.training)\n",
    "        x = self.fc2(x)\n",
    "        return F.log_softmax(x, dim=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "After creating a LeNet class, let's create an object and move it to gpu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MNIST_net model:\n",
      "\n",
      "LeNet(\n",
      "  (conv1): Conv2d(1, 10, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2): Conv2d(10, 20, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2_drop): Dropout2d(p=0.5)\n",
      "  (fc1): Linear(in_features=320, out_features=50, bias=True)\n",
      "  (fc2): Linear(in_features=50, out_features=10, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "model = LeNet()\n",
    "if cuda_gpu:\n",
    "    model.cuda()\n",
    "    \n",
    "print ('MNIST_net model:\\n')\n",
    "print (model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "To train this model we're going to use SGD with learning rate=0.01 and momentum=0.5 \n",
    "Check this link for SGD and [mometum](http://ruder.io/optimizing-gradient-descent/index.html#momentum) explanation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss()  \n",
    "optimizer = optim.SGD(model.parameters(),lr = 0.005, momentum = 0.9)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "We can train a pretty accurate LeNet model just after 5 epochs (one epoch means you use the whole training dataset for updating the training model weights one time). This code checks to see if you have a pre-trained model already stored in a file. If so, it just loads it. Otherwise, it trains one and saves it to disk."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading model\n",
      "\n",
      "Test set: Average loss: 0.0471, Accuracy: 9859/10000 (99%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "\n",
    "epochs = 5\n",
    "if (os.path.isfile('pretrained/MNIST_net.t7')):\n",
    "    print ('Loading model')\n",
    "    model.load_state_dict(torch.load('pretrained/MNIST_net.t7', map_location=lambda storage, loc: storage))\n",
    "    acc, loss = test(model, 1, criterion, test_loader)\n",
    "else:\n",
    "    print ('Training model')\n",
    "    for epoch in range(1, epochs + 1):\n",
    "        train(model, epoch, criterion, optimizer, train_loader)\n",
    "        acc, loss = test(model, 1, criterion, test_loader)\n",
    "    torch.save(model.state_dict(), 'pretrained/MNIST_net.t7')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Now let's look at our model. First let's print out the information about it. The print functions show all the layers (E.g., Dropout is implemented as a separate layer), their name and parameters. It also prints the names for all the layers which will be used in the next steps."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MNIST_net model:\n",
      "\n",
      "LeNet(\n",
      "  (conv1): Conv2d(1, 10, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2): Conv2d(10, 20, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2_drop): Dropout2d(p=0.5)\n",
      "  (fc1): Linear(in_features=320, out_features=50, bias=True)\n",
      "  (fc2): Linear(in_features=50, out_features=10, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print ('MNIST_net model:\\n')\n",
    "print (model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "\n",
    "\n",
    "There is also an iterator that runs over all the named modules in the model. This is helpful when you have a complex DNN that consists of multiple \"internal\" models. Iterating over all named modules allows us to create model parsers that can read model parameters and create modules similar to ones that are used in the network.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Internal models:\n",
      "0 -> ('', LeNet(\n",
      "  (conv1): Conv2d(1, 10, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2): Conv2d(10, 20, kernel_size=(5, 5), stride=(1, 1))\n",
      "  (conv2_drop): Dropout2d(p=0.5)\n",
      "  (fc1): Linear(in_features=320, out_features=50, bias=True)\n",
      "  (fc2): Linear(in_features=50, out_features=10, bias=True)\n",
      "))\n",
      "-------------------------------------------------------------------------\n",
      "1 -> ('conv1', Conv2d(1, 10, kernel_size=(5, 5), stride=(1, 1)))\n",
      "-------------------------------------------------------------------------\n",
      "2 -> ('conv2', Conv2d(10, 20, kernel_size=(5, 5), stride=(1, 1)))\n",
      "-------------------------------------------------------------------------\n",
      "3 -> ('conv2_drop', Dropout2d(p=0.5))\n",
      "-------------------------------------------------------------------------\n",
      "4 -> ('fc1', Linear(in_features=320, out_features=50, bias=True))\n",
      "-------------------------------------------------------------------------\n",
      "5 -> ('fc2', Linear(in_features=50, out_features=10, bias=True))\n",
      "-------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print ('Internal models:')\n",
    "for idx, m in enumerate(model.named_modules()):\n",
    "    print(idx, '->', m)\n",
    "    print ('-------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Similar to named modules, we can iterate over a model's internal parameters. This code prints out information about all the data containers that a specific model uses. It's helpful to know how to access these parameters for updating weight tensors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "internal parameters:\n",
      "conv1.weight:  <class 'torch.nn.parameter.Parameter'> torch.Size([10, 1, 5, 5])\n",
      "---------------------------------------------------------------------------------------\n",
      "conv1.bias:  <class 'torch.nn.parameter.Parameter'> torch.Size([10])\n",
      "---------------------------------------------------------------------------------------\n",
      "conv2.weight:  <class 'torch.nn.parameter.Parameter'> torch.Size([20, 10, 5, 5])\n",
      "---------------------------------------------------------------------------------------\n",
      "conv2.bias:  <class 'torch.nn.parameter.Parameter'> torch.Size([20])\n",
      "---------------------------------------------------------------------------------------\n",
      "fc1.weight:  <class 'torch.nn.parameter.Parameter'> torch.Size([50, 320])\n",
      "---------------------------------------------------------------------------------------\n",
      "fc1.bias:  <class 'torch.nn.parameter.Parameter'> torch.Size([50])\n",
      "---------------------------------------------------------------------------------------\n",
      "fc2.weight:  <class 'torch.nn.parameter.Parameter'> torch.Size([10, 50])\n",
      "---------------------------------------------------------------------------------------\n",
      "fc2.bias:  <class 'torch.nn.parameter.Parameter'> torch.Size([10])\n",
      "---------------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "print('internal parameters:')\n",
    "for name,param in model.named_parameters():\n",
    "    print(name + \": \", type(param), param.size())\n",
    "    print('---------------------------------------------------------------------------------------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Let's try to access a weight tensor and print it out. We can see the tensor type and size along with data. Also, we can just print the tensor size. Weight tensors in DNNs are usually very large, so checking the size (as opposed to the contents) is very useful when debugging."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "scrolled": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'torch.nn.parameter.Parameter'>\n",
      "torch.Size([10, 1, 5, 5])\n",
      "Parameter containing:\n",
      "(0 ,0 ,.,.) = \n",
      " -0.3326  0.0619  0.1446  0.0613  0.0869\n",
      " -0.4763 -0.0131  0.1291  0.3374  0.2129\n",
      " -0.1188  0.1643  0.2340  0.1542 -0.0067\n",
      "  0.0897  0.2317 -0.0397 -0.1650 -0.1677\n",
      "  0.0560 -0.1716 -0.1474  0.0969 -0.1119\n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      "  0.1474  0.3115  0.4264  0.2165 -0.0203\n",
      " -0.1965 -0.1293 -0.2629 -0.0161 -0.0762\n",
      " -0.2161 -0.4547 -0.4099 -0.2752 -0.0581\n",
      " -0.0212 -0.0789 -0.0444  0.0544  0.0801\n",
      "  0.2962  0.2762  0.3134  0.1210  0.0436\n",
      "\n",
      "(2 ,0 ,.,.) = \n",
      " -0.0276  0.1624 -0.0489  0.3204  0.1842\n",
      " -0.1948  0.0329  0.0998  0.0817  0.4210\n",
      "  0.1177 -0.0811 -0.0034 -0.0733  0.1315\n",
      " -0.0694 -0.1168 -0.1301 -0.0334  0.2707\n",
      " -0.3927 -0.1096 -0.1386 -0.2398  0.0943\n",
      "\n",
      "(3 ,0 ,.,.) = \n",
      "  0.1664  0.1382 -0.1063 -0.1228 -0.3840\n",
      " -0.0894 -0.1955 -0.2881 -0.3713 -0.0840\n",
      "  0.0021 -0.2913 -0.2737 -0.0415  0.2981\n",
      " -0.3226 -0.2793  0.0191  0.1750  0.0479\n",
      " -0.3260  0.0261 -0.0394  0.1033  0.1065\n",
      "\n",
      "(4 ,0 ,.,.) = \n",
      "  0.0569  0.3361  0.3601 -0.1554 -0.2146\n",
      " -0.1108  0.3457  0.2381 -0.2242 -0.3739\n",
      "  0.1349  0.2011  0.1508 -0.1235 -0.1359\n",
      "  0.0282  0.2195 -0.0515 -0.0662 -0.2621\n",
      "  0.0739 -0.0391 -0.1506 -0.2479 -0.2894\n",
      "\n",
      "(5 ,0 ,.,.) = \n",
      " -0.2313 -0.0644 -0.2631 -0.1022 -0.0290\n",
      " -0.2088 -0.1734  0.0198 -0.0561  0.1800\n",
      " -0.2025 -0.0044 -0.1250 -0.1381 -0.1612\n",
      " -0.1609 -0.0325 -0.2306 -0.2220 -0.1447\n",
      "  0.1475  0.1293  0.0892  0.0208 -0.2559\n",
      "\n",
      "(6 ,0 ,.,.) = \n",
      "  0.3296  0.4071  0.2384  0.1520  0.2367\n",
      "  0.0133  0.2080  0.2095  0.3014  0.1548\n",
