NN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wCdWj_Ww36Bq"
   },
   "source": [
    "# Implementation of baseline for DANN\n",
    "Reference: https://arxiv.org/pdf/1505.07818.pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "55U-5Nwk36Bw"
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "id": "O5SHeDH24HnX",
    "outputId": "18ece497-0d25-4b69-e041-d280ca5ddc8e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.0.0'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "cGHr9W5o36Bz"
   },
   "outputs": [],
   "source": [
    "from torch.utils.data import DataLoader\n",
    "from torchvision import datasets\n",
    "from torchvision import transforms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "3S7AK3Pa36B1"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import datetime\n",
    "import os, sys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "iTNg8OmY36B4"
   },
   "outputs": [],
   "source": [
    "from matplotlib.pyplot import imshow, imsave\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "os.environ[\"CUDA_DEVICE_ORDER\"] = \"PCI_BUS_ID\"\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"{}\".format(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "d_hVsBuF36B7"
   },
   "outputs": [],
   "source": [
    "MODEL_NAME = 'NN'\n",
    "DEVICE = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "9OcpKYmx36CH"
   },
   "outputs": [],
   "source": [
    "class FeatureExtractor(nn.Module):\n",
    "    \"\"\"\n",
    "        Feature Extractor\n",
    "    \"\"\"\n",
    "    def __init__(self, in_channel=1, hidden_dims=512):\n",
    "        super(FeatureExtractor, self).__init__()\n",
    "        self.conv = nn.Sequential(\n",
    "            nn.Conv2d(in_channel, 64, 3, padding=1),\n",
    "            nn.BatchNorm2d(64),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(64, 128, 3, padding=1),\n",
    "            nn.BatchNorm2d(128),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(128, 256, 3, padding=1),\n",
    "            nn.BatchNorm2d(256),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(256, 256, 3, padding=1),\n",
    "            nn.BatchNorm2d(256),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Conv2d(256, hidden_dims, 3, padding=1),\n",
    "            nn.BatchNorm2d(hidden_dims),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.AdaptiveAvgPool2d((1,1)),\n",
    "        )\n",
    "        \n",
    "    def forward(self, x):\n",
    "        h = self.conv(x).squeeze() # (N, hidden_dims)\n",
    "        return h"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "c-1JVlFr95C8"
   },
   "outputs": [],
   "source": [
    "class Classifier(nn.Module):\n",
    "    \"\"\"\n",
    "        Classifier\n",
    "    \"\"\"\n",
    "    def __init__(self, input_size=512, num_classes=10):\n",
    "        super(Classifier, self).__init__()\n",
    "        self.layer = nn.Sequential(\n",
    "            nn.Linear(input_size, 256),\n",
    "            nn.ReLU(inplace=True),\n",
    "            nn.Linear(256, num_classes),\n",
    "        )\n",
    "        \n",
    "    def forward(self, h):\n",
    "        c = self.layer(h)\n",
    "        return c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "rmaidaPO36CQ"
   },
   "outputs": [],
   "source": [
    "F = FeatureExtractor().to(DEVICE)\n",
    "C = Classifier().to(DEVICE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "-zhMZpPc36CT"
   },
   "outputs": [],
   "source": [
    "transform = transforms.Compose([\n",
    "    transforms.Grayscale(1),\n",
    "    transforms.Resize(28),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize(mean=[0.5],\n",
    "                         std=[0.5])\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "dctQjVmy36CX"
   },
   "outputs": [],
   "source": [
    "mnist_train = datasets.MNIST(root='../data/', train=True, transform=transform, download=True)\n",
    "mnist_test = datasets.MNIST(root='../data/', train=False, transform=transform, download=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 53
    },
    "colab_type": "code",
    "id": "0GRgUyko4Yf8",
    "outputId": "8d5d6e2f-8a02-431f-80ef-f1f1583bc5b6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using downloaded and verified file: ../data/train_32x32.mat\n",
      "Using downloaded and verified file: ../data/test_32x32.mat\n"
     ]
    }
   ],
   "source": [
    "svhn_train = datasets.SVHN(root='../data/', split='train', transform=transform, download=True)\n",
    "svhn_test = datasets.SVHN(root='../data/', split='test', transform=transform, download=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "k-yZsYK_36Ca"
   },
   "outputs": [],
   "source": [
    "batch_size = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "id": "Nf_BypcP62yb",
    "outputId": "840bf031-34c1-4e69-859d-1665f4a7b86e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(73257, 3, 32, 32)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svhn_train.data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "bNGJwRYq36Cc"
   },
   "outputs": [],
   "source": [
    "svhn_loader = DataLoader(dataset=svhn_train, batch_size=batch_size, shuffle=True, drop_last=True)\n",
