train.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MONAI version: 1.3.0\n",
      "Numpy version: 1.26.1\n",
      "Pytorch version: 2.1.0+cu121\n",
      "MONAI flags: HAS_EXT = False, USE_COMPILED = False, USE_META_DICT = False\n",
      "MONAI rev id: 865972f7a791bf7b42efbcd87c8402bd865b329e\n",
      "MONAI __file__: /home/<username>/enviroments/mri-classifier/lib/python3.10/site-packages/monai/__init__.py\n",
      "\n",
      "Optional dependencies:\n",
      "Pytorch Ignite version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "ITK version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "Nibabel version: 5.1.0\n",
      "scikit-image version: 0.22.0\n",
      "scipy version: 1.11.3\n",
      "Pillow version: 10.0.1\n",
      "Tensorboard version: 2.14.1\n",
      "gdown version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "TorchVision version: 0.16.0+cu121\n",
      "tqdm version: 4.66.1\n",
      "lmdb version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "psutil version: 5.9.6\n",
      "pandas version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "einops version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "transformers version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "mlflow version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "pynrrd version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "clearml version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "\n",
      "For details about installing the optional dependencies, please visit:\n",
      "    https://docs.monai.io/en/latest/installation.html#installing-the-recommended-dependencies\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import logging\n",
    "import os\n",
    "import sys\n",
    "import shutil\n",
    "import tempfile\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "from torch.utils.tensorboard import SummaryWriter\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "import monai\n",
    "from monai.apps import download_and_extract\n",
    "from monai.config import print_config\n",
    "from monai.data import DataLoader, ImageDataset\n",
    "from monai.transforms import (\n",
    "    EnsureChannelFirst,\n",
    "    Compose,\n",
    "    RandRotate90,\n",
    "    Resize,\n",
    "    ScaleIntensity,\n",
    ")\n",
    "\n",
    "pin_memory= torch.cuda.is_available()\n",
    "device= torch.device(\"cuda\"if torch.cuda.is_available()else \"cpu\")\n",
    "\n",
    "logging.basicConfig(stream=sys.stdout, level=logging.INFO)\n",
    "print_config()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#set data directory\n",
    "parent_directory = r\"data/ADNIDenoise\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#create filepaths and labels\n",
    "dir_labels = [\"AD\", \"MCI\", \"CN\"]\n",
    "images = []\n",
    "labels = []\n",
    "\n",
    "label_count = 0\n",
    "\n",
    "for i in dir_labels:\n",
    "    dir_path = os.path.join(parent_directory, i)\n",
    "    file_names = os.listdir(dir_path)\n",
    "    for f in file_names:\n",
    "        images.append(os.path.join(dir_path, f))\n",
    "    for j in range(0, len(file_names)):\n",
    "        labels.append(label_count)\n",
    "    label_count += 1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define transforms\n",
    "train_transforms = Compose([ScaleIntensity(), EnsureChannelFirst(), Resize((96, 96, 96)), RandRotate90()])\n",
    "val_transforms = Compose([ScaleIntensity(), EnsureChannelFirst(), Resize((96, 96, 96))])\n",
    "\n",
    "#divide train, test and validate\n",
    "train_paths, temp_paths, train_labels, temp_labels = train_test_split(images, labels, test_size=0.4, random_state=42)\n",
    "valid_paths, test_paths, valid_labels, test_labels = train_test_split(temp_paths, temp_labels, test_size=0.5, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create a training data loader\n",
    "train_ds = ImageDataset(image_files=train_paths, labels=train_labels, transform=train_transforms)\n",
    "train_loader = DataLoader(train_ds, batch_size=2, shuffle=True, num_workers=2, pin_memory=pin_memory)\n",
    "\n",
    "# create a validation data loader\n",
    "val_ds = ImageDataset(image_files=valid_paths, labels=valid_labels, transform=val_transforms)\n",
    "val_loader = DataLoader(val_ds, batch_size=2, num_workers=2, pin_memory=pin_memory)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = monai.networks.nets.DenseNet121(spatial_dims=3, in_channels=1, out_channels = 3).to(device)\n",
    "checkpoint = torch.load(\"best_metric_model_classification3d_array.pth\")\n",
    "model.load_state_dict(checkpoint)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------\n",
      "epoch 1/20\n",
      "1/484, train_loss: 0.0021\n",
      "2/484, train_loss: 0.0020\n",
      "3/484, train_loss: 0.0038\n",
      "4/484, train_loss: 0.0073\n",
      "5/484, train_loss: 0.0011\n",
      "6/484, train_loss: 0.0030\n",
      "7/484, train_loss: 0.0083\n",
      "8/484, train_loss: 0.0018\n",
      "9/484, train_loss: 0.0011\n",
      "10/484, train_loss: 0.0086\n",
      "11/484, train_loss: 0.0036\n",
      "12/484, train_loss: 0.0035\n",
      "13/484, train_loss: 0.3569\n",
      "14/484, train_loss: 0.0051\n",
      "15/484, train_loss: 0.4789\n",
      "16/484, train_loss: 0.0013\n",
      "17/484, train_loss: 0.0039\n",
      "18/484, train_loss: 0.0028\n",
      "19/484, train_loss: 0.0016\n",
      "20/484, train_loss: 0.1176\n",
      "21/484, train_loss: 0.0031\n",
      "22/484, train_loss: 0.0158\n",
      "23/484, train_loss: 0.0697\n",
      "24/484, train_loss: 0.0054\n",
      "25/484, train_loss: 0.0327\n",
      "26/484, train_loss: 0.0096\n",
      "27/484, train_loss: 0.0260\n",
      "28/484, train_loss: 0.0044\n",
      "29/484, train_loss: 0.0017\n",
      "30/484, train_loss: 0.5916\n",
      "31/484, train_loss: 0.3266\n",
      "32/484, train_loss: 0.0020\n",
      "33/484, train_loss: 1.4066\n",
      "34/484, train_loss: 0.2761\n",
      "35/484, train_loss: 0.0030\n",
      "36/484, train_loss: 0.0010\n",
      "37/484, train_loss: 0.0014\n",
      "38/484, train_loss: 0.2774\n",
      "39/484, train_loss: 0.0026\n",
      "40/484, train_loss: 0.0863\n",
      "41/484, train_loss: 0.4901\n",
      "42/484, train_loss: 0.0014\n",
      "43/484, train_loss: 0.0034\n",
      "44/484, train_loss: 0.3062\n",
      "45/484, train_loss: 0.0025\n",
      "46/484, train_loss: 0.0025\n",
      "47/484, train_loss: 0.0052\n",
      "48/484, train_loss: 0.0118\n",
      "49/484, train_loss: 0.0033\n",
      "50/484, train_loss: 0.0456\n",
      "51/484, train_loss: 0.4813\n",
      "52/484, train_loss: 0.0084\n",
      "53/484, train_loss: 0.0018\n",
      "54/484, train_loss: 0.0073\n",
      "55/484, train_loss: 0.2248\n",
      "56/484, train_loss: 0.3200\n",
      "57/484, train_loss: 0.2545\n",
      "58/484, train_loss: 0.0016\n",
      "59/484, train_loss: 0.0024\n",
      "60/484, train_loss: 0.2790\n",
      "61/484, train_loss: 0.0138\n",
      "62/484, train_loss: 0.0014\n",
      "63/484, train_loss: 0.0011\n",
      "64/484, train_loss: 0.0008\n",
      "65/484, train_loss: 1.2300\n",
      "66/484, train_loss: 0.0109\n",
      "67/484, train_loss: 0.0344\n",
      "68/484, train_loss: 0.0023\n",
      "69/484, train_loss: 0.0066\n",
      "70/484, train_loss: 0.0048\n",
      "71/484, train_loss: 0.0027\n",
      "72/484, train_loss: 0.0016\n",
      "73/484, train_loss: 0.0027\n",
      "74/484, train_loss: 0.0113\n",
      "75/484, train_loss: 0.0011\n",
      "76/484, train_loss: 0.0041\n",
      "77/484, train_loss: 0.0010\n",
      "78/484, train_loss: 0.0007\n",
      "79/484, train_loss: 0.0023\n",
      "80/484, train_loss: 0.0020\n",
      "81/484, train_loss: 0.0009\n",
      "82/484, train_loss: 0.0015\n",
      "83/484, train_loss: 0.0010\n",
      "84/484, train_loss: 0.0034\n",
      "85/484, train_loss: 0.1760\n",
      "86/484, train_loss: 0.0029\n",
      "87/484, train_loss: 0.0031\n",
      "88/484, train_loss: 0.0012\n",
      "89/484, train_loss: 0.0924\n",
      "90/484, train_loss: 0.1552\n",
      "91/484, train_loss: 0.0040\n",
      "92/484, train_loss: 0.0009\n",
      "93/484, train_loss: 0.0016\n",
      "94/484, train_loss: 0.2878\n",
      "95/484, train_loss: 0.4221\n",
      "96/484, train_loss: 0.0190\n",
      "97/484, train_loss: 0.0011\n",
      "98/484, train_loss: 0.0062\n",
      "99/484, train_loss: 0.8681\n",
      "100/484, train_loss: 0.0016\n",
      "101/484, train_loss: 0.0021\n",
      "102/484, train_loss: 0.5104\n",
      "103/484, train_loss: 0.1351\n",
      "104/484, train_loss: 0.0046\n",
      "105/484, train_loss: 0.3619\n",
      "106/484, train_loss: 0.0012\n",
      "107/484, train_loss: 0.0418\n",
      "108/484, train_loss: 0.1859\n",
      "109/484, train_loss: 0.0011\n",
      "110/484, train_loss: 0.0009\n",
      "111/484, train_loss: 0.0037\n",
      "112/484, train_loss: 2.6748\n",
      "113/484, train_loss: 0.0031\n",
      "114/484, train_loss: 0.0175\n",
      "115/484, train_loss: 0.1299\n",
      "116/484, train_loss: 0.0010\n",
      "117/484, train_loss: 0.1745\n",
      "118/484, train_loss: 0.0049\n",
      "119/484, train_loss: 0.0012\n",
      "120/484, train_loss: 0.0068\n",
      "121/484, train_loss: 0.0018\n",
      "122/484, train_loss: 0.2495\n",
      "123/484, train_loss: 0.0015\n",
      "124/484, train_loss: 0.0188\n",
      "125/484, train_loss: 0.0011\n",
      "126/484, train_loss: 0.0014\n",
      "127/484, train_loss: 0.0250\n",
      "128/484, train_loss: 0.0118\n",
      "129/484, train_loss: 0.0110\n",
      "130/484, train_loss: 0.0011\n",
      "131/484, train_loss: 0.7421\n",
      "132/484, train_loss: 0.0017\n",
      "133/484, train_loss: 0.0010\n",
      "134/484, train_loss: 0.0041\n",
      "135/484, train_loss: 0.0019\n",
      "136/484, train_loss: 0.0044\n",
      "137/484, train_loss: 0.0053\n",
      "138/484, train_loss: 0.0009\n",
      "139/484, train_loss: 0.0112\n",
      "140/484, train_loss: 0.0110\n",
      "141/484, train_loss: 0.0016\n",
      "142/484, train_loss: 0.0012\n",
      "143/484, train_loss: 0.0052\n",
      "144/484, train_loss: 0.0019\n",
      "145/484, train_loss: 0.0030\n",
      "146/484, train_loss: 1.0346\n",
      "147/484, train_loss: 0.0052\n",
      "148/484, train_loss: 0.1556\n",
      "149/484, train_loss: 0.0040\n",
      "150/484, train_loss: 0.0333\n",
      "151/484, train_loss: 0.0042\n",
      "152/484, train_loss: 1.3202\n",
      "153/484, train_loss: 0.0034\n",
      "154/484, train_loss: 0.0055\n",
      "155/484, train_loss: 0.0015\n",
      "156/484, train_loss: 1.1328\n",
      "157/484, train_loss: 0.0035\n",
      "158/484, train_loss: 0.2146\n",
      "159/484, train_loss: 0.0344\n",
      "160/484, train_loss: 0.3721\n",
      "161/484, train_loss: 0.0039\n",
      "162/484, train_loss: 0.0009\n",
      "163/484, train_loss: 0.0046\n",
      "164/484, train_loss: 0.0028\n",
      "165/484, train_loss: 0.0047\n",
      "166/484, train_loss: 1.7113\n",
      "167/484, train_loss: 0.0022\n",
      "168/484, train_loss: 0.0048\n",
      "169/484, train_loss: 0.0161\n",
      "170/484, train_loss: 0.0066\n",
      "171/484, train_loss: 0.2753\n",
      "172/484, train_loss: 0.0015\n",
      "173/484, train_loss: 0.0550\n",
      "174/484, train_loss: 0.0034\n",
      "175/484, train_loss: 0.0034\n",
      "176/484, train_loss: 0.0020\n",
      "177/484, train_loss: 0.1917\n",
      "178/484, train_loss: 0.0166\n",
      "179/484, train_loss: 0.0061\n",
      "180/484, train_loss: 0.0041\n",
      "181/484, train_loss: 0.0017\n",
      "182/484, train_loss: 0.0059\n",
      "183/484, train_loss: 0.0031\n",
      "184/484, train_loss: 0.0023\n",
      "185/484, train_loss: 0.0043\n",
      "186/484, train_loss: 0.0036\n",
      "187/484, train_loss: 0.0033\n",
      "188/484, train_loss: 0.0015\n",
      "189/484, train_loss: 0.0055\n",
      "190/484, train_loss: 0.0051\n",
      "191/484, train_loss: 0.0040\n",
      "192/484, train_loss: 0.0025\n",
      "193/484, train_loss: 0.6173\n",
      "194/484, train_loss: 0.7448\n",
      "195/484, train_loss: 0.0337\n",
      "196/484, train_loss: 0.1029\n",
      "197/484, train_loss: 0.0033\n",
      "198/484, train_loss: 0.0026\n",
      "199/484, train_loss: 0.0030\n",
      "200/484, train_loss: 1.2540\n",
      "201/484, train_loss: 1.1721\n",
      "202/484, train_loss: 0.0020\n",
      "203/484, train_loss: 0.0012\n",
      "204/484, train_loss: 0.0053\n",
      "205/484, train_loss: 0.0034\n",
      "206/484, train_loss: 0.0051\n",
      "207/484, train_loss: 0.0038\n",
      "208/484, train_loss: 0.0134\n",
      "209/484, train_loss: 0.0012\n",
      "210/484, train_loss: 0.0009\n",
      "211/484, train_loss: 0.0023\n",
      "212/484, train_loss: 0.0234\n",
      "213/484, train_loss: 0.0139\n",
      "214/484, train_loss: 0.0014\n",
      "215/484, train_loss: 0.3454\n",
      "216/484, train_loss: 0.0050\n",
      "217/484, train_loss: 0.0927\n",
      "218/484, train_loss: 0.0011\n",
      "219/484, train_loss: 0.0047\n",
      "220/484, train_loss: 0.0027\n",
      "221/484, train_loss: 0.0039\n",
      "222/484, train_loss: 0.0056\n",
      "223/484, train_loss: 0.0044\n",
      "224/484, train_loss: 0.0031\n",
      "225/484, train_loss: 0.0375\n",
      "226/484, train_loss: 0.0008\n",
      "227/484, train_loss: 0.0013\n",
      "228/484, train_loss: 0.0048\n",
      "229/484, train_loss: 0.0012\n",
      "230/484, train_loss: 0.0013\n",
      "231/484, train_loss: 0.0047\n",
      "232/484, train_loss: 0.0011\n",
      "233/484, train_loss: 0.3424\n",
      "234/484, train_loss: 0.0026\n",
      "235/484, train_loss: 0.0025\n",
      "236/484, train_loss: 0.0015\n",
      "237/484, train_loss: 0.0041\n",
      "238/484, train_loss: 0.1203\n",
      "239/484, train_loss: 0.0033\n",
      "240/484, train_loss: 0.3315\n",
      "241/484, train_loss: 0.0018\n",
      "242/484, train_loss: 0.0025\n",
      "243/484, train_loss: 0.0013\n",
      "244/484, train_loss: 0.0035\n",
      "245/484, train_loss: 0.0053\n",
      "246/484, train_loss: 0.0018\n",
      "247/484, train_loss: 0.0202\n",
      "248/484, train_loss: 0.0137\n",
      "249/484, train_loss: 0.2221\n",
      "250/484, train_loss: 0.0047\n",
      "251/484, train_loss: 0.0038\n",
      "252/484, train_loss: 0.0025\n",
      "253/484, train_loss: 0.9795\n",
      "254/484, train_loss: 0.0075\n",
      "255/484, train_loss: 0.0058\n",
      "256/484, train_loss: 0.0011\n",
      "257/484, train_loss: 0.0014\n",
      "258/484, train_loss: 0.0062\n",
      "259/484, train_loss: 0.0803\n",
      "260/484, train_loss: 0.0041\n",
      "261/484, train_loss: 0.0034\n",
      "262/484, train_loss: 0.9870\n",
      "263/484, train_loss: 0.0047\n",
      "264/484, train_loss: 0.0034\n",
      "265/484, train_loss: 0.0086\n",
      "266/484, train_loss: 0.0049\n",
      "267/484, train_loss: 0.5464\n",
      "268/484, train_loss: 0.0343\n",
      "269/484, train_loss: 0.0017\n",
      "270/484, train_loss: 0.0011\n",
      "271/484, train_loss: 0.2138\n",
      "272/484, train_loss: 0.0879\n",
      "273/484, train_loss: 0.1467\n",
      "274/484, train_loss: 0.0069\n",
      "275/484, train_loss: 0.0263\n",
      "276/484, train_loss: 0.0045\n",
      "277/484, train_loss: 0.0020\n",
      "278/484, train_loss: 0.0011\n",
      "279/484, train_loss: 0.0041\n",
      "280/484, train_loss: 0.1343\n",
      "281/484, train_loss: 0.0009\n",
      "282/484, train_loss: 0.0035\n",
      "283/484, train_loss: 0.0051\n",
      "284/484, train_loss: 0.0021\n",
      "285/484, train_loss: 0.0016\n",
      "286/484, train_loss: 0.0020\n",
      "287/484, train_loss: 0.0046\n",
      "288/484, train_loss: 0.0031\n",
      "289/484, train_loss: 0.0080\n",
      "290/484, train_loss: 0.0623\n",
      "291/484, train_loss: 0.0079\n",
      "292/484, train_loss: 0.0028\n",
      "293/484, train_loss: 0.0014\n",
      "294/484, train_loss: 0.0041\n",
      "295/484, train_loss: 0.0842\n",
      "296/484, train_loss: 0.0042\n",
      "297/484, train_loss: 0.0153\n",
      "298/484, train_loss: 0.0014\n",
      "299/484, train_loss: 1.3978\n",
      "300/484, train_loss: 0.0023\n",
      "301/484, train_loss: 0.0011\n",
      "302/484, train_loss: 0.0054\n",
      "303/484, train_loss: 0.0085\n",
      "304/484, train_loss: 0.0050\n",
      "305/484, train_loss: 0.0010\n",
      "306/484, train_loss: 0.0042\n",
      "307/484, train_loss: 1.5029\n",
      "308/484, train_loss: 0.0010\n",
      "309/484, train_loss: 0.1780\n",
      "310/484, train_loss: 0.0035\n",
      "311/484, train_loss: 0.0025\n",
      "312/484, train_loss: 0.0008\n",
      "313/484, train_loss: 0.0063\n",
      "314/484, train_loss: 0.0028\n",
      "315/484, train_loss: 0.1814\n",
      "316/484, train_loss: 0.0431\n",
      "317/484, train_loss: 0.0011\n",
      "318/484, train_loss: 1.1040\n",
      "319/484, train_loss: 0.4895\n",
      "320/484, train_loss: 0.0011\n",
      "321/484, train_loss: 0.0052\n",
      "322/484, train_loss: 0.8595\n",
      "323/484, train_loss: 0.6823\n",
      "324/484, train_loss: 0.0046\n",
      "325/484, train_loss: 0.0041\n",
      "326/484, train_loss: 0.2864\n",
      "327/484, train_loss: 1.1468\n",
      "328/484, train_loss: 0.0051\n",
      "329/484, train_loss: 0.0013\n",
      "330/484, train_loss: 0.0030\n",
      "331/484, train_loss: 0.0013\n",
      "332/484, train_loss: 2.3804\n",
      "333/484, train_loss: 0.0078\n",
      "334/484, train_loss: 0.0049\n",
      "335/484, train_loss: 1.3877\n",
      "336/484, train_loss: 0.0017\n",
      "337/484, train_loss: 1.4705\n",
      "338/484, train_loss: 0.0335\n",
      "339/484, train_loss: 0.0028\n",
      "340/484, train_loss: 1.4219\n",
      "341/484, train_loss: 0.0026\n",
      "342/484, train_loss: 0.8986\n",
      "343/484, train_loss: 0.0045\n",
      "344/484, train_loss: 0.4616\n",
      "345/484, train_loss: 0.0029\n",
      "346/484, train_loss: 0.0096\n",
      "347/484, train_loss: 0.0057\n",
      "348/484, train_loss: 0.0024\n",
      "349/484, train_loss: 0.0682\n",
      "350/484, train_loss: 0.0071\n",
      "351/484, train_loss: 0.0041\n",
      "352/484, train_loss: 0.0050\n",
      "353/484, train_loss: 0.1783\n",
      "354/484, train_loss: 0.0071\n",
      "355/484, train_loss: 0.0228\n",
      "356/484, train_loss: 0.0021\n",
      "357/484, train_loss: 0.9200\n",
      "358/484, train_loss: 0.0035\n",
      "359/484, train_loss: 0.0066\n",
      "360/484, train_loss: 0.4780\n",
      "361/484, train_loss: 0.0062\n",
      "362/484, train_loss: 0.0708\n",
      "363/484, train_loss: 0.0016\n",
      "364/484, train_loss: 0.0054\n",
      "365/484, train_loss: 0.7273\n",
      "366/484, train_loss: 1.1863\n",
      "367/484, train_loss: 0.0049\n",
      "368/484, train_loss: 1.0525\n",
      "369/484, train_loss: 0.0060\n",
      "370/484, train_loss: 0.0040\n",
      "371/484, train_loss: 0.7629\n",
      "372/484, train_loss: 0.0176\n",
      "373/484, train_loss: 1.0627\n",
      "374/484, train_loss: 0.0100\n",
      "375/484, train_loss: 0.0035\n",
      "376/484, train_loss: 0.0109\n",
      "377/484, train_loss: 1.8229\n",
      "378/484, train_loss: 0.0031\n",
      "379/484, train_loss: 0.9778\n",
      "380/484, train_loss: 0.0087\n",
      "381/484, train_loss: 0.3461\n",
      "382/484, train_loss: 0.3721\n",
      "383/484, train_loss: 1.4834\n",
      "384/484, train_loss: 0.0471\n",
      "385/484, train_loss: 0.2336\n",
      "386/484, train_loss: 0.0260\n",
      "387/484, train_loss: 0.3218\n",
      "388/484, train_loss: 0.0989\n",
      "389/484, train_loss: 0.9347\n",
      "390/484, train_loss: 0.0596\n",
      "391/484, train_loss: 0.0223\n",
      "392/484, train_loss: 0.0036\n",
      "393/484, train_loss: 0.0176\n",
      "394/484, train_loss: 0.0127\n",
      "395/484, train_loss: 0.6530\n",
      "396/484, train_loss: 0.0114\n",
      "397/484, train_loss: 0.0113\n",
      "398/484, train_loss: 0.0116\n",
      "399/484, train_loss: 0.0043\n",
      "400/484, train_loss: 0.0150\n",
      "401/484, train_loss: 0.6726\n",
      "402/484, train_loss: 0.1684\n",
      "403/484, train_loss: 0.0054\n",
      "404/484, train_loss: 0.0418\n",
      "405/484, train_loss: 0.0062\n",
      "406/484, train_loss: 0.4237\n",
      "407/484, train_loss: 0.0066\n",
      "408/484, train_loss: 0.0713\n",
      "409/484, train_loss: 0.0062\n",
      "410/484, train_loss: 0.6387\n",
      "411/484, train_loss: 0.0525\n",
      "412/484, train_loss: 0.0316\n",
      "413/484, train_loss: 0.0621\n",
      "414/484, train_loss: 0.0114\n",
      "415/484, train_loss: 0.0190\n",
      "416/484, train_loss: 2.3151\n",
      "417/484, train_loss: 0.0065\n",
      "418/484, train_loss: 0.0323\n",
      "419/484, train_loss: 0.0049\n",
      "420/484, train_loss: 0.5580\n",
      "421/484, train_loss: 0.0138\n",
      "422/484, train_loss: 0.0293\n",
      "423/484, train_loss: 2.8332\n",
      "424/484, train_loss: 0.0137\n",
      "425/484, train_loss: 0.0062\n",
      "426/484, train_loss: 0.0381\n",
      "427/484, train_loss: 0.3997\n",
      "428/484, train_loss: 0.1370\n",
      "429/484, train_loss: 1.7956\n",
      "430/484, train_loss: 0.0049\n",
      "431/484, train_loss: 0.0086\n",
      "432/484, train_loss: 0.0045\n",
      "433/484, train_loss: 0.0168\n",
      "434/484, train_loss: 0.0037\n",
      "435/484, train_loss: 0.4664\n",
      "436/484, train_loss: 0.0042\n",
      "437/484, train_loss: 0.4900\n",
      "438/484, train_loss: 0.0041\n",
      "439/484, train_loss: 0.0050\n",
      "440/484, train_loss: 0.1169\n",
      "441/484, train_loss: 0.0049\n",
      "442/484, train_loss: 0.0038\n",
      "443/484, train_loss: 0.0099\n",
      "444/484, train_loss: 0.0033\n",
      "445/484, train_loss: 0.0078\n",
      "446/484, train_loss: 0.0039\n",
      "447/484, train_loss: 0.2803\n",
      "448/484, train_loss: 0.2457\n",
      "449/484, train_loss: 0.0059\n",
      "450/484, train_loss: 0.0252\n",
      "451/484, train_loss: 0.0200\n",
      "452/484, train_loss: 0.5614\n",
      "453/484, train_loss: 0.5034\n",
      "454/484, train_loss: 0.0180\n",
      "455/484, train_loss: 0.0052\n",
      "456/484, train_loss: 0.0392\n",
      "457/484, train_loss: 0.0055\n",
      "458/484, train_loss: 0.4311\n",
      "459/484, train_loss: 0.0056\n",
      "460/484, train_loss: 0.0051\n",
      "461/484, train_loss: 0.0054\n",
      "462/484, train_loss: 0.2448\n",
      "463/484, train_loss: 0.0258\n",
      "464/484, train_loss: 0.0037\n",
      "465/484, train_loss: 0.3381\n",
      "466/484, train_loss: 0.0068\n",
      "467/484, train_loss: 0.0045\n",
      "468/484, train_loss: 0.0184\n",
      "469/484, train_loss: 0.0041\n",
      "470/484, train_loss: 0.0065\n",
      "471/484, train_loss: 0.0043\n",
      "472/484, train_loss: 0.7248\n",
      "473/484, train_loss: 0.0220\n",
      "474/484, train_loss: 0.0050\n",
      "475/484, train_loss: 0.0204\n",
      "476/484, train_loss: 0.0172\n",
      "477/484, train_loss: 0.0178\n",
      "478/484, train_loss: 0.0039\n",
      "479/484, train_loss: 0.4926\n",
      "480/484, train_loss: 0.5535\n",
      "481/484, train_loss: 0.0032\n",
      "482/484, train_loss: 0.2770\n",
      "483/484, train_loss: 0.0046\n",
      "484/484, train_loss: 0.0070\n",
      "epoch 1 average loss: 0.1610\n",
      "----------\n",
      "epoch 2/20\n",
      "1/484, train_loss: 0.0079\n",
      "2/484, train_loss: 0.4372\n",
      "3/484, train_loss: 0.7321\n",
      "4/484, train_loss: 0.0069\n",
      "5/484, train_loss: 0.0030\n",
      "6/484, train_loss: 0.0120\n",
      "7/484, train_loss: 0.0086\n",
      "8/484, train_loss: 0.2516\n",
      "9/484, train_loss: 0.0123\n",
      "10/484, train_loss: 0.6166\n",
      "11/484, train_loss: 0.0080\n",
      "12/484, train_loss: 0.0063\n",
      "13/484, train_loss: 0.0079\n",
      "14/484, train_loss: 0.0093\n",
      "15/484, train_loss: 0.3371\n",
      "16/484, train_loss: 0.0049\n",
      "17/484, train_loss: 0.0220\n",
      "18/484, train_loss: 0.3790\n",
      "19/484, train_loss: 0.0081\n",
      "20/484, train_loss: 0.0581\n",
      "21/484, train_loss: 0.0037\n",
      "22/484, train_loss: 0.0062\n",
      "23/484, train_loss: 0.0063\n",
      "24/484, train_loss: 0.0037\n",
      "25/484, train_loss: 0.0034\n",
      "26/484, train_loss: 0.0022\n",
      "27/484, train_loss: 0.0227\n",
      "28/484, train_loss: 0.0078\n",
      "29/484, train_loss: 0.0108\n",
      "30/484, train_loss: 0.0022\n",
      "31/484, train_loss: 0.0083\n",
      "32/484, train_loss: 0.0104\n",
      "33/484, train_loss: 0.3735\n",
      "34/484, train_loss: 0.7146\n",
      "35/484, train_loss: 0.0147\n",
      "36/484, train_loss: 0.0096\n",
      "37/484, train_loss: 0.0034\n",
      "38/484, train_loss: 0.0077\n",
      "39/484, train_loss: 0.0152\n",
      "40/484, train_loss: 0.0030\n",
      "41/484, train_loss: 0.0085\n",
      "42/484, train_loss: 0.0144\n",
      "43/484, train_loss: 0.0074\n",
      "44/484, train_loss: 0.0043\n",
      "45/484, train_loss: 0.0041\n",
      "46/484, train_loss: 0.0031\n",
      "47/484, train_loss: 0.0097\n",
      "48/484, train_loss: 0.0063\n",
      "49/484, train_loss: 0.0028\n",
      "50/484, train_loss: 0.0071\n",
      "51/484, train_loss: 0.0102\n",
      "52/484, train_loss: 0.0087\n",
      "53/484, train_loss: 0.0025\n",
      "54/484, train_loss: 0.5198\n",
      "55/484, train_loss: 0.6434\n",
      "56/484, train_loss: 0.0049\n",
      "57/484, train_loss: 0.0047\n",
      "58/484, train_loss: 0.0263\n",
      "59/484, train_loss: 0.0059\n",
      "60/484, train_loss: 0.0058\n",
      "61/484, train_loss: 0.0157\n",
      "62/484, train_loss: 0.0095\n",
      "63/484, train_loss: 0.2076\n",
      "64/484, train_loss: 0.1254\n",
      "65/484, train_loss: 0.0034\n",
      "66/484, train_loss: 0.3930\n",
      "67/484, train_loss: 0.0063\n",
      "68/484, train_loss: 0.2518\n",
      "69/484, train_loss: 0.0029\n",
      "70/484, train_loss: 0.0055\n",
      "71/484, train_loss: 0.0050\n",
      "72/484, train_loss: 0.0146\n",
      "73/484, train_loss: 0.0057\n",
      "74/484, train_loss: 0.0015\n",
      "75/484, train_loss: 0.0057\n",
      "76/484, train_loss: 0.6746\n",
      "77/484, train_loss: 0.0055\n",
      "78/484, train_loss: 0.0066\n",
      "79/484, train_loss: 0.0072\n",
      "80/484, train_loss: 0.0080\n",
      "81/484, train_loss: 0.0046\n",
      "82/484, train_loss: 0.0021\n",
      "83/484, train_loss: 0.0542\n",
      "84/484, train_loss: 0.0163\n",
      "85/484, train_loss: 0.0016\n",
      "86/484, train_loss: 0.0031\n",
      "87/484, train_loss: 0.0511\n",
      "88/484, train_loss: 0.4758\n",
      "89/484, train_loss: 0.0014\n",
      "90/484, train_loss: 0.6849\n",
      "91/484, train_loss: 0.0068\n",
      "92/484, train_loss: 0.0085\n",
      "93/484, train_loss: 0.0032\n",
      "94/484, train_loss: 0.0028\n",
      "95/484, train_loss: 0.0017\n",
      "96/484, train_loss: 0.0084\n",
      "97/484, train_loss: 0.0075\n",
      "98/484, train_loss: 0.2528\n",
      "99/484, train_loss: 0.0036\n",
      "100/484, train_loss: 0.0034\n",
      "101/484, train_loss: 0.0049\n",
      "102/484, train_loss: 0.0063\n",
      "103/484, train_loss: 0.0036\n",
      "104/484, train_loss: 0.0115\n",
      "105/484, train_loss: 0.0078\n",
      "106/484, train_loss: 0.0054\n",
      "107/484, train_loss: 0.0119\n",
      "108/484, train_loss: 0.0028\n",
      "109/484, train_loss: 0.0034\n",
      "110/484, train_loss: 0.0021\n",
      "111/484, train_loss: 0.4466\n",
      "112/484, train_loss: 0.0034\n",
      "113/484, train_loss: 0.0022\n",
      "114/484, train_loss: 0.4110\n",
      "115/484, train_loss: 0.0050\n",
      "116/484, train_loss: 0.0015\n",
      "117/484, train_loss: 0.0049\n",
      "118/484, train_loss: 0.7025\n",
      "119/484, train_loss: 0.0065\n",
      "120/484, train_loss: 0.0054\n",
      "121/484, train_loss: 0.0017\n",
      "122/484, train_loss: 0.0063\n",
      "123/484, train_loss: 0.0037\n",
      "124/484, train_loss: 0.0019\n",
      "125/484, train_loss: 0.5167\n",
      "126/484, train_loss: 0.0037\n",
      "127/484, train_loss: 0.0012\n",
      "128/484, train_loss: 0.0116\n",
      "129/484, train_loss: 0.5869\n",
      "130/484, train_loss: 0.0031\n",
      "131/484, train_loss: 0.9413\n",
      "132/484, train_loss: 0.0070\n",
      "133/484, train_loss: 0.0086\n",
      "134/484, train_loss: 0.0320\n",
      "135/484, train_loss: 0.0026\n",
      "136/484, train_loss: 0.0224\n",
      "137/484, train_loss: 0.0041\n",
      "138/484, train_loss: 0.2809\n",
      "139/484, train_loss: 0.0034\n",
      "140/484, train_loss: 0.0192\n",
      "141/484, train_loss: 0.0076\n",
      "142/484, train_loss: 0.0079\n",
      "143/484, train_loss: 0.0077\n",
      "144/484, train_loss: 0.4491\n",
      "145/484, train_loss: 0.3103\n",
      "146/484, train_loss: 0.0056\n",
      "147/484, train_loss: 0.1264\n",
      "148/484, train_loss: 0.0022\n",
      "149/484, train_loss: 0.0030\n",
      "150/484, train_loss: 0.0116\n",
      "151/484, train_loss: 0.0138\n",
      "152/484, train_loss: 0.0025\n",
      "153/484, train_loss: 0.0040\n",
      "154/484, train_loss: 0.0054\n",
      "155/484, train_loss: 0.0028\n",
      "156/484, train_loss: 0.0064\n",
      "157/484, train_loss: 0.0036\n",
      "158/484, train_loss: 0.0052\n",
      "159/484, train_loss: 0.1944\n",
      "160/484, train_loss: 0.2349\n",
      "161/484, train_loss: 0.0049\n",
      "162/484, train_loss: 0.0044\n",
      "163/484, train_loss: 0.0054\n",
      "164/484, train_loss: 0.0064\n",
      "165/484, train_loss: 0.0019\n",
      "166/484, train_loss: 0.6610\n",
      "167/484, train_loss: 0.0072\n",
      "168/484, train_loss: 0.0060\n",
      "169/484, train_loss: 0.0306\n",
      "170/484, train_loss: 0.0025\n",
      "171/484, train_loss: 0.0057\n",
      "172/484, train_loss: 0.0037\n",
      "173/484, train_loss: 0.0097\n",
      "174/484, train_loss: 0.0184\n",
      "175/484, train_loss: 0.0038\n",
      "176/484, train_loss: 0.0018\n",
      "177/484, train_loss: 0.0017\n",
      "178/484, train_loss: 0.9574\n",
      "179/484, train_loss: 0.0020\n",
      "180/484, train_loss: 0.0059\n",
      "181/484, train_loss: 0.0066\n",
      "182/484, train_loss: 0.0042\n",
      "183/484, train_loss: 0.0044\n",
      "184/484, train_loss: 0.0037\n",
      "185/484, train_loss: 0.0063\n",
      "186/484, train_loss: 0.6360\n",
      "187/484, train_loss: 0.0024\n",
      "188/484, train_loss: 0.0043\n",
      "189/484, train_loss: 0.0053\n",
      "190/484, train_loss: 0.0054\n",
      "191/484, train_loss: 0.0034\n",
      "192/484, train_loss: 0.0053\n",
      "193/484, train_loss: 0.0048\n",
      "194/484, train_loss: 0.0036\n",
      "195/484, train_loss: 0.0100\n",
      "196/484, train_loss: 0.0069\n",
      "197/484, train_loss: 0.0016\n",
      "198/484, train_loss: 0.0027\n",
      "199/484, train_loss: 0.0029\n",
      "200/484, train_loss: 0.0035\n",
      "201/484, train_loss: 0.0031\n",
      "202/484, train_loss: 0.0032\n",
      "203/484, train_loss: 0.3215\n",
      "204/484, train_loss: 0.0027\n",
      "205/484, train_loss: 0.5173\n",
      "206/484, train_loss: 0.0031\n",
      "207/484, train_loss: 0.0038\n",
      "208/484, train_loss: 0.0032\n",
      "209/484, train_loss: 0.0063\n",
      "210/484, train_loss: 0.0029\n",
      "211/484, train_loss: 0.0051\n",
      "212/484, train_loss: 0.0031\n",
      "213/484, train_loss: 0.0086\n",
      "214/484, train_loss: 0.0055\n",
      "215/484, train_loss: 0.0019\n",
      "216/484, train_loss: 0.0042\n",
      "217/484, train_loss: 0.0061\n",
      "218/484, train_loss: 0.0037\n",
      "219/484, train_loss: 0.0032\n",
      "220/484, train_loss: 0.0030\n",
      "221/484, train_loss: 0.0027\n",
      "222/484, train_loss: 0.0024\n",
      "223/484, train_loss: 0.0041\n",
      "224/484, train_loss: 0.0907\n",
      "225/484, train_loss: 0.0102\n",
      "226/484, train_loss: 0.0230\n",
      "227/484, train_loss: 0.0030\n",
      "228/484, train_loss: 0.0026\n",
      "229/484, train_loss: 0.0049\n",
      "230/484, train_loss: 0.0033\n",
      "231/484, train_loss: 0.0043\n",
      "232/484, train_loss: 0.0043\n",
      "233/484, train_loss: 0.0052\n",
      "234/484, train_loss: 0.0063\n",
      "235/484, train_loss: 0.8566\n",
      "236/484, train_loss: 0.2720\n",
      "237/484, train_loss: 0.0061\n",
      "238/484, train_loss: 0.0012\n",
      "239/484, train_loss: 0.2995\n",
      "240/484, train_loss: 0.0048\n",
      "241/484, train_loss: 0.2615\n",
      "242/484, train_loss: 0.0018\n",
      "243/484, train_loss: 0.0029\n",
      "244/484, train_loss: 0.0088\n",
      "245/484, train_loss: 0.0033\n",
      "246/484, train_loss: 0.0071\n",
      "247/484, train_loss: 0.0128\n",
      "248/484, train_loss: 0.0048\n",
      "249/484, train_loss: 0.4260\n",
      "250/484, train_loss: 0.2176\n",
      "251/484, train_loss: 0.0032\n",
      "252/484, train_loss: 0.6250\n",
      "253/484, train_loss: 0.9200\n",
      "254/484, train_loss: 0.0018\n",
      "255/484, train_loss: 0.0012\n",
      "256/484, train_loss: 0.0020\n",
      "257/484, train_loss: 0.0031\n",
      "258/484, train_loss: 0.0036\n",
      "259/484, train_loss: 0.0011\n",
      "260/484, train_loss: 0.0029\n",
      "261/484, train_loss: 0.1683\n",
      "262/484, train_loss: 0.0037\n",
      "263/484, train_loss: 0.0147\n",
      "264/484, train_loss: 0.0033\n",
      "265/484, train_loss: 0.0036\n",
      "266/484, train_loss: 0.0024\n",
      "267/484, train_loss: 0.0028\n",
      "268/484, train_loss: 0.1534\n",
      "269/484, train_loss: 0.7750\n",
      "270/484, train_loss: 0.2896\n",
      "271/484, train_loss: 0.0060\n",
      "272/484, train_loss: 0.2404\n",
      "273/484, train_loss: 0.0041\n",
      "274/484, train_loss: 0.0018\n",
      "275/484, train_loss: 0.7011\n",
      "276/484, train_loss: 0.2525\n",
      "277/484, train_loss: 0.0017\n",
      "278/484, train_loss: 1.3332\n",
      "279/484, train_loss: 0.0012\n",
      "280/484, train_loss: 0.0035\n",
      "281/484, train_loss: 0.0023\n",
      "282/484, train_loss: 0.0088\n",
      "283/484, train_loss: 0.0183\n",
      "284/484, train_loss: 0.9243\n",
      "285/484, train_loss: 0.0039\n",
      "286/484, train_loss: 0.0032\n",
      "287/484, train_loss: 0.0036\n",
      "288/484, train_loss: 0.0293\n",
      "289/484, train_loss: 0.0027\n",
      "290/484, train_loss: 0.3502\n",
      "291/484, train_loss: 0.0034\n",
      "292/484, train_loss: 0.1448\n",
      "293/484, train_loss: 0.0024\n",
      "294/484, train_loss: 0.0031\n",
      "295/484, train_loss: 0.0081\n",
      "296/484, train_loss: 0.7564\n",
      "297/484, train_loss: 0.0054\n",
      "298/484, train_loss: 0.0013\n",
      "299/484, train_loss: 0.0047\n",
      "300/484, train_loss: 0.7466\n",
      "301/484, train_loss: 0.2737\n",
      "302/484, train_loss: 0.0015\n",
      "303/484, train_loss: 0.0022\n",
      "304/484, train_loss: 0.0086\n",
      "305/484, train_loss: 0.0058\n",
      "306/484, train_loss: 0.0050\n",
      "307/484, train_loss: 0.0053\n",
      "308/484, train_loss: 0.0383\n",
      "309/484, train_loss: 0.0019\n",
      "310/484, train_loss: 0.0127\n",
      "311/484, train_loss: 0.0026\n",
      "312/484, train_loss: 0.0912\n",
      "313/484, train_loss: 0.0040\n",
      "314/484, train_loss: 0.3685\n",
      "315/484, train_loss: 0.0161\n",
      "316/484, train_loss: 0.0056\n",
      "317/484, train_loss: 0.0078\n",
      "318/484, train_loss: 0.0025\n",
      "319/484, train_loss: 0.0032\n",
      "320/484, train_loss: 0.0058\n",
      "321/484, train_loss: 0.0051\n",
      "322/484, train_loss: 0.0111\n",
      "323/484, train_loss: 0.0344\n",
      "324/484, train_loss: 0.0032\n",
      "325/484, train_loss: 0.0070\n",
      "326/484, train_loss: 0.0019\n",
      "327/484, train_loss: 0.0018\n",
      "328/484, train_loss: 0.0038\n",
      "329/484, train_loss: 0.3228\n",
      "330/484, train_loss: 0.5329\n",
      "331/484, train_loss: 0.0013\n",
      "332/484, train_loss: 0.4549\n",
      "333/484, train_loss: 0.0023\n",
      "334/484, train_loss: 0.0075\n",
      "335/484, train_loss: 0.8203\n",
      "336/484, train_loss: 0.0060\n",
