09-경쟁하며_학습하는_GAN/conditional_gan.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# cGAN으로 생성 제어하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "import torchvision\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchvision import transforms, datasets\n",
    "from torchvision.utils import save_image\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using Device: cuda\n"
     ]
    }
   ],
   "source": [
    "# 하이퍼파라미터\n",
    "EPOCHS = 300\n",
    "BATCH_SIZE = 100\n",
    "USE_CUDA = torch.cuda.is_available()\n",
    "DEVICE = torch.device(\"cuda\" if USE_CUDA else \"cpu\")\n",
    "print(\"Using Device:\", DEVICE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Fashion MNIST 데이터셋\n",
    "trainset = datasets.FashionMNIST(\n",
    "    './.data',\n",
    "    train=True,\n",
    "    download=True,\n",
    "    transform=transforms.Compose([\n",
    "       transforms.ToTensor(),\n",
    "       transforms.Normalize((0.5,), (0.5,))\n",
    "    ])\n",
    ")\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    dataset     = trainset,\n",
    "    batch_size  = BATCH_SIZE,\n",
    "    shuffle     = True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 생성자 (Generator)\n",
    "class Generator(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        \n",
    "        self.embed = nn.Embedding(10, 10)\n",
    "        \n",
    "        self.model = nn.Sequential(\n",
    "            nn.Linear(110, 256),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Linear(256, 512),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Linear(512, 1024),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Linear(1024, 784),\n",
    "            nn.Tanh()\n",
    "        )\n",
    "    \n",
    "    def forward(self, z, labels):\n",
    "        c = self.embed(labels)\n",
    "        x = torch.cat([z, c], 1)\n",
    "        return self.model(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 판별자 (Discriminator)\n",
    "class Discriminator(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        \n",
    "        self.embed = nn.Embedding(10, 10)\n",
    "        \n",
    "        self.model = nn.Sequential(\n",
    "            nn.Linear(794, 1024),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Dropout(0.3),\n",
    "            nn.Linear(1024, 512),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Dropout(0.3),\n",
    "            nn.Linear(512, 256),\n",
    "            nn.LeakyReLU(0.2, inplace=True),\n",
    "            nn.Dropout(0.3),\n",
    "            nn.Linear(256, 1),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "    \n",
    "    def forward(self, x, labels):\n",
    "        c = self.embed(labels)\n",
    "        x = torch.cat([x, c], 1)\n",
    "        return self.model(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 인스턴스를 만들고 모델의 가중치를 지정한 장치로 보내기\n",
    "D = Discriminator().to(DEVICE)\n",
    "G = Generator().to(DEVICE)\n",
    "\n",
    "# 이진 교차 엔트로피 함수와\n",
    "# 생성자와 판별자를 최적화할 Adam 모듈\n",
    "criterion = nn.BCELoss()\n",
    "d_optimizer = optim.Adam(D.parameters(), lr =0.0002)\n",
    "g_optimizer = optim.Adam(G.parameters(), lr =0.0002)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "이폭 [0/300] d_loss:0.3160 g_loss: 6.8871 D(x):0.89 D(G(z)):0.04\n",
      "이폭 [1/300] d_loss:0.5910 g_loss: 3.9762 D(x):0.87 D(G(z)):0.19\n",
      "이폭 [2/300] d_loss:0.3091 g_loss: 4.8783 D(x):0.92 D(G(z)):0.08\n",
      "이폭 [3/300] d_loss:0.1595 g_loss: 4.3155 D(x):0.95 D(G(z)):0.06\n",
      "이폭 [4/300] d_loss:0.5849 g_loss: 3.0070 D(x):0.78 D(G(z)):0.13\n",
      "이폭 [5/300] d_loss:0.4532 g_loss: 3.0106 D(x):0.82 D(G(z)):0.10\n",
      "이폭 [6/300] d_loss:1.0214 g_loss: 2.8349 D(x):0.64 D(G(z)):0.09\n",
      "이폭 [7/300] d_loss:0.6247 g_loss: 2.1058 D(x):0.79 D(G(z)):0.20\n",
      "이폭 [8/300] d_loss:0.6507 g_loss: 1.7656 D(x):0.82 D(G(z)):0.27\n",
      "이폭 [9/300] d_loss:0.6669 g_loss: 1.7414 D(x):0.81 D(G(z)):0.25\n",
      "이폭 [10/300] d_loss:0.6982 g_loss: 2.1408 D(x):0.81 D(G(z)):0.26\n",
      "이폭 [11/300] d_loss:0.8622 g_loss: 1.8217 D(x):0.74 D(G(z)):0.25\n",
      "이폭 [12/300] d_loss:0.8173 g_loss: 2.0123 D(x):0.77 D(G(z)):0.26\n",
      "이폭 [13/300] d_loss:0.7775 g_loss: 1.7512 D(x):0.75 D(G(z)):0.28\n",
      "이폭 [14/300] d_loss:0.7232 g_loss: 1.5495 D(x):0.76 D(G(z)):0.27\n",