    "mnist_loader = DataLoader(dataset=mnist_train, batch_size=batch_size, shuffle=True, drop_last=True)\n",
    "eval_loader = DataLoader(dataset=svhn_test, batch_size=batch_size, shuffle=False, drop_last=False)\n",
    "test_loader = DataLoader(dataset=mnist_test, batch_size=batch_size, shuffle=False, drop_last=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "-BvTO5_CCoRj"
   },
   "outputs": [],
   "source": [
    "xe = nn.CrossEntropyLoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "dx-hu7Kx36Ch"
   },
   "outputs": [],
   "source": [
    "F_opt = torch.optim.Adam(F.parameters())\n",
    "C_opt = torch.optim.Adam(C.parameters())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "UZ4BHnuu36Co"
   },
   "outputs": [],
   "source": [
    "max_epoch = 50\n",
    "step = 0\n",
    "n_critic = 1 # for training more k steps about Discriminator\n",
    "n_batches = len(mnist_train)//batch_size\n",
    "# lamda = 0.01"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uiCXVKq236Cz"
   },
   "source": [
    "### Training Code\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "KrX9TTe8BLLP"
   },
   "outputs": [],
   "source": [
    "mnist_set = iter(mnist_loader)\n",
    "\n",
    "def sample_mnist(step, n_batches):\n",
    "    global mnist_set\n",
    "    if step % n_batches == 0:\n",
    "        mnist_set = iter(mnist_loader)\n",
    "    return mnist_set.next()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ll_c = []\n",
    "acc_lst = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "Q_htVI_VFwJ3",
    "outputId": "d571460c-dff5-46c6-dffa-fbd7396aba16",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1/50, Step: 0, C Loss: 2.3101 ---- 05:08:53\n",
      "***** Eval Result: 0.0759, Step: 0\n",
      "***** Test Result: 0.0958, Step: 0\n",
      "Epoch: 1/50, Step: 100, C Loss: 2.1153 ---- 05:09:10\n",
      "Epoch: 1/50, Step: 200, C Loss: 1.9948 ---- 05:09:16\n",
      "Epoch: 1/50, Step: 300, C Loss: 1.9285 ---- 05:09:22\n",
      "Epoch: 1/50, Step: 400, C Loss: 1.6654 ---- 05:09:28\n",
      "Epoch: 1/50, Step: 500, C Loss: 1.8997 ---- 05:09:34\n",
      "***** Eval Result: 0.3834, Step: 500\n",
      "***** Test Result: 0.2259, Step: 500\n",
      "Epoch: 1/50, Step: 600, C Loss: 1.5960 ---- 05:09:51\n",
      "Epoch: 1/50, Step: 700, C Loss: 1.6066 ---- 05:09:57\n",
      "Epoch: 1/50, Step: 800, C Loss: 1.2442 ---- 05:10:03\n",
      "Epoch: 1/50, Step: 900, C Loss: 1.4055 ---- 05:10:10\n",
      "Epoch: 1/50, Step: 1000, C Loss: 1.1821 ---- 05:10:16\n",
      "***** Eval Result: 0.5699, Step: 1000\n",
      "***** Test Result: 0.2789, Step: 1000\n",
      "Epoch: 1/50, Step: 1100, C Loss: 1.2983 ---- 05:10:34\n",
      "Epoch: 2/50, Step: 1200, C Loss: 0.9590 ---- 05:10:40\n",
      "Epoch: 2/50, Step: 1300, C Loss: 0.8810 ---- 05:10:46\n",
      "Epoch: 2/50, Step: 1400, C Loss: 0.7639 ---- 05:10:53\n",
      "Epoch: 2/50, Step: 1500, C Loss: 0.8699 ---- 05:10:59\n",
      "***** Eval Result: 0.6916, Step: 1500\n",
      "***** Test Result: 0.3601, Step: 1500\n",
      "Epoch: 2/50, Step: 1600, C Loss: 0.6450 ---- 05:11:17\n",
      "Epoch: 2/50, Step: 1700, C Loss: 0.7307 ---- 05:11:23\n",
      "Epoch: 2/50, Step: 1800, C Loss: 0.7665 ---- 05:11:29\n",
      "Epoch: 2/50, Step: 1900, C Loss: 0.8508 ---- 05:11:36\n",
      "Epoch: 2/50, Step: 2000, C Loss: 0.5347 ---- 05:11:42\n",
      "***** Eval Result: 0.7794, Step: 2000\n",
      "***** Test Result: 0.4373, Step: 2000\n",
      "Epoch: 2/50, Step: 2100, C Loss: 0.4717 ---- 05:12:00\n",
      "Epoch: 2/50, Step: 2200, C Loss: 0.6236 ---- 05:12:06\n",
      "Epoch: 3/50, Step: 2300, C Loss: 0.7120 ---- 05:12:13\n",
      "Epoch: 3/50, Step: 2400, C Loss: 0.5704 ---- 05:12:19\n",
      "Epoch: 3/50, Step: 2500, C Loss: 0.3494 ---- 05:12:25\n",
      "***** Eval Result: 0.8203, Step: 2500\n",
      "***** Test Result: 0.3740, Step: 2500\n",
      "Epoch: 3/50, Step: 2600, C Loss: 0.4971 ---- 05:12:43\n",
      "Epoch: 3/50, Step: 2700, C Loss: 0.5572 ---- 05:12:50\n",
      "Epoch: 3/50, Step: 2800, C Loss: 0.4639 ---- 05:12:56\n",
      "Epoch: 3/50, Step: 2900, C Loss: 0.4915 ---- 05:13:02\n",
      "Epoch: 3/50, Step: 3000, C Loss: 0.9108 ---- 05:13:09\n",
      "***** Eval Result: 0.8416, Step: 3000\n",
      "***** Test Result: 0.3577, Step: 3000\n",
      "Epoch: 3/50, Step: 3100, C Loss: 0.2872 ---- 05:13:27\n",
      "Epoch: 3/50, Step: 3200, C Loss: 0.4752 ---- 05:13:33\n",
      "Epoch: 3/50, Step: 3300, C Loss: 0.6527 ---- 05:13:39\n",
      "Epoch: 3/50, Step: 3400, C Loss: 0.4394 ---- 05:13:46\n",
      "Epoch: 4/50, Step: 3500, C Loss: 0.3660 ---- 05:13:52\n",
      "***** Eval Result: 0.7965, Step: 3500\n",
      "***** Test Result: 0.2836, Step: 3500\n",
      "Epoch: 4/50, Step: 3600, C Loss: 0.4516 ---- 05:14:10\n",
      "Epoch: 4/50, Step: 3700, C Loss: 0.5929 ---- 05:14:17\n",
      "Epoch: 4/50, Step: 3800, C Loss: 0.4181 ---- 05:14:23\n",
      "Epoch: 4/50, Step: 3900, C Loss: 0.2867 ---- 05:14:29\n",
      "Epoch: 4/50, Step: 4000, C Loss: 0.5138 ---- 05:14:36\n",
      "***** Eval Result: 0.8625, Step: 4000\n",
      "***** Test Result: 0.3361, Step: 4000\n",
      "Epoch: 4/50, Step: 4100, C Loss: 0.4716 ---- 05:14:54\n",
      "Epoch: 4/50, Step: 4200, C Loss: 0.6185 ---- 05:15:00\n",