      "337/484, train_loss: 0.0089\n",
      "338/484, train_loss: 0.3802\n",
      "339/484, train_loss: 0.0034\n",
      "340/484, train_loss: 0.0022\n",
      "341/484, train_loss: 0.0100\n",
      "342/484, train_loss: 0.0042\n",
      "343/484, train_loss: 0.0015\n",
      "344/484, train_loss: 0.0034\n",
      "345/484, train_loss: 0.0061\n",
      "346/484, train_loss: 0.0041\n",
      "347/484, train_loss: 0.0049\n",
      "348/484, train_loss: 0.0020\n",
      "349/484, train_loss: 0.0032\n",
      "350/484, train_loss: 0.0044\n",
      "351/484, train_loss: 1.0181\n",
      "352/484, train_loss: 0.0110\n",
      "353/484, train_loss: 0.0032\n",
      "354/484, train_loss: 0.0036\n",
      "355/484, train_loss: 0.0993\n",
      "356/484, train_loss: 0.8182\n",
      "357/484, train_loss: 0.0018\n",
      "358/484, train_loss: 0.0031\n",
      "359/484, train_loss: 0.0027\n",
      "360/484, train_loss: 0.0040\n",
      "361/484, train_loss: 0.0126\n",
      "362/484, train_loss: 0.0019\n",
      "363/484, train_loss: 0.0029\n",
      "364/484, train_loss: 0.0026\n",
      "365/484, train_loss: 0.0040\n",
      "366/484, train_loss: 0.0109\n",
      "367/484, train_loss: 0.0038\n",
      "368/484, train_loss: 0.0019\n",
      "369/484, train_loss: 0.4363\n",
      "370/484, train_loss: 0.0039\n",
      "371/484, train_loss: 0.0094\n",
      "372/484, train_loss: 1.7998\n",
      "373/484, train_loss: 0.9481\n",
      "374/484, train_loss: 0.0070\n",
      "375/484, train_loss: 0.0038\n",
      "376/484, train_loss: 0.0147\n",
      "377/484, train_loss: 0.0101\n",
      "378/484, train_loss: 0.0020\n",
      "379/484, train_loss: 0.0335\n",
      "380/484, train_loss: 0.0025\n",
      "381/484, train_loss: 0.0184\n",
      "382/484, train_loss: 0.0049\n",
      "383/484, train_loss: 0.0031\n",
      "384/484, train_loss: 0.0114\n",
      "385/484, train_loss: 1.7896\n",
      "386/484, train_loss: 0.7650\n",
      "387/484, train_loss: 2.6526\n",
      "388/484, train_loss: 1.9404\n",
      "389/484, train_loss: 0.0102\n",
      "390/484, train_loss: 0.0030\n",
      "391/484, train_loss: 0.0042\n",
      "392/484, train_loss: 0.4438\n",
      "393/484, train_loss: 0.0202\n",
      "394/484, train_loss: 2.1838\n",
      "395/484, train_loss: 0.0951\n",
      "396/484, train_loss: 0.0189\n",
      "397/484, train_loss: 0.7192\n",
      "398/484, train_loss: 0.0120\n",
      "399/484, train_loss: 1.1505\n",
      "400/484, train_loss: 0.0210\n",
      "401/484, train_loss: 0.0065\n",
      "402/484, train_loss: 0.0573\n",
      "403/484, train_loss: 0.0030\n",
      "404/484, train_loss: 0.0082\n",
      "405/484, train_loss: 0.0194\n",
      "406/484, train_loss: 0.0182\n",
      "407/484, train_loss: 0.0063\n",
      "408/484, train_loss: 0.5679\n",
      "409/484, train_loss: 0.0943\n",
      "410/484, train_loss: 0.2093\n",
      "411/484, train_loss: 0.0064\n",
      "412/484, train_loss: 0.0096\n",
      "413/484, train_loss: 1.0550\n",
      "414/484, train_loss: 0.9293\n",
      "415/484, train_loss: 0.0141\n",
      "416/484, train_loss: 0.0142\n",
      "417/484, train_loss: 0.0116\n",
      "418/484, train_loss: 0.0186\n",
      "419/484, train_loss: 0.0298\n",
      "420/484, train_loss: 0.0063\n",
      "421/484, train_loss: 0.0140\n",
      "422/484, train_loss: 0.7012\n",
      "423/484, train_loss: 0.0078\n",
      "424/484, train_loss: 0.0149\n",
      "425/484, train_loss: 0.5773\n",
      "426/484, train_loss: 0.0201\n",
      "427/484, train_loss: 0.2433\n",
      "428/484, train_loss: 0.0069\n",
      "429/484, train_loss: 0.0057\n",
      "430/484, train_loss: 0.0082\n",
      "431/484, train_loss: 0.6109\n",
      "432/484, train_loss: 0.0103\n",
      "433/484, train_loss: 0.0038\n",
      "434/484, train_loss: 0.6669\n",
      "435/484, train_loss: 0.5697\n",
      "436/484, train_loss: 0.0359\n",
      "437/484, train_loss: 0.7003\n",
      "438/484, train_loss: 0.5972\n",
      "439/484, train_loss: 1.1218\n",
      "440/484, train_loss: 0.9816\n",
      "441/484, train_loss: 0.4574\n",
      "442/484, train_loss: 0.0044\n",
      "443/484, train_loss: 0.0062\n",
      "444/484, train_loss: 0.0936\n",
      "445/484, train_loss: 0.6661\n",
      "446/484, train_loss: 0.4201\n",
      "447/484, train_loss: 0.0070\n",
      "448/484, train_loss: 0.0303\n",
      "449/484, train_loss: 0.0780\n",
      "450/484, train_loss: 0.0064\n",
      "451/484, train_loss: 0.2026\n",
      "452/484, train_loss: 0.0046\n",
      "453/484, train_loss: 0.0357\n",
      "454/484, train_loss: 0.0109\n",
      "455/484, train_loss: 0.4110\n",
      "456/484, train_loss: 0.0050\n",
      "457/484, train_loss: 0.0071\n",
      "458/484, train_loss: 0.0051\n",
      "459/484, train_loss: 0.0252\n",
      "460/484, train_loss: 0.4239\n",
      "461/484, train_loss: 0.5619\n",
      "462/484, train_loss: 0.0119\n",
      "463/484, train_loss: 0.0886\n",
      "464/484, train_loss: 0.0681\n",
      "465/484, train_loss: 0.5261\n",
      "466/484, train_loss: 0.6517\n",
      "467/484, train_loss: 0.0086\n",
      "468/484, train_loss: 0.2742\n",
      "469/484, train_loss: 0.0039\n",
      "470/484, train_loss: 0.1206\n",
      "471/484, train_loss: 0.2965\n",
      "472/484, train_loss: 0.0172\n",
      "473/484, train_loss: 0.0683\n",
      "474/484, train_loss: 0.5272\n",
      "475/484, train_loss: 0.0256\n",
      "476/484, train_loss: 0.0122\n",
      "477/484, train_loss: 0.0085\n",
      "478/484, train_loss: 0.0044\n",
      "479/484, train_loss: 0.0095\n",
      "480/484, train_loss: 0.0031\n",
      "481/484, train_loss: 0.0045\n",
      "482/484, train_loss: 0.0090\n",
      "483/484, train_loss: 0.0030\n",
      "484/484, train_loss: 0.0194\n",
      "epoch 2 average loss: 0.1357\n",
      "saved new best metric model\n",
      "Current epoch: 2 current accuracy: 0.7461 \n",
      "Best accuracy: 0.7461 at epoch 2\n",
      "----------\n",
      "epoch 3/20\n",
      "1/484, train_loss: 0.5679\n",
      "2/484, train_loss: 0.0160\n",
      "3/484, train_loss: 0.0919\n",
      "4/484, train_loss: 0.0109\n",
      "5/484, train_loss: 0.0214\n",
      "6/484, train_loss: 0.1544\n",
      "7/484, train_loss: 0.3029\n",
      "8/484, train_loss: 0.8496\n",
      "9/484, train_loss: 0.0100\n",
      "10/484, train_loss: 0.0122\n",
      "11/484, train_loss: 0.0097\n",
      "12/484, train_loss: 0.0031\n",
      "13/484, train_loss: 0.6040\n",
      "14/484, train_loss: 0.0066\n",
      "15/484, train_loss: 0.0052\n",
      "16/484, train_loss: 0.4395\n",
      "17/484, train_loss: 0.0153\n",
      "18/484, train_loss: 0.0038\n",
      "19/484, train_loss: 0.0037\n",
      "20/484, train_loss: 0.0108\n",
      "21/484, train_loss: 0.0071\n",
      "22/484, train_loss: 0.8743\n",
      "23/484, train_loss: 0.0050\n",
      "24/484, train_loss: 0.0069\n",
      "25/484, train_loss: 0.4407\n",
      "26/484, train_loss: 0.0034\n",
      "27/484, train_loss: 0.0042\n",
      "28/484, train_loss: 0.0025\n",
      "29/484, train_loss: 0.0484\n",
      "30/484, train_loss: 0.0151\n",
      "31/484, train_loss: 0.0363\n",
      "32/484, train_loss: 0.0030\n",
      "33/484, train_loss: 0.0029\n",
      "34/484, train_loss: 0.0044\n",
      "35/484, train_loss: 0.0059\n",
      "36/484, train_loss: 0.2450\n",
      "37/484, train_loss: 0.0076\n",
      "38/484, train_loss: 0.1284\n",
      "39/484, train_loss: 0.0205\n",
      "40/484, train_loss: 0.0056\n",
      "41/484, train_loss: 0.0071\n",
      "42/484, train_loss: 0.3547\n",
      "43/484, train_loss: 0.0054\n",
      "44/484, train_loss: 0.0147\n",
      "45/484, train_loss: 0.0152\n",
      "46/484, train_loss: 0.0067\n",
      "47/484, train_loss: 0.0135\n",
      "48/484, train_loss: 0.0999\n",
      "49/484, train_loss: 0.0034\n",
      "50/484, train_loss: 0.0058\n",
      "51/484, train_loss: 0.0933\n",
      "52/484, train_loss: 0.0175\n",
      "53/484, train_loss: 0.0037\n",
      "54/484, train_loss: 0.0032\n",
      "55/484, train_loss: 0.0053\n",
      "56/484, train_loss: 0.0091\n",
      "57/484, train_loss: 0.0120\n",
      "58/484, train_loss: 0.0030\n",
      "59/484, train_loss: 0.0021\n",
      "60/484, train_loss: 0.0051\n",
      "61/484, train_loss: 0.0194\n",
      "62/484, train_loss: 0.3460\n",
      "63/484, train_loss: 0.0075\n",
      "64/484, train_loss: 0.0272\n",
      "65/484, train_loss: 0.0084\n",
      "66/484, train_loss: 0.0051\n",
      "67/484, train_loss: 0.0050\n",
      "68/484, train_loss: 0.0035\n",
      "69/484, train_loss: 0.1521\n",
      "70/484, train_loss: 0.1047\n",
      "71/484, train_loss: 0.0055\n",
      "72/484, train_loss: 0.0093\n",
      "73/484, train_loss: 0.0041\n",
      "74/484, train_loss: 0.0110\n",
      "75/484, train_loss: 0.0067\n",
      "76/484, train_loss: 0.0134\n",
      "77/484, train_loss: 0.0026\n",
      "78/484, train_loss: 0.0616\n",
      "79/484, train_loss: 0.0804\n",
      "80/484, train_loss: 0.0201\n",
      "81/484, train_loss: 0.0059\n",
      "82/484, train_loss: 0.0064\n",
      "83/484, train_loss: 0.0048\n",
      "84/484, train_loss: 0.0035\n",
      "85/484, train_loss: 0.0067\n",
      "86/484, train_loss: 1.6744\n",
      "87/484, train_loss: 0.0059\n",
      "88/484, train_loss: 0.0164\n",
      "89/484, train_loss: 0.0126\n",
      "90/484, train_loss: 0.0058\n",
      "91/484, train_loss: 0.0261\n",
      "92/484, train_loss: 0.0122\n",
      "93/484, train_loss: 0.0049\n",
      "94/484, train_loss: 0.0039\n",
      "95/484, train_loss: 0.0068\n",
      "96/484, train_loss: 0.0070\n",
      "97/484, train_loss: 0.0086\n",
      "98/484, train_loss: 0.0035\n",
      "99/484, train_loss: 0.0081\n",
      "100/484, train_loss: 0.0033\n",
      "101/484, train_loss: 0.0255\n",
      "102/484, train_loss: 0.0029\n",
      "103/484, train_loss: 0.0049\n",
      "104/484, train_loss: 0.0037\n",
      "105/484, train_loss: 0.0199\n",
      "106/484, train_loss: 0.0054\n",
      "107/484, train_loss: 0.0031\n",
      "108/484, train_loss: 0.4221\n",
      "109/484, train_loss: 0.0046\n",
      "110/484, train_loss: 0.0053\n",
      "111/484, train_loss: 0.0081\n",
      "112/484, train_loss: 0.0036\n",
      "113/484, train_loss: 0.3106\n",
      "114/484, train_loss: 0.0049\n",
      "115/484, train_loss: 0.0058\n",
      "116/484, train_loss: 0.0021\n",
      "117/484, train_loss: 0.0061\n",
      "118/484, train_loss: 0.0116\n",
      "119/484, train_loss: 0.2014\n",
      "120/484, train_loss: 0.0041\n",
      "121/484, train_loss: 0.0085\n",
      "122/484, train_loss: 0.0049\n",
      "123/484, train_loss: 0.9214\n",
      "124/484, train_loss: 0.0103\n",
      "125/484, train_loss: 0.0030\n",
      "126/484, train_loss: 0.0075\n",
      "127/484, train_loss: 0.0029\n",
      "128/484, train_loss: 0.0113\n",
      "129/484, train_loss: 0.7549\n",
      "130/484, train_loss: 0.0082\n",
      "131/484, train_loss: 0.0070\n",
      "132/484, train_loss: 0.5276\n",
      "133/484, train_loss: 0.0053\n",
      "134/484, train_loss: 0.7226\n",
      "135/484, train_loss: 0.0035\n",
      "136/484, train_loss: 0.3410\n",
      "137/484, train_loss: 0.1684\n",
      "138/484, train_loss: 0.0053\n",
      "139/484, train_loss: 0.0122\n",
      "140/484, train_loss: 0.0076\n",
      "141/484, train_loss: 0.0155\n",
      "142/484, train_loss: 0.4653\n",
      "143/484, train_loss: 0.0052\n",
      "144/484, train_loss: 0.0057\n",
      "145/484, train_loss: 0.0025\n",
      "146/484, train_loss: 0.3424\n",
      "147/484, train_loss: 0.0046\n",
      "148/484, train_loss: 0.0026\n",
      "149/484, train_loss: 0.0040\n",
      "150/484, train_loss: 0.0037\n",
      "151/484, train_loss: 0.0034\n",
      "152/484, train_loss: 0.0052\n",
      "153/484, train_loss: 0.0109\n",
      "154/484, train_loss: 0.3426\n",
      "155/484, train_loss: 0.0071\n",
      "156/484, train_loss: 0.7149\n",
      "157/484, train_loss: 0.0029\n",
      "158/484, train_loss: 0.0258\n",
      "159/484, train_loss: 0.0096\n",
      "160/484, train_loss: 0.0116\n",
      "161/484, train_loss: 0.0056\n",
      "162/484, train_loss: 1.2320\n",
      "163/484, train_loss: 0.0038\n",
      "164/484, train_loss: 0.0022\n",
      "165/484, train_loss: 0.1422\n",
      "166/484, train_loss: 0.0061\n",
      "167/484, train_loss: 0.0033\n",
      "168/484, train_loss: 0.0047\n",
      "169/484, train_loss: 0.0112\n",
      "170/484, train_loss: 0.0132\n",
      "171/484, train_loss: 0.0027\n",
      "172/484, train_loss: 0.0154\n",
      "173/484, train_loss: 0.0033\n",
      "174/484, train_loss: 0.0104\n",
      "175/484, train_loss: 0.0045\n",
      "176/484, train_loss: 0.0101\n",
      "177/484, train_loss: 0.0155\n",
      "178/484, train_loss: 0.1849\n",
      "179/484, train_loss: 0.0210\n",
      "180/484, train_loss: 0.0021\n",
      "181/484, train_loss: 0.3377\n",
      "182/484, train_loss: 0.0079\n",
      "183/484, train_loss: 0.0078\n",
      "184/484, train_loss: 0.0071\n",
      "185/484, train_loss: 0.0067\n",
      "186/484, train_loss: 0.0068\n",
      "187/484, train_loss: 0.1123\n",
      "188/484, train_loss: 0.0044\n",
      "189/484, train_loss: 0.0054\n",
      "190/484, train_loss: 0.5324\n",
      "191/484, train_loss: 0.0023\n",
      "192/484, train_loss: 0.0029\n",
      "193/484, train_loss: 0.4743\n",
      "194/484, train_loss: 0.0055\n",
      "195/484, train_loss: 0.0200\n",
      "196/484, train_loss: 0.2107\n",
      "197/484, train_loss: 0.0025\n",
      "198/484, train_loss: 0.0406\n",
      "199/484, train_loss: 0.0037\n",
      "200/484, train_loss: 0.0054\n",
      "201/484, train_loss: 0.3519\n",
      "202/484, train_loss: 0.0019\n",
      "203/484, train_loss: 0.4940\n",
      "204/484, train_loss: 0.0066\n",
      "205/484, train_loss: 0.0092\n",
      "206/484, train_loss: 0.0039\n",
      "207/484, train_loss: 0.0202\n",
      "208/484, train_loss: 0.0133\n",
      "209/484, train_loss: 0.0089\n",
      "210/484, train_loss: 0.0028\n",
      "211/484, train_loss: 0.0327\n",
      "212/484, train_loss: 0.0038\n",
      "213/484, train_loss: 0.0038\n",
      "214/484, train_loss: 0.3757\n",
      "215/484, train_loss: 0.0066\n",
      "216/484, train_loss: 0.0036\n",
      "217/484, train_loss: 0.0043\n",
      "218/484, train_loss: 0.0050\n",
      "219/484, train_loss: 0.0053\n",
      "220/484, train_loss: 0.2286\n",
      "221/484, train_loss: 0.0046\n",
      "222/484, train_loss: 0.0031\n",
      "223/484, train_loss: 0.0041\n",
      "224/484, train_loss: 0.0129\n",
      "225/484, train_loss: 0.0253\n",
      "226/484, train_loss: 0.0053\n",
      "227/484, train_loss: 0.0183\n",
      "228/484, train_loss: 0.0081\n",
      "229/484, train_loss: 0.0034\n",
      "230/484, train_loss: 0.0040\n",
      "231/484, train_loss: 0.0164\n",
      "232/484, train_loss: 0.0026\n",
      "233/484, train_loss: 0.0178\n",
      "234/484, train_loss: 0.0058\n",
      "235/484, train_loss: 0.0145\n",
      "236/484, train_loss: 0.0080\n",
      "237/484, train_loss: 0.1174\n",
      "238/484, train_loss: 0.0033\n",
      "239/484, train_loss: 0.0029\n",
      "240/484, train_loss: 0.0043\n",
      "241/484, train_loss: 0.0436\n",
      "242/484, train_loss: 0.0081\n",
      "243/484, train_loss: 0.0027\n",
      "244/484, train_loss: 0.0024\n",
      "245/484, train_loss: 0.0061\n",
      "246/484, train_loss: 0.2285\n",
      "247/484, train_loss: 0.0021\n",
      "248/484, train_loss: 0.1659\n",
      "249/484, train_loss: 0.0034\n",
      "250/484, train_loss: 0.0042\n",
      "251/484, train_loss: 0.0028\n",
      "252/484, train_loss: 0.0036\n",
      "253/484, train_loss: 0.0096\n",
      "254/484, train_loss: 0.0158\n",
      "255/484, train_loss: 0.0041\n",
      "256/484, train_loss: 0.0542\n",
      "257/484, train_loss: 0.0032\n",
      "258/484, train_loss: 0.0033\n",
      "259/484, train_loss: 0.0034\n",
      "260/484, train_loss: 0.0032\n",
      "261/484, train_loss: 0.0553\n",
      "262/484, train_loss: 0.2651\n",
      "263/484, train_loss: 0.0156\n",
      "264/484, train_loss: 0.0036\n",
      "265/484, train_loss: 0.0957\n",
      "266/484, train_loss: 0.0094\n",
      "267/484, train_loss: 0.0055\n",
      "268/484, train_loss: 0.2225\n",
      "269/484, train_loss: 0.2206\n",
      "270/484, train_loss: 0.2969\n",
      "271/484, train_loss: 0.0083\n",
      "272/484, train_loss: 0.0269\n",
      "273/484, train_loss: 0.0031\n",
      "274/484, train_loss: 0.0036\n",
      "275/484, train_loss: 0.0035\n",
      "276/484, train_loss: 0.0077\n",
      "277/484, train_loss: 0.0034\n",
      "278/484, train_loss: 0.1571\n",
      "279/484, train_loss: 0.0957\n",
      "280/484, train_loss: 0.0057\n",
      "281/484, train_loss: 0.0034\n",
      "282/484, train_loss: 0.0093\n",
      "283/484, train_loss: 0.0045\n",
      "284/484, train_loss: 0.0025\n",
      "285/484, train_loss: 0.0026\n",
      "286/484, train_loss: 0.0038\n",
      "287/484, train_loss: 1.1704\n",
      "288/484, train_loss: 0.1719\n",
      "289/484, train_loss: 0.0026\n",
      "290/484, train_loss: 0.0033\n",
      "291/484, train_loss: 0.0637\n",
      "292/484, train_loss: 0.0043\n",
      "293/484, train_loss: 0.0052\n",
      "294/484, train_loss: 0.0113\n",
      "295/484, train_loss: 1.6726\n",
      "296/484, train_loss: 0.0081\n",
      "297/484, train_loss: 0.0094\n",
      "298/484, train_loss: 0.0067\n",
      "299/484, train_loss: 0.0035\n",
      "300/484, train_loss: 0.0077\n",
      "301/484, train_loss: 1.3643\n",
      "302/484, train_loss: 0.0093\n",
      "303/484, train_loss: 0.0086\n",
      "304/484, train_loss: 0.0079\n",
      "305/484, train_loss: 0.0035\n",
      "306/484, train_loss: 0.0024\n",
      "307/484, train_loss: 0.0039\n",
      "308/484, train_loss: 0.2316\n",
      "309/484, train_loss: 0.0019\n",
      "310/484, train_loss: 0.0018\n",
      "311/484, train_loss: 0.0035\n",
      "312/484, train_loss: 0.0025\n",
      "313/484, train_loss: 0.0038\n",
      "314/484, train_loss: 0.0062\n",
      "315/484, train_loss: 0.0038\n",
      "316/484, train_loss: 0.0051\n",
      "317/484, train_loss: 0.0073\n",
      "318/484, train_loss: 0.0105\n",
      "319/484, train_loss: 0.0015\n",
      "320/484, train_loss: 0.0030\n",
      "321/484, train_loss: 0.0021\n",
      "322/484, train_loss: 0.0025\n",
      "323/484, train_loss: 0.0024\n",
      "324/484, train_loss: 0.0025\n",
      "325/484, train_loss: 0.0069\n",
      "326/484, train_loss: 0.0030\n",
      "327/484, train_loss: 0.0043\n",
      "328/484, train_loss: 0.0023\n",
      "329/484, train_loss: 0.0067\n",
      "330/484, train_loss: 0.0028\n",
      "331/484, train_loss: 0.0038\n",
      "332/484, train_loss: 0.0087\n",
      "333/484, train_loss: 0.0039\n",
      "334/484, train_loss: 0.0125\n",
      "335/484, train_loss: 0.0023\n",
      "336/484, train_loss: 0.0097\n",
      "337/484, train_loss: 0.6790\n",
      "338/484, train_loss: 0.0029\n",
      "339/484, train_loss: 0.0033\n",
      "340/484, train_loss: 0.0027\n",
      "341/484, train_loss: 0.0056\n",
      "342/484, train_loss: 0.0215\n",
      "343/484, train_loss: 0.0040\n",
      "344/484, train_loss: 0.0080\n",
      "345/484, train_loss: 0.0171\n",
      "346/484, train_loss: 0.0043\n",
      "347/484, train_loss: 0.0050\n",
      "348/484, train_loss: 0.0042\n",
      "349/484, train_loss: 0.0040\n",
      "350/484, train_loss: 0.0062\n",
      "351/484, train_loss: 0.0044\n",
      "352/484, train_loss: 0.0035\n",
      "353/484, train_loss: 0.0070\n",
      "354/484, train_loss: 0.0032\n",
      "355/484, train_loss: 0.0037\n",
      "356/484, train_loss: 0.0057\n",
      "357/484, train_loss: 0.0065\n",
      "358/484, train_loss: 0.0032\n",
      "359/484, train_loss: 0.0136\n",
      "360/484, train_loss: 0.0158\n",
      "361/484, train_loss: 0.0106\n",
      "362/484, train_loss: 0.0026\n",
      "363/484, train_loss: 0.0020\n",
      "364/484, train_loss: 0.2814\n",
      "365/484, train_loss: 0.0025\n",
      "366/484, train_loss: 0.0471\n",
      "367/484, train_loss: 0.0111\n",
      "368/484, train_loss: 0.0041\n",
      "369/484, train_loss: 0.0022\n",
      "370/484, train_loss: 0.0051\n",
      "371/484, train_loss: 0.0033\n",
      "372/484, train_loss: 0.0043\n",
      "373/484, train_loss: 0.8745\n",
      "374/484, train_loss: 0.0058\n",
      "375/484, train_loss: 0.0054\n",
      "376/484, train_loss: 0.0021\n",
      "377/484, train_loss: 0.0103\n",
      "378/484, train_loss: 0.0040\n",
      "379/484, train_loss: 0.0021\n",
      "380/484, train_loss: 0.0074\n",
      "381/484, train_loss: 0.0031\n",
      "382/484, train_loss: 0.0024\n",
      "383/484, train_loss: 0.0035\n",
      "384/484, train_loss: 0.0106\n",
      "385/484, train_loss: 0.2633\n",
      "386/484, train_loss: 0.0090\n",
      "387/484, train_loss: 0.0036\n",
      "388/484, train_loss: 0.5734\n",
      "389/484, train_loss: 0.3458\n",
      "390/484, train_loss: 0.5329\n",
      "391/484, train_loss: 0.0048\n",
      "392/484, train_loss: 0.0022\n",
      "393/484, train_loss: 0.0045\n",
      "394/484, train_loss: 0.0015\n",
      "395/484, train_loss: 0.0138\n",
      "396/484, train_loss: 0.0026\n",
      "397/484, train_loss: 0.0096\n",
      "398/484, train_loss: 0.0095\n",
      "399/484, train_loss: 0.0046\n",
      "400/484, train_loss: 0.0045\n",
      "401/484, train_loss: 0.0053\n",
      "402/484, train_loss: 0.0052\n",
      "403/484, train_loss: 0.0021\n",
      "404/484, train_loss: 0.4065\n",
      "405/484, train_loss: 0.0042\n",
      "406/484, train_loss: 0.0105\n",
      "407/484, train_loss: 0.0052\n",
      "408/484, train_loss: 0.0040\n",
      "409/484, train_loss: 0.0034\n",
      "410/484, train_loss: 0.0020\n",
      "411/484, train_loss: 0.0047\n",
      "412/484, train_loss: 0.0030\n",
      "413/484, train_loss: 0.0039\n",
      "414/484, train_loss: 0.0041\n",
      "415/484, train_loss: 0.9658\n",
      "416/484, train_loss: 0.0022\n",
      "417/484, train_loss: 0.0039\n",
      "418/484, train_loss: 0.0088\n",
      "419/484, train_loss: 0.0050\n",
      "420/484, train_loss: 0.0105\n",
      "421/484, train_loss: 0.0042\n",
      "422/484, train_loss: 0.0028\n",
      "423/484, train_loss: 0.0136\n",
      "424/484, train_loss: 0.0032\n",
      "425/484, train_loss: 0.0141\n",
      "426/484, train_loss: 0.0042\n",
      "427/484, train_loss: 0.0019\n",
      "428/484, train_loss: 0.0117\n",
      "429/484, train_loss: 0.0042\n",
      "430/484, train_loss: 0.0020\n",
      "431/484, train_loss: 0.0310\n",
      "432/484, train_loss: 1.1190\n",
      "433/484, train_loss: 0.0057\n",
      "434/484, train_loss: 0.0038\n",
      "435/484, train_loss: 0.0039\n",
      "436/484, train_loss: 0.0035\n",
      "437/484, train_loss: 0.0028\n",
      "438/484, train_loss: 0.0051\n",
      "439/484, train_loss: 0.0060\n",
      "440/484, train_loss: 0.0055\n",
      "441/484, train_loss: 0.0068\n",
      "442/484, train_loss: 0.0059\n",
      "443/484, train_loss: 0.0035\n",
      "444/484, train_loss: 0.2986\n",
      "445/484, train_loss: 0.0064\n",
      "446/484, train_loss: 0.0032\n",
      "447/484, train_loss: 0.0056\n",
      "448/484, train_loss: 0.0039\n",
      "449/484, train_loss: 0.0036\n",
      "450/484, train_loss: 0.0031\n",
      "451/484, train_loss: 0.0047\n",
      "452/484, train_loss: 0.0019\n",
      "453/484, train_loss: 0.0029\n",
      "454/484, train_loss: 0.0038\n",
      "455/484, train_loss: 0.0194\n",
      "456/484, train_loss: 0.0598\n",
      "457/484, train_loss: 0.0032\n",
      "458/484, train_loss: 0.0030\n",
      "459/484, train_loss: 0.0067\n",
      "460/484, train_loss: 0.0023\n",
      "461/484, train_loss: 0.0685\n",
      "462/484, train_loss: 0.0054\n",
      "463/484, train_loss: 0.0044\n",
      "464/484, train_loss: 0.0113\n",
      "465/484, train_loss: 0.1652\n",
      "466/484, train_loss: 0.1985\n",
      "467/484, train_loss: 0.0085\n",
      "468/484, train_loss: 0.0025\n",
      "469/484, train_loss: 0.0051\n",
      "470/484, train_loss: 0.0047\n",
      "471/484, train_loss: 0.0025\n",
      "472/484, train_loss: 0.0027\n",
      "473/484, train_loss: 0.0033\n",
      "474/484, train_loss: 0.0067\n",
      "475/484, train_loss: 0.0054\n",
      "476/484, train_loss: 0.0589\n",
      "477/484, train_loss: 0.0032\n",
      "478/484, train_loss: 0.8262\n",
      "479/484, train_loss: 0.5676\n",
      "480/484, train_loss: 0.2459\n",
      "481/484, train_loss: 0.0054\n",
      "482/484, train_loss: 0.0123\n",
      "483/484, train_loss: 0.0044\n",
      "484/484, train_loss: 0.0038\n",
      "epoch 3 average loss: 0.0757\n",
      "----------\n",
      "epoch 4/20\n",
      "1/484, train_loss: 0.0026\n",
      "2/484, train_loss: 0.3272\n",
      "3/484, train_loss: 0.0067\n",
      "4/484, train_loss: 0.0050\n",
      "5/484, train_loss: 0.0018\n",
      "6/484, train_loss: 0.1967\n",
      "7/484, train_loss: 0.1607\n",
      "8/484, train_loss: 0.0025\n",
      "9/484, train_loss: 0.0046\n",
      "10/484, train_loss: 0.0031\n",
      "11/484, train_loss: 0.0069\n",
      "12/484, train_loss: 0.0029\n",
      "13/484, train_loss: 0.0037\n",
      "14/484, train_loss: 0.2600\n",
      "15/484, train_loss: 0.1851\n",
      "16/484, train_loss: 0.0087\n",
      "17/484, train_loss: 0.0021\n",
      "18/484, train_loss: 0.0048\n",
      "19/484, train_loss: 0.0062\n",
      "20/484, train_loss: 0.0017\n",
      "21/484, train_loss: 0.0026\n",
      "22/484, train_loss: 0.0037\n",
      "23/484, train_loss: 0.0044\n",
      "24/484, train_loss: 0.0021\n",
      "25/484, train_loss: 0.5784\n",
      "26/484, train_loss: 0.0038\n",
      "27/484, train_loss: 0.0023\n",
      "28/484, train_loss: 0.0031\n",
      "29/484, train_loss: 0.0035\n",
      "30/484, train_loss: 0.0025\n",
      "31/484, train_loss: 0.0021\n",
      "32/484, train_loss: 0.0039\n",
      "33/484, train_loss: 0.0016\n",
      "34/484, train_loss: 0.0057\n",
      "35/484, train_loss: 0.1215\n",
      "36/484, train_loss: 0.0026\n",
      "37/484, train_loss: 0.0066\n",
      "38/484, train_loss: 1.3536\n",
      "39/484, train_loss: 0.0055\n",
      "40/484, train_loss: 0.0042\n",
      "41/484, train_loss: 0.0040\n",
      "42/484, train_loss: 1.2779\n",
      "43/484, train_loss: 0.0139\n",
      "44/484, train_loss: 0.0023\n",
      "45/484, train_loss: 0.0021\n",
      "46/484, train_loss: 0.0050\n",
      "47/484, train_loss: 0.0703\n",
      "48/484, train_loss: 0.0082\n",
      "49/484, train_loss: 0.0106\n",
      "50/484, train_loss: 0.0040\n",
      "51/484, train_loss: 0.0019\n",
      "52/484, train_loss: 0.0051\n",
      "53/484, train_loss: 0.5019\n",
      "54/484, train_loss: 0.0041\n",
      "55/484, train_loss: 0.0058\n",
      "56/484, train_loss: 0.0035\n",
      "57/484, train_loss: 0.0026\n",
      "58/484, train_loss: 0.4815\n",
      "59/484, train_loss: 0.0035\n",
      "60/484, train_loss: 0.1901\n",
      "61/484, train_loss: 0.0061\n",
      "62/484, train_loss: 0.0077\n",
      "63/484, train_loss: 0.0031\n",
      "64/484, train_loss: 0.0027\n",
      "65/484, train_loss: 0.0154\n",
      "66/484, train_loss: 0.3172\n",
      "67/484, train_loss: 0.0040\n",
      "68/484, train_loss: 0.6180\n",
      "69/484, train_loss: 0.0055\n",
      "70/484, train_loss: 0.0025\n",
      "71/484, train_loss: 0.0039\n",
      "72/484, train_loss: 0.3048\n",
      "73/484, train_loss: 0.0022\n",
      "74/484, train_loss: 0.0572\n",
      "75/484, train_loss: 0.0082\n",
      "76/484, train_loss: 0.0050\n",
      "77/484, train_loss: 0.0041\n",
      "78/484, train_loss: 0.0041\n",
      "79/484, train_loss: 0.0109\n",
      "80/484, train_loss: 0.0038\n",
      "81/484, train_loss: 0.0026\n",
      "82/484, train_loss: 0.0156\n",
      "83/484, train_loss: 0.0032\n",
      "84/484, train_loss: 0.0066\n",
      "85/484, train_loss: 0.0026\n",
      "86/484, train_loss: 0.0052\n",
      "87/484, train_loss: 0.0028\n",
      "88/484, train_loss: 0.0044\n",
      "89/484, train_loss: 0.0022\n",
      "90/484, train_loss: 0.0022\n",
      "91/484, train_loss: 0.0033\n",
      "92/484, train_loss: 0.0073\n",
      "93/484, train_loss: 0.0059\n",
      "94/484, train_loss: 0.0031\n",
      "95/484, train_loss: 0.0062\n",
      "96/484, train_loss: 0.0028\n",
      "97/484, train_loss: 0.4424\n",
      "98/484, train_loss: 0.0016\n",
      "99/484, train_loss: 0.0657\n",
      "100/484, train_loss: 0.0047\n",
      "101/484, train_loss: 0.0036\n",
      "102/484, train_loss: 0.0033\n",
      "103/484, train_loss: 0.0105\n",
      "104/484, train_loss: 0.0199\n",
      "105/484, train_loss: 0.0022\n",
      "106/484, train_loss: 0.0075\n",
      "107/484, train_loss: 0.0035\n",
      "108/484, train_loss: 0.0041\n",
      "109/484, train_loss: 0.0027\n",
      "110/484, train_loss: 0.0025\n",
      "111/484, train_loss: 0.2093\n",
      "112/484, train_loss: 0.0025\n",
      "113/484, train_loss: 0.0024\n",
      "114/484, train_loss: 0.3129\n",
      "115/484, train_loss: 0.0037\n",
      "116/484, train_loss: 0.0026\n",
      "117/484, train_loss: 0.0080\n",
      "118/484, train_loss: 0.0046\n",
      "119/484, train_loss: 0.0054\n",
      "120/484, train_loss: 0.0053\n",
      "121/484, train_loss: 0.0074\n",
      "122/484, train_loss: 0.0135\n",
      "123/484, train_loss: 0.0017\n",
      "124/484, train_loss: 0.0025\n",
      "125/484, train_loss: 0.8200\n",
      "126/484, train_loss: 0.8420\n",
      "127/484, train_loss: 0.0130\n",
      "128/484, train_loss: 0.0029\n",
      "129/484, train_loss: 0.7396\n",
      "130/484, train_loss: 0.0016\n",
      "131/484, train_loss: 0.0040\n",
      "132/484, train_loss: 0.1693\n",
      "133/484, train_loss: 0.0030\n",
      "134/484, train_loss: 0.0035\n",
      "135/484, train_loss: 0.2318\n",
      "136/484, train_loss: 0.0019\n",
      "137/484, train_loss: 0.0034\n",
      "138/484, train_loss: 0.0026\n",
      "139/484, train_loss: 0.4578\n",
      "140/484, train_loss: 0.0027\n",
      "141/484, train_loss: 0.0039\n",
      "142/484, train_loss: 0.0028\n",
      "143/484, train_loss: 0.0032\n",
      "144/484, train_loss: 0.4163\n",
      "145/484, train_loss: 0.0073\n",
      "146/484, train_loss: 0.0031\n",
      "147/484, train_loss: 0.0059\n",
      "148/484, train_loss: 0.0020\n",
      "149/484, train_loss: 0.0018\n",
      "150/484, train_loss: 0.0058\n",
      "151/484, train_loss: 0.4046\n",
      "152/484, train_loss: 0.0019\n",
      "153/484, train_loss: 0.0026\n",
      "154/484, train_loss: 0.0083\n",
      "155/484, train_loss: 0.0020\n",
      "156/484, train_loss: 0.0029\n",
      "157/484, train_loss: 0.0029\n",
      "158/484, train_loss: 0.0019\n",
      "159/484, train_loss: 0.0026\n",
      "160/484, train_loss: 0.0075\n",
      "161/484, train_loss: 0.0035\n",
      "162/484, train_loss: 0.1922\n",
      "163/484, train_loss: 0.0053\n",
      "164/484, train_loss: 0.0024\n",
      "165/484, train_loss: 0.5567\n",
      "166/484, train_loss: 0.0025\n",
      "167/484, train_loss: 0.0699\n",
      "168/484, train_loss: 0.0032\n",
      "169/484, train_loss: 0.0074\n",
      "170/484, train_loss: 0.0022\n",
      "171/484, train_loss: 0.5599\n",
      "172/484, train_loss: 0.0025\n",
      "173/484, train_loss: 0.0050\n",
      "174/484, train_loss: 0.0017\n",
      "175/484, train_loss: 0.1655\n",
      "176/484, train_loss: 0.2378\n",
      "177/484, train_loss: 0.0030\n",
      "178/484, train_loss: 0.4717\n",
      "179/484, train_loss: 0.0037\n",
      "180/484, train_loss: 0.0014\n",
      "181/484, train_loss: 0.0044\n",
      "182/484, train_loss: 0.0034\n",
      "183/484, train_loss: 0.0222\n",
      "184/484, train_loss: 0.5035\n",
      "185/484, train_loss: 0.0031\n",
      "186/484, train_loss: 0.8272\n",
      "187/484, train_loss: 0.1084\n",
      "188/484, train_loss: 0.0041\n",
      "189/484, train_loss: 0.0064\n",
      "190/484, train_loss: 0.0078\n",
      "191/484, train_loss: 0.4519\n",
      "192/484, train_loss: 0.0042\n",
      "193/484, train_loss: 0.0043\n",
      "194/484, train_loss: 0.0042\n",
      "195/484, train_loss: 0.0016\n",
      "196/484, train_loss: 0.1053\n",
      "197/484, train_loss: 0.0107\n",
      "198/484, train_loss: 0.0079\n",
      "199/484, train_loss: 0.0026\n",
      "200/484, train_loss: 0.0032\n",
      "201/484, train_loss: 0.0088\n",
      "202/484, train_loss: 0.0170\n",
      "203/484, train_loss: 0.0028\n",
      "204/484, train_loss: 0.0940\n",
      "205/484, train_loss: 0.0126\n",
      "206/484, train_loss: 0.0023\n",
      "207/484, train_loss: 0.0051\n",
      "208/484, train_loss: 0.7866\n",
      "209/484, train_loss: 0.0074\n",
      "210/484, train_loss: 0.0035\n",
      "211/484, train_loss: 0.5081\n",
      "212/484, train_loss: 0.0031\n",
      "213/484, train_loss: 0.1291\n",
      "214/484, train_loss: 0.0046\n",
      "215/484, train_loss: 0.0061\n",
      "216/484, train_loss: 0.0060\n",
      "217/484, train_loss: 0.0080\n",
      "218/484, train_loss: 0.0015\n",
      "219/484, train_loss: 0.0039\n",
      "220/484, train_loss: 0.0047\n",
      "221/484, train_loss: 0.0019\n",
      "222/484, train_loss: 0.0014\n",
      "223/484, train_loss: 0.0091\n",
      "224/484, train_loss: 0.0145\n",
      "225/484, train_loss: 0.0087\n",
      "226/484, train_loss: 0.0017\n",
      "227/484, train_loss: 0.1869\n",
      "228/484, train_loss: 0.0021\n",
      "229/484, train_loss: 0.0076\n",
      "230/484, train_loss: 0.0037\n",
      "231/484, train_loss: 0.0049\n",
      "232/484, train_loss: 0.0125\n",
      "233/484, train_loss: 0.0038\n",
      "234/484, train_loss: 0.0031\n",
      "235/484, train_loss: 0.0044\n",
      "236/484, train_loss: 0.0036\n",
      "237/484, train_loss: 0.0047\n",
      "238/484, train_loss: 0.0037\n",
      "239/484, train_loss: 0.5495\n",
      "240/484, train_loss: 0.1954\n",
      "241/484, train_loss: 0.0018\n",
      "242/484, train_loss: 0.3780\n",
      "243/484, train_loss: 0.0054\n",
      "244/484, train_loss: 0.0294\n",
      "245/484, train_loss: 0.0049\n",
      "246/484, train_loss: 0.0028\n",
      "247/484, train_loss: 0.0034\n",
      "248/484, train_loss: 0.0031\n",
      "249/484, train_loss: 0.0029\n",
      "250/484, train_loss: 0.0021\n",
      "251/484, train_loss: 0.0057\n",
      "252/484, train_loss: 0.0050\n",
      "253/484, train_loss: 0.4787\n",
      "254/484, train_loss: 0.4988\n",
      "255/484, train_loss: 0.0026\n",
      "256/484, train_loss: 0.0038\n",
      "257/484, train_loss: 0.0069\n",
      "258/484, train_loss: 0.0054\n",
      "259/484, train_loss: 0.0033\n",
      "260/484, train_loss: 0.0047\n",
      "261/484, train_loss: 0.0060\n",
      "262/484, train_loss: 0.0037\n",
      "263/484, train_loss: 0.0016\n",
      "264/484, train_loss: 0.0042\n",
      "265/484, train_loss: 0.0031\n",
      "266/484, train_loss: 0.0032\n",
      "267/484, train_loss: 0.4563\n",
      "268/484, train_loss: 0.0080\n",
      "269/484, train_loss: 0.7305\n",
      "270/484, train_loss: 0.0110\n",
      "271/484, train_loss: 0.0031\n",
      "272/484, train_loss: 0.0042\n",
      "273/484, train_loss: 0.1945\n",
      "274/484, train_loss: 0.0022\n",
      "275/484, train_loss: 0.0078\n",
      "276/484, train_loss: 0.0079\n",
      "277/484, train_loss: 0.0081\n",
      "278/484, train_loss: 0.0023\n",
      "279/484, train_loss: 0.0039\n",
      "280/484, train_loss: 0.0053\n",
      "281/484, train_loss: 0.0026\n",
      "282/484, train_loss: 0.0063\n",
      "283/484, train_loss: 0.0056\n",
      "284/484, train_loss: 0.0333\n",
      "285/484, train_loss: 0.0032\n",
      "286/484, train_loss: 0.3385\n",
      "287/484, train_loss: 0.0018\n",
      "288/484, train_loss: 0.0038\n",
      "289/484, train_loss: 0.0038\n",
      "290/484, train_loss: 0.7697\n",
      "291/484, train_loss: 0.0041\n",
      "292/484, train_loss: 0.0016\n",
      "293/484, train_loss: 0.0049\n",
      "294/484, train_loss: 0.0035\n",
      "295/484, train_loss: 0.3324\n",
      "296/484, train_loss: 0.0016\n",
      "297/484, train_loss: 0.0036\n",
      "298/484, train_loss: 0.0013\n",
      "299/484, train_loss: 0.5876\n",
      "300/484, train_loss: 0.0037\n",
      "301/484, train_loss: 0.0022\n",
      "302/484, train_loss: 0.2161\n",
      "303/484, train_loss: 1.4050\n",