      "이폭 [15/300] d_loss:0.8877 g_loss: 1.4896 D(x):0.69 D(G(z)):0.27\n",
      "이폭 [16/300] d_loss:0.7954 g_loss: 2.0424 D(x):0.69 D(G(z)):0.21\n",
      "이폭 [17/300] d_loss:0.7658 g_loss: 1.6869 D(x):0.74 D(G(z)):0.25\n",
      "이폭 [18/300] d_loss:0.9213 g_loss: 1.4370 D(x):0.69 D(G(z)):0.29\n",
      "이폭 [19/300] d_loss:0.8899 g_loss: 1.3454 D(x):0.73 D(G(z)):0.35\n",
      "이폭 [20/300] d_loss:1.1175 g_loss: 1.2468 D(x):0.58 D(G(z)):0.33\n",
      "이폭 [21/300] d_loss:0.8845 g_loss: 1.4447 D(x):0.73 D(G(z)):0.32\n",
      "이폭 [22/300] d_loss:1.0204 g_loss: 1.1669 D(x):0.63 D(G(z)):0.32\n",
      "이폭 [23/300] d_loss:1.2346 g_loss: 1.2664 D(x):0.63 D(G(z)):0.40\n",
      "이폭 [24/300] d_loss:1.0757 g_loss: 1.5536 D(x):0.62 D(G(z)):0.30\n",
      "이폭 [25/300] d_loss:0.8524 g_loss: 1.7157 D(x):0.66 D(G(z)):0.26\n",
      "이폭 [26/300] d_loss:1.0937 g_loss: 1.1672 D(x):0.64 D(G(z)):0.37\n",
      "이폭 [27/300] d_loss:0.9598 g_loss: 1.2177 D(x):0.67 D(G(z)):0.36\n",
      "이폭 [28/300] d_loss:0.9920 g_loss: 1.1852 D(x):0.66 D(G(z)):0.34\n",
      "이폭 [29/300] d_loss:1.0648 g_loss: 1.2556 D(x):0.67 D(G(z)):0.37\n",
      "이폭 [30/300] d_loss:1.1545 g_loss: 1.5034 D(x):0.56 D(G(z)):0.28\n",
      "이폭 [31/300] d_loss:1.2270 g_loss: 1.1542 D(x):0.58 D(G(z)):0.39\n",
      "이폭 [32/300] d_loss:1.1263 g_loss: 1.3266 D(x):0.65 D(G(z)):0.36\n",
      "이폭 [33/300] d_loss:1.1248 g_loss: 1.1094 D(x):0.62 D(G(z)):0.38\n",
      "이폭 [34/300] d_loss:1.1967 g_loss: 1.2937 D(x):0.56 D(G(z)):0.35\n",
      "이폭 [35/300] d_loss:1.1043 g_loss: 1.3641 D(x):0.60 D(G(z)):0.34\n",
      "이폭 [36/300] d_loss:0.8997 g_loss: 1.5824 D(x):0.68 D(G(z)):0.29\n",
      "이폭 [37/300] d_loss:1.1376 g_loss: 1.1206 D(x):0.59 D(G(z)):0.39\n",
      "이폭 [38/300] d_loss:1.3034 g_loss: 1.3584 D(x):0.63 D(G(z)):0.39\n",
      "이폭 [39/300] d_loss:0.9956 g_loss: 1.3172 D(x):0.68 D(G(z)):0.34\n",
      "이폭 [40/300] d_loss:0.9860 g_loss: 1.2805 D(x):0.67 D(G(z)):0.35\n",
      "이폭 [41/300] d_loss:1.0009 g_loss: 1.2975 D(x):0.69 D(G(z)):0.35\n",
      "이폭 [42/300] d_loss:1.1486 g_loss: 1.2332 D(x):0.58 D(G(z)):0.35\n",
      "이폭 [43/300] d_loss:1.0234 g_loss: 1.2544 D(x):0.66 D(G(z)):0.35\n",
      "이폭 [44/300] d_loss:1.1792 g_loss: 1.0267 D(x):0.59 D(G(z)):0.41\n",
      "이폭 [45/300] d_loss:1.0645 g_loss: 1.5280 D(x):0.70 D(G(z)):0.37\n",
      "이폭 [46/300] d_loss:1.1528 g_loss: 1.2533 D(x):0.58 D(G(z)):0.33\n",
      "이폭 [47/300] d_loss:1.1689 g_loss: 1.1853 D(x):0.59 D(G(z)):0.36\n",
      "이폭 [48/300] d_loss:0.9911 g_loss: 1.1729 D(x):0.64 D(G(z)):0.35\n",
      "이폭 [49/300] d_loss:1.2479 g_loss: 1.1463 D(x):0.59 D(G(z)):0.39\n",
      "이폭 [50/300] d_loss:1.0531 g_loss: 1.4914 D(x):0.69 D(G(z)):0.33\n",
      "이폭 [51/300] d_loss:1.2084 g_loss: 1.3741 D(x):0.60 D(G(z)):0.35\n",
      "이폭 [52/300] d_loss:1.1491 g_loss: 0.9025 D(x):0.60 D(G(z)):0.42\n",
      "이폭 [53/300] d_loss:1.1795 g_loss: 1.0152 D(x):0.63 D(G(z)):0.42\n",
      "이폭 [54/300] d_loss:1.2368 g_loss: 0.9881 D(x):0.59 D(G(z)):0.43\n",
      "이폭 [55/300] d_loss:1.1144 g_loss: 1.0084 D(x):0.61 D(G(z)):0.41\n",
      "이폭 [56/300] d_loss:0.9742 g_loss: 1.3360 D(x):0.62 D(G(z)):0.30\n",
      "이폭 [57/300] d_loss:1.1111 g_loss: 1.2013 D(x):0.62 D(G(z)):0.37\n",
      "이폭 [58/300] d_loss:1.1293 g_loss: 0.9425 D(x):0.61 D(G(z)):0.42\n",
      "이폭 [59/300] d_loss:1.1115 g_loss: 1.0971 D(x):0.64 D(G(z)):0.40\n",
      "이폭 [60/300] d_loss:1.1711 g_loss: 0.8514 D(x):0.64 D(G(z)):0.46\n",
      "이폭 [61/300] d_loss:1.2574 g_loss: 1.1688 D(x):0.53 D(G(z)):0.35\n",
      "이폭 [62/300] d_loss:1.1047 g_loss: 1.1473 D(x):0.62 D(G(z)):0.37\n",
      "이폭 [63/300] d_loss:1.1258 g_loss: 1.2802 D(x):0.62 D(G(z)):0.35\n",
      "이폭 [64/300] d_loss:1.1038 g_loss: 1.1615 D(x):0.70 D(G(z)):0.40\n",
      "이폭 [65/300] d_loss:1.1447 g_loss: 1.1335 D(x):0.58 D(G(z)):0.34\n",
      "이폭 [66/300] d_loss:1.2183 g_loss: 0.9577 D(x):0.60 D(G(z)):0.43\n",
      "이폭 [67/300] d_loss:1.0147 g_loss: 1.1334 D(x):0.67 D(G(z)):0.38\n",
      "이폭 [68/300] d_loss:0.9647 g_loss: 1.4295 D(x):0.69 D(G(z)):0.31\n",
      "이폭 [69/300] d_loss:1.1435 g_loss: 1.2024 D(x):0.64 D(G(z)):0.40\n",
      "이폭 [70/300] d_loss:1.1723 g_loss: 0.9909 D(x):0.61 D(G(z)):0.42\n",
      "이폭 [71/300] d_loss:1.2682 g_loss: 1.0710 D(x):0.58 D(G(z)):0.41\n",
      "이폭 [72/300] d_loss:1.1716 g_loss: 1.0253 D(x):0.56 D(G(z)):0.39\n",
