nb/cvae/LabeledCVAE.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Labeled CVAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "19d0f047",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.utils.data import DataLoader, SubsetRandomSampler\n",
    "from torchvision.transforms.functional import to_pil_image\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load pallets library from project root\n",
    "import sys, os\n",
    "sys.path.append(os.path.abspath(os.path.join(os.getcwd(), '..', '..')))\n",
    "\n",
    "from pallets import (\n",
    "    images as I,\n",
    "    datasets as DS,\n",
    "    models as M,\n",
    "    logging as L\n",
    ")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "844bbd48",
   "metadata": {},
   "source": [
    "# Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9b8ec428",
   "metadata": {},
   "outputs": [],
   "source": [
    "SAVE_NAME = 'cvae.labeled'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "ae1b5c2f",
   "metadata": {},
   "outputs": [],
   "source": [
    "USE_GPU = True\n",
    "LOG_LEVEL = 'INFO'\n",
    "LABELS_FILE = 'pallets_labels.json'\n",
    "\n",
    "TEST_SIZE = 1000\n",
    "EPOCHS = 50\n",
    "LR = 1e-03\n",
    "BATCH_SIZE = 32\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Logger pallets (INFO)>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# To GPU, or not to GPU\n",
    "device = M.get_device(require_gpu=USE_GPU)\n",
    "\n",
    "# Logging\n",
    "L.init_logger(level=LOG_LEVEL, notebook=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prepare Datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "35fef1c0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO | loading punk images\n",
      "INFO | - image 0\n",
      "INFO | - image 1000\n",
      "INFO | - image 2000\n",
      "INFO | - image 3000\n",
      "INFO | - image 4000\n",
      "INFO | - image 5000\n",
      "INFO | - image 6000\n",
      "INFO | - image 7000\n",
      "INFO | - image 8000\n",
      "INFO | - image 9000\n",
      "INFO | punk images complete\n",
      "INFO | loading punk labels: pallets_labels.json\n"
     ]
    }
   ],
   "source": [
    "all_colors = I.get_punk_colors()\n",
    "mapper = DS.ColorOneHotMapper(all_colors)\n",
    "dataset = DS.OneHotCPunksDataset(\n",
    "    mapper, labels_file=LABELS_FILE, test_size=TEST_SIZE\n",
    ")\n",
    "\n",
    "# Uncomment to put entire dataset in GPU RAM\n",
    "# dataset = DS.FastOneHotCPunksDataset(\n",
    "#     device, mapper, labels_file=LABELS_FILE, test_size=TEST_SIZE\n",
    "# )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_sampler = SubsetRandomSampler(dataset.train_idx)\n",
    "test_sampler = SubsetRandomSampler(dataset.test_idx)\n",
    "\n",
    "train_loader = DataLoader(dataset, batch_size=BATCH_SIZE, sampler=train_sampler)\n",
    "test_loader = DataLoader(dataset, batch_size=BATCH_SIZE, sampler=test_sampler)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b39e0222",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7e4d5fd0",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_dim = 24 * 24 * 222\n",
    "hidden_dim = 576\n",
    "latent_dim = 32\n",
    "classes_dim = len(dataset._label_keys)  # 92\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9521e7fe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO | model: pallets.models.cvae.LabeledCVAE\n",
      "INFO | criterion: pallets.models.cvae.Loss\n",
      "INFO | learn rate: 0.001\n",
      "INFO | epochs: 50\n",
      "INFO | epoch 1 (  0%) loss: 1423593.125000\n",
      "INFO | epoch 1 ( 35%) loss: 88232.447633\n",
      "INFO | epoch 1 ( 70%) loss: 49998.685532\n",
      "INFO | epoch 1 (100%) loss: 37908.333248\n",