      "304/484, train_loss: 0.2115\n",
      "305/484, train_loss: 0.1427\n",
      "306/484, train_loss: 0.0049\n",
      "307/484, train_loss: 0.2019\n",
      "308/484, train_loss: 0.0026\n",
      "309/484, train_loss: 0.0017\n",
      "310/484, train_loss: 0.5074\n",
      "311/484, train_loss: 0.0063\n",
      "312/484, train_loss: 0.0030\n",
      "313/484, train_loss: 0.0038\n",
      "314/484, train_loss: 0.0026\n",
      "315/484, train_loss: 0.0017\n",
      "316/484, train_loss: 0.0039\n",
      "317/484, train_loss: 0.0015\n",
      "318/484, train_loss: 0.0028\n",
      "319/484, train_loss: 0.0039\n",
      "320/484, train_loss: 0.0020\n",
      "321/484, train_loss: 0.0016\n",
      "322/484, train_loss: 0.0649\n",
      "323/484, train_loss: 0.0021\n",
      "324/484, train_loss: 0.0141\n",
      "325/484, train_loss: 0.0116\n",
      "326/484, train_loss: 0.0040\n",
      "327/484, train_loss: 0.0205\n",
      "328/484, train_loss: 0.0023\n",
      "329/484, train_loss: 0.0048\n",
      "330/484, train_loss: 0.0026\n",
      "331/484, train_loss: 0.0039\n",
      "332/484, train_loss: 0.0099\n",
      "333/484, train_loss: 0.0014\n",
      "334/484, train_loss: 0.0021\n",
      "335/484, train_loss: 0.0013\n",
      "336/484, train_loss: 0.0011\n",
      "337/484, train_loss: 0.0060\n",
      "338/484, train_loss: 1.1607\n",
      "339/484, train_loss: 0.4485\n",
      "340/484, train_loss: 0.1791\n",
      "341/484, train_loss: 0.0026\n",
      "342/484, train_loss: 0.0024\n",
      "343/484, train_loss: 0.5201\n",
      "344/484, train_loss: 0.0026\n",
      "345/484, train_loss: 0.2528\n",
      "346/484, train_loss: 0.0016\n",
      "347/484, train_loss: 0.0019\n",
      "348/484, train_loss: 0.5720\n",
      "349/484, train_loss: 0.0036\n",
      "350/484, train_loss: 0.0018\n",
      "351/484, train_loss: 0.0026\n",
      "352/484, train_loss: 0.0018\n",
      "353/484, train_loss: 0.0031\n",
      "354/484, train_loss: 0.0022\n",
      "355/484, train_loss: 0.0027\n",
      "356/484, train_loss: 0.0029\n",
      "357/484, train_loss: 0.0014\n",
      "358/484, train_loss: 0.0020\n",
      "359/484, train_loss: 0.0016\n",
      "360/484, train_loss: 0.0052\n",
      "361/484, train_loss: 1.1030\n",
      "362/484, train_loss: 0.4228\n",
      "363/484, train_loss: 0.0013\n",
      "364/484, train_loss: 0.0026\n",
      "365/484, train_loss: 0.0022\n",
      "366/484, train_loss: 0.0052\n",
      "367/484, train_loss: 0.0105\n",
      "368/484, train_loss: 0.0019\n",
      "369/484, train_loss: 0.0108\n",
      "370/484, train_loss: 0.0014\n",
      "371/484, train_loss: 0.0043\n",
      "372/484, train_loss: 0.0038\n",
      "373/484, train_loss: 0.0025\n",
      "374/484, train_loss: 0.2317\n",
      "375/484, train_loss: 0.0020\n",
      "376/484, train_loss: 0.5038\n",
      "377/484, train_loss: 0.2223\n",
      "378/484, train_loss: 0.0058\n",
      "379/484, train_loss: 0.2987\n",
      "380/484, train_loss: 0.0058\n",
      "381/484, train_loss: 0.0056\n",
      "382/484, train_loss: 0.0027\n",
      "383/484, train_loss: 0.0089\n",
      "384/484, train_loss: 0.0034\n",
      "385/484, train_loss: 0.0047\n",
      "386/484, train_loss: 0.0058\n",
      "387/484, train_loss: 1.2138\n",
      "388/484, train_loss: 0.9454\n",
      "389/484, train_loss: 0.0061\n",
      "390/484, train_loss: 0.0089\n",
      "391/484, train_loss: 0.0110\n",
      "392/484, train_loss: 0.0027\n",
      "393/484, train_loss: 1.2897\n",
      "394/484, train_loss: 0.0017\n",
      "395/484, train_loss: 0.5103\n",
      "396/484, train_loss: 0.0032\n",
      "397/484, train_loss: 0.0058\n",
      "398/484, train_loss: 0.0019\n",
      "399/484, train_loss: 0.0047\n",
      "400/484, train_loss: 0.0031\n",
      "401/484, train_loss: 0.0041\n",
      "402/484, train_loss: 0.0069\n",
      "403/484, train_loss: 0.2947\n",
      "404/484, train_loss: 0.0054\n",
      "405/484, train_loss: 0.0025\n",
      "406/484, train_loss: 0.0029\n",
      "407/484, train_loss: 0.0047\n",
      "408/484, train_loss: 0.0018\n",
      "409/484, train_loss: 0.0082\n",
      "410/484, train_loss: 0.0359\n",
      "411/484, train_loss: 0.0519\n",
      "412/484, train_loss: 0.5498\n",
      "413/484, train_loss: 0.0699\n",
      "414/484, train_loss: 0.6459\n",
      "415/484, train_loss: 1.5829\n",
      "416/484, train_loss: 1.3569\n",
      "417/484, train_loss: 0.0054\n",
      "418/484, train_loss: 1.4511\n",
      "419/484, train_loss: 0.0070\n",
      "420/484, train_loss: 0.0418\n",
      "421/484, train_loss: 0.3092\n",
      "422/484, train_loss: 0.1839\n",
      "423/484, train_loss: 0.4843\n",
      "424/484, train_loss: 0.1628\n",
      "425/484, train_loss: 2.1971\n",
      "426/484, train_loss: 0.4113\n",
      "427/484, train_loss: 0.1764\n",
      "428/484, train_loss: 0.0078\n",
      "429/484, train_loss: 0.0155\n",
      "430/484, train_loss: 0.0627\n",
      "431/484, train_loss: 0.1089\n",
      "432/484, train_loss: 0.0275\n",
      "433/484, train_loss: 0.0823\n",
      "434/484, train_loss: 0.0170\n",
      "435/484, train_loss: 0.0095\n",
      "436/484, train_loss: 0.0088\n",
      "437/484, train_loss: 0.0060\n",
      "438/484, train_loss: 0.0052\n",
      "439/484, train_loss: 0.0144\n",
      "440/484, train_loss: 0.0781\n",
      "441/484, train_loss: 0.1401\n",
      "442/484, train_loss: 0.8963\n",
      "443/484, train_loss: 0.0615\n",
      "444/484, train_loss: 2.1690\n",
      "445/484, train_loss: 0.4933\n",
      "446/484, train_loss: 0.4877\n",
      "447/484, train_loss: 0.0128\n",
      "448/484, train_loss: 0.7087\n",
      "449/484, train_loss: 0.0077\n",
      "450/484, train_loss: 0.0210\n",
      "451/484, train_loss: 0.0706\n",
      "452/484, train_loss: 0.6336\n",
      "453/484, train_loss: 0.0055\n",
      "454/484, train_loss: 0.0065\n",
      "455/484, train_loss: 0.2652\n",
      "456/484, train_loss: 0.3895\n",
      "457/484, train_loss: 0.0025\n",
      "458/484, train_loss: 1.5692\n",
      "459/484, train_loss: 0.0056\n",
      "460/484, train_loss: 1.6050\n",
      "461/484, train_loss: 0.0062\n",
      "462/484, train_loss: 0.0061\n",
      "463/484, train_loss: 0.7830\n",
      "464/484, train_loss: 0.0065\n",
      "465/484, train_loss: 0.1617\n",
      "466/484, train_loss: 0.0791\n",
      "467/484, train_loss: 0.0072\n",
      "468/484, train_loss: 0.0107\n",
      "469/484, train_loss: 0.0636\n",
      "470/484, train_loss: 0.0051\n",
      "471/484, train_loss: 1.5301\n",
      "472/484, train_loss: 0.0028\n",
      "473/484, train_loss: 0.0201\n",
      "474/484, train_loss: 0.0051\n",
      "475/484, train_loss: 0.0048\n",
      "476/484, train_loss: 0.0092\n",
      "477/484, train_loss: 0.2942\n",
      "478/484, train_loss: 0.0125\n",
      "479/484, train_loss: 0.1450\n",
      "480/484, train_loss: 0.0102\n",
      "481/484, train_loss: 0.0193\n",
      "482/484, train_loss: 0.2116\n",
      "483/484, train_loss: 0.1243\n",
      "484/484, train_loss: 0.3832\n",
      "epoch 4 average loss: 0.1274\n",
      "saved new best metric model\n",
      "Current epoch: 4 current accuracy: 0.7895 \n",
      "Best accuracy: 0.7895 at epoch 4\n",
      "----------\n",
      "epoch 5/20\n",
      "1/484, train_loss: 0.0078\n",
      "2/484, train_loss: 0.0041\n",
      "3/484, train_loss: 0.0052\n",
      "4/484, train_loss: 0.0333\n",
      "5/484, train_loss: 0.2574\n",
      "6/484, train_loss: 0.0090\n",
      "7/484, train_loss: 0.0686\n",
      "8/484, train_loss: 0.0277\n",
      "9/484, train_loss: 0.0053\n",
      "10/484, train_loss: 1.9826\n",
      "11/484, train_loss: 0.0054\n",
      "12/484, train_loss: 0.0141\n",
      "13/484, train_loss: 0.0281\n",
      "14/484, train_loss: 0.0034\n",
      "15/484, train_loss: 0.0191\n",
      "16/484, train_loss: 0.3897\n",
      "17/484, train_loss: 0.4493\n",
      "18/484, train_loss: 0.0377\n",
      "19/484, train_loss: 0.0072\n",
      "20/484, train_loss: 0.0087\n",
      "21/484, train_loss: 0.2029\n",
      "22/484, train_loss: 0.3818\n",
      "23/484, train_loss: 0.0518\n",
      "24/484, train_loss: 0.0076\n",
      "25/484, train_loss: 0.0768\n",
      "26/484, train_loss: 0.0070\n",
      "27/484, train_loss: 0.0181\n",
      "28/484, train_loss: 0.2245\n",
      "29/484, train_loss: 0.6840\n",
      "30/484, train_loss: 0.3245\n",
      "31/484, train_loss: 1.7481\n",
      "32/484, train_loss: 0.0037\n",
      "33/484, train_loss: 0.0066\n",
      "34/484, train_loss: 0.0069\n",
      "35/484, train_loss: 0.2723\n",
      "36/484, train_loss: 0.2907\n",
      "37/484, train_loss: 0.0043\n",
      "38/484, train_loss: 0.4790\n",
      "39/484, train_loss: 0.3289\n",
      "40/484, train_loss: 0.0079\n",
      "41/484, train_loss: 0.0253\n",
      "42/484, train_loss: 0.0035\n",
      "43/484, train_loss: 0.0162\n",
      "44/484, train_loss: 0.0041\n",
      "45/484, train_loss: 0.5665\n",
      "46/484, train_loss: 0.0079\n",
      "47/484, train_loss: 0.0076\n",
      "48/484, train_loss: 0.1559\n",
      "49/484, train_loss: 0.0448\n",
      "50/484, train_loss: 0.3505\n",
      "51/484, train_loss: 0.0296\n",
      "52/484, train_loss: 0.0207\n",
      "53/484, train_loss: 0.0096\n",
      "54/484, train_loss: 0.0199\n",
      "55/484, train_loss: 0.0034\n",
      "56/484, train_loss: 0.0149\n",
      "57/484, train_loss: 0.0231\n",
      "58/484, train_loss: 0.0356\n",
      "59/484, train_loss: 0.1158\n",
      "60/484, train_loss: 0.0071\n",
      "61/484, train_loss: 0.0296\n",
      "62/484, train_loss: 0.0031\n",
      "63/484, train_loss: 0.0099\n",
      "64/484, train_loss: 0.0020\n",
      "65/484, train_loss: 0.0136\n",
      "66/484, train_loss: 0.0195\n",
      "67/484, train_loss: 0.1667\n",
      "68/484, train_loss: 0.0594\n",
      "69/484, train_loss: 0.0305\n",
      "70/484, train_loss: 0.3599\n",
      "71/484, train_loss: 0.1666\n",
      "72/484, train_loss: 0.0487\n",
      "73/484, train_loss: 1.1177\n",
      "74/484, train_loss: 1.1270\n",
      "75/484, train_loss: 0.0177\n",
      "76/484, train_loss: 0.3153\n",
      "77/484, train_loss: 0.0072\n",
      "78/484, train_loss: 0.0048\n",
      "79/484, train_loss: 0.0136\n",
      "80/484, train_loss: 0.0137\n",
      "81/484, train_loss: 0.6473\n",
      "82/484, train_loss: 0.3460\n",
      "83/484, train_loss: 0.0038\n",
      "84/484, train_loss: 0.0090\n",
      "85/484, train_loss: 0.0603\n",
      "86/484, train_loss: 0.2894\n",
      "87/484, train_loss: 0.0659\n",
      "88/484, train_loss: 0.0178\n",
      "89/484, train_loss: 0.0443\n",
      "90/484, train_loss: 0.0095\n",
      "91/484, train_loss: 0.1810\n",
      "92/484, train_loss: 0.0427\n",
      "93/484, train_loss: 0.0050\n",
      "94/484, train_loss: 0.0038\n",
      "95/484, train_loss: 0.0025\n",
      "96/484, train_loss: 0.1605\n",
      "97/484, train_loss: 0.0812\n",
      "98/484, train_loss: 0.1528\n",
      "99/484, train_loss: 0.3752\n",
      "100/484, train_loss: 0.3587\n",
      "101/484, train_loss: 0.0029\n",
      "102/484, train_loss: 0.4114\n",
      "103/484, train_loss: 0.0073\n",
      "104/484, train_loss: 0.0049\n",
      "105/484, train_loss: 1.0477\n",
      "106/484, train_loss: 1.0237\n",
      "107/484, train_loss: 0.0922\n",
      "108/484, train_loss: 0.0044\n",
      "109/484, train_loss: 4.5070\n",
      "110/484, train_loss: 2.5964\n",
      "111/484, train_loss: 1.0408\n",
      "112/484, train_loss: 0.0179\n",
      "113/484, train_loss: 0.0050\n",
      "114/484, train_loss: 0.2855\n",
      "115/484, train_loss: 0.3869\n",
      "116/484, train_loss: 1.8215\n",
      "117/484, train_loss: 0.0848\n",
      "118/484, train_loss: 0.0388\n",
      "119/484, train_loss: 0.0110\n",
      "120/484, train_loss: 0.0115\n",
      "121/484, train_loss: 0.1009\n",
      "122/484, train_loss: 0.0545\n",
      "123/484, train_loss: 1.8070\n",
      "124/484, train_loss: 0.5044\n",
      "125/484, train_loss: 0.0033\n",
      "126/484, train_loss: 0.0410\n",
      "127/484, train_loss: 0.0146\n",
      "128/484, train_loss: 0.0270\n",
      "129/484, train_loss: 0.0644\n",
      "130/484, train_loss: 0.0572\n",
      "131/484, train_loss: 0.3290\n",
      "132/484, train_loss: 0.2432\n",
      "133/484, train_loss: 0.0141\n",
      "134/484, train_loss: 1.3726\n",
      "135/484, train_loss: 0.0266\n",
      "136/484, train_loss: 0.0283\n",
      "137/484, train_loss: 0.0557\n",
      "138/484, train_loss: 0.0226\n",
      "139/484, train_loss: 0.3094\n",
      "140/484, train_loss: 0.0258\n",
      "141/484, train_loss: 0.0519\n",
      "142/484, train_loss: 0.0256\n",
      "143/484, train_loss: 0.9433\n",
      "144/484, train_loss: 0.8840\n",
      "145/484, train_loss: 0.0099\n",
      "146/484, train_loss: 0.0703\n",
      "147/484, train_loss: 0.0196\n",
      "148/484, train_loss: 0.0157\n",
      "149/484, train_loss: 0.0168\n",
      "150/484, train_loss: 0.0102\n",
      "151/484, train_loss: 0.0818\n",
      "152/484, train_loss: 0.4778\n",
      "153/484, train_loss: 0.3959\n",
      "154/484, train_loss: 0.0314\n",
      "155/484, train_loss: 0.2057\n",
      "156/484, train_loss: 0.3670\n",
      "157/484, train_loss: 0.0114\n",
      "158/484, train_loss: 0.0117\n",
      "159/484, train_loss: 0.0099\n",
      "160/484, train_loss: 0.0691\n",
      "161/484, train_loss: 0.0287\n",
      "162/484, train_loss: 0.0268\n",
      "163/484, train_loss: 0.4280\n",
      "164/484, train_loss: 0.0391\n",
      "165/484, train_loss: 0.0056\n",
      "166/484, train_loss: 0.0077\n",
      "167/484, train_loss: 0.5840\n",
      "168/484, train_loss: 0.0134\n",
      "169/484, train_loss: 0.0065\n",
      "170/484, train_loss: 0.0910\n",
      "171/484, train_loss: 1.1286\n",
      "172/484, train_loss: 0.0138\n",
      "173/484, train_loss: 0.0229\n",
      "174/484, train_loss: 0.0241\n",
      "175/484, train_loss: 1.1320\n",
      "176/484, train_loss: 0.2864\n",
      "177/484, train_loss: 0.0141\n",
      "178/484, train_loss: 0.5709\n",
      "179/484, train_loss: 0.0158\n",
      "180/484, train_loss: 0.0085\n",
      "181/484, train_loss: 0.0100\n",
      "182/484, train_loss: 0.0034\n",
      "183/484, train_loss: 0.1914\n",
      "184/484, train_loss: 0.4523\n",
      "185/484, train_loss: 0.0064\n",
      "186/484, train_loss: 0.4274\n",
      "187/484, train_loss: 0.0367\n",
      "188/484, train_loss: 0.0213\n",
      "189/484, train_loss: 0.0069\n",
      "190/484, train_loss: 0.5409\n",
      "191/484, train_loss: 0.0155\n",
      "192/484, train_loss: 0.0092\n",
      "193/484, train_loss: 0.0234\n",
      "194/484, train_loss: 0.5304\n",
      "195/484, train_loss: 0.0194\n",
      "196/484, train_loss: 0.0635\n",
      "197/484, train_loss: 0.0179\n",
      "198/484, train_loss: 0.0149\n",
      "199/484, train_loss: 0.6796\n",
      "200/484, train_loss: 0.0079\n",
      "201/484, train_loss: 0.5055\n",
      "202/484, train_loss: 0.0164\n",
      "203/484, train_loss: 0.4392\n",
      "204/484, train_loss: 0.0127\n",
      "205/484, train_loss: 0.0570\n",
      "206/484, train_loss: 0.2318\n",
      "207/484, train_loss: 0.0278\n",
      "208/484, train_loss: 0.2906\n",
      "209/484, train_loss: 0.0082\n",
      "210/484, train_loss: 0.1680\n",
      "211/484, train_loss: 0.0061\n",
      "212/484, train_loss: 0.1015\n",
      "213/484, train_loss: 0.0064\n",
      "214/484, train_loss: 0.8331\n",
      "215/484, train_loss: 0.0184\n",
      "216/484, train_loss: 0.0164\n",
      "217/484, train_loss: 0.2585\n",
      "218/484, train_loss: 0.0398\n",
      "219/484, train_loss: 0.1298\n",
      "220/484, train_loss: 0.1761\n",
      "221/484, train_loss: 0.0074\n",
      "222/484, train_loss: 0.2204\n",
      "223/484, train_loss: 0.0280\n",
      "224/484, train_loss: 0.2924\n",
      "225/484, train_loss: 0.0035\n",
      "226/484, train_loss: 0.0288\n",
      "227/484, train_loss: 0.2664\n",
      "228/484, train_loss: 0.0070\n",
      "229/484, train_loss: 0.0093\n",
      "230/484, train_loss: 0.3301\n",
      "231/484, train_loss: 0.0243\n",
      "232/484, train_loss: 1.0122\n",
      "233/484, train_loss: 0.0179\n",
      "234/484, train_loss: 0.9688\n",
      "235/484, train_loss: 0.1909\n",
      "236/484, train_loss: 0.0528\n",
      "237/484, train_loss: 0.1335\n",
      "238/484, train_loss: 0.0170\n",
      "239/484, train_loss: 0.0118\n",
      "240/484, train_loss: 0.5436\n",
      "241/484, train_loss: 0.0102\n",
      "242/484, train_loss: 0.0060\n",
      "243/484, train_loss: 0.3155\n",
      "244/484, train_loss: 0.1040\n",
      "245/484, train_loss: 0.0051\n",
      "246/484, train_loss: 0.0247\n",
      "247/484, train_loss: 0.0173\n",
      "248/484, train_loss: 0.0056\n",
      "249/484, train_loss: 0.0034\n",
      "250/484, train_loss: 0.0032\n",
      "251/484, train_loss: 0.0089\n",
      "252/484, train_loss: 0.0070\n",
      "253/484, train_loss: 0.0829\n",
      "254/484, train_loss: 0.0094\n",
      "255/484, train_loss: 0.0085\n",
      "256/484, train_loss: 0.0037\n",
      "257/484, train_loss: 0.0061\n",
      "258/484, train_loss: 0.0171\n",
      "259/484, train_loss: 0.4557\n",
      "260/484, train_loss: 0.0226\n",
      "261/484, train_loss: 0.0187\n",
      "262/484, train_loss: 0.0804\n",
      "263/484, train_loss: 0.2605\n",
      "264/484, train_loss: 0.0408\n",
      "265/484, train_loss: 0.0106\n",
      "266/484, train_loss: 0.0073\n",
      "267/484, train_loss: 0.0037\n",
      "268/484, train_loss: 0.6683\n",
      "269/484, train_loss: 0.6046\n",
      "270/484, train_loss: 0.4710\n",
      "271/484, train_loss: 0.0162\n",
      "272/484, train_loss: 0.0083\n",
      "273/484, train_loss: 0.0157\n",
      "274/484, train_loss: 0.0101\n",
      "275/484, train_loss: 0.0080\n",
      "276/484, train_loss: 0.0271\n",
      "277/484, train_loss: 0.0050\n",
      "278/484, train_loss: 0.2254\n",
      "279/484, train_loss: 0.0139\n",
      "280/484, train_loss: 0.0183\n",
      "281/484, train_loss: 0.0905\n",
      "282/484, train_loss: 0.0054\n",
      "283/484, train_loss: 0.0168\n",
      "284/484, train_loss: 0.0116\n",
      "285/484, train_loss: 0.0107\n",
      "286/484, train_loss: 0.3960\n",
      "287/484, train_loss: 0.5662\n",
      "288/484, train_loss: 0.2031\n",
      "289/484, train_loss: 0.0098\n",
      "290/484, train_loss: 0.0536\n",
      "291/484, train_loss: 0.0058\n",
      "292/484, train_loss: 0.0023\n",
      "293/484, train_loss: 0.0207\n",
      "294/484, train_loss: 0.0029\n",
      "295/484, train_loss: 0.0069\n",
      "296/484, train_loss: 0.0040\n",
      "297/484, train_loss: 0.0018\n",
      "298/484, train_loss: 0.0049\n",
      "299/484, train_loss: 0.7468\n",
      "300/484, train_loss: 0.0127\n",
      "301/484, train_loss: 0.0039\n",
      "302/484, train_loss: 0.0078\n",
      "303/484, train_loss: 0.3496\n",
      "304/484, train_loss: 0.0045\n",
      "305/484, train_loss: 0.0076\n",
      "306/484, train_loss: 0.7517\n",
      "307/484, train_loss: 0.0315\n",
      "308/484, train_loss: 0.0075\n",
      "309/484, train_loss: 0.0073\n",
      "310/484, train_loss: 0.0025\n",
      "311/484, train_loss: 0.0965\n",
      "312/484, train_loss: 0.0114\n",
      "313/484, train_loss: 0.0048\n",
      "314/484, train_loss: 0.2804\n",
      "315/484, train_loss: 0.0022\n",
      "316/484, train_loss: 0.2605\n",
      "317/484, train_loss: 0.0031\n",
      "318/484, train_loss: 0.0042\n",
      "319/484, train_loss: 0.0069\n",
      "320/484, train_loss: 0.0019\n",
      "321/484, train_loss: 0.3603\n",
      "322/484, train_loss: 0.0105\n",
      "323/484, train_loss: 0.1472\n",
      "324/484, train_loss: 0.0116\n",
      "325/484, train_loss: 0.2277\n",
      "326/484, train_loss: 1.1864\n",
      "327/484, train_loss: 0.0046\n",
      "328/484, train_loss: 0.0198\n",
      "329/484, train_loss: 0.0055\n",
      "330/484, train_loss: 0.0037\n",
      "331/484, train_loss: 0.4726\n",
      "332/484, train_loss: 0.0045\n",
      "333/484, train_loss: 0.0067\n",
      "334/484, train_loss: 0.0050\n",
      "335/484, train_loss: 0.0192\n",
      "336/484, train_loss: 0.0968\n",
      "337/484, train_loss: 0.0104\n",
      "338/484, train_loss: 0.0039\n",
      "339/484, train_loss: 0.0061\n",
      "340/484, train_loss: 0.0716\n",
      "341/484, train_loss: 0.0173\n",
      "342/484, train_loss: 0.3167\n",
      "343/484, train_loss: 0.0130\n",
      "344/484, train_loss: 0.0019\n",
      "345/484, train_loss: 0.7038\n",
      "346/484, train_loss: 0.0222\n",
      "347/484, train_loss: 0.1752\n",
      "348/484, train_loss: 0.0050\n",
      "349/484, train_loss: 0.0070\n",
      "350/484, train_loss: 0.0121\n",
      "351/484, train_loss: 0.2359\n",
      "352/484, train_loss: 0.4820\n",
      "353/484, train_loss: 0.0227\n",
      "354/484, train_loss: 0.0044\n",
      "355/484, train_loss: 0.0098\n",
      "356/484, train_loss: 0.0092\n",
      "357/484, train_loss: 0.0246\n",
      "358/484, train_loss: 0.3220\n",
      "359/484, train_loss: 0.0313\n",
      "360/484, train_loss: 0.0250\n",
      "361/484, train_loss: 0.0031\n",
      "362/484, train_loss: 0.0160\n",
      "363/484, train_loss: 0.0057\n",
      "364/484, train_loss: 0.0050\n",
      "365/484, train_loss: 0.0093\n",
      "366/484, train_loss: 0.0047\n",
      "367/484, train_loss: 0.0117\n",
      "368/484, train_loss: 0.0121\n",
      "369/484, train_loss: 0.0041\n",
      "370/484, train_loss: 0.1561\n",
      "371/484, train_loss: 0.0123\n",
      "372/484, train_loss: 0.0023\n",
      "373/484, train_loss: 0.6938\n",
      "374/484, train_loss: 0.0049\n",
      "375/484, train_loss: 0.2912\n",
      "376/484, train_loss: 0.0052\n",
      "377/484, train_loss: 0.0050\n",
      "378/484, train_loss: 1.0156\n",
      "379/484, train_loss: 0.0013\n",
      "380/484, train_loss: 0.4095\n",
      "381/484, train_loss: 0.1148\n",
      "382/484, train_loss: 0.0036\n",
      "383/484, train_loss: 0.1341\n",
      "384/484, train_loss: 0.0065\n",
      "385/484, train_loss: 1.8062\n",
      "386/484, train_loss: 0.0074\n",
      "387/484, train_loss: 0.0872\n",
      "388/484, train_loss: 0.0144\n",
      "389/484, train_loss: 0.0041\n",
      "390/484, train_loss: 0.0097\n",
      "391/484, train_loss: 0.1291\n",
      "392/484, train_loss: 0.0222\n",
      "393/484, train_loss: 0.5556\n",
      "394/484, train_loss: 0.0067\n",
      "395/484, train_loss: 0.0030\n",
      "396/484, train_loss: 0.2302\n",
      "397/484, train_loss: 0.0020\n",
      "398/484, train_loss: 0.0056\n",
      "399/484, train_loss: 0.0044\n",
      "400/484, train_loss: 0.0078\n",
      "401/484, train_loss: 0.2288\n",
      "402/484, train_loss: 0.0050\n",
      "403/484, train_loss: 0.0106\n",
      "404/484, train_loss: 0.0094\n",
      "405/484, train_loss: 0.0030\n",
      "406/484, train_loss: 0.6355\n",
      "407/484, train_loss: 0.0036\n",
      "408/484, train_loss: 0.0044\n",
      "409/484, train_loss: 0.0052\n",
      "410/484, train_loss: 0.0182\n",
      "411/484, train_loss: 0.0065\n",
      "412/484, train_loss: 0.2388\n",
      "413/484, train_loss: 0.0063\n",
      "414/484, train_loss: 0.0090\n",
      "415/484, train_loss: 0.0316\n",
      "416/484, train_loss: 0.0121\n",
      "417/484, train_loss: 0.0124\n",
      "418/484, train_loss: 0.1446\n",
      "419/484, train_loss: 0.0011\n",
      "420/484, train_loss: 0.0032\n",
      "421/484, train_loss: 0.0020\n",
      "422/484, train_loss: 0.0551\n",
      "423/484, train_loss: 0.9273\n",
      "424/484, train_loss: 0.0073\n",
      "425/484, train_loss: 0.7302\n",
      "426/484, train_loss: 0.0043\n",
      "427/484, train_loss: 0.0050\n",
      "428/484, train_loss: 0.0063\n",
      "429/484, train_loss: 0.0682\n",
      "430/484, train_loss: 0.0044\n",
      "431/484, train_loss: 1.3698\n",
      "432/484, train_loss: 0.0035\n",
      "433/484, train_loss: 0.5950\n",
      "434/484, train_loss: 0.0070\n",
      "435/484, train_loss: 0.0085\n",
      "436/484, train_loss: 0.0011\n",
      "437/484, train_loss: 0.0100\n",
      "438/484, train_loss: 0.0047\n",
      "439/484, train_loss: 0.0163\n",
      "440/484, train_loss: 0.0052\n",
      "441/484, train_loss: 0.0014\n",
      "442/484, train_loss: 0.0131\n",
      "443/484, train_loss: 0.0033\n",
      "444/484, train_loss: 0.0031\n",
      "445/484, train_loss: 0.0261\n",
      "446/484, train_loss: 0.0048\n",
      "447/484, train_loss: 0.1767\n",
      "448/484, train_loss: 0.0053\n",
      "449/484, train_loss: 0.0033\n",
      "450/484, train_loss: 0.0056\n",
      "451/484, train_loss: 0.0034\n",
      "452/484, train_loss: 0.1600\n",
      "453/484, train_loss: 0.0092\n",
      "454/484, train_loss: 0.0186\n",
      "455/484, train_loss: 0.0028\n",
      "456/484, train_loss: 0.0047\n",
      "457/484, train_loss: 0.0032\n",
      "458/484, train_loss: 0.0052\n",
      "459/484, train_loss: 0.0011\n",
      "460/484, train_loss: 0.3594\n",
      "461/484, train_loss: 0.5682\n",
      "462/484, train_loss: 0.0225\n",
      "463/484, train_loss: 0.0075\n",
      "464/484, train_loss: 0.6381\n",
      "465/484, train_loss: 0.0210\n",
      "466/484, train_loss: 0.0069\n",
      "467/484, train_loss: 0.0041\n",
      "468/484, train_loss: 0.0074\n",
      "469/484, train_loss: 0.0242\n",
      "470/484, train_loss: 0.0134\n",
      "471/484, train_loss: 0.0020\n",
      "472/484, train_loss: 0.0090\n",
      "473/484, train_loss: 0.0056\n",
      "474/484, train_loss: 0.0027\n",
      "475/484, train_loss: 0.0039\n",
      "476/484, train_loss: 0.0070\n",
      "477/484, train_loss: 0.0035\n",
      "478/484, train_loss: 0.0173\n",
      "479/484, train_loss: 0.0124\n",
      "480/484, train_loss: 0.0043\n",
      "481/484, train_loss: 0.4347\n",
      "482/484, train_loss: 0.0039\n",
      "483/484, train_loss: 0.0047\n",
      "484/484, train_loss: 0.0084\n",
      "epoch 5 average loss: 0.1654\n",
      "----------\n",
      "epoch 6/20\n",
      "1/484, train_loss: 0.0115\n",
      "2/484, train_loss: 0.0291\n",
      "3/484, train_loss: 0.0070\n",
      "4/484, train_loss: 0.0153\n",
      "5/484, train_loss: 0.0059\n",
      "6/484, train_loss: 0.0036\n",
      "7/484, train_loss: 1.5318\n",
      "8/484, train_loss: 0.0060\n",
      "9/484, train_loss: 0.0079\n",
      "10/484, train_loss: 0.0047\n",
      "11/484, train_loss: 0.1025\n",
      "12/484, train_loss: 0.5007\n",
      "13/484, train_loss: 0.3481\n",
      "14/484, train_loss: 1.0754\n",
      "15/484, train_loss: 0.0097\n",
      "16/484, train_loss: 0.5379\n",
      "17/484, train_loss: 0.0062\n",
      "18/484, train_loss: 0.2134\n",
      "19/484, train_loss: 0.0030\n",
      "20/484, train_loss: 0.1954\n",
      "21/484, train_loss: 0.3531\n",
      "22/484, train_loss: 0.0036\n",
      "23/484, train_loss: 0.3156\n",
      "24/484, train_loss: 0.0018\n",
      "25/484, train_loss: 0.0136\n",
      "26/484, train_loss: 0.0022\n",
      "27/484, train_loss: 0.0054\n",
      "28/484, train_loss: 0.0019\n",
      "29/484, train_loss: 0.5626\n",
      "30/484, train_loss: 0.0170\n",
      "31/484, train_loss: 0.0042\n",
      "32/484, train_loss: 0.0596\n",
      "33/484, train_loss: 1.2898\n",
      "34/484, train_loss: 0.0058\n",
      "35/484, train_loss: 0.0689\n",
      "36/484, train_loss: 0.0028\n",
      "37/484, train_loss: 0.0011\n",
      "38/484, train_loss: 0.8191\n",
      "39/484, train_loss: 0.0047\n",
      "40/484, train_loss: 0.0235\n",
      "41/484, train_loss: 0.0035\n",
      "42/484, train_loss: 1.2953\n",
      "43/484, train_loss: 0.0016\n",
      "44/484, train_loss: 0.0244\n",
      "45/484, train_loss: 0.0054\n",
      "46/484, train_loss: 0.0188\n",
      "47/484, train_loss: 0.0076\n",
      "48/484, train_loss: 0.0036\n",
      "49/484, train_loss: 0.0024\n",
      "50/484, train_loss: 0.0212\n",
      "51/484, train_loss: 0.0142\n",
      "52/484, train_loss: 0.0090\n",
      "53/484, train_loss: 0.2838\n",
      "54/484, train_loss: 0.0042\n",
      "55/484, train_loss: 0.0010\n",
      "56/484, train_loss: 0.0047\n",
      "57/484, train_loss: 0.0059\n",
      "58/484, train_loss: 0.2259\n",
      "59/484, train_loss: 0.0028\n",
      "60/484, train_loss: 0.0015\n",
      "61/484, train_loss: 0.0024\n",
      "62/484, train_loss: 0.0033\n",
      "63/484, train_loss: 0.0049\n",
      "64/484, train_loss: 0.0017\n",
      "65/484, train_loss: 0.0138\n",
      "66/484, train_loss: 0.0048\n",
      "67/484, train_loss: 0.0016\n",
      "68/484, train_loss: 0.0039\n",
      "69/484, train_loss: 0.0084\n",
      "70/484, train_loss: 0.0021\n",
      "71/484, train_loss: 0.1916\n",
      "72/484, train_loss: 0.5484\n",
      "73/484, train_loss: 0.0104\n",
      "74/484, train_loss: 0.0038\n",
      "75/484, train_loss: 0.0062\n",
      "76/484, train_loss: 0.0077\n",
      "77/484, train_loss: 0.0051\n",
      "78/484, train_loss: 0.0146\n",
      "79/484, train_loss: 0.0058\n",
      "80/484, train_loss: 0.0161\n",
      "81/484, train_loss: 0.0032\n",
      "82/484, train_loss: 0.0101\n",
      "83/484, train_loss: 0.0026\n",
      "84/484, train_loss: 0.0012\n",
      "85/484, train_loss: 0.2735\n",
      "86/484, train_loss: 0.0025\n",
      "87/484, train_loss: 0.0035\n",
      "88/484, train_loss: 0.4883\n",
      "89/484, train_loss: 0.0052\n",
      "90/484, train_loss: 0.0046\n",
      "91/484, train_loss: 0.2255\n",
      "92/484, train_loss: 0.0360\n",
      "93/484, train_loss: 0.0032\n",
      "94/484, train_loss: 0.0024\n",
      "95/484, train_loss: 0.0044\n",
      "96/484, train_loss: 0.3495\n",
      "97/484, train_loss: 0.0146\n",
      "98/484, train_loss: 0.0102\n",
      "99/484, train_loss: 0.0033\n",
      "100/484, train_loss: 0.0312\n",
      "101/484, train_loss: 0.0044\n",
      "102/484, train_loss: 0.0057\n",
      "103/484, train_loss: 0.0019\n",
      "104/484, train_loss: 0.0031\n",
      "105/484, train_loss: 0.0008\n",
      "106/484, train_loss: 0.0026\n",
      "107/484, train_loss: 0.0025\n",
      "108/484, train_loss: 0.3599\n",
      "109/484, train_loss: 0.0017\n",
      "110/484, train_loss: 0.0399\n",
      "111/484, train_loss: 0.2131\n",
      "112/484, train_loss: 0.0037\n",
      "113/484, train_loss: 0.0040\n",
      "114/484, train_loss: 0.0129\n",
      "115/484, train_loss: 0.0064\n",
      "116/484, train_loss: 0.8493\n",
      "117/484, train_loss: 1.2362\n",
      "118/484, train_loss: 0.0124\n",
      "119/484, train_loss: 0.0020\n",
      "120/484, train_loss: 0.0101\n",
      "121/484, train_loss: 0.0010\n",
      "122/484, train_loss: 0.8059\n",
      "123/484, train_loss: 0.1580\n",
      "124/484, train_loss: 0.0037\n",
      "125/484, train_loss: 0.4952\n",
      "126/484, train_loss: 0.0018\n",
      "127/484, train_loss: 0.0183\n",
      "128/484, train_loss: 0.0059\n",
      "129/484, train_loss: 0.1552\n",
      "130/484, train_loss: 0.1478\n",
      "131/484, train_loss: 0.0613\n",
      "132/484, train_loss: 0.0042\n",
      "133/484, train_loss: 0.4487\n",
      "134/484, train_loss: 0.0031\n",
      "135/484, train_loss: 0.0059\n",
      "136/484, train_loss: 1.0794\n",
      "137/484, train_loss: 0.3564\n",
      "138/484, train_loss: 0.5250\n",
      "139/484, train_loss: 0.0050\n",
      "140/484, train_loss: 0.0030\n",
      "141/484, train_loss: 0.0821\n",
      "142/484, train_loss: 0.0040\n",
      "143/484, train_loss: 0.3971\n",
      "144/484, train_loss: 0.0084\n",
      "145/484, train_loss: 0.0038\n",
      "146/484, train_loss: 0.0114\n",
      "147/484, train_loss: 0.3769\n",
      "148/484, train_loss: 2.3936\n",
      "149/484, train_loss: 0.0024\n",
      "150/484, train_loss: 0.2707\n",
      "151/484, train_loss: 0.0061\n",
      "152/484, train_loss: 0.0048\n",
      "153/484, train_loss: 0.0067\n",
      "154/484, train_loss: 0.0018\n",
      "155/484, train_loss: 0.0056\n",
      "156/484, train_loss: 0.1269\n",
      "157/484, train_loss: 0.0028\n",
      "158/484, train_loss: 1.3144\n",
      "159/484, train_loss: 0.3552\n",
      "160/484, train_loss: 0.0034\n",
      "161/484, train_loss: 0.0084\n",
      "162/484, train_loss: 0.4172\n",
      "163/484, train_loss: 0.0597\n",
      "164/484, train_loss: 0.2088\n",
      "165/484, train_loss: 0.0121\n",
      "166/484, train_loss: 0.1886\n",
      "167/484, train_loss: 0.0034\n",
      "168/484, train_loss: 1.9968\n",
      "169/484, train_loss: 0.4728\n",
      "170/484, train_loss: 0.0032\n",
      "171/484, train_loss: 0.0787\n",
      "172/484, train_loss: 0.0103\n",
      "173/484, train_loss: 0.0084\n",
      "174/484, train_loss: 0.5789\n",
      "175/484, train_loss: 0.0150\n",
      "176/484, train_loss: 0.2966\n",
      "177/484, train_loss: 0.8464\n",
      "178/484, train_loss: 0.0025\n",
      "179/484, train_loss: 0.0033\n",
      "180/484, train_loss: 0.0117\n",
      "181/484, train_loss: 0.0325\n",
      "182/484, train_loss: 0.0120\n",
      "183/484, train_loss: 0.0105\n",
      "184/484, train_loss: 0.0989\n",
      "185/484, train_loss: 1.5145\n",
      "186/484, train_loss: 0.3215\n",
      "187/484, train_loss: 0.1930\n",
      "188/484, train_loss: 0.0149\n",
      "189/484, train_loss: 0.0916\n",
      "190/484, train_loss: 0.2952\n",
      "191/484, train_loss: 0.1418\n",
      "192/484, train_loss: 0.0036\n",
      "193/484, train_loss: 0.5587\n",
      "194/484, train_loss: 1.0404\n",
      "195/484, train_loss: 0.0029\n",
      "196/484, train_loss: 0.1938\n",
      "197/484, train_loss: 0.0973\n",
      "198/484, train_loss: 0.0038\n",
      "199/484, train_loss: 0.0155\n",
      "200/484, train_loss: 0.0059\n",
      "201/484, train_loss: 0.0937\n",
      "202/484, train_loss: 0.0308\n",
      "203/484, train_loss: 0.0027\n",
      "204/484, train_loss: 1.5372\n",
      "205/484, train_loss: 0.0162\n",
      "206/484, train_loss: 0.0088\n",
      "207/484, train_loss: 0.0241\n",
      "208/484, train_loss: 0.0960\n",
      "209/484, train_loss: 0.1438\n",
      "210/484, train_loss: 0.0991\n",
      "211/484, train_loss: 0.4169\n",
      "212/484, train_loss: 0.0028\n",
      "213/484, train_loss: 0.0185\n",
      "214/484, train_loss: 0.0140\n",
      "215/484, train_loss: 0.1445\n",
      "216/484, train_loss: 0.0091\n",
      "217/484, train_loss: 0.0253\n",
      "218/484, train_loss: 0.4706\n",
      "219/484, train_loss: 0.0063\n",
      "220/484, train_loss: 0.0049\n",
      "221/484, train_loss: 0.2672\n",
      "222/484, train_loss: 0.5364\n",
      "223/484, train_loss: 0.3502\n",
      "224/484, train_loss: 1.8587\n",
      "225/484, train_loss: 1.7309\n",
      "226/484, train_loss: 0.0039\n",
      "227/484, train_loss: 0.2108\n",
      "228/484, train_loss: 0.0028\n",
      "229/484, train_loss: 1.5174\n",
      "230/484, train_loss: 0.0046\n",
      "231/484, train_loss: 0.0025\n",
      "232/484, train_loss: 0.0114\n",
      "233/484, train_loss: 0.0021\n",
      "234/484, train_loss: 0.0146\n",
      "235/484, train_loss: 0.3388\n",
      "236/484, train_loss: 0.0111\n",
      "237/484, train_loss: 0.0270\n",
      "238/484, train_loss: 0.0185\n",
      "239/484, train_loss: 0.0031\n",
      "240/484, train_loss: 0.0032\n",
      "241/484, train_loss: 0.3148\n",
      "242/484, train_loss: 0.0031\n",
      "243/484, train_loss: 0.0040\n",
      "244/484, train_loss: 0.0356\n",
      "245/484, train_loss: 0.1985\n",
      "246/484, train_loss: 0.0181\n",
      "247/484, train_loss: 0.0418\n",
      "248/484, train_loss: 0.5813\n",
      "249/484, train_loss: 0.1979\n",
      "250/484, train_loss: 0.0057\n",
      "251/484, train_loss: 0.2810\n",
      "252/484, train_loss: 0.2555\n",
      "253/484, train_loss: 0.0057\n",
      "254/484, train_loss: 0.0337\n",
      "255/484, train_loss: 0.0178\n",
      "256/484, train_loss: 0.0204\n",
      "257/484, train_loss: 3.1792\n",
      "258/484, train_loss: 0.0035\n",
      "259/484, train_loss: 0.0360\n",
      "260/484, train_loss: 0.0063\n",
      "261/484, train_loss: 0.0124\n",
      "262/484, train_loss: 0.0058\n",
      "263/484, train_loss: 0.0114\n",
      "264/484, train_loss: 1.5075\n",
      "265/484, train_loss: 0.0070\n",
      "266/484, train_loss: 0.0312\n",
      "267/484, train_loss: 0.3939\n",
      "268/484, train_loss: 0.0438\n",
      "269/484, train_loss: 0.3628\n",
      "270/484, train_loss: 0.0307\n",