      "이폭 [73/300] d_loss:1.2004 g_loss: 1.0096 D(x):0.55 D(G(z)):0.41\n",
      "이폭 [74/300] d_loss:1.0970 g_loss: 1.2437 D(x):0.66 D(G(z)):0.37\n",
      "이폭 [75/300] d_loss:1.0555 g_loss: 1.1143 D(x):0.64 D(G(z)):0.37\n",
      "이폭 [76/300] d_loss:1.1444 g_loss: 1.1805 D(x):0.61 D(G(z)):0.38\n",
      "이폭 [77/300] d_loss:1.1058 g_loss: 1.0923 D(x):0.61 D(G(z)):0.39\n",
      "이폭 [78/300] d_loss:1.1779 g_loss: 1.0541 D(x):0.63 D(G(z)):0.43\n",
      "이폭 [79/300] d_loss:1.2472 g_loss: 1.1688 D(x):0.63 D(G(z)):0.42\n",
      "이폭 [80/300] d_loss:1.2288 g_loss: 1.0327 D(x):0.58 D(G(z)):0.41\n",
      "이폭 [81/300] d_loss:0.9920 g_loss: 1.1429 D(x):0.63 D(G(z)):0.36\n",
      "이폭 [82/300] d_loss:1.0643 g_loss: 1.1235 D(x):0.66 D(G(z)):0.39\n",
      "이폭 [83/300] d_loss:1.1819 g_loss: 1.0258 D(x):0.61 D(G(z)):0.42\n",
      "이폭 [84/300] d_loss:1.1638 g_loss: 1.1043 D(x):0.60 D(G(z)):0.39\n",
      "이폭 [85/300] d_loss:1.0785 g_loss: 1.1045 D(x):0.62 D(G(z)):0.37\n",
      "이폭 [86/300] d_loss:1.2417 g_loss: 0.8960 D(x):0.58 D(G(z)):0.45\n",
      "이폭 [87/300] d_loss:1.1268 g_loss: 1.2213 D(x):0.60 D(G(z)):0.37\n",
      "이폭 [88/300] d_loss:1.2939 g_loss: 1.0583 D(x):0.58 D(G(z)):0.42\n",
      "이폭 [89/300] d_loss:1.1326 g_loss: 1.1059 D(x):0.62 D(G(z)):0.38\n",
      "이폭 [90/300] d_loss:1.3805 g_loss: 1.1491 D(x):0.48 D(G(z)):0.40\n",
      "이폭 [91/300] d_loss:1.2216 g_loss: 1.2522 D(x):0.58 D(G(z)):0.34\n",
      "이폭 [92/300] d_loss:1.2549 g_loss: 1.0024 D(x):0.65 D(G(z)):0.46\n",
      "이폭 [93/300] d_loss:1.2608 g_loss: 0.9609 D(x):0.56 D(G(z)):0.43\n",
      "이폭 [94/300] d_loss:1.1010 g_loss: 1.0334 D(x):0.63 D(G(z)):0.40\n",
      "이폭 [95/300] d_loss:1.1408 g_loss: 1.1788 D(x):0.59 D(G(z)):0.37\n",
      "이폭 [96/300] d_loss:1.0771 g_loss: 1.1025 D(x):0.63 D(G(z)):0.38\n",
      "이폭 [97/300] d_loss:1.2462 g_loss: 1.0300 D(x):0.59 D(G(z)):0.43\n",
      "이폭 [98/300] d_loss:1.3169 g_loss: 1.3007 D(x):0.57 D(G(z)):0.37\n",
      "이폭 [99/300] d_loss:1.0622 g_loss: 1.1779 D(x):0.61 D(G(z)):0.36\n",
      "이폭 [100/300] d_loss:1.1634 g_loss: 1.0386 D(x):0.58 D(G(z)):0.38\n",
      "이폭 [101/300] d_loss:1.2804 g_loss: 1.0972 D(x):0.55 D(G(z)):0.38\n",
      "이폭 [102/300] d_loss:1.3521 g_loss: 0.9992 D(x):0.53 D(G(z)):0.43\n",
      "이폭 [103/300] d_loss:1.0172 g_loss: 1.1565 D(x):0.67 D(G(z)):0.38\n",
      "이폭 [104/300] d_loss:1.0991 g_loss: 1.1267 D(x):0.61 D(G(z)):0.37\n",
      "이폭 [105/300] d_loss:1.1062 g_loss: 0.9505 D(x):0.62 D(G(z)):0.42\n",
      "이폭 [106/300] d_loss:1.0160 g_loss: 1.2985 D(x):0.65 D(G(z)):0.36\n",
      "이폭 [107/300] d_loss:1.1161 g_loss: 1.2177 D(x):0.59 D(G(z)):0.32\n",
      "이폭 [108/300] d_loss:1.1991 g_loss: 1.1204 D(x):0.57 D(G(z)):0.37\n",
      "이폭 [109/300] d_loss:1.2836 g_loss: 1.0011 D(x):0.58 D(G(z)):0.44\n",
      "이폭 [110/300] d_loss:1.1443 g_loss: 0.9716 D(x):0.63 D(G(z)):0.43\n",
      "이폭 [111/300] d_loss:1.2482 g_loss: 0.9895 D(x):0.55 D(G(z)):0.40\n",
      "이폭 [112/300] d_loss:1.0871 g_loss: 0.9745 D(x):0.62 D(G(z)):0.40\n",
      "이폭 [113/300] d_loss:1.2977 g_loss: 0.9649 D(x):0.57 D(G(z)):0.45\n",
      "이폭 [114/300] d_loss:1.2286 g_loss: 1.0941 D(x):0.63 D(G(z)):0.42\n",
      "이폭 [115/300] d_loss:1.2376 g_loss: 1.1070 D(x):0.57 D(G(z)):0.38\n",
      "이폭 [116/300] d_loss:1.1855 g_loss: 0.8995 D(x):0.60 D(G(z)):0.43\n",
      "이폭 [117/300] d_loss:1.0689 g_loss: 1.3804 D(x):0.63 D(G(z)):0.36\n",
      "이폭 [118/300] d_loss:1.1096 g_loss: 1.0656 D(x):0.63 D(G(z)):0.38\n",
      "이폭 [119/300] d_loss:1.2109 g_loss: 0.9245 D(x):0.57 D(G(z)):0.41\n",
      "이폭 [120/300] d_loss:1.0274 g_loss: 1.1375 D(x):0.63 D(G(z)):0.37\n",
      "이폭 [121/300] d_loss:1.1511 g_loss: 1.1668 D(x):0.61 D(G(z)):0.36\n",
      "이폭 [122/300] d_loss:1.2541 g_loss: 1.1607 D(x):0.54 D(G(z)):0.39\n",
      "이폭 [123/300] d_loss:1.2176 g_loss: 1.1589 D(x):0.58 D(G(z)):0.37\n",
      "이폭 [124/300] d_loss:1.2434 g_loss: 1.0104 D(x):0.60 D(G(z)):0.42\n",
      "이폭 [125/300] d_loss:1.2500 g_loss: 0.9446 D(x):0.57 D(G(z)):0.44\n",
      "이폭 [126/300] d_loss:1.1338 g_loss: 1.0269 D(x):0.59 D(G(z)):0.39\n",
      "이폭 [127/300] d_loss:1.3256 g_loss: 0.8861 D(x):0.55 D(G(z)):0.47\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "이폭 [128/300] d_loss:1.1599 g_loss: 1.0357 D(x):0.60 D(G(z)):0.39\n",
      "이폭 [129/300] d_loss:1.2836 g_loss: 0.9070 D(x):0.55 D(G(z)):0.43\n",