      "INFO | epoch 1 (test) loss: 6926.361279\n",
      "INFO | epoch 2 (  0%) loss: 6745.611328\n",
      "INFO | epoch 2 ( 35%) loss: 6538.608302\n",
      "INFO | epoch 2 ( 70%) loss: 5885.737992\n",
      "INFO | epoch 2 (100%) loss: 5409.353846\n",
      "INFO | epoch 2 (test) loss: 3903.990690\n",
      "INFO | epoch 3 (  0%) loss: 3624.968750\n",
      "INFO | epoch 3 ( 35%) loss: 3598.175984\n",
      "INFO | epoch 3 ( 70%) loss: 3366.212178\n",
      "INFO | epoch 3 (100%) loss: 3210.196091\n",
      "INFO | epoch 3 (test) loss: 2801.327785\n",
      "INFO | epoch 4 (  0%) loss: 2804.128418\n",
      "INFO | epoch 4 ( 35%) loss: 2578.875919\n",
      "INFO | epoch 4 ( 70%) loss: 2449.443076\n",
      "INFO | epoch 4 (100%) loss: 2357.474783\n",
      "INFO | epoch 4 (test) loss: 2153.213619\n",
      "INFO | epoch 5 (  0%) loss: 2202.340820\n",
      "INFO | epoch 5 ( 35%) loss: 1958.569778\n",
      "INFO | epoch 5 ( 70%) loss: 1892.043015\n",
      "INFO | epoch 5 (100%) loss: 1840.395952\n",
      "INFO | epoch 5 (test) loss: 1751.493816\n",
      "INFO | epoch 6 (  0%) loss: 1624.591309\n",
      "INFO | epoch 6 ( 35%) loss: 1595.367297\n",
      "INFO | epoch 6 ( 70%) loss: 1549.535858\n",
      "INFO | epoch 6 (100%) loss: 1508.776043\n",
      "INFO | epoch 6 (test) loss: 1503.972304\n",
      "INFO | epoch 7 (  0%) loss: 1503.257568\n",
      "INFO | epoch 7 ( 35%) loss: 1353.744496\n",
      "INFO | epoch 7 ( 70%) loss: 1317.228660\n",
      "INFO | epoch 7 (100%) loss: 1285.220787\n",
      "INFO | epoch 7 (test) loss: 1249.438839\n",
      "INFO | epoch 8 (  0%) loss: 1270.046875\n",
      "INFO | epoch 8 ( 35%) loss: 1156.732979\n",
      "INFO | epoch 8 ( 70%) loss: 1128.621205\n",
      "INFO | epoch 8 (100%) loss: 1112.047168\n",
      "INFO | epoch 8 (test) loss: 1083.952080\n",
      "INFO | epoch 9 (  0%) loss: 1010.633606\n",
      "INFO | epoch 9 ( 35%) loss: 1015.082472\n",
      "INFO | epoch 9 ( 70%) loss: 986.145595\n",
      "INFO | epoch 9 (100%) loss: 966.514770\n",
      "INFO | epoch 9 (test) loss: 964.279748\n",
      "INFO | epoch 10 (  0%) loss: 987.024536\n",
      "INFO | epoch 10 ( 35%) loss: 889.271629\n",
      "INFO | epoch 10 ( 70%) loss: 874.473702\n",
      "INFO | epoch 10 (100%) loss: 857.283829\n",
      "INFO | epoch 10 (test) loss: 852.884037\n",
      "INFO | epoch 11 (  0%) loss: 813.513184\n",
      "INFO | epoch 11 ( 35%) loss: 793.097092\n",
      "INFO | epoch 11 ( 70%) loss: 783.527383\n",
      "INFO | epoch 11 (100%) loss: 767.677538\n",
      "INFO | epoch 11 (test) loss: 778.400311\n",
      "INFO | epoch 12 (  0%) loss: 697.741821\n",
      "INFO | epoch 12 ( 35%) loss: 712.583953\n",
      "INFO | epoch 12 ( 70%) loss: 700.236983\n",
      "INFO | epoch 12 (100%) loss: 691.994792\n",
      "INFO | epoch 12 (test) loss: 681.602884\n",
      "INFO | epoch 13 (  0%) loss: 662.583130\n",
      "INFO | epoch 13 ( 35%) loss: 643.752205\n",
      "INFO | epoch 13 ( 70%) loss: 632.602151\n",
      "INFO | epoch 13 (100%) loss: 623.880382\n",
      "INFO | epoch 13 (test) loss: 621.008989\n",
      "INFO | epoch 14 (  0%) loss: 612.496948\n",
      "INFO | epoch 14 ( 35%) loss: 575.534647\n",
      "INFO | epoch 14 ( 70%) loss: 574.567179\n",
      "INFO | epoch 14 (100%) loss: 566.479248\n",
      "INFO | epoch 14 (test) loss: 563.400029\n",
      "INFO | epoch 15 (  0%) loss: 578.859741\n",
      "INFO | epoch 15 ( 35%) loss: 531.553232\n",
      "INFO | epoch 15 ( 70%) loss: 527.001037\n",
      "INFO | epoch 15 (100%) loss: 517.466317\n",
      "INFO | epoch 15 (test) loss: 532.165802\n",
      "INFO | epoch 16 (  0%) loss: 464.354431\n",
      "INFO | epoch 16 ( 35%) loss: 481.348696\n",
