Models/VAE/Demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A Demo of VAE for Handwritten Number Generation\n",
    "-------\n",
    "\n",
    "## Variational Autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T12:44:44.563387Z",
     "start_time": "2020-07-13T12:44:44.249180Z"
    }
   },
   "outputs": [],
   "source": [
    "!python train_VAE.py #skip this if the model has been trained before"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T13:04:43.076736Z",
     "start_time": "2020-07-13T13:04:43.068736Z"
    }
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "from matplotlib import pyplot as plt\n",
    "from matplotlib import cm\n",
    "from model import VAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T13:05:10.269826Z",
     "start_time": "2020-07-13T13:05:09.507782Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "VAE(\n",
       "  (encoder): Encoder(\n",
       "    (encode): Sequential(\n",
       "      (0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1))\n",
       "      (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (2): ReLU(inplace)\n",
       "      (3): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1))\n",
       "      (4): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (5): ReLU(inplace)\n",
       "      (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "      (7): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1))\n",
       "      (8): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (9): ReLU(inplace)\n",
       "      (10): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1))\n",
       "      (11): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (12): ReLU(inplace)\n",
       "      (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "      (14): Flatten()\n",
       "      (15): MLP(\n",
       "        (mlp): Sequential(\n",
       "          (Linear_0): Linear(in_features=576, out_features=256, bias=True)\n",
       "          (BatchNorm_0): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_0): ReLU(inplace)\n",
       "          (Linear_1): Linear(in_features=256, out_features=128, bias=True)\n",
       "          (BatchNorm_1): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_1): ReLU(inplace)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (calc_mean): MLP(\n",
       "      (mlp): Sequential(\n",
       "        (Linear_0): Linear(in_features=128, out_features=64, bias=True)\n",
       "        (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (ReLU_0): ReLU(inplace)\n",
       "        (Linear_1): Linear(in_features=64, out_features=4, bias=True)\n",
       "      )\n",
       "    )\n",
       "    (calc_logvar): MLP(\n",
       "      (mlp): Sequential(\n",
       "        (Linear_0): Linear(in_features=128, out_features=64, bias=True)\n",
       "        (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (ReLU_0): ReLU(inplace)\n",
       "        (Linear_1): Linear(in_features=64, out_features=4, bias=True)\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (decoder): Decoder(\n",
       "    (decode): Sequential(\n",
       "      (0): MLP(\n",
       "        (mlp): Sequential(\n",
       "          (Linear_0): Linear(in_features=4, out_features=64, bias=True)\n",
       "          (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_0): ReLU(inplace)\n",
       "          (Linear_1): Linear(in_features=64, out_features=128, bias=True)\n",
       "          (BatchNorm_1): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_1): ReLU(inplace)\n",
       "          (Linear_2): Linear(in_features=128, out_features=256, bias=True)\n",
       "          (BatchNorm_2): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_2): ReLU(inplace)\n",
       "          (Linear_3): Linear(in_features=256, out_features=784, bias=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Sigmoid()\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "\n",
    "net = VAE((1, 28, 28), nhid = 4)\n",
    "checkpoint = torch.load(\"VAE.pt\", map_location = device)\n",
    "net.load_state_dict(checkpoint[\"net\"])\n",
    "net.to(device)\n",
    "net.eval()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T12:51:23.609246Z",
     "start_time": "2020-07-13T12:51:23.592245Z"
    }
   },
   "source": [
    "Generate a single image."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T13:05:14.167635Z",
     "start_time": "2020-07-13T13:05:13.556600Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2b88aaa1320>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    x = net.generate()\n",
    "plt.imshow(x.squeeze(0).cpu().numpy(), cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T11:57:47.308498Z",
     "start_time": "2020-07-13T11:57:47.292898Z"
    }
   },
   "source": [
    "We illustrate 15 generated fake images here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-07-13T13:05:49.182767Z",
     "start_time": "2020-07-13T13:05:43.275036Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fake images\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 15 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    x = net.generate(batch_size=15)\n",
    "    \n",
    "print(\"fake images\")\n",
    "for i in range(15):\n",
    "    plt.subplot(3, 5, i+1)\n",
    "    plt.axis(\"off\")\n",