      "이폭 [130/300] d_loss:1.3021 g_loss: 1.0356 D(x):0.58 D(G(z)):0.41\n",
      "이폭 [131/300] d_loss:1.0849 g_loss: 1.2057 D(x):0.59 D(G(z)):0.35\n",
      "이폭 [132/300] d_loss:1.1975 g_loss: 0.9894 D(x):0.54 D(G(z)):0.39\n",
      "이폭 [133/300] d_loss:1.2037 g_loss: 1.0071 D(x):0.56 D(G(z)):0.40\n",
      "이폭 [134/300] d_loss:1.3586 g_loss: 0.7803 D(x):0.55 D(G(z)):0.48\n",
      "이폭 [135/300] d_loss:1.0960 g_loss: 1.3504 D(x):0.56 D(G(z)):0.29\n",
      "이폭 [136/300] d_loss:1.2298 g_loss: 0.9643 D(x):0.62 D(G(z)):0.42\n",
      "이폭 [137/300] d_loss:1.1276 g_loss: 1.0070 D(x):0.62 D(G(z)):0.41\n",
      "이폭 [138/300] d_loss:1.0824 g_loss: 1.0769 D(x):0.63 D(G(z)):0.40\n",
      "이폭 [139/300] d_loss:1.0641 g_loss: 1.1952 D(x):0.64 D(G(z)):0.39\n",
      "이폭 [140/300] d_loss:1.3613 g_loss: 0.9529 D(x):0.52 D(G(z)):0.45\n",
      "이폭 [141/300] d_loss:1.2217 g_loss: 1.0125 D(x):0.62 D(G(z)):0.42\n",
      "이폭 [142/300] d_loss:1.2494 g_loss: 0.8275 D(x):0.58 D(G(z)):0.46\n",
      "이폭 [143/300] d_loss:1.1680 g_loss: 0.9340 D(x):0.64 D(G(z)):0.45\n",
      "이폭 [144/300] d_loss:1.2364 g_loss: 1.0078 D(x):0.60 D(G(z)):0.40\n",
      "이폭 [145/300] d_loss:1.2444 g_loss: 0.9622 D(x):0.54 D(G(z)):0.40\n",
      "이폭 [146/300] d_loss:1.3496 g_loss: 0.9978 D(x):0.51 D(G(z)):0.40\n",
      "이폭 [147/300] d_loss:1.3775 g_loss: 0.9300 D(x):0.52 D(G(z)):0.45\n",
      "이폭 [148/300] d_loss:1.1927 g_loss: 0.9561 D(x):0.59 D(G(z)):0.42\n",
      "이폭 [149/300] d_loss:1.2422 g_loss: 0.8285 D(x):0.57 D(G(z)):0.45\n",
      "이폭 [150/300] d_loss:1.3064 g_loss: 0.9320 D(x):0.59 D(G(z)):0.45\n",
      "이폭 [151/300] d_loss:1.2686 g_loss: 0.9476 D(x):0.59 D(G(z)):0.43\n",
      "이폭 [152/300] d_loss:1.2767 g_loss: 0.9308 D(x):0.57 D(G(z)):0.42\n",
      "이폭 [153/300] d_loss:1.3285 g_loss: 0.9270 D(x):0.53 D(G(z)):0.43\n",
      "이폭 [154/300] d_loss:1.2871 g_loss: 0.9036 D(x):0.53 D(G(z)):0.43\n",
      "이폭 [155/300] d_loss:1.1348 g_loss: 1.0673 D(x):0.59 D(G(z)):0.39\n",
      "이폭 [156/300] d_loss:1.2569 g_loss: 1.2044 D(x):0.55 D(G(z)):0.39\n",
      "이폭 [157/300] d_loss:1.1620 g_loss: 1.0272 D(x):0.60 D(G(z)):0.40\n",
      "이폭 [158/300] d_loss:1.2275 g_loss: 0.9917 D(x):0.57 D(G(z)):0.41\n",
      "이폭 [159/300] d_loss:1.2050 g_loss: 0.9738 D(x):0.56 D(G(z)):0.40\n",
      "이폭 [160/300] d_loss:1.1526 g_loss: 1.2333 D(x):0.60 D(G(z)):0.36\n",
      "이폭 [161/300] d_loss:1.2868 g_loss: 0.9614 D(x):0.59 D(G(z)):0.46\n",
      "이폭 [162/300] d_loss:1.3186 g_loss: 0.8249 D(x):0.59 D(G(z)):0.47\n",
      "이폭 [163/300] d_loss:1.3044 g_loss: 0.9417 D(x):0.53 D(G(z)):0.42\n",
      "이폭 [164/300] d_loss:1.1137 g_loss: 1.0092 D(x):0.61 D(G(z)):0.40\n",
      "이폭 [165/300] d_loss:1.2838 g_loss: 0.9069 D(x):0.60 D(G(z)):0.46\n",
      "이폭 [166/300] d_loss:1.2802 g_loss: 0.9738 D(x):0.58 D(G(z)):0.43\n",
      "이폭 [167/300] d_loss:1.2483 g_loss: 0.9931 D(x):0.54 D(G(z)):0.40\n",
      "이폭 [168/300] d_loss:1.0827 g_loss: 1.1990 D(x):0.59 D(G(z)):0.34\n",
      "이폭 [169/300] d_loss:1.2997 g_loss: 0.8047 D(x):0.57 D(G(z)):0.48\n",
      "이폭 [170/300] d_loss:1.1657 g_loss: 0.9691 D(x):0.62 D(G(z)):0.42\n",
      "이폭 [171/300] d_loss:1.2189 g_loss: 0.9322 D(x):0.58 D(G(z)):0.43\n",
      "이폭 [172/300] d_loss:1.3357 g_loss: 0.7715 D(x):0.55 D(G(z)):0.48\n",
      "이폭 [173/300] d_loss:1.2662 g_loss: 1.0085 D(x):0.54 D(G(z)):0.40\n",
      "이폭 [174/300] d_loss:1.1604 g_loss: 1.2816 D(x):0.59 D(G(z)):0.34\n",
      "이폭 [175/300] d_loss:1.3354 g_loss: 0.8050 D(x):0.58 D(G(z)):0.48\n",
      "이폭 [176/300] d_loss:1.2668 g_loss: 0.8402 D(x):0.56 D(G(z)):0.46\n",
      "이폭 [177/300] d_loss:1.1373 g_loss: 1.1312 D(x):0.60 D(G(z)):0.38\n",
      "이폭 [178/300] d_loss:1.3249 g_loss: 1.1571 D(x):0.53 D(G(z)):0.39\n",
      "이폭 [179/300] d_loss:1.2576 g_loss: 0.9544 D(x):0.54 D(G(z)):0.42\n",
      "이폭 [180/300] d_loss:1.0759 g_loss: 1.0976 D(x):0.62 D(G(z)):0.38\n",
      "이폭 [181/300] d_loss:1.1209 g_loss: 1.0663 D(x):0.60 D(G(z)):0.38\n",
      "이폭 [182/300] d_loss:1.1964 g_loss: 1.0737 D(x):0.62 D(G(z)):0.41\n",
      "이폭 [183/300] d_loss:1.2747 g_loss: 0.9663 D(x):0.57 D(G(z)):0.42\n",
      "이폭 [184/300] d_loss:1.2308 g_loss: 0.8984 D(x):0.58 D(G(z)):0.43\n",
      "이폭 [185/300] d_loss:1.2885 g_loss: 0.8884 D(x):0.57 D(G(z)):0.43\n",
      "이폭 [186/300] d_loss:1.2854 g_loss: 0.8067 D(x):0.57 D(G(z)):0.46\n",