      "271/484, train_loss: 0.0409\n",
      "272/484, train_loss: 0.0932\n",
      "273/484, train_loss: 0.0064\n",
      "274/484, train_loss: 0.0226\n",
      "275/484, train_loss: 0.0032\n",
      "276/484, train_loss: 0.0564\n",
      "277/484, train_loss: 0.0272\n",
      "278/484, train_loss: 0.0079\n",
      "279/484, train_loss: 0.0045\n",
      "280/484, train_loss: 0.0366\n",
      "281/484, train_loss: 0.0269\n",
      "282/484, train_loss: 0.0041\n",
      "283/484, train_loss: 0.0827\n",
      "284/484, train_loss: 0.2776\n",
      "285/484, train_loss: 0.0840\n",
      "286/484, train_loss: 0.0022\n",
      "287/484, train_loss: 0.0047\n",
      "288/484, train_loss: 0.8036\n",
      "289/484, train_loss: 0.0094\n",
      "290/484, train_loss: 0.0062\n",
      "291/484, train_loss: 0.0178\n",
      "292/484, train_loss: 0.0068\n",
      "293/484, train_loss: 0.3154\n",
      "294/484, train_loss: 0.0082\n",
      "295/484, train_loss: 0.0171\n",
      "296/484, train_loss: 0.0052\n",
      "297/484, train_loss: 0.1830\n",
      "298/484, train_loss: 0.0732\n",
      "299/484, train_loss: 0.0125\n",
      "300/484, train_loss: 0.0032\n",
      "301/484, train_loss: 0.0038\n",
      "302/484, train_loss: 0.4997\n",
      "303/484, train_loss: 0.0112\n",
      "304/484, train_loss: 0.0112\n",
      "305/484, train_loss: 0.0093\n",
      "306/484, train_loss: 0.0310\n",
      "307/484, train_loss: 0.1660\n",
      "308/484, train_loss: 0.0084\n",
      "309/484, train_loss: 0.2119\n",
      "310/484, train_loss: 0.0229\n",
      "311/484, train_loss: 0.0104\n",
      "312/484, train_loss: 0.0029\n",
      "313/484, train_loss: 0.0058\n",
      "314/484, train_loss: 0.9365\n",
      "315/484, train_loss: 0.0291\n",
      "316/484, train_loss: 0.0065\n",
      "317/484, train_loss: 0.0075\n",
      "318/484, train_loss: 0.0522\n",
      "319/484, train_loss: 0.0187\n",
      "320/484, train_loss: 0.0100\n",
      "321/484, train_loss: 0.1459\n",
      "322/484, train_loss: 0.0168\n",
      "323/484, train_loss: 0.0485\n",
      "324/484, train_loss: 0.0069\n",
      "325/484, train_loss: 0.0047\n",
      "326/484, train_loss: 0.0048\n",
      "327/484, train_loss: 0.0177\n",
      "328/484, train_loss: 0.0031\n",
      "329/484, train_loss: 1.9175\n",
      "330/484, train_loss: 0.0034\n",
      "331/484, train_loss: 0.0042\n",
      "332/484, train_loss: 0.0031\n",
      "333/484, train_loss: 0.0107\n",
      "334/484, train_loss: 0.0043\n",
      "335/484, train_loss: 0.0045\n",
      "336/484, train_loss: 0.0118\n",
      "337/484, train_loss: 0.0239\n",
      "338/484, train_loss: 0.0056\n",
      "339/484, train_loss: 0.0080\n",
      "340/484, train_loss: 0.0071\n",
      "341/484, train_loss: 0.0054\n",
      "342/484, train_loss: 0.0581\n",
      "343/484, train_loss: 0.0115\n",
      "344/484, train_loss: 0.0073\n",
      "345/484, train_loss: 0.0700\n",
      "346/484, train_loss: 0.5069\n",
      "347/484, train_loss: 0.0041\n",
      "348/484, train_loss: 0.0188\n",
      "349/484, train_loss: 0.0090\n",
      "350/484, train_loss: 0.0045\n",
      "351/484, train_loss: 0.0017\n",
      "352/484, train_loss: 0.0038\n",
      "353/484, train_loss: 0.0825\n",
      "354/484, train_loss: 0.0064\n",
      "355/484, train_loss: 1.0222\n",
      "356/484, train_loss: 0.0514\n",
      "357/484, train_loss: 0.2467\n",
      "358/484, train_loss: 0.0052\n",
      "359/484, train_loss: 0.0061\n",
      "360/484, train_loss: 0.0034\n",
      "361/484, train_loss: 0.0116\n",
      "362/484, train_loss: 0.0197\n",
      "363/484, train_loss: 0.2405\n",
      "364/484, train_loss: 2.5810\n",
      "365/484, train_loss: 0.0048\n",
      "366/484, train_loss: 0.0060\n",
      "367/484, train_loss: 0.0082\n",
      "368/484, train_loss: 0.3715\n",
      "369/484, train_loss: 0.1383\n",
      "370/484, train_loss: 0.0264\n",
      "371/484, train_loss: 0.0314\n",
      "372/484, train_loss: 0.0234\n",
      "373/484, train_loss: 0.1183\n",
      "374/484, train_loss: 0.0073\n",
      "375/484, train_loss: 0.0023\n",
      "376/484, train_loss: 0.0251\n",
      "377/484, train_loss: 0.2857\n",
      "378/484, train_loss: 0.0763\n",
      "379/484, train_loss: 1.2704\n",
      "380/484, train_loss: 0.1021\n",
      "381/484, train_loss: 0.0050\n",
      "382/484, train_loss: 0.0073\n",
      "383/484, train_loss: 0.0073\n",
      "384/484, train_loss: 0.6892\n",
      "385/484, train_loss: 0.0086\n",
      "386/484, train_loss: 0.2936\n",
      "387/484, train_loss: 1.3901\n",
      "388/484, train_loss: 0.0072\n",
      "389/484, train_loss: 0.0048\n",
      "390/484, train_loss: 0.0058\n",
      "391/484, train_loss: 0.0053\n",
      "392/484, train_loss: 0.0105\n",
      "393/484, train_loss: 0.5892\n",
      "394/484, train_loss: 0.0055\n",
      "395/484, train_loss: 0.0115\n",
      "396/484, train_loss: 0.0046\n",
      "397/484, train_loss: 0.0407\n",
      "398/484, train_loss: 0.0067\n",
      "399/484, train_loss: 0.0081\n",
      "400/484, train_loss: 0.3613\n",
      "401/484, train_loss: 0.0209\n",
      "402/484, train_loss: 0.0071\n",
      "403/484, train_loss: 0.0028\n",
      "404/484, train_loss: 0.0070\n",
      "405/484, train_loss: 1.0405\n",
      "406/484, train_loss: 0.0087\n",
      "407/484, train_loss: 0.9892\n",
      "408/484, train_loss: 0.0087\n",
      "409/484, train_loss: 0.0054\n",
      "410/484, train_loss: 0.0148\n",
      "411/484, train_loss: 0.0082\n",
      "412/484, train_loss: 1.5614\n",
      "413/484, train_loss: 0.0127\n",
      "414/484, train_loss: 0.0028\n",
      "415/484, train_loss: 0.0559\n",
      "416/484, train_loss: 0.0187\n",
      "417/484, train_loss: 0.2039\n",
      "418/484, train_loss: 0.0122\n",
      "419/484, train_loss: 0.0404\n",
      "420/484, train_loss: 0.0112\n",
      "421/484, train_loss: 0.0080\n",
      "422/484, train_loss: 0.3208\n",
      "423/484, train_loss: 0.6712\n",
      "424/484, train_loss: 0.0096\n",
      "425/484, train_loss: 0.6092\n",
      "426/484, train_loss: 0.0153\n",
      "427/484, train_loss: 0.0125\n",
      "428/484, train_loss: 1.4710\n",
      "429/484, train_loss: 0.2384\n",
      "430/484, train_loss: 0.0707\n",
      "431/484, train_loss: 0.0391\n",
      "432/484, train_loss: 0.5233\n",
      "433/484, train_loss: 0.5414\n",
      "434/484, train_loss: 0.0110\n",
      "435/484, train_loss: 0.0056\n",
      "436/484, train_loss: 0.0247\n",
      "437/484, train_loss: 0.1778\n",
      "438/484, train_loss: 0.0285\n",
      "439/484, train_loss: 0.4516\n",
      "440/484, train_loss: 0.0289\n",
      "441/484, train_loss: 0.0121\n",
      "442/484, train_loss: 0.0024\n",
      "443/484, train_loss: 0.0049\n",
      "444/484, train_loss: 0.0090\n",
      "445/484, train_loss: 0.8311\n",
      "446/484, train_loss: 0.0113\n",
      "447/484, train_loss: 0.0099\n",
      "448/484, train_loss: 0.0454\n",
      "449/484, train_loss: 0.0079\n",
      "450/484, train_loss: 0.0367\n",
      "451/484, train_loss: 0.0084\n",
      "452/484, train_loss: 0.0021\n",
      "453/484, train_loss: 0.0097\n",
      "454/484, train_loss: 0.0110\n",
      "455/484, train_loss: 0.0170\n",
      "456/484, train_loss: 0.3531\n",
      "457/484, train_loss: 0.0074\n",
      "458/484, train_loss: 0.0141\n",
      "459/484, train_loss: 0.0353\n",
      "460/484, train_loss: 0.0029\n",
      "461/484, train_loss: 1.2578\n",
      "462/484, train_loss: 0.0036\n",
      "463/484, train_loss: 0.0081\n",
      "464/484, train_loss: 0.0071\n",
      "465/484, train_loss: 0.0053\n",
      "466/484, train_loss: 0.1188\n",
      "467/484, train_loss: 0.0199\n",
      "468/484, train_loss: 0.0160\n",
      "469/484, train_loss: 0.0086\n",
      "470/484, train_loss: 0.2448\n",
      "471/484, train_loss: 1.5747\n",
      "472/484, train_loss: 0.0203\n",
      "473/484, train_loss: 0.0071\n",
      "474/484, train_loss: 0.0054\n",
      "475/484, train_loss: 0.0034\n",
      "476/484, train_loss: 0.0094\n",
      "477/484, train_loss: 0.0130\n",
      "478/484, train_loss: 0.0028\n",
      "479/484, train_loss: 0.0024\n",
      "480/484, train_loss: 0.0186\n",
      "481/484, train_loss: 0.0046\n",
      "482/484, train_loss: 0.0170\n",
      "483/484, train_loss: 0.0053\n",
      "484/484, train_loss: 0.0117\n",
      "epoch 6 average loss: 0.1737\n",
      "saved new best metric model\n",
      "Current epoch: 6 current accuracy: 0.8297 \n",
      "Best accuracy: 0.8297 at epoch 6\n",
      "----------\n",
      "epoch 7/20\n",
      "1/484, train_loss: 0.0298\n",
      "2/484, train_loss: 0.0379\n",
      "3/484, train_loss: 0.0491\n",
      "4/484, train_loss: 0.0061\n",
      "5/484, train_loss: 0.4575\n",
      "6/484, train_loss: 0.6123\n",
      "7/484, train_loss: 0.0081\n",
      "8/484, train_loss: 0.0169\n",
      "9/484, train_loss: 0.1936\n",
      "10/484, train_loss: 0.0076\n",
      "11/484, train_loss: 0.0056\n",
      "12/484, train_loss: 0.2409\n",
      "13/484, train_loss: 0.0076\n",
      "14/484, train_loss: 0.0204\n",
      "15/484, train_loss: 0.0153\n",
      "16/484, train_loss: 0.0067\n",
      "17/484, train_loss: 0.2474\n",
      "18/484, train_loss: 0.1911\n",
      "19/484, train_loss: 0.0033\n",
      "20/484, train_loss: 0.0046\n",
      "21/484, train_loss: 0.5526\n",
      "22/484, train_loss: 0.0031\n",
      "23/484, train_loss: 0.0049\n",
      "24/484, train_loss: 0.0057\n",
      "25/484, train_loss: 0.0018\n",
      "26/484, train_loss: 0.3345\n",
      "27/484, train_loss: 0.0165\n",
      "28/484, train_loss: 0.1927\n",
      "29/484, train_loss: 0.0111\n",
      "30/484, train_loss: 0.4954\n",
      "31/484, train_loss: 0.0038\n",
      "32/484, train_loss: 0.0064\n",
      "33/484, train_loss: 0.0050\n",
      "34/484, train_loss: 0.0046\n",
      "35/484, train_loss: 0.2256\n",
      "36/484, train_loss: 0.0031\n",
      "37/484, train_loss: 0.0790\n",
      "38/484, train_loss: 0.0088\n",
      "39/484, train_loss: 0.0040\n",
      "40/484, train_loss: 0.0034\n",
      "41/484, train_loss: 0.0074\n",
      "42/484, train_loss: 0.0104\n",
      "43/484, train_loss: 0.0065\n",
      "44/484, train_loss: 0.3687\n",
      "45/484, train_loss: 0.3939\n",
      "46/484, train_loss: 0.0100\n",
      "47/484, train_loss: 0.0044\n",
      "48/484, train_loss: 0.0085\n",
      "49/484, train_loss: 0.0263\n",
      "50/484, train_loss: 0.0053\n",
      "51/484, train_loss: 0.6349\n",
      "52/484, train_loss: 0.0016\n",
      "53/484, train_loss: 0.0015\n",
      "54/484, train_loss: 0.6279\n",
      "55/484, train_loss: 0.0071\n",
      "56/484, train_loss: 0.6948\n",
      "57/484, train_loss: 0.0044\n",
      "58/484, train_loss: 0.0261\n",
      "59/484, train_loss: 0.0074\n",
      "60/484, train_loss: 0.0063\n",
      "61/484, train_loss: 0.0082\n",
      "62/484, train_loss: 0.0080\n",
      "63/484, train_loss: 0.0223\n",
      "64/484, train_loss: 0.0076\n",
      "65/484, train_loss: 0.4932\n",
      "66/484, train_loss: 0.0025\n",
      "67/484, train_loss: 0.0143\n",
      "68/484, train_loss: 0.2720\n",
      "69/484, train_loss: 0.2392\n",
      "70/484, train_loss: 0.0130\n",
      "71/484, train_loss: 0.0215\n",
      "72/484, train_loss: 0.3588\n",
      "73/484, train_loss: 0.3388\n",
      "74/484, train_loss: 0.0087\n",
      "75/484, train_loss: 0.0093\n",
      "76/484, train_loss: 0.0181\n",
      "77/484, train_loss: 0.4059\n",
      "78/484, train_loss: 0.0045\n",
      "79/484, train_loss: 0.6660\n",
      "80/484, train_loss: 0.0327\n",
      "81/484, train_loss: 0.0009\n",
      "82/484, train_loss: 0.0073\n",
      "83/484, train_loss: 0.0683\n",
      "84/484, train_loss: 0.0080\n",
      "85/484, train_loss: 1.0408\n",
      "86/484, train_loss: 0.3914\n",
      "87/484, train_loss: 0.0068\n",
      "88/484, train_loss: 0.0030\n",
      "89/484, train_loss: 0.0102\n",
      "90/484, train_loss: 0.0545\n",
      "91/484, train_loss: 0.0066\n",
      "92/484, train_loss: 0.0041\n",
      "93/484, train_loss: 0.0090\n",
      "94/484, train_loss: 0.0015\n",
      "95/484, train_loss: 0.0126\n",
      "96/484, train_loss: 0.0232\n",
      "97/484, train_loss: 0.0041\n",
      "98/484, train_loss: 0.0022\n",
      "99/484, train_loss: 0.0065\n",
      "100/484, train_loss: 0.8057\n",
      "101/484, train_loss: 1.5330\n",
      "102/484, train_loss: 0.1044\n",
      "103/484, train_loss: 0.0071\n",
      "104/484, train_loss: 0.0067\n",
      "105/484, train_loss: 0.0033\n",
      "106/484, train_loss: 0.0034\n",
      "107/484, train_loss: 0.0026\n",
      "108/484, train_loss: 0.0044\n",
      "109/484, train_loss: 0.0064\n",
      "110/484, train_loss: 0.0045\n",
      "111/484, train_loss: 0.0039\n",
      "112/484, train_loss: 0.0097\n",
      "113/484, train_loss: 0.0032\n",
      "114/484, train_loss: 0.0031\n",
      "115/484, train_loss: 0.0033\n",
      "116/484, train_loss: 0.0097\n",
      "117/484, train_loss: 0.0064\n",
      "118/484, train_loss: 0.0107\n",
      "119/484, train_loss: 0.0025\n",
      "120/484, train_loss: 0.0040\n",
      "121/484, train_loss: 0.0346\n",
      "122/484, train_loss: 0.0050\n",
      "123/484, train_loss: 0.0042\n",
      "124/484, train_loss: 0.0031\n",
      "125/484, train_loss: 0.0143\n",
      "126/484, train_loss: 0.4501\n",
      "127/484, train_loss: 0.0067\n",
      "128/484, train_loss: 0.0019\n",
      "129/484, train_loss: 0.0047\n",
      "130/484, train_loss: 0.0021\n",
      "131/484, train_loss: 0.0209\n",
      "132/484, train_loss: 0.0017\n",
      "133/484, train_loss: 0.0032\n",
      "134/484, train_loss: 0.0041\n",
      "135/484, train_loss: 0.0086\n",
      "136/484, train_loss: 0.0162\n",
      "137/484, train_loss: 0.0057\n",
      "138/484, train_loss: 0.4042\n",
      "139/484, train_loss: 0.0028\n",
      "140/484, train_loss: 0.0035\n",
      "141/484, train_loss: 0.0048\n",
      "142/484, train_loss: 0.0049\n",
      "143/484, train_loss: 0.0043\n",
      "144/484, train_loss: 0.4262\n",
      "145/484, train_loss: 0.0072\n",
      "146/484, train_loss: 0.0046\n",
      "147/484, train_loss: 0.6018\n",
      "148/484, train_loss: 0.0122\n",
      "149/484, train_loss: 0.0072\n",
      "150/484, train_loss: 0.6767\n",
      "151/484, train_loss: 0.0127\n",
      "152/484, train_loss: 0.1507\n",
      "153/484, train_loss: 0.1720\n",
      "154/484, train_loss: 0.5064\n",
      "155/484, train_loss: 0.0080\n",
      "156/484, train_loss: 0.3115\n",
      "157/484, train_loss: 0.0135\n",
      "158/484, train_loss: 0.0027\n",
      "159/484, train_loss: 0.4007\n",
      "160/484, train_loss: 0.0033\n",
      "161/484, train_loss: 0.0046\n",
      "162/484, train_loss: 0.2835\n",
      "163/484, train_loss: 0.0036\n",
      "164/484, train_loss: 0.0030\n",
      "165/484, train_loss: 0.0064\n",
      "166/484, train_loss: 0.0032\n",
      "167/484, train_loss: 0.9392\n",
      "168/484, train_loss: 0.0041\n",
      "169/484, train_loss: 0.8281\n",
      "170/484, train_loss: 0.0022\n",
      "171/484, train_loss: 0.5844\n",
      "172/484, train_loss: 0.0026\n",
      "173/484, train_loss: 0.0049\n",
      "174/484, train_loss: 0.0028\n",
      "175/484, train_loss: 0.0042\n",
      "176/484, train_loss: 0.0040\n",
      "177/484, train_loss: 0.0070\n",
      "178/484, train_loss: 0.1713\n",
      "179/484, train_loss: 0.0137\n",
      "180/484, train_loss: 0.2875\n",
      "181/484, train_loss: 0.0034\n",
      "182/484, train_loss: 0.0094\n",
      "183/484, train_loss: 0.0068\n",
      "184/484, train_loss: 0.0014\n",
      "185/484, train_loss: 0.0040\n",
      "186/484, train_loss: 0.0042\n",
      "187/484, train_loss: 0.0154\n",
      "188/484, train_loss: 0.0039\n",
      "189/484, train_loss: 0.0034\n",
      "190/484, train_loss: 0.0048\n",
      "191/484, train_loss: 0.0046\n",
      "192/484, train_loss: 0.0042\n",
      "193/484, train_loss: 0.0032\n",
      "194/484, train_loss: 0.0042\n",
      "195/484, train_loss: 0.0038\n",
      "196/484, train_loss: 0.3966\n",
      "197/484, train_loss: 0.0045\n",
      "198/484, train_loss: 0.0101\n",
      "199/484, train_loss: 0.0059\n",
      "200/484, train_loss: 0.0489\n",
      "201/484, train_loss: 0.0053\n",
      "202/484, train_loss: 0.0989\n",
      "203/484, train_loss: 0.0040\n",
      "204/484, train_loss: 0.0037\n",
      "205/484, train_loss: 0.2722\n",
      "206/484, train_loss: 0.0013\n",
      "207/484, train_loss: 0.0199\n",
      "208/484, train_loss: 0.0033\n",
      "209/484, train_loss: 0.0368\n",
      "210/484, train_loss: 0.8066\n",
      "211/484, train_loss: 0.3142\n",
      "212/484, train_loss: 0.0047\n",
      "213/484, train_loss: 0.0112\n",
      "214/484, train_loss: 0.0066\n",
      "215/484, train_loss: 0.0054\n",
      "216/484, train_loss: 0.2382\n",
      "217/484, train_loss: 0.0028\n",
      "218/484, train_loss: 1.5450\n",
      "219/484, train_loss: 0.0110\n",
      "220/484, train_loss: 0.0020\n",
      "221/484, train_loss: 0.0025\n",
      "222/484, train_loss: 0.0029\n",
      "223/484, train_loss: 0.0074\n",
      "224/484, train_loss: 0.0023\n",
      "225/484, train_loss: 0.0029\n",
      "226/484, train_loss: 0.1549\n",
      "227/484, train_loss: 0.0035\n",
      "228/484, train_loss: 0.0049\n",
      "229/484, train_loss: 1.2849\n",
      "230/484, train_loss: 0.0048\n",
      "231/484, train_loss: 0.1006\n",
      "232/484, train_loss: 0.0067\n",
      "233/484, train_loss: 0.0028\n",
      "234/484, train_loss: 0.0058\n",
      "235/484, train_loss: 0.0034\n",
      "236/484, train_loss: 0.0046\n",
      "237/484, train_loss: 0.0046\n",
      "238/484, train_loss: 0.0054\n",
      "239/484, train_loss: 0.0019\n",
      "240/484, train_loss: 0.0032\n",
      "241/484, train_loss: 0.0172\n",
      "242/484, train_loss: 0.5071\n",
      "243/484, train_loss: 0.0113\n",
      "244/484, train_loss: 0.0031\n",
      "245/484, train_loss: 0.0018\n",
      "246/484, train_loss: 0.0046\n",
      "247/484, train_loss: 0.0031\n",
      "248/484, train_loss: 0.3220\n",
      "249/484, train_loss: 0.0037\n",
      "250/484, train_loss: 0.1240\n",
      "251/484, train_loss: 0.0047\n",
      "252/484, train_loss: 0.0081\n",
      "253/484, train_loss: 0.0025\n",
      "254/484, train_loss: 0.0352\n",
      "255/484, train_loss: 0.0033\n",
      "256/484, train_loss: 0.0019\n",
      "257/484, train_loss: 0.1126\n",
      "258/484, train_loss: 0.1818\n",
      "259/484, train_loss: 0.0065\n",
      "260/484, train_loss: 0.0026\n",
      "261/484, train_loss: 0.0078\n",
      "262/484, train_loss: 1.7468\n",
      "263/484, train_loss: 0.0035\n",
      "264/484, train_loss: 0.0048\n",
      "265/484, train_loss: 0.0041\n",
      "266/484, train_loss: 0.6061\n",
      "267/484, train_loss: 0.3989\n",
      "268/484, train_loss: 0.0037\n",
      "269/484, train_loss: 0.0050\n",
      "270/484, train_loss: 0.0049\n",
      "271/484, train_loss: 0.0036\n",
      "272/484, train_loss: 0.0023\n",
      "273/484, train_loss: 0.1513\n",
      "274/484, train_loss: 0.0090\n",
      "275/484, train_loss: 0.0115\n",
      "276/484, train_loss: 0.0112\n",
      "277/484, train_loss: 0.0039\n",
      "278/484, train_loss: 0.0037\n",
      "279/484, train_loss: 0.0033\n",
      "280/484, train_loss: 0.5511\n",
      "281/484, train_loss: 0.0038\n",
      "282/484, train_loss: 0.1273\n",
      "283/484, train_loss: 0.0075\n",
      "284/484, train_loss: 0.0059\n",
      "285/484, train_loss: 0.0045\n",
      "286/484, train_loss: 0.0038\n",
      "287/484, train_loss: 0.0020\n",
      "288/484, train_loss: 0.0090\n",
      "289/484, train_loss: 0.0541\n",
      "290/484, train_loss: 0.0033\n",
      "291/484, train_loss: 0.0019\n",
      "292/484, train_loss: 0.7709\n",
      "293/484, train_loss: 0.0016\n",
      "294/484, train_loss: 0.0017\n",
      "295/484, train_loss: 0.0109\n",
      "296/484, train_loss: 0.0041\n",
      "297/484, train_loss: 1.1541\n",
      "298/484, train_loss: 0.4331\n",
      "299/484, train_loss: 0.0095\n",
      "300/484, train_loss: 0.0025\n",
      "301/484, train_loss: 0.0018\n",
      "302/484, train_loss: 0.3215\n",
      "303/484, train_loss: 0.0032\n",
      "304/484, train_loss: 0.0073\n",
      "305/484, train_loss: 0.0064\n",
      "306/484, train_loss: 0.0024\n",
      "307/484, train_loss: 0.0022\n",
      "308/484, train_loss: 0.0044\n",
      "309/484, train_loss: 0.1689\n",
      "310/484, train_loss: 0.0149\n",
      "311/484, train_loss: 0.0025\n",
      "312/484, train_loss: 0.0101\n",
      "313/484, train_loss: 0.0028\n",
      "314/484, train_loss: 0.0049\n",
      "315/484, train_loss: 0.0066\n",
      "316/484, train_loss: 0.0047\n",
      "317/484, train_loss: 0.0033\n",
      "318/484, train_loss: 0.0063\n",
      "319/484, train_loss: 0.0044\n",
      "320/484, train_loss: 0.0024\n",
      "321/484, train_loss: 0.0024\n",
      "322/484, train_loss: 0.0041\n",
      "323/484, train_loss: 0.4122\n",
      "324/484, train_loss: 0.0028\n",
      "325/484, train_loss: 0.0069\n",
      "326/484, train_loss: 0.0071\n",
      "327/484, train_loss: 0.0109\n",
      "328/484, train_loss: 0.0092\n",
      "329/484, train_loss: 0.3873\n",
      "330/484, train_loss: 0.0077\n",
      "331/484, train_loss: 0.0159\n",
      "332/484, train_loss: 0.0036\n",
      "333/484, train_loss: 0.0090\n",
      "334/484, train_loss: 0.0031\n",
      "335/484, train_loss: 0.0041\n",
      "336/484, train_loss: 0.0164\n",
      "337/484, train_loss: 0.0032\n",
      "338/484, train_loss: 0.0172\n",
      "339/484, train_loss: 0.0033\n",
      "340/484, train_loss: 0.0086\n",
      "341/484, train_loss: 0.0133\n",
      "342/484, train_loss: 0.0048\n",
      "343/484, train_loss: 0.1735\n",
      "344/484, train_loss: 0.0044\n",
      "345/484, train_loss: 0.0028\n",
      "346/484, train_loss: 0.0030\n",
      "347/484, train_loss: 0.0047\n",
      "348/484, train_loss: 0.0317\n",
      "349/484, train_loss: 0.0033\n",
      "350/484, train_loss: 0.0043\n",
      "351/484, train_loss: 0.0130\n",
      "352/484, train_loss: 0.0033\n",
      "353/484, train_loss: 0.9357\n",
      "354/484, train_loss: 0.0041\n",
      "355/484, train_loss: 0.0028\n",
      "356/484, train_loss: 0.0035\n",
      "357/484, train_loss: 0.4077\n",
      "358/484, train_loss: 0.1750\n",
      "359/484, train_loss: 0.0131\n",
      "360/484, train_loss: 0.3868\n",
      "361/484, train_loss: 0.0120\n",
      "362/484, train_loss: 0.0070\n",
      "363/484, train_loss: 0.7500\n",
      "364/484, train_loss: 0.0110\n",
      "365/484, train_loss: 0.0053\n",
      "366/484, train_loss: 0.0070\n",
      "367/484, train_loss: 0.0042\n",
      "368/484, train_loss: 0.0258\n",
      "369/484, train_loss: 0.0045\n",
      "370/484, train_loss: 0.0027\n",
      "371/484, train_loss: 0.6228\n",
      "372/484, train_loss: 0.7344\n",
      "373/484, train_loss: 0.0035\n",
      "374/484, train_loss: 0.0070\n",
      "375/484, train_loss: 0.0070\n",
      "376/484, train_loss: 0.0027\n",
      "377/484, train_loss: 0.0019\n",
      "378/484, train_loss: 0.0072\n",
      "379/484, train_loss: 0.0061\n",
      "380/484, train_loss: 0.0049\n",
      "381/484, train_loss: 0.5002\n",
      "382/484, train_loss: 0.2204\n",
      "383/484, train_loss: 0.0020\n",
      "384/484, train_loss: 0.2995\n",
      "385/484, train_loss: 0.0159\n",
      "386/484, train_loss: 0.0069\n",
      "387/484, train_loss: 0.0043\n",
      "388/484, train_loss: 0.0055\n",
      "389/484, train_loss: 0.6159\n",
      "390/484, train_loss: 0.0075\n",
      "391/484, train_loss: 0.0240\n",
      "392/484, train_loss: 0.0045\n",
      "393/484, train_loss: 0.2310\n",
      "394/484, train_loss: 0.0058\n",
      "395/484, train_loss: 0.0539\n",
      "396/484, train_loss: 0.0045\n",
      "397/484, train_loss: 0.0041\n",
      "398/484, train_loss: 0.0031\n",
      "399/484, train_loss: 0.0036\n",
      "400/484, train_loss: 0.0038\n",
      "401/484, train_loss: 0.0159\n",
      "402/484, train_loss: 0.0018\n",
      "403/484, train_loss: 0.0071\n",
      "404/484, train_loss: 0.0043\n",
      "405/484, train_loss: 0.2916\n",
      "406/484, train_loss: 0.0026\n",
      "407/484, train_loss: 0.0116\n",
      "408/484, train_loss: 0.0065\n",
      "409/484, train_loss: 0.0037\n",
      "410/484, train_loss: 0.0023\n",
      "411/484, train_loss: 0.0079\n",
      "412/484, train_loss: 0.0032\n",
      "413/484, train_loss: 0.0204\n",
      "414/484, train_loss: 0.0034\n",
      "415/484, train_loss: 0.2452\n",
      "416/484, train_loss: 0.0037\n",
      "417/484, train_loss: 0.5563\n",
      "418/484, train_loss: 0.0035\n",
      "419/484, train_loss: 0.4534\n",
      "420/484, train_loss: 0.0013\n",
      "421/484, train_loss: 0.0087\n",
      "422/484, train_loss: 0.0098\n",
      "423/484, train_loss: 0.0040\n",
      "424/484, train_loss: 0.0033\n",
      "425/484, train_loss: 0.0024\n",
      "426/484, train_loss: 0.0036\n",
      "427/484, train_loss: 0.1729\n",
      "428/484, train_loss: 0.0028\n",
      "429/484, train_loss: 0.0015\n",
      "430/484, train_loss: 0.0149\n",
      "431/484, train_loss: 0.0019\n",
      "432/484, train_loss: 0.0034\n",
      "433/484, train_loss: 0.0028\n",
      "434/484, train_loss: 0.0080\n",
      "435/484, train_loss: 0.0052\n",
      "436/484, train_loss: 0.0048\n",
      "437/484, train_loss: 0.2666\n",
      "438/484, train_loss: 0.0016\n",
      "439/484, train_loss: 0.0415\n",
      "440/484, train_loss: 0.0016\n",
      "441/484, train_loss: 0.7028\n",
      "442/484, train_loss: 0.6545\n",
      "443/484, train_loss: 0.0098\n",
      "444/484, train_loss: 0.0024\n",
      "445/484, train_loss: 0.0084\n",
      "446/484, train_loss: 0.2747\n",
      "447/484, train_loss: 0.0022\n",
      "448/484, train_loss: 0.0021\n",
      "449/484, train_loss: 0.0035\n",
      "450/484, train_loss: 0.0048\n",
      "451/484, train_loss: 1.5029\n",
      "452/484, train_loss: 0.0015\n",
      "453/484, train_loss: 0.0033\n",
      "454/484, train_loss: 0.0074\n",
      "455/484, train_loss: 0.0015\n",
      "456/484, train_loss: 0.2089\n",
      "457/484, train_loss: 0.0016\n",
      "458/484, train_loss: 0.0026\n",
      "459/484, train_loss: 0.1846\n",
      "460/484, train_loss: 0.0015\n",
      "461/484, train_loss: 0.0021\n",
      "462/484, train_loss: 0.0032\n",
      "463/484, train_loss: 0.3250\n",
      "464/484, train_loss: 0.0032\n",
      "465/484, train_loss: 0.0033\n",
      "466/484, train_loss: 0.3955\n",
      "467/484, train_loss: 0.5693\n",
      "468/484, train_loss: 0.0042\n",
      "469/484, train_loss: 0.3495\n",
      "470/484, train_loss: 0.0038\n",
      "471/484, train_loss: 0.2912\n",
      "472/484, train_loss: 0.0032\n",
      "473/484, train_loss: 0.0025\n",
      "474/484, train_loss: 0.0024\n",
      "475/484, train_loss: 0.0190\n",
      "476/484, train_loss: 0.0240\n",
      "477/484, train_loss: 0.0187\n",
      "478/484, train_loss: 0.0057\n",
      "479/484, train_loss: 0.5443\n",
      "480/484, train_loss: 0.0024\n",
      "481/484, train_loss: 0.0045\n",
      "482/484, train_loss: 0.0033\n",
      "483/484, train_loss: 0.0047\n",
      "484/484, train_loss: 0.0122\n",
      "epoch 7 average loss: 0.1054\n",
      "----------\n",
      "epoch 8/20\n",
      "1/484, train_loss: 0.0035\n",
      "2/484, train_loss: 0.0024\n",
      "3/484, train_loss: 0.1952\n",
      "4/484, train_loss: 0.0901\n",
      "5/484, train_loss: 0.0047\n",
      "6/484, train_loss: 0.0034\n",
      "7/484, train_loss: 1.8053\n",
      "8/484, train_loss: 0.0047\n",
      "9/484, train_loss: 0.0060\n",
      "10/484, train_loss: 0.0089\n",
      "11/484, train_loss: 1.9614\n",
      "12/484, train_loss: 0.0024\n",
      "13/484, train_loss: 0.0034\n",
      "14/484, train_loss: 0.1226\n",
      "15/484, train_loss: 0.0052\n",
      "16/484, train_loss: 0.0134\n",
      "17/484, train_loss: 0.0052\n",
      "18/484, train_loss: 1.7226\n",
      "19/484, train_loss: 0.0020\n",
      "20/484, train_loss: 0.0064\n",
      "21/484, train_loss: 0.4121\n",
      "22/484, train_loss: 0.0040\n",
      "23/484, train_loss: 0.1195\n",
      "24/484, train_loss: 0.0044\n",
      "25/484, train_loss: 0.0027\n",
      "26/484, train_loss: 0.8424\n",
      "27/484, train_loss: 0.0027\n",
      "28/484, train_loss: 0.0032\n",
      "29/484, train_loss: 0.0026\n",
      "30/484, train_loss: 0.0037\n",
      "31/484, train_loss: 0.0036\n",
      "32/484, train_loss: 0.0043\n",
      "33/484, train_loss: 0.0096\n",
      "34/484, train_loss: 0.0082\n",
      "35/484, train_loss: 0.5978\n",
      "36/484, train_loss: 0.0025\n",
      "37/484, train_loss: 0.2901\n",
      "38/484, train_loss: 0.0027\n",
      "39/484, train_loss: 0.0042\n",
      "40/484, train_loss: 0.0030\n",
      "41/484, train_loss: 0.0043\n",
      "42/484, train_loss: 0.0080\n",
      "43/484, train_loss: 0.0023\n",
      "44/484, train_loss: 0.0061\n",
      "45/484, train_loss: 0.0031\n",
      "46/484, train_loss: 0.0051\n",
      "47/484, train_loss: 0.2309\n",
      "48/484, train_loss: 0.0032\n",
      "49/484, train_loss: 0.3357\n",
      "50/484, train_loss: 0.0096\n",
      "51/484, train_loss: 0.0028\n",
      "52/484, train_loss: 0.1017\n",
      "53/484, train_loss: 0.0033\n",
      "54/484, train_loss: 0.0019\n",
      "55/484, train_loss: 0.0747\n",
      "56/484, train_loss: 0.0080\n",
      "57/484, train_loss: 0.0026\n",
      "58/484, train_loss: 0.1403\n",
      "59/484, train_loss: 0.0023\n",
      "60/484, train_loss: 0.0112\n",
      "61/484, train_loss: 0.0036\n",
      "62/484, train_loss: 0.4711\n",
      "63/484, train_loss: 0.0017\n",
      "64/484, train_loss: 0.0037\n",
      "65/484, train_loss: 0.0056\n",
      "66/484, train_loss: 0.0053\n",
      "67/484, train_loss: 0.0023\n",
      "68/484, train_loss: 0.0028\n",
      "69/484, train_loss: 0.0038\n",
      "70/484, train_loss: 0.1024\n",
      "71/484, train_loss: 0.0031\n",
      "72/484, train_loss: 0.0249\n",
      "73/484, train_loss: 0.0036\n",
      "74/484, train_loss: 0.1665\n",
      "75/484, train_loss: 0.2522\n",
      "76/484, train_loss: 0.0043\n",
      "77/484, train_loss: 0.0028\n",
      "78/484, train_loss: 0.0023\n",
      "79/484, train_loss: 0.0020\n",
      "80/484, train_loss: 0.0025\n",
      "81/484, train_loss: 0.0044\n",
      "82/484, train_loss: 0.0187\n",
      "83/484, train_loss: 0.0060\n",
      "84/484, train_loss: 0.0102\n",
      "85/484, train_loss: 0.0026\n",
      "86/484, train_loss: 0.0048\n",
      "87/484, train_loss: 0.1579\n",
      "88/484, train_loss: 0.0014\n",
      "89/484, train_loss: 0.8139\n",
      "90/484, train_loss: 0.0018\n",
      "91/484, train_loss: 0.0019\n",
      "92/484, train_loss: 0.0177\n",
      "93/484, train_loss: 0.0021\n",
      "94/484, train_loss: 0.0083\n",
      "95/484, train_loss: 0.0027\n",
      "96/484, train_loss: 0.0030\n",
      "97/484, train_loss: 0.0061\n",
      "98/484, train_loss: 0.0222\n",
      "99/484, train_loss: 0.0026\n",
      "100/484, train_loss: 0.0075\n",
      "101/484, train_loss: 0.0098\n",
      "102/484, train_loss: 0.0041\n",
      "103/484, train_loss: 0.0028\n",
      "104/484, train_loss: 0.2700\n",
      "105/484, train_loss: 0.0076\n",
      "106/484, train_loss: 0.0016\n",
      "107/484, train_loss: 0.6550\n",
      "108/484, train_loss: 0.0041\n",
      "109/484, train_loss: 0.0064\n",
      "110/484, train_loss: 0.0037\n",
      "111/484, train_loss: 0.0022\n",
      "112/484, train_loss: 0.0046\n",
      "113/484, train_loss: 0.0031\n",
      "114/484, train_loss: 0.0025\n",
      "115/484, train_loss: 0.0052\n",
      "116/484, train_loss: 0.0039\n",
      "117/484, train_loss: 0.0034\n",
      "118/484, train_loss: 0.0038\n",
      "119/484, train_loss: 0.2444\n",
      "120/484, train_loss: 0.0039\n",
      "121/484, train_loss: 0.0045\n",
      "122/484, train_loss: 0.0042\n",
      "123/484, train_loss: 0.0017\n",
      "124/484, train_loss: 0.5744\n",
      "125/484, train_loss: 0.0097\n",
      "126/484, train_loss: 0.0170\n",
      "127/484, train_loss: 0.0021\n",
      "128/484, train_loss: 0.0025\n",
      "129/484, train_loss: 0.4063\n",
      "130/484, train_loss: 0.0023\n",
      "131/484, train_loss: 0.0313\n",
      "132/484, train_loss: 0.0019\n",
      "133/484, train_loss: 0.3818\n",
      "134/484, train_loss: 0.3240\n",
      "135/484, train_loss: 0.0086\n",
      "136/484, train_loss: 0.0091\n",
      "137/484, train_loss: 0.0100\n",
      "138/484, train_loss: 0.1918\n",
      "139/484, train_loss: 0.0058\n",
      "140/484, train_loss: 0.2091\n",
      "141/484, train_loss: 0.0027\n",
      "142/484, train_loss: 0.0017\n",
      "143/484, train_loss: 0.0020\n",
      "144/484, train_loss: 0.0034\n",
      "145/484, train_loss: 0.0098\n",
      "146/484, train_loss: 0.0034\n",
      "147/484, train_loss: 0.0025\n",
      "148/484, train_loss: 0.0031\n",
      "149/484, train_loss: 0.0040\n",
      "150/484, train_loss: 0.0096\n",
      "151/484, train_loss: 0.0135\n",
      "152/484, train_loss: 0.0057\n",
      "153/484, train_loss: 0.0027\n",
      "154/484, train_loss: 0.2294\n",
      "155/484, train_loss: 0.0021\n",
      "156/484, train_loss: 0.0068\n",
      "157/484, train_loss: 0.0125\n",
      "158/484, train_loss: 0.0020\n",
      "159/484, train_loss: 0.0033\n",
      "160/484, train_loss: 0.0040\n",
      "161/484, train_loss: 0.0033\n",
      "162/484, train_loss: 0.0019\n",
      "163/484, train_loss: 0.0024\n",
      "164/484, train_loss: 0.1210\n",
      "165/484, train_loss: 0.0019\n",
      "166/484, train_loss: 0.0032\n",
      "167/484, train_loss: 0.0053\n",
      "168/484, train_loss: 0.0018\n",
      "169/484, train_loss: 0.0026\n",
      "170/484, train_loss: 0.0019\n",
      "171/484, train_loss: 0.1016\n",
      "172/484, train_loss: 0.0036\n",
      "173/484, train_loss: 0.1380\n",
      "174/484, train_loss: 0.0016\n",
      "175/484, train_loss: 0.0021\n",
      "176/484, train_loss: 0.0064\n",
      "177/484, train_loss: 0.0021\n",
      "178/484, train_loss: 0.0016\n",
      "179/484, train_loss: 0.1273\n",
      "180/484, train_loss: 0.0030\n",
      "181/484, train_loss: 1.3743\n",
      "182/484, train_loss: 0.0010\n",
      "183/484, train_loss: 0.0087\n",
      "184/484, train_loss: 0.2712\n",
      "185/484, train_loss: 0.0016\n",
      "186/484, train_loss: 0.0266\n",
      "187/484, train_loss: 0.0075\n",
      "188/484, train_loss: 0.0107\n",
      "189/484, train_loss: 0.0053\n",
      "190/484, train_loss: 0.0050\n",
      "191/484, train_loss: 0.0016\n",
      "192/484, train_loss: 0.0028\n",
      "193/484, train_loss: 0.4415\n",
      "194/484, train_loss: 0.0067\n",
      "195/484, train_loss: 0.1291\n",
      "196/484, train_loss: 0.1485\n",
      "197/484, train_loss: 0.0035\n",
      "198/484, train_loss: 0.0030\n",
      "199/484, train_loss: 1.9876\n",
      "200/484, train_loss: 0.0074\n",
      "201/484, train_loss: 0.0141\n",
      "202/484, train_loss: 0.4339\n",
      "203/484, train_loss: 0.0017\n",
      "204/484, train_loss: 0.0038\n",
      "205/484, train_loss: 0.0063\n",
      "206/484, train_loss: 0.0085\n",
      "207/484, train_loss: 0.1625\n",
      "208/484, train_loss: 0.0024\n",
      "209/484, train_loss: 0.0035\n",
      "210/484, train_loss: 0.0018\n",
      "211/484, train_loss: 1.4709\n",
      "212/484, train_loss: 0.7019\n",
      "213/484, train_loss: 0.0042\n",
      "214/484, train_loss: 0.0032\n",
      "215/484, train_loss: 0.0259\n",
      "216/484, train_loss: 0.0033\n",
      "217/484, train_loss: 0.1987\n",
      "218/484, train_loss: 0.5973\n",
      "219/484, train_loss: 0.0031\n",
      "220/484, train_loss: 0.0055\n",
      "221/484, train_loss: 0.0131\n",
      "222/484, train_loss: 0.1298\n",
      "223/484, train_loss: 0.0018\n",
      "224/484, train_loss: 0.8652\n",
      "225/484, train_loss: 0.9805\n",
      "226/484, train_loss: 0.2735\n",
      "227/484, train_loss: 0.1593\n",
      "228/484, train_loss: 0.0076\n",
      "229/484, train_loss: 0.0038\n",
      "230/484, train_loss: 0.0012\n",
      "231/484, train_loss: 0.0046\n",
      "232/484, train_loss: 0.6533\n",
      "233/484, train_loss: 0.1630\n",
      "234/484, train_loss: 0.0118\n",
      "235/484, train_loss: 0.4779\n",
      "236/484, train_loss: 0.0103\n",
      "237/484, train_loss: 0.0086\n",