      " -0.4288 -0.1705  0.0070 -0.0356  0.0398\n",
      " -0.3152 -0.3320 -0.3219 -0.2408 -0.0034\n",
      " -0.1924 -0.1633 -0.2200 -0.2028 -0.1924\n",
      "\n",
      "(7 ,0 ,.,.) = \n",
      " -0.2054 -0.2413 -0.2034  0.0382  0.2969\n",
      " -0.2966 -0.0589 -0.1505  0.2017  0.1148\n",
      " -0.1718 -0.3054 -0.0744  0.2358  0.2504\n",
      " -0.3152 -0.2195 -0.1975  0.0525  0.1366\n",
      " -0.0837 -0.0439 -0.0255  0.0283  0.3430\n",
      "\n",
      "(8 ,0 ,.,.) = \n",
      " -0.3190 -0.2982 -0.2073 -0.1974 -0.1053\n",
      " -0.3045 -0.1606 -0.1677 -0.2339 -0.3507\n",
      " -0.0489 -0.1328 -0.0687 -0.1963 -0.1884\n",
      "  0.1958  0.0153  0.0875  0.2082  0.2719\n",
      "  0.2808 -0.0002 -0.1506  0.2590  0.2161\n",
      "\n",
      "(9 ,0 ,.,.) = \n",
      "  0.1021 -0.0717 -0.0703  0.0410 -0.0408\n",
      " -0.2936 -0.1835 -0.0009 -0.1231  0.0390\n",
      " -0.2049 -0.1676 -0.2479  0.0554  0.0900\n",
      "  0.1618 -0.0602  0.2054  0.1968 -0.0757\n",
      "  0.3526  0.3580  0.0918 -0.1536 -0.0027\n",
      "[torch.cuda.FloatTensor of size 10x1x5x5 (GPU 0)]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "t = model.conv1.weight\n",
    "print (type(t))\n",
    "print (t.size())\n",
    "print (t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Here the object printed is a Variable and not a tensor. It is a wrapper around the tensor used to build the computation graph for the forward/backward path. To access the actual tensor, we need to use .data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "scrolled": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'torch.cuda.FloatTensor'>\n",
      "torch.Size([20, 10, 5, 5])\n",
      "\n",
      "(0 ,0 ,.,.) = \n",
      " -8.2037e-05  1.3582e-02  2.0244e-02 -1.9943e-02 -7.7893e-02\n",
      " -1.0396e-01 -8.3978e-02 -1.2411e-01 -1.2931e-01 -4.0826e-02\n",
      " -1.2326e-01 -9.8568e-02 -5.3306e-02  1.0149e-01  1.2639e-01\n",
      " -8.6661e-02 -6.2155e-02 -6.5777e-02 -5.7857e-02 -6.2671e-02\n",
      " -6.5787e-02 -1.8636e-01 -1.6291e-01 -1.7345e-01 -1.2998e-01\n",
      "\n",
      "(0 ,1 ,.,.) = \n",
      " -5.8389e-02 -2.2088e-02 -6.8403e-02  2.6269e-02  5.0578e-02\n",
      "  2.9074e-02  7.5517e-02  1.2157e-01  1.3417e-01  7.3019e-02\n",
      "  3.9684e-02 -7.8071e-02 -3.4606e-02 -2.5784e-02  1.5061e-02\n",
      "  4.4148e-02 -2.4495e-02 -3.2028e-02  7.8683e-02  6.5671e-02\n",
      " -1.0896e-04  5.9706e-02  1.8927e-02  3.1335e-02 -3.6424e-02\n",
      "\n",
      "(0 ,2 ,.,.) = \n",
      " -4.5042e-02 -1.5815e-02 -1.0917e-02 -4.0813e-02  2.7118e-03\n",
      "  3.1682e-02  4.8692e-02 -4.9076e-02 -7.4701e-02 -6.0333e-02\n",
      " -4.1446e-02 -1.2763e-01 -8.7152e-02  1.8507e-02  6.4115e-02\n",
      " -5.9946e-02 -3.4380e-02  7.6220e-02  9.3114e-03  4.0778e-02\n",
      "  6.4881e-02 -1.5272e-02  8.1760e-02  7.2702e-04  1.8256e-02\n",
      "   ...\n",
      "\n",
      "(0 ,7 ,.,.) = \n",
      "  1.5341e-02 -6.5551e-02 -2.2489e-03 -3.8212e-03 -2.9176e-02\n",
      " -8.5326e-02 -4.5779e-02 -2.7835e-02 -9.0050e-02 -5.2610e-02\n",
      " -7.2369e-02 -1.1728e-02  2.9702e-02  1.8628e-02 -4.2310e-02\n",
      " -7.9920e-02  4.1981e-02  1.8630e-02  1.7634e-03 -8.1894e-02\n",
      " -4.3635e-02 -1.2626e-02  3.4980e-02 -1.1856e-01 -9.3455e-02\n",
      "\n",
      "(0 ,8 ,.,.) = \n",
      " -1.1751e-02  3.2204e-02 -3.9134e-02 -5.5944e-02 -5.0121e-02\n",
      " -2.4856e-02  5.5534e-02  7.4951e-02  8.4154e-02  9.3987e-02\n",
      " -3.6105e-03  5.1196e-03 -1.6284e-02  3.7331e-02  1.7574e-02\n",
      " -2.4077e-02 -5.4070e-02  1.4837e-02 -7.0221e-02 -3.8622e-02\n",
      "  7.7082e-02 -4.4997e-02  7.3446e-02 -4.2818e-02 -3.9459e-02\n",
      "\n",
      "(0 ,9 ,.,.) = \n",
      " -7.4102e-02 -9.2458e-03 -4.2408e-02 -4.6486e-02 -8.7267e-02\n",
      " -4.6227e-02 -1.5664e-02  2.2344e-02  1.6561e-02  9.4998e-02\n",
      " -3.6270e-02 -3.9993e-02  3.3930e-04 -2.1563e-02 -1.9965e-02\n",
      " -4.6755e-02  7.3500e-03  2.3813e-02 -9.7210e-02 -2.5895e-02\n",
      "  6.7023e-02  8.1926e-02  1.1507e-02  3.4635e-02  2.3324e-02\n",
      "     ⋮ \n",
      "\n",
      "(1 ,0 ,.,.) = \n",
      "  6.0016e-02  5.0006e-02  1.7948e-01  4.7997e-02 -4.2356e-02\n",
      " -5.1257e-02  9.4135e-02  9.5144e-02 -1.4788e-01 -8.6359e-02\n",
      " -1.2853e-01 -6.1892e-02 -2.7254e-01 -8.2618e-02  9.5560e-02\n",
      " -6.8764e-02 -8.8780e-02 -9.2987e-02  1.1164e-01  3.1770e-02\n",
      " -5.3992e-02 -3.8700e-02 -5.9118e-02 -9.1183e-02 -1.0023e-01\n",
      "\n",
      "(1 ,1 ,.,.) = \n",
      " -4.6365e-02 -1.2775e-02 -3.2748e-03 -1.3364e-01 -1.1349e-01\n",
      " -1.3393e-02 -2.7892e-02 -5.2732e-02  5.0289e-02  1.0809e-01\n",
      "  6.9444e-03  9.7858e-02  1.1983e-01  1.1519e-01  4.1913e-02\n",
      " -2.4429e-02 -1.6398e-02 -2.9657e-02  4.0157e-03 -1.1136e-02\n",
      " -4.6395e-02 -5.6350e-02 -8.5068e-02  1.0065e-02 -6.6433e-04\n",
      "\n",
      "(1 ,2 ,.,.) = \n",
      " -5.6754e-03 -5.7463e-02  5.5209e-02 -9.7076e-02  1.0835e-02\n",
      " -7.4181e-02 -8.4499e-02 -1.2707e-02 -1.3180e-01 -1.3710e-01\n",
      " -1.7485e-02 -8.2239e-02 -1.5625e-01 -1.9209e-01 -2.9264e-02\n",
      " -3.2297e-02 -1.2461e-01 -1.2771e-01 -6.7789e-02  2.0350e-02\n",
      " -4.2816e-02 -1.0220e-01 -1.9046e-02 -9.0565e-02  2.3398e-02\n",
      "   ...\n",
      "\n",
      "(1 ,7 ,.,.) = \n",
      " -3.5395e-03  1.0102e-01  3.8646e-02  2.2807e-02  9.0554e-03\n",
      " -5.3194e-02 -2.3986e-02 -3.3721e-02 -3.0369e-02  4.9882e-02\n",
      " -6.5125e-02 -9.9803e-02 -1.8218e-02  2.7164e-02  1.3753e-02\n",
      " -7.2606e-02 -5.2061e-02 -3.0226e-02 -5.7344e-02  2.5905e-03\n",
      "  1.1660e-01  2.9398e-03  6.4503e-03 -9.0392e-02 -5.3284e-02\n",
      "\n",
      "(1 ,8 ,.,.) = \n",
      "  5.3767e-02  2.2759e-02  1.4102e-02  8.6102e-03 -1.3645e-01\n",
      " -1.8307e-02 -8.0410e-02  4.8464e-02  1.1102e-02  8.0966e-03\n",
      "  3.4969e-02  2.0080e-02  1.4002e-01  1.2117e-01  1.4945e-01\n",
      "  3.8591e-02 -2.6471e-02  1.0782e-02  1.5258e-02  6.5592e-03\n",
      "  5.8438e-03 -2.9101e-02 -2.2256e-02 -1.1759e-01 -4.2572e-02\n",
      "\n",
      "(1 ,9 ,.,.) = \n",
      " -4.2657e-02  8.4783e-02  4.3867e-02  1.7813e-03 -2.2578e-02\n",
      " -1.8723e-02 -3.1019e-02 -4.1105e-02 -3.2504e-02  3.7309e-02\n",
      " -9.6160e-02  4.2823e-02  1.7815e-02  4.7939e-02  1.3663e-02\n",
      " -4.8860e-02 -5.2605e-02  6.5742e-02 -5.6223e-02  1.2583e-02\n",
      " -2.1883e-02  2.4538e-03  5.1331e-02 -3.7048e-03 -2.8235e-02\n",
      "     ⋮ \n",
      "\n",
      "(2 ,0 ,.,.) = \n",
      " -1.1343e-02 -1.6515e-01  1.3750e-02 -7.3038e-04  4.1279e-02\n",
      " -9.3457e-02 -2.0358e-01 -7.5808e-02  1.0439e-01  1.8264e-01\n",
      " -5.8241e-02 -1.6343e-01  1.2275e-01  8.3715e-02 -3.4118e-02\n",
      " -1.3959e-01  5.0841e-02  4.4182e-02 -3.8282e-02 -1.5186e-01\n",
      " -6.4689e-02  4.3519e-02 -3.8929e-04 -3.0478e-02 -1.0092e-01\n",
      "\n",
      "(2 ,1 ,.,.) = \n",
      " -1.1887e-01 -1.0894e-01 -4.4239e-02  8.1807e-02 -6.4569e-02\n",
      " -1.2791e-01 -4.3896e-02 -4.7125e-02 -4.3031e-02 -1.2612e-01\n",
      " -5.4641e-02 -7.7896e-02 -3.2407e-02 -6.8002e-02 -1.4188e-01\n",
      " -2.8161e-02  8.1560e-03 -1.3922e-01 -5.9659e-02  8.4232e-02\n",
      "  8.8565e-02 -3.9376e-02 -7.8187e-02  3.9106e-02 -2.8383e-02\n",
      "\n",
      "(2 ,2 ,.,.) = \n",
      " -1.0880e-01  4.3608e-03  3.9058e-02  7.1572e-02  2.6427e-02\n",
      " -1.0294e-01 -4.1599e-02 -4.4224e-02 -4.5941e-02 -1.4797e-02\n",
      " -9.2298e-02 -5.0403e-02 -4.0677e-02 -5.8107e-02 -9.4755e-02\n",
      " -7.0369e-02 -5.4460e-02  3.3144e-02 -2.4758e-02 -4.3179e-02\n",
      " -2.1640e-01 -9.1549e-02 -2.2661e-03 -1.9908e-02  3.9288e-02\n",
      "   ...\n",
      "\n",
      "(2 ,7 ,.,.) = \n",
      "  6.9271e-03 -5.0982e-03 -1.5981e-02 -3.1967e-02  5.7127e-02\n",
      " -1.2256e-02  5.1788e-02  6.9832e-02  8.0015e-02  5.2999e-02\n",
      " -8.8297e-02  7.4222e-02  4.9427e-02  1.0904e-01 -4.6003e-02\n",
      " -9.5128e-02  7.4223e-02  5.3610e-02  1.7359e-01 -1.4072e-01\n",
      " -5.0380e-02 -2.2467e-02  2.4300e-02  1.0812e-01 -5.4894e-02\n",
      "\n",
      "(2 ,8 ,.,.) = \n",
      " -1.1985e-02 -1.7361e-02 -2.1129e-02  1.6763e-02 -3.6996e-02\n",
      "  3.5981e-02  7.3382e-02 -3.2001e-02  1.0201e-01 -6.1857e-02\n",
      "  7.5560e-03  1.0652e-02 -3.6130e-02 -6.1376e-02 -1.3672e-01\n",
      "  4.6432e-02 -1.2657e-01 -7.6716e-02 -6.9796e-02 -6.5647e-02\n",
      "  5.2731e-03 -7.7747e-02 -1.1689e-01 -2.7526e-02  4.6108e-02\n",
      "\n",
      "(2 ,9 ,.,.) = \n",
      " -7.8076e-02 -8.1092e-02 -1.0337e-02 -4.4480e-02  1.2494e-02\n",
      "  4.7213e-02 -3.2783e-02  5.2397e-02  3.6568e-02 -4.1201e-02\n",
      " -4.1849e-02 -6.7098e-02 -3.0186e-02  8.2873e-03 -1.0584e-01\n",
      "  4.9345e-02 -4.7663e-02 -1.2037e-04 -5.4613e-02 -1.3521e-02\n",
      "  3.6178e-02  7.0070e-02 -4.1157e-02 -7.4993e-03 -4.0017e-02\n",
      "...   \n",
      "     ⋮ \n",
      "\n",
      "(17,0 ,.,.) = \n",
      "  8.3150e-02 -6.0100e-02 -3.6816e-03  2.6451e-02  1.8018e-01\n",
      " -7.9064e-02 -2.5469e-01 -9.4042e-02 -6.8061e-03 -6.6482e-02\n",
      " -1.7161e-01 -2.1320e-01 -2.5275e-01 -2.0759e-01 -2.9297e-01\n",
      " -8.4972e-02 -7.5225e-02 -9.4782e-02 -9.4855e-02 -1.2089e-01\n",
      "  5.7906e-02  6.1543e-02  2.6089e-02  1.4314e-02  4.4270e-02\n",
      "\n",
      "(17,1 ,.,.) = \n",
      " -9.7270e-02 -8.3440e-02 -4.3678e-02  3.3496e-02 -1.8743e-02\n",
      "  8.8732e-02  7.5752e-02  1.7899e-02 -6.7135e-02  3.2662e-03\n",
      "  7.5458e-02 -2.3177e-02  4.4386e-03  5.4769e-02  1.1419e-01\n",