      "Epoch: 4/50, Step: 4300, C Loss: 0.4696 ---- 05:15:07\n",
      "Epoch: 4/50, Step: 4400, C Loss: 0.4005 ---- 05:15:13\n",
      "Epoch: 4/50, Step: 4500, C Loss: 0.3726 ---- 05:15:20\n",
      "***** Eval Result: 0.8869, Step: 4500\n",
      "***** Test Result: 0.2803, Step: 4500\n",
      "Epoch: 5/50, Step: 4600, C Loss: 0.5453 ---- 05:15:38\n",
      "Epoch: 5/50, Step: 4700, C Loss: 0.6047 ---- 05:15:44\n",
      "Epoch: 5/50, Step: 4800, C Loss: 0.5229 ---- 05:15:50\n",
      "Epoch: 5/50, Step: 4900, C Loss: 0.2887 ---- 05:15:57\n",
      "Epoch: 5/50, Step: 5000, C Loss: 0.4958 ---- 05:16:03\n",
      "***** Eval Result: 0.8895, Step: 5000\n",
      "***** Test Result: 0.5511, Step: 5000\n",
      "Epoch: 5/50, Step: 5100, C Loss: 0.4499 ---- 05:16:21\n",
      "Epoch: 5/50, Step: 5200, C Loss: 0.4005 ---- 05:16:28\n",
      "Epoch: 5/50, Step: 5300, C Loss: 0.3493 ---- 05:16:34\n",
      "Epoch: 5/50, Step: 5400, C Loss: 0.3062 ---- 05:16:41\n",
      "Epoch: 5/50, Step: 5500, C Loss: 0.2911 ---- 05:16:47\n",
      "***** Eval Result: 0.8985, Step: 5500\n",
      "***** Test Result: 0.3781, Step: 5500\n",
      "Epoch: 5/50, Step: 5600, C Loss: 0.2583 ---- 05:17:05\n",
      "Epoch: 5/50, Step: 5700, C Loss: 0.4022 ---- 05:17:11\n",
      "Epoch: 6/50, Step: 5800, C Loss: 0.2760 ---- 05:17:18\n",
      "Epoch: 6/50, Step: 5900, C Loss: 0.4282 ---- 05:17:24\n",
      "Epoch: 6/50, Step: 6000, C Loss: 0.4242 ---- 05:17:31\n",
      "***** Eval Result: 0.8863, Step: 6000\n",
      "***** Test Result: 0.4638, Step: 6000\n",
      "Epoch: 6/50, Step: 6100, C Loss: 0.5048 ---- 05:17:49\n",
      "Epoch: 6/50, Step: 6200, C Loss: 0.4057 ---- 05:17:55\n",
      "Epoch: 6/50, Step: 6300, C Loss: 0.3749 ---- 05:18:02\n",
      "Epoch: 6/50, Step: 6400, C Loss: 0.5247 ---- 05:18:08\n",
      "Epoch: 6/50, Step: 6500, C Loss: 0.2675 ---- 05:18:14\n",
      "***** Eval Result: 0.9029, Step: 6500\n",
      "***** Test Result: 0.4154, Step: 6500\n",
      "Epoch: 6/50, Step: 6600, C Loss: 0.2420 ---- 05:18:32\n",
      "Epoch: 6/50, Step: 6700, C Loss: 0.4487 ---- 05:18:39\n",
      "Epoch: 6/50, Step: 6800, C Loss: 0.2060 ---- 05:18:45\n",
      "Epoch: 7/50, Step: 6900, C Loss: 0.1542 ---- 05:18:52\n",
      "Epoch: 7/50, Step: 7000, C Loss: 0.0924 ---- 05:18:58\n",
      "***** Eval Result: 0.8684, Step: 7000\n",
      "***** Test Result: 0.4824, Step: 7000\n",
      "Epoch: 7/50, Step: 7100, C Loss: 0.4566 ---- 05:19:16\n",
      "Epoch: 7/50, Step: 7200, C Loss: 0.2574 ---- 05:19:23\n",
      "Epoch: 7/50, Step: 7300, C Loss: 0.2215 ---- 05:19:29\n",
      "Epoch: 7/50, Step: 7400, C Loss: 0.6018 ---- 05:19:35\n",
      "Epoch: 7/50, Step: 7500, C Loss: 0.3835 ---- 05:19:42\n",
      "***** Eval Result: 0.9046, Step: 7500\n",
      "***** Test Result: 0.4366, Step: 7500\n",
      "Epoch: 7/50, Step: 7600, C Loss: 0.2335 ---- 05:20:00\n",
      "Epoch: 7/50, Step: 7700, C Loss: 0.3034 ---- 05:20:06\n",
      "Epoch: 7/50, Step: 7800, C Loss: 0.2697 ---- 05:20:13\n",
      "Epoch: 7/50, Step: 7900, C Loss: 0.2036 ---- 05:20:19\n",
      "Epoch: 7/50, Step: 8000, C Loss: 0.3531 ---- 05:20:25\n",
      "***** Eval Result: 0.9033, Step: 8000\n",
      "***** Test Result: 0.3287, Step: 8000\n",
      "Epoch: 8/50, Step: 8100, C Loss: 0.2578 ---- 05:20:43\n",
      "Epoch: 8/50, Step: 8200, C Loss: 0.2338 ---- 05:20:50\n",
      "Epoch: 8/50, Step: 8300, C Loss: 0.2221 ---- 05:20:56\n",
      "Epoch: 8/50, Step: 8400, C Loss: 0.2098 ---- 05:21:02\n",
      "Epoch: 8/50, Step: 8500, C Loss: 0.3476 ---- 05:21:09\n",
      "***** Eval Result: 0.9155, Step: 8500\n",
      "***** Test Result: 0.3397, Step: 8500\n",
      "Epoch: 8/50, Step: 8600, C Loss: 0.2871 ---- 05:21:27\n",
      "Epoch: 8/50, Step: 8700, C Loss: 0.4675 ---- 05:21:33\n",
      "Epoch: 8/50, Step: 8800, C Loss: 0.1984 ---- 05:21:40\n",
      "Epoch: 8/50, Step: 8900, C Loss: 0.1358 ---- 05:21:46\n",
      "Epoch: 8/50, Step: 9000, C Loss: 0.2463 ---- 05:21:53\n",
      "***** Eval Result: 0.9186, Step: 9000\n",
      "***** Test Result: 0.5453, Step: 9000\n",
      "Epoch: 8/50, Step: 9100, C Loss: 0.2802 ---- 05:22:11\n",
      "Epoch: 9/50, Step: 9200, C Loss: 0.3357 ---- 05:22:17\n",
      "Epoch: 9/50, Step: 9300, C Loss: 0.1713 ---- 05:22:23\n",
      "Epoch: 9/50, Step: 9400, C Loss: 0.5229 ---- 05:22:30\n",
      "Epoch: 9/50, Step: 9500, C Loss: 0.1786 ---- 05:22:36\n",
      "***** Eval Result: 0.9125, Step: 9500\n",
      "***** Test Result: 0.4767, Step: 9500\n",
      "Epoch: 9/50, Step: 9600, C Loss: 0.2466 ---- 05:22:54\n",
      "Epoch: 9/50, Step: 9700, C Loss: 0.2692 ---- 05:23:01\n",
      "Epoch: 9/50, Step: 9800, C Loss: 0.4295 ---- 05:23:07\n",
      "Epoch: 9/50, Step: 9900, C Loss: 0.5643 ---- 05:23:13\n",
      "Epoch: 9/50, Step: 10000, C Loss: 0.2152 ---- 05:23:20\n",
      "***** Eval Result: 0.9172, Step: 10000\n",
      "***** Test Result: 0.5070, Step: 10000\n",
      "Epoch: 9/50, Step: 10100, C Loss: 0.2127 ---- 05:23:38\n",
      "Epoch: 9/50, Step: 10200, C Loss: 0.2968 ---- 05:23:44\n",
      "Epoch: 10/50, Step: 10300, C Loss: 0.3067 ---- 05:23:51\n",
      "Epoch: 10/50, Step: 10400, C Loss: 0.1516 ---- 05:23:57\n",
      "Epoch: 10/50, Step: 10500, C Loss: 0.2920 ---- 05:24:04\n",