      "238/484, train_loss: 0.0150\n",
      "239/484, train_loss: 0.0020\n",
      "240/484, train_loss: 0.0072\n",
      "241/484, train_loss: 0.0026\n",
      "242/484, train_loss: 0.0031\n",
      "243/484, train_loss: 0.0426\n",
      "244/484, train_loss: 0.0109\n",
      "245/484, train_loss: 0.0054\n",
      "246/484, train_loss: 0.0046\n",
      "247/484, train_loss: 0.0046\n",
      "248/484, train_loss: 0.0061\n",
      "249/484, train_loss: 0.0038\n",
      "250/484, train_loss: 0.0025\n",
      "251/484, train_loss: 0.0045\n",
      "252/484, train_loss: 0.0126\n",
      "253/484, train_loss: 0.3948\n",
      "254/484, train_loss: 0.0049\n",
      "255/484, train_loss: 0.0040\n",
      "256/484, train_loss: 0.0043\n",
      "257/484, train_loss: 0.0032\n",
      "258/484, train_loss: 0.0038\n",
      "259/484, train_loss: 0.0111\n",
      "260/484, train_loss: 0.0175\n",
      "261/484, train_loss: 0.0119\n",
      "262/484, train_loss: 0.0089\n",
      "263/484, train_loss: 0.6015\n",
      "264/484, train_loss: 0.0028\n",
      "265/484, train_loss: 0.0033\n",
      "266/484, train_loss: 0.0089\n",
      "267/484, train_loss: 0.0146\n",
      "268/484, train_loss: 0.0025\n",
      "269/484, train_loss: 0.0524\n",
      "270/484, train_loss: 0.0028\n",
      "271/484, train_loss: 0.0062\n",
      "272/484, train_loss: 0.0023\n",
      "273/484, train_loss: 0.0029\n",
      "274/484, train_loss: 0.0058\n",
      "275/484, train_loss: 0.0062\n",
      "276/484, train_loss: 0.0054\n",
      "277/484, train_loss: 0.2208\n",
      "278/484, train_loss: 0.0048\n",
      "279/484, train_loss: 0.0035\n",
      "280/484, train_loss: 0.0024\n",
      "281/484, train_loss: 0.0022\n",
      "282/484, train_loss: 0.0040\n",
      "283/484, train_loss: 0.1884\n",
      "284/484, train_loss: 0.0027\n",
      "285/484, train_loss: 0.0062\n",
      "286/484, train_loss: 0.0036\n",
      "287/484, train_loss: 0.0033\n",
      "288/484, train_loss: 0.0025\n",
      "289/484, train_loss: 0.0060\n",
      "290/484, train_loss: 0.0017\n",
      "291/484, train_loss: 0.2294\n",
      "292/484, train_loss: 0.0104\n",
      "293/484, train_loss: 0.0029\n",
      "294/484, train_loss: 0.9678\n",
      "295/484, train_loss: 0.8271\n",
      "296/484, train_loss: 0.0026\n",
      "297/484, train_loss: 1.2932\n",
      "298/484, train_loss: 0.0113\n",
      "299/484, train_loss: 0.0183\n",
      "300/484, train_loss: 0.0040\n",
      "301/484, train_loss: 0.0033\n",
      "302/484, train_loss: 0.0021\n",
      "303/484, train_loss: 0.0030\n",
      "304/484, train_loss: 0.0090\n",
      "305/484, train_loss: 0.0021\n",
      "306/484, train_loss: 0.0090\n",
      "307/484, train_loss: 0.2764\n",
      "308/484, train_loss: 0.0045\n",
      "309/484, train_loss: 0.1388\n",
      "310/484, train_loss: 0.0033\n",
      "311/484, train_loss: 1.0418\n",
      "312/484, train_loss: 0.0021\n",
      "313/484, train_loss: 0.2659\n",
      "314/484, train_loss: 0.0021\n",
      "315/484, train_loss: 0.2406\n",
      "316/484, train_loss: 0.0018\n",
      "317/484, train_loss: 0.0112\n",
      "318/484, train_loss: 0.0049\n",
      "319/484, train_loss: 0.0032\n",
      "320/484, train_loss: 0.0056\n",
      "321/484, train_loss: 0.0043\n",
      "322/484, train_loss: 0.0406\n",
      "323/484, train_loss: 0.0045\n",
      "324/484, train_loss: 0.2112\n",
      "325/484, train_loss: 0.0054\n",
      "326/484, train_loss: 0.0064\n",
      "327/484, train_loss: 0.0071\n",
      "328/484, train_loss: 0.3252\n",
      "329/484, train_loss: 0.0032\n",
      "330/484, train_loss: 0.0018\n",
      "331/484, train_loss: 0.0015\n",
      "332/484, train_loss: 0.0051\n",
      "333/484, train_loss: 0.0044\n",
      "334/484, train_loss: 0.0064\n",
      "335/484, train_loss: 0.0042\n",
      "336/484, train_loss: 0.8401\n",
      "337/484, train_loss: 0.0021\n",
      "338/484, train_loss: 0.0027\n",
      "339/484, train_loss: 0.5063\n",
      "340/484, train_loss: 0.0042\n",
      "341/484, train_loss: 0.0024\n",
      "342/484, train_loss: 0.0036\n",
      "343/484, train_loss: 0.0017\n",
      "344/484, train_loss: 0.0023\n",
      "345/484, train_loss: 1.1730\n",
      "346/484, train_loss: 0.2814\n",
      "347/484, train_loss: 0.0117\n",
      "348/484, train_loss: 0.0028\n",
      "349/484, train_loss: 0.0028\n",
      "350/484, train_loss: 0.0025\n",
      "351/484, train_loss: 0.0219\n",
      "352/484, train_loss: 0.0024\n",
      "353/484, train_loss: 0.3419\n",
      "354/484, train_loss: 0.0066\n",
      "355/484, train_loss: 0.0041\n",
      "356/484, train_loss: 0.0028\n",
      "357/484, train_loss: 0.3034\n",
      "358/484, train_loss: 0.0060\n",
      "359/484, train_loss: 0.0039\n",
      "360/484, train_loss: 0.0093\n",
      "361/484, train_loss: 0.0030\n",
      "362/484, train_loss: 0.0049\n",
      "363/484, train_loss: 0.0041\n",
      "364/484, train_loss: 0.0021\n",
      "365/484, train_loss: 0.0015\n",
      "366/484, train_loss: 0.9399\n",
      "367/484, train_loss: 0.0032\n",
      "368/484, train_loss: 0.3280\n",
      "369/484, train_loss: 0.0035\n",
      "370/484, train_loss: 0.0073\n",
      "371/484, train_loss: 0.0041\n",
      "372/484, train_loss: 0.0059\n",
      "373/484, train_loss: 0.0062\n",
      "374/484, train_loss: 0.0044\n",
      "375/484, train_loss: 0.0020\n",
      "376/484, train_loss: 0.0045\n",
      "377/484, train_loss: 0.0016\n",
      "378/484, train_loss: 0.0062\n",
      "379/484, train_loss: 0.3308\n",
      "380/484, train_loss: 0.0115\n",
      "381/484, train_loss: 0.0011\n",
      "382/484, train_loss: 0.0031\n",
      "383/484, train_loss: 1.0482\n",
      "384/484, train_loss: 0.0061\n",
      "385/484, train_loss: 0.0034\n",
      "386/484, train_loss: 0.1959\n",
      "387/484, train_loss: 0.0039\n",
      "388/484, train_loss: 0.0165\n",
      "389/484, train_loss: 0.0026\n",
      "390/484, train_loss: 0.0051\n",
      "391/484, train_loss: 0.0038\n",
      "392/484, train_loss: 0.0036\n",
      "393/484, train_loss: 0.0892\n",
      "394/484, train_loss: 0.0084\n",
      "395/484, train_loss: 0.0019\n",
      "396/484, train_loss: 0.0067\n",
      "397/484, train_loss: 0.0137\n",
      "398/484, train_loss: 1.0412\n",
      "399/484, train_loss: 0.0220\n",
      "400/484, train_loss: 0.0048\n",
      "401/484, train_loss: 0.0038\n",
      "402/484, train_loss: 0.0056\n",
      "403/484, train_loss: 0.0030\n",
      "404/484, train_loss: 0.0028\n",
      "405/484, train_loss: 0.0045\n",
      "406/484, train_loss: 0.0090\n",
      "407/484, train_loss: 0.0041\n",
      "408/484, train_loss: 0.5206\n",
      "409/484, train_loss: 0.0032\n",
      "410/484, train_loss: 0.0034\n",
      "411/484, train_loss: 0.0038\n",
      "412/484, train_loss: 0.0020\n",
      "413/484, train_loss: 0.0016\n",
      "414/484, train_loss: 0.1966\n",
      "415/484, train_loss: 0.0031\n",
      "416/484, train_loss: 0.0037\n",
      "417/484, train_loss: 0.0055\n",
      "418/484, train_loss: 0.0052\n",
      "419/484, train_loss: 0.0055\n",
      "420/484, train_loss: 0.0023\n",
      "421/484, train_loss: 0.0053\n",
      "422/484, train_loss: 0.2770\n",
      "423/484, train_loss: 0.0054\n",
      "424/484, train_loss: 0.0207\n",
      "425/484, train_loss: 1.4812\n",
      "426/484, train_loss: 0.6831\n",
      "427/484, train_loss: 0.0065\n",
      "428/484, train_loss: 0.2620\n",
      "429/484, train_loss: 0.0056\n",
      "430/484, train_loss: 0.0049\n",
      "431/484, train_loss: 0.0846\n",
      "432/484, train_loss: 0.0055\n",
      "433/484, train_loss: 0.0020\n",
      "434/484, train_loss: 0.8008\n",
      "435/484, train_loss: 0.0050\n",
      "436/484, train_loss: 0.0102\n",
      "437/484, train_loss: 0.0046\n",
      "438/484, train_loss: 0.0241\n",
      "439/484, train_loss: 0.0215\n",
      "440/484, train_loss: 0.0019\n",
      "441/484, train_loss: 3.3171\n",
      "442/484, train_loss: 0.0258\n",
      "443/484, train_loss: 0.0069\n",
      "444/484, train_loss: 0.0053\n",
      "445/484, train_loss: 0.0022\n",
      "446/484, train_loss: 0.0183\n",
      "447/484, train_loss: 0.0024\n",
      "448/484, train_loss: 0.0064\n",
      "449/484, train_loss: 0.0049\n",
      "450/484, train_loss: 0.0028\n",
      "451/484, train_loss: 0.0035\n",
      "452/484, train_loss: 0.0100\n",
      "453/484, train_loss: 0.0056\n",
      "454/484, train_loss: 0.0017\n",
      "455/484, train_loss: 0.0041\n",
      "456/484, train_loss: 0.3545\n",
      "457/484, train_loss: 0.0077\n",
      "458/484, train_loss: 0.9914\n",
      "459/484, train_loss: 0.1235\n",
      "460/484, train_loss: 0.0644\n",
      "461/484, train_loss: 0.0085\n",
      "462/484, train_loss: 1.1474\n",
      "463/484, train_loss: 0.0044\n",
      "464/484, train_loss: 0.0057\n",
      "465/484, train_loss: 0.0031\n",
      "466/484, train_loss: 0.1811\n",
      "467/484, train_loss: 0.0054\n",
      "468/484, train_loss: 0.0083\n",
      "469/484, train_loss: 0.0583\n",
      "470/484, train_loss: 0.1379\n",
      "471/484, train_loss: 0.0048\n",
      "472/484, train_loss: 0.0040\n",
      "473/484, train_loss: 0.0153\n",
      "474/484, train_loss: 0.0092\n",
      "475/484, train_loss: 0.0095\n",
      "476/484, train_loss: 0.0091\n",
      "477/484, train_loss: 0.1373\n",
      "478/484, train_loss: 0.0313\n",
      "479/484, train_loss: 0.0014\n",
      "480/484, train_loss: 0.0169\n",
      "481/484, train_loss: 0.3577\n",
      "482/484, train_loss: 0.0605\n",
      "483/484, train_loss: 0.0019\n",
      "484/484, train_loss: 0.0041\n",
      "epoch 8 average loss: 0.1131\n",
      "Current epoch: 8 current accuracy: 0.8266 \n",
      "Best accuracy: 0.8297 at epoch 6\n",
      "----------\n",
      "epoch 9/20\n",
      "1/484, train_loss: 0.0159\n",
      "2/484, train_loss: 0.0136\n",
      "3/484, train_loss: 0.8046\n",
      "4/484, train_loss: 0.0111\n",
      "5/484, train_loss: 0.0162\n",
      "6/484, train_loss: 0.0168\n",
      "7/484, train_loss: 0.0042\n",
      "8/484, train_loss: 0.0025\n",
      "9/484, train_loss: 0.0034\n",
      "10/484, train_loss: 0.0017\n",
      "11/484, train_loss: 0.0660\n",
      "12/484, train_loss: 0.0154\n",
      "13/484, train_loss: 0.0034\n",
      "14/484, train_loss: 0.0018\n",
      "15/484, train_loss: 0.0083\n",
      "16/484, train_loss: 0.0038\n",
      "17/484, train_loss: 0.0163\n",
      "18/484, train_loss: 0.0170\n",
      "19/484, train_loss: 0.0024\n",
      "20/484, train_loss: 0.7054\n",
      "21/484, train_loss: 0.0071\n",
      "22/484, train_loss: 0.0042\n",
      "23/484, train_loss: 0.0054\n",
      "24/484, train_loss: 0.4833\n",
      "25/484, train_loss: 0.9447\n",
      "26/484, train_loss: 0.3323\n",
      "27/484, train_loss: 0.1886\n",
      "28/484, train_loss: 0.0607\n",
      "29/484, train_loss: 0.4249\n",
      "30/484, train_loss: 0.0085\n",
      "31/484, train_loss: 0.0029\n",
      "32/484, train_loss: 0.0058\n",
      "33/484, train_loss: 0.0049\n",
      "34/484, train_loss: 0.0038\n",
      "35/484, train_loss: 0.0046\n",
      "36/484, train_loss: 0.0082\n",
      "37/484, train_loss: 0.4866\n",
      "38/484, train_loss: 0.0045\n",
      "39/484, train_loss: 0.0016\n",
      "40/484, train_loss: 0.8256\n",
      "41/484, train_loss: 0.0036\n",
      "42/484, train_loss: 0.0054\n",
      "43/484, train_loss: 0.2858\n",
      "44/484, train_loss: 0.0134\n",
      "45/484, train_loss: 0.0016\n",
      "46/484, train_loss: 0.0150\n",
      "47/484, train_loss: 0.0052\n",
      "48/484, train_loss: 0.0193\n",
      "49/484, train_loss: 0.0182\n",
      "50/484, train_loss: 1.4030\n",
      "51/484, train_loss: 0.0526\n",
      "52/484, train_loss: 0.0041\n",
      "53/484, train_loss: 0.0056\n",
      "54/484, train_loss: 0.0066\n",
      "55/484, train_loss: 0.0110\n",
      "56/484, train_loss: 0.0067\n",
      "57/484, train_loss: 0.0034\n",
      "58/484, train_loss: 0.0047\n",
      "59/484, train_loss: 0.0035\n",
      "60/484, train_loss: 0.0028\n",
      "61/484, train_loss: 0.0020\n",
      "62/484, train_loss: 0.0254\n",
      "63/484, train_loss: 0.8745\n",
      "64/484, train_loss: 0.0028\n",
      "65/484, train_loss: 0.0029\n",
      "66/484, train_loss: 0.4910\n",
      "67/484, train_loss: 0.0068\n",
      "68/484, train_loss: 0.3376\n",
      "69/484, train_loss: 0.0020\n",
      "70/484, train_loss: 0.0025\n",
      "71/484, train_loss: 0.9235\n",
      "72/484, train_loss: 0.2802\n",
      "73/484, train_loss: 0.0043\n",
      "74/484, train_loss: 0.0022\n",
      "75/484, train_loss: 0.0013\n",
      "76/484, train_loss: 0.0032\n",
      "77/484, train_loss: 0.0020\n",
      "78/484, train_loss: 0.0016\n",
      "79/484, train_loss: 0.0033\n",
      "80/484, train_loss: 1.2337\n",
      "81/484, train_loss: 0.4190\n",
      "82/484, train_loss: 0.0026\n",
      "83/484, train_loss: 1.2368\n",
      "84/484, train_loss: 0.0043\n",
      "85/484, train_loss: 0.0037\n",
      "86/484, train_loss: 0.1006\n",
      "87/484, train_loss: 0.0044\n",
      "88/484, train_loss: 0.0483\n",
      "89/484, train_loss: 0.0029\n",
      "90/484, train_loss: 0.0021\n",
      "91/484, train_loss: 0.0017\n",
      "92/484, train_loss: 0.0032\n",
      "93/484, train_loss: 0.0762\n",
      "94/484, train_loss: 0.0021\n",
      "95/484, train_loss: 0.0055\n",
      "96/484, train_loss: 0.0054\n",
      "97/484, train_loss: 0.0117\n",
      "98/484, train_loss: 0.0018\n",
      "99/484, train_loss: 0.0045\n",
      "100/484, train_loss: 0.4081\n",
      "101/484, train_loss: 0.0029\n",
      "102/484, train_loss: 0.0042\n",
      "103/484, train_loss: 0.0096\n",
      "104/484, train_loss: 0.0047\n",
      "105/484, train_loss: 0.0036\n",
      "106/484, train_loss: 0.0019\n",
      "107/484, train_loss: 0.0073\n",
      "108/484, train_loss: 0.0018\n",
      "109/484, train_loss: 0.0080\n",
      "110/484, train_loss: 0.0017\n",
      "111/484, train_loss: 0.0034\n",
      "112/484, train_loss: 0.0047\n",
      "113/484, train_loss: 0.0047\n",
      "114/484, train_loss: 0.0088\n",
      "115/484, train_loss: 0.0041\n",
      "116/484, train_loss: 0.2124\n",
      "117/484, train_loss: 0.0068\n",
      "118/484, train_loss: 0.3747\n",
      "119/484, train_loss: 0.0068\n",
      "120/484, train_loss: 0.0096\n",
      "121/484, train_loss: 0.0034\n",
      "122/484, train_loss: 0.3289\n",
      "123/484, train_loss: 0.0030\n",
      "124/484, train_loss: 0.0026\n",
      "125/484, train_loss: 0.0028\n",
      "126/484, train_loss: 0.4557\n",
      "127/484, train_loss: 0.0037\n",
      "128/484, train_loss: 0.3434\n",
      "129/484, train_loss: 0.0019\n",
      "130/484, train_loss: 0.0056\n",
      "131/484, train_loss: 0.0269\n",
      "132/484, train_loss: 0.0055\n",
      "133/484, train_loss: 0.0045\n",
      "134/484, train_loss: 0.0029\n",
      "135/484, train_loss: 0.0025\n",
      "136/484, train_loss: 0.0022\n",
      "137/484, train_loss: 0.0254\n",
      "138/484, train_loss: 0.0038\n",
      "139/484, train_loss: 0.0041\n",
      "140/484, train_loss: 0.0034\n",
      "141/484, train_loss: 0.0037\n",
      "142/484, train_loss: 0.0021\n",
      "143/484, train_loss: 1.1073\n",
      "144/484, train_loss: 0.0132\n",
      "145/484, train_loss: 0.0045\n",
      "146/484, train_loss: 0.0019\n",
      "147/484, train_loss: 0.7002\n",
      "148/484, train_loss: 0.1995\n",
      "149/484, train_loss: 0.0115\n",
      "150/484, train_loss: 0.0075\n",
      "151/484, train_loss: 0.0018\n",
      "152/484, train_loss: 0.0024\n",
      "153/484, train_loss: 0.4512\n",
      "154/484, train_loss: 0.0015\n",
      "155/484, train_loss: 0.0039\n",
      "156/484, train_loss: 0.8247\n",
      "157/484, train_loss: 0.0119\n",
      "158/484, train_loss: 0.0023\n",
      "159/484, train_loss: 0.0025\n",
      "160/484, train_loss: 0.0016\n",
      "161/484, train_loss: 0.0024\n",
      "162/484, train_loss: 0.2668\n",
      "163/484, train_loss: 0.0045\n",
      "164/484, train_loss: 0.0032\n",
      "165/484, train_loss: 0.0041\n",
      "166/484, train_loss: 0.7471\n",
      "167/484, train_loss: 0.0036\n",
      "168/484, train_loss: 0.0040\n",
      "169/484, train_loss: 0.0020\n",
      "170/484, train_loss: 0.0037\n",
      "171/484, train_loss: 0.0029\n",
      "172/484, train_loss: 0.0020\n",
      "173/484, train_loss: 0.2619\n",
      "174/484, train_loss: 0.0139\n",
      "175/484, train_loss: 0.5356\n",
      "176/484, train_loss: 0.1927\n",
      "177/484, train_loss: 0.0425\n",
      "178/484, train_loss: 0.0037\n",
      "179/484, train_loss: 0.0038\n",
      "180/484, train_loss: 0.0038\n",
      "181/484, train_loss: 0.1049\n",
      "182/484, train_loss: 0.0052\n",
      "183/484, train_loss: 0.5983\n",
      "184/484, train_loss: 0.0360\n",
      "185/484, train_loss: 1.6703\n",
      "186/484, train_loss: 2.4963\n",
      "187/484, train_loss: 0.5085\n",
      "188/484, train_loss: 0.0030\n",
      "189/484, train_loss: 0.4058\n",
      "190/484, train_loss: 0.0041\n",
      "191/484, train_loss: 0.0052\n",
      "192/484, train_loss: 0.4620\n",
      "193/484, train_loss: 0.6818\n",
      "194/484, train_loss: 0.0072\n",
      "195/484, train_loss: 0.0042\n",
      "196/484, train_loss: 0.0074\n",
      "197/484, train_loss: 0.0028\n",
      "198/484, train_loss: 0.0041\n",
      "199/484, train_loss: 0.0043\n",
      "200/484, train_loss: 0.0050\n",
      "201/484, train_loss: 1.1637\n",
      "202/484, train_loss: 0.0031\n",
      "203/484, train_loss: 0.3317\n",
      "204/484, train_loss: 0.8644\n",
      "205/484, train_loss: 0.2739\n",
      "206/484, train_loss: 0.0076\n",
      "207/484, train_loss: 0.3370\n",
      "208/484, train_loss: 0.0029\n",
      "209/484, train_loss: 0.0028\n",
      "210/484, train_loss: 0.0030\n",
      "211/484, train_loss: 0.0031\n",
      "212/484, train_loss: 0.0028\n",
      "213/484, train_loss: 0.0115\n",
      "214/484, train_loss: 0.4067\n",
      "215/484, train_loss: 0.0029\n",
      "216/484, train_loss: 0.0074\n",
      "217/484, train_loss: 0.0098\n",
      "218/484, train_loss: 0.0051\n",
      "219/484, train_loss: 0.0019\n",
      "220/484, train_loss: 0.0031\n",
      "221/484, train_loss: 0.0024\n",
      "222/484, train_loss: 0.5277\n",
      "223/484, train_loss: 0.0044\n",
      "224/484, train_loss: 0.0039\n",
      "225/484, train_loss: 0.0020\n",
      "226/484, train_loss: 0.0032\n",
      "227/484, train_loss: 0.4225\n",
      "228/484, train_loss: 0.0028\n",
      "229/484, train_loss: 0.0153\n",
      "230/484, train_loss: 0.0023\n",
      "231/484, train_loss: 0.2334\n",
      "232/484, train_loss: 0.0102\n",
      "233/484, train_loss: 0.4589\n",
      "234/484, train_loss: 0.0035\n",
      "235/484, train_loss: 0.0034\n",
      "236/484, train_loss: 0.0030\n",
      "237/484, train_loss: 0.0035\n",
      "238/484, train_loss: 0.0030\n",
      "239/484, train_loss: 0.0122\n",
      "240/484, train_loss: 0.0250\n",
      "241/484, train_loss: 0.2904\n",
      "242/484, train_loss: 0.0037\n",
      "243/484, train_loss: 1.0392\n",
      "244/484, train_loss: 0.0025\n",
      "245/484, train_loss: 0.0046\n",
      "246/484, train_loss: 0.0027\n",
      "247/484, train_loss: 0.0068\n",
      "248/484, train_loss: 0.0135\n",
      "249/484, train_loss: 0.1531\n",
      "250/484, train_loss: 0.0142\n",
      "251/484, train_loss: 0.9262\n",
      "252/484, train_loss: 0.2321\n",
      "253/484, train_loss: 0.0050\n",
      "254/484, train_loss: 0.0043\n",
      "255/484, train_loss: 0.0229\n",
      "256/484, train_loss: 0.0061\n",
      "257/484, train_loss: 0.0039\n",
      "258/484, train_loss: 0.0036\n",
      "259/484, train_loss: 0.0193\n",
      "260/484, train_loss: 0.0052\n",
      "261/484, train_loss: 1.3337\n",
      "262/484, train_loss: 0.0012\n",
      "263/484, train_loss: 0.8905\n",
      "264/484, train_loss: 0.0438\n",
      "265/484, train_loss: 0.0042\n",
      "266/484, train_loss: 0.0058\n",
      "267/484, train_loss: 0.0126\n",
      "268/484, train_loss: 0.0059\n",
      "269/484, train_loss: 0.0042\n",
      "270/484, train_loss: 0.5283\n",
      "271/484, train_loss: 0.0031\n",
      "272/484, train_loss: 0.0038\n",
      "273/484, train_loss: 0.0117\n",
      "274/484, train_loss: 0.0030\n",
      "275/484, train_loss: 0.3124\n",
      "276/484, train_loss: 0.4317\n",
      "277/484, train_loss: 0.0038\n",
      "278/484, train_loss: 0.4397\n",
      "279/484, train_loss: 0.0361\n",
      "280/484, train_loss: 0.0062\n",
      "281/484, train_loss: 0.0015\n",
      "282/484, train_loss: 0.0025\n",
      "283/484, train_loss: 0.3967\n",
      "284/484, train_loss: 0.0088\n",
      "285/484, train_loss: 0.0017\n",
      "286/484, train_loss: 0.0054\n",
      "287/484, train_loss: 0.0053\n",
      "288/484, train_loss: 0.0049\n",
      "289/484, train_loss: 0.0017\n",
      "290/484, train_loss: 0.0087\n",
      "291/484, train_loss: 0.0241\n",
      "292/484, train_loss: 0.0028\n",
      "293/484, train_loss: 0.2623\n",
      "294/484, train_loss: 0.7149\n",
      "295/484, train_loss: 0.0027\n",
      "296/484, train_loss: 0.8409\n",
      "297/484, train_loss: 0.0017\n",
      "298/484, train_loss: 0.0015\n",
      "299/484, train_loss: 0.0032\n",
      "300/484, train_loss: 0.0056\n",
      "301/484, train_loss: 0.0047\n",
      "302/484, train_loss: 0.0074\n",
      "303/484, train_loss: 0.0019\n",
      "304/484, train_loss: 0.2968\n",
      "305/484, train_loss: 0.0035\n",
      "306/484, train_loss: 0.0245\n",
      "307/484, train_loss: 0.1902\n",
      "308/484, train_loss: 0.0065\n",
      "309/484, train_loss: 0.0022\n",
      "310/484, train_loss: 0.2917\n",
      "311/484, train_loss: 0.0053\n",
      "312/484, train_loss: 0.0015\n",
      "313/484, train_loss: 0.0029\n",
      "314/484, train_loss: 0.0036\n",
      "315/484, train_loss: 0.0020\n",
      "316/484, train_loss: 0.0041\n",
      "317/484, train_loss: 0.5842\n",
      "318/484, train_loss: 0.0022\n",
      "319/484, train_loss: 0.3216\n",
      "320/484, train_loss: 0.0070\n",
      "321/484, train_loss: 0.0017\n",
      "322/484, train_loss: 0.0026\n",
      "323/484, train_loss: 0.0333\n",
      "324/484, train_loss: 0.0033\n",
      "325/484, train_loss: 0.1767\n",
      "326/484, train_loss: 0.0044\n",
      "327/484, train_loss: 0.0064\n",
      "328/484, train_loss: 0.0111\n",
      "329/484, train_loss: 0.2682\n",
      "330/484, train_loss: 0.0021\n",
      "331/484, train_loss: 0.0010\n",
      "332/484, train_loss: 0.0095\n",
      "333/484, train_loss: 0.0037\n",
      "334/484, train_loss: 0.0047\n",
      "335/484, train_loss: 0.0029\n",
      "336/484, train_loss: 0.0024\n",
      "337/484, train_loss: 1.1000\n",
      "338/484, train_loss: 0.0014\n",
      "339/484, train_loss: 0.1994\n",
      "340/484, train_loss: 0.0043\n",
      "341/484, train_loss: 0.0049\n",
      "342/484, train_loss: 0.0015\n",
      "343/484, train_loss: 0.0037\n",
      "344/484, train_loss: 0.0145\n",
      "345/484, train_loss: 0.0027\n",
      "346/484, train_loss: 0.0039\n",
      "347/484, train_loss: 0.4514\n",
      "348/484, train_loss: 0.2797\n",
      "349/484, train_loss: 0.7448\n",
      "350/484, train_loss: 0.0131\n",
      "351/484, train_loss: 0.0326\n",
      "352/484, train_loss: 0.0094\n",
      "353/484, train_loss: 0.0042\n",
      "354/484, train_loss: 0.2342\n",
      "355/484, train_loss: 0.1529\n",
      "356/484, train_loss: 0.0031\n",
      "357/484, train_loss: 0.4090\n",
      "358/484, train_loss: 0.0014\n",
      "359/484, train_loss: 0.6819\n",
      "360/484, train_loss: 0.0061\n",
      "361/484, train_loss: 0.0063\n",
      "362/484, train_loss: 0.0048\n",
      "363/484, train_loss: 0.0041\n",
      "364/484, train_loss: 0.0073\n",
      "365/484, train_loss: 0.0048\n",
      "366/484, train_loss: 0.0033\n",
      "367/484, train_loss: 0.0020\n",
      "368/484, train_loss: 0.0149\n",
      "369/484, train_loss: 0.0023\n",
      "370/484, train_loss: 0.7645\n",
      "371/484, train_loss: 0.0136\n",
      "372/484, train_loss: 0.0015\n",
      "373/484, train_loss: 0.0026\n",
      "374/484, train_loss: 0.0017\n",
      "375/484, train_loss: 0.0164\n",
      "376/484, train_loss: 0.0018\n",
      "377/484, train_loss: 1.1224\n",
      "378/484, train_loss: 0.0018\n",
      "379/484, train_loss: 0.0025\n",
      "380/484, train_loss: 0.0028\n",
      "381/484, train_loss: 0.0010\n",
      "382/484, train_loss: 0.0016\n",
      "383/484, train_loss: 0.6431\n",
      "384/484, train_loss: 0.0020\n",
      "385/484, train_loss: 0.0093\n",
      "386/484, train_loss: 0.0029\n",
      "387/484, train_loss: 0.0012\n",
      "388/484, train_loss: 0.0011\n",
      "389/484, train_loss: 0.0018\n",
      "390/484, train_loss: 0.6658\n",
      "391/484, train_loss: 0.2524\n",
      "392/484, train_loss: 0.2766\n",
      "393/484, train_loss: 0.0011\n",
      "394/484, train_loss: 0.0015\n",
      "395/484, train_loss: 0.2040\n",
      "396/484, train_loss: 0.0156\n",
      "397/484, train_loss: 0.0299\n",
      "398/484, train_loss: 0.0082\n",
      "399/484, train_loss: 0.0023\n",
      "400/484, train_loss: 0.0018\n",
      "401/484, train_loss: 0.0026\n",
      "402/484, train_loss: 0.0034\n",
      "403/484, train_loss: 0.0035\n",
      "404/484, train_loss: 0.0044\n",
      "405/484, train_loss: 0.0023\n",
      "406/484, train_loss: 0.0063\n",
      "407/484, train_loss: 0.0022\n",
      "408/484, train_loss: 0.3196\n",
      "409/484, train_loss: 0.0028\n",
      "410/484, train_loss: 0.0013\n",
      "411/484, train_loss: 0.6968\n",
      "412/484, train_loss: 0.0053\n",
      "413/484, train_loss: 0.1123\n",
      "414/484, train_loss: 0.8257\n",
      "415/484, train_loss: 0.0015\n",
      "416/484, train_loss: 0.0051\n",
      "417/484, train_loss: 0.0135\n",
      "418/484, train_loss: 0.0013\n",
      "419/484, train_loss: 0.0026\n",
      "420/484, train_loss: 0.0017\n",
      "421/484, train_loss: 0.0106\n",
      "422/484, train_loss: 0.0021\n",
      "423/484, train_loss: 0.0023\n",
      "424/484, train_loss: 0.4137\n",
      "425/484, train_loss: 0.0009\n",
      "426/484, train_loss: 0.0038\n",
      "427/484, train_loss: 0.1324\n",
      "428/484, train_loss: 0.3748\n",
      "429/484, train_loss: 0.0045\n",
      "430/484, train_loss: 0.0012\n",
      "431/484, train_loss: 0.5951\n",
      "432/484, train_loss: 0.0707\n",
      "433/484, train_loss: 0.0019\n",
      "434/484, train_loss: 0.0042\n",
      "435/484, train_loss: 0.0057\n",
      "436/484, train_loss: 0.2991\n",
      "437/484, train_loss: 0.0062\n",
      "438/484, train_loss: 0.0030\n",
      "439/484, train_loss: 0.0025\n",
      "440/484, train_loss: 0.0026\n",
      "441/484, train_loss: 0.0031\n",
      "442/484, train_loss: 0.0050\n",
      "443/484, train_loss: 0.0046\n",
      "444/484, train_loss: 0.0088\n",
      "445/484, train_loss: 0.0047\n",
      "446/484, train_loss: 0.5423\n",
      "447/484, train_loss: 0.5883\n",
      "448/484, train_loss: 0.3386\n",
      "449/484, train_loss: 0.5364\n",
      "450/484, train_loss: 1.1482\n",
      "451/484, train_loss: 0.0334\n",
      "452/484, train_loss: 0.0023\n",
      "453/484, train_loss: 0.0016\n",
      "454/484, train_loss: 0.0014\n",
      "455/484, train_loss: 0.0061\n",
      "456/484, train_loss: 0.5771\n",
      "457/484, train_loss: 0.0050\n",
      "458/484, train_loss: 2.5902\n",
      "459/484, train_loss: 0.2669\n",
      "460/484, train_loss: 0.0028\n",
      "461/484, train_loss: 0.0172\n",
      "462/484, train_loss: 0.0030\n",
      "463/484, train_loss: 0.0052\n",
      "464/484, train_loss: 0.0227\n",
      "465/484, train_loss: 0.0025\n",
      "466/484, train_loss: 0.0030\n",
      "467/484, train_loss: 0.4445\n",
      "468/484, train_loss: 0.0028\n",
      "469/484, train_loss: 0.0018\n",
      "470/484, train_loss: 0.0127\n",
      "471/484, train_loss: 0.9774\n",
      "472/484, train_loss: 0.0049\n",
      "473/484, train_loss: 0.0111\n",
      "474/484, train_loss: 0.3328\n",
      "475/484, train_loss: 0.0148\n",
      "476/484, train_loss: 0.4996\n",
      "477/484, train_loss: 0.8957\n",
      "478/484, train_loss: 0.0032\n",
      "479/484, train_loss: 0.0013\n",
      "480/484, train_loss: 0.0075\n",
      "481/484, train_loss: 0.0027\n",
      "482/484, train_loss: 0.0013\n",
      "483/484, train_loss: 0.0078\n",
      "484/484, train_loss: 0.3681\n",
      "epoch 9 average loss: 0.1399\n",
      "----------\n",
      "epoch 10/20\n",
      "1/484, train_loss: 0.0011\n",
      "2/484, train_loss: 0.0011\n",
      "3/484, train_loss: 0.4776\n",
      "4/484, train_loss: 0.0018\n",
      "5/484, train_loss: 0.0012\n",
      "6/484, train_loss: 0.0024\n",
      "7/484, train_loss: 0.0031\n",
      "8/484, train_loss: 0.0020\n",
      "9/484, train_loss: 0.0034\n",
      "10/484, train_loss: 0.0053\n",
      "11/484, train_loss: 0.0094\n",
      "12/484, train_loss: 0.0018\n",
      "13/484, train_loss: 0.0012\n",
      "14/484, train_loss: 0.0027\n",
      "15/484, train_loss: 0.0027\n",
      "16/484, train_loss: 0.5069\n",
      "17/484, train_loss: 0.0071\n",
      "18/484, train_loss: 0.0018\n",
      "19/484, train_loss: 0.6186\n",
      "20/484, train_loss: 0.0038\n",
      "21/484, train_loss: 0.0021\n",
      "22/484, train_loss: 0.0115\n",
      "23/484, train_loss: 0.3691\n",
      "24/484, train_loss: 0.9191\n",
      "25/484, train_loss: 0.4978\n",
      "26/484, train_loss: 0.0032\n",
      "27/484, train_loss: 0.1056\n",
      "28/484, train_loss: 0.0195\n",
      "29/484, train_loss: 0.0037\n",
      "30/484, train_loss: 0.0030\n",
      "31/484, train_loss: 0.0050\n",
      "32/484, train_loss: 0.0025\n",
      "33/484, train_loss: 0.1916\n",
      "34/484, train_loss: 0.0043\n",
      "35/484, train_loss: 0.0024\n",
      "36/484, train_loss: 0.0028\n",
      "37/484, train_loss: 0.0050\n",
      "38/484, train_loss: 0.2213\n",
      "39/484, train_loss: 0.0019\n",
      "40/484, train_loss: 0.0028\n",
      "41/484, train_loss: 0.0343\n",
      "42/484, train_loss: 0.5450\n",
      "43/484, train_loss: 0.0028\n",
      "44/484, train_loss: 0.0018\n",
      "45/484, train_loss: 0.0056\n",
      "46/484, train_loss: 0.0011\n",
      "47/484, train_loss: 0.0064\n",
      "48/484, train_loss: 0.0015\n",
      "49/484, train_loss: 0.0039\n",
      "50/484, train_loss: 0.0023\n",
      "51/484, train_loss: 0.0023\n",
      "52/484, train_loss: 0.0036\n",
      "53/484, train_loss: 0.0062\n",
      "54/484, train_loss: 0.0026\n",
      "55/484, train_loss: 0.0018\n",
      "56/484, train_loss: 0.3325\n",
      "57/484, train_loss: 0.0021\n",
      "58/484, train_loss: 0.1041\n",
      "59/484, train_loss: 0.0973\n",
      "60/484, train_loss: 0.0034\n",
      "61/484, train_loss: 0.0014\n",
      "62/484, train_loss: 0.0014\n",
      "63/484, train_loss: 1.3111\n",
      "64/484, train_loss: 0.2643\n",
      "65/484, train_loss: 0.0054\n",
      "66/484, train_loss: 0.0941\n",
      "67/484, train_loss: 0.0175\n",
      "68/484, train_loss: 0.0035\n",
      "69/484, train_loss: 0.4249\n",
      "70/484, train_loss: 0.0019\n",
      "71/484, train_loss: 0.0055\n",
      "72/484, train_loss: 0.0054\n",
      "73/484, train_loss: 0.0498\n",
      "74/484, train_loss: 0.0038\n",
      "75/484, train_loss: 0.0021\n",
      "76/484, train_loss: 0.0035\n",
      "77/484, train_loss: 0.2073\n",
      "78/484, train_loss: 0.0019\n",
      "79/484, train_loss: 0.0068\n",
      "80/484, train_loss: 0.0036\n",
      "81/484, train_loss: 0.0024\n",
      "82/484, train_loss: 0.0668\n",
      "83/484, train_loss: 0.0239\n",
      "84/484, train_loss: 0.0199\n",
      "85/484, train_loss: 0.1675\n",
      "86/484, train_loss: 0.0084\n",
      "87/484, train_loss: 0.0155\n",
      "88/484, train_loss: 0.0020\n",
      "89/484, train_loss: 0.0066\n",
      "90/484, train_loss: 0.0027\n",
      "91/484, train_loss: 0.2228\n",
      "92/484, train_loss: 0.0014\n",
      "93/484, train_loss: 0.0015\n",
      "94/484, train_loss: 0.0035\n",
      "95/484, train_loss: 0.0042\n",
      "96/484, train_loss: 0.0540\n",
      "97/484, train_loss: 0.8652\n",
      "98/484, train_loss: 0.7174\n",
      "99/484, train_loss: 0.0965\n",
      "100/484, train_loss: 0.0206\n",
      "101/484, train_loss: 0.0133\n",
      "102/484, train_loss: 0.0112\n",
      "103/484, train_loss: 0.0148\n",
      "104/484, train_loss: 0.0107\n",
      "105/484, train_loss: 1.3319\n",
      "106/484, train_loss: 0.0045\n",
      "107/484, train_loss: 0.0587\n",
      "108/484, train_loss: 0.0364\n",
      "109/484, train_loss: 0.0319\n",
      "110/484, train_loss: 0.0249\n",
      "111/484, train_loss: 0.0165\n",
      "112/484, train_loss: 0.0073\n",
      "113/484, train_loss: 0.2567\n",
      "114/484, train_loss: 0.0008\n",
      "115/484, train_loss: 0.0157\n",
      "116/484, train_loss: 0.0047\n",
      "117/484, train_loss: 0.6300\n",
      "118/484, train_loss: 0.0143\n",
      "119/484, train_loss: 1.8093\n",
      "120/484, train_loss: 0.0026\n",
      "121/484, train_loss: 1.5576\n",
      "122/484, train_loss: 0.0029\n",
      "123/484, train_loss: 0.0270\n",
      "124/484, train_loss: 0.0022\n",
      "125/484, train_loss: 0.3368\n",
      "126/484, train_loss: 0.0105\n",
      "127/484, train_loss: 0.2297\n",
      "128/484, train_loss: 0.0300\n",
      "129/484, train_loss: 0.0705\n",
      "130/484, train_loss: 0.7840\n",
      "131/484, train_loss: 1.7792\n",
      "132/484, train_loss: 0.5579\n",
      "133/484, train_loss: 0.0142\n",
      "134/484, train_loss: 0.8697\n",
      "135/484, train_loss: 0.0031\n",
      "136/484, train_loss: 0.0094\n",
      "137/484, train_loss: 0.0050\n",
      "138/484, train_loss: 0.0027\n",
      "139/484, train_loss: 0.6559\n",
      "140/484, train_loss: 0.0038\n",
      "141/484, train_loss: 0.3687\n",
      "142/484, train_loss: 0.0168\n",
      "143/484, train_loss: 0.0129\n",
      "144/484, train_loss: 0.0300\n",
      "145/484, train_loss: 0.0308\n",
      "146/484, train_loss: 0.8316\n",
      "147/484, train_loss: 1.9499\n",
      "148/484, train_loss: 0.1461\n",
      "149/484, train_loss: 0.0064\n",
      "150/484, train_loss: 0.1480\n",
      "151/484, train_loss: 0.0154\n",
      "152/484, train_loss: 0.0030\n",
      "153/484, train_loss: 0.0101\n",
      "154/484, train_loss: 0.0348\n",
      "155/484, train_loss: 0.9819\n",
      "156/484, train_loss: 0.7705\n",
      "157/484, train_loss: 0.8267\n",
      "158/484, train_loss: 0.0070\n",
      "159/484, train_loss: 0.0155\n",
      "160/484, train_loss: 0.0128\n",
      "161/484, train_loss: 0.0801\n",
      "162/484, train_loss: 0.0160\n",
      "163/484, train_loss: 0.0143\n",
      "164/484, train_loss: 0.0220\n",
      "165/484, train_loss: 0.4958\n",
      "166/484, train_loss: 0.0061\n",
      "167/484, train_loss: 0.0974\n",
      "168/484, train_loss: 1.2385\n",
      "169/484, train_loss: 0.3108\n",
      "170/484, train_loss: 0.9501\n",
      "171/484, train_loss: 0.2735\n",
      "172/484, train_loss: 0.3459\n",
      "173/484, train_loss: 0.0336\n",
      "174/484, train_loss: 0.0043\n",
      "175/484, train_loss: 0.0242\n",
      "176/484, train_loss: 0.0057\n",
      "177/484, train_loss: 0.0061\n",
      "178/484, train_loss: 0.2383\n",
      "179/484, train_loss: 0.0121\n",
      "180/484, train_loss: 0.0190\n",
      "181/484, train_loss: 0.4252\n",
      "182/484, train_loss: 0.0483\n",
      "183/484, train_loss: 0.0070\n",
      "184/484, train_loss: 0.4544\n",
      "185/484, train_loss: 0.3107\n",
      "186/484, train_loss: 0.0049\n",
      "187/484, train_loss: 0.1153\n",
      "188/484, train_loss: 0.3607\n",
      "189/484, train_loss: 0.0087\n",
      "190/484, train_loss: 0.0023\n",
      "191/484, train_loss: 0.0570\n",
      "192/484, train_loss: 0.0121\n",
      "193/484, train_loss: 0.0177\n",
      "194/484, train_loss: 0.0110\n",
      "195/484, train_loss: 0.0049\n",
      "196/484, train_loss: 0.0081\n",
      "197/484, train_loss: 0.0043\n",
      "198/484, train_loss: 0.0098\n",
      "199/484, train_loss: 0.0119\n",
      "200/484, train_loss: 0.0472\n",
      "201/484, train_loss: 0.0137\n",
      "202/484, train_loss: 0.0329\n",
      "203/484, train_loss: 0.0029\n",
      "204/484, train_loss: 0.0132\n",
      "205/484, train_loss: 0.0126\n",
      "206/484, train_loss: 0.4134\n",