      "이폭 [187/300] d_loss:1.1997 g_loss: 0.8304 D(x):0.55 D(G(z)):0.42\n",
      "이폭 [188/300] d_loss:1.1353 g_loss: 1.0545 D(x):0.59 D(G(z)):0.40\n",
      "이폭 [189/300] d_loss:1.2585 g_loss: 1.1410 D(x):0.60 D(G(z)):0.41\n",
      "이폭 [190/300] d_loss:1.2669 g_loss: 0.7520 D(x):0.58 D(G(z)):0.48\n",
      "이폭 [191/300] d_loss:1.3322 g_loss: 0.8776 D(x):0.55 D(G(z)):0.46\n",
      "이폭 [192/300] d_loss:1.1893 g_loss: 0.8112 D(x):0.61 D(G(z)):0.46\n",
      "이폭 [193/300] d_loss:1.1830 g_loss: 0.7791 D(x):0.61 D(G(z)):0.47\n",
      "이폭 [194/300] d_loss:1.3332 g_loss: 0.8339 D(x):0.51 D(G(z)):0.46\n",
      "이폭 [195/300] d_loss:1.2623 g_loss: 1.0741 D(x):0.54 D(G(z)):0.40\n",
      "이폭 [196/300] d_loss:1.1428 g_loss: 1.0090 D(x):0.62 D(G(z)):0.41\n",
      "이폭 [197/300] d_loss:1.3870 g_loss: 0.9004 D(x):0.52 D(G(z)):0.45\n",
      "이폭 [198/300] d_loss:1.3383 g_loss: 0.9571 D(x):0.52 D(G(z)):0.43\n",
      "이폭 [199/300] d_loss:1.3075 g_loss: 0.7501 D(x):0.56 D(G(z)):0.49\n",
      "이폭 [200/300] d_loss:1.2396 g_loss: 0.9617 D(x):0.62 D(G(z)):0.46\n",
      "이폭 [201/300] d_loss:1.3438 g_loss: 1.1092 D(x):0.52 D(G(z)):0.38\n",
      "이폭 [202/300] d_loss:1.2730 g_loss: 0.8511 D(x):0.57 D(G(z)):0.47\n",
      "이폭 [203/300] d_loss:1.1790 g_loss: 0.9788 D(x):0.59 D(G(z)):0.41\n",
      "이폭 [204/300] d_loss:1.2273 g_loss: 0.8437 D(x):0.58 D(G(z)):0.47\n",
      "이폭 [205/300] d_loss:1.2920 g_loss: 0.8673 D(x):0.55 D(G(z)):0.45\n",
      "이폭 [206/300] d_loss:1.2945 g_loss: 0.9297 D(x):0.60 D(G(z)):0.46\n",
      "이폭 [207/300] d_loss:1.2715 g_loss: 1.0305 D(x):0.65 D(G(z)):0.45\n",
      "이폭 [208/300] d_loss:1.3115 g_loss: 0.9542 D(x):0.55 D(G(z)):0.44\n",
      "이폭 [209/300] d_loss:1.3051 g_loss: 1.2393 D(x):0.55 D(G(z)):0.38\n",
      "이폭 [210/300] d_loss:1.1815 g_loss: 0.9364 D(x):0.59 D(G(z)):0.43\n",
      "이폭 [211/300] d_loss:1.2948 g_loss: 1.0206 D(x):0.54 D(G(z)):0.42\n",
      "이폭 [212/300] d_loss:1.1876 g_loss: 0.9778 D(x):0.63 D(G(z)):0.44\n",
      "이폭 [213/300] d_loss:1.2058 g_loss: 0.8672 D(x):0.59 D(G(z)):0.45\n",
      "이폭 [214/300] d_loss:1.4025 g_loss: 0.9284 D(x):0.52 D(G(z)):0.45\n",
      "이폭 [215/300] d_loss:1.2316 g_loss: 0.8712 D(x):0.60 D(G(z)):0.45\n",
      "이폭 [216/300] d_loss:1.3874 g_loss: 0.9454 D(x):0.52 D(G(z)):0.43\n",
      "이폭 [217/300] d_loss:1.1225 g_loss: 1.0231 D(x):0.59 D(G(z)):0.38\n",
      "이폭 [218/300] d_loss:1.2485 g_loss: 0.9867 D(x):0.56 D(G(z)):0.41\n",
      "이폭 [219/300] d_loss:1.2633 g_loss: 1.1601 D(x):0.55 D(G(z)):0.38\n",
      "이폭 [220/300] d_loss:1.4121 g_loss: 0.9077 D(x):0.50 D(G(z)):0.44\n",
      "이폭 [221/300] d_loss:1.3511 g_loss: 0.8124 D(x):0.54 D(G(z)):0.46\n",
      "이폭 [222/300] d_loss:1.2591 g_loss: 0.9181 D(x):0.57 D(G(z)):0.45\n",
      "이폭 [223/300] d_loss:1.2789 g_loss: 0.8000 D(x):0.58 D(G(z)):0.47\n",
      "이폭 [224/300] d_loss:1.3491 g_loss: 0.9742 D(x):0.52 D(G(z)):0.43\n",
      "이폭 [225/300] d_loss:1.2939 g_loss: 0.9227 D(x):0.55 D(G(z)):0.42\n",
      "이폭 [226/300] d_loss:1.2688 g_loss: 0.9572 D(x):0.54 D(G(z)):0.39\n",
      "이폭 [227/300] d_loss:1.1632 g_loss: 1.0091 D(x):0.58 D(G(z)):0.40\n",
      "이폭 [228/300] d_loss:1.1878 g_loss: 0.9685 D(x):0.62 D(G(z)):0.44\n",
      "이폭 [229/300] d_loss:1.1764 g_loss: 1.0931 D(x):0.60 D(G(z)):0.40\n",
      "이폭 [230/300] d_loss:1.1633 g_loss: 1.0254 D(x):0.63 D(G(z)):0.42\n",
      "이폭 [231/300] d_loss:1.3779 g_loss: 0.9117 D(x):0.53 D(G(z)):0.45\n",
      "이폭 [232/300] d_loss:1.3449 g_loss: 0.8536 D(x):0.57 D(G(z)):0.48\n",
      "이폭 [233/300] d_loss:1.3516 g_loss: 0.7557 D(x):0.54 D(G(z)):0.47\n",
      "이폭 [234/300] d_loss:1.3285 g_loss: 1.1418 D(x):0.56 D(G(z)):0.43\n",
      "이폭 [235/300] d_loss:1.3615 g_loss: 1.0132 D(x):0.56 D(G(z)):0.45\n",
      "이폭 [236/300] d_loss:1.3904 g_loss: 0.8201 D(x):0.51 D(G(z)):0.46\n",
      "이폭 [237/300] d_loss:1.3223 g_loss: 0.8283 D(x):0.54 D(G(z)):0.45\n",
      "이폭 [238/300] d_loss:1.2615 g_loss: 0.8339 D(x):0.57 D(G(z)):0.45\n",
      "이폭 [239/300] d_loss:1.2030 g_loss: 0.9740 D(x):0.59 D(G(z)):0.42\n",
      "이폭 [240/300] d_loss:1.1859 g_loss: 1.0911 D(x):0.61 D(G(z)):0.39\n",
      "이폭 [241/300] d_loss:1.2214 g_loss: 0.8406 D(x):0.59 D(G(z)):0.45\n",
      "이폭 [242/300] d_loss:1.2384 g_loss: 0.9231 D(x):0.56 D(G(z)):0.40\n",
      "이폭 [243/300] d_loss:1.1728 g_loss: 1.0237 D(x):0.58 D(G(z)):0.39\n",