      "***** Eval Result: 0.9237, Step: 10500\n",
      "***** Test Result: 0.4127, Step: 10500\n",
      "Epoch: 10/50, Step: 10600, C Loss: 0.2341 ---- 05:24:22\n",
      "Epoch: 10/50, Step: 10700, C Loss: 0.3020 ---- 05:24:28\n",
      "Epoch: 10/50, Step: 10800, C Loss: 0.1450 ---- 05:24:34\n",
      "Epoch: 10/50, Step: 10900, C Loss: 0.2803 ---- 05:24:41\n",
      "Epoch: 10/50, Step: 11000, C Loss: 0.4618 ---- 05:24:47\n",
      "***** Eval Result: 0.9216, Step: 11000\n",
      "***** Test Result: 0.4825, Step: 11000\n",
      "Epoch: 10/50, Step: 11100, C Loss: 0.2443 ---- 05:25:05\n",
      "Epoch: 10/50, Step: 11200, C Loss: 0.1502 ---- 05:25:12\n",
      "Epoch: 10/50, Step: 11300, C Loss: 0.1335 ---- 05:25:18\n",
      "Epoch: 10/50, Step: 11400, C Loss: 0.1666 ---- 05:25:24\n",
      "Epoch: 11/50, Step: 11500, C Loss: 0.2195 ---- 05:25:31\n",
      "***** Eval Result: 0.9193, Step: 11500\n",
      "***** Test Result: 0.4177, Step: 11500\n",
      "Epoch: 11/50, Step: 11600, C Loss: 0.2073 ---- 05:25:49\n",
      "Epoch: 11/50, Step: 11700, C Loss: 0.0849 ---- 05:25:55\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 11/50, Step: 11800, C Loss: 0.2066 ---- 05:26:02\n",
      "Epoch: 11/50, Step: 11900, C Loss: 0.3356 ---- 05:26:08\n",
      "Epoch: 11/50, Step: 12000, C Loss: 0.2478 ---- 05:26:15\n",
      "***** Eval Result: 0.9117, Step: 12000\n",
      "***** Test Result: 0.4888, Step: 12000\n",
      "Epoch: 11/50, Step: 12100, C Loss: 0.1986 ---- 05:26:33\n",
      "Epoch: 11/50, Step: 12200, C Loss: 0.1354 ---- 05:26:39\n",
      "Epoch: 11/50, Step: 12300, C Loss: 0.1408 ---- 05:26:46\n",
      "Epoch: 11/50, Step: 12400, C Loss: 0.1469 ---- 05:26:52\n",
      "Epoch: 11/50, Step: 12500, C Loss: 0.2274 ---- 05:26:58\n",
      "***** Eval Result: 0.9224, Step: 12500\n",
      "***** Test Result: 0.4555, Step: 12500\n",
      "Epoch: 12/50, Step: 12600, C Loss: 0.1501 ---- 05:27:16\n",
      "Epoch: 12/50, Step: 12700, C Loss: 0.2082 ---- 05:27:23\n",
      "Epoch: 12/50, Step: 12800, C Loss: 0.1478 ---- 05:27:29\n",
      "Epoch: 12/50, Step: 12900, C Loss: 0.2060 ---- 05:27:36\n",
      "Epoch: 12/50, Step: 13000, C Loss: 0.1271 ---- 05:27:42\n",
      "***** Eval Result: 0.9078, Step: 13000\n",
      "***** Test Result: 0.5270, Step: 13000\n",
      "Epoch: 12/50, Step: 13100, C Loss: 0.5025 ---- 05:28:00\n",
      "Epoch: 12/50, Step: 13200, C Loss: 0.3010 ---- 05:28:06\n",
      "Epoch: 12/50, Step: 13300, C Loss: 0.1918 ---- 05:28:13\n",
      "Epoch: 12/50, Step: 13400, C Loss: 0.3435 ---- 05:28:19\n",
      "Epoch: 12/50, Step: 13500, C Loss: 0.1789 ---- 05:28:26\n",
      "***** Eval Result: 0.9253, Step: 13500\n",
      "***** Test Result: 0.5453, Step: 13500\n",
      "Epoch: 12/50, Step: 13600, C Loss: 0.1806 ---- 05:28:44\n",
      "Epoch: 12/50, Step: 13700, C Loss: 0.3411 ---- 05:28:50\n",
      "Epoch: 13/50, Step: 13800, C Loss: 0.1208 ---- 05:28:57\n",
      "Epoch: 13/50, Step: 13900, C Loss: 0.3258 ---- 05:29:03\n",
      "Epoch: 13/50, Step: 14000, C Loss: 0.3066 ---- 05:29:09\n",
      "***** Eval Result: 0.9302, Step: 14000\n",
      "***** Test Result: 0.5207, Step: 14000\n",
      "Epoch: 13/50, Step: 14100, C Loss: 0.2376 ---- 05:29:27\n",
      "Epoch: 13/50, Step: 14200, C Loss: 0.2700 ---- 05:29:34\n",
      "Epoch: 13/50, Step: 14300, C Loss: 0.3610 ---- 05:29:40\n",
      "Epoch: 13/50, Step: 14400, C Loss: 0.2234 ---- 05:29:47\n",
      "Epoch: 13/50, Step: 14500, C Loss: 0.0655 ---- 05:29:53\n",
      "***** Eval Result: 0.9111, Step: 14500\n",
      "***** Test Result: 0.5453, Step: 14500\n",
      "Epoch: 13/50, Step: 14600, C Loss: 0.1381 ---- 05:30:11\n",
      "Epoch: 13/50, Step: 14700, C Loss: 0.1552 ---- 05:30:18\n",
      "Epoch: 13/50, Step: 14800, C Loss: 0.2627 ---- 05:30:24\n",
      "Epoch: 14/50, Step: 14900, C Loss: 0.1712 ---- 05:30:31\n",
      "Epoch: 14/50, Step: 15000, C Loss: 0.1602 ---- 05:30:37\n",
      "***** Eval Result: 0.9255, Step: 15000\n",
      "***** Test Result: 0.4032, Step: 15000\n",
      "Epoch: 14/50, Step: 15100, C Loss: 0.1072 ---- 05:30:55\n",
      "Epoch: 14/50, Step: 15200, C Loss: 0.2377 ---- 05:31:01\n",
      "Epoch: 14/50, Step: 15300, C Loss: 0.2460 ---- 05:31:08\n",
      "Epoch: 14/50, Step: 15400, C Loss: 0.2210 ---- 05:31:14\n",
      "Epoch: 14/50, Step: 15500, C Loss: 0.2100 ---- 05:31:21\n",
      "***** Eval Result: 0.9235, Step: 15500\n",
      "***** Test Result: 0.5209, Step: 15500\n",
      "Epoch: 14/50, Step: 15600, C Loss: 0.1568 ---- 05:31:39\n",
      "Epoch: 14/50, Step: 15700, C Loss: 0.1963 ---- 05:31:45\n",
      "Epoch: 14/50, Step: 15800, C Loss: 0.1913 ---- 05:31:51\n",
      "Epoch: 14/50, Step: 15900, C Loss: 0.2967 ---- 05:31:58\n",
      "Epoch: 14/50, Step: 16000, C Loss: 0.1811 ---- 05:32:04\n",
      "***** Eval Result: 0.9122, Step: 16000\n",
      "***** Test Result: 0.4512, Step: 16000\n",
      "Epoch: 15/50, Step: 16100, C Loss: 0.0520 ---- 05:32:22\n",
      "Epoch: 15/50, Step: 16200, C Loss: 0.0905 ---- 05:32:29\n",
      "Epoch: 15/50, Step: 16300, C Loss: 0.0781 ---- 05:32:35\n",
      "Epoch: 15/50, Step: 16400, C Loss: 0.1601 ---- 05:32:42\n",