      "207/484, train_loss: 0.0391\n",
      "208/484, train_loss: 0.3913\n",
      "209/484, train_loss: 0.0870\n",
      "210/484, train_loss: 0.0156\n",
      "211/484, train_loss: 0.0066\n",
      "212/484, train_loss: 0.6805\n",
      "213/484, train_loss: 0.0046\n",
      "214/484, train_loss: 0.0198\n",
      "215/484, train_loss: 0.2368\n",
      "216/484, train_loss: 0.0058\n",
      "217/484, train_loss: 0.0038\n",
      "218/484, train_loss: 0.6480\n",
      "219/484, train_loss: 0.0344\n",
      "220/484, train_loss: 0.0674\n",
      "221/484, train_loss: 0.0041\n",
      "222/484, train_loss: 0.1114\n",
      "223/484, train_loss: 0.1426\n",
      "224/484, train_loss: 0.0538\n",
      "225/484, train_loss: 0.0159\n",
      "226/484, train_loss: 0.1613\n",
      "227/484, train_loss: 0.0028\n",
      "228/484, train_loss: 0.0969\n",
      "229/484, train_loss: 1.1756\n",
      "230/484, train_loss: 0.7034\n",
      "231/484, train_loss: 0.0016\n",
      "232/484, train_loss: 0.1804\n",
      "233/484, train_loss: 0.1409\n",
      "234/484, train_loss: 0.0086\n",
      "235/484, train_loss: 0.6457\n",
      "236/484, train_loss: 0.0093\n",
      "237/484, train_loss: 0.0176\n",
      "238/484, train_loss: 0.1660\n",
      "239/484, train_loss: 0.2681\n",
      "240/484, train_loss: 2.8974\n",
      "241/484, train_loss: 0.0101\n",
      "242/484, train_loss: 1.9516\n",
      "243/484, train_loss: 0.6019\n",
      "244/484, train_loss: 3.6675\n",
      "245/484, train_loss: 3.4319\n",
      "246/484, train_loss: 0.6361\n",
      "247/484, train_loss: 1.1011\n",
      "248/484, train_loss: 1.5527\n",
      "249/484, train_loss: 0.0319\n",
      "250/484, train_loss: 0.0747\n",
      "251/484, train_loss: 2.6613\n",
      "252/484, train_loss: 0.2005\n",
      "253/484, train_loss: 0.0621\n",
      "254/484, train_loss: 0.3187\n",
      "255/484, train_loss: 2.8727\n",
      "256/484, train_loss: 2.1706\n",
      "257/484, train_loss: 2.9163\n",
      "258/484, train_loss: 1.4645\n",
      "259/484, train_loss: 0.1230\n",
      "260/484, train_loss: 0.3462\n",
      "261/484, train_loss: 0.0176\n",
      "262/484, train_loss: 0.7940\n",
      "263/484, train_loss: 0.3542\n",
      "264/484, train_loss: 0.5385\n",
      "265/484, train_loss: 0.0339\n",
      "266/484, train_loss: 0.4695\n",
      "267/484, train_loss: 0.5438\n",
      "268/484, train_loss: 0.5641\n",
      "269/484, train_loss: 2.6742\n",
      "270/484, train_loss: 1.1777\n",
      "271/484, train_loss: 0.4772\n",
      "272/484, train_loss: 0.9769\n",
      "273/484, train_loss: 1.2054\n",
      "274/484, train_loss: 3.1381\n",
      "275/484, train_loss: 0.1437\n",
      "276/484, train_loss: 0.0517\n",
      "277/484, train_loss: 0.0366\n",
      "278/484, train_loss: 0.9758\n",
      "279/484, train_loss: 0.1548\n",
      "280/484, train_loss: 0.6449\n",
      "281/484, train_loss: 0.0254\n",
      "282/484, train_loss: 0.0536\n",
      "283/484, train_loss: 0.1998\n",
      "284/484, train_loss: 0.5460\n",
      "285/484, train_loss: 0.2547\n",
      "286/484, train_loss: 0.1223\n",
      "287/484, train_loss: 0.0319\n",
      "288/484, train_loss: 0.1438\n",
      "289/484, train_loss: 0.3972\n",
      "290/484, train_loss: 0.7565\n",
      "291/484, train_loss: 0.0477\n",
      "292/484, train_loss: 0.3162\n",
      "293/484, train_loss: 0.0393\n",
      "294/484, train_loss: 0.4183\n",
      "295/484, train_loss: 0.0150\n",
      "296/484, train_loss: 0.9703\n",
      "297/484, train_loss: 0.4596\n",
      "298/484, train_loss: 0.1486\n",
      "299/484, train_loss: 0.0447\n",
      "300/484, train_loss: 0.3857\n",
      "301/484, train_loss: 0.1099\n",
      "302/484, train_loss: 1.0512\n",
      "303/484, train_loss: 0.1886\n",
      "304/484, train_loss: 0.0735\n",
      "305/484, train_loss: 0.4305\n",
      "306/484, train_loss: 1.2468\n",
      "307/484, train_loss: 0.3852\n",
      "308/484, train_loss: 0.0061\n",
      "309/484, train_loss: 0.0534\n",
      "310/484, train_loss: 0.0229\n",
      "311/484, train_loss: 0.0313\n",
      "312/484, train_loss: 0.0479\n",
      "313/484, train_loss: 0.0068\n",
      "314/484, train_loss: 0.0493\n",
      "315/484, train_loss: 0.1470\n",
      "316/484, train_loss: 0.0141\n",
      "317/484, train_loss: 0.0371\n",
      "318/484, train_loss: 0.0131\n",
      "319/484, train_loss: 0.0548\n",
      "320/484, train_loss: 0.0129\n",
      "321/484, train_loss: 0.0092\n",
      "322/484, train_loss: 0.0773\n",
      "323/484, train_loss: 0.0315\n",
      "324/484, train_loss: 0.0256\n",
      "325/484, train_loss: 1.0254\n",
      "326/484, train_loss: 0.0074\n",
      "327/484, train_loss: 0.0090\n",
      "328/484, train_loss: 0.0179\n",
      "329/484, train_loss: 0.0262\n",
      "330/484, train_loss: 0.2599\n",
      "331/484, train_loss: 0.3107\n",
      "332/484, train_loss: 0.1184\n",
      "333/484, train_loss: 0.0544\n",
      "334/484, train_loss: 0.0087\n",
      "335/484, train_loss: 0.0119\n",
      "336/484, train_loss: 0.3598\n",
      "337/484, train_loss: 0.1363\n",
      "338/484, train_loss: 0.3418\n",
      "339/484, train_loss: 0.0339\n",
      "340/484, train_loss: 0.4542\n",
      "341/484, train_loss: 0.0100\n",
      "342/484, train_loss: 0.0970\n",
      "343/484, train_loss: 0.0081\n",
      "344/484, train_loss: 0.7006\n",
      "345/484, train_loss: 0.1015\n",
      "346/484, train_loss: 0.3673\n",
      "347/484, train_loss: 0.0492\n",
      "348/484, train_loss: 0.0073\n",
      "349/484, train_loss: 0.1588\n",
      "350/484, train_loss: 0.0065\n",
      "351/484, train_loss: 1.2857\n",
      "352/484, train_loss: 0.0065\n",
      "353/484, train_loss: 0.0143\n",
      "354/484, train_loss: 0.0067\n",
      "355/484, train_loss: 0.0044\n",
      "356/484, train_loss: 0.0051\n",
      "357/484, train_loss: 0.0133\n",
      "358/484, train_loss: 0.0070\n",
      "359/484, train_loss: 0.0230\n",
      "360/484, train_loss: 0.0131\n",
      "361/484, train_loss: 0.8270\n",
      "362/484, train_loss: 0.0930\n",
      "363/484, train_loss: 0.0615\n",
      "364/484, train_loss: 0.0203\n",
      "365/484, train_loss: 0.0099\n",
      "366/484, train_loss: 0.0349\n",
      "367/484, train_loss: 0.0123\n",
      "368/484, train_loss: 0.0084\n",
      "369/484, train_loss: 1.0501\n",
      "370/484, train_loss: 0.3374\n",
      "371/484, train_loss: 0.0079\n",
      "372/484, train_loss: 0.0440\n",
      "373/484, train_loss: 0.0071\n",
      "374/484, train_loss: 0.0201\n",
      "375/484, train_loss: 0.2555\n",
      "376/484, train_loss: 0.0130\n",
      "377/484, train_loss: 0.0206\n",
      "378/484, train_loss: 0.0090\n",
      "379/484, train_loss: 0.2198\n",
      "380/484, train_loss: 0.0059\n",
      "381/484, train_loss: 0.0071\n",
      "382/484, train_loss: 0.0043\n",
      "383/484, train_loss: 0.0479\n",
      "384/484, train_loss: 0.0452\n",
      "385/484, train_loss: 0.0133\n",
      "386/484, train_loss: 0.2402\n",
      "387/484, train_loss: 0.0064\n",
      "388/484, train_loss: 0.1660\n",
      "389/484, train_loss: 0.0052\n",
      "390/484, train_loss: 0.0384\n",
      "391/484, train_loss: 0.0141\n",
      "392/484, train_loss: 0.0093\n",
      "393/484, train_loss: 0.3369\n",
      "394/484, train_loss: 0.0297\n",
      "395/484, train_loss: 0.0065\n",
      "396/484, train_loss: 0.0451\n",
      "397/484, train_loss: 0.1318\n",
      "398/484, train_loss: 0.0076\n",
      "399/484, train_loss: 0.0075\n",
      "400/484, train_loss: 0.8709\n",
      "401/484, train_loss: 0.0438\n",
      "402/484, train_loss: 0.1455\n",
      "403/484, train_loss: 0.0147\n",
      "404/484, train_loss: 0.0637\n",
      "405/484, train_loss: 0.0182\n",
      "406/484, train_loss: 0.2446\n",
      "407/484, train_loss: 0.0093\n",
      "408/484, train_loss: 0.0763\n",
      "409/484, train_loss: 0.0910\n",
      "410/484, train_loss: 0.0039\n",
      "411/484, train_loss: 0.1000\n",
      "412/484, train_loss: 0.0107\n",
      "413/484, train_loss: 0.0065\n",
      "414/484, train_loss: 0.0048\n",
      "415/484, train_loss: 0.0034\n",
      "416/484, train_loss: 0.0049\n",
      "417/484, train_loss: 0.0640\n",
      "418/484, train_loss: 0.0049\n",
      "419/484, train_loss: 0.0123\n",
      "420/484, train_loss: 0.0069\n",
      "421/484, train_loss: 0.0115\n",
      "422/484, train_loss: 0.1330\n",
      "423/484, train_loss: 0.6192\n",
      "424/484, train_loss: 0.0044\n",
      "425/484, train_loss: 1.2019\n",
      "426/484, train_loss: 0.0050\n",
      "427/484, train_loss: 0.0030\n",
      "428/484, train_loss: 0.0811\n",
      "429/484, train_loss: 0.0066\n",
      "430/484, train_loss: 0.2302\n",
      "431/484, train_loss: 0.0085\n",
      "432/484, train_loss: 0.0049\n",
      "433/484, train_loss: 0.0086\n",
      "434/484, train_loss: 0.0052\n",
      "435/484, train_loss: 0.0214\n",
      "436/484, train_loss: 0.4104\n",
      "437/484, train_loss: 0.0197\n",
      "438/484, train_loss: 0.0130\n",
      "439/484, train_loss: 0.0116\n",
      "440/484, train_loss: 0.0333\n",
      "441/484, train_loss: 0.2096\n",
      "442/484, train_loss: 0.0115\n",
      "443/484, train_loss: 0.6598\n",
      "444/484, train_loss: 0.0031\n",
      "445/484, train_loss: 0.0044\n",
      "446/484, train_loss: 0.0046\n",
      "447/484, train_loss: 0.0124\n",
      "448/484, train_loss: 0.0138\n",
      "449/484, train_loss: 0.0099\n",
      "450/484, train_loss: 0.0053\n",
      "451/484, train_loss: 0.0108\n",
      "452/484, train_loss: 0.0351\n",
      "453/484, train_loss: 0.0143\n",
      "454/484, train_loss: 0.0041\n",
      "455/484, train_loss: 0.0076\n",
      "456/484, train_loss: 0.3898\n",
      "457/484, train_loss: 1.1884\n",
      "458/484, train_loss: 0.3065\n",
      "459/484, train_loss: 0.0193\n",
      "460/484, train_loss: 0.0068\n",
      "461/484, train_loss: 0.0056\n",
      "462/484, train_loss: 0.0238\n",
      "463/484, train_loss: 0.6139\n",
      "464/484, train_loss: 0.0048\n",
      "465/484, train_loss: 0.6099\n",
      "466/484, train_loss: 0.0030\n",
      "467/484, train_loss: 0.0496\n",
      "468/484, train_loss: 0.0393\n",
      "469/484, train_loss: 0.0038\n",
      "470/484, train_loss: 0.0076\n",
      "471/484, train_loss: 0.0085\n",
      "472/484, train_loss: 0.0101\n",
      "473/484, train_loss: 0.0053\n",
      "474/484, train_loss: 0.0079\n",
      "475/484, train_loss: 0.0185\n",
      "476/484, train_loss: 0.0248\n",
      "477/484, train_loss: 0.0076\n",
      "478/484, train_loss: 0.2831\n",
      "479/484, train_loss: 0.0086\n",
      "480/484, train_loss: 0.0033\n",
      "481/484, train_loss: 0.2280\n",
      "482/484, train_loss: 0.0150\n",
      "483/484, train_loss: 0.0139\n",
      "484/484, train_loss: 0.0095\n",
      "epoch 10 average loss: 0.2346\n",
      "Current epoch: 10 current accuracy: 0.7988 \n",
      "Best accuracy: 0.8297 at epoch 6\n",
      "----------\n",
      "epoch 11/20\n",
      "1/484, train_loss: 0.0082\n",
      "2/484, train_loss: 0.1784\n",
      "3/484, train_loss: 0.0035\n",
      "4/484, train_loss: 0.1866\n",
      "5/484, train_loss: 0.0099\n",
      "6/484, train_loss: 0.0104\n",
      "7/484, train_loss: 0.0064\n",
      "8/484, train_loss: 0.0168\n",
      "9/484, train_loss: 0.0050\n",
      "10/484, train_loss: 0.0271\n",
      "11/484, train_loss: 0.0349\n",
      "12/484, train_loss: 0.1091\n",
      "13/484, train_loss: 0.0099\n",
      "14/484, train_loss: 0.0103\n",
      "15/484, train_loss: 0.0038\n",
      "16/484, train_loss: 1.2117\n",
      "17/484, train_loss: 0.1436\n",
      "18/484, train_loss: 0.0058\n",
      "19/484, train_loss: 1.5975\n",
      "20/484, train_loss: 0.1636\n",
      "21/484, train_loss: 0.6139\n",
      "22/484, train_loss: 0.0266\n",
      "23/484, train_loss: 0.0291\n",
      "24/484, train_loss: 0.0122\n",
      "25/484, train_loss: 0.0200\n",
      "26/484, train_loss: 0.0037\n",
      "27/484, train_loss: 0.9195\n",
      "28/484, train_loss: 0.0102\n",
      "29/484, train_loss: 1.3112\n",
      "30/484, train_loss: 0.0084\n",
      "31/484, train_loss: 0.0163\n",
      "32/484, train_loss: 0.0047\n",
      "33/484, train_loss: 0.3317\n",
      "34/484, train_loss: 0.0046\n",
      "35/484, train_loss: 0.0445\n",
      "36/484, train_loss: 0.0200\n",
      "37/484, train_loss: 0.4005\n",
      "38/484, train_loss: 0.0119\n",
      "39/484, train_loss: 0.0089\n",
      "40/484, train_loss: 0.0137\n",
      "41/484, train_loss: 0.7031\n",
      "42/484, train_loss: 0.0051\n",
      "43/484, train_loss: 0.0106\n",
      "44/484, train_loss: 0.3179\n",
      "45/484, train_loss: 0.0424\n",
      "46/484, train_loss: 0.0079\n",
      "47/484, train_loss: 0.0046\n",
      "48/484, train_loss: 0.5106\n",
      "49/484, train_loss: 0.4508\n",
      "50/484, train_loss: 0.0482\n",
      "51/484, train_loss: 0.0032\n",
      "52/484, train_loss: 0.0060\n",
      "53/484, train_loss: 0.2440\n",
      "54/484, train_loss: 0.3789\n",
      "55/484, train_loss: 0.0043\n",
      "56/484, train_loss: 0.0659\n",
      "57/484, train_loss: 0.4592\n",
      "58/484, train_loss: 0.0142\n",
      "59/484, train_loss: 0.0063\n",
      "60/484, train_loss: 0.0061\n",
      "61/484, train_loss: 0.0128\n",
      "62/484, train_loss: 0.0144\n",
      "63/484, train_loss: 0.3344\n",
      "64/484, train_loss: 0.0069\n",
      "65/484, train_loss: 0.0069\n",
      "66/484, train_loss: 0.0031\n",
      "67/484, train_loss: 0.4793\n",
      "68/484, train_loss: 0.0048\n",
      "69/484, train_loss: 0.0025\n",
      "70/484, train_loss: 0.0036\n",
      "71/484, train_loss: 0.4352\n",
      "72/484, train_loss: 0.2175\n",
      "73/484, train_loss: 1.9321\n",
      "74/484, train_loss: 0.0094\n",
      "75/484, train_loss: 0.0069\n",
      "76/484, train_loss: 0.0223\n",
      "77/484, train_loss: 0.0275\n",
      "78/484, train_loss: 0.0082\n",
      "79/484, train_loss: 0.0029\n",
      "80/484, train_loss: 0.0050\n",
      "81/484, train_loss: 0.9193\n",
      "82/484, train_loss: 0.6494\n",
      "83/484, train_loss: 0.0094\n",
      "84/484, train_loss: 0.0094\n",
      "85/484, train_loss: 0.0220\n",
      "86/484, train_loss: 0.0127\n",
      "87/484, train_loss: 0.0104\n",
      "88/484, train_loss: 0.0068\n",
      "89/484, train_loss: 0.1649\n",
      "90/484, train_loss: 0.4500\n",
      "91/484, train_loss: 0.1527\n",
      "92/484, train_loss: 0.0084\n",
      "93/484, train_loss: 0.0097\n",
      "94/484, train_loss: 0.4321\n",
      "95/484, train_loss: 0.0088\n",
      "96/484, train_loss: 0.0026\n",
      "97/484, train_loss: 0.6561\n",
      "98/484, train_loss: 0.1961\n",
      "99/484, train_loss: 0.0086\n",
      "100/484, train_loss: 0.0071\n",
      "101/484, train_loss: 0.0221\n",
      "102/484, train_loss: 0.0123\n",
      "103/484, train_loss: 0.0263\n",
      "104/484, train_loss: 0.0049\n",
      "105/484, train_loss: 1.0471\n",
      "106/484, train_loss: 0.2597\n",
      "107/484, train_loss: 0.0059\n",
      "108/484, train_loss: 1.3837\n",
      "109/484, train_loss: 1.6621\n",
      "110/484, train_loss: 0.0056\n",
      "111/484, train_loss: 0.8979\n",
      "112/484, train_loss: 1.4767\n",
      "113/484, train_loss: 0.0502\n",
      "114/484, train_loss: 0.5383\n",
      "115/484, train_loss: 0.0218\n",
      "116/484, train_loss: 0.0396\n",
      "117/484, train_loss: 0.0111\n",
      "118/484, train_loss: 0.5471\n",
      "119/484, train_loss: 0.0054\n",
      "120/484, train_loss: 0.0063\n",
      "121/484, train_loss: 0.0069\n",
      "122/484, train_loss: 1.4429\n",
      "123/484, train_loss: 0.1980\n",
      "124/484, train_loss: 0.0248\n",
      "125/484, train_loss: 0.0128\n",
      "126/484, train_loss: 0.0059\n",
      "127/484, train_loss: 0.0100\n",
      "128/484, train_loss: 0.0132\n",
      "129/484, train_loss: 0.8928\n",
      "130/484, train_loss: 0.0067\n",
      "131/484, train_loss: 0.0049\n",
      "132/484, train_loss: 0.6633\n",
      "133/484, train_loss: 0.0080\n",
      "134/484, train_loss: 0.0054\n",
      "135/484, train_loss: 0.0148\n",
      "136/484, train_loss: 0.0105\n",
      "137/484, train_loss: 1.0782\n",
      "138/484, train_loss: 0.1204\n",
      "139/484, train_loss: 0.0276\n",
      "140/484, train_loss: 0.0096\n",
      "141/484, train_loss: 0.7913\n",
      "142/484, train_loss: 0.4314\n",
      "143/484, train_loss: 0.0040\n",
      "144/484, train_loss: 0.0096\n",
      "145/484, train_loss: 0.0367\n",
      "146/484, train_loss: 0.3336\n",
      "147/484, train_loss: 0.0101\n",
      "148/484, train_loss: 0.0095\n",
      "149/484, train_loss: 0.0083\n",
      "150/484, train_loss: 0.0215\n",
      "151/484, train_loss: 0.0065\n",
      "152/484, train_loss: 0.5718\n",
      "153/484, train_loss: 0.0174\n",
      "154/484, train_loss: 0.0144\n",
      "155/484, train_loss: 0.0255\n",
      "156/484, train_loss: 0.0212\n",
      "157/484, train_loss: 0.0377\n",
      "158/484, train_loss: 0.0119\n",
      "159/484, train_loss: 0.0142\n",
      "160/484, train_loss: 0.0427\n",
      "161/484, train_loss: 0.0071\n",
      "162/484, train_loss: 0.0194\n",
      "163/484, train_loss: 0.0038\n",
      "164/484, train_loss: 0.0460\n",
      "165/484, train_loss: 0.0124\n",
      "166/484, train_loss: 0.0038\n",
      "167/484, train_loss: 0.0072\n",
      "168/484, train_loss: 0.0115\n",
      "169/484, train_loss: 0.0071\n",
      "170/484, train_loss: 0.0031\n",
      "171/484, train_loss: 0.0127\n",
      "172/484, train_loss: 0.0032\n",
      "173/484, train_loss: 0.0074\n",
      "174/484, train_loss: 0.0135\n",
      "175/484, train_loss: 0.0194\n",
      "176/484, train_loss: 0.0238\n",
      "177/484, train_loss: 0.5184\n",
      "178/484, train_loss: 0.0029\n",
      "179/484, train_loss: 0.0095\n",
      "180/484, train_loss: 0.0327\n",
      "181/484, train_loss: 0.0039\n",
      "182/484, train_loss: 0.0129\n",
      "183/484, train_loss: 0.0062\n",
      "184/484, train_loss: 0.2226\n",
      "185/484, train_loss: 0.0051\n",
      "186/484, train_loss: 0.0085\n",
      "187/484, train_loss: 0.0048\n",
      "188/484, train_loss: 0.0278\n",
      "189/484, train_loss: 0.0051\n",
      "190/484, train_loss: 0.0187\n",
      "191/484, train_loss: 0.0182\n",
      "192/484, train_loss: 0.0058\n",
      "193/484, train_loss: 0.5846\n",
      "194/484, train_loss: 0.0081\n",
      "195/484, train_loss: 0.0047\n",
      "196/484, train_loss: 0.0213\n",
      "197/484, train_loss: 0.0031\n",
      "198/484, train_loss: 0.0036\n",
      "199/484, train_loss: 0.0196\n",
      "200/484, train_loss: 0.0218\n",
      "201/484, train_loss: 0.0025\n",
      "202/484, train_loss: 0.0053\n",
      "203/484, train_loss: 0.0048\n",
      "204/484, train_loss: 0.0037\n",
      "205/484, train_loss: 0.5482\n",
      "206/484, train_loss: 0.0058\n",
      "207/484, train_loss: 0.0056\n",
      "208/484, train_loss: 0.0057\n",
      "209/484, train_loss: 0.0054\n",
      "210/484, train_loss: 0.4287\n",
      "211/484, train_loss: 0.0142\n",
      "212/484, train_loss: 0.0095\n",
      "213/484, train_loss: 0.3415\n",
      "214/484, train_loss: 0.0113\n",
      "215/484, train_loss: 0.3511\n",
      "216/484, train_loss: 0.0123\n",
      "217/484, train_loss: 0.5879\n",
      "218/484, train_loss: 0.0034\n",
      "219/484, train_loss: 0.0030\n",
      "220/484, train_loss: 0.0205\n",
      "221/484, train_loss: 0.0111\n",
      "222/484, train_loss: 0.3791\n",
      "223/484, train_loss: 0.3822\n",
      "224/484, train_loss: 0.0152\n",
      "225/484, train_loss: 0.0094\n",
      "226/484, train_loss: 0.0045\n",
      "227/484, train_loss: 0.0035\n",
      "228/484, train_loss: 0.0029\n",
      "229/484, train_loss: 0.0082\n",
      "230/484, train_loss: 0.3687\n",
      "231/484, train_loss: 0.0332\n",
      "232/484, train_loss: 0.0028\n",
      "233/484, train_loss: 0.0132\n",
      "234/484, train_loss: 0.0141\n",
      "235/484, train_loss: 0.0088\n",
      "236/484, train_loss: 0.0019\n",
      "237/484, train_loss: 0.0148\n",
      "238/484, train_loss: 0.0099\n",
      "239/484, train_loss: 0.0060\n",
      "240/484, train_loss: 0.0071\n",
      "241/484, train_loss: 0.0052\n",
      "242/484, train_loss: 0.0083\n",
      "243/484, train_loss: 0.0044\n",
      "244/484, train_loss: 0.0054\n",
      "245/484, train_loss: 0.0076\n",
      "246/484, train_loss: 0.0054\n",
      "247/484, train_loss: 0.0058\n",
      "248/484, train_loss: 0.0136\n",
      "249/484, train_loss: 0.0068\n",
      "250/484, train_loss: 0.0047\n",
      "251/484, train_loss: 0.0094\n",
      "252/484, train_loss: 0.0041\n",
      "253/484, train_loss: 0.0093\n",
      "254/484, train_loss: 0.0030\n",
      "255/484, train_loss: 0.0049\n",
      "256/484, train_loss: 0.0087\n",
      "257/484, train_loss: 0.0114\n",
      "258/484, train_loss: 0.5669\n",
      "259/484, train_loss: 0.0115\n",
      "260/484, train_loss: 0.0045\n",
      "261/484, train_loss: 0.0093\n",
      "262/484, train_loss: 0.0041\n",
      "263/484, train_loss: 0.0184\n",
      "264/484, train_loss: 0.5981\n",
      "265/484, train_loss: 0.0066\n",
      "266/484, train_loss: 0.5724\n",
      "267/484, train_loss: 0.0090\n",
      "268/484, train_loss: 0.0067\n",
      "269/484, train_loss: 0.0020\n",
      "270/484, train_loss: 0.0162\n",
      "271/484, train_loss: 0.3280\n",
      "272/484, train_loss: 0.2873\n",
      "273/484, train_loss: 0.0024\n",
      "274/484, train_loss: 0.1497\n",
      "275/484, train_loss: 0.0041\n",
      "276/484, train_loss: 0.0055\n",
      "277/484, train_loss: 0.0046\n",
      "278/484, train_loss: 0.0165\n",
      "279/484, train_loss: 0.0107\n",
      "280/484, train_loss: 0.0091\n",
      "281/484, train_loss: 0.0619\n",
      "282/484, train_loss: 0.0038\n",
      "283/484, train_loss: 0.0037\n",
      "284/484, train_loss: 0.0072\n",
      "285/484, train_loss: 0.3817\n",
      "286/484, train_loss: 0.0188\n",
      "287/484, train_loss: 0.0056\n",
      "288/484, train_loss: 0.0596\n",
      "289/484, train_loss: 0.0044\n",
      "290/484, train_loss: 0.0022\n",
      "291/484, train_loss: 1.5027\n",
      "292/484, train_loss: 0.2319\n",
      "293/484, train_loss: 0.0231\n",
      "294/484, train_loss: 0.0042\n",
      "295/484, train_loss: 2.4422\n",
      "296/484, train_loss: 0.0120\n",
      "297/484, train_loss: 0.3978\n",
      "298/484, train_loss: 0.0047\n",
      "299/484, train_loss: 0.5865\n",
      "300/484, train_loss: 0.7284\n",
      "301/484, train_loss: 0.0322\n",
      "302/484, train_loss: 0.3127\n",
      "303/484, train_loss: 0.0155\n",
      "304/484, train_loss: 0.8945\n",
      "305/484, train_loss: 0.0079\n",
      "306/484, train_loss: 0.7222\n",
      "307/484, train_loss: 0.0279\n",
      "308/484, train_loss: 0.0169\n",
      "309/484, train_loss: 0.1110\n",
      "310/484, train_loss: 1.9779\n",
      "311/484, train_loss: 0.1864\n",
      "312/484, train_loss: 0.0580\n",
      "313/484, train_loss: 0.3256\n",
      "314/484, train_loss: 1.6950\n",
      "315/484, train_loss: 2.6143\n",
      "316/484, train_loss: 0.0247\n",
      "317/484, train_loss: 0.4931\n",
      "318/484, train_loss: 1.1077\n",
      "319/484, train_loss: 0.1536\n",
      "320/484, train_loss: 0.2691\n",
      "321/484, train_loss: 0.2375\n",
      "322/484, train_loss: 0.0554\n",
      "323/484, train_loss: 0.1527\n",
      "324/484, train_loss: 0.0620\n",
      "325/484, train_loss: 0.2609\n",
      "326/484, train_loss: 0.0324\n",
      "327/484, train_loss: 0.1371\n",
      "328/484, train_loss: 0.0314\n",
      "329/484, train_loss: 0.0414\n",
      "330/484, train_loss: 0.7800\n",
      "331/484, train_loss: 1.2164\n",
      "332/484, train_loss: 0.2470\n",
      "333/484, train_loss: 0.1739\n",
      "334/484, train_loss: 0.0681\n",
      "335/484, train_loss: 0.7273\n",
      "336/484, train_loss: 0.1905\n",
      "337/484, train_loss: 0.0223\n",
      "338/484, train_loss: 0.2528\n",
      "339/484, train_loss: 0.0163\n",
      "340/484, train_loss: 0.0775\n",
      "341/484, train_loss: 0.3344\n",
      "342/484, train_loss: 0.3229\n",
      "343/484, train_loss: 0.0290\n",
      "344/484, train_loss: 0.0251\n",
      "345/484, train_loss: 1.6283\n",
      "346/484, train_loss: 0.0198\n",
      "347/484, train_loss: 0.0492\n",
      "348/484, train_loss: 0.5726\n",
      "349/484, train_loss: 0.1226\n",
      "350/484, train_loss: 0.2063\n",
      "351/484, train_loss: 1.7599\n",
      "352/484, train_loss: 0.7799\n",
      "353/484, train_loss: 0.0109\n",
      "354/484, train_loss: 0.0680\n",
      "355/484, train_loss: 0.7071\n",
      "356/484, train_loss: 0.0122\n",
      "357/484, train_loss: 0.0697\n",
      "358/484, train_loss: 0.0126\n",
      "359/484, train_loss: 0.0433\n",
      "360/484, train_loss: 0.0852\n",
      "361/484, train_loss: 0.0281\n",
      "362/484, train_loss: 0.7835\n",
      "363/484, train_loss: 0.7595\n",
      "364/484, train_loss: 0.4745\n",
      "365/484, train_loss: 1.0557\n",
      "366/484, train_loss: 0.0085\n",
      "367/484, train_loss: 0.4347\n",
      "368/484, train_loss: 0.0243\n",
      "369/484, train_loss: 0.0212\n",
      "370/484, train_loss: 0.0400\n",
      "371/484, train_loss: 0.0205\n",
      "372/484, train_loss: 0.5575\n",
      "373/484, train_loss: 0.0349\n",
      "374/484, train_loss: 0.0240\n",
      "375/484, train_loss: 0.0055\n",
      "376/484, train_loss: 0.0159\n",
      "377/484, train_loss: 0.0606\n",
      "378/484, train_loss: 0.0375\n",
      "379/484, train_loss: 0.6823\n",
      "380/484, train_loss: 0.0358\n",
      "381/484, train_loss: 0.5298\n",
      "382/484, train_loss: 0.3359\n",
      "383/484, train_loss: 0.0529\n",
      "384/484, train_loss: 0.0410\n",
      "385/484, train_loss: 0.0189\n",
      "386/484, train_loss: 1.1259\n",
      "387/484, train_loss: 0.0142\n",
      "388/484, train_loss: 0.0063\n",
      "389/484, train_loss: 0.4186\n",
      "390/484, train_loss: 0.1226\n",
      "391/484, train_loss: 0.0105\n",
      "392/484, train_loss: 0.5516\n",
      "393/484, train_loss: 0.0064\n",
      "394/484, train_loss: 0.0343\n",
      "395/484, train_loss: 0.0854\n",
      "396/484, train_loss: 1.5033\n",
      "397/484, train_loss: 0.0604\n",
      "398/484, train_loss: 0.0898\n",
      "399/484, train_loss: 0.0146\n",
      "400/484, train_loss: 0.0499\n",
      "401/484, train_loss: 1.2577\n",
      "402/484, train_loss: 0.0226\n",
      "403/484, train_loss: 0.0242\n",
      "404/484, train_loss: 0.1859\n",
      "405/484, train_loss: 0.0280\n",
      "406/484, train_loss: 0.0519\n",
      "407/484, train_loss: 0.1342\n",
      "408/484, train_loss: 0.0335\n",
      "409/484, train_loss: 0.0105\n",
      "410/484, train_loss: 0.0873\n",
      "411/484, train_loss: 0.2789\n",
      "412/484, train_loss: 0.0287\n",
      "413/484, train_loss: 0.2119\n",
      "414/484, train_loss: 0.0083\n",
      "415/484, train_loss: 0.3512\n",
      "416/484, train_loss: 0.0073\n",
      "417/484, train_loss: 0.3998\n",
      "418/484, train_loss: 0.1249\n",
      "419/484, train_loss: 0.0422\n",
      "420/484, train_loss: 0.0059\n",
      "421/484, train_loss: 0.0104\n",
      "422/484, train_loss: 0.3316\n",
      "423/484, train_loss: 0.0313\n",
      "424/484, train_loss: 0.0052\n",
      "425/484, train_loss: 0.0200\n",
      "426/484, train_loss: 0.0650\n",
      "427/484, train_loss: 0.0097\n",
      "428/484, train_loss: 0.0048\n",
      "429/484, train_loss: 0.0114\n",
      "430/484, train_loss: 0.0054\n",
      "431/484, train_loss: 0.0323\n",
      "432/484, train_loss: 0.0118\n",
      "433/484, train_loss: 0.0333\n",
      "434/484, train_loss: 0.1023\n",
      "435/484, train_loss: 0.0069\n",
      "436/484, train_loss: 0.5302\n",
      "437/484, train_loss: 0.0120\n",
      "438/484, train_loss: 0.6401\n",
      "439/484, train_loss: 0.0246\n",
      "440/484, train_loss: 0.0076\n",
      "441/484, train_loss: 0.3492\n",
      "442/484, train_loss: 0.0114\n",
      "443/484, train_loss: 0.0062\n",
      "444/484, train_loss: 0.0057\n",
      "445/484, train_loss: 0.0064\n",
      "446/484, train_loss: 0.0122\n",
      "447/484, train_loss: 0.0120\n",
      "448/484, train_loss: 0.0101\n",
      "449/484, train_loss: 0.3446\n",
      "450/484, train_loss: 0.9757\n",
      "451/484, train_loss: 0.1236\n",
      "452/484, train_loss: 0.0087\n",
      "453/484, train_loss: 0.0092\n",
      "454/484, train_loss: 0.0161\n",
      "455/484, train_loss: 0.0238\n",
      "456/484, train_loss: 0.0028\n",
      "457/484, train_loss: 0.0309\n",
      "458/484, train_loss: 0.0035\n",
      "459/484, train_loss: 0.1547\n",
      "460/484, train_loss: 0.0148\n",
      "461/484, train_loss: 0.0268\n",
      "462/484, train_loss: 0.1752\n",
      "463/484, train_loss: 0.0066\n",
      "464/484, train_loss: 0.0044\n",
      "465/484, train_loss: 0.0046\n",
      "466/484, train_loss: 0.0097\n",
      "467/484, train_loss: 0.0029\n",
      "468/484, train_loss: 0.1453\n",
      "469/484, train_loss: 0.0086\n",
      "470/484, train_loss: 0.0052\n",
      "471/484, train_loss: 0.2617\n",
      "472/484, train_loss: 0.0104\n",
      "473/484, train_loss: 0.0089\n",
      "474/484, train_loss: 0.3676\n",
      "475/484, train_loss: 0.0137\n",
      "476/484, train_loss: 0.0034\n",
      "477/484, train_loss: 0.1788\n",
      "478/484, train_loss: 0.0073\n",
      "479/484, train_loss: 1.4182\n",
      "480/484, train_loss: 0.0382\n",
      "481/484, train_loss: 0.0171\n",
      "482/484, train_loss: 0.0035\n",
      "483/484, train_loss: 0.0206\n",
      "484/484, train_loss: 0.0692\n",
      "epoch 11 average loss: 0.1876\n",
      "----------\n",
      "epoch 12/20\n",
      "1/484, train_loss: 0.0047\n",
      "2/484, train_loss: 0.0042\n",
      "3/484, train_loss: 0.0108\n",
      "4/484, train_loss: 0.3957\n",
      "5/484, train_loss: 0.0087\n",
      "6/484, train_loss: 0.1933\n",
      "7/484, train_loss: 0.0095\n",
      "8/484, train_loss: 0.6526\n",
      "9/484, train_loss: 0.0125\n",
      "10/484, train_loss: 0.6148\n",
      "11/484, train_loss: 0.5205\n",
      "12/484, train_loss: 0.2040\n",
      "13/484, train_loss: 0.0527\n",
      "14/484, train_loss: 0.0102\n",
      "15/484, train_loss: 0.0031\n",
      "16/484, train_loss: 0.3254\n",
      "17/484, train_loss: 0.2728\n",
      "18/484, train_loss: 0.7152\n",
      "19/484, train_loss: 0.0245\n",
      "20/484, train_loss: 0.0042\n",
      "21/484, train_loss: 0.0051\n",
      "22/484, train_loss: 0.0207\n",
      "23/484, train_loss: 0.5173\n",
      "24/484, train_loss: 0.0405\n",
      "25/484, train_loss: 0.0047\n",
      "26/484, train_loss: 0.0046\n",
      "27/484, train_loss: 0.0824\n",
      "28/484, train_loss: 0.2680\n",
      "29/484, train_loss: 0.0593\n",
      "30/484, train_loss: 0.0844\n",
      "31/484, train_loss: 0.4801\n",
      "32/484, train_loss: 0.1735\n",
      "33/484, train_loss: 0.0119\n",
      "34/484, train_loss: 0.3521\n",
      "35/484, train_loss: 0.2012\n",
      "36/484, train_loss: 0.0121\n",
      "37/484, train_loss: 0.0298\n",
      "38/484, train_loss: 0.2048\n",
      "39/484, train_loss: 0.0795\n",
      "40/484, train_loss: 0.0187\n",
      "41/484, train_loss: 0.0161\n",
      "42/484, train_loss: 0.8254\n",
      "43/484, train_loss: 0.3407\n",
      "44/484, train_loss: 0.0065\n",
      "45/484, train_loss: 0.0023\n",
      "46/484, train_loss: 0.0047\n",
      "47/484, train_loss: 0.0026\n",
      "48/484, train_loss: 0.0435\n",
      "49/484, train_loss: 0.0091\n",
      "50/484, train_loss: 0.0144\n",
      "51/484, train_loss: 0.0073\n",
      "52/484, train_loss: 0.1563\n",
      "53/484, train_loss: 0.7713\n",
      "54/484, train_loss: 0.0335\n",
      "55/484, train_loss: 0.4931\n",
      "56/484, train_loss: 0.0676\n",
      "57/484, train_loss: 0.0024\n",
      "58/484, train_loss: 0.0062\n",
      "59/484, train_loss: 0.7607\n",
      "60/484, train_loss: 0.0062\n",
      "61/484, train_loss: 0.0079\n",
      "62/484, train_loss: 0.7208\n",
      "63/484, train_loss: 0.0053\n",
      "64/484, train_loss: 0.5993\n",
      "65/484, train_loss: 0.0296\n",
      "66/484, train_loss: 0.1126\n",
      "67/484, train_loss: 0.0054\n",
      "68/484, train_loss: 0.0070\n",
      "69/484, train_loss: 0.2245\n",
      "70/484, train_loss: 0.0053\n",
      "71/484, train_loss: 0.0123\n",
      "72/484, train_loss: 0.1009\n",
      "73/484, train_loss: 0.1256\n",
      "74/484, train_loss: 1.1542\n",
      "75/484, train_loss: 0.0244\n",
      "76/484, train_loss: 0.0145\n",
      "77/484, train_loss: 0.0027\n",
      "78/484, train_loss: 0.0671\n",
      "79/484, train_loss: 0.0101\n",
      "80/484, train_loss: 0.0112\n",
      "81/484, train_loss: 0.1210\n",
      "82/484, train_loss: 0.0077\n",
      "83/484, train_loss: 0.0145\n",
      "84/484, train_loss: 0.0071\n",
      "85/484, train_loss: 0.5163\n",
      "86/484, train_loss: 0.3850\n",
      "87/484, train_loss: 0.0044\n",
      "88/484, train_loss: 0.0449\n",
      "89/484, train_loss: 0.0025\n",
      "90/484, train_loss: 0.0129\n",
      "91/484, train_loss: 0.3687\n",
      "92/484, train_loss: 0.3422\n",
      "93/484, train_loss: 0.0124\n",
      "94/484, train_loss: 0.8548\n",
      "95/484, train_loss: 0.0054\n",
      "96/484, train_loss: 0.0085\n",
      "97/484, train_loss: 0.0109\n",
      "98/484, train_loss: 0.0098\n",
      "99/484, train_loss: 0.0063\n",
      "100/484, train_loss: 0.0125\n",
      "101/484, train_loss: 0.0053\n",
      "102/484, train_loss: 0.0052\n",
      "103/484, train_loss: 0.0056\n",
      "104/484, train_loss: 0.0044\n",
      "105/484, train_loss: 0.0107\n",
      "106/484, train_loss: 0.0064\n",
      "107/484, train_loss: 0.2223\n",
      "108/484, train_loss: 0.0994\n",
      "109/484, train_loss: 0.0036\n",
      "110/484, train_loss: 0.0027\n",
      "111/484, train_loss: 0.0218\n",
      "112/484, train_loss: 0.0156\n",
      "113/484, train_loss: 0.0066\n",
      "114/484, train_loss: 0.2397\n",
      "115/484, train_loss: 0.0053\n",
      "116/484, train_loss: 0.1015\n",
      "117/484, train_loss: 0.0048\n",
      "118/484, train_loss: 0.6244\n",
      "119/484, train_loss: 0.0080\n",
      "120/484, train_loss: 0.0022\n",
      "121/484, train_loss: 0.0048\n",
      "122/484, train_loss: 0.7328\n",
      "123/484, train_loss: 0.0197\n",
      "124/484, train_loss: 1.3005\n",
      "125/484, train_loss: 0.0222\n",
      "126/484, train_loss: 2.1086\n",
      "127/484, train_loss: 0.0077\n",
      "128/484, train_loss: 0.7076\n",
      "129/484, train_loss: 0.0035\n",
      "130/484, train_loss: 0.4411\n",
      "131/484, train_loss: 0.0029\n",
      "132/484, train_loss: 0.0057\n",
      "133/484, train_loss: 0.4844\n",
      "134/484, train_loss: 0.0073\n",
      "135/484, train_loss: 0.0061\n",
      "136/484, train_loss: 0.0101\n",
      "137/484, train_loss: 0.0031\n",
      "138/484, train_loss: 0.0232\n",
      "139/484, train_loss: 0.0018\n",
      "140/484, train_loss: 0.0050\n",
      "141/484, train_loss: 0.0095\n",
      "142/484, train_loss: 0.0073\n",
      "143/484, train_loss: 0.0046\n",
      "144/484, train_loss: 0.0088\n",
      "145/484, train_loss: 0.0072\n",
      "146/484, train_loss: 0.0171\n",
      "147/484, train_loss: 0.1119\n",
      "148/484, train_loss: 0.0068\n",
      "149/484, train_loss: 0.0151\n",
      "150/484, train_loss: 0.0454\n",
      "151/484, train_loss: 0.0117\n",
      "152/484, train_loss: 0.0050\n",
      "153/484, train_loss: 0.0056\n",
      "154/484, train_loss: 0.0028\n",
      "155/484, train_loss: 0.0212\n",
      "156/484, train_loss: 0.0074\n",
      "157/484, train_loss: 0.0034\n",
      "158/484, train_loss: 0.0210\n",
      "159/484, train_loss: 0.7452\n",
      "160/484, train_loss: 0.4007\n",
      "161/484, train_loss: 0.0031\n",
      "162/484, train_loss: 0.0189\n",
      "163/484, train_loss: 0.0059\n",
      "164/484, train_loss: 0.3808\n",
      "165/484, train_loss: 0.0022\n",
      "166/484, train_loss: 0.0056\n",
      "167/484, train_loss: 0.0035\n",
      "168/484, train_loss: 0.4677\n",
      "169/484, train_loss: 0.0122\n",
      "170/484, train_loss: 0.0347\n",
      "171/484, train_loss: 0.0135\n",
      "172/484, train_loss: 0.0947\n",
      "173/484, train_loss: 0.0034\n",
      "174/484, train_loss: 0.0081\n",