      "이폭 [244/300] d_loss:1.2932 g_loss: 0.9684 D(x):0.58 D(G(z)):0.44\n",
      "이폭 [245/300] d_loss:1.0582 g_loss: 0.9457 D(x):0.69 D(G(z)):0.40\n",
      "이폭 [246/300] d_loss:1.3432 g_loss: 0.9335 D(x):0.56 D(G(z)):0.45\n",
      "이폭 [247/300] d_loss:1.2782 g_loss: 0.8679 D(x):0.52 D(G(z)):0.43\n",
      "이폭 [248/300] d_loss:1.2220 g_loss: 1.0346 D(x):0.53 D(G(z)):0.36\n",
      "이폭 [249/300] d_loss:1.2484 g_loss: 1.0295 D(x):0.58 D(G(z)):0.44\n",
      "이폭 [250/300] d_loss:1.2475 g_loss: 0.9172 D(x):0.54 D(G(z)):0.42\n",
      "이폭 [251/300] d_loss:1.3299 g_loss: 0.9635 D(x):0.54 D(G(z)):0.41\n",
      "이폭 [252/300] d_loss:1.2654 g_loss: 0.7947 D(x):0.58 D(G(z)):0.47\n",
      "이폭 [253/300] d_loss:1.2718 g_loss: 0.7681 D(x):0.56 D(G(z)):0.47\n",
      "이폭 [254/300] d_loss:1.2711 g_loss: 0.8470 D(x):0.57 D(G(z)):0.45\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "이폭 [255/300] d_loss:1.3796 g_loss: 0.8272 D(x):0.57 D(G(z)):0.47\n",
      "이폭 [256/300] d_loss:1.3426 g_loss: 0.9632 D(x):0.52 D(G(z)):0.42\n",
      "이폭 [257/300] d_loss:1.2546 g_loss: 1.0157 D(x):0.57 D(G(z)):0.42\n",
      "이폭 [258/300] d_loss:1.2047 g_loss: 0.9986 D(x):0.62 D(G(z)):0.44\n",
      "이폭 [259/300] d_loss:1.3171 g_loss: 1.0010 D(x):0.54 D(G(z)):0.40\n",
      "이폭 [260/300] d_loss:1.3425 g_loss: 0.8781 D(x):0.55 D(G(z)):0.46\n",
      "이폭 [261/300] d_loss:1.2267 g_loss: 1.0366 D(x):0.62 D(G(z)):0.44\n",
      "이폭 [262/300] d_loss:1.1505 g_loss: 1.1203 D(x):0.61 D(G(z)):0.39\n",
      "이폭 [263/300] d_loss:1.2370 g_loss: 0.9579 D(x):0.57 D(G(z)):0.44\n",
      "이폭 [264/300] d_loss:1.1778 g_loss: 0.9074 D(x):0.58 D(G(z)):0.40\n",
      "이폭 [265/300] d_loss:1.3648 g_loss: 0.9658 D(x):0.53 D(G(z)):0.44\n",
      "이폭 [266/300] d_loss:1.2038 g_loss: 0.9570 D(x):0.57 D(G(z)):0.42\n",
      "이폭 [267/300] d_loss:1.2762 g_loss: 0.8381 D(x):0.58 D(G(z)):0.49\n",
      "이폭 [268/300] d_loss:1.0946 g_loss: 1.3682 D(x):0.65 D(G(z)):0.37\n",
      "이폭 [269/300] d_loss:1.5071 g_loss: 0.8394 D(x):0.51 D(G(z)):0.50\n",
      "이폭 [270/300] d_loss:1.1830 g_loss: 0.8900 D(x):0.59 D(G(z)):0.44\n",
      "이폭 [271/300] d_loss:1.2162 g_loss: 0.9783 D(x):0.61 D(G(z)):0.44\n",
      "이폭 [272/300] d_loss:1.2841 g_loss: 0.9193 D(x):0.57 D(G(z)):0.44\n",
      "이폭 [273/300] d_loss:1.3742 g_loss: 0.9791 D(x):0.53 D(G(z)):0.41\n",
      "이폭 [274/300] d_loss:1.1375 g_loss: 1.2430 D(x):0.62 D(G(z)):0.39\n",
      "이폭 [275/300] d_loss:1.1785 g_loss: 1.1205 D(x):0.61 D(G(z)):0.40\n",
      "이폭 [276/300] d_loss:1.4276 g_loss: 1.0428 D(x):0.51 D(G(z)):0.42\n",
      "이폭 [277/300] d_loss:1.3847 g_loss: 1.0324 D(x):0.55 D(G(z)):0.44\n",
      "이폭 [278/300] d_loss:1.1892 g_loss: 1.1183 D(x):0.62 D(G(z)):0.40\n",
      "이폭 [279/300] d_loss:1.2214 g_loss: 1.0615 D(x):0.56 D(G(z)):0.38\n",
      "이폭 [280/300] d_loss:1.2796 g_loss: 1.0563 D(x):0.54 D(G(z)):0.39\n",
      "이폭 [281/300] d_loss:1.1867 g_loss: 0.8633 D(x):0.63 D(G(z)):0.45\n",
      "이폭 [282/300] d_loss:1.1535 g_loss: 1.0292 D(x):0.59 D(G(z)):0.40\n",
      "이폭 [283/300] d_loss:1.2170 g_loss: 0.9809 D(x):0.58 D(G(z)):0.42\n",
      "이폭 [284/300] d_loss:1.2118 g_loss: 0.9556 D(x):0.60 D(G(z)):0.44\n",
      "이폭 [285/300] d_loss:1.2876 g_loss: 1.0758 D(x):0.53 D(G(z)):0.40\n",
      "이폭 [286/300] d_loss:1.4553 g_loss: 0.9380 D(x):0.54 D(G(z)):0.46\n",
      "이폭 [287/300] d_loss:1.3330 g_loss: 0.8979 D(x):0.53 D(G(z)):0.44\n",
      "이폭 [288/300] d_loss:1.3713 g_loss: 0.8559 D(x):0.50 D(G(z)):0.44\n",
      "이폭 [289/300] d_loss:1.3561 g_loss: 0.8846 D(x):0.54 D(G(z)):0.46\n",
      "이폭 [290/300] d_loss:1.2901 g_loss: 0.9300 D(x):0.53 D(G(z)):0.42\n",
      "이폭 [291/300] d_loss:1.3586 g_loss: 1.0188 D(x):0.53 D(G(z)):0.41\n",
      "이폭 [292/300] d_loss:1.2751 g_loss: 0.8963 D(x):0.57 D(G(z)):0.44\n",
      "이폭 [293/300] d_loss:1.1592 g_loss: 1.0320 D(x):0.64 D(G(z)):0.43\n",
      "이폭 [294/300] d_loss:1.2472 g_loss: 0.8309 D(x):0.61 D(G(z)):0.47\n",
      "이폭 [295/300] d_loss:1.2423 g_loss: 1.0166 D(x):0.63 D(G(z)):0.43\n",
      "이폭 [296/300] d_loss:1.3765 g_loss: 0.9225 D(x):0.49 D(G(z)):0.44\n",
      "이폭 [297/300] d_loss:1.2616 g_loss: 0.8626 D(x):0.56 D(G(z)):0.45\n",
      "이폭 [298/300] d_loss:1.2170 g_loss: 0.8461 D(x):0.57 D(G(z)):0.43\n",
      "이폭 [299/300] d_loss:1.1471 g_loss: 1.2162 D(x):0.60 D(G(z)):0.38\n"
     ]
    }
   ],
   "source": [
    "total_step = len(train_loader)\n",