      "  7.8623e-02  1.3667e-02  2.1315e-02 -3.6077e-02  9.9734e-02\n",
      " -6.6446e-02 -7.0974e-02  4.1828e-03  1.5787e-03 -4.7972e-02\n",
      "\n",
      "(17,2 ,.,.) = \n",
      " -6.0340e-02  2.2019e-02  1.2102e-02  4.5135e-03  8.7463e-02\n",
      " -4.4256e-02 -5.1448e-02 -9.7400e-03 -3.4540e-02 -9.8068e-05\n",
      " -2.8136e-02 -1.6089e-02 -1.5177e-02 -7.3941e-02 -1.0003e-01\n",
      "  1.6406e-02 -4.3388e-02 -1.5189e-02 -3.4609e-02  1.2368e-02\n",
      "  1.2132e-03  6.9856e-03  1.5866e-02  6.4759e-02  1.0814e-01\n",
      "   ...\n",
      "\n",
      "(17,7 ,.,.) = \n",
      " -1.2338e-01 -4.4809e-02  2.9071e-02  2.4744e-02  3.2710e-02\n",
      " -1.5079e-01 -1.3598e-02 -5.1865e-02  7.6386e-04 -6.4213e-02\n",
      " -5.9107e-02  1.8223e-02 -3.0056e-02 -1.5562e-03 -7.1045e-02\n",
      " -2.4272e-02  2.8340e-02 -2.8921e-02  3.0744e-02 -6.1466e-02\n",
      " -2.4578e-02  1.1963e-02  7.3569e-03  1.0648e-02 -7.5783e-02\n",
      "\n",
      "(17,8 ,.,.) = \n",
      " -8.4110e-02 -1.1641e-01 -2.0302e-02  1.4232e-02  9.8737e-03\n",
      " -3.5106e-03 -1.8508e-03  7.9643e-02 -3.9945e-02 -8.7312e-02\n",
      "  1.0906e-02  7.6149e-02  1.5230e-01  8.2442e-02  9.6480e-02\n",
      "  7.4120e-02  7.6459e-02  8.3624e-02  1.5280e-01  1.5167e-01\n",
      "  1.9025e-02  1.7967e-02  3.1701e-02  4.4394e-02 -7.9279e-03\n",
      "\n",
      "(17,9 ,.,.) = \n",
      " -6.6649e-02 -4.0174e-02  2.9158e-02  2.8468e-02 -5.2543e-02\n",
      " -4.6994e-02  2.1548e-02  8.4623e-02  3.2145e-03 -2.0821e-02\n",
      "  4.4747e-02 -1.7806e-02 -1.0138e-02  2.3851e-02  1.2106e-02\n",
      "  7.5885e-02  2.2751e-02  4.1703e-02  6.9400e-02  7.9342e-02\n",
      "  4.9225e-02 -7.6066e-03  3.5458e-02  1.3309e-02 -8.3993e-02\n",
      "     ⋮ \n",
      "\n",
      "(18,0 ,.,.) = \n",
      " -7.1817e-02 -1.1957e-01  7.4977e-02  1.6294e-02 -6.0131e-02\n",
      "  1.2204e-01 -3.0933e-02 -1.9166e-01 -9.4832e-02 -2.7473e-03\n",
      "  5.4187e-02  1.2425e-01 -7.8450e-02 -1.6364e-01 -4.1271e-02\n",
      "  1.2294e-01  3.2215e-02  2.7863e-02  4.7096e-02  1.5970e-01\n",
      "  6.4909e-02  1.0095e-01  9.9266e-02  1.4109e-01  4.7272e-03\n",
      "\n",
      "(18,1 ,.,.) = \n",
      " -1.6598e-02  8.0243e-02  5.2195e-02  3.6322e-02 -5.7024e-02\n",
      "  3.9262e-02 -4.0285e-02  5.0997e-02  1.3852e-01  5.5447e-03\n",
      " -3.3964e-02  7.7902e-02 -4.1124e-02  1.4838e-02  2.2140e-02\n",
      "  3.8708e-02  4.6558e-02  3.3423e-02 -4.7968e-02  2.7906e-02\n",
      " -3.9625e-02  2.0845e-02  6.0486e-03 -7.0882e-02 -1.5979e-01\n",
      "\n",
      "(18,2 ,.,.) = \n",
      " -6.6559e-02 -1.6328e-01 -7.4153e-03 -4.7527e-02 -3.7007e-02\n",
      " -6.4731e-02 -7.3354e-02 -5.9842e-02 -1.2620e-02 -2.8734e-02\n",
      " -1.1607e-02  8.6068e-02 -1.8978e-02 -3.6458e-02 -1.1055e-01\n",
      " -4.9359e-02  7.7014e-02  6.8650e-02 -5.4042e-02  2.7854e-02\n",
      "  5.9719e-02 -1.5676e-02  9.0832e-02 -1.7772e-03  4.2265e-02\n",
      "   ...\n",
      "\n",
      "(18,7 ,.,.) = \n",
      " -3.9311e-02 -6.1909e-02 -1.9868e-02  8.0381e-03 -2.2486e-02\n",
      " -9.3393e-02 -1.3517e-02 -1.4140e-02  6.6068e-02  1.1294e-01\n",
      " -4.0678e-02 -3.9167e-02  3.6690e-02  4.8973e-02  1.1758e-01\n",
      " -2.0564e-02  4.8462e-04  1.1999e-01 -1.8976e-02  7.1856e-02\n",
      " -7.4342e-02 -2.7098e-02  2.6179e-02 -1.4321e-02 -4.6212e-02\n",
      "\n",
      "(18,8 ,.,.) = \n",
      "  4.9845e-02  4.9602e-02  7.1783e-03  4.0600e-02  5.6701e-02\n",
      " -7.4566e-02 -1.9180e-02 -2.2553e-02  6.9294e-02  9.7918e-02\n",
      " -1.6064e-01 -1.6368e-01 -1.0125e-01  3.4873e-02  1.5692e-01\n",
      " -1.7756e-01 -2.1067e-01 -5.6298e-02  3.3489e-02  1.5823e-02\n",
      " -6.0866e-02  1.1981e-02  1.1453e-01  8.0696e-03 -1.4874e-02\n",
      "\n",
      "(18,9 ,.,.) = \n",
      "  2.1613e-02  4.4673e-02  8.3889e-02  2.5231e-02  1.4184e-02\n",
      "  6.8755e-02  7.2049e-02  1.3462e-01  1.5100e-01 -7.1492e-03\n",
      " -2.0170e-02 -9.1103e-02 -7.2628e-02 -6.2418e-02  8.6701e-02\n",
      "  2.2754e-02  9.2777e-02  4.4084e-02 -4.1038e-02 -2.6783e-03\n",
      "  6.1834e-02  1.0630e-01  1.5752e-02 -1.0129e-03 -5.3364e-02\n",
      "     ⋮ \n",
      "\n",
      "(19,0 ,.,.) = \n",
      " -1.2684e-02  3.4092e-02  1.8731e-02  4.2164e-02  1.4505e-02\n",
      "  9.9883e-02  6.7916e-02 -4.0652e-02 -5.6912e-02 -5.1977e-02\n",
      "  2.5387e-01 -6.6274e-03 -1.8041e-01 -1.4370e-01 -8.4390e-02\n",
      "  1.7071e-01  4.4959e-02 -4.5476e-02 -1.5569e-01 -1.2023e-01\n",
      "  1.1366e-01  1.8017e-01  6.2143e-02 -1.0338e-01 -1.1375e-01\n",
      "\n",
      "(19,1 ,.,.) = \n",
      " -2.9638e-02 -7.5691e-02 -9.4371e-02 -7.8026e-02  2.1120e-02\n",
      " -6.6078e-02 -1.0392e-01 -4.4055e-02 -3.2623e-02 -9.0596e-02\n",
      "  8.2151e-03 -8.0977e-02 -5.2392e-02 -2.1398e-03 -8.7618e-02\n",
      "  2.0529e-02 -1.1911e-02 -5.0446e-02 -1.0117e-01 -1.0656e-01\n",
      "  8.7415e-02  9.7045e-03  1.9666e-02 -3.4298e-02 -8.7867e-02\n",
      "\n",
      "(19,2 ,.,.) = \n",
      " -1.3448e-01 -9.7368e-02  1.0858e-02  1.7286e-02  3.8594e-03\n",
      " -5.0855e-02 -6.6390e-02 -5.9236e-04 -2.4002e-03 -5.4883e-02\n",
      "  1.0509e-02  4.0477e-03 -4.3656e-02  3.2645e-02 -1.7178e-02\n",
      " -2.4795e-02 -1.1388e-02  8.5563e-03 -4.5076e-02  4.4743e-02\n",
      "  1.1880e-01  1.6352e-02 -3.0909e-03 -2.0871e-02 -3.7492e-02\n",
      "   ...\n",
      "\n",
      "(19,7 ,.,.) = \n",
      "  5.8455e-02 -7.1192e-03 -3.4398e-02 -4.1017e-02  5.6500e-02\n",
      "  4.7691e-02 -4.1814e-03 -3.8130e-02 -6.0821e-02  7.6834e-02\n",
      "  2.0336e-02  7.4208e-02 -4.3493e-02 -7.1544e-02 -2.5517e-02\n",
      "  8.0602e-02  1.2416e-01 -1.0061e-02  6.6043e-02  6.8603e-02\n",
      "  2.9698e-02  3.9892e-02  4.9861e-02  7.1898e-02  2.5071e-02\n",
      "\n",
      "(19,8 ,.,.) = \n",
      " -6.0901e-02  2.3743e-02 -2.0248e-02 -5.6985e-02 -2.6070e-02\n",
      " -3.2225e-02  1.0882e-01  8.7429e-02 -4.7888e-02 -7.2084e-02\n",
      "  4.6041e-02  6.9497e-03  4.6355e-02 -1.3494e-02  7.1231e-03\n",
      " -2.5826e-02 -6.2310e-02 -1.7675e-02  2.9606e-02  2.5995e-02\n",
      " -8.5864e-02  1.0975e-02 -3.6584e-02 -5.4181e-02  6.7833e-02\n",
      "\n",
      "(19,9 ,.,.) = \n",
      " -3.3611e-02  5.3481e-03 -4.0087e-02  2.8521e-03 -7.7641e-02\n",
      "  6.4816e-02  3.6758e-02  4.8816e-02 -3.3225e-03 -5.1894e-03\n",
      "  3.8270e-02 -5.2504e-02 -5.3306e-02  3.2807e-02 -6.9571e-02\n",
      " -1.7032e-02 -8.0857e-02 -5.3255e-02 -3.0414e-02 -1.6542e-02\n",
      " -4.3085e-02  4.0040e-03 -1.8141e-02 -1.9958e-02 -5.6711e-03\n",
      "[torch.cuda.FloatTensor of size 20x10x5x5 (GPU 0)]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "t = model.conv2.weight \n",
    "print(type(t.data))\n",
    "print(t.data.size())\n",
    "print(t.data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "\n",
    "\n",
    "You always can move tensor to the CPU (or make sure it's there) using the .cpu() method. Alternatively, you can always move a tensor to the GPU, if a GPU is available (torch.cuda.is_available()) using .cuda() method. You can see that if the tensor is on GPU, its type is 'torch.cuda.FloatTensor'. If tensor is on the CPU, its type is just 'torch.FloatTensor'.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'torch.FloatTensor'>\n",
      "Cuda is available\n",
      "<class 'torch.cuda.FloatTensor'>\n"
     ]
    }
   ],
   "source": [
    "print (type(t.cpu().data))\n",
    "if torch.cuda.is_available():\n",
    "    print (\"Cuda is available\")\n",
    "    print (type(t.cuda().data))\n",
    "else:\n",
    "    print (\"Cuda is NOT available\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "If the tensor is on the CPU, we can transform it to numpy array sharing the same memory location and changing the one will change the other. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'numpy.ndarray'>\n",
      "(20, 10, 5, 5)\n",
      "[[[[ -8.20374626e-05   1.35821616e-02   2.02440582e-02  -1.99429858e-02\n",
      "     -7.78933167e-02]\n",
      "   [ -1.03958428e-01  -8.39783996e-02  -1.24106504e-01  -1.29314303e-01\n",
      "     -4.08263505e-02]\n",
      "   [ -1.23256370e-01  -9.85683054e-02  -5.33063151e-02   1.01492234e-01\n",
      "      1.26388997e-01]\n",
      "   [ -8.66612867e-02  -6.21546768e-02  -6.57773316e-02  -5.78570701e-02\n",
      "     -6.26707450e-02]\n",
      "   [ -6.57874942e-02  -1.86357662e-01  -1.62909701e-01  -1.73445672e-01\n",
      "     -1.29978418e-01]]\n",
      "\n",
      "  [[ -5.83886914e-02  -2.20877603e-02  -6.84031472e-02   2.62686014e-02\n",
      "      5.05783781e-02]\n",
      "   [  2.90736072e-02   7.55167156e-02   1.21565565e-01   1.34170964e-01\n",
      "      7.30192587e-02]\n",
      "   [  3.96843553e-02  -7.80714527e-02  -3.46058942e-02  -2.57841907e-02\n",
      "      1.50611680e-02]\n",
      "   [  4.41484600e-02  -2.44949162e-02  -3.20279710e-02   7.86829293e-02\n",
      "      6.56709597e-02]\n",
      "   [ -1.08956403e-04   5.97056448e-02   1.89267155e-02   3.13353091e-02\n",
      "     -3.64244208e-02]]\n",
      "\n",
      "  [[ -4.50419225e-02  -1.58146527e-02  -1.09172007e-02  -4.08128016e-02\n",
      "      2.71183695e-03]\n",
      "   [  3.16815116e-02   4.86920401e-02  -4.90760021e-02  -7.47012123e-02\n",
      "     -6.03332147e-02]\n",
      "   [ -4.14458811e-02  -1.27627999e-01  -8.71521086e-02   1.85069144e-02\n",
      "      6.41146973e-02]\n",
      "   [ -5.99461384e-02  -3.43797281e-02   7.62198791e-02   9.31143109e-03\n",
      "      4.07778285e-02]\n",