      "175/484, train_loss: 0.6155\n",
      "176/484, train_loss: 0.5757\n",
      "177/484, train_loss: 0.3129\n",
      "178/484, train_loss: 0.3089\n",
      "179/484, train_loss: 0.7912\n",
      "180/484, train_loss: 0.0224\n",
      "181/484, train_loss: 0.3936\n",
      "182/484, train_loss: 0.0251\n",
      "183/484, train_loss: 0.0198\n",
      "184/484, train_loss: 0.0119\n",
      "185/484, train_loss: 0.0015\n",
      "186/484, train_loss: 0.0042\n",
      "187/484, train_loss: 0.0031\n",
      "188/484, train_loss: 0.0043\n",
      "189/484, train_loss: 0.0048\n",
      "190/484, train_loss: 0.0037\n",
      "191/484, train_loss: 0.0054\n",
      "192/484, train_loss: 0.9745\n",
      "193/484, train_loss: 0.8752\n",
      "194/484, train_loss: 0.0015\n",
      "195/484, train_loss: 0.9680\n",
      "196/484, train_loss: 0.0074\n",
      "197/484, train_loss: 0.0099\n",
      "198/484, train_loss: 0.0196\n",
      "199/484, train_loss: 0.0024\n",
      "200/484, train_loss: 0.0035\n",
      "201/484, train_loss: 0.0453\n",
      "202/484, train_loss: 0.0026\n",
      "203/484, train_loss: 0.2394\n",
      "204/484, train_loss: 0.0216\n",
      "205/484, train_loss: 0.0232\n",
      "206/484, train_loss: 0.0044\n",
      "207/484, train_loss: 0.0046\n",
      "208/484, train_loss: 0.0143\n",
      "209/484, train_loss: 0.0044\n",
      "210/484, train_loss: 0.3628\n",
      "211/484, train_loss: 0.2471\n",
      "212/484, train_loss: 0.5994\n",
      "213/484, train_loss: 0.0149\n",
      "214/484, train_loss: 0.3406\n",
      "215/484, train_loss: 0.0060\n",
      "216/484, train_loss: 0.0194\n",
      "217/484, train_loss: 0.0035\n",
      "218/484, train_loss: 0.1989\n",
      "219/484, train_loss: 1.4116\n",
      "220/484, train_loss: 0.0169\n",
      "221/484, train_loss: 0.6567\n",
      "222/484, train_loss: 0.0075\n",
      "223/484, train_loss: 0.9619\n",
      "224/484, train_loss: 1.6458\n",
      "225/484, train_loss: 1.1516\n",
      "226/484, train_loss: 0.0276\n",
      "227/484, train_loss: 0.0419\n",
      "228/484, train_loss: 0.0019\n",
      "229/484, train_loss: 0.0134\n",
      "230/484, train_loss: 0.0091\n",
      "231/484, train_loss: 0.0349\n",
      "232/484, train_loss: 0.8388\n",
      "233/484, train_loss: 0.0017\n",
      "234/484, train_loss: 0.0776\n",
      "235/484, train_loss: 0.9921\n",
      "236/484, train_loss: 0.4994\n",
      "237/484, train_loss: 0.0247\n",
      "238/484, train_loss: 0.0028\n",
      "239/484, train_loss: 0.0077\n",
      "240/484, train_loss: 0.0113\n",
      "241/484, train_loss: 0.0080\n",
      "242/484, train_loss: 0.0176\n",
      "243/484, train_loss: 0.0081\n",
      "244/484, train_loss: 0.4232\n",
      "245/484, train_loss: 0.0392\n",
      "246/484, train_loss: 0.0218\n",
      "247/484, train_loss: 0.0164\n",
      "248/484, train_loss: 0.0020\n",
      "249/484, train_loss: 0.3053\n",
      "250/484, train_loss: 0.0072\n",
      "251/484, train_loss: 0.0137\n",
      "252/484, train_loss: 0.2563\n",
      "253/484, train_loss: 0.0356\n",
      "254/484, train_loss: 0.0152\n",
      "255/484, train_loss: 0.0481\n",
      "256/484, train_loss: 0.0077\n",
      "257/484, train_loss: 1.5952\n",
      "258/484, train_loss: 0.0192\n",
      "259/484, train_loss: 0.5598\n",
      "260/484, train_loss: 0.0102\n",
      "261/484, train_loss: 0.0053\n",
      "262/484, train_loss: 0.0285\n",
      "263/484, train_loss: 0.0139\n",
      "264/484, train_loss: 0.0305\n",
      "265/484, train_loss: 0.0175\n",
      "266/484, train_loss: 0.4224\n",
      "267/484, train_loss: 0.0128\n",
      "268/484, train_loss: 0.0428\n",
      "269/484, train_loss: 0.0153\n",
      "270/484, train_loss: 0.0099\n",
      "271/484, train_loss: 0.2683\n",
      "272/484, train_loss: 0.0068\n",
      "273/484, train_loss: 0.0337\n",
      "274/484, train_loss: 0.0092\n",
      "275/484, train_loss: 2.9773\n",
      "276/484, train_loss: 0.0142\n",
      "277/484, train_loss: 0.1893\n",
      "278/484, train_loss: 0.0038\n",
      "279/484, train_loss: 0.0067\n",
      "280/484, train_loss: 0.0333\n",
      "281/484, train_loss: 0.5639\n",
      "282/484, train_loss: 0.0037\n",
      "283/484, train_loss: 0.8566\n",
      "284/484, train_loss: 0.0059\n",
      "285/484, train_loss: 0.0072\n",
      "286/484, train_loss: 0.0063\n",
      "287/484, train_loss: 0.0046\n",
      "288/484, train_loss: 0.0383\n",
      "289/484, train_loss: 0.5387\n",
      "290/484, train_loss: 0.0041\n",
      "291/484, train_loss: 0.2683\n",
      "292/484, train_loss: 0.0125\n",
      "293/484, train_loss: 0.2686\n",
      "294/484, train_loss: 0.0071\n",
      "295/484, train_loss: 0.0115\n",
      "296/484, train_loss: 0.3779\n",
      "297/484, train_loss: 0.6162\n",
      "298/484, train_loss: 0.0141\n",
      "299/484, train_loss: 0.0182\n",
      "300/484, train_loss: 0.0360\n",
      "301/484, train_loss: 0.0112\n",
      "302/484, train_loss: 0.0511\n",
      "303/484, train_loss: 0.0156\n",
      "304/484, train_loss: 0.0065\n",
      "305/484, train_loss: 0.2417\n",
      "306/484, train_loss: 0.0040\n",
      "307/484, train_loss: 0.0081\n",
      "308/484, train_loss: 0.0041\n",
      "309/484, train_loss: 0.0130\n",
      "310/484, train_loss: 0.0132\n",
      "311/484, train_loss: 0.0153\n",
      "312/484, train_loss: 0.0089\n",
      "313/484, train_loss: 0.0115\n",
      "314/484, train_loss: 0.0061\n",
      "315/484, train_loss: 0.0099\n",
      "316/484, train_loss: 0.2831\n",
      "317/484, train_loss: 0.0050\n",
      "318/484, train_loss: 0.0520\n",
      "319/484, train_loss: 0.0861\n",
      "320/484, train_loss: 0.0142\n",
      "321/484, train_loss: 0.4836\n",
      "322/484, train_loss: 0.0097\n",
      "323/484, train_loss: 0.0124\n",
      "324/484, train_loss: 0.0029\n",
      "325/484, train_loss: 0.2177\n",
      "326/484, train_loss: 0.0071\n",
      "327/484, train_loss: 0.0572\n",
      "328/484, train_loss: 0.0066\n",
      "329/484, train_loss: 0.2810\n",
      "330/484, train_loss: 0.0519\n",
      "331/484, train_loss: 0.0136\n",
      "332/484, train_loss: 0.5506\n",
      "333/484, train_loss: 0.0161\n",
      "334/484, train_loss: 0.0043\n",
      "335/484, train_loss: 0.4815\n",
      "336/484, train_loss: 0.0042\n",
      "337/484, train_loss: 0.4487\n",
      "338/484, train_loss: 0.0057\n",
      "339/484, train_loss: 0.4681\n",
      "340/484, train_loss: 0.3428\n",
      "341/484, train_loss: 1.2525\n",
      "342/484, train_loss: 0.0064\n",
      "343/484, train_loss: 0.0103\n",
      "344/484, train_loss: 0.0058\n",
      "345/484, train_loss: 0.8179\n",
      "346/484, train_loss: 0.0421\n",
      "347/484, train_loss: 0.0324\n",
      "348/484, train_loss: 0.0139\n",
      "349/484, train_loss: 0.0066\n",
      "350/484, train_loss: 0.0774\n",
      "351/484, train_loss: 0.1823\n",
      "352/484, train_loss: 0.3996\n",
      "353/484, train_loss: 0.1503\n",
      "354/484, train_loss: 0.0026\n",
      "355/484, train_loss: 0.0048\n",
      "356/484, train_loss: 0.0213\n",
      "357/484, train_loss: 0.0499\n",
      "358/484, train_loss: 0.2117\n",
      "359/484, train_loss: 0.0411\n",
      "360/484, train_loss: 0.2848\n",
      "361/484, train_loss: 0.0176\n",
      "362/484, train_loss: 0.0049\n",
      "363/484, train_loss: 0.4592\n",
      "364/484, train_loss: 0.0152\n",
      "365/484, train_loss: 0.0589\n",
      "366/484, train_loss: 0.0876\n",
      "367/484, train_loss: 0.0082\n",
      "368/484, train_loss: 0.0294\n",
      "369/484, train_loss: 0.0021\n",
      "370/484, train_loss: 0.0169\n",
      "371/484, train_loss: 0.0078\n",
      "372/484, train_loss: 0.0115\n",
      "373/484, train_loss: 0.6987\n",
      "374/484, train_loss: 0.0049\n",
      "375/484, train_loss: 0.0052\n",
      "376/484, train_loss: 0.7848\n",
      "377/484, train_loss: 0.0046\n",
      "378/484, train_loss: 0.0071\n",
      "379/484, train_loss: 0.0039\n",
      "380/484, train_loss: 0.0630\n",
      "381/484, train_loss: 0.0133\n",
      "382/484, train_loss: 0.0041\n",
      "383/484, train_loss: 0.0297\n",
      "384/484, train_loss: 0.0065\n",
      "385/484, train_loss: 0.0072\n",
      "386/484, train_loss: 0.1889\n",
      "387/484, train_loss: 0.0378\n",
      "388/484, train_loss: 0.0035\n",
      "389/484, train_loss: 0.0376\n",
      "390/484, train_loss: 0.0043\n",
      "391/484, train_loss: 0.5877\n",
      "392/484, train_loss: 0.2723\n",
      "393/484, train_loss: 0.0046\n",
      "394/484, train_loss: 0.0617\n",
      "395/484, train_loss: 0.0056\n",
      "396/484, train_loss: 0.4250\n",
      "397/484, train_loss: 0.0233\n",
      "398/484, train_loss: 0.0136\n",
      "399/484, train_loss: 0.3578\n",
      "400/484, train_loss: 0.0020\n",
      "401/484, train_loss: 0.1093\n",
      "402/484, train_loss: 0.0646\n",
      "403/484, train_loss: 0.3490\n",
      "404/484, train_loss: 0.0035\n",
      "405/484, train_loss: 1.8637\n",
      "406/484, train_loss: 0.0074\n",
      "407/484, train_loss: 0.0067\n",
      "408/484, train_loss: 0.0031\n",
      "409/484, train_loss: 0.5640\n",
      "410/484, train_loss: 0.0071\n",
      "411/484, train_loss: 0.0076\n",
      "412/484, train_loss: 0.5417\n",
      "413/484, train_loss: 0.0233\n",
      "414/484, train_loss: 0.0039\n",
      "415/484, train_loss: 0.6495\n",
      "416/484, train_loss: 0.0033\n",
      "417/484, train_loss: 0.0050\n",
      "418/484, train_loss: 0.5333\n",
      "419/484, train_loss: 0.8367\n",
      "420/484, train_loss: 0.0063\n",
      "421/484, train_loss: 0.3030\n",
      "422/484, train_loss: 0.0137\n",
      "423/484, train_loss: 0.0058\n",
      "424/484, train_loss: 0.0066\n",
      "425/484, train_loss: 0.0076\n",
      "426/484, train_loss: 0.0045\n",
      "427/484, train_loss: 0.0050\n",
      "428/484, train_loss: 0.0185\n",
      "429/484, train_loss: 0.0033\n",
      "430/484, train_loss: 0.0048\n",
      "431/484, train_loss: 1.4234\n",
      "432/484, train_loss: 0.0576\n",
      "433/484, train_loss: 0.0103\n",
      "434/484, train_loss: 0.1569\n",
      "435/484, train_loss: 0.0177\n",
      "436/484, train_loss: 0.3600\n",
      "437/484, train_loss: 0.2674\n",
      "438/484, train_loss: 0.0160\n",
      "439/484, train_loss: 0.0059\n",
      "440/484, train_loss: 0.0066\n",
      "441/484, train_loss: 0.0042\n",
      "442/484, train_loss: 0.0061\n",
      "443/484, train_loss: 0.0997\n",
      "444/484, train_loss: 0.0035\n",
      "445/484, train_loss: 0.0236\n",
      "446/484, train_loss: 0.3295\n",
      "447/484, train_loss: 0.0129\n",
      "448/484, train_loss: 0.3506\n",
      "449/484, train_loss: 0.0034\n",
      "450/484, train_loss: 0.0505\n",
      "451/484, train_loss: 0.0076\n",
      "452/484, train_loss: 0.0402\n",
      "453/484, train_loss: 0.0027\n",
      "454/484, train_loss: 0.0081\n",
      "455/484, train_loss: 0.2688\n",
      "456/484, train_loss: 0.0122\n",
      "457/484, train_loss: 0.0136\n",
      "458/484, train_loss: 0.0135\n",
      "459/484, train_loss: 0.0032\n",
      "460/484, train_loss: 0.0035\n",
      "461/484, train_loss: 0.6275\n",
      "462/484, train_loss: 0.9294\n",
      "463/484, train_loss: 0.0126\n",
      "464/484, train_loss: 0.0032\n",
      "465/484, train_loss: 0.0046\n",
      "466/484, train_loss: 0.0164\n",
      "467/484, train_loss: 0.0220\n",
      "468/484, train_loss: 0.0059\n",
      "469/484, train_loss: 0.7565\n",
      "470/484, train_loss: 0.0027\n",
      "471/484, train_loss: 0.0074\n",
      "472/484, train_loss: 0.0333\n",
      "473/484, train_loss: 0.0033\n",
      "474/484, train_loss: 0.0040\n",
      "475/484, train_loss: 0.0042\n",
      "476/484, train_loss: 0.2794\n",
      "477/484, train_loss: 0.0025\n",
      "478/484, train_loss: 0.0332\n",
      "479/484, train_loss: 0.0064\n",
      "480/484, train_loss: 0.0149\n",
      "481/484, train_loss: 0.0135\n",
      "482/484, train_loss: 0.3586\n",
      "483/484, train_loss: 0.0093\n",
      "484/484, train_loss: 0.0202\n",
      "epoch 12 average loss: 0.1629\n",
      "Current epoch: 12 current accuracy: 0.8173 \n",
      "Best accuracy: 0.8297 at epoch 6\n",
      "----------\n",
      "epoch 13/20\n",
      "1/484, train_loss: 0.0076\n",
      "2/484, train_loss: 0.0063\n",
      "3/484, train_loss: 0.0069\n",
      "4/484, train_loss: 0.3384\n",
      "5/484, train_loss: 0.0029\n",
      "6/484, train_loss: 0.0110\n",
      "7/484, train_loss: 0.8268\n",
      "8/484, train_loss: 0.0084\n",
      "9/484, train_loss: 0.0049\n",
      "10/484, train_loss: 0.0344\n",
      "11/484, train_loss: 0.0120\n",
      "12/484, train_loss: 0.0097\n",
      "13/484, train_loss: 0.0108\n",
      "14/484, train_loss: 0.0054\n",
      "15/484, train_loss: 0.0102\n",
      "16/484, train_loss: 0.0122\n",
      "17/484, train_loss: 0.0234\n",
      "18/484, train_loss: 0.0059\n",
      "19/484, train_loss: 0.0080\n",
      "20/484, train_loss: 0.7311\n",
      "21/484, train_loss: 0.0058\n",
      "22/484, train_loss: 0.0085\n",
      "23/484, train_loss: 0.0118\n",
      "24/484, train_loss: 0.0218\n",
      "25/484, train_loss: 0.7387\n",
      "26/484, train_loss: 0.0217\n",
      "27/484, train_loss: 0.0136\n",
      "28/484, train_loss: 0.0072\n",
      "29/484, train_loss: 0.0056\n",
      "30/484, train_loss: 0.0296\n",
      "31/484, train_loss: 0.0036\n",
      "32/484, train_loss: 0.0291\n",
      "33/484, train_loss: 0.7123\n",
      "34/484, train_loss: 0.0037\n",
      "35/484, train_loss: 0.1494\n",
      "36/484, train_loss: 0.0932\n",
      "37/484, train_loss: 0.9383\n",
      "38/484, train_loss: 0.4800\n",
      "39/484, train_loss: 0.0054\n",
      "40/484, train_loss: 0.0080\n",
      "41/484, train_loss: 0.0029\n",
      "42/484, train_loss: 0.0207\n",
      "43/484, train_loss: 0.0063\n",
      "44/484, train_loss: 0.0104\n",
      "45/484, train_loss: 0.0122\n",
      "46/484, train_loss: 0.3625\n",
      "47/484, train_loss: 0.3307\n",
      "48/484, train_loss: 0.0116\n",
      "49/484, train_loss: 0.0075\n",
      "50/484, train_loss: 0.0063\n",
      "51/484, train_loss: 0.0119\n",
      "52/484, train_loss: 0.0047\n",
      "53/484, train_loss: 1.0913\n",
      "54/484, train_loss: 0.1637\n",
      "55/484, train_loss: 0.0115\n",
      "56/484, train_loss: 0.0193\n",
      "57/484, train_loss: 0.0053\n",
      "58/484, train_loss: 0.0030\n",
      "59/484, train_loss: 0.0679\n",
      "60/484, train_loss: 0.0071\n",
      "61/484, train_loss: 0.0045\n",
      "62/484, train_loss: 0.0065\n",
      "63/484, train_loss: 0.0103\n",
      "64/484, train_loss: 0.0076\n",
      "65/484, train_loss: 0.0034\n",
      "66/484, train_loss: 0.0042\n",
      "67/484, train_loss: 0.0088\n",
      "68/484, train_loss: 0.0042\n",
      "69/484, train_loss: 0.0458\n",
      "70/484, train_loss: 0.0108\n",
      "71/484, train_loss: 0.0049\n",
      "72/484, train_loss: 0.0038\n",
      "73/484, train_loss: 0.0037\n",
      "74/484, train_loss: 0.5302\n",
      "75/484, train_loss: 0.0060\n",
      "76/484, train_loss: 0.0045\n",
      "77/484, train_loss: 0.0060\n",
      "78/484, train_loss: 0.0029\n",
      "79/484, train_loss: 0.0170\n",
      "80/484, train_loss: 0.0070\n",
      "81/484, train_loss: 0.0055\n",
      "82/484, train_loss: 0.0215\n",
      "83/484, train_loss: 0.0188\n",
      "84/484, train_loss: 0.0062\n",
      "85/484, train_loss: 0.0211\n",
      "86/484, train_loss: 0.0046\n",
      "87/484, train_loss: 0.0023\n",
      "88/484, train_loss: 0.0048\n",
      "89/484, train_loss: 0.2216\n",
      "90/484, train_loss: 0.0046\n",
      "91/484, train_loss: 0.0113\n",
      "92/484, train_loss: 0.0042\n",
      "93/484, train_loss: 0.0216\n",
      "94/484, train_loss: 0.0051\n",
      "95/484, train_loss: 0.0169\n",
      "96/484, train_loss: 0.4422\n",
      "97/484, train_loss: 0.0030\n",
      "98/484, train_loss: 0.1108\n",
      "99/484, train_loss: 0.4732\n",
      "100/484, train_loss: 0.1304\n",
      "101/484, train_loss: 0.0114\n",
      "102/484, train_loss: 0.0236\n",
      "103/484, train_loss: 0.0033\n",
      "104/484, train_loss: 0.0095\n",
      "105/484, train_loss: 0.0024\n",
      "106/484, train_loss: 0.7959\n",
      "107/484, train_loss: 0.2271\n",
      "108/484, train_loss: 1.2155\n",
      "109/484, train_loss: 0.0241\n",
      "110/484, train_loss: 0.0098\n",
      "111/484, train_loss: 0.0029\n",
      "112/484, train_loss: 0.0049\n",
      "113/484, train_loss: 0.0174\n",
      "114/484, train_loss: 0.0090\n",
      "115/484, train_loss: 0.0108\n",
      "116/484, train_loss: 0.0027\n",
      "117/484, train_loss: 0.0020\n",
      "118/484, train_loss: 0.0040\n",
      "119/484, train_loss: 0.0039\n",
      "120/484, train_loss: 0.0121\n",
      "121/484, train_loss: 0.0067\n",
      "122/484, train_loss: 0.8302\n",
      "123/484, train_loss: 0.0056\n",
      "124/484, train_loss: 0.0045\n",
      "125/484, train_loss: 0.0075\n",
      "126/484, train_loss: 1.3706\n",
      "127/484, train_loss: 0.0748\n",
      "128/484, train_loss: 0.0117\n",
      "129/484, train_loss: 0.0062\n",
      "130/484, train_loss: 0.0048\n",
      "131/484, train_loss: 0.2725\n",
      "132/484, train_loss: 0.0104\n",
      "133/484, train_loss: 0.0052\n",
      "134/484, train_loss: 0.0040\n",
      "135/484, train_loss: 0.0072\n",
      "136/484, train_loss: 0.0042\n",
      "137/484, train_loss: 0.0102\n",
      "138/484, train_loss: 0.0075\n",
      "139/484, train_loss: 0.4443\n",
      "140/484, train_loss: 0.0053\n",
      "141/484, train_loss: 0.0241\n",
      "142/484, train_loss: 0.0070\n",
      "143/484, train_loss: 0.0029\n",
      "144/484, train_loss: 0.0512\n",
      "145/484, train_loss: 0.0046\n",
      "146/484, train_loss: 0.0065\n",
      "147/484, train_loss: 0.0150\n",
      "148/484, train_loss: 0.0048\n",
      "149/484, train_loss: 0.7335\n",
      "150/484, train_loss: 0.0053\n",
      "151/484, train_loss: 0.0028\n",
      "152/484, train_loss: 0.5515\n",
      "153/484, train_loss: 0.0202\n",
      "154/484, train_loss: 0.9642\n",
      "155/484, train_loss: 0.0061\n",
      "156/484, train_loss: 0.0242\n",
      "157/484, train_loss: 0.0023\n",
      "158/484, train_loss: 0.0032\n",
      "159/484, train_loss: 0.2823\n",
      "160/484, train_loss: 0.0086\n",
      "161/484, train_loss: 0.0050\n",
      "162/484, train_loss: 0.0146\n",
      "163/484, train_loss: 0.0163\n",
      "164/484, train_loss: 0.0023\n",
      "165/484, train_loss: 0.0125\n",
      "166/484, train_loss: 0.0025\n",
      "167/484, train_loss: 0.0036\n",
      "168/484, train_loss: 0.0449\n",
      "169/484, train_loss: 0.0097\n",
      "170/484, train_loss: 0.0039\n",
      "171/484, train_loss: 0.0047\n",
      "172/484, train_loss: 0.0028\n",
      "173/484, train_loss: 0.4492\n",
      "174/484, train_loss: 0.3498\n",
      "175/484, train_loss: 0.0042\n",
      "176/484, train_loss: 0.2489\n",
      "177/484, train_loss: 0.0046\n",
      "178/484, train_loss: 0.0046\n",
      "179/484, train_loss: 0.0099\n",
      "180/484, train_loss: 0.0036\n",
      "181/484, train_loss: 0.0900\n",
      "182/484, train_loss: 0.6785\n",
      "183/484, train_loss: 0.0189\n",
      "184/484, train_loss: 0.0152\n",
      "185/484, train_loss: 0.0062\n",
      "186/484, train_loss: 0.0093\n",
      "187/484, train_loss: 0.0022\n",
      "188/484, train_loss: 0.0084\n",
      "189/484, train_loss: 0.0048\n",
      "190/484, train_loss: 0.0108\n",
      "191/484, train_loss: 0.0108\n",
      "192/484, train_loss: 0.2308\n",
      "193/484, train_loss: 0.0091\n",
      "194/484, train_loss: 0.0076\n",
      "195/484, train_loss: 0.0093\n",
      "196/484, train_loss: 0.0046\n",
      "197/484, train_loss: 0.0088\n",
      "198/484, train_loss: 0.0052\n",
      "199/484, train_loss: 0.0032\n",
      "200/484, train_loss: 0.0042\n",
      "201/484, train_loss: 0.0060\n",
      "202/484, train_loss: 0.0064\n",
      "203/484, train_loss: 0.0084\n",
      "204/484, train_loss: 0.0021\n",
      "205/484, train_loss: 0.0280\n",
      "206/484, train_loss: 0.0038\n",
      "207/484, train_loss: 0.0031\n",
      "208/484, train_loss: 0.0037\n",
      "209/484, train_loss: 0.0060\n",
      "210/484, train_loss: 0.0062\n",
      "211/484, train_loss: 0.0040\n",
      "212/484, train_loss: 0.0055\n",
      "213/484, train_loss: 0.0021\n",
      "214/484, train_loss: 0.0072\n",
      "215/484, train_loss: 0.0037\n",
      "216/484, train_loss: 0.3185\n",
      "217/484, train_loss: 0.0083\n",
      "218/484, train_loss: 0.0046\n",
      "219/484, train_loss: 0.0085\n",
      "220/484, train_loss: 0.0021\n",
      "221/484, train_loss: 0.3474\n",
      "222/484, train_loss: 0.0045\n",
      "223/484, train_loss: 0.0044\n",
      "224/484, train_loss: 0.0033\n",
      "225/484, train_loss: 0.0044\n",
      "226/484, train_loss: 0.1734\n",
      "227/484, train_loss: 0.2872\n",
      "228/484, train_loss: 0.0066\n",
      "229/484, train_loss: 0.0024\n",
      "230/484, train_loss: 0.0373\n",
      "231/484, train_loss: 0.0114\n",
      "232/484, train_loss: 0.0088\n",
      "233/484, train_loss: 0.0040\n",
      "234/484, train_loss: 0.0037\n",
      "235/484, train_loss: 0.0014\n",
      "236/484, train_loss: 0.0064\n",
      "237/484, train_loss: 0.0035\n",
      "238/484, train_loss: 0.0053\n",
      "239/484, train_loss: 0.0120\n",
      "240/484, train_loss: 0.0132\n",
      "241/484, train_loss: 0.0032\n",
      "242/484, train_loss: 0.4404\n",
      "243/484, train_loss: 0.0038\n",
      "244/484, train_loss: 0.0039\n",
      "245/484, train_loss: 0.3962\n",
      "246/484, train_loss: 0.0051\n",
      "247/484, train_loss: 0.5787\n",
      "248/484, train_loss: 0.0059\n",
      "249/484, train_loss: 0.0042\n",
      "250/484, train_loss: 0.0024\n",
      "251/484, train_loss: 0.0103\n",
      "252/484, train_loss: 0.0044\n",
      "253/484, train_loss: 0.0035\n",
      "254/484, train_loss: 0.0056\n",
      "255/484, train_loss: 0.0026\n",
      "256/484, train_loss: 0.3612\n",
      "257/484, train_loss: 0.0040\n",
      "258/484, train_loss: 0.0052\n",
      "259/484, train_loss: 0.0033\n",
      "260/484, train_loss: 0.0088\n",
      "261/484, train_loss: 0.0048\n",
      "262/484, train_loss: 0.0116\n",
      "263/484, train_loss: 0.0025\n",
      "264/484, train_loss: 0.0019\n",
      "265/484, train_loss: 0.0053\n",
      "266/484, train_loss: 0.0045\n",
      "267/484, train_loss: 0.6024\n",
      "268/484, train_loss: 0.0027\n",
      "269/484, train_loss: 0.3747\n",
      "270/484, train_loss: 0.0056\n",
      "271/484, train_loss: 0.0023\n",
      "272/484, train_loss: 0.0082\n",
      "273/484, train_loss: 0.2908\n",
      "274/484, train_loss: 0.0059\n",
      "275/484, train_loss: 0.0058\n",
      "276/484, train_loss: 0.2378\n",
      "277/484, train_loss: 0.0040\n",
      "278/484, train_loss: 0.0017\n",
      "279/484, train_loss: 0.1812\n",
      "280/484, train_loss: 0.0024\n",
      "281/484, train_loss: 0.0067\n",
      "282/484, train_loss: 0.0027\n",
      "283/484, train_loss: 0.0115\n",
      "284/484, train_loss: 0.0068\n",
      "285/484, train_loss: 0.0029\n",
      "286/484, train_loss: 0.5732\n",
      "287/484, train_loss: 0.1772\n",
      "288/484, train_loss: 0.0032\n",
      "289/484, train_loss: 0.0040\n",
      "290/484, train_loss: 0.0145\n",
      "291/484, train_loss: 0.0018\n",
      "292/484, train_loss: 0.0097\n",
      "293/484, train_loss: 0.0032\n",
      "294/484, train_loss: 0.0095\n",
      "295/484, train_loss: 0.0024\n",
      "296/484, train_loss: 0.1771\n",
      "297/484, train_loss: 0.0101\n",
      "298/484, train_loss: 0.0037\n",
      "299/484, train_loss: 0.0040\n",
      "300/484, train_loss: 0.0035\n",
      "301/484, train_loss: 0.0045\n",
      "302/484, train_loss: 1.2387\n",
      "303/484, train_loss: 0.6923\n",
      "304/484, train_loss: 0.0117\n",
      "305/484, train_loss: 0.6019\n",
      "306/484, train_loss: 0.0024\n",
      "307/484, train_loss: 0.0021\n",
      "308/484, train_loss: 0.0066\n",
      "309/484, train_loss: 0.0107\n",
      "310/484, train_loss: 0.0123\n",
      "311/484, train_loss: 0.0056\n",
      "312/484, train_loss: 0.0034\n",
      "313/484, train_loss: 0.0083\n",
      "314/484, train_loss: 0.0022\n",
      "315/484, train_loss: 0.0023\n",
      "316/484, train_loss: 0.0081\n",
      "317/484, train_loss: 0.6049\n",
      "318/484, train_loss: 0.0021\n",
      "319/484, train_loss: 0.0105\n",
      "320/484, train_loss: 0.0066\n",
      "321/484, train_loss: 0.0072\n",
      "322/484, train_loss: 0.0028\n",
      "323/484, train_loss: 0.0032\n",
      "324/484, train_loss: 0.0040\n",
      "325/484, train_loss: 0.0075\n",
      "326/484, train_loss: 0.0045\n",
      "327/484, train_loss: 0.0066\n",
      "328/484, train_loss: 0.0020\n",
      "329/484, train_loss: 0.6064\n",
      "330/484, train_loss: 0.0072\n",
      "331/484, train_loss: 0.0025\n",
      "332/484, train_loss: 0.6249\n",
      "333/484, train_loss: 0.0023\n",
      "334/484, train_loss: 0.0203\n",
      "335/484, train_loss: 0.0014\n",
      "336/484, train_loss: 0.5756\n",
      "337/484, train_loss: 0.7861\n",
      "338/484, train_loss: 0.0027\n",
      "339/484, train_loss: 0.0033\n",
      "340/484, train_loss: 0.2261\n",
      "341/484, train_loss: 0.0032\n",
      "342/484, train_loss: 0.2289\n",
      "343/484, train_loss: 0.0035\n",
      "344/484, train_loss: 0.0027\n",
      "345/484, train_loss: 0.0052\n",
      "346/484, train_loss: 0.0164\n",
      "347/484, train_loss: 1.1866\n",
      "348/484, train_loss: 0.0062\n",
      "349/484, train_loss: 0.6914\n",
      "350/484, train_loss: 0.0024\n",
      "351/484, train_loss: 0.0127\n",
      "352/484, train_loss: 0.0014\n",
      "353/484, train_loss: 0.0062\n",
      "354/484, train_loss: 0.7206\n",
      "355/484, train_loss: 0.0081\n",
      "356/484, train_loss: 0.0072\n",
      "357/484, train_loss: 0.0121\n",
      "358/484, train_loss: 0.0077\n",
      "359/484, train_loss: 0.0049\n",
      "360/484, train_loss: 0.7460\n",
      "361/484, train_loss: 0.2637\n",
      "362/484, train_loss: 0.0037\n",
      "363/484, train_loss: 0.0016\n",
      "364/484, train_loss: 0.0062\n",
      "365/484, train_loss: 0.0027\n",
      "366/484, train_loss: 0.0024\n",
      "367/484, train_loss: 0.0195\n",
      "368/484, train_loss: 0.0048\n",
      "369/484, train_loss: 0.3553\n",
      "370/484, train_loss: 0.0050\n",
      "371/484, train_loss: 0.0017\n",
      "372/484, train_loss: 0.0027\n",
      "373/484, train_loss: 0.0058\n",
      "374/484, train_loss: 0.0023\n",
      "375/484, train_loss: 0.0050\n",
      "376/484, train_loss: 0.0151\n",
      "377/484, train_loss: 0.0033\n",
      "378/484, train_loss: 0.0029\n",
      "379/484, train_loss: 0.0054\n",
      "380/484, train_loss: 0.0035\n",
      "381/484, train_loss: 0.2145\n",
      "382/484, train_loss: 0.0037\n",
      "383/484, train_loss: 0.0023\n",
      "384/484, train_loss: 0.0022\n",
      "385/484, train_loss: 0.0030\n",
      "386/484, train_loss: 0.5375\n",
      "387/484, train_loss: 0.0047\n",
      "388/484, train_loss: 0.7531\n",
      "389/484, train_loss: 0.0045\n",
      "390/484, train_loss: 0.1453\n",
      "391/484, train_loss: 0.0044\n",
      "392/484, train_loss: 0.0043\n",
      "393/484, train_loss: 0.0048\n",
      "394/484, train_loss: 0.3800\n",
      "395/484, train_loss: 0.1617\n",
      "396/484, train_loss: 0.0032\n",
      "397/484, train_loss: 0.0071\n",
      "398/484, train_loss: 1.1803\n",
      "399/484, train_loss: 0.0040\n",
      "400/484, train_loss: 0.0015\n",
      "401/484, train_loss: 0.0031\n",
      "402/484, train_loss: 0.0561\n",
      "403/484, train_loss: 0.0052\n",
      "404/484, train_loss: 0.0034\n",
      "405/484, train_loss: 0.0037\n",
      "406/484, train_loss: 0.0025\n",
      "407/484, train_loss: 0.0068\n",
      "408/484, train_loss: 0.0025\n",
      "409/484, train_loss: 0.1924\n",
      "410/484, train_loss: 0.0023\n",
      "411/484, train_loss: 0.0136\n",
      "412/484, train_loss: 0.0080\n",
      "413/484, train_loss: 0.0035\n",
      "414/484, train_loss: 1.0729\n",
      "415/484, train_loss: 0.0028\n",
      "416/484, train_loss: 0.0024\n",
      "417/484, train_loss: 0.0024\n",
      "418/484, train_loss: 0.0043\n",
      "419/484, train_loss: 0.0170\n",
      "420/484, train_loss: 0.0033\n",
      "421/484, train_loss: 0.0066\n",
      "422/484, train_loss: 0.0024\n",
      "423/484, train_loss: 0.0081\n",
      "424/484, train_loss: 0.0017\n",
      "425/484, train_loss: 0.0817\n",
      "426/484, train_loss: 0.0124\n",
      "427/484, train_loss: 0.0037\n",
      "428/484, train_loss: 0.0064\n",
      "429/484, train_loss: 0.0044\n",
      "430/484, train_loss: 0.1572\n",
      "431/484, train_loss: 0.0017\n",
      "432/484, train_loss: 0.0027\n",
      "433/484, train_loss: 0.0070\n",
      "434/484, train_loss: 0.0026\n",
      "435/484, train_loss: 0.0090\n",
      "436/484, train_loss: 0.3942\n",
      "437/484, train_loss: 0.0018\n",
      "438/484, train_loss: 0.0049\n",
      "439/484, train_loss: 0.0053\n",
      "440/484, train_loss: 0.0035\n",
      "441/484, train_loss: 0.0033\n",
      "442/484, train_loss: 0.0105\n",
      "443/484, train_loss: 0.0024\n",
      "444/484, train_loss: 0.0058\n",
      "445/484, train_loss: 0.0028\n",
      "446/484, train_loss: 0.0208\n",
      "447/484, train_loss: 0.0072\n",
      "448/484, train_loss: 0.0023\n",
      "449/484, train_loss: 0.0039\n",
      "450/484, train_loss: 0.1784\n",
      "451/484, train_loss: 0.3050\n",
      "452/484, train_loss: 0.0066\n",
      "453/484, train_loss: 0.0015\n",
      "454/484, train_loss: 0.0067\n",
      "455/484, train_loss: 0.0170\n",
      "456/484, train_loss: 0.0031\n",
      "457/484, train_loss: 0.0025\n",
      "458/484, train_loss: 0.0120\n",
      "459/484, train_loss: 0.3836\n",
      "460/484, train_loss: 0.1894\n",
      "461/484, train_loss: 0.6749\n",
      "462/484, train_loss: 0.0032\n",
      "463/484, train_loss: 0.0053\n",
      "464/484, train_loss: 0.0025\n",
      "465/484, train_loss: 0.3705\n",
      "466/484, train_loss: 0.0022\n",
      "467/484, train_loss: 0.0025\n",
      "468/484, train_loss: 0.2969\n",
      "469/484, train_loss: 0.0041\n",
      "470/484, train_loss: 0.0039\n",
      "471/484, train_loss: 0.0040\n",
      "472/484, train_loss: 0.2075\n",
      "473/484, train_loss: 0.2693\n",
      "474/484, train_loss: 0.0032\n",
      "475/484, train_loss: 0.0020\n",
      "476/484, train_loss: 0.0010\n",
      "477/484, train_loss: 0.3598\n",
      "478/484, train_loss: 0.0025\n",
      "479/484, train_loss: 0.0033\n",
      "480/484, train_loss: 0.0671\n",
      "481/484, train_loss: 0.0017\n",
      "482/484, train_loss: 0.1290\n",
      "483/484, train_loss: 0.4929\n",
      "484/484, train_loss: 0.0017\n",
      "epoch 13 average loss: 0.0960\n",
      "----------\n",
      "epoch 14/20\n",
      "1/484, train_loss: 0.0038\n",
      "2/484, train_loss: 0.0046\n",
      "3/484, train_loss: 0.0027\n",
      "4/484, train_loss: 0.9718\n",
      "5/484, train_loss: 0.0037\n",
      "6/484, train_loss: 0.0098\n",
      "7/484, train_loss: 0.0105\n",
      "8/484, train_loss: 0.0040\n",
      "9/484, train_loss: 0.1381\n",
      "10/484, train_loss: 0.0011\n",
      "11/484, train_loss: 0.0022\n",
      "12/484, train_loss: 0.0029\n",
      "13/484, train_loss: 0.1321\n",
      "14/484, train_loss: 0.0061\n",
      "15/484, train_loss: 0.0046\n",
      "16/484, train_loss: 0.0033\n",
      "17/484, train_loss: 0.0027\n",
      "18/484, train_loss: 0.4631\n",
      "19/484, train_loss: 0.0037\n",
      "20/484, train_loss: 0.9373\n",
      "21/484, train_loss: 0.0019\n",
      "22/484, train_loss: 0.0045\n",
      "23/484, train_loss: 0.0026\n",
      "24/484, train_loss: 0.0197\n",
      "25/484, train_loss: 0.0089\n",
      "26/484, train_loss: 0.2734\n",
      "27/484, train_loss: 0.0026\n",
      "28/484, train_loss: 0.0058\n",
      "29/484, train_loss: 0.0388\n",
      "30/484, train_loss: 0.0047\n",
      "31/484, train_loss: 0.0046\n",
      "32/484, train_loss: 0.0037\n",
      "33/484, train_loss: 0.1488\n",
      "34/484, train_loss: 0.0026\n",
      "35/484, train_loss: 0.2491\n",
      "36/484, train_loss: 0.0027\n",
      "37/484, train_loss: 1.5040\n",
      "38/484, train_loss: 0.3360\n",
      "39/484, train_loss: 0.0020\n",
      "40/484, train_loss: 0.0018\n",
      "41/484, train_loss: 0.0120\n",
      "42/484, train_loss: 0.0517\n",
      "43/484, train_loss: 0.0021\n",
      "44/484, train_loss: 0.0055\n",
      "45/484, train_loss: 0.0016\n",
      "46/484, train_loss: 0.0031\n",
      "47/484, train_loss: 0.0062\n",
      "48/484, train_loss: 0.0041\n",
      "49/484, train_loss: 0.0051\n",
      "50/484, train_loss: 0.0009\n",
      "51/484, train_loss: 0.0074\n",
      "52/484, train_loss: 0.0053\n",
      "53/484, train_loss: 0.0077\n",
      "54/484, train_loss: 0.0027\n",
      "55/484, train_loss: 0.0306\n",
      "56/484, train_loss: 0.0013\n",
      "57/484, train_loss: 0.0126\n",
      "58/484, train_loss: 0.0017\n",
      "59/484, train_loss: 0.0010\n",
      "60/484, train_loss: 0.0089\n",
      "61/484, train_loss: 0.0030\n",
      "62/484, train_loss: 0.0051\n",
      "63/484, train_loss: 0.0009\n",
      "64/484, train_loss: 0.0014\n",
      "65/484, train_loss: 0.0075\n",
      "66/484, train_loss: 0.0010\n",
      "67/484, train_loss: 0.5489\n",
      "68/484, train_loss: 0.0016\n",
      "69/484, train_loss: 0.0045\n",
      "70/484, train_loss: 0.0047\n",
      "71/484, train_loss: 0.2523\n",
      "72/484, train_loss: 0.0009\n",
      "73/484, train_loss: 0.0051\n",
      "74/484, train_loss: 0.0129\n",
      "75/484, train_loss: 0.0039\n",
      "76/484, train_loss: 0.0029\n",
      "77/484, train_loss: 0.0027\n",
      "78/484, train_loss: 0.1514\n",
      "79/484, train_loss: 0.0041\n",
      "80/484, train_loss: 0.0047\n",
      "81/484, train_loss: 0.0104\n",
      "82/484, train_loss: 0.2165\n",
      "83/484, train_loss: 0.0015\n",
      "84/484, train_loss: 0.0039\n",
      "85/484, train_loss: 0.0034\n",
      "86/484, train_loss: 0.0044\n",
      "87/484, train_loss: 0.0018\n",
      "88/484, train_loss: 0.0020\n",
      "89/484, train_loss: 0.0180\n",
      "90/484, train_loss: 0.0046\n",
      "91/484, train_loss: 0.0029\n",
      "92/484, train_loss: 0.0042\n",
      "93/484, train_loss: 0.0018\n",
      "94/484, train_loss: 0.8370\n",
      "95/484, train_loss: 0.1535\n",
      "96/484, train_loss: 0.0043\n",
      "97/484, train_loss: 0.0046\n",
      "98/484, train_loss: 0.0061\n",
      "99/484, train_loss: 0.0012\n",
      "100/484, train_loss: 0.0016\n",
      "101/484, train_loss: 0.0030\n",
      "102/484, train_loss: 0.0020\n",
      "103/484, train_loss: 0.0067\n",
      "104/484, train_loss: 1.4294\n",
      "105/484, train_loss: 0.0056\n",
      "106/484, train_loss: 0.0116\n",
      "107/484, train_loss: 0.0017\n",
      "108/484, train_loss: 0.0021\n",
      "109/484, train_loss: 0.0059\n",
      "110/484, train_loss: 1.0714\n",
      "111/484, train_loss: 0.0185\n",
      "112/484, train_loss: 0.0084\n",
      "113/484, train_loss: 0.0039\n",
      "114/484, train_loss: 0.0029\n",
      "115/484, train_loss: 0.0918\n",
      "116/484, train_loss: 0.0096\n",
      "117/484, train_loss: 0.5044\n",
      "118/484, train_loss: 0.0020\n",
      "119/484, train_loss: 0.1616\n",
      "120/484, train_loss: 0.0038\n",
      "121/484, train_loss: 0.0030\n",
      "122/484, train_loss: 0.6854\n",
      "123/484, train_loss: 0.0099\n",
      "124/484, train_loss: 0.0032\n",
      "125/484, train_loss: 0.0010\n",
      "126/484, train_loss: 0.0035\n",
      "127/484, train_loss: 0.0130\n",
      "128/484, train_loss: 0.0038\n",
      "129/484, train_loss: 0.0031\n",
      "130/484, train_loss: 0.9920\n",
      "131/484, train_loss: 0.0045\n",
      "132/484, train_loss: 0.0055\n",
      "133/484, train_loss: 0.5302\n",
      "134/484, train_loss: 0.0043\n",
      "135/484, train_loss: 0.0036\n",
      "136/484, train_loss: 0.0024\n",
      "137/484, train_loss: 0.0285\n",
      "138/484, train_loss: 0.0015\n",
      "139/484, train_loss: 0.2755\n",
      "140/484, train_loss: 0.0014\n",
      "141/484, train_loss: 0.0041\n",
      "142/484, train_loss: 0.0017\n",