      "INFO | epoch 16 ( 70%) loss: 476.008826\n",
      "INFO | epoch 16 (100%) loss: 469.466468\n",
      "INFO | epoch 16 (test) loss: 474.755593\n",
      "INFO | epoch 17 (  0%) loss: 462.633911\n",
      "INFO | epoch 17 ( 35%) loss: 437.593804\n",
      "INFO | epoch 17 ( 70%) loss: 430.927656\n",
      "INFO | epoch 17 (100%) loss: 426.590969\n",
      "INFO | epoch 17 (test) loss: 433.493756\n",
      "INFO | epoch 18 (  0%) loss: 384.860931\n",
      "INFO | epoch 18 ( 35%) loss: 407.335699\n",
      "INFO | epoch 18 ( 70%) loss: 399.077353\n",
      "INFO | epoch 18 (100%) loss: 393.285407\n",
      "INFO | epoch 18 (test) loss: 410.833937\n",
      "INFO | epoch 19 (  0%) loss: 369.829041\n",
      "INFO | epoch 19 ( 35%) loss: 367.973700\n",
      "INFO | epoch 19 ( 70%) loss: 366.323484\n",
      "INFO | epoch 19 (100%) loss: 363.608412\n",
      "INFO | epoch 19 (test) loss: 378.902534\n",
      "INFO | epoch 20 (  0%) loss: 331.694824\n",
      "INFO | epoch 20 ( 35%) loss: 338.923622\n",
      "INFO | epoch 20 ( 70%) loss: 335.263673\n",
      "INFO | epoch 20 (100%) loss: 329.626902\n",
      "INFO | epoch 20 (test) loss: 341.378279\n",
      "INFO | epoch 21 (  0%) loss: 321.057251\n",
      "INFO | epoch 21 ( 35%) loss: 323.793287\n",
      "INFO | epoch 21 ( 70%) loss: 316.992491\n",
      "INFO | epoch 21 (100%) loss: 310.902686\n",
      "INFO | epoch 21 (test) loss: 319.931873\n",
      "INFO | epoch 22 (  0%) loss: 274.352112\n",
      "INFO | epoch 22 ( 35%) loss: 292.504896\n",
      "INFO | epoch 22 ( 70%) loss: 293.752615\n",
      "INFO | epoch 22 (100%) loss: 288.550342\n",
      "INFO | epoch 22 (test) loss: 343.671114\n",
      "INFO | epoch 23 (  0%) loss: 242.184067\n",
      "INFO | epoch 23 ( 35%) loss: 269.326588\n",
      "INFO | epoch 23 ( 70%) loss: 270.246749\n",
      "INFO | epoch 23 (100%) loss: 267.905264\n",
      "INFO | epoch 23 (test) loss: 284.621200\n",
      "INFO | epoch 24 (  0%) loss: 234.094162\n",
      "INFO | epoch 24 ( 35%) loss: 252.132084\n",
      "INFO | epoch 24 ( 70%) loss: 246.209898\n",
      "INFO | epoch 24 (100%) loss: 244.383371\n",
      "INFO | epoch 24 (test) loss: 280.030351\n",
      "INFO | epoch 25 (  0%) loss: 222.521088\n",
      "INFO | epoch 25 ( 35%) loss: 234.934850\n",
      "INFO | epoch 25 ( 70%) loss: 232.638656\n",
      "INFO | epoch 25 (100%) loss: 231.579614\n",
      "INFO | epoch 25 (test) loss: 255.487929\n",
      "INFO | epoch 26 (  0%) loss: 204.916351\n",
      "INFO | epoch 26 ( 35%) loss: 213.405231\n",
      "INFO | epoch 26 ( 70%) loss: 214.541745\n",
      "INFO | epoch 26 (100%) loss: 210.498102\n",
      "INFO | epoch 26 (test) loss: 226.181442\n",
      "INFO | epoch 27 (  0%) loss: 231.852249\n",
      "INFO | epoch 27 ( 35%) loss: 203.576061\n",
      "INFO | epoch 27 ( 70%) loss: 200.490627\n",
      "INFO | epoch 27 (100%) loss: 195.427320\n",
      "INFO | epoch 27 (test) loss: 215.910060\n",
      "INFO | epoch 28 (  0%) loss: 168.696426\n",
      "INFO | epoch 28 ( 35%) loss: 184.562257\n",
      "INFO | epoch 28 ( 70%) loss: 181.561444\n",
      "INFO | epoch 28 (100%) loss: 178.527718\n",
      "INFO | epoch 28 (test) loss: 192.886151\n",
      "INFO | epoch 29 (  0%) loss: 188.046906\n",
      "INFO | epoch 29 ( 35%) loss: 174.136394\n",
      "INFO | epoch 29 ( 70%) loss: 172.828047\n",
      "INFO | epoch 29 (100%) loss: 171.886030\n",
      "INFO | epoch 29 (test) loss: 187.196449\n",
      "INFO | epoch 30 (  0%) loss: 163.618942\n",
      "INFO | epoch 30 ( 35%) loss: 164.417872\n",
      "INFO | epoch 30 ( 70%) loss: 161.146361\n",
      "INFO | epoch 30 (100%) loss: 159.417004\n",
      "INFO | epoch 30 (test) loss: 173.732220\n",