      "Epoch: 15/50, Step: 16500, C Loss: 0.0898 ---- 05:32:48\n",
      "***** Eval Result: 0.9265, Step: 16500\n",
      "***** Test Result: 0.5402, Step: 16500\n",
      "Epoch: 15/50, Step: 16600, C Loss: 0.1698 ---- 05:33:06\n",
      "Epoch: 15/50, Step: 16700, C Loss: 0.2887 ---- 05:33:13\n",
      "Epoch: 15/50, Step: 16800, C Loss: 0.1305 ---- 05:33:19\n",
      "Epoch: 15/50, Step: 16900, C Loss: 0.2903 ---- 05:33:25\n",
      "Epoch: 15/50, Step: 17000, C Loss: 0.1810 ---- 05:33:32\n",
      "***** Eval Result: 0.9325, Step: 17000\n",
      "***** Test Result: 0.5780, Step: 17000\n",
      "Epoch: 15/50, Step: 17100, C Loss: 0.1319 ---- 05:33:50\n",
      "Epoch: 16/50, Step: 17200, C Loss: 0.1307 ---- 05:33:56\n",
      "Epoch: 16/50, Step: 17300, C Loss: 0.0980 ---- 05:34:03\n",
      "Epoch: 16/50, Step: 17400, C Loss: 0.2094 ---- 05:34:09\n",
      "Epoch: 16/50, Step: 17500, C Loss: 0.1346 ---- 05:34:15\n",
      "***** Eval Result: 0.9207, Step: 17500\n",
      "***** Test Result: 0.5144, Step: 17500\n",
      "Epoch: 16/50, Step: 17600, C Loss: 0.1134 ---- 05:34:34\n",
      "Epoch: 16/50, Step: 17700, C Loss: 0.1220 ---- 05:34:40\n",
      "Epoch: 16/50, Step: 17800, C Loss: 0.2317 ---- 05:34:46\n",
      "Epoch: 16/50, Step: 17900, C Loss: 0.0911 ---- 05:34:53\n",
      "Epoch: 16/50, Step: 18000, C Loss: 0.1377 ---- 05:34:59\n",
      "***** Eval Result: 0.9193, Step: 18000\n",
      "***** Test Result: 0.5245, Step: 18000\n",
      "Epoch: 16/50, Step: 18100, C Loss: 0.1689 ---- 05:35:17\n",
      "Epoch: 16/50, Step: 18200, C Loss: 0.1609 ---- 05:35:24\n",
      "Epoch: 16/50, Step: 18300, C Loss: 0.0618 ---- 05:35:30\n",
      "Epoch: 17/50, Step: 18400, C Loss: 0.1935 ---- 05:35:36\n",
      "Epoch: 17/50, Step: 18500, C Loss: 0.0856 ---- 05:35:43\n",
      "***** Eval Result: 0.9277, Step: 18500\n",
      "***** Test Result: 0.5451, Step: 18500\n",
      "Epoch: 17/50, Step: 18600, C Loss: 0.1559 ---- 05:36:01\n",
      "Epoch: 17/50, Step: 18700, C Loss: 0.1837 ---- 05:36:07\n",
      "Epoch: 17/50, Step: 18800, C Loss: 0.0352 ---- 05:36:14\n",
      "Epoch: 17/50, Step: 18900, C Loss: 0.0556 ---- 05:36:20\n",
      "Epoch: 17/50, Step: 19000, C Loss: 0.1078 ---- 05:36:26\n",
      "***** Eval Result: 0.9196, Step: 19000\n",
      "***** Test Result: 0.4310, Step: 19000\n",
      "Epoch: 17/50, Step: 19100, C Loss: 0.1668 ---- 05:36:45\n",
      "Epoch: 17/50, Step: 19200, C Loss: 0.0676 ---- 05:36:51\n",
      "Epoch: 17/50, Step: 19300, C Loss: 0.0851 ---- 05:36:57\n",
      "Epoch: 17/50, Step: 19400, C Loss: 0.2705 ---- 05:37:04\n",
      "Epoch: 18/50, Step: 19500, C Loss: 0.0542 ---- 05:37:10\n",
      "***** Eval Result: 0.9248, Step: 19500\n",
      "***** Test Result: 0.4142, Step: 19500\n",
      "Epoch: 18/50, Step: 19600, C Loss: 0.1114 ---- 05:37:28\n",
      "Epoch: 18/50, Step: 19700, C Loss: 0.0304 ---- 05:37:35\n",
      "Epoch: 18/50, Step: 19800, C Loss: 0.1634 ---- 05:37:41\n",
      "Epoch: 18/50, Step: 19900, C Loss: 0.1760 ---- 05:37:48\n",
      "Epoch: 18/50, Step: 20000, C Loss: 0.1702 ---- 05:37:54\n",
      "***** Eval Result: 0.9261, Step: 20000\n",
      "***** Test Result: 0.4987, Step: 20000\n",
      "Epoch: 18/50, Step: 20100, C Loss: 0.1511 ---- 05:38:12\n",
      "Epoch: 18/50, Step: 20200, C Loss: 0.1358 ---- 05:38:19\n",
      "Epoch: 18/50, Step: 20300, C Loss: 0.2099 ---- 05:38:25\n",
      "Epoch: 18/50, Step: 20400, C Loss: 0.1088 ---- 05:38:31\n",
      "Epoch: 18/50, Step: 20500, C Loss: 0.1869 ---- 05:38:38\n",
      "***** Eval Result: 0.9224, Step: 20500\n",
      "***** Test Result: 0.3913, Step: 20500\n",
      "Epoch: 19/50, Step: 20600, C Loss: 0.1566 ---- 05:38:56\n",
      "Epoch: 19/50, Step: 20700, C Loss: 0.1877 ---- 05:39:02\n",
      "Epoch: 19/50, Step: 20800, C Loss: 0.0961 ---- 05:39:09\n",
      "Epoch: 19/50, Step: 20900, C Loss: 0.1183 ---- 05:39:15\n",
      "Epoch: 19/50, Step: 21000, C Loss: 0.1384 ---- 05:39:22\n",
      "***** Eval Result: 0.9167, Step: 21000\n",
      "***** Test Result: 0.3781, Step: 21000\n",
      "Epoch: 19/50, Step: 21100, C Loss: 0.1563 ---- 05:39:40\n",
      "Epoch: 19/50, Step: 21200, C Loss: 0.0309 ---- 05:39:46\n",
      "Epoch: 19/50, Step: 21300, C Loss: 0.2499 ---- 05:39:53\n",
      "Epoch: 19/50, Step: 21400, C Loss: 0.1191 ---- 05:39:59\n",
      "Epoch: 19/50, Step: 21500, C Loss: 0.2343 ---- 05:40:05\n",
      "***** Eval Result: 0.9316, Step: 21500\n",
      "***** Test Result: 0.5357, Step: 21500\n",
      "Epoch: 19/50, Step: 21600, C Loss: 0.0940 ---- 05:40:23\n",
      "Epoch: 19/50, Step: 21700, C Loss: 0.1010 ---- 05:40:30\n",
      "Epoch: 20/50, Step: 21800, C Loss: 0.0586 ---- 05:40:36\n",
      "Epoch: 20/50, Step: 21900, C Loss: 0.0783 ---- 05:40:43\n",
      "Epoch: 20/50, Step: 22000, C Loss: 0.0852 ---- 05:40:49\n",
      "***** Eval Result: 0.9199, Step: 22000\n",
      "***** Test Result: 0.4269, Step: 22000\n",
      "Epoch: 20/50, Step: 22100, C Loss: 0.0749 ---- 05:41:07\n",
      "Epoch: 20/50, Step: 22200, C Loss: 0.0994 ---- 05:41:14\n",
      "Epoch: 20/50, Step: 22300, C Loss: 0.1209 ---- 05:41:20\n",
      "Epoch: 20/50, Step: 22400, C Loss: 0.1501 ---- 05:41:27\n",
      "Epoch: 20/50, Step: 22500, C Loss: 0.1468 ---- 05:41:33\n",