      "   [  6.48806393e-02  -1.52715398e-02   8.17599595e-02   7.27016712e-04\n",
      "      1.82558019e-02]]\n",
      "\n",
      "  ..., \n",
      "  [[  1.53412800e-02  -6.55514747e-02  -2.24889512e-03  -3.82118509e-03\n",
      "     -2.91758776e-02]\n",
      "   [ -8.53257179e-02  -4.57785279e-02  -2.78347544e-02  -9.00501236e-02\n",
      "     -5.26098646e-02]\n",
      "   [ -7.23687187e-02  -1.17277084e-02   2.97023430e-02   1.86282005e-02\n",
      "     -4.23102938e-02]\n",
      "   [ -7.99198449e-02   4.19813171e-02   1.86300799e-02   1.76339247e-03\n",
      "     -8.18940774e-02]\n",
      "   [ -4.36349511e-02  -1.26260864e-02   3.49804722e-02  -1.18564665e-01\n",
      "     -9.34550762e-02]]\n",
      "\n",
      "  [[ -1.17509654e-02   3.22040804e-02  -3.91338877e-02  -5.59439808e-02\n",
      "     -5.01212962e-02]\n",
      "   [ -2.48563476e-02   5.55336364e-02   7.49514103e-02   8.41538236e-02\n",
      "      9.39874053e-02]\n",
      "   [ -3.61047359e-03   5.11956401e-03  -1.62837785e-02   3.73309478e-02\n",
      "      1.75742637e-02]\n",
      "   [ -2.40772795e-02  -5.40698133e-02   1.48370536e-02  -7.02210888e-02\n",
      "     -3.86224426e-02]\n",
      "   [  7.70820975e-02  -4.49969731e-02   7.34456703e-02  -4.28184308e-02\n",
      "     -3.94586362e-02]]\n",
      "\n",
      "  [[ -7.41018802e-02  -9.24578402e-03  -4.24078479e-02  -4.64858897e-02\n",
      "     -8.72670114e-02]\n",
      "   [ -4.62272875e-02  -1.56644844e-02   2.23442111e-02   1.65609419e-02\n",
      "      9.49979126e-02]\n",
      "   [ -3.62699591e-02  -3.99925075e-02   3.39298276e-04  -2.15633586e-02\n",
      "     -1.99651048e-02]\n",
      "   [ -4.67554405e-02   7.34997308e-03   2.38130596e-02  -9.72098559e-02\n",
      "     -2.58948635e-02]\n",
      "   [  6.70231581e-02   8.19264129e-02   1.15070688e-02   3.46349441e-02\n",
      "      2.33243667e-02]]]\n",
      "\n",
      "\n",
      " [[[  6.00155741e-02   5.00063449e-02   1.79480731e-01   4.79969382e-02\n",
      "     -4.23564650e-02]\n",
      "   [ -5.12573980e-02   9.41352621e-02   9.51435268e-02  -1.47884011e-01\n",
      "     -8.63592327e-02]\n",
      "   [ -1.28532544e-01  -6.18918166e-02  -2.72543430e-01  -8.26183408e-02\n",
      "      9.55599099e-02]\n",
      "   [ -6.87635913e-02  -8.87800083e-02  -9.29865167e-02   1.11639276e-01\n",
      "      3.17699686e-02]\n",
      "   [ -5.39916717e-02  -3.87000516e-02  -5.91179468e-02  -9.11827609e-02\n",
      "     -1.00233361e-01]]\n",
      "\n",
      "  [[ -4.63648811e-02  -1.27751287e-02  -3.27479793e-03  -1.33635491e-01\n",
      "     -1.13487199e-01]\n",
      "   [ -1.33931404e-02  -2.78923791e-02  -5.27315252e-02   5.02887554e-02\n",
      "      1.08088218e-01]\n",
      "   [  6.94443751e-03   9.78580192e-02   1.19826972e-01   1.15193859e-01\n",
      "      4.19126526e-02]\n",
      "   [ -2.44287327e-02  -1.63984224e-02  -2.96571497e-02   4.01569344e-03\n",
      "     -1.11360727e-02]\n",
      "   [ -4.63950001e-02  -5.63501753e-02  -8.50678161e-02   1.00645740e-02\n",
      "     -6.64327526e-04]]\n",
      "\n",
      "  [[ -5.67536941e-03  -5.74634001e-02   5.52086942e-02  -9.70758572e-02\n",
      "      1.08345784e-02]\n",
      "   [ -7.41806179e-02  -8.44987854e-02  -1.27070285e-02  -1.31803885e-01\n",
      "     -1.37096688e-01]\n",
      "   [ -1.74852703e-02  -8.22390392e-02  -1.56251892e-01  -1.92091152e-01\n",
      "     -2.92636864e-02]\n",
      "   [ -3.22972536e-02  -1.24609098e-01  -1.27706349e-01  -6.77894726e-02\n",
      "      2.03496832e-02]\n",
      "   [ -4.28161807e-02  -1.02202512e-01  -1.90459471e-02  -9.05647501e-02\n",
      "      2.33979281e-02]]\n",
      "\n",
      "  ..., \n",
      "  [[ -3.53950961e-03   1.01021893e-01   3.86457965e-02   2.28073113e-02\n",
      "      9.05536115e-03]\n",
      "   [ -5.31937368e-02  -2.39859466e-02  -3.37214470e-02  -3.03688534e-02\n",
      "      4.98819165e-02]\n",
      "   [ -6.51248544e-02  -9.98034775e-02  -1.82180703e-02   2.71638166e-02\n",
      "      1.37532987e-02]\n",
      "   [ -7.26061910e-02  -5.20608798e-02  -3.02263256e-02  -5.73440082e-02\n",
      "      2.59051146e-03]\n",
      "   [  1.16600521e-01   2.93984287e-03   6.45032153e-03  -9.03923661e-02\n",
      "     -5.32839186e-02]]\n",
      "\n",
      "  [[  5.37668169e-02   2.27589477e-02   1.41019635e-02   8.61022435e-03\n",
      "     -1.36448652e-01]\n",
      "   [ -1.83070805e-02  -8.04104954e-02   4.84639034e-02   1.11017944e-02\n",
      "      8.09658132e-03]\n",
      "   [  3.49693000e-02   2.00802647e-02   1.40018657e-01   1.21169209e-01\n",
      "      1.49445310e-01]\n",
      "   [  3.85908261e-02  -2.64714304e-02   1.07820053e-02   1.52584240e-02\n",
      "      6.55916426e-03]\n",
      "   [  5.84379071e-03  -2.91013084e-02  -2.22564116e-02  -1.17588028e-01\n",
      "     -4.25722077e-02]]\n",
      "\n",
      "  [[ -4.26571406e-02   8.47827122e-02   4.38667387e-02   1.78132555e-03\n",
      "     -2.25779954e-02]\n",
      "   [ -1.87227018e-02  -3.10194530e-02  -4.11049798e-02  -3.25036645e-02\n",
      "      3.73086706e-02]\n",
      "   [ -9.61597785e-02   4.28227074e-02   1.78151410e-02   4.79387864e-02\n",
      "      1.36627601e-02]\n",
      "   [ -4.88603003e-02  -5.26054017e-02   6.57415688e-02  -5.62230162e-02\n",
      "      1.25827398e-02]\n",
      "   [ -2.18827426e-02   2.45375629e-03   5.13307601e-02  -3.70475533e-03\n",
      "     -2.82348283e-02]]]\n",
      "\n",
      "\n",
      " [[[ -1.13432901e-02  -1.65147081e-01   1.37504693e-02  -7.30381405e-04\n",
      "      4.12788726e-02]\n",
      "   [ -9.34568644e-02  -2.03576490e-01  -7.58082345e-02   1.04391187e-01\n",
      "      1.82641357e-01]\n",
      "   [ -5.82408942e-02  -1.63434297e-01   1.22746140e-01   8.37150589e-02\n",
      "     -3.41176949e-02]\n",
      "   [ -1.39590457e-01   5.08410595e-02   4.41823006e-02  -3.82821150e-02\n",
      "     -1.51857138e-01]\n",
      "   [ -6.46891817e-02   4.35186885e-02  -3.89285298e-04  -3.04784141e-02\n",
      "     -1.00916877e-01]]\n",
      "\n",
      "  [[ -1.18873656e-01  -1.08937830e-01  -4.42392938e-02   8.18073303e-02\n",
      "     -6.45687580e-02]\n",
      "   [ -1.27912000e-01  -4.38959524e-02  -4.71252464e-02  -4.30311970e-02\n",
      "     -1.26122907e-01]\n",
      "   [ -5.46406433e-02  -7.78956711e-02  -3.24070677e-02  -6.80018142e-02\n",
      "     -1.41882464e-01]\n",
      "   [ -2.81611178e-02   8.15599971e-03  -1.39224038e-01  -5.96593171e-02\n",
      "      8.42316076e-02]\n",
      "   [  8.85649249e-02  -3.93756256e-02  -7.81870931e-02   3.91058400e-02\n",
      "     -2.83834934e-02]]\n",
      "\n",
      "  [[ -1.08800001e-01   4.36081737e-03   3.90583761e-02   7.15717971e-02\n",
      "      2.64271125e-02]\n",
      "   [ -1.02938585e-01  -4.15987782e-02  -4.42243256e-02  -4.59406972e-02\n",
      "     -1.47973429e-02]\n",
      "   [ -9.22977105e-02  -5.04026748e-02  -4.06769104e-02  -5.81070334e-02\n",
      "     -9.47554708e-02]\n",
      "   [ -7.03690201e-02  -5.44602387e-02   3.31436209e-02  -2.47582477e-02\n",
      "     -4.31794263e-02]\n",
      "   [ -2.16397241e-01  -9.15488899e-02  -2.26614671e-03  -1.99081022e-02\n",
      "      3.92876975e-02]]\n",
      "\n",
      "  ..., \n",
      "  [[  6.92713587e-03  -5.09818783e-03  -1.59813762e-02  -3.19672264e-02\n",
      "      5.71267158e-02]\n",
      "   [ -1.22562489e-02   5.17876633e-02   6.98315725e-02   8.00152421e-02\n",
      "      5.29988334e-02]\n",
      "   [ -8.82966071e-02   7.42221624e-02   4.94273119e-02   1.09039493e-01\n",
      "     -4.60029393e-02]\n",
      "   [ -9.51279849e-02   7.42227957e-02   5.36096059e-02   1.73593193e-01\n",
      "     -1.40723124e-01]\n",
      "   [ -5.03800400e-02  -2.24669967e-02   2.42999792e-02   1.08120993e-01\n",
      "     -5.48940636e-02]]\n",
      "\n",
      "  [[ -1.19847106e-02  -1.73605792e-02  -2.11292766e-02   1.67630482e-02\n",
      "     -3.69961336e-02]\n",
      "   [  3.59808989e-02   7.33823702e-02  -3.20010260e-02   1.02014683e-01\n",
      "     -6.18569963e-02]\n",
      "   [  7.55599141e-03   1.06523344e-02  -3.61297615e-02  -6.13762923e-02\n",
      "     -1.36718377e-01]\n",
      "   [  4.64320667e-02  -1.26569450e-01  -7.67161846e-02  -6.97960034e-02\n",
      "     -6.56472519e-02]\n",
      "   [  5.27308229e-03  -7.77473301e-02  -1.16894446e-01  -2.75262259e-02\n",
      "      4.61079888e-02]]\n",
      "\n",
      "  [[ -7.80758262e-02  -8.10922310e-02  -1.03368871e-02  -4.44799550e-02\n",
      "      1.24936048e-02]\n",
      "   [  4.72132228e-02  -3.27833965e-02   5.23973033e-02   3.65681499e-02\n",
      "     -4.12011631e-02]\n",
      "   [ -4.18493822e-02  -6.70980886e-02  -3.01863551e-02   8.28727428e-03\n",
      "     -1.05844595e-01]\n",
      "   [  4.93454412e-02  -4.76631448e-02  -1.20374039e-04  -5.46130165e-02\n",
      "     -1.35212746e-02]\n",
      "   [  3.61775197e-02   7.00696334e-02  -4.11574058e-02  -7.49932183e-03\n",
      "     -4.00166549e-02]]]\n",
      "\n",
      "\n",
      " ..., \n",
      " [[[  8.31502900e-02  -6.01003878e-02  -3.68162151e-03   2.64509656e-02\n",
      "      1.80176064e-01]\n",
      "   [ -7.90639296e-02  -2.54694402e-01  -9.40423012e-02  -6.80610212e-03\n",
      "     -6.64819479e-02]\n",
      "   [ -1.71607465e-01  -2.13202581e-01  -2.52754509e-01  -2.07591638e-01\n",
      "     -2.92971879e-01]\n",
      "   [ -8.49724561e-02  -7.52245188e-02  -9.47817862e-02  -9.48553830e-02\n",
      "     -1.20887093e-01]\n",
      "   [  5.79060316e-02   6.15428612e-02   2.60893106e-02   1.43141747e-02\n",