      "INFO | epoch 31 (  0%) loss: 158.911423\n",
      "INFO | epoch 31 ( 35%) loss: 156.782076\n",
      "INFO | epoch 31 ( 70%) loss: 158.052061\n",
      "INFO | epoch 31 (100%) loss: 156.172856\n",
      "INFO | epoch 31 (test) loss: 164.857191\n",
      "INFO | epoch 32 (  0%) loss: 152.964233\n",
      "INFO | epoch 32 ( 35%) loss: 146.868982\n",
      "INFO | epoch 32 ( 70%) loss: 144.957282\n",
      "INFO | epoch 32 (100%) loss: 145.541081\n",
      "INFO | epoch 32 (test) loss: 171.861038\n",
      "INFO | epoch 33 (  0%) loss: 180.447235\n",
      "INFO | epoch 33 ( 35%) loss: 147.732850\n",
      "INFO | epoch 33 ( 70%) loss: 144.863666\n",
      "INFO | epoch 33 (100%) loss: 139.864216\n",
      "INFO | epoch 33 (test) loss: 153.596599\n",
      "INFO | epoch 34 (  0%) loss: 108.483932\n",
      "INFO | epoch 34 ( 35%) loss: 135.367098\n",
      "INFO | epoch 34 ( 70%) loss: 132.776199\n",
      "INFO | epoch 34 (100%) loss: 131.002763\n",
      "INFO | epoch 34 (test) loss: 154.371211\n",
      "INFO | epoch 35 (  0%) loss: 127.404327\n",
      "INFO | epoch 35 ( 35%) loss: 135.575457\n",
      "INFO | epoch 35 ( 70%) loss: 134.838896\n",
      "INFO | epoch 35 (100%) loss: 130.106412\n",
      "INFO | epoch 35 (test) loss: 150.672075\n",
      "INFO | epoch 36 (  0%) loss: 104.423157\n",
      "INFO | epoch 36 ( 35%) loss: 124.714414\n",
      "INFO | epoch 36 ( 70%) loss: 121.693635\n",
      "INFO | epoch 36 (100%) loss: 121.048705\n",
      "INFO | epoch 36 (test) loss: 132.706560\n",
      "INFO | epoch 37 (  0%) loss: 117.166916\n",
      "INFO | epoch 37 ( 35%) loss: 110.020113\n",
      "INFO | epoch 37 ( 70%) loss: 124.751838\n",
      "INFO | epoch 37 (100%) loss: 124.106765\n",
      "INFO | epoch 37 (test) loss: 128.066795\n",
      "INFO | epoch 38 (  0%) loss: 108.804214\n",
      "INFO | epoch 38 ( 35%) loss: 123.258123\n",
      "INFO | epoch 38 ( 70%) loss: 118.432704\n",
      "INFO | epoch 38 (100%) loss: 115.412265\n",
      "INFO | epoch 38 (test) loss: 122.522269\n",
      "INFO | epoch 39 (  0%) loss: 99.556801\n",
      "INFO | epoch 39 ( 35%) loss: 102.064411\n",
      "INFO | epoch 39 ( 70%) loss: 112.463783\n",
      "INFO | epoch 39 (100%) loss: 112.089799\n",
      "INFO | epoch 39 (test) loss: 129.163205\n",
      "INFO | epoch 40 (  0%) loss: 105.966957\n",
      "INFO | epoch 40 ( 35%) loss: 110.495510\n",
      "INFO | epoch 40 ( 70%) loss: 103.961245\n",
      "INFO | epoch 40 (100%) loss: 101.879302\n",
      "INFO | epoch 40 (test) loss: 115.056815\n",
      "INFO | epoch 41 (  0%) loss: 89.217529\n",
      "INFO | epoch 41 ( 35%) loss: 99.189637\n",
      "INFO | epoch 41 ( 70%) loss: 99.848603\n",
      "INFO | epoch 41 (100%) loss: 100.821031\n",
      "INFO | epoch 41 (test) loss: 142.022021\n",
      "INFO | epoch 42 (  0%) loss: 87.306885\n",
      "INFO | epoch 42 ( 35%) loss: 91.986686\n",
      "INFO | epoch 42 ( 70%) loss: 93.891756\n",
      "INFO | epoch 42 (100%) loss: 94.629059\n",
      "INFO | epoch 42 (test) loss: 114.020569\n",
      "INFO | epoch 43 (  0%) loss: 87.965149\n",
      "INFO | epoch 43 ( 35%) loss: 97.270919\n",
      "INFO | epoch 43 ( 70%) loss: 97.294978\n",
      "INFO | epoch 43 (100%) loss: 94.629595\n",
      "INFO | epoch 43 (test) loss: 112.782597\n",
      "INFO | epoch 44 (  0%) loss: 79.808472\n",
      "INFO | epoch 44 ( 35%) loss: 93.721593\n",
      "INFO | epoch 44 ( 70%) loss: 95.911725\n",
      "INFO | epoch 44 (100%) loss: 94.016952\n",
      "INFO | epoch 44 (test) loss: 121.840981\n",
      "INFO | epoch 45 (  0%) loss: 103.587662\n",
      "INFO | epoch 45 ( 35%) loss: 94.517772\n",
      "INFO | epoch 45 ( 70%) loss: 97.714617\n",