    "for epoch in range(EPOCHS):\n",
    "    for i, (images, labels) in enumerate(train_loader):\n",
    "        images = images.reshape(BATCH_SIZE, -1).to(DEVICE)\n",
    "        \n",
    "        # '진짜'와 '가짜' 레이블 생성\n",
    "        real_labels = torch.ones(BATCH_SIZE, 1).to(DEVICE)\n",
    "        fake_labels = torch.zeros(BATCH_SIZE, 1).to(DEVICE)\n",
    "\n",
    "        # 판별자가 진짜 이미지를 진짜로 인식하는 오차 계산 (데이터셋 레이블 입력)\n",
    "        labels = labels.to(DEVICE)\n",
    "        outputs = D(images, labels)\n",
    "        d_loss_real = criterion(outputs, real_labels)\n",
    "        real_score = outputs\n",
    "    \n",
    "        # 무작위 텐서와 무작위 레이블을 생성자에 입력해 가짜 이미지 생성\n",
    "        z = torch.randn(BATCH_SIZE, 100).to(DEVICE)\n",
    "        g_label = torch.randint(0, 10, (BATCH_SIZE,)).to(DEVICE)\n",
    "        fake_images = G(z, g_label)\n",
    "        \n",
    "        # 판별자가 가짜 이미지를 가짜로 인식하는 오차 계산\n",
    "        outputs = D(fake_images, g_label)\n",
    "        d_loss_fake = criterion(outputs, fake_labels)\n",
    "        fake_score = outputs\n",
    "        \n",
    "        # 진짜와 가짜 이미지를 갖고 낸 오차를 더해서 판별자의 오차 계산\n",
    "        d_loss = d_loss_real + d_loss_fake\n",
    "        \n",
    "        # 역전파 알고리즘으로 판별자 모델의 학습을 진행\n",
    "        d_optimizer.zero_grad()\n",
    "        g_optimizer.zero_grad()\n",
    "        d_loss.backward()\n",
    "        d_optimizer.step()\n",
    "        \n",
    "        # 생성자가 판별자를 속였는지에 대한 오차 계산(무작위 레이블 입력)\n",
    "        fake_images = G(z, g_label)\n",
    "        outputs = D(fake_images, g_label)\n",
    "        g_loss = criterion(outputs, real_labels)\n",
    "\n",
    "        # 역전파 알고리즘으로 생성자 모델의 학습을 진행\n",
    "        d_optimizer.zero_grad()\n",
    "        g_optimizer.zero_grad()\n",
    "        g_loss.backward()\n",
    "        g_optimizer.step()\n",
    "        \n",
    "    print('이폭 [{}/{}] d_loss:{:.4f} g_loss: {:.4f} D(x):{:.2f} D(G(z)):{:.2f}'\n",
    "          .format(epoch,\n",
    "                  EPOCHS,\n",
    "                  d_loss.item(),\n",
    "                  g_loss.item(),\n",
    "                  real_score.mean().item(),\n",
    "                  fake_score.mean().item()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAEVBJREFUeJzt3XuMlFWax/HfA7ZcRUSwA4gwq2gc0WUEDUZCxrgQVide/tBITGRdIkbHZE32jzVu4pKsaya6M+pfJngJuM46bAJGM9HVwZh1Nm5GLrrKxQsgQndaWsIdFG362T/6ZbbFfp/TVFVXFZzvJyFdVU+deg8v/eOtqvO+55i7C0B+BjW6AwAag/ADmSL8QKYIP5Apwg9kivADmSL8QKYIP5Apwg9k6ox6bszMOJ0QGGDubv15XlVHfjObb2afmtkWM3uomtcCUF9W6bn9ZjZY0meS5kpqk7RG0gJ33xS04cgPDLB6HPmvlrTF3be5+3eSfifp5ipeD0AdVRP+iZJ29rrfVjz2A2a22MzWmtnaKrYFoMYG/As/d18qaanE236gmVRz5G+XNKnX/fOLxwCcAqoJ/xpJU83sJ2Z2pqQ7JL1Wm24BGGgVv+139y4ze0DSm5IGS3rB3TfWrGcABlTFQ30VbYzP/MCAq8tJPgBOXYQfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU3Vdohunn0GD4uNHd3f3gG3bLJ6kdtiwYaW1I0eO1Lo7J+WMM8qj19XVFbaN9vnJ7G+O/ECmCD+QKcIPZIrwA5ki/ECmCD+QKcIPZKqqcX4z2y7poKRjkrrcfWYtOoVTR2pcORrPbmlpCdtOnjw5rO/cuTOsX3XVVaW1DRs2hG0PHz4c1r/55puw3t7eHtZHjhxZWtu6dWvY9rbbbiuttbW1hW17q8VJPte5++4avA6AOuJtP5CpasPvkt4ys3VmtrgWHQJQH9W+7Z/t7u1mdp6kP5jZJ+7+bu8nFP8p8B8D0GSqOvK7e3vxs1PSK5Ku7uM5S919Jl8GAs2l4vCb2QgzO+v4bUnzJMVfoQJoGtW87W+V9EpxWeUZkv7d3f+zJr0CMOAqDr+7b5P0lzXsC05D1113XWktdU39J598EtZT5wmMGjWqtHbgwIGw7ejRo8N6apx/1qxZYX358uWltVTfOjo6Smvff/992LY3hvqATBF+IFOEH8gU4QcyRfiBTBF+IFNM3Y2qRJemStKKFStKa6kpqlOXp77xxhth/cEHH6yoX5I0derUsL5p06awvmPHjrA+bdq00lo05bgkDR06tLR29OjRsG1vHPmBTBF+IFOEH8gU4QcyRfiBTBF+IFOEH8iUuXv9NmZWv42hLu64446w/swzz5TWUr97w4cPD+upacOj8fJ9+/aFbQ8dOhTW33zzzbC+aNGisB5J7ZdoOvTu7m65e7x2eYEjP5Apwg9kivADmSL8QKYIP5Apwg9kivADmWKcH1U566yzwvpzzz1XWrv11lvDtseOHQvr0Xi3JA0aVH5s27t3b9g2mvZbkvbv3x/WU9OOT5w4sbSWmhb8sssuC+uM8wMIEX4gU4QfyBThBzJF+IFMEX4gU4QfyFRy3n4ze0HSLyR1uvu04rExklZImiJpu6Tb3T0eOMVpqbW1NawPHjy4tHbw4MGw7dlnnx3WU8tRR/PbDxkyJGwb9VuSzj333LCeWuI7WpPgpptuCtvWSn+O/MskzT/hsYckve3uUyW9XdwHcApJht/d35W054SHb5a0vLi9XNItNe4XgAFW6Wf+VnfvKG5/JSl+7weg6VS9Vp+7e3TOvpktlrS42u0AqK1Kj/y7zGy8JBU/O8ue6O5L3X2mu8+scFsABkCl4X9N0sLi9kJJr9amOwDqJRl+M3tZ0v9IusTM2sxskaRfSZprZp9L+qviPoBTSPIzv7svKCldX+O+oAktWFD2z99j3rx5Yf2SSy4praXmAkitcT9ixIiwHl3vv2fPiQNYJ7ft119/PazPn3/i6PgPXXzxxaW11H5JrTnQX5zhB2SK8AOZIvxApgg/kCnCD2SK8AOZqvr0XpzaUpfkTp06Naxff3084rtx48bS2nnnnVdxWyk9hfW4ceMq3nZ0ObAk3X333WF9zJgxYT1aXjx1OXGtcOQHMkX4gUwRfiBThB/IFOEHMkX4gUwRfiBTjPOf5szi1ZqPHj0a1seOHRvWU+