      "      4.42696847e-02]]\n",
      "\n",
      "  [[ -9.72698256e-02  -8.34404901e-02  -4.36783321e-02   3.34956273e-02\n",
      "     -1.87434424e-02]\n",
      "   [  8.87315348e-02   7.57521465e-02   1.78986304e-02  -6.71348721e-02\n",
      "      3.26617551e-03]\n",
      "   [  7.54582882e-02  -2.31768154e-02   4.43861680e-03   5.47687598e-02\n",
      "      1.14186369e-01]\n",
      "   [  7.86225051e-02   1.36670610e-02   2.13152040e-02  -3.60767581e-02\n",
      "      9.97344851e-02]\n",
      "   [ -6.64460585e-02  -7.09741861e-02   4.18284535e-03   1.57872750e-03\n",
      "     -4.79722470e-02]]\n",
      "\n",
      "  [[ -6.03395030e-02   2.20189095e-02   1.21018374e-02   4.51353146e-03\n",
      "      8.74627084e-02]\n",
      "   [ -4.42555025e-02  -5.14476709e-02  -9.74003319e-03  -3.45400386e-02\n",
      "     -9.80682671e-05]\n",
      "   [ -2.81357523e-02  -1.60891823e-02  -1.51767079e-02  -7.39409029e-02\n",
      "     -1.00031614e-01]\n",
      "   [  1.64058041e-02  -4.33877073e-02  -1.51891885e-02  -3.46090309e-02\n",
      "      1.23679535e-02]\n",
      "   [  1.21321995e-03   6.98564528e-03   1.58656444e-02   6.47586435e-02\n",
      "      1.08142853e-01]]\n",
      "\n",
      "  ..., \n",
      "  [[ -1.23381719e-01  -4.48085666e-02   2.90706865e-02   2.47439686e-02\n",
      "      3.27103436e-02]\n",
      "   [ -1.50786966e-01  -1.35982810e-02  -5.18653691e-02   7.63864547e-04\n",
      "     -6.42127395e-02]\n",
      "   [ -5.91070764e-02   1.82225946e-02  -3.00556421e-02  -1.55618764e-03\n",
      "     -7.10452646e-02]\n",
      "   [ -2.42717303e-02   2.83402093e-02  -2.89214011e-02   3.07442024e-02\n",
      "     -6.14664666e-02]\n",
      "   [ -2.45778803e-02   1.19632147e-02   7.35689839e-03   1.06481100e-02\n",
      "     -7.57829994e-02]]\n",
      "\n",
      "  [[ -8.41095150e-02  -1.16410717e-01  -2.03015432e-02   1.42322928e-02\n",
      "      9.87365097e-03]\n",
      "   [ -3.51055805e-03  -1.85078918e-03   7.96433613e-02  -3.99449132e-02\n",
      "     -8.73122662e-02]\n",
      "   [  1.09060975e-02   7.61488304e-02   1.52304098e-01   8.24424177e-02\n",
      "      9.64798182e-02]\n",
      "   [  7.41204843e-02   7.64587000e-02   8.36235881e-02   1.52796254e-01\n",
      "      1.51670814e-01]\n",
      "   [  1.90251544e-02   1.79672260e-02   3.17013487e-02   4.43938337e-02\n",
      "     -7.92793743e-03]]\n",
      "\n",
      "  [[ -6.66492656e-02  -4.01740074e-02   2.91575287e-02   2.84684952e-02\n",
      "     -5.25429808e-02]\n",
      "   [ -4.69936021e-02   2.15478037e-02   8.46225321e-02   3.21447919e-03\n",
      "     -2.08208747e-02]\n",
      "   [  4.47466113e-02  -1.78063009e-02  -1.01376446e-02   2.38508042e-02\n",
      "      1.21057900e-02]\n",
      "   [  7.58853480e-02   2.27512121e-02   4.17034812e-02   6.94001392e-02\n",
      "      7.93415755e-02]\n",
      "   [  4.92254309e-02  -7.60664279e-03   3.54582295e-02   1.33093037e-02\n",
      "     -8.39925185e-02]]]\n",
      "\n",
      "\n",
      " [[[ -7.18170404e-02  -1.19570263e-01   7.49770850e-02   1.62943546e-02\n",
      "     -6.01310134e-02]\n",
      "   [  1.22044794e-01  -3.09330635e-02  -1.91663936e-01  -9.48323533e-02\n",
      "     -2.74725189e-03]\n",
      "   [  5.41867800e-02   1.24250948e-01  -7.84497410e-02  -1.63638532e-01\n",
      "     -4.12709713e-02]\n",
      "   [  1.22939050e-01   3.22151855e-02   2.78631914e-02   4.70955744e-02\n",
      "      1.59703642e-01]\n",
      "   [  6.49094433e-02   1.00947015e-01   9.92663875e-02   1.41091108e-01\n",
      "      4.72717313e-03]]\n",
      "\n",
      "  [[ -1.65980626e-02   8.02425146e-02   5.21947481e-02   3.63216959e-02\n",
      "     -5.70243709e-02]\n",
      "   [  3.92623655e-02  -4.02854457e-02   5.09970821e-02   1.38521224e-01\n",
      "      5.54469647e-03]\n",
      "   [ -3.39636318e-02   7.79024884e-02  -4.11241762e-02   1.48379216e-02\n",
      "      2.21401863e-02]\n",
      "   [  3.87083814e-02   4.65576127e-02   3.34232673e-02  -4.79677059e-02\n",
      "      2.79056728e-02]\n",
      "   [ -3.96253020e-02   2.08452176e-02   6.04856806e-03  -7.08815306e-02\n",
      "     -1.59794778e-01]]\n",
      "\n",
      "  [[ -6.65592030e-02  -1.63277566e-01  -7.41532119e-03  -4.75268625e-02\n",
      "     -3.70069519e-02]\n",
      "   [ -6.47313595e-02  -7.33540952e-02  -5.98420091e-02  -1.26197869e-02\n",
      "     -2.87337210e-02]\n",
      "   [ -1.16068413e-02   8.60683843e-02  -1.89782996e-02  -3.64581533e-02\n",
      "     -1.10548139e-01]\n",
      "   [ -4.93588448e-02   7.70138875e-02   6.86497688e-02  -5.40418886e-02\n",
      "      2.78540794e-02]\n",
      "   [  5.97189479e-02  -1.56758800e-02   9.08316374e-02  -1.77715614e-03\n",
      "      4.22653519e-02]]\n",
      "\n",
      "  ..., \n",
      "  [[ -3.93114127e-02  -6.19087294e-02  -1.98676195e-02   8.03806074e-03\n",
      "     -2.24862117e-02]\n",
      "   [ -9.33932289e-02  -1.35173416e-02  -1.41398953e-02   6.60676509e-02\n",
      "      1.12935878e-01]\n",
      "   [ -4.06784639e-02  -3.91674377e-02   3.66901048e-02   4.89725769e-02\n",
      "      1.17575981e-01]\n",
      "   [ -2.05638334e-02   4.84617194e-04   1.19994514e-01  -1.89755987e-02\n",
      "      7.18562007e-02]\n",
      "   [ -7.43421167e-02  -2.70975046e-02   2.61789653e-02  -1.43208327e-02\n",
      "     -4.62118387e-02]]\n",
      "\n",
      "  [[  4.98447157e-02   4.96022142e-02   7.17834150e-03   4.05999795e-02\n",
      "      5.67013919e-02]\n",
      "   [ -7.45661706e-02  -1.91795267e-02  -2.25533806e-02   6.92940652e-02\n",
      "      9.79183391e-02]\n",
      "   [ -1.60637677e-01  -1.63675353e-01  -1.01245537e-01   3.48732956e-02\n",
      "      1.56918988e-01]\n",
      "   [ -1.77560240e-01  -2.10667312e-01  -5.62977269e-02   3.34889479e-02\n",
      "      1.58232283e-02]\n",
      "   [ -6.08660206e-02   1.19806463e-02   1.14528365e-01   8.06961767e-03\n",
      "     -1.48743233e-02]]\n",
      "\n",
      "  [[  2.16131490e-02   4.46726643e-02   8.38894248e-02   2.52310578e-02\n",
      "      1.41836470e-02]\n",
      "   [  6.87546730e-02   7.20493868e-02   1.34622410e-01   1.50996655e-01\n",
      "     -7.14915385e-03]\n",
      "   [ -2.01703589e-02  -9.11028460e-02  -7.26275593e-02  -6.24181479e-02\n",
      "      8.67013559e-02]\n",
      "   [  2.27544457e-02   9.27766263e-02   4.40841131e-02  -4.10381258e-02\n",
      "     -2.67833308e-03]\n",
      "   [  6.18339032e-02   1.06301829e-01   1.57524720e-02  -1.01286813e-03\n",
      "     -5.33636250e-02]]]\n",
      "\n",
      "\n",
      " [[[ -1.26836626e-02   3.40916477e-02   1.87310465e-02   4.21637185e-02\n",
      "      1.45046199e-02]\n",
      "   [  9.98833105e-02   6.79155439e-02  -4.06518690e-02  -5.69119677e-02\n",
      "     -5.19773550e-02]\n",
      "   [  2.53865480e-01  -6.62742369e-03  -1.80407569e-01  -1.43699452e-01\n",
      "     -8.43903050e-02]\n",
      "   [  1.70709759e-01   4.49592024e-02  -4.54755239e-02  -1.55688331e-01\n",
      "     -1.20233089e-01]\n",
      "   [  1.13655627e-01   1.80172235e-01   6.21429496e-02  -1.03381939e-01\n",
      "     -1.13745421e-01]]\n",
      "\n",
      "  [[ -2.96379756e-02  -7.56910667e-02  -9.43707824e-02  -7.80255869e-02\n",
      "      2.11196933e-02]\n",
      "   [ -6.60778806e-02  -1.03922069e-01  -4.40546237e-02  -3.26233767e-02\n",
      "     -9.05961320e-02]\n",
      "   [  8.21509212e-03  -8.09769705e-02  -5.23917340e-02  -2.13982398e-03\n",
      "     -8.76180753e-02]\n",
      "   [  2.05293354e-02  -1.19106248e-02  -5.04455343e-02  -1.01169430e-01\n",
      "     -1.06564239e-01]\n",
      "   [  8.74150321e-02   9.70446412e-03   1.96657497e-02  -3.42981219e-02\n",
      "     -8.78666639e-02]]\n",
      "\n",
      "  [[ -1.34484634e-01  -9.73683223e-02   1.08584929e-02   1.72863472e-02\n",
      "      3.85938818e-03]\n",
      "   [ -5.08553013e-02  -6.63895980e-02  -5.92358061e-04  -2.40023690e-03\n",
      "     -5.48834428e-02]\n",
      "   [  1.05088204e-02   4.04773001e-03  -4.36562113e-02   3.26447673e-02\n",
      "     -1.71783604e-02]\n",
      "   [ -2.47949716e-02  -1.13878511e-02   8.55627470e-03  -4.50764187e-02\n",
      "      4.47429121e-02]\n",
      "   [  1.18802510e-01   1.63518023e-02  -3.09092947e-03  -2.08711457e-02\n",
      "     -3.74923907e-02]]\n",
      "\n",
      "  ..., \n",
      "  [[  5.84548488e-02  -7.11920625e-03  -3.43984701e-02  -4.10172865e-02\n",
      "      5.64996898e-02]\n",
      "   [  4.76912148e-02  -4.18138038e-03  -3.81300636e-02  -6.08209558e-02\n",
      "      7.68335164e-02]\n",
      "   [  2.03356501e-02   7.42083117e-02  -4.34930660e-02  -7.15441555e-02\n",
      "     -2.55171917e-02]\n",
      "   [  8.06021541e-02   1.24158055e-01  -1.00612789e-02   6.60433099e-02\n",
      "      6.86030909e-02]\n",
      "   [  2.96977051e-02   3.98916006e-02   4.98613119e-02   7.18980581e-02\n",
      "      2.50710342e-02]]\n",
      "\n",
      "  [[ -6.09009899e-02   2.37426721e-02  -2.02478953e-02  -5.69846220e-02\n",
      "     -2.60700583e-02]\n",
      "   [ -3.22246775e-02   1.08824559e-01   8.74285474e-02  -4.78877723e-02\n",
      "     -7.20841363e-02]\n",
      "   [  4.60410416e-02   6.94972137e-03   4.63546924e-02  -1.34936590e-02\n",
      "      7.12305121e-03]\n",
      "   [ -2.58257855e-02  -6.23104423e-02  -1.76750757e-02   2.96056923e-02\n",
      "      2.59947851e-02]\n",
      "   [ -8.58641192e-02   1.09745553e-02  -3.65842953e-02  -5.41808605e-02\n",
      "      6.78327233e-02]]\n",
      "\n",