      "INFO | epoch 45 (100%) loss: 103.256909\n",
      "INFO | epoch 45 (test) loss: 135.704670\n",
      "INFO | epoch 46 (  0%) loss: 122.366310\n",
      "INFO | epoch 46 ( 35%) loss: 99.197133\n",
      "INFO | epoch 46 ( 70%) loss: 104.771804\n",
      "INFO | epoch 46 (100%) loss: 108.661398\n",
      "INFO | epoch 46 (test) loss: 127.948620\n",
      "INFO | epoch 47 (  0%) loss: 92.457016\n",
      "INFO | epoch 47 ( 35%) loss: 98.338374\n",
      "INFO | epoch 47 ( 70%) loss: 101.311106\n",
      "INFO | epoch 47 (100%) loss: 100.139573\n",
      "INFO | epoch 47 (test) loss: 114.110178\n",
      "INFO | epoch 48 (  0%) loss: 83.564224\n",
      "INFO | epoch 48 ( 35%) loss: 91.647272\n",
      "INFO | epoch 48 ( 70%) loss: 95.689671\n",
      "INFO | epoch 48 (100%) loss: 98.926598\n",
      "INFO | epoch 48 (test) loss: 118.242511\n",
      "INFO | epoch 49 (  0%) loss: 90.806107\n",
      "INFO | epoch 49 ( 35%) loss: 98.092199\n",
      "INFO | epoch 49 ( 70%) loss: 101.583362\n",
      "INFO | epoch 49 (100%) loss: 98.367951\n",
      "INFO | epoch 49 (test) loss: 103.866755\n",
      "INFO | epoch 50 (  0%) loss: 84.068741\n",
      "INFO | epoch 50 ( 35%) loss: 106.924756\n",
      "INFO | epoch 50 ( 70%) loss: 106.063829\n",
      "INFO | epoch 50 (100%) loss: 105.052579\n",
      "INFO | epoch 50 (test) loss: 142.632611\n",
      "INFO | model blob saved to /home/jmsdnns/ML/pallets/saved/cvae.labeled.pkl\n",
      "INFO | training losses saved to /home/jmsdnns/ML/pallets/saved/cvae.labeled.json\n"
     ]
    }
   ],
   "source": [
    "model = M.cvae.LabeledCVAE(input_dim, hidden_dim, latent_dim, classes_dim)\n",
    "criterion = M.cvae.Loss()\n",
    "\n",
    "train_losses, test_losses = M.cvae.train(\n",
    "    device, model, criterion, train_loader, test_loader,\n",
    "    learn_rate=LR, epochs=EPOCHS, conditional_loss=True\n",
    ")\n",
    "\n",
    "M.save(SAVE_NAME, model, train_losses, test_losses)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "77fb32ad",
   "metadata": {},
   "source": [
    "# Results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bce33f5a",
   "metadata": {},
   "source": [
    "## Random Inputs w/ Real Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "f2b9d212",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "def rand_label():\n",
    "    label_idx = int(random.random() * len(dataset._labels))\n",
    "    features = dataset._labels[label_idx]\n",
    "    enabled_names = [k for k,v in zip(dataset._label_keys, features) if v.item() == 1]\n",
    "    return features.to(device), enabled_names\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "38c47d81",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate new image\n",
    "\n",
    "def rand_punk():\n",
    "    z = torch.randn(1, latent_dim).to(device)\n",
    "    # print(z.shape)\n",
    "    features, names = rand_label()\n",
    "    print(f\"Features: {', '.join(names)}\")\n",
    "\n",
    "    with torch.no_grad():\n",
    "        model.eval()\n",
    "        generated_image = model.decoder(z, features.unsqueeze(0))\n",
    "        # print(generated_image.shape)\n",
    "\n",
    "    decoded_one_hot = generated_image[0]\n",
    "    # print(decoded_one_hot.shape)\n",
    "    decoded_one_hot = decoded_one_hot[:-classes_dim].view((222, 24, 24))\n",
    "    # print(decoded_one_hot.shape)\n",
    "    decoded = DS.one_hot_to_rgba(decoded_one_hot, dataset.mapper)\n",
    "    # print(decoded.shape)\n",
    "    return decoded\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31e74ceb",
   "metadata": {},
   "source": [
    "### 5 Randoms w/ Real Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: male, shavedHead, smallShades\n",