    "    plt.imshow(x[i].squeeze(0).cpu().numpy(), cm.gray)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "real images\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 15 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import torchvision\n",
    "train_data = torchvision.datasets.MNIST(root='../../Datasets', train=True, download=True, transform=torchvision.transforms.ToTensor())\n",
    "print(\"real images\")\n",
    "for i in range(15):\n",
    "    plt.subplot(3, 5, i+1)\n",
    "    plt.axis(\"off\")\n",
    "    plt.imshow(train_data[i][0].squeeze(0).cpu().numpy(), cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conditional Variational Autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!python train_cVAE.py #skip this if the model has been trained before"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from model import cVAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "cVAE(\n",
       "  (encoder): Encoder(\n",
       "    (encode): Sequential(\n",
       "      (0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1))\n",
       "      (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (2): ReLU(inplace)\n",
       "      (3): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1))\n",
       "      (4): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (5): ReLU(inplace)\n",
       "      (6): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "      (7): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1))\n",
       "      (8): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (9): ReLU(inplace)\n",
       "      (10): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1))\n",
       "      (11): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (12): ReLU(inplace)\n",
       "      (13): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "      (14): Flatten()\n",
       "      (15): MLP(\n",
       "        (mlp): Sequential(\n",
       "          (Linear_0): Linear(in_features=576, out_features=256, bias=True)\n",
       "          (BatchNorm_0): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_0): ReLU(inplace)\n",
       "          (Linear_1): Linear(in_features=256, out_features=128, bias=True)\n",
       "          (BatchNorm_1): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_1): ReLU(inplace)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (calc_mean): MLP(\n",
       "      (mlp): Sequential(\n",
       "        (Linear_0): Linear(in_features=144, out_features=64, bias=True)\n",
       "        (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (ReLU_0): ReLU(inplace)\n",
       "        (Linear_1): Linear(in_features=64, out_features=2, bias=True)\n",
       "      )\n",
       "    )\n",
       "    (calc_logvar): MLP(\n",
       "      (mlp): Sequential(\n",
       "        (Linear_0): Linear(in_features=144, out_features=64, bias=True)\n",
       "        (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (ReLU_0): ReLU(inplace)\n",
       "        (Linear_1): Linear(in_features=64, out_features=2, bias=True)\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (decoder): Decoder(\n",
       "    (decode): Sequential(\n",
       "      (0): MLP(\n",
       "        (mlp): Sequential(\n",
       "          (Linear_0): Linear(in_features=18, out_features=64, bias=True)\n",
       "          (BatchNorm_0): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_0): ReLU(inplace)\n",
       "          (Linear_1): Linear(in_features=64, out_features=128, bias=True)\n",
       "          (BatchNorm_1): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_1): ReLU(inplace)\n",
       "          (Linear_2): Linear(in_features=128, out_features=256, bias=True)\n",
       "          (BatchNorm_2): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (ReLU_2): ReLU(inplace)\n",
       "          (Linear_3): Linear(in_features=256, out_features=784, bias=True)\n",
       "        )\n",
       "      )\n",
       "      (1): Sigmoid()\n",
       "    )\n",
       "  )\n",
       "  (label_embedding): Embedding(10, 16)\n",
       ")"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "\n",
    "net = cVAE((1, 28, 28), nclass = 10, nhid = 2, ncond = 16)\n",
    "checkpoint = torch.load(\"cVAE.pt\", map_location = device)\n",
    "net.load_state_dict(checkpoint[\"net\"])\n",
    "net.to(device)\n",
    "net.eval()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an illustration of a generated handwritten `number`. You can assign an arbitrary digit from 0~9 for `number`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "number = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2b8dfb189b0>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    x = net.generate(number)\n",
    "plt.imshow(x.squeeze(0).cpu().numpy(), cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here are 40 samples generated by the model, from which we can see the variation of its generation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 50 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    x = net.generate(torch.tensor([list(range(10))] * 5).clone().view(-1))\n",
    "    \n",
    "for i in range(50):\n",
    "    plt.subplot(5, 10, i+1)\n",
    "    plt.axis(\"off\")\n",
    "    plt.imshow(x[i].squeeze(0).cpu().numpy(), cm.gray)"
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}