      "***** Eval Result: 0.9123, Step: 22500\n",
      "***** Test Result: 0.2952, Step: 22500\n",
      "Epoch: 20/50, Step: 22600, C Loss: 0.1845 ---- 05:41:51\n",
      "Epoch: 20/50, Step: 22700, C Loss: 0.0550 ---- 05:41:58\n",
      "Epoch: 20/50, Step: 22800, C Loss: 0.0864 ---- 05:42:04\n",
      "Epoch: 21/50, Step: 22900, C Loss: 0.0578 ---- 05:42:10\n",
      "Epoch: 21/50, Step: 23000, C Loss: 0.0625 ---- 05:42:17\n",
      "***** Eval Result: 0.9224, Step: 23000\n",
      "***** Test Result: 0.5403, Step: 23000\n",
      "Epoch: 21/50, Step: 23100, C Loss: 0.0462 ---- 05:42:35\n",
      "Epoch: 21/50, Step: 23200, C Loss: 0.0366 ---- 05:42:41\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 21/50, Step: 23300, C Loss: 0.0402 ---- 05:42:48\n",
      "Epoch: 21/50, Step: 23400, C Loss: 0.1669 ---- 05:42:54\n",
      "Epoch: 21/50, Step: 23500, C Loss: 0.0607 ---- 05:43:01\n",
      "***** Eval Result: 0.9317, Step: 23500\n",
      "***** Test Result: 0.5417, Step: 23500\n",
      "Epoch: 21/50, Step: 23600, C Loss: 0.0598 ---- 05:43:19\n",
      "Epoch: 21/50, Step: 23700, C Loss: 0.0679 ---- 05:43:25\n",
      "Epoch: 21/50, Step: 23800, C Loss: 0.0608 ---- 05:43:31\n",
      "Epoch: 21/50, Step: 23900, C Loss: 0.1545 ---- 05:43:38\n",
      "Epoch: 21/50, Step: 24000, C Loss: 0.1994 ---- 05:43:44\n",
      "***** Eval Result: 0.9174, Step: 24000\n",
      "***** Test Result: 0.4973, Step: 24000\n",
      "Epoch: 22/50, Step: 24100, C Loss: 0.0410 ---- 05:44:02\n",
      "Epoch: 22/50, Step: 24200, C Loss: 0.0265 ---- 05:44:09\n",
      "Epoch: 22/50, Step: 24300, C Loss: 0.1230 ---- 05:44:15\n",
      "Epoch: 22/50, Step: 24400, C Loss: 0.0827 ---- 05:44:22\n",
      "Epoch: 22/50, Step: 24500, C Loss: 0.0252 ---- 05:44:28\n",
      "***** Eval Result: 0.9072, Step: 24500\n",
      "***** Test Result: 0.4130, Step: 24500\n",
      "Epoch: 22/50, Step: 24600, C Loss: 0.0388 ---- 05:44:46\n",
      "Epoch: 22/50, Step: 24700, C Loss: 0.0554 ---- 05:44:53\n",
      "Epoch: 22/50, Step: 24800, C Loss: 0.0930 ---- 05:44:59\n",
      "Epoch: 22/50, Step: 24900, C Loss: 0.1010 ---- 05:45:05\n",
      "Epoch: 22/50, Step: 25000, C Loss: 0.1147 ---- 05:45:12\n",
      "***** Eval Result: 0.9140, Step: 25000\n",
      "***** Test Result: 0.4427, Step: 25000\n",
      "Epoch: 22/50, Step: 25100, C Loss: 0.1742 ---- 05:45:30\n",
      "Epoch: 23/50, Step: 25200, C Loss: 0.0270 ---- 05:45:36\n",
      "Epoch: 23/50, Step: 25300, C Loss: 0.0262 ---- 05:45:43\n",
      "Epoch: 23/50, Step: 25400, C Loss: 0.0312 ---- 05:45:49\n",
      "Epoch: 23/50, Step: 25500, C Loss: 0.0406 ---- 05:45:56\n",
      "***** Eval Result: 0.9348, Step: 25500\n",
      "***** Test Result: 0.5003, Step: 25500\n",
      "Epoch: 23/50, Step: 25600, C Loss: 0.1108 ---- 05:46:14\n",
      "Epoch: 23/50, Step: 25700, C Loss: 0.0632 ---- 05:46:20\n",
      "Epoch: 23/50, Step: 25800, C Loss: 0.0878 ---- 05:46:26\n",
      "Epoch: 23/50, Step: 25900, C Loss: 0.0244 ---- 05:46:33\n",
      "Epoch: 23/50, Step: 26000, C Loss: 0.0539 ---- 05:46:39\n",
      "***** Eval Result: 0.9078, Step: 26000\n",
      "***** Test Result: 0.4136, Step: 26000\n",
      "Epoch: 23/50, Step: 26100, C Loss: 0.1074 ---- 05:46:57\n",
      "Epoch: 23/50, Step: 26200, C Loss: 0.0850 ---- 05:47:04\n",
      "Epoch: 23/50, Step: 26300, C Loss: 0.1661 ---- 05:47:10\n",
      "Epoch: 24/50, Step: 26400, C Loss: 0.0635 ---- 05:47:16\n",
      "Epoch: 24/50, Step: 26500, C Loss: 0.0237 ---- 05:47:23\n",
      "***** Eval Result: 0.9276, Step: 26500\n",
      "***** Test Result: 0.4777, Step: 26500\n",
      "Epoch: 24/50, Step: 26600, C Loss: 0.0906 ---- 05:47:41\n",
      "Epoch: 24/50, Step: 26700, C Loss: 0.0651 ---- 05:47:48\n",
      "Epoch: 24/50, Step: 26800, C Loss: 0.0487 ---- 05:47:54\n",
      "Epoch: 24/50, Step: 26900, C Loss: 0.0378 ---- 05:48:00\n",
      "Epoch: 24/50, Step: 27000, C Loss: 0.0141 ---- 05:48:07\n",
      "***** Eval Result: 0.9001, Step: 27000\n",
      "***** Test Result: 0.3056, Step: 27000\n",
      "Epoch: 24/50, Step: 27100, C Loss: 0.0137 ---- 05:48:25\n",
      "Epoch: 24/50, Step: 27200, C Loss: 0.0304 ---- 05:48:31\n",
      "Epoch: 24/50, Step: 27300, C Loss: 0.1276 ---- 05:48:38\n",
      "Epoch: 24/50, Step: 27400, C Loss: 0.1972 ---- 05:48:44\n",
      "Epoch: 25/50, Step: 27500, C Loss: 0.0785 ---- 05:48:51\n",
      "***** Eval Result: 0.9241, Step: 27500\n",
      "***** Test Result: 0.5002, Step: 27500\n",
      "Epoch: 25/50, Step: 27600, C Loss: 0.0567 ---- 05:49:09\n",
      "Epoch: 25/50, Step: 27700, C Loss: 0.0154 ---- 05:49:15\n",
      "Epoch: 25/50, Step: 27800, C Loss: 0.0556 ---- 05:49:21\n",
      "Epoch: 25/50, Step: 27900, C Loss: 0.0847 ---- 05:49:28\n",
      "Epoch: 25/50, Step: 28000, C Loss: 0.0990 ---- 05:49:34\n",
      "***** Eval Result: 0.9251, Step: 28000\n",
      "***** Test Result: 0.4244, Step: 28000\n",
      "Epoch: 25/50, Step: 28100, C Loss: 0.0207 ---- 05:49:52\n",
      "Epoch: 25/50, Step: 28200, C Loss: 0.1057 ---- 05:49:59\n",
      "Epoch: 25/50, Step: 28300, C Loss: 0.0381 ---- 05:50:05\n",
      "Epoch: 25/50, Step: 28400, C Loss: 0.0866 ---- 05:50:12\n",
      "Epoch: 25/50, Step: 28500, C Loss: 0.2022 ---- 05:50:18\n",