      "torch.Size([127964])\n",
      "Shape: torch.Size([4, 24, 24])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "decoded = rand_punk()\n",
    "\n",
    "plt.imshow(to_pil_image(decoded))\n",
    "plt.axis('off')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "6744444a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, purpleEyeShadow, frumpyHair, mole\n",
      "torch.Size([127964])\n",
      "Shape: torch.Size([4, 24, 24])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "decoded = rand_punk()\n",
    "\n",
    "plt.imshow(to_pil_image(decoded))\n",
    "plt.axis('off')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "e190c583",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, pigtails, purpleLipstick, earring\n",
      "torch.Size([127964])\n",
      "Shape: torch.Size([4, 24, 24])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "decoded = rand_punk()\n",
    "\n",
    "plt.imshow(to_pil_image(decoded))\n",
    "plt.axis('off')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "be357c27",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: male, doRag, regularShades, mole\n",
      "torch.Size([127964])\n",
      "Shape: torch.Size([4, 24, 24])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "decoded = rand_punk()\n",
    "\n",
    "plt.imshow(to_pil_image(decoded))\n",
    "plt.axis('off')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d8f339ae",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, tassleHat, nerdGlasses\n",
      "torch.Size([127964])\n",
      "Shape: torch.Size([4, 24, 24])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "decoded = rand_punk()\n",
    "\n",
    "plt.imshow(to_pil_image(decoded))\n",
    "plt.axis('off')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e1bd761",
   "metadata": {},
   "source": [
    "## Reconstruction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a150efab",
   "metadata": {},
   "outputs": [],
   "source": [
    "def reconstruct_punk(idx):\n",
    "    punk = I.get_punk_tensor(idx)\n",
    "    p = DS.rgba_to_one_hot(punk, dataset.mapper)\n",
    "    p = p.unsqueeze(0)\n",
    "    p = p.to(device)\n",
    "\n",
    "    _, labels = dataset[idx]\n",
    "    l = labels.unsqueeze(0)\n",
    "    l = l.to(device)\n",
    "    enabled_features = [k for k,v in zip(dataset._label_keys, labels) if v.item() == 1]\n",
    "\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        reconstructed, mu, logvar = model.forward(p, l)\n",
    "\n",
    "    recon_punk = reconstructed[0].cpu()\n",
    "    recon_punk = recon_punk[:-classes_dim].view((222, 24, 24))\n",
    "    recon_punk = DS.one_hot_to_rgba(recon_punk, dataset.mapper)\n",
    "\n",
    "    return punk, recon_punk, enabled_features\n",
    "\n",
    "\n",
    "def draw_two(img1, img2):\n",
    "    page_size = 2\n",
    "    view_x, view_y = 4*page_size, 2*page_size\n",
    "    fig = plt.figure(figsize=(view_x, view_y))\n",
    "    fig.add_subplot(1, 2, 1)\n",
    "    plt.imshow(to_pil_image(img1))\n",
    "    plt.axis('off')\n",
    "    fig.add_subplot(1, 2, 2)\n",
    "    plt.imshow(to_pil_image(img2))\n",
    "    plt.axis('off')\n",
    "    plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a12b220",
   "metadata": {},
   "source": [
    "### 5 Reconstructions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "142e642d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, rosyCheeks, darkHair, cigarette\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "punk, recon_punk, features = reconstruct_punk(1000)\n",
    "\n",
    "print(f\"Features: {', '.join(features)}\")\n",
    "draw_two(punk, recon_punk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, stringyHair, purpleLipstick, earring\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAn8AAAEtCAYAAAB54AaaAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/SrBM8AAAACXBIWXMAAA9hAAAPYQGoP6dpAAAHMUlEQVR4nO3dMY5jRRRAUX/U+yADcpZAQN476JS8E8QOnJN2CFHvgCWQAxlspCZDQpqRXIy/f9n3nPirVA78dFXJ28YY4wQAQMIXR18AAIDbEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIU9HX4Dr2rbt6CssaYxx9BWAO2GOfpw5+ji8/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABCyDcv6lja7Y/L9/HLxt8+vb7udvff5s2fP8reAx2GOXufsWebourz8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAEGK3743tuWPydJrb1bjSjsnZ8/e++yx/I7gdc/Q655ujXV7+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI9W43NruWaNbMup6V1gzNnr/a3f2N4HbM0eucv9rdzdHb8fIHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiN2+Nza7k3J2lyLXYSclrMscvQ/m6Lq8/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABDydPQF7t1qOyb/+P233c7+5tvvdjt71p6/E7gtc/QY5miXlz8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABC7PZ9MD/++s/F3/7y0/c73mRfM7/zdPo/v3XufOBxmKMfZ44+Di9/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBkG2OMoy+xmm3bLv52/PnV3Nlf/zX1/fv5Zep7ruP59W3qe38j+C9zFHN0XV7+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACHk6+gIrmtkvOLO/EqDCHIV1efkDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgxG7fzzSzv/J0au2wfH59m/r+/fyy002AlZmjn2aOsgcvfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAITY7ctu7JgE+DzmKHvw8gcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQqx342Jfvv5w9BX+9ff556OvADDNHGUFXv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAI2cYY4+hL8Gnbtk19/35+2ekm9+359W3qe38LeBzm6HWYo4/Dyx8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAhT0dfgOPM7mlcad+lHZPACsxR7pGXPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELs9g2b3TE5uwdyT3ZMAiswR7lHXv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj1blxs7zVGVg0Bj84cZQVe/gAAQsQfAECI+AMACBF/AAAh4g8AIET8AQCEiD8AgBDxBwAQIv4AAELEHwBAiPgDAAgRfwAAIeIPACBE/AEAhIg/AIAQ8QcAECL+AABCxB8AQIj4AwAIEX8AACHiDwAgRPwBAISIPwCAkG2MMY6+BAAAt+HlDwAgRPwBAISIPwCAEPEHABAi/gAAQsQfAECI+AMACBF/AAAh4g8AIOQDWUkFvnFjJJcAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "punk, recon_punk, features = reconstruct_punk(2001)\n",
    "\n",
    "print(f\"Features: {', '.join(features)}\")\n",
    "draw_two(punk, recon_punk)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "8e401eca",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: female, wildWhiteHair, greenEyeShadow, hotLipstick\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "punk, recon_punk, features = reconstruct_punk(5000)\n",
    "\n",
    "print(f\"Features: {', '.join(features)}\")\n",
    "draw_two(punk, recon_punk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "99f3ac30",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: male, rosyCheeks, mustache, vr, cap, cigarette\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "punk, recon_punk, features = reconstruct_punk(8000)\n",
    "\n",
    "print(f\"Features: {', '.join(features)}\")\n",
    "draw_two(punk, recon_punk)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "eb402159",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features: male, handlebars, bandana\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "punk, recon_punk, features = reconstruct_punk(1337)\n",
    "\n",
    "print(f\"Features: {', '.join(features)}\")\n",
    "draw_two(punk, recon_punk)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "04a97510",
   "metadata": {},
   "source": [
    "## Losses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function matplotlib.pyplot.show(close=None, block=None)>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.title(\"Train & Test loss\")\n",
    "plt.plot(train_losses, label='train loss')\n",