      "  [[ -3.36112641e-02   5.34814503e-03  -4.00868505e-02   2.85205198e-03\n",
      "     -7.76412934e-02]\n",
      "   [  6.48164749e-02   3.67582105e-02   4.88162786e-02  -3.32253240e-03\n",
      "     -5.18940156e-03]\n",
      "   [  3.82703356e-02  -5.25044613e-02  -5.33058867e-02   3.28069814e-02\n",
      "     -6.95708990e-02]\n",
      "   [ -1.70316547e-02  -8.08569342e-02  -5.32546453e-02  -3.04136761e-02\n",
      "     -1.65421665e-02]\n",
      "   [ -4.30854000e-02   4.00404632e-03  -1.81407295e-02  -1.99580118e-02\n",
      "     -5.67107042e-03]]]]\n"
     ]
    }
   ],
   "source": [
    "if torch.cuda.is_available():\n",
    "    try:\n",
    "        print(t.data.numpy())\n",
    "    except RuntimeError as e:\n",
    "        \"you can't transform a GPU tensor to a numpy nd array, you have to copy your weight tendor to cpu and then get the numpy array\"\n",
    "\n",
    "print(type(t.cpu().data.numpy()))\n",
    "print(t.cpu().data.numpy().shape)\n",
    "print(t.cpu().data.numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "\n",
    "\n",
    "Now that we know how to convert tensors to numpy arrays, we can use this knowledge to vizualise them with matplotlib! Let's print out the conv filters of the first convolutional layer.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10, 1, 5, 5)\n",
      "(10, 5, 5)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABjcAAACyCAYAAAAOE51uAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAHF4IRr7ZJDLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20a397f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = model.conv1.weight.cpu().data.numpy()\n",
    "print (data.shape)\n",
    "print (data[:, 0].shape)\n",
    "\n",
    "kernel_num = data.shape[0]\n",
    "\n",
    "fig, axes = plt.subplots(ncols=kernel_num, figsize=(2*kernel_num, 2))\n",
    "\n",
    "for col in range(kernel_num):\n",
    "    axes[col].imshow(data[col, 0, :, :], cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "\n",
    "\n",
    "These plots have a large amount of contrast, but that is because they are normalized by the maximum/minimum weight in each filter, not across all the filters. After normalization, the plots look like the figures below.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABjcAAACyCAYAAAAOE51uAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ntS+EOdIkLQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20a3ec7b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(ncols=kernel_num, figsize=(2*kernel_num, 2))\n",
    "\n",
    "for col in range(kernel_num):\n",
    "    axes[col].imshow(data[col, 0, :, :], vmin=np.amin(data), vmax=np.amax(data), cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "\n",
    "\n",
    "If you want also zero-center the plots, you need to add this condition to the plot properties. After zero-centering, the plots look almost identical, which tells us that the weights are relatively balanced with respect to zero.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Minimal weight -0.476305\n",
      "Maximal weight 0.426415\n",
      "Absolute maximal weight 0.476305\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABjcAAACyCAYAAAAOE51uAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAHZ5JREFUeJzt3V1s1uX9+PFPaWnLQ0WRgToYY5FE\nDTM6YBlOhUWHU7NNt2RbpiZb3BYjOggnmzPZ0wnZyWIyHyLGeGKcLHFGD5xZfQA0TiWI82Eb29wS\nqqiAiwUKtAXu/8H+5XcjbrOl13Xd36+vV9IDar0+32/75rrv9qJtW6PRaAQAAAAAAEBFTCh9AQAA\nAAAAAKPhcAMAAAAAAKgUhxsAAAAAAEClONwAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAA\nAAAAVIrDDQAAAAAAoFIcbgAAAAAAAJXicAMAAAAAAKgUhxsAAAAAAECldOQcdvjw4di+fXv09PRE\nW1tbztG0oEajEXv27InTTjstJkxId86mO0bkai5Cd/wfex0l6I7cPMZSgr2OEnRHbpqjBN2R21ib\ny3q4sX379pgzZ07OkVRAX19fzJ49O9n6uuO9UjcXoTuOZa+jBN2Rm8dYSrDXUYLuyE1zlKA7chtt\nc1kPN3p6eiIiYt68ecn/Rdd3v/vdpOs3mzFjRpY5u3btyjInIqKzszP5jAMHDsSPfvSjI12kMrL+\nrbfeGpMmTUo666yzzkq6frNcm/+mTZuyzImIeOqpp5KuPzg4GLfddlvy5iL+r7vrr78+urq6ks46\n55xzkq7f7Lnnnssy51vf+laWORERr732WtL19+3bF9dee222ve7OO+9MvtcNDg4mXb/Zxo0bs8wZ\nHh7OMici4oknnkg+4/Dhw7Fr165s3f3kJz+J7u7upLMuueSSpOs3mzhxYpY5P/7xj7PMiYjo7+9P\nuv7Bgwdj48aNWR9jlyxZEh0daT+dyfncbvPmzVnm5Lynk08+Oen6g4ODcfvtt2fb69avXx9Tp05N\nOiv1+6zZn/70pyxz3n777SxzIiJmzpyZfMa+ffvia1/7Wrbuli1blnyv279/f9L1m6V+njriwgsv\nzDInIuJzn/tc0vUHBgbiC1/4QrbmTj/99Ghvb086K9dzrYiIU045JdusXE499dTkM4aGhmLdunXZ\nuvvFL36RfH945ZVXkq7fLNd3oaT+R0XNdu7cmXT9wcHBuPPOO0fdXNbDjZEP7IQJE5JvlKk/wW42\nefLkLHNy3lPqL8Y2S/0XfmT9SZMmJf9Ypf5kp9kJJ5yQZc6UKVOyzInI112OB5mRGV1dXcnvK9ce\nFJHvY5Tz71Ku91+d9rrUj+HNchy2R+R78hkRyf+BR7Nc3XV3dyd/npJzX8jVXc5P7FN/YWxEzsfY\njo6O5PeV8zlxro9Rrr4j6vPcbmT9qVOnJt+Lcj3Hj8j3PD/nc9Wcn7vk6i7HXpdr/4nI99iX8+sn\nuZ6j5Gquvb09+fP9nJ9P1Ok50Iicj+U5P49NfbhRp/fbiDp+rXi07zu/UBwAAAAAAKgUhxsAAAAA\nAEClONwAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAAAAAAVMqYDjduv/32mDdvXnR3d8fC\nhQvjqaeeGu/rgmPojtw0Rwm6IzfNUYLuKEF35KY5StAdJeiOUkZ9uLFu3bpYtWpV3HzzzbFly5a4\n4IIL4tJLL41t27aluD6ICN2Rn+YoQXfkpjlK0B0l6I7cNEcJuqME3VHSqA83fvnLX8a1114b3/nO\nd+LMM8+MW265JebMmRN33HFHiuuDiNAd+WmOEnRHbpqjBN1Rgu7ITXOUoDtK0B0ljepwY2hoKDZv\n3hzLly8/6vXLly+PZ5555pi3HxwcjN27dx/1AqOlO3IbbXMRuuP42evIzV5HCbqjBI+x5GavowR7\nHSXojtJGdbixa9euOHToUMyaNeuo18+aNSveeuutY95+zZo1MW3atCMvc+bMOb6r5UNJd+Q22uYi\ndMfxs9eRm72OEnRHCR5jyc1eRwn2OkrQHaWN6ReKt7W1HfXnRqNxzOsiIm666abo7+8/8tLX1ze2\nq4TQHfl90OYidMf4sdeRm72OEnRHCR5jyc1eRwn2OkrQHaV0jOaNZ8yYEe3t7cecvO3YseOYE7qI\niK6urujq6jq+K+RDT3fkNtrmInTH8bPXkZu9jhJ0RwkeY8nNXkcJ9jpK0B2ljeo7Nzo7O2PhwoXR\n29t71Ot7e3vjvPPOG9cLgxG6IzfNUYLuyE1zlKA7StAduWmOEnRHCbqjtFF950ZExOrVq+Oaa66J\nRYsWxZIlS2Lt2rWxbdu2uO6661JcH0SE7shPc5SgO3LTHCXojhJ0R26aowTdUYLuKGnUhxtf//rX\n45133omf//zn8eabb8aCBQvikUceiblz56a4PogI3ZGf5ihBd+SmOUrQHSXojtw0Rwm6owTdUdKo\nDzciIq6//vq4/vrrx/ta4L/SHblpjhJ0R26aowTdUYLuyE1zlKA7StAdpYzqd24AAAAAAACU5nAD\nAAAAAACoFIcbAAAAAABApTjcAAAAAAAAKsXhBgAAAAAAUCkdJYa+9tpryWfs2rUr+YwR06dPzzLn\nsssuyzInIuKf//xn8hkDAwPJZzRbt25dTJw4MemMXC1ERPztb3/LMueiiy7KMiciYvfu3UnXHxoa\nSrr++3n11VeTd3fiiScmXb/ZmWeemWXOk08+mWVORMTVV1+ddP3UXb/Xk08+GZ2dnUlnvP7660nX\nb/bmm29mmbNjx44scyIirrzyyuQzhoaG4u67704+Z8Qrr7ySvLsvfvGLSddvluN5UETEO++8k2VO\nRMRf/vKXpOsfPnw46frvp7OzMzo60n4609fXl3T9Zrka37lzZ5Y5ERGf+cxnkq6/b9++pOu/14MP\nPhjd3d1JZyxatCjp+s3a2tqyzDnjjDOyzImI2Lp1a/IZubt7++23o729PemMZcuWJV2/Wa7ndgsW\nLMgyJyJiy5YtSdffv39/0vXfq6OjI3lz8+fPT7p+s0OHDmWZ8+6772aZExHx+OOPJ5+R+7ndXXfd\nlby7j33sY0nXb3buuedmmfPyyy9nmRORfl89cODAmP4/37kBAAAAAABUisMNAAAAAACgUhxuAAAA\nAAAAleJwAwAAAAAAqBSHGwAAAAAAQKU43AAAAAAAACrF4QYAAAAAAFApDjcAAAAAAIBKcbgBAAAA\nAABUisMNAAAAAACgUhxuAAAAAAAAleJwAwAAAAAAqBSHGwAAAAAAQKU43AAAAAAAACrF4QYAAAAA\nAFApDjcAAAAAAIBKcbgBAAAAAABUisMNAAAAAACgUhxuAAAAAAAAleJwAwAAAAAAqBSHGwAAAAAA\nQKU43AAAAAAAACrF4QYAAAAAAFApDjcAAAAAAIBKcbgBAAAAAABUisMNAAAAAACgUjpKDP3mN78Z\nnZ2dSWecc845SddvNn369CxztmzZkmVORMSrr76afMbg4GDyGc2Gh4ej0WgknXH//fcnXb/Z1KlT\ns8w5fPhwljkREeedd17S9XPey4grrrgiJk2alHTG888/n3T9Zvv3788y55133skyJyLi4YcfTrp+\nrvfZiFmzZkV3d3fSGc8991zS9Zulfr4w4hvf+EaWORERV199dfIZe/fujbvvvjv5nBFLly5Nvtf9\n9a9/Tbp+s927d2eZ89GPfjTLnIiIU045Jen6w8PD8Zvf/CbpjPdqa2uLCRPS/lutT37yk0nXb7Z1\n69Ysc4aHh7PMiYi49957k66f814iIm655ZZoa2tLOuPzn/980vWb5fqceWBgIMuciIgZM2Ykn3Hg\nwIHkM5rNnTs3Jk6cmHTGY489lnT9ZsuWLcsyJ/Xzkmbz589Pun7Ov0MR/35e19XVlXTGzp07k67f\n7KSTTsoyZ2hoKMuciDzPT4aHh2Pbtm3J54yYNWtWdHSk/TJ1e3t70vWbPf7441nmfOUrX8kyJyJi\n8uTJSdcf62Od79wAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAAAAAAVIrDDQAAAAAAoFIc\nbgAAAAAAAJXicAMAAAAAAKgUhxsAAAAAAEClONwAAAAAAAAqZVSHG2vWrInFixdHT09PzJw5M664\n4orYunVrqmuDiNAd+WmOEnRHCbojN81Rgu7ITXOUoDtK0B2ljepwY8OGDbFixYp49tlno7e3Nw4e\nPBjLly+PgYGBVNcHuiM7zVGC7ihBd+SmOUrQHblpjhJ0Rwm6o7SO0bzxo48+etSf77nnnpg5c2Zs\n3rw5LrzwwnG9MBihO3LTHCXojhJ0R26aowTdkZvmKEF3lKA7ShvV4cZ79ff3R0TE9OnT3/e/Dw4O\nxuDg4JE/7969+3jGQUTojvz+V3MRumP82esoQXfk5jGWEux15GavowR7HSXojtzG/AvFG41GrF69\nOs4///xYsGDB+77NmjVrYtq0aUde5syZM+YLhQjdkd8HaS5Cd4wvex0l6I7cPMZSgr2O3Ox1lGCv\nowTdUcKYDzduuOGGeOmll+LXv/71f3ybm266Kfr7+4+89PX1jXUcRITuyO+DNBehO8aXvY4SdEdu\nHmMpwV5HbvY6SrDXUYLuKGFMP5bqxhtvjIcffjg2btwYs2fP/o9v19XVFV1dXWO+OGimO3L7oM1F\n6I7xY6+jBN2Rm8dYSrDXkZu9jhLsdZSgO0oZ1eFGo9GIG2+8MR588MFYv359zJs3L9V1wRG6IzfN\nUYLuKEF35KY5StAduWmOEnRHCbqjtFEdbqxYsSLuu+++eOihh6KnpyfeeuutiIiYNm1aTJo0KckF\ngu7ITXOUoDtK0B25aY4SdEdumqME3VGC7ihtVL9z44477oj+/v5YtmxZnHrqqUde1q1bl+r6QHdk\npzlK0B0l6I7cNEcJuiM3zVGC7ihBd5Q26h9LBbnpjtw0Rwm6owTdkZvmKEF35KY5StAdJeiO0kb1\nnRsAAAAAAAClOdwAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAAAAAAVEpHiaGLFy+OSZMm\nJZ0xffr0pOs3+93vfpdlzj/+8Y8scyIiDh8+nHzG8PBw8hnNzj333Ojq6ko64/TTT0+6frMpU6Zk\nmXPaaadlmRMRyT8++/fvT7r++/njH/8YnZ2dSWfkvK9c+1DO7ubPn590/YGBgaTrv9cDDzwQ7e3t\nSWdMnTo16frNvvzlL2eZc9ZZZ2WZExHx2GOPJZ9x4MCB5DOavfHGG9Hd3Z10xqxZs5Ku3+y5557L\nMufEE0/MMici4vnnn0+6/sGDB5Ou/36mTJkSEydOTDpjcHAw6frN+vr6ssxJ/RjR7CMf+Ui2WTnM\nmTMn+fvv1VdfTbp+sz//+c9Z5uTcv5ctW5Z8Rs59ISLi73//e/Lucj73fu2117LM2bJlS5Y5Een3\nutyfxy5dujQmT56cdMaLL76YdP1mM2bMyDLnsssuyzInIs/XPPfu3RuPPvpo8jkjrrnmmuTdPf30\n00nXb/a9730vy5ycz8FT9zA0NDSm/893bgAAAAAAAJXicAMAAAAAAKgUhxsAAAAAAEClONwAAAAA\nAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAAAAAAVIrDDQAAAAAAoFIcbgAAAAAAAJXicAMAAAAA\nAKgUhxsAAAAAAEClONwAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACASnG4AQAAAAAAVIrDDQAAAAAA\noFIcbgAAAAAAAJXicAMAAAAAAKgUhxsAAAAAAEClONwAAAAAAAAqxeEGAAAAAABQKQ43AAAAAACA\nSnG4AQAAAAAAVIrDDQAAAAAAoFIcbgAAAAAAAJXicAMAAAAAAKiUjhJDDxw4kHzGr371q+QzRnR1\ndWWZc8kll2SZExHxhz/8IfmMoaGh5DOaXXnllTF16tSkM4aHh5Ou32zSpElZ5mzevDnLnIiInTt3\nJl0/x97zXg888EBMmJD2HPnyyy9Pun6zgwcPZpnz7LPPZpkTEfGJT3wi6fr79+9Pun4JX/3qV7PN\n+te//pVlzq5du7LMiYg48cQTk8/I3d2UKVOSPy719vYmXb/Z9u3bs8x54403ssyJSP+869ChQ0nX\nfz9f+tKXYvLkyUln5Hxu9/GPfzzLnFz7akTEI488knT9XM9LRsydOzcmTpyYdEZ3d3fS9Zu9++67\nWeakfp81y/G5S859ISLis5/9bHR2diadMWXKlKTrN8vVw+DgYJY5ERFLlixJuv6ePXuSrv9eM2fO\nTP61k5z79/r167PM+f73v59lTkREf39/8hkDAwPJZzQ755xzoqenJ+mMxYsXJ12/2aOPPpplzhNP\nPJFlTkTEmWeemXT9se7bvnMDAAAAAACoFIcbAAAAAABApTjcAAAAAAAAKsXhBgAAAAAAUCkONwAA\nAAAAgEpxuAEAAAAAAFSKww0AAAAAAKBSHG4AAAAAAACVclyHG2vWrIm2trZYtWrVeF0P/FeaowTd\nUYLuyE1zlKA7ctMcJeiOEnRHbpqjhDEfbmzatCnWrl0bZ5999nheD/xHmqME3VGC7shNc5SgO3LT\nHCXojhJ0R26ao5QxHW7s3bs3rrrqqrjrrrvipJNOGu9rgmNojhJ0Rwm6IzfNUYLuyE1zlKA7StAd\nuWmOksZ0uLFixYq4/PLL4+KLL/6vbzc4OBi7d+8+6gXG4oM2F6E7xo/uKMFjLLnZ6yhBd+SmOUrQ\nHSX4fILc7HWU1DHa/+H++++PF154ITZt2vQ/33bNmjXxs5/9bEwXBiNG01yE7hgfuqMEj7HkZq+j\nBN2Rm+YoQXeU4PMJcrPXUdqovnOjr68vVq5cGffee290d3f/z7e/6aabor+//8hLX1/fmC+UD6fR\nNhehO46f7ijBYyy52esoQXfkpjlK0B0l+HyC3Ox1tIJRfefG5s2bY8eOHbFw4cIjrzt06FBs3Lgx\nbr311hgcHIz29vYj/62rqyu6urrG72r50BltcxG64/jpjhI8xpKbvY4SdEdumqME3VGCzyfIzV5H\nKxjV4cZFF10UL7/88lGv+/a3vx1nnHFG/OAHPzgmWDhemqME3VGC7shNc5SgO3LTHCXojhJ0R26a\noxWM6nCjp6cnFixYcNTrpkyZEieffPIxr4fxoDlK0B0l6I7cNEcJuiM3zVGC7ihBd+SmOVrBqH7n\nBgAAAAAAQGmj+s6N97N+/fpxuAz44DRHCbqjBN2Rm+YoQXfkpjlK0B0l6I7cNEduvnMDAAAAAACo\nFIcbAAAAAABApTjcAAAAAAAAKsXhBgAAAAAAUCkONwAAAAAAgErpyDms0WhERMSBAweSzxoeHk4+\nY8SECXnOiPbv359lTkTE0NBQ8hkjH6ORLlIZWX9gYCDpnIi83R06dCjLnJzdpd4bRtZP3VzzjMOH\nDyeflePv64hcjR88eDDLnIj0jefqbmT9HHtDjsfxEYODg1nm5Nzr2traks8YuZ9c3eVoIlcLEfXc\n61LvDSPr53yMzfH3tk6PRyNy7uGp338j6+fa63L00N7ennzGiJx955LjuXfu7nI81+/oyPeloRwf\no4i8e92ePXuSrr93796IqNfXTur4vC7H+y3nrH379kVEvu5GOk8p19dvI/I9r8v5dcjUf29H1h9t\nc22NHJ+B/H+vv/56zJkzJ9c4KqKvry9mz56dbH3d8V6pm4vQHcey11GC7sjNYywl2OsoQXfkpjlK\n0B25jba5rIcbhw8fju3bt0dPT8+o/tXi7t27Y86cOdHX1xcnnHBCwivMo273EzG2e2o0GrFnz544\n7bTTkp6ejqU7H6PW18rNReguon73E9Ha3Wnu39zTv+kur7rdUys3F6G7iPrdT0Rrd6e5f6vbPY31\nfnSXV93uqW57XYSPURXUrTsfo9aXs7msP5ZqwoQJx3Xad8IJJ9TiAzyibvcTMfp7mjZtWsKr+bfj\n6c7HqPW1YnMRumtWt/uJaM3uNHc096S7Eup2T63YXITumtXtfiJaszvNHa1u9zSW+9FdfnW7p7rt\ndRE+RlVQt+58jFpfjub8QnEAAAAAAKBSHG4AAAAAAACV0v7Tn/70p6Uv4oNob2+PZcuWRUdH1p+k\nlUzd7ieifvdUt/uJqN891e1+Iup3T3W7n4j63VPd7ifCPVVB3e4non73VLf7iajfPdXtfiLqd091\nu5+I+t1T3e4nwj1VQd3uJ6J+91S3+4mo3z3V7X4i6ndPue4n6y8UBwAAAAAAOF5+LBUAAAAAAFAp\nDjcAAAAAAIBKcbgBAAAAAABUisMNAAAAAACgUhxuAAAAAAAAldLyhxu33357zJs3L7q7u2PhwoXx\n1FNPlb6kMVuzZk0sXrw4enp6YubMmXHFFVfE1q1bS1/WuFmzZk20tbXFqlWrSl/KcatLd3VvLqI+\n3dWluYj6d1eX5iJ0VyW6az11by6iPt3VpbmI+ndXl+YidFcldelOc9VRl+YidFclums9dW8uIk93\nLX24sW7duli1alXcfPPNsWXLlrjgggvi0ksvjW3btpW+tDHZsGFDrFixIp599tno7e2NgwcPxvLl\ny2NgYKD0pR23TZs2xdq1a+Pss88ufSnHrU7d1bm5iPp0V6fmIurdXV2ai9BdleiuNdW5uYj6dFen\n5iLq3V1dmovQXZXUpTvNVUddmovQXZXorjXVubmIjN01WtinP/3pxnXXXXfU684444zGD3/4w0JX\nNL527NjRiIjGhg0bSl/KcdmzZ09j/vz5jd7e3sbSpUsbK1euLH1Jx6XO3dWluUajXt3VublGoz7d\n1am5RkN3VaG76qhLc41Gvbqrc3ONRn26q1NzjYbuqqJO3WmuGurUXKOhu6rQXXXUpblGI293Lfud\nG0NDQ7F58+ZYvnz5Ua9fvnx5PPPMM4Wuanz19/dHRMT06dMLX8nxWbFiRVx++eVx8cUXl76U41b3\n7urSXER9uqt7cxH16a4uzUXorkp0Vx11aS6iPt3VvbmI+nRXl+YidFcldelOc9VRl+YidFcluquO\nujQXkbe7juQTxmjXrl1x6NChmDVr1lGvnzVrVrz11luFrmr8NBqNWL16dZx//vmxYMGC0pczZvff\nf3+88MILsWnTptKXMi7q3F1dmouoV3d1bi6iPt3VqbkI3VWF7qqjLs1F1Ku7OjcXUZ/u6tRchO6q\nok7daa4a6tRchO6qQnfVUZfmIvJ317KHGyPa2tqO+nOj0TjmdVV0ww03xEsvvRRPP/106UsZs76+\nvli5cmX8/ve/j+7u7tKXM67q2F0dmouob3d1bC6iHt3VtbkI3bUy3VVLHZqLqG93dWwuoh7d1bW5\nCN21srp2p7nWVdfmInTXynRXLXVoLqJMdy17uDFjxoxob28/5uRtx44dx5zQVc2NN94YDz/8cGzc\nuDFmz55d+nLGbPPmzbFjx45YuHDhkdcdOnQoNm7cGLfeemsMDg5Ge3t7wSscvbp2V5fmIurXXV2b\ni6hPd3VrLkJ3VaC76qhLcxH1666uzUXUp7u6NRehuyqoW3eaa311ay5Cd1Wgu+qoS3MRZbpr2d+5\n0dnZGQsXLoze3t6jXt/b2xvnnXdeoas6Po1GI2644Yb47W9/G0888UTMmzev9CUdl4suuihefvnl\nePHFF4+8LFq0KK666qp48cUXK7dJRtSvu7o1F1G/7urWXET9uqtbcxG6qwLdtb66NRdRv+7q1lxE\n/bqrW3MRuquCunWnudZXt+YidFcFumt9dWsuokx3LfudGxERq1evjmuuuSYWLVoUS5YsibVr18a2\nbdviuuuuK31pY7JixYq477774qGHHoqenp4jJ43Tpk2LSZMmFb660evp6Tnm58BNmTIlTj755Er/\nfLg6dVe35iLq2V2dmouoX3d1bC5Cd61Od62vbs1F1LO7OjUXUb/u6thchO5aXR2701xrq2NzEbpr\ndbprfXVrLqJQd40Wd9tttzXmzp3b6OzsbHzqU59qbNiwofQljVlEvO/LPffcU/rSxs3SpUsbK1eu\nLH0Zx60u3X0Ymms06tFdXZprND4c3dWhuUZDd1Wju9byYWiu0ahHd3VprtH4cHRXh+YaDd1VTR26\n01y11KG5RkN3VaO71vJhaK7RSN9dW6PRaIzPMQkAAAAAAEB6Lfs7NwAAAAAAAN6Pww0AAAAAAKBS\nHG4AAAAAAACV4nADAAAAAACoFIcbAAAAAABApTjcAAAAAAAAKsXhBgAAAAAAUCkONwAAAAAAgEpx\nuAEAAAAAAFSKww0AAAAAAKBSHG4AAAAAAACV8v8Ao/whyzOD1iMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe209e24ba8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(ncols=kernel_num, figsize=(2*kernel_num, 2))\n",
    "\n",
    "abs_max = np.max(np.fabs(data))\n",
    "print ('Minimal weight', np.amin(data))\n",
    "print ('Maximal weight', np.amax(data))\n",
    "print ('Absolute maximal weight', abs_max)\n",
    "for col in range(kernel_num):\n",
    "    axes[col].imshow(data[col, 0, :, :], vmin=-abs_max, vmax=abs_max, cmap=plt.cm.gray)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 4. Application on surface finish quality inspection\n",
    "\n",
    "* Surface finish is important to improve appearance as well as quality (rough surfaces wear more quickly).\n",
    "* The surface roughness is a good predictor of the mechanical interactions and can be due to a non-optimal manufacturing process.\n",
    "* Roughness is mostly measured using a manual comparison against a surface roughness gage, it's also possible to use a profilometer.\n",
    "* **Goal:** Automate surface finish quality inspection on aluminum pieces manufactured at Georgia Tech.\n",
    "* Replicate human judgement of good/bad surface finish => Image classification => CNN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "In the goal of moving towards smart manufacturing and fully automated machine shops, we looked into finding a way to automate surface finish quality inspection. We are trying to replicate human judgement without using surface roughness gage. The generated pieces are in aluminum produced from turning operations from a CNC machine. \n",
    "If a human can look at the surface and decide whether the surface is good or bad, can an algorithm do so based on a picture, mainly that now, most of us have good quality cameras on our devices ? \n",
    "This problem goes back to a binary image classification task. \n",
    "The rest of the tutorial will show you how we retrained a CNN to classify our images and reached --- % accuracy. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "from torch.optim import lr_scheduler\n",
    "from skimage import io, transform\n",
    "import torchvision\n",
    "from torchvision import models\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "\n",
    "plt.ion()   # interactive mode"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "We load our dataset that you can find publically available at .... . Our dataset has 570 pictures per class, divided as follows: 490 per class for training and 80 per class for validation. It is recommended to keep 10 to 30 % of the data as validation data in order to tune the model. Remember that in Machine Learning/Deep learning, the testing dataset is precious, so it should be different from the validation and the training. We used 70 pictures per class for Testing.  \n",
    "We do some pre-processsing on the images: Resize them to 256, then crop at center to get an image of size 224.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "# Data augmentation and normalization for training\n",
    "# Just normalization for validation and testing\n",
    "data_transforms = {\n",
    "    'train': transforms.Compose([\n",
    "        transforms.Resize(256),\n",
    "        transforms.CenterCrop(224),\n",
    "        transforms.RandomHorizontalFlip(),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'val': transforms.Compose([\n",
    "        transforms.Resize(256),\n",
    "        transforms.CenterCrop(224),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "    'test': transforms.Compose([\n",
    "        transforms.Resize(256),\n",
    "        transforms.CenterCrop(224),\n",
    "        transforms.ToTensor(),\n",
    "        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ]),\n",
    "}\n",
    "\n",
    "data_dir = 'data/surface_quality'\n",
    "image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),\n",
    "                                          data_transforms[x])\n",
    "                  for x in ['train', 'val','test']}\n",
    "dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=batch_num_size,\n",
    "                                             shuffle=True, num_workers=4)\n",
    "              for x in ['train', 'val','test']}\n",
    "dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val','test']}\n",
    "class_names = image_datasets['train'].classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## 4.1 Dataset:\n",
    "* Training: 490 per class\n",
    "* Validation: 80 per class\n",
    "* Testing: 70 per class "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Let's write a function to visualize the images that we using from our dataet. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "def imshow(batch, class_names=None, num_images=4):\n",
    "    plt.figure(figsize=(1.7*num_images, 1.7))\n",
    "    img, classes = batch\n",
    "    img_num = min(num_images, img.shape[0])\n",
    "\n",
    "    grid = torchvision.utils.make_grid(img[:img_num], nrow=img_num, padding=10, pad_value=10)\n",
    "    grid = grid.cpu().numpy().transpose((1, 2, 0))\n",
    "    mean = np.array([0.485, 0.456, 0.406])\n",
    "    std = np.array([0.229, 0.224, 0.225])\n",
    "    grid = std * grid + mean\n",
    "    grid = np.clip(grid, 0, 1)\n",
    "    plt.imshow(grid)\n",
    "    if class_names:\n",
    "        titles = [class_names[x] for x in classes[:img_num]]\n",
    "        plt.title(titles)\n",
