deep_learning_pytorch/Part 8 - Transfer Learning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transfer Learning\n",
    "\n",
    "In this notebook, you'll learn how to use pre-trained networks to solved challenging problems in computer vision. Specifically, you'll use networks trained on [ImageNet](http://www.image-net.org/) [available from torchvision](http://pytorch.org/docs/0.3.0/torchvision/models.html). \n",
    "\n",
    "ImageNet is a massive dataset with over 1 million labeled images in 1000 categories. It's used to train deep neural networks using an architecture called convolutional layers. I'm not going to get into the details of convolutional networks here, but if you want to learn more about them, please [watch this](https://www.youtube.com/watch?v=2-Ol7ZB0MmU).\n",
    "\n",
    "Once trained, these models work astonishingly well as feature detectors for images they weren't trained on. Using a pre-trained network on images not in the training set is called transfer learning. Here we'll use transfer learning to train a network that can classify our cat and dog photos with near perfect accuracy.\n",
    "\n",
    "With `torchvision.models` you can download these pre-trained networks and use them in your applications. We'll include `models` in our imports now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch import optim\n",
    "import torch.nn.functional as F\n",
    "from torchvision import datasets, transforms, models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Most of the pretrained models require the input to be 224x224 images. Also, we'll need to match the normalization used when the models were trained. Each color channel was normalized separately, the means are `[0.485, 0.456, 0.406]` and the standard deviations are `[0.229, 0.224, 0.225]`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dir = 'Cat_Dog_data'\n",
    "\n",
    "# TODO: Define transforms for the training data and testing data\n",
    "# Define a transform to normalize the data\n",
    "train_transforms = transforms.Compose([transforms.Resize((224,224)),                                      \n",
    "                                       transforms.ToTensor(),\n",
    "                                       transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),\n",
    "                                                                    \n",
    "                             ])\n",
    "\n",
    "# Define a transform to normalize the data\n",
    "test_transforms = transforms.Compose([transforms.Resize((224,224)),                                       \n",
    "                                      transforms.ToTensor(),\n",
    "                                      transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),\n",
    "                                       \n",
    "                             ])\n",
    "\n",
    "# Pass transforms in here, then run the next cell to see how the transforms look\n",
    "train_data = datasets.ImageFolder( data_dir + '/train', transform=train_transforms)\n",
    "test_data = datasets.ImageFolder(data_dir + '/test', transform=test_transforms)\n",
    "\n",
    "trainloader = torch.utils.data.DataLoader(train_data, batch_size=128, shuffle=True,num_workers=32)\n",
    "testloader = torch.utils.data.DataLoader(test_data, batch_size=64,num_workers=32)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can load in a model such as [DenseNet](http://pytorch.org/docs/0.3.0/torchvision/models.html#id5). Let's print out the model architecture so we can see what's going on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/rh/rh-python35/root/usr/lib/python3.5/site-packages/torchvision/models/densenet.py:212: UserWarning: nn.init.kaiming_normal is now deprecated in favor of nn.init.kaiming_normal_.\n",
      "  nn.init.kaiming_normal(m.weight.data)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "DenseNet(\n",
       "  (features): Sequential(\n",
       "    (conv0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "    (norm0): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (relu0): ReLU(inplace)\n",
       "    (pool0): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "    (denseblock1): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(96, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "    )\n",
       "    (transition1): _Transition(\n",
       "      (norm): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace)\n",
       "      (conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock2): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "    )\n",
       "    (transition2): _Transition(\n",
       "      (norm): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace)\n",
       "      (conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock3): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer13): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(640, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer14): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(672, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer15): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(704, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer16): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(736, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer17): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer18): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(800, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer19): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(832, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer20): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(864, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer21): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer22): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer23): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer24): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "    )\n",
       "    (transition3): _Transition(\n",
       "      (norm): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace)\n",
       "      (conv): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock4): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(640, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(672, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(704, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(736, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(800, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(832, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(864, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer13): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(896, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer14): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(928, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer15): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(960, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "      (denselayer16): _DenseLayer(\n",
       "        (norm1): BatchNorm2d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu1): ReLU(inplace)\n",
       "        (conv1): Conv2d(992, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (relu2): ReLU(inplace)\n",