cJDB8+vLSWWop6+vTpYf2LL74I61OmTCmtdXV1hW1T4/Sp5cFTy4tH2z9w4EDYtlY48gOZIvxApgg/kCnCD2SK8AOZIvxApgg/kClzL11pq/YbC5b1QuWi67+vvfbasO0111wT1u+///6wnlpGO1qGO7XMdeochdR174cOHSqtpZbojpbQlqQLL7wwrKdyFY3zr1q1Kmx75513prYd77gCR34gU4QfyBThBzJF+IFMEX4gU4QfyBThBzKVvJ7fzF6Q9AtJne4+rXhsiaR7JH1dPO1hd4/XLEap1Hj2oEHx/9HRNfP33ntv2HbChAlhfd26dWE9tRR1dB5A6pr41N87mvteqm4ugcmTJ4f1avt25plnltYmTZoUtq2V/hz5l0nq61/4SXefXvwh+MApJhl+d39X0p469AVAHVXzmf8BM/vIzF4ws3Nq1iMAdVFp+J+RdKGk6ZI6JP267IlmttjM1prZ2gq3BWAAVBR+d9/l7sfcvVvSs5KuDp671N1nuvvMSjsJoPYqCr+Zje9191ZJG2rTHQD10p+hvpcl/VzSWDNrk/RPkn5uZtMluaTtkuLxJABNJxl+d1/Qx8PPD0BfTlmpMd/Utd1z584N64sWLQrrTzzxRGlt9OjRYdvU9fzPPvtsWE9dF9/e3l5amzhxYtg2Nfd9atstLS2ltdQ4fGqugJTUuRup7dcDZ/gBmSL8QKYIP5Apwg9kivADmSL8QKZOm6m7U1NIR8s1S9LGjRvD+n333Vda27JlS9j2nnvuCeupYZ/U5aezZs0qrUXDXZL0wQcfhPXUpa2p4bpoyCw1DJka6oumBZfiS3pXr14dtp03b15YT12OfOTIkbAe/d1Sr33OOeWX0nR1dTF1N4AY4QcyRfiBTBF+IFOEH8gU4QcyRfiBTDXVOH9q2eORI0eW1i699NKw7dChQ8P6iBEjwvqTTz5ZWtu/f3/Y9q233grrqUtTL7/88rC+bNmy0lpqie1hw4aF9c2bN4f1GTNmhPUDBw6U1kaNGhW23bp1a1i/4IILwnpHR0dpLToHQEovH56SGuc/fPhwaS117kT0+9Ld3c04P4AY4QcyRfiBTBF+IFOEH8gU4QcyRfiBTDXVOP/jjz8etr/yyitLa1dccUVlnSqkxpy//vrr0tq+ffvCttu3bw/r48aNC+vvvfdeWL/llltKa4cOHQrbRktFS1Jra2tYT01RHY1Jp8bC33///bCe6nt07sb06dPDtqlr6lNzDUTj+FJ8TX5KNLfEypUr1dnZyTg/gHKEH8gU4QcyRfiBTBF+IFOEH8gU4QcylVyi28wmSXpRUqskl7TU3Z82szGSVkiaImm7pNvdfW8/Xq+0lporfc2aNamXL7V3b9y1G2+8MaxPmDChtLZhw4aw7Z49e8J66jyA1Lz/0XkG3377bdh2/vz5Yf3pp58O67Nnzw7r0X4fM2ZM2Da11kJqrDwa59+xY0fY9qKLLgrrqWXZU+tIRGsxpOZYiNZ5OJnzdvpz5O+S9Pfu/lNJsyT90sx+KukhSW+7+1RJbxf3AZwikuF39w53X1/cPihps6SJkm6WtLx42nJJ5aeZAWg6J/WZ38ymSPqZpD9JanX34/MkfaWejwUAThHJz/zHmdlISSslPejuB3p/dnd3Lztv38wWS1pcbUcB1Fa/jvxm1qKe4P/W3VcVD+8ys/FFfbykzr7auvtSd5/p7jNr0WEAtZEMv/Uc4p+XtNndf9Or9JqkhcXthZJerX33AAyU5CW9ZjZb0h8lfSzp+BjDw+r53P8fki6Q9KV6hvrCMa2WlhYfO3Zsaf3LL78M+/Ldd9+V1j799NOw7TvvvBPW77rrrrAeDUulloqO+i2lh3ZSw0rRpa2pIc7UlOapYcrzzz8/rEeXvqYuyU1dLtzZ2eebzT+Lpt9O/d5HS4tL6X+TgfTSSy+V1h555BFt27atX5f0Jj/zu/t/Syp7sev7sxEAzYcz/IBMEX4gU4QfyBThBzJF+IFMEX4gU3WfujuaEjnVl2gZ7tWrV4dtU2Pp0fLfKakx366urrC+c+fOsJ5aPjwaz25rawvbTp48OaxHS2xL0u7du8P6pEmTSmuPPvpo2LaaS5kladq0aaW1aqc0T9VTot/11PkNUd/nzJmj9evXM3U3gHKEH8gU4QcyRfiBTBF+IFOEH8gU4Qcy1VRLdFf52mH9scceC+sLFiwI69FS1altp6SWe37qqafC+pw5c0prM2bMCNsuWbIkrI8ePTqsp6bPXrZsWWnts88+C9um9ms0hbUUzw+ROn8hNU/BQErNUxCdH7FixQrt2rWLcX4A5Qg/kCnCD2SK8AOZIvxApgg/kCnCD2TqtBnnb2ap8ep6/hugf4YMGRLWU9fzp+YLGDVqVGlt//79Ydvo98nd5e6M8wMoR/iBTBF+IFOEH8gU4QcyRfiBTBF+IFPJcX4zmyTpRUmtklzSUnd/2syWSLpH0tfFUx9299cTr8WANjDA+jvO35/wj5c03t3Xm9lZktZJukXS7ZIOufu/9rdThB8YeP0Nf/nyOf//Qh2SOorbB81ss6SJ1XUPQKOd1Gd+M5si6WeS/lQ89ICZfWRmL5hZn/M5mdliM1trZmur6imAmur3uf1mNlLSf0n6F3dfZWatknar53uAf1bPR4O/TbwGb/uBAVazz/ySZGYtkn4v6U13/00f9SmSfu/u5SsjivAD9VCzC3us5xKi5yVt7h384ovA426VtOFkOwmgcfrzbf9sSX+U9LGk43MlPyxpgaTp6nnbv13SvcWXg9FrceQHBlhN3/bXCuEHBh7X8wMIEX4gU4QfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU8kJPGtst6Qve90fWzzWjJq1b83aL4m+VaqWfZvc3yfW9Xr+H23cbK27z2xYBwLN2rdm7ZdE3yrVqL7xth/IFOEHMtXo8C9t8PYjzdq3Zu2XRN8q1ZC+NfQzP4DGafSRH0CDNCT8ZjbfzD41sy1m9lAj+lDGzLab2cdm9mGjlxgrlkHrNLMNvR4bY2Z/MLPPi599LpPWoL4tMbP2Yt99aGY3NKhvk8zsHTPbZGYbzezviscbuu+CfjVkv9X9bb+ZDZb0maS5ktokrZG0wN031bUjJcxsu6SZ7t7wMWEzmyPpkKQXj6+GZGaPS9rj7r8q/uM8x93/oUn6tkQnuXLzAPWtbGXpv1ED910tV7yuhUYc+a+WtMXdt7n7d5J+J+nmBvSj6bn7u5L2nPDwzZKWF7eXq+eXp+5K+tYU3L3D3dcXtw9KOr6ydEP3XdCvhmhE+CdK2tnrfpuaa8lvl/SWma0zs8WN7kwfWnutjPSVpNZGdqYPyZWb6+mElaWbZt9VsuJ1rfGF34/NdvcrJf21pF8Wb2+bkvd8Zmum4ZpnJF2onmXcOiT9upGdKVaWXinpQXc/0LvWyH3XR78ast8aEf52SZN63T+/eKwpuHt78bNT0ivq+ZjSTHYdXyS1+NnZ4P78mbvvcvdj7t4t6Vk1cN8VK0uvlPRbd19VPNzwfddXvxq13xoR/jWSpprZT8zsTEl3SHqtAf34ETMbUXwRIzMbIWmemm/14dckLSxuL5T0agP78gPNsnJz2crSavC+a7oVr9297n8k3aCeb/y3SvrHRvShpF9/Iel/iz8bG903SS+r523g9+r5bmSRpHMlvS3pc0mrJY1por79m3pWc/5IPUEb36C+zVbPW/qPJH1Y/Lmh0fsu6FdD9htn+AGZ4gs/IFOEH8gU4QcyRfiBTBF+IFOEH8gU4QcyRfiBTP0fxpCv333MDS4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 만들고 싶은 아이템 생성하고 시각화하기\n",
    "item_number = 9 # 아이템 번호\n",
    "z = torch.randn(1, 100).to(DEVICE) # 배치 크기 1\n",
    "g_label = torch.full((1,), item_number, dtype=torch.long).to(DEVICE)\n",
    "sample_images = G(z, g_label)\n",
    "\n",
    "sample_images_img = np.reshape(sample_images.data.cpu().numpy()\n",
    "                               [0],(28, 28))\n",
    "plt.imshow(sample_images_img, cmap = 'gray')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}