    "    plt.pause(0.001)  # pause a bit so that plots are updated"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Now let's visualize 4 random images from the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAzgAAADCCAYAAACFbWdgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYCcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe209e57e10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "imshow(next(iter(dataloaders['train'])), class_names, num_images=6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Let's now write our training function: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "def train_model(model, criterion, optimizer, scheduler, train_loader, test_loader, num_epochs=25):\n",
    "    best_model_wts = model.state_dict()\n",
    "    best_acc = 0.0\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        print('Epoch {}/{}'.format(epoch, num_epochs - 1))\n",
    "        print('-' * 80)\n",
    "\n",
    "        scheduler.step()\n",
    "        train(model, epoch, criterion, optimizer, train_loader)\n",
    "        acc, loss = test(model, epoch, criterion, test_loader)\n",
    "        if acc > best_acc:\n",
    "            best_acc = acc\n",
    "            best_model_wts = model.state_dict()\n",
    "\n",
    "    print('Best val Acc: {:4f}'.format(best_acc))\n",
    "\n",
    "    # load best model weights\n",
    "    model.load_state_dict(best_model_wts)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "def visualize_model(model, dataset, class_names=None, num_images=6):\n",
    "    inputs, target = next(iter(dataset))\n",
    "    inputs, target = Variable(inputs), Variable(target)\n",
    "    if cuda_gpu:\n",
    "        inputs, labels = inputs.cuda(), target.cuda()\n",
    "        model.cuda()\n",
    "    outputs = model(inputs)\n",
    "    _, preds = torch.max(outputs.data, 1)\n",
    "    imshow((inputs.data, preds), class_names=class_names, num_images=num_images)\n",
    "    if class_names:\n",
    "        print ('Actual labels:\\n',[class_names[x] for x in target.data[:num_images]])\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "## Transfer Learning\n",
    "\n",
    "Training a deep network from scratch is tricky because of limited dataset and computationally expensive.\n",
    "\n",
    "**Transfer Learning** is a method of using a deep net pretrained on a larger dataset as feature extractor. We  need to change classification layers according to our dataset and retrain the network. For re-training we have two options: \n",
    "* Finetuning -- retrain all the layers.\n",
    "* **Fixed feature extractor** -- only retrain classifier (usually, last fully connected layers of the network).\n",
    "\n",
    "We used **ResNet18** which was trained on ImageNet, we set pretrained=True\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [],
   "source": [
    "model_ft = models.resnet18(pretrained=True)\n",
    "for param in model.parameters():\n",
    "    param.requires_grad = False\n",
    "num_ftrs = model_ft.fc.in_features\n",
    "model_ft.fc = nn.Linear(num_ftrs, 2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "Let's set our training parameters and fine-tune the last layer of the function.\n",
    "* Optimization Method: SGD\n",
    "* Loss function: CrossEntropyLoss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [],
   "source": [
    "cuda_gpu = torch.cuda.is_available()\n",
    "#cuda_gpu = False\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "\n",
    "# Observe that all parameters are being optimized\n",
    "optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, momentum=0.9)\n",
    "\n",
    "# Decay LR by a factor of 0.1 every 7 epochs\n",
    "exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.2126, Accuracy: 153/160 (96%)\n",
      "\n",
      "Epoch 1/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.1146, Accuracy: 155/160 (97%)\n",
      "\n",
      "Epoch 2/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.1029, Accuracy: 156/160 (98%)\n",
      "\n",
      "Epoch 3/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0888, Accuracy: 155/160 (97%)\n",
      "\n",
      "Epoch 4/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0620, Accuracy: 156/160 (98%)\n",
      "\n",
      "Epoch 5/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0703, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 6/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0638, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 7/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0670, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 8/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0482, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 9/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0483, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 10/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0536, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 11/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0475, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 12/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0555, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 13/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0545, Accuracy: 156/160 (98%)\n",
      "\n",
      "Epoch 14/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0689, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 15/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0563, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 16/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0528, Accuracy: 156/160 (98%)\n",
      "\n",
      "Epoch 17/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0555, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 18/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0570, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 19/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0516, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 20/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0505, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 21/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0461, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 22/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0563, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 23/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0555, Accuracy: 157/160 (98%)\n",
      "\n",
      "Epoch 24/24\n",
      "--------------------------------------------------------------------------------\n",
      "\n",
      "Test set: Average loss: 0.0526, Accuracy: 157/160 (98%)\n",
      "\n",
      "Best val Acc: 0.981250\n",
      "CPU times: user 30.4 s, sys: 40.3 s, total: 1min 10s\n",
      "Wall time: 1min 42s\n"
     ]
    }
   ],
   "source": [
    "%time model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler, \\\n",
    "                             dataloaders['train'], dataloaders['val'], num_epochs=25)\n",
    "# Save the trained model\n",
    "torch.save(model_ft, 'pretrained/resnet18_surface.t7')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Training Results: \n",
    "\n",
    "* Training on CPU took 1h 21min 55s with a test accuracy = 97.14 %\n",
    "* Training on GPU took 1min 47s with a test accuracy = 97.86 %"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Test set: Average loss: 0.1052, Accuracy: 136/140 (97%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "acc, loss = test(model_ft, 0, criterion, dataloaders['test'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAzgAAADCCAYAAACFbWdgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nenEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe20a6ed828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Actual labels:\n",
      " ['good_surface', 'bad_surface', 'bad_surface', 'bad_surface', 'good_surface', 'bad_surface']\n"
     ]
    }
   ],
   "source": [
    "visualize_model(model_ft, dataloaders['test'], num_images=6, class_names=class_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "  (relu): ReLU(inplace)\n",
      "  (maxpool): MaxPool2d(kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), dilation=(1, 1), ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "      (relu): ReLU(inplace)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AvgPool2d(kernel_size=7, stride=1, padding=0, ceil_mode=False, count_include_pad=True)\n",
      "  (fc): Linear(in_features=512, out_features=2, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(model_ft)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "# Future work\n",
    "\n",
    "1. Adding more variety to dataset\n",
    "2. Adding more classes for surface quality\n",
    "3. Propagating class labels back to pixels to aid visual quality inspection\n",
    "4. Write an Android application to run surface quality inspection on the phone directly in the machine shop "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "**Related Python-friendly deep learning frameworks.**\n",
    "* PyTorch\n",
    "* Caffe2\n",
    "* Keras\n",
    "* Tensorflow\n",
    "* MXNet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "skip"
    }
   },
   "source": [
    "**Acknowledgements.** We'd like to thank [Facebook Research's Caffe2 team](https://caffe2.ai/) for their generous support of some of our research, which motivated us to learn PyTorch and develop this tutorial."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "fragment"
    }
   },
   "source": [
    "# Thank You !"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "celltoolbar": "Slideshow",
  "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}