      "***** Eval Result: 0.9216, Step: 28500\n",
      "***** Test Result: 0.4933, Step: 28500\n",
      "Epoch: 26/50, Step: 28600, C Loss: 0.0471 ---- 05:50:36\n",
      "Epoch: 26/50, Step: 28700, C Loss: 0.0060 ---- 05:50:42\n",
      "Epoch: 26/50, Step: 28800, C Loss: 0.0345 ---- 05:50:49\n",
      "Epoch: 26/50, Step: 28900, C Loss: 0.0274 ---- 05:50:55\n",
      "Epoch: 26/50, Step: 29000, C Loss: 0.0210 ---- 05:51:02\n",
      "***** Eval Result: 0.9147, Step: 29000\n",
      "***** Test Result: 0.4292, Step: 29000\n",
      "Epoch: 26/50, Step: 29100, C Loss: 0.0412 ---- 05:51:20\n",
      "Epoch: 26/50, Step: 29200, C Loss: 0.0766 ---- 05:51:26\n",
      "Epoch: 26/50, Step: 29300, C Loss: 0.0622 ---- 05:51:33\n",
      "Epoch: 26/50, Step: 29400, C Loss: 0.0920 ---- 05:51:39\n",
      "Epoch: 26/50, Step: 29500, C Loss: 0.0433 ---- 05:51:46\n",
      "***** Eval Result: 0.9184, Step: 29500\n",
      "***** Test Result: 0.4725, Step: 29500\n",
      "Epoch: 26/50, Step: 29600, C Loss: 0.0508 ---- 05:52:04\n",
      "Epoch: 26/50, Step: 29700, C Loss: 0.1408 ---- 05:52:10\n",
      "Epoch: 27/50, Step: 29800, C Loss: 0.0253 ---- 05:52:16\n",
      "Epoch: 27/50, Step: 29900, C Loss: 0.0618 ---- 05:52:23\n",
      "Epoch: 27/50, Step: 30000, C Loss: 0.0486 ---- 05:52:29\n",
      "***** Eval Result: 0.9282, Step: 30000\n",
      "***** Test Result: 0.3906, Step: 30000\n",
      "Epoch: 27/50, Step: 30100, C Loss: 0.0523 ---- 05:52:47\n",
      "Epoch: 27/50, Step: 30200, C Loss: 0.0242 ---- 05:52:54\n",
      "Epoch: 27/50, Step: 30300, C Loss: 0.0398 ---- 05:53:00\n",
      "Epoch: 27/50, Step: 30400, C Loss: 0.0365 ---- 05:53:07\n",
      "Epoch: 27/50, Step: 30500, C Loss: 0.2105 ---- 05:53:13\n",
      "***** Eval Result: 0.8879, Step: 30500\n",
      "***** Test Result: 0.3528, Step: 30500\n",
      "Epoch: 27/50, Step: 30600, C Loss: 0.0567 ---- 05:53:31\n",
      "Epoch: 27/50, Step: 30700, C Loss: 0.1782 ---- 05:53:37\n",
      "Epoch: 27/50, Step: 30800, C Loss: 0.0589 ---- 05:53:44\n",
      "Epoch: 28/50, Step: 30900, C Loss: 0.0218 ---- 05:53:50\n",
      "Epoch: 28/50, Step: 31000, C Loss: 0.0814 ---- 05:53:57\n",
      "***** Eval Result: 0.9299, Step: 31000\n",
      "***** Test Result: 0.4450, Step: 31000\n",
      "Epoch: 28/50, Step: 31100, C Loss: 0.0209 ---- 05:54:15\n",
      "Epoch: 28/50, Step: 31200, C Loss: 0.0243 ---- 05:54:21\n",
      "Epoch: 28/50, Step: 31300, C Loss: 0.0133 ---- 05:54:28\n",
      "Epoch: 28/50, Step: 31400, C Loss: 0.0174 ---- 05:54:34\n",
      "Epoch: 28/50, Step: 31500, C Loss: 0.0442 ---- 05:54:40\n",
      "***** Eval Result: 0.9194, Step: 31500\n",
      "***** Test Result: 0.4287, Step: 31500\n",
      "Epoch: 28/50, Step: 31600, C Loss: 0.1871 ---- 05:54:59\n",
      "Epoch: 28/50, Step: 31700, C Loss: 0.0157 ---- 05:55:05\n",
      "Epoch: 28/50, Step: 31800, C Loss: 0.0542 ---- 05:55:11\n",
      "Epoch: 28/50, Step: 31900, C Loss: 0.0107 ---- 05:55:18\n",
      "Epoch: 28/50, Step: 32000, C Loss: 0.0288 ---- 05:55:24\n",
      "***** Eval Result: 0.9203, Step: 32000\n",
      "***** Test Result: 0.5562, Step: 32000\n",
      "Epoch: 29/50, Step: 32100, C Loss: 0.0453 ---- 05:55:42\n",
      "Epoch: 29/50, Step: 32200, C Loss: 0.1094 ---- 05:55:49\n",
      "Epoch: 29/50, Step: 32300, C Loss: 0.0173 ---- 05:55:55\n",
      "Epoch: 29/50, Step: 32400, C Loss: 0.1174 ---- 05:56:02\n",
      "Epoch: 29/50, Step: 32500, C Loss: 0.0527 ---- 05:56:08\n",
      "***** Eval Result: 0.9240, Step: 32500\n",
      "***** Test Result: 0.4559, Step: 32500\n",
      "Epoch: 29/50, Step: 32600, C Loss: 0.0357 ---- 05:56:26\n",
      "Epoch: 29/50, Step: 32700, C Loss: 0.0141 ---- 05:56:32\n",
      "Epoch: 29/50, Step: 32800, C Loss: 0.0549 ---- 05:56:39\n",
      "Epoch: 29/50, Step: 32900, C Loss: 0.0264 ---- 05:56:45\n",
      "Epoch: 29/50, Step: 33000, C Loss: 0.0332 ---- 05:56:52\n",
      "***** Eval Result: 0.9281, Step: 33000\n",
      "***** Test Result: 0.4812, Step: 33000\n",
      "Epoch: 29/50, Step: 33100, C Loss: 0.0611 ---- 05:57:10\n",
      "Epoch: 30/50, Step: 33200, C Loss: 0.0553 ---- 05:57:16\n",
      "Epoch: 30/50, Step: 33300, C Loss: 0.0316 ---- 05:57:23\n",
      "Epoch: 30/50, Step: 33400, C Loss: 0.0095 ---- 05:57:29\n",
      "Epoch: 30/50, Step: 33500, C Loss: 0.0599 ---- 05:57:36\n",
      "***** Eval Result: 0.9271, Step: 33500\n",
      "***** Test Result: 0.5519, Step: 33500\n",
      "Epoch: 30/50, Step: 33600, C Loss: 0.0061 ---- 05:57:54\n",
      "Epoch: 30/50, Step: 33700, C Loss: 0.0154 ---- 05:58:00\n",
      "Epoch: 30/50, Step: 33800, C Loss: 0.0202 ---- 05:58:07\n",
      "Epoch: 30/50, Step: 33900, C Loss: 0.0892 ---- 05:58:13\n",
      "Epoch: 30/50, Step: 34000, C Loss: 0.0043 ---- 05:58:19\n",
      "***** Eval Result: 0.9225, Step: 34000\n",
      "***** Test Result: 0.4235, Step: 34000\n",
      "Epoch: 30/50, Step: 34100, C Loss: 0.2450 ---- 05:58:37\n",
      "Epoch: 30/50, Step: 34200, C Loss: 0.0296 ---- 05:58:44\n",
      "Epoch: 30/50, Step: 34300, C Loss: 0.1301 ---- 05:58:50\n",
      "Epoch: 31/50, Step: 34400, C Loss: 0.0238 ---- 05:58:57\n",
      "Epoch: 31/50, Step: 34500, C Loss: 0.0658 ---- 05:59:03\n",
