Visualizing model predictions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualizing model predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For displaying images, we will be using `matplotlib`, which works wonderfully even from within Jupyter Notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "from matplotlib.pyplot import imshow\n",
    "\n",
    "def show_images(images):\n",
    "    # Convert grayscale image into RGB channels.\n",
    "    images = images.expand(-1, 3, -1, -1)\n",
    "    \n",
    "    # Make a grid view image of this batch.\n",
    "    images = torchvision.utils.make_grid(images)\n",
    "    \n",
    "    # Unnormalize.\n",
    "    images = images / 2 + 0.5\n",
    "    \n",
    "    # Convert PyTorch tensor to a NumPy array.\n",
    "    images = images.numpy().transpose([1, 2, 0])\n",
    "    \n",
    "    # Display images.\n",
    "    imshow(images)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's fetch 8 images from the test dataset and show them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from main import get_data\n",
    "\n",
    "data_train, data_val, data_test = get_data(batch_size=8, num_workers=0)\n",
    "inputs, targets = iter(data_test).next()\n",
    "\n",
    "show_images(inputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And the actual labels for each image:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([9, 2, 1, 1, 6, 1, 4, 6])\n"
     ]
    }
   ],
   "source": [
    "print(targets)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Oops! The labels are in numbers, of course. They are the label indices, and fortunately, we can find the translated human words from https://github.com/zalandoresearch/fashion-mnist."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Truth:\n",
      "   ankle boot pullover trouser trouser shirt trouser coat shirt\n"
     ]
    }
   ],
   "source": [
    "LABELS = ('t-shirt/top', 'trouser', 'pullover', 'dress', 'coat',\n",
    "          'sandal', 'shirt', 'sneaker', 'bag', 'ankle boot')\n",
    "\n",
    "print('Truth:\\n  ', ' '.join(LABELS[i] for i in targets))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Make predictions with the trained model\n",
    "\n",
    "First, let's build the model and load the trained state from a checkpoint file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from main import get_model, load_checkpoint\n",
    "\n",
    "model = get_model()\n",
    "load_checkpoint('checkpoints/29.pth', model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The magic happens here! Run the inputs from the test dataset through the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[29.7212, 18.2632, 26.6118,  ..., -0.7714, -1.4926, -0.2872],\n",
      "        [32.5628, 13.8593, 50.0357,  ...,  1.2142, -0.3287, -1.0400],\n",
      "        [47.4349, 81.4266, 40.4236,  ..., -1.4161,  0.2404,  1.7200],\n",
      "        ...,\n",
      "        [36.5342, 53.4745, 32.0289,  ..., -0.5849, -0.2114,  2.1639],\n",
      "        [30.1944, 19.1829, 44.3080,  ...,  1.1572,  0.0380,  1.0208],\n",
      "        [33.7547, 16.3837, 31.8395,  ...,  1.3180, -0.2662, -0.1007]],\n",
      "       grad_fn=<ThAddmmBackward>)\n"
     ]
    }
   ],
   "source": [
    "outputs = model(inputs)\n",
    "\n",
    "print(outputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's pause for a bit; you'll notice that there is a weird `grad_fn` attached to our tensor. That is the place where PyTorch keeps the history for automatic gradient calculation. If you don't need them, you can tell PyTorch so to save up memory:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[29.7212, 18.2632, 26.6118,  ..., -0.7714, -1.4926, -0.2872],\n",
      "        [32.5628, 13.8593, 50.0357,  ...,  1.2142, -0.3287, -1.0400],\n",
      "        [47.4349, 81.4266, 40.4236,  ..., -1.4161,  0.2404,  1.7200],\n",
      "        ...,\n",
      "        [36.5342, 53.4745, 32.0289,  ..., -0.5849, -0.2114,  2.1639],\n",
      "        [30.1944, 19.1829, 44.3080,  ...,  1.1572,  0.0380,  1.0208],\n",
      "        [33.7547, 16.3837, 31.8395,  ...,  1.3180, -0.2662, -0.1007]])\n"
     ]
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    outputs = model(inputs)\n",
    "\n",
    "print(outputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bye-bye, `grad_fn`!\n",
    "\n",
    "In the above tensor, each row corresponds to an image, and each column denotes the predicted likelihood for each labels. For our use, we need to get the indices of the column with the highest likelihood, which is easy in PyTorch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([9, 2, 1, 1, 6, 1, 4, 6])\n"
     ]
    }
   ],
   "source": [
    "_, predictions = torch.max(outputs, 1)\n",
    "\n",
    "print(predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's turn those numbers into words!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction:\n",
      "   ankle boot pullover trouser trouser shirt trouser coat shirt\n"
     ]
    }
   ],
   "source": [
    "print('Prediction:\\n  ', ' '.join(LABELS[i] for i in predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## All-in-one"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_images(inputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Truth:\n",
      "   ankle boot pullover trouser trouser shirt trouser coat shirt\n",
      "Prediction:\n",
      "   ankle boot pullover trouser trouser shirt trouser coat shirt\n"
     ]
    }
   ],
   "source": [
    "print('Truth:\\n  ', ' '.join(LABELS[i] for i in targets))\n",
    "print('Prediction:\\n  ', ' '.join(LABELS[i] for i in predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (pytorch-v0.4.1)",
   "language": "python",
   "name": "pytorch-v0.4.1"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}