    "plt.plot(test_losses, label='test loss')\n",
    "plt.legend()\n",
    "plt.show\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "c1261252",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[75816.66649524878,\n",
       " 10818.70769268063,\n",
       " 6420.392181612921,\n",
       " 4714.949565914506,\n",
       " 3680.791904747063,\n",
       " 3017.552086796321,\n",
       " 2570.441573988461,\n",
       " 2224.0943350284656,\n",
       " 1933.0295397170046,\n",
       " 1714.5676587693235,\n",
       " 1535.3550761337822,\n",
       " 1383.9895837662068,\n",
       " 1247.7607630736438,\n",
       " 1132.95849674306,\n",
       " 1034.932633582582,\n",
       " 938.9329368104326,\n",
       " 853.1819370350939,\n",
       " 786.570814524982,\n",
       " 727.2168233884987,\n",
       " 659.2538031179009,\n",
       " 621.8053717105946,\n",
       " 577.1006844162096,\n",
       " 535.8105273686402,\n",
       " 488.7667427333534,\n",
       " 463.1592271412518,\n",
       " 420.99620391629264,\n",
       " 390.8546402843286,\n",
       " 357.0554353971008,\n",
       " 343.77206036723254,\n",
       " 318.8340080748213,\n",
       " 312.34571270232505,\n",
       " 291.08216211305444,\n",
       " 279.7284327838438,\n",
       " 262.0055263329905,\n",
       " 260.21282472001747,\n",
       " 242.09740960682538,\n",
       " 248.2135298573379,\n",
       " 230.82453079764724,\n",
       " 224.17959876094304,\n",
       " 203.75860419848286,\n",
       " 201.64206236981332,\n",
       " 189.25811703999838,\n",
       " 189.25919049851439,\n",
       " 188.03390394711326,\n",
       " 206.51381818622562,\n",
       " 217.32279638195715,\n",
       " 200.27914557220242,\n",
       " 197.85319524453888,\n",
       " 196.73590201519906,\n",
       " 210.10515843384655]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_losses\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "15d6f86b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[13852.722557067871,\n",
       " 7807.9813804626465,\n",
       " 5602.655570983887,\n",
       " 4306.4272384643555,\n",
       " 3502.987632751465,\n",
       " 3007.9446086883545,\n",
       " 2498.8776779174805,\n",
       " 2167.9041595458984,\n",
       " 1928.5594968795776,\n",
       " 1705.7680749893188,\n",
       " 1556.8006219863892,\n",
       " 1363.205768585205,\n",
       " 1242.017978668213,\n",
       " 1126.8000583648682,\n",
       " 1064.331603050232,\n",
       " 949.5111865997314,\n",
       " 866.987512588501,\n",
       " 821.6678743362427,\n",
       " 757.8050684928894,\n",
       " 682.7565588951111,\n",
       " 639.8637466430664,\n",
       " 687.342227935791,\n",
       " 569.2424001693726,\n",
       " 560.0607013702393,\n",
       " 510.97585821151733,\n",
       " 452.36288499832153,\n",
       " 431.82012033462524,\n",
       " 385.7723021507263,\n",
       " 374.3928985595703,\n",
       " 347.46443915367126,\n",
       " 329.7143819332123,\n",
       " 343.72207617759705,\n",
       " 307.19319701194763,\n",
       " 308.7424211502075,\n",
       " 301.3441495895386,\n",
       " 265.41312050819397,\n",
       " 256.13358902931213,\n",
       " 245.04453706741333,\n",
       " 258.32641077041626,\n",
       " 230.1136291027069,\n",
       " 284.0440423488617,\n",
       " 228.04113841056824,\n",
       " 225.56519484519958,\n",
       " 243.68196153640747,\n",
       " 271.4093405008316,\n",
       " 255.89724028110504,\n",
       " 228.2203551530838,\n",
       " 236.48502254486084,\n",
       " 207.73351073265076,\n",
       " 285.265221118927]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_losses\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}