      "***** Eval Result: 0.9306, Step: 34500\n",
      "***** Test Result: 0.4531, Step: 34500\n",
      "Epoch: 31/50, Step: 34600, C Loss: 0.0569 ---- 05:59:21\n",
      "Epoch: 31/50, Step: 34700, C Loss: 0.0334 ---- 05:59:27\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 31/50, Step: 34800, C Loss: 0.0952 ---- 05:59:34\n",
      "Epoch: 31/50, Step: 34900, C Loss: 0.0284 ---- 05:59:40\n",
      "Epoch: 31/50, Step: 35000, C Loss: 0.0564 ---- 05:59:47\n",
      "***** Eval Result: 0.9249, Step: 35000\n",
      "***** Test Result: 0.4800, Step: 35000\n",
      "Epoch: 31/50, Step: 35100, C Loss: 0.0548 ---- 06:00:05\n",
      "Epoch: 31/50, Step: 35200, C Loss: 0.1651 ---- 06:00:11\n",
      "Epoch: 31/50, Step: 35300, C Loss: 0.0931 ---- 06:00:18\n",
      "Epoch: 31/50, Step: 35400, C Loss: 0.1654 ---- 06:00:24\n",
      "Epoch: 32/50, Step: 35500, C Loss: 0.0404 ---- 06:00:30\n",
      "***** Eval Result: 0.9188, Step: 35500\n",
      "***** Test Result: 0.4325, Step: 35500\n",
      "Epoch: 32/50, Step: 35600, C Loss: 0.0614 ---- 06:00:48\n",
      "Epoch: 32/50, Step: 35700, C Loss: 0.0294 ---- 06:00:55\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-29-3c2ff3ca10ee>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m         \u001b[0mtgt_images\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msample_mnist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_batches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m         \u001b[0;31m# Training Discriminator\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m         \u001b[0msrc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtgt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msrc_images\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mDEVICE\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mDEVICE\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtgt_images\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mDEVICE\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m         \u001b[0mc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mF\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "for epoch in range(1, max_epoch+1):\n",
    "    for idx, (src_images, labels) in enumerate(svhn_loader):\n",
    "        tgt_images, _ = sample_mnist(step, n_batches)\n",
    "        # Training Discriminator\n",
    "        src, labels, tgt = src_images.to(DEVICE), labels.to(DEVICE), tgt_images.to(DEVICE)\n",
    "        \n",
    "        c = C(F(src))\n",
    "        Lc = xe(c, labels)\n",
    "        \n",
    "        F.zero_grad()\n",
    "        C.zero_grad()\n",
    "        \n",
    "        Lc.backward()\n",
    "        \n",
    "        C_opt.step()\n",
    "        F_opt.step()\n",
    "        \n",
    "        if step % 100 == 0:\n",
    "            dt = datetime.datetime.now().strftime('%H:%M:%S')\n",
    "            print('Epoch: {}/{}, Step: {}, C Loss: {:.4f} ---- {}'.format(epoch, max_epoch, step, Lc.item(), dt))\n",
    "            ll_c.append(Lc)\n",
    "        \n",
    "        if step % 500 == 0:\n",
    "            F.eval()\n",
    "            C.eval()\n",
    "            with torch.no_grad():\n",
    "                corrects = torch.zeros(1).to(DEVICE)\n",
    "                for idx, (src, labels) in enumerate(eval_loader):\n",
    "                    src, labels = src.to(DEVICE), labels.to(DEVICE)\n",
    "                    c = C(F(src))\n",
    "                    _, preds = torch.max(c, 1)\n",
    "                    corrects += (preds == labels).sum()\n",
    "                acc = corrects.item() / len(eval_loader.dataset)\n",
    "                print('***** Eval Result: {:.4f}, Step: {}'.format(acc, step))\n",
    "                \n",
    "                corrects = torch.zeros(1).to(DEVICE)\n",
    "                for idx, (tgt, labels) in enumerate(test_loader):\n",
    "                    tgt, labels = tgt.to(DEVICE), labels.to(DEVICE)\n",
    "                    c = C(F(tgt))\n",
    "                    _, preds = torch.max(c, 1)\n",
    "                    corrects += (preds == labels).sum()\n",
    "                acc = corrects.item() / len(test_loader.dataset)\n",
    "                print('***** Test Result: {:.4f}, Step: {}'.format(acc, step))\n",
    "                acc_lst.append(acc)\n",
    "                \n",
    "            F.train()\n",
    "            C.train()\n",
    "        step += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "a8rIVAtx36C5"
   },
   "source": [
    "## Visualize Sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f0820033d68>]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f0820029908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(len(ll_c)), ll_c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ZjunDavN36C_"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f07d0c0bba8>]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f08200296d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(len(acc_lst)), acc_lst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.578"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max(acc_lst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "DANN.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}