       "        (conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      )\n",
       "    )\n",
       "    (norm5): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  )\n",
       "  (classifier): Linear(in_features=1024, out_features=1000, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = models.densenet121(pretrained=True)\n",
    "model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This model is built out of two main parts, the features and the classifier. The features part is a stack of convolutional layers and overall works as a feature detector that can be fed into a classifier. The classifier part is a single fully-connected layer `(classifier): Linear(in_features=1024, out_features=1000)`. This layer was trained on the ImageNet dataset, so it won't work for our specific problem. That means we need to replace the classifier, but the features will work perfectly on their own. In general, I think about pre-trained networks as amazingly good feature detectors that can be used as the input for simple feed-forward classifiers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Freeze parameters so we don't backprop through them\n",
    "for param in model.parameters():\n",
    "    param.requires_grad = False\n",
    "\n",
    "from collections import OrderedDict\n",
    "classifier = nn.Sequential(OrderedDict([\n",
    "                          ('fc1', nn.Linear(1024, 500)),\n",
    "                          ('relu', nn.ReLU()),\n",
    "                          ('fc2', nn.Linear(500, 2)),\n",
    "                          ('output', nn.LogSoftmax(dim=1))\n",
    "                          ]))\n",
    "    \n",
    "model.classifier = classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With our model built, we need to train the classifier. However, now we're using a **really deep** neural network. If you try to train this on a CPU like normal, it will take a long, long time. Instead, we're going to use the GPU to do the calculations. The linear algebra computations are done in parallel on the GPU leading to 100x increased training speeds. It's also possible to train on multiple GPUs, further decreasing training time.\n",
    "\n",
    "PyTorch, along with pretty much every other deep learning framework, uses [CUDA](https://developer.nvidia.com/cuda-zone) to efficiently compute the forward and backwards passes on the GPU. In PyTorch, you move your model parameters and other tensors to the GPU memory using `model.to('cuda')`. You can move them back from the GPU with `model.to('cpu')` which you'll commonly do when you need to operate on the network output outside of PyTorch. As a demonstration of the increased speed, I'll compare how long it takes to perform a forward and backward pass with and without a GPU."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "\n",
    "for device in ['cpu']:\n",
    "\n",
    "    criterion = nn.NLLLoss()\n",
    "    # Only train the classifier parameters, feature parameters are frozen\n",
    "    optimizer = optim.Adam(model.classifier.parameters(), lr=0.001)\n",
    "\n",
    "    model.to(device)\n",
    "    count=0\n",
    "    for ii, (inputs, labels) in enumerate(trainloader):\n",
    "        \n",
    "        # Move input and label tensors to the GPU\n",
    "        inputs, labels = inputs.to(device), labels.to(device)\n",
    "        start = time.time()\n",
    "\n",
    "        outputs = model.forward(inputs)\n",
    "        loss = criterion(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        if ii==3:\n",
    "            break\n",
    "\n",
    "    time_taken = (time.time() - start)/3\n",
    "    print(\"Device = {}; Time per batch: {} seconds\".format(device,time_taken))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can write device agnostic code which will automatically use CUDA if it's enabled like so:\n",
    "```python\n",
    "# at beginning of the script\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "...\n",
    "\n",
    "# then whenever you get a new Tensor or Module\n",
    "# this won't copy if they are already on the desired device\n",
    "input = data.to(device)\n",
    "model = MyModule(...).to(device)\n",
    "```\n",
    "\n",
    "From here, I'll let you finish training the model. The process is the same as before except now your model is much more powerful. You should get better than 95% accuracy easily.\n",
    "\n",
    ">**Exercise:** Train a pretrained models to classify the cat and dog images. Continue with the DenseNet model, or try ResNet, it's also a good model to try out first. Make sure you are only training the classifier and the parameters for the features part are frozen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.NLLLoss()\n",
    "optimizer = optim.Adam(model.classifier.parameters(), lr=0.001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Implement a function for the validation pass\n",
    "def validation(model, testloader, criterion):\n",
    "    test_loss = 0\n",
    "    accuracy = 0\n",
    "    for images, labels in testloader:\n",
    "\n",
    "        output = model.forward(images)\n",
    "        test_loss += criterion(output, labels).item()\n",
    "\n",
    "        ps = torch.exp(output)\n",
    "        equality = (labels.data == ps.max(dim=1)[1])\n",
    "        accuracy += equality.type(torch.FloatTensor).mean()\n",
    "    \n",
    "    return test_loss, accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1/1..  Training Loss: 4.105..  Test Loss: 10.862..  Test Accuracy: 0.489\n",
      "Epoch: 1/1..  Training Loss: 4.860..  Test Loss: 1.645..  Test Accuracy: 0.930\n"
     ]
    }
   ],
   "source": [
    "# TODO: Train a model with a pre-trained network\n",
    "import time\n",
    "\n",
    "epoch = 1\n",
    "steps = 0\n",
    "running_loss = 0\n",
    "print_every =  80\n",
    "\n",
    "for e in range(epoch):\n",
    "    model.train()\n",
    "    \n",
    "    for images, labels in trainloader :\n",
    "        steps +=1\n",
    "        \n",
    "        outputs = model.forward(images)\n",
    "        loss =  criterion(outputs,labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        running_loss += loss.item()\n",
    "        \n",
    "        if steps % print_every == 0:\n",
    "            # Make sure network is in eval mode for inference\n",
    "            model.eval()\n",
    "            \n",
    "            # Turn off gradients for validation, saves memory and computations\n",
    "            with torch.no_grad():\n",
    "                test_loss, accuracy = validation(model, testloader, criterion)\n",
    "                \n",
    "            print(\"Epoch: {}/{}.. \".format(e+1, epoch),\n",
    "                  \"Training Loss: {:.3f}.. \".format(running_loss/print_every),\n",
    "                  \"Test Loss: {:.3f}.. \".format(test_loss/len(testloader)),\n",
    "                  \"Test Accuracy: {:.3f}\".format(accuracy/len(testloader)))\n",
    "            \n",
    "            running_loss = 0\n",
    "            \n",
    "            # Make sure training is back on\n",
    "            model.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}