      "143/484, train_loss: 0.4598\n",
      "144/484, train_loss: 0.0029\n",
      "145/484, train_loss: 0.0096\n",
      "146/484, train_loss: 0.0022\n",
      "147/484, train_loss: 0.0037\n",
      "148/484, train_loss: 0.0047\n",
      "149/484, train_loss: 0.0024\n",
      "150/484, train_loss: 0.2685\n",
      "151/484, train_loss: 0.0257\n",
      "152/484, train_loss: 0.3717\n",
      "153/484, train_loss: 0.2821\n",
      "154/484, train_loss: 0.0038\n",
      "155/484, train_loss: 0.0023\n",
      "156/484, train_loss: 0.0057\n",
      "157/484, train_loss: 0.0026\n",
      "158/484, train_loss: 0.2325\n",
      "159/484, train_loss: 0.0038\n",
      "160/484, train_loss: 0.0042\n",
      "161/484, train_loss: 0.0028\n",
      "162/484, train_loss: 0.0068\n",
      "163/484, train_loss: 0.0023\n",
      "164/484, train_loss: 0.0050\n",
      "165/484, train_loss: 0.0048\n",
      "166/484, train_loss: 0.0107\n",
      "167/484, train_loss: 0.0031\n",
      "168/484, train_loss: 1.1609\n",
      "169/484, train_loss: 0.0013\n",
      "170/484, train_loss: 0.0026\n",
      "171/484, train_loss: 0.0090\n",
      "172/484, train_loss: 0.2913\n",
      "173/484, train_loss: 0.0044\n",
      "174/484, train_loss: 0.0026\n",
      "175/484, train_loss: 0.0121\n",
      "176/484, train_loss: 0.2233\n",
      "177/484, train_loss: 0.0100\n",
      "178/484, train_loss: 0.0102\n",
      "179/484, train_loss: 0.0068\n",
      "180/484, train_loss: 0.0069\n",
      "181/484, train_loss: 1.1852\n",
      "182/484, train_loss: 0.0131\n",
      "183/484, train_loss: 0.0037\n",
      "184/484, train_loss: 0.0028\n",
      "185/484, train_loss: 0.0045\n",
      "186/484, train_loss: 0.0030\n",
      "187/484, train_loss: 0.0020\n",
      "188/484, train_loss: 0.0037\n",
      "189/484, train_loss: 0.0020\n",
      "190/484, train_loss: 0.0091\n",
      "191/484, train_loss: 0.0025\n",
      "192/484, train_loss: 0.0053\n",
      "193/484, train_loss: 0.0029\n",
      "194/484, train_loss: 0.0036\n",
      "195/484, train_loss: 0.0141\n",
      "196/484, train_loss: 0.0024\n",
      "197/484, train_loss: 0.0028\n",
      "198/484, train_loss: 0.0038\n",
      "199/484, train_loss: 0.0050\n",
      "200/484, train_loss: 0.0034\n",
      "201/484, train_loss: 0.0012\n",
      "202/484, train_loss: 0.0025\n",
      "203/484, train_loss: 0.0020\n",
      "204/484, train_loss: 0.0033\n",
      "205/484, train_loss: 0.0024\n",
      "206/484, train_loss: 0.4550\n",
      "207/484, train_loss: 0.0036\n",
      "208/484, train_loss: 0.0034\n",
      "209/484, train_loss: 0.0222\n",
      "210/484, train_loss: 0.0054\n",
      "211/484, train_loss: 0.0023\n",
      "212/484, train_loss: 0.0012\n",
      "213/484, train_loss: 0.2620\n",
      "214/484, train_loss: 0.0030\n",
      "215/484, train_loss: 0.0113\n",
      "216/484, train_loss: 0.0018\n",
      "217/484, train_loss: 1.9857\n",
      "218/484, train_loss: 0.0041\n",
      "219/484, train_loss: 0.0032\n",
      "220/484, train_loss: 0.0046\n",
      "221/484, train_loss: 0.0101\n",
      "222/484, train_loss: 0.0020\n",
      "223/484, train_loss: 1.6332\n",
      "224/484, train_loss: 0.3425\n",
      "225/484, train_loss: 0.0021\n",
      "226/484, train_loss: 0.0049\n",
      "227/484, train_loss: 0.9062\n",
      "228/484, train_loss: 0.0031\n",
      "229/484, train_loss: 0.0030\n",
      "230/484, train_loss: 0.0052\n",
      "231/484, train_loss: 0.0019\n",
      "232/484, train_loss: 0.0013\n",
      "233/484, train_loss: 0.8673\n",
      "234/484, train_loss: 0.0036\n",
      "235/484, train_loss: 0.0047\n",
      "236/484, train_loss: 0.2410\n",
      "237/484, train_loss: 0.0053\n",
      "238/484, train_loss: 0.0028\n",
      "239/484, train_loss: 0.0040\n",
      "240/484, train_loss: 0.0045\n",
      "241/484, train_loss: 0.0234\n",
      "242/484, train_loss: 0.4995\n",
      "243/484, train_loss: 0.0460\n",
      "244/484, train_loss: 0.7493\n",
      "245/484, train_loss: 0.0016\n",
      "246/484, train_loss: 0.0121\n",
      "247/484, train_loss: 0.0240\n",
      "248/484, train_loss: 0.0053\n",
      "249/484, train_loss: 0.0042\n",
      "250/484, train_loss: 0.6652\n",
      "251/484, train_loss: 0.0084\n",
      "252/484, train_loss: 0.0392\n",
      "253/484, train_loss: 0.5468\n",
      "254/484, train_loss: 0.0120\n",
      "255/484, train_loss: 0.0051\n",
      "256/484, train_loss: 0.0882\n",
      "257/484, train_loss: 0.0102\n",
      "258/484, train_loss: 0.2948\n",
      "259/484, train_loss: 0.0031\n",
      "260/484, train_loss: 0.0028\n",
      "261/484, train_loss: 0.0064\n",
      "262/484, train_loss: 0.0244\n",
      "263/484, train_loss: 0.0041\n",
      "264/484, train_loss: 0.0022\n",
      "265/484, train_loss: 0.0048\n",
      "266/484, train_loss: 0.7771\n",
      "267/484, train_loss: 0.0047\n",
      "268/484, train_loss: 0.0019\n",
      "269/484, train_loss: 0.0013\n",
      "270/484, train_loss: 0.0034\n",
      "271/484, train_loss: 0.0038\n",
      "272/484, train_loss: 0.2240\n",
      "273/484, train_loss: 0.0040\n",
      "274/484, train_loss: 0.0050\n",
      "275/484, train_loss: 0.0014\n",
      "276/484, train_loss: 0.0040\n",
      "277/484, train_loss: 0.0112\n",
      "278/484, train_loss: 0.0051\n",
      "279/484, train_loss: 0.0042\n",
      "280/484, train_loss: 0.0015\n",
      "281/484, train_loss: 0.0020\n",
      "282/484, train_loss: 0.0043\n",
      "283/484, train_loss: 0.0009\n",
      "284/484, train_loss: 0.0063\n",
      "285/484, train_loss: 0.0038\n",
      "286/484, train_loss: 0.0036\n",
      "287/484, train_loss: 0.0055\n",
      "288/484, train_loss: 0.0034\n",
      "289/484, train_loss: 0.8463\n",
      "290/484, train_loss: 0.3476\n",
      "291/484, train_loss: 0.4659\n",
      "292/484, train_loss: 0.0008\n",
      "293/484, train_loss: 0.0053\n",
      "294/484, train_loss: 0.2880\n",
      "295/484, train_loss: 0.0010\n",
      "296/484, train_loss: 0.0020\n",
      "297/484, train_loss: 0.0042\n",
      "298/484, train_loss: 0.0036\n",
      "299/484, train_loss: 0.0060\n",
      "300/484, train_loss: 0.0010\n",
      "301/484, train_loss: 1.1627\n",
      "302/484, train_loss: 0.0009\n",
      "303/484, train_loss: 0.2387\n",
      "304/484, train_loss: 0.0019\n",
      "305/484, train_loss: 0.8498\n",
      "306/484, train_loss: 0.0016\n",
      "307/484, train_loss: 0.0043\n",
      "308/484, train_loss: 0.0042\n",
      "309/484, train_loss: 0.2935\n",
      "310/484, train_loss: 0.0038\n",
      "311/484, train_loss: 0.0018\n",
      "312/484, train_loss: 0.6429\n",
      "313/484, train_loss: 0.0021\n",
      "314/484, train_loss: 0.2247\n",
      "315/484, train_loss: 0.4269\n",
      "316/484, train_loss: 0.6484\n",
      "317/484, train_loss: 0.0023\n",
      "318/484, train_loss: 0.0078\n",
      "319/484, train_loss: 0.0034\n",
      "320/484, train_loss: 0.0010\n",
      "321/484, train_loss: 0.0199\n",
      "322/484, train_loss: 0.0016\n",
      "323/484, train_loss: 0.5297\n",
      "324/484, train_loss: 0.0010\n",
      "325/484, train_loss: 0.3992\n",
      "326/484, train_loss: 1.0979\n",
      "327/484, train_loss: 0.0017\n",
      "328/484, train_loss: 0.3958\n",
      "329/484, train_loss: 0.0037\n",
      "330/484, train_loss: 0.8338\n",
      "331/484, train_loss: 0.3269\n",
      "332/484, train_loss: 0.0067\n",
      "333/484, train_loss: 0.5773\n",
      "334/484, train_loss: 0.0022\n",
      "335/484, train_loss: 0.3326\n",
      "336/484, train_loss: 0.0026\n",
      "337/484, train_loss: 0.0057\n",
      "338/484, train_loss: 0.0077\n",
      "339/484, train_loss: 0.0124\n",
      "340/484, train_loss: 0.0064\n",
      "341/484, train_loss: 0.0055\n",
      "342/484, train_loss: 0.0569\n",
      "343/484, train_loss: 0.0026\n",
      "344/484, train_loss: 0.0014\n",
      "345/484, train_loss: 0.0008\n",
      "346/484, train_loss: 0.0064\n",
      "347/484, train_loss: 0.4236\n",
      "348/484, train_loss: 0.5493\n",
      "349/484, train_loss: 0.0048\n",
      "350/484, train_loss: 0.0035\n",
      "351/484, train_loss: 0.0093\n",
      "352/484, train_loss: 0.0030\n",
      "353/484, train_loss: 0.0047\n",
      "354/484, train_loss: 0.0010\n",
      "355/484, train_loss: 0.0081\n",
      "356/484, train_loss: 0.0041\n",
      "357/484, train_loss: 0.0023\n",
      "358/484, train_loss: 0.0035\n",
      "359/484, train_loss: 0.0024\n",
      "360/484, train_loss: 0.0037\n",
      "361/484, train_loss: 0.2574\n",
      "362/484, train_loss: 0.1281\n",
      "363/484, train_loss: 0.0044\n",
      "364/484, train_loss: 0.0029\n",
      "365/484, train_loss: 0.0020\n",
      "366/484, train_loss: 0.0022\n",
      "367/484, train_loss: 0.0037\n",
      "368/484, train_loss: 0.6558\n",
      "369/484, train_loss: 0.0106\n",
      "370/484, train_loss: 0.0025\n",
      "371/484, train_loss: 0.0018\n",
      "372/484, train_loss: 0.0031\n",
      "373/484, train_loss: 0.0033\n",
      "374/484, train_loss: 0.0470\n",
      "375/484, train_loss: 0.0164\n",
      "376/484, train_loss: 0.0473\n",
      "377/484, train_loss: 0.0031\n",
      "378/484, train_loss: 0.0010\n",
      "379/484, train_loss: 0.3712\n",
      "380/484, train_loss: 0.7821\n",
      "381/484, train_loss: 0.6839\n",
      "382/484, train_loss: 0.2056\n",
      "383/484, train_loss: 0.0920\n",
      "384/484, train_loss: 0.0030\n",
      "385/484, train_loss: 0.0103\n",
      "386/484, train_loss: 0.0027\n",
      "387/484, train_loss: 0.0039\n",
      "388/484, train_loss: 0.0019\n",
      "389/484, train_loss: 0.0071\n",
      "390/484, train_loss: 0.0015\n",
      "391/484, train_loss: 0.0090\n",
      "392/484, train_loss: 0.0023\n",
      "393/484, train_loss: 0.0047\n",
      "394/484, train_loss: 0.0039\n",
      "395/484, train_loss: 0.0012\n",
      "396/484, train_loss: 0.0044\n",
      "397/484, train_loss: 0.0008\n",
      "398/484, train_loss: 0.0022\n",
      "399/484, train_loss: 0.0029\n",
      "400/484, train_loss: 0.0017\n",
      "401/484, train_loss: 0.0039\n",
      "402/484, train_loss: 0.0011\n",
      "403/484, train_loss: 0.0055\n",
      "404/484, train_loss: 0.0026\n",
      "405/484, train_loss: 0.0956\n",
      "406/484, train_loss: 0.0019\n",
      "407/484, train_loss: 1.0815\n",
      "408/484, train_loss: 0.0057\n",
      "409/484, train_loss: 0.0056\n",
      "410/484, train_loss: 0.0083\n",
      "411/484, train_loss: 0.2177\n",
      "412/484, train_loss: 0.0021\n",
      "413/484, train_loss: 0.0062\n",
      "414/484, train_loss: 0.0022\n",
      "415/484, train_loss: 0.0083\n",
      "416/484, train_loss: 0.0055\n",
      "417/484, train_loss: 0.0081\n",
      "418/484, train_loss: 0.0014\n",
      "419/484, train_loss: 0.0084\n",
      "420/484, train_loss: 0.2925\n",
      "421/484, train_loss: 0.0039\n",
      "422/484, train_loss: 0.0110\n",
      "423/484, train_loss: 0.0009\n",
      "424/484, train_loss: 0.0012\n",
      "425/484, train_loss: 0.0017\n",
      "426/484, train_loss: 0.0049\n",
      "427/484, train_loss: 0.0281\n",
      "428/484, train_loss: 0.0065\n",
      "429/484, train_loss: 0.0145\n",
      "430/484, train_loss: 0.0078\n",
      "431/484, train_loss: 0.0103\n",
      "432/484, train_loss: 0.6761\n",
      "433/484, train_loss: 0.0541\n",
      "434/484, train_loss: 0.0065\n",
      "435/484, train_loss: 0.0080\n",
      "436/484, train_loss: 1.4091\n",
      "437/484, train_loss: 0.0013\n",
      "438/484, train_loss: 0.2468\n",
      "439/484, train_loss: 0.0028\n",
      "440/484, train_loss: 0.0072\n",
      "441/484, train_loss: 0.0044\n",
      "442/484, train_loss: 0.0020\n",
      "443/484, train_loss: 0.0058\n",
      "444/484, train_loss: 0.0105\n",
      "445/484, train_loss: 0.0025\n",
      "446/484, train_loss: 0.0010\n",
      "447/484, train_loss: 0.0111\n",
      "448/484, train_loss: 0.0011\n",
      "449/484, train_loss: 0.1989\n",
      "450/484, train_loss: 0.0698\n",
      "451/484, train_loss: 0.0015\n",
      "452/484, train_loss: 0.4686\n",
      "453/484, train_loss: 0.0025\n",
      "454/484, train_loss: 0.0025\n",
      "455/484, train_loss: 0.0062\n",
      "456/484, train_loss: 0.7883\n",
      "457/484, train_loss: 0.0063\n",
      "458/484, train_loss: 0.0031\n",
      "459/484, train_loss: 1.6639\n",
      "460/484, train_loss: 0.0047\n",
      "461/484, train_loss: 0.0097\n",
      "462/484, train_loss: 0.0070\n",
      "463/484, train_loss: 0.0028\n",
      "464/484, train_loss: 0.0074\n",
      "465/484, train_loss: 0.0026\n",
      "466/484, train_loss: 0.0039\n",
      "467/484, train_loss: 0.0007\n",
      "468/484, train_loss: 0.0026\n",
      "469/484, train_loss: 0.9189\n",
      "470/484, train_loss: 0.0014\n",
      "471/484, train_loss: 0.1927\n",
      "472/484, train_loss: 0.0027\n",
      "473/484, train_loss: 0.0011\n",
      "474/484, train_loss: 0.4655\n",
      "475/484, train_loss: 0.0027\n",
      "476/484, train_loss: 0.0035\n",
      "477/484, train_loss: 0.0011\n",
      "478/484, train_loss: 0.0012\n",
      "479/484, train_loss: 0.0021\n",
      "480/484, train_loss: 0.3142\n",
      "481/484, train_loss: 0.0078\n",
      "482/484, train_loss: 0.7296\n",
      "483/484, train_loss: 0.0105\n",
      "484/484, train_loss: 0.0025\n",
      "epoch 14 average loss: 0.1141\n",
      "saved new best metric model\n",
      "Current epoch: 14 current accuracy: 0.8669 \n",
      "Best accuracy: 0.8669 at epoch 14\n",
      "----------\n",
      "epoch 15/20\n",
      "1/484, train_loss: 0.0028\n",
      "2/484, train_loss: 0.1079\n",
      "3/484, train_loss: 0.0167\n",
      "4/484, train_loss: 0.0021\n",
      "5/484, train_loss: 0.3997\n",
      "6/484, train_loss: 0.0009\n",
      "7/484, train_loss: 0.0038\n",
      "8/484, train_loss: 0.0153\n",
      "9/484, train_loss: 0.7164\n",
      "10/484, train_loss: 0.0028\n",
      "11/484, train_loss: 0.0033\n",
      "12/484, train_loss: 0.0031\n",
      "13/484, train_loss: 0.0030\n",
      "14/484, train_loss: 1.1691\n",
      "15/484, train_loss: 0.0042\n",
      "16/484, train_loss: 0.0066\n",
      "17/484, train_loss: 0.0020\n",
      "18/484, train_loss: 0.0029\n",
      "19/484, train_loss: 0.0012\n",
      "20/484, train_loss: 0.0039\n",
      "21/484, train_loss: 0.0040\n",
      "22/484, train_loss: 0.0030\n",
      "23/484, train_loss: 0.0018\n",
      "24/484, train_loss: 0.0068\n",
      "25/484, train_loss: 0.0022\n",
      "26/484, train_loss: 0.0170\n",
      "27/484, train_loss: 0.0009\n",
      "28/484, train_loss: 0.0024\n",
      "29/484, train_loss: 0.0048\n",
      "30/484, train_loss: 0.0016\n",
      "31/484, train_loss: 0.0013\n",
      "32/484, train_loss: 0.0037\n",
      "33/484, train_loss: 0.0154\n",
      "34/484, train_loss: 0.0089\n",
      "35/484, train_loss: 0.0021\n",
      "36/484, train_loss: 0.0025\n",
      "37/484, train_loss: 0.0034\n",
      "38/484, train_loss: 0.0260\n",
      "39/484, train_loss: 0.2938\n",
      "40/484, train_loss: 0.1406\n",
      "41/484, train_loss: 0.0057\n",
      "42/484, train_loss: 0.0016\n",
      "43/484, train_loss: 0.0017\n",
      "44/484, train_loss: 0.0014\n",
      "45/484, train_loss: 0.4687\n",
      "46/484, train_loss: 0.8469\n",
      "47/484, train_loss: 0.0040\n",
      "48/484, train_loss: 0.0024\n",
      "49/484, train_loss: 0.9190\n",
      "50/484, train_loss: 0.2414\n",
      "51/484, train_loss: 0.6321\n",
      "52/484, train_loss: 0.0081\n",
      "53/484, train_loss: 0.2042\n",
      "54/484, train_loss: 0.0039\n",
      "55/484, train_loss: 0.5043\n",
      "56/484, train_loss: 0.0011\n",
      "57/484, train_loss: 0.0009\n",
      "58/484, train_loss: 0.0013\n",
      "59/484, train_loss: 0.0059\n",
      "60/484, train_loss: 0.0170\n",
      "61/484, train_loss: 0.0067\n",
      "62/484, train_loss: 0.0024\n",
      "63/484, train_loss: 0.0078\n",
      "64/484, train_loss: 0.0050\n",
      "65/484, train_loss: 0.1727\n",
      "66/484, train_loss: 0.9943\n",
      "67/484, train_loss: 0.0029\n",
      "68/484, train_loss: 0.0021\n",
      "69/484, train_loss: 0.0031\n",
      "70/484, train_loss: 0.0010\n",
      "71/484, train_loss: 0.0026\n",
      "72/484, train_loss: 0.0028\n",
      "73/484, train_loss: 0.3772\n",
      "74/484, train_loss: 0.0046\n",
      "75/484, train_loss: 0.0009\n",
      "76/484, train_loss: 0.0018\n",
      "77/484, train_loss: 0.0026\n",
      "78/484, train_loss: 0.2186\n",
      "79/484, train_loss: 0.0023\n",
      "80/484, train_loss: 0.0024\n",
      "81/484, train_loss: 0.0019\n",
      "82/484, train_loss: 0.0030\n",
      "83/484, train_loss: 0.0009\n",
      "84/484, train_loss: 0.0021\n",
      "85/484, train_loss: 0.0050\n",
      "86/484, train_loss: 0.0019\n",
      "87/484, train_loss: 0.0045\n",
      "88/484, train_loss: 0.0030\n",
      "89/484, train_loss: 0.0017\n",
      "90/484, train_loss: 0.0016\n",
      "91/484, train_loss: 0.0025\n",
      "92/484, train_loss: 0.0032\n",
      "93/484, train_loss: 0.0023\n",
      "94/484, train_loss: 0.0032\n",
      "95/484, train_loss: 0.1213\n",
      "96/484, train_loss: 0.0098\n",
      "97/484, train_loss: 0.0022\n",
      "98/484, train_loss: 0.6833\n",
      "99/484, train_loss: 0.1288\n",
      "100/484, train_loss: 0.0048\n",
      "101/484, train_loss: 0.0039\n",
      "102/484, train_loss: 0.0019\n",
      "103/484, train_loss: 0.0029\n",
      "104/484, train_loss: 0.2868\n",
      "105/484, train_loss: 0.0018\n",
      "106/484, train_loss: 1.7451\n",
      "107/484, train_loss: 0.0082\n",
      "108/484, train_loss: 0.0021\n",
      "109/484, train_loss: 0.0015\n",
      "110/484, train_loss: 0.0019\n",
      "111/484, train_loss: 0.3755\n",
      "112/484, train_loss: 0.3624\n",
      "113/484, train_loss: 0.0014\n",
      "114/484, train_loss: 0.0091\n",
      "115/484, train_loss: 0.0027\n",
      "116/484, train_loss: 0.0075\n",
      "117/484, train_loss: 0.0017\n",
      "118/484, train_loss: 0.0016\n",
      "119/484, train_loss: 0.0061\n",
      "120/484, train_loss: 0.0029\n",
      "121/484, train_loss: 0.0016\n",
      "122/484, train_loss: 0.0066\n",
      "123/484, train_loss: 0.0017\n",
      "124/484, train_loss: 0.0019\n",
      "125/484, train_loss: 0.0129\n",
      "126/484, train_loss: 0.0027\n",
      "127/484, train_loss: 1.4347\n",
      "128/484, train_loss: 0.0027\n",
      "129/484, train_loss: 0.2801\n",
      "130/484, train_loss: 0.0014\n",
      "131/484, train_loss: 0.0020\n",
      "132/484, train_loss: 0.0023\n",
      "133/484, train_loss: 0.0020\n",
      "134/484, train_loss: 0.0024\n",
      "135/484, train_loss: 0.5295\n",
      "136/484, train_loss: 0.0017\n",
      "137/484, train_loss: 0.0033\n",
      "138/484, train_loss: 0.0008\n",
      "139/484, train_loss: 0.3770\n",
      "140/484, train_loss: 0.0011\n",
      "141/484, train_loss: 0.0034\n",
      "142/484, train_loss: 0.0025\n",
      "143/484, train_loss: 0.0024\n",
      "144/484, train_loss: 0.0388\n",
      "145/484, train_loss: 0.0025\n",
      "146/484, train_loss: 0.0029\n",
      "147/484, train_loss: 0.0007\n",
      "148/484, train_loss: 0.0018\n",
      "149/484, train_loss: 0.0112\n",
      "150/484, train_loss: 0.1318\n",
      "151/484, train_loss: 0.0025\n",
      "152/484, train_loss: 0.0020\n",
      "153/484, train_loss: 0.6075\n",
      "154/484, train_loss: 0.4374\n",
      "155/484, train_loss: 0.0010\n",
      "156/484, train_loss: 0.1998\n",
      "157/484, train_loss: 0.0023\n",
      "158/484, train_loss: 0.0015\n",
      "159/484, train_loss: 0.0009\n",
      "160/484, train_loss: 0.0018\n",
      "161/484, train_loss: 0.0022\n",
      "162/484, train_loss: 0.0045\n",
      "163/484, train_loss: 0.2734\n",
      "164/484, train_loss: 0.1891\n",
      "165/484, train_loss: 0.0025\n",
      "166/484, train_loss: 0.0008\n",
      "167/484, train_loss: 0.0025\n",
      "168/484, train_loss: 1.3861\n",
      "169/484, train_loss: 0.2443\n",
      "170/484, train_loss: 0.0016\n",
      "171/484, train_loss: 0.0036\n",
      "172/484, train_loss: 0.0049\n",
      "173/484, train_loss: 0.0026\n",
      "174/484, train_loss: 0.0042\n",
      "175/484, train_loss: 0.0022\n",
      "176/484, train_loss: 0.0087\n",
      "177/484, train_loss: 0.0028\n",
      "178/484, train_loss: 0.0024\n",
      "179/484, train_loss: 0.0023\n",
      "180/484, train_loss: 0.0010\n",
      "181/484, train_loss: 0.1159\n",
      "182/484, train_loss: 0.0022\n",
      "183/484, train_loss: 0.2684\n",
      "184/484, train_loss: 0.0008\n",
      "185/484, train_loss: 0.0055\n",
      "186/484, train_loss: 0.0038\n",
      "187/484, train_loss: 0.0027\n",
      "188/484, train_loss: 0.0021\n",
      "189/484, train_loss: 0.0429\n",
      "190/484, train_loss: 0.0762\n",
      "191/484, train_loss: 0.0015\n",
      "192/484, train_loss: 0.1807\n",
      "193/484, train_loss: 0.0099\n",
      "194/484, train_loss: 0.0064\n",
      "195/484, train_loss: 0.0042\n",
      "196/484, train_loss: 2.3205\n",
      "197/484, train_loss: 0.0258\n",
      "198/484, train_loss: 0.0028\n",
      "199/484, train_loss: 1.5209\n",
      "200/484, train_loss: 0.0120\n",
      "201/484, train_loss: 0.0012\n",
      "202/484, train_loss: 0.0020\n",
      "203/484, train_loss: 0.0059\n",
      "204/484, train_loss: 0.0554\n",
      "205/484, train_loss: 0.0028\n",
      "206/484, train_loss: 0.9199\n",
      "207/484, train_loss: 0.0021\n",
      "208/484, train_loss: 0.0105\n",
      "209/484, train_loss: 0.1347\n",
      "210/484, train_loss: 0.0071\n",
      "211/484, train_loss: 0.0028\n",
      "212/484, train_loss: 0.0051\n",
      "213/484, train_loss: 0.0046\n",
      "214/484, train_loss: 0.0751\n",
      "215/484, train_loss: 0.0019\n",
      "216/484, train_loss: 0.0031\n",
      "217/484, train_loss: 0.0088\n",
      "218/484, train_loss: 0.0012\n",
      "219/484, train_loss: 0.4159\n",
      "220/484, train_loss: 0.0017\n",
      "221/484, train_loss: 0.0030\n",
      "222/484, train_loss: 0.0038\n",
      "223/484, train_loss: 0.0018\n",
      "224/484, train_loss: 0.7419\n",
      "225/484, train_loss: 0.0022\n",
      "226/484, train_loss: 0.0112\n",
      "227/484, train_loss: 0.0272\n",
      "228/484, train_loss: 0.0011\n",
      "229/484, train_loss: 0.0011\n",
      "230/484, train_loss: 0.4192\n",
      "231/484, train_loss: 0.0026\n",
      "232/484, train_loss: 0.0135\n",
      "233/484, train_loss: 0.0015\n",
      "234/484, train_loss: 0.2926\n",
      "235/484, train_loss: 1.4848\n",
      "236/484, train_loss: 0.0056\n",
      "237/484, train_loss: 0.0204\n",
      "238/484, train_loss: 0.8471\n",
      "239/484, train_loss: 0.0015\n",
      "240/484, train_loss: 0.0210\n",
      "241/484, train_loss: 0.0053\n",
      "242/484, train_loss: 0.5996\n",
      "243/484, train_loss: 0.0021\n",
      "244/484, train_loss: 0.0030\n",
      "245/484, train_loss: 0.0361\n",
      "246/484, train_loss: 0.0086\n",
      "247/484, train_loss: 0.0312\n",
      "248/484, train_loss: 0.1041\n",
      "249/484, train_loss: 0.0135\n",
      "250/484, train_loss: 0.0075\n",
      "251/484, train_loss: 1.0185\n",
      "252/484, train_loss: 0.0033\n",
      "253/484, train_loss: 0.0016\n",
      "254/484, train_loss: 0.3063\n",
      "255/484, train_loss: 2.3578\n",
      "256/484, train_loss: 0.0060\n",
      "257/484, train_loss: 0.0088\n",
      "258/484, train_loss: 0.1607\n",
      "259/484, train_loss: 0.0101\n",
      "260/484, train_loss: 0.5196\n",
      "261/484, train_loss: 0.0054\n",
      "262/484, train_loss: 0.0037\n",
      "263/484, train_loss: 0.0036\n",
      "264/484, train_loss: 0.0062\n",
      "265/484, train_loss: 0.0032\n",
      "266/484, train_loss: 0.0107\n",
      "267/484, train_loss: 1.2042\n",
      "268/484, train_loss: 0.0391\n",
      "269/484, train_loss: 0.0032\n",
      "270/484, train_loss: 0.8414\n",
      "271/484, train_loss: 0.0041\n",
      "272/484, train_loss: 0.0020\n",
      "273/484, train_loss: 0.0043\n",
      "274/484, train_loss: 0.0034\n",
      "275/484, train_loss: 0.0242\n",
      "276/484, train_loss: 0.0073\n",
      "277/484, train_loss: 0.0063\n",
      "278/484, train_loss: 0.0059\n",
      "279/484, train_loss: 0.0095\n",
      "280/484, train_loss: 0.0074\n",
      "281/484, train_loss: 0.0088\n",
      "282/484, train_loss: 0.0144\n",
      "283/484, train_loss: 0.0021\n",
      "284/484, train_loss: 0.0021\n",
      "285/484, train_loss: 0.3361\n",
      "286/484, train_loss: 0.0052\n",
      "287/484, train_loss: 0.0052\n",
      "288/484, train_loss: 0.5929\n",
      "289/484, train_loss: 0.0125\n",
      "290/484, train_loss: 0.3202\n",
      "291/484, train_loss: 0.0020\n",
      "292/484, train_loss: 0.0031\n",
      "293/484, train_loss: 0.0119\n",
      "294/484, train_loss: 0.0296\n",
      "295/484, train_loss: 0.9722\n",
      "296/484, train_loss: 0.0049\n",
      "297/484, train_loss: 0.0015\n",
      "298/484, train_loss: 0.0081\n",
      "299/484, train_loss: 0.0449\n",
      "300/484, train_loss: 0.6302\n",
      "301/484, train_loss: 0.0201\n",
      "302/484, train_loss: 1.1168\n",
      "303/484, train_loss: 0.0233\n",
      "304/484, train_loss: 0.8362\n",
      "305/484, train_loss: 0.0029\n",
      "306/484, train_loss: 0.0636\n",
      "307/484, train_loss: 0.1223\n",
      "308/484, train_loss: 0.0013\n",
      "309/484, train_loss: 0.0017\n",
      "310/484, train_loss: 0.0026\n",
      "311/484, train_loss: 0.0169\n",
      "312/484, train_loss: 2.0374\n",
      "313/484, train_loss: 1.0436\n",
      "314/484, train_loss: 0.0030\n",
      "315/484, train_loss: 0.0051\n",
      "316/484, train_loss: 0.4344\n",
      "317/484, train_loss: 0.0059\n",
      "318/484, train_loss: 0.0034\n",
      "319/484, train_loss: 0.0572\n",
      "320/484, train_loss: 0.7857\n",
      "321/484, train_loss: 0.3175\n",
      "322/484, train_loss: 0.0151\n",
      "323/484, train_loss: 0.0173\n",
      "324/484, train_loss: 0.3226\n",
      "325/484, train_loss: 0.0044\n",
      "326/484, train_loss: 0.0034\n",
      "327/484, train_loss: 0.0094\n",
      "328/484, train_loss: 0.0044\n",
      "329/484, train_loss: 0.0361\n",
      "330/484, train_loss: 0.0086\n",
      "331/484, train_loss: 0.0514\n",
      "332/484, train_loss: 0.0086\n",
      "333/484, train_loss: 0.0328\n",
      "334/484, train_loss: 0.0022\n",
      "335/484, train_loss: 0.0143\n",
      "336/484, train_loss: 0.0045\n",
      "337/484, train_loss: 0.0274\n",
      "338/484, train_loss: 0.0590\n",
      "339/484, train_loss: 0.4747\n",
      "340/484, train_loss: 0.2335\n",
      "341/484, train_loss: 0.0019\n",
      "342/484, train_loss: 0.0032\n",
      "343/484, train_loss: 0.0152\n",
      "344/484, train_loss: 0.0021\n",
      "345/484, train_loss: 0.0052\n",
      "346/484, train_loss: 0.0112\n",
      "347/484, train_loss: 0.0306\n",
      "348/484, train_loss: 0.0612\n",
      "349/484, train_loss: 2.3080\n",
      "350/484, train_loss: 0.0077\n",
      "351/484, train_loss: 0.0425\n",
      "352/484, train_loss: 0.0049\n",
      "353/484, train_loss: 0.0027\n",
      "354/484, train_loss: 0.0017\n",
      "355/484, train_loss: 0.3646\n",
      "356/484, train_loss: 0.0665\n",
      "357/484, train_loss: 0.0088\n",
      "358/484, train_loss: 0.0084\n",
      "359/484, train_loss: 0.0019\n",
      "360/484, train_loss: 0.0036\n",
      "361/484, train_loss: 0.0546\n",
      "362/484, train_loss: 0.0080\n",
      "363/484, train_loss: 0.0196\n",
      "364/484, train_loss: 0.0190\n",
      "365/484, train_loss: 0.0099\n",
      "366/484, train_loss: 0.0070\n",
      "367/484, train_loss: 0.0126\n",
      "368/484, train_loss: 0.2967\n",
      "369/484, train_loss: 0.0555\n",
      "370/484, train_loss: 0.0826\n",
      "371/484, train_loss: 0.0128\n",
      "372/484, train_loss: 1.7257\n",
      "373/484, train_loss: 0.0124\n",
      "374/484, train_loss: 0.0083\n",
      "375/484, train_loss: 0.0882\n",
      "376/484, train_loss: 0.9471\n",
      "377/484, train_loss: 0.0100\n",
      "378/484, train_loss: 0.0080\n",
      "379/484, train_loss: 1.4260\n",
      "380/484, train_loss: 0.0069\n",
      "381/484, train_loss: 0.0066\n",
      "382/484, train_loss: 0.4678\n",
      "383/484, train_loss: 0.0491\n",
      "384/484, train_loss: 0.0048\n",
      "385/484, train_loss: 0.0324\n",
      "386/484, train_loss: 0.0052\n",
      "387/484, train_loss: 0.0045\n",
      "388/484, train_loss: 0.0054\n",
      "389/484, train_loss: 0.8109\n",
      "390/484, train_loss: 0.0073\n",
      "391/484, train_loss: 0.0089\n",
      "392/484, train_loss: 0.0790\n",
      "393/484, train_loss: 0.0489\n",
      "394/484, train_loss: 0.0525\n",
      "395/484, train_loss: 0.0188\n",
      "396/484, train_loss: 0.0027\n",
      "397/484, train_loss: 0.0051\n",
      "398/484, train_loss: 0.6230\n",
      "399/484, train_loss: 0.0061\n",
      "400/484, train_loss: 0.1558\n",
      "401/484, train_loss: 0.0094\n",
      "402/484, train_loss: 0.6965\n",
      "403/484, train_loss: 0.5717\n",
      "404/484, train_loss: 0.0564\n",
      "405/484, train_loss: 0.0044\n",
      "406/484, train_loss: 0.0691\n",
      "407/484, train_loss: 0.2170\n",
      "408/484, train_loss: 0.0106\n",
      "409/484, train_loss: 0.0109\n",
      "410/484, train_loss: 0.2063\n",
      "411/484, train_loss: 0.9399\n",
      "412/484, train_loss: 0.0292\n",
      "413/484, train_loss: 1.7857\n",
      "414/484, train_loss: 0.0020\n",
      "415/484, train_loss: 0.1803\n",
      "416/484, train_loss: 0.0035\n",
      "417/484, train_loss: 0.0090\n",
      "418/484, train_loss: 0.0068\n",
      "419/484, train_loss: 0.0037\n",
      "420/484, train_loss: 0.0195\n",
      "421/484, train_loss: 0.0044\n",
      "422/484, train_loss: 0.0983\n",
      "423/484, train_loss: 0.0065\n",
      "424/484, train_loss: 0.1651\n",
      "425/484, train_loss: 0.0194\n",
      "426/484, train_loss: 0.0086\n",
      "427/484, train_loss: 0.0761\n",
      "428/484, train_loss: 0.9427\n",
      "429/484, train_loss: 0.4010\n",
      "430/484, train_loss: 0.0052\n",
      "431/484, train_loss: 0.0036\n",
      "432/484, train_loss: 0.3096\n",
      "433/484, train_loss: 0.2210\n",
      "434/484, train_loss: 0.0025\n",
      "435/484, train_loss: 0.0038\n",
      "436/484, train_loss: 0.0070\n",
      "437/484, train_loss: 0.0055\n",
      "438/484, train_loss: 0.0014\n",
      "439/484, train_loss: 0.3718\n",
      "440/484, train_loss: 0.0051\n",
      "441/484, train_loss: 0.8884\n",
      "442/484, train_loss: 0.4091\n",
      "443/484, train_loss: 0.0042\n",
      "444/484, train_loss: 0.0062\n",
      "445/484, train_loss: 0.0296\n",
      "446/484, train_loss: 0.0101\n",
      "447/484, train_loss: 0.0342\n",
      "448/484, train_loss: 0.0021\n",
      "449/484, train_loss: 0.0267\n",
      "450/484, train_loss: 0.5641\n",
      "451/484, train_loss: 0.0049\n",
      "452/484, train_loss: 0.0072\n",
      "453/484, train_loss: 0.0404\n",
      "454/484, train_loss: 0.3110\n",
      "455/484, train_loss: 0.0089\n",
      "456/484, train_loss: 0.0028\n",
      "457/484, train_loss: 0.0085\n",
      "458/484, train_loss: 0.0026\n",
      "459/484, train_loss: 0.0191\n",
      "460/484, train_loss: 0.0126\n",
      "461/484, train_loss: 0.0160\n",
      "462/484, train_loss: 0.0214\n",
      "463/484, train_loss: 0.0120\n",
      "464/484, train_loss: 0.0128\n",
      "465/484, train_loss: 0.6499\n",
      "466/484, train_loss: 0.0229\n",
      "467/484, train_loss: 0.0036\n",
      "468/484, train_loss: 0.0029\n",
      "469/484, train_loss: 0.0316\n",
      "470/484, train_loss: 0.0174\n",
      "471/484, train_loss: 0.0027\n",
      "472/484, train_loss: 0.0122\n",
      "473/484, train_loss: 0.0034\n",
      "474/484, train_loss: 0.0014\n",
      "475/484, train_loss: 0.0314\n",
      "476/484, train_loss: 0.0071\n",
      "477/484, train_loss: 0.0048\n",
      "478/484, train_loss: 0.2514\n",
      "479/484, train_loss: 0.7553\n",
      "480/484, train_loss: 0.0179\n",
      "481/484, train_loss: 1.4197\n",
      "482/484, train_loss: 0.0020\n",
      "483/484, train_loss: 0.0055\n",
      "484/484, train_loss: 0.0018\n",
      "epoch 15 average loss: 0.1465\n",
      "----------\n",
      "epoch 16/20\n",
      "1/484, train_loss: 0.4578\n",
      "2/484, train_loss: 0.0338\n",
      "3/484, train_loss: 0.0028\n",
      "4/484, train_loss: 0.2195\n",
      "5/484, train_loss: 0.0185\n",
      "6/484, train_loss: 0.1580\n",
      "7/484, train_loss: 0.0048\n",
      "8/484, train_loss: 0.0457\n",
      "9/484, train_loss: 0.0049\n",
      "10/484, train_loss: 0.1040\n",
      "11/484, train_loss: 0.0109\n",
      "12/484, train_loss: 0.0026\n",
      "13/484, train_loss: 0.0082\n",
      "14/484, train_loss: 0.2655\n",
      "15/484, train_loss: 0.0067\n",
      "16/484, train_loss: 0.0056\n",
      "17/484, train_loss: 0.0025\n",
      "18/484, train_loss: 0.0049\n",
      "19/484, train_loss: 0.0046\n",
      "20/484, train_loss: 0.0055\n",
      "21/484, train_loss: 0.0028\n",
      "22/484, train_loss: 0.5540\n",
      "23/484, train_loss: 0.0165\n",
      "24/484, train_loss: 0.0023\n",
      "25/484, train_loss: 0.0064\n",
      "26/484, train_loss: 0.0152\n",
      "27/484, train_loss: 0.0310\n",
      "28/484, train_loss: 0.0176\n",
      "29/484, train_loss: 0.6210\n",
      "30/484, train_loss: 0.0016\n",
      "31/484, train_loss: 0.0532\n",
      "32/484, train_loss: 0.2559\n",
      "33/484, train_loss: 0.0157\n",
      "34/484, train_loss: 0.0031\n",
      "35/484, train_loss: 0.0126\n",
      "36/484, train_loss: 0.0066\n",
      "37/484, train_loss: 0.0043\n",
      "38/484, train_loss: 0.0038\n",
      "39/484, train_loss: 0.0021\n",
      "40/484, train_loss: 0.0035\n",
      "41/484, train_loss: 0.0132\n",
      "42/484, train_loss: 0.0035\n",
      "43/484, train_loss: 0.2443\n",
      "44/484, train_loss: 0.0781\n",
      "45/484, train_loss: 0.0020\n",
      "46/484, train_loss: 0.0120\n",
      "47/484, train_loss: 0.0066\n",
      "48/484, train_loss: 0.0022\n",
      "49/484, train_loss: 0.0095\n",
      "50/484, train_loss: 0.4150\n",
      "51/484, train_loss: 0.0100\n",
      "52/484, train_loss: 0.0033\n",
      "53/484, train_loss: 0.5372\n",
      "54/484, train_loss: 0.0027\n",
      "55/484, train_loss: 0.0049\n",
      "56/484, train_loss: 0.0053\n",
      "57/484, train_loss: 0.0030\n",
      "58/484, train_loss: 0.0018\n",
      "59/484, train_loss: 0.0134\n",
      "60/484, train_loss: 0.0026\n",
      "61/484, train_loss: 0.1097\n",
      "62/484, train_loss: 0.0044\n",
      "63/484, train_loss: 0.0045\n",
      "64/484, train_loss: 0.3465\n",
      "65/484, train_loss: 0.3537\n",
      "66/484, train_loss: 0.2481\n",
      "67/484, train_loss: 0.0044\n",
      "68/484, train_loss: 0.0034\n",
      "69/484, train_loss: 0.0018\n",
      "70/484, train_loss: 0.0159\n",
      "71/484, train_loss: 0.6376\n",
      "72/484, train_loss: 0.0079\n",
      "73/484, train_loss: 0.0033\n",
      "74/484, train_loss: 0.0093\n",
      "75/484, train_loss: 0.0023\n",
      "76/484, train_loss: 0.0081\n",
      "77/484, train_loss: 0.3328\n",
      "78/484, train_loss: 0.3180\n",
      "79/484, train_loss: 0.0051\n",
      "80/484, train_loss: 0.0033\n",
      "81/484, train_loss: 0.0121\n",
      "82/484, train_loss: 0.0058\n",
      "83/484, train_loss: 0.0024\n",
      "84/484, train_loss: 0.0039\n",
      "85/484, train_loss: 0.4207\n",
      "86/484, train_loss: 0.0095\n",
      "87/484, train_loss: 0.0284\n",
      "88/484, train_loss: 0.0037\n",
      "89/484, train_loss: 0.0028\n",
      "90/484, train_loss: 0.0077\n",
      "91/484, train_loss: 0.0027\n",
      "92/484, train_loss: 0.0057\n",
      "93/484, train_loss: 0.7739\n",
      "94/484, train_loss: 0.0071\n",
      "95/484, train_loss: 0.0071\n",
      "96/484, train_loss: 0.0024\n",
      "97/484, train_loss: 0.0042\n",
      "98/484, train_loss: 1.0542\n",
      "99/484, train_loss: 0.0045\n",
      "100/484, train_loss: 0.0106\n",
      "101/484, train_loss: 0.0027\n",
      "102/484, train_loss: 0.0039\n",
      "103/484, train_loss: 0.0036\n",
      "104/484, train_loss: 0.0021\n",
      "105/484, train_loss: 0.0030\n",
      "106/484, train_loss: 0.0027\n",
      "107/484, train_loss: 0.0044\n",
      "108/484, train_loss: 0.0081\n",
      "109/484, train_loss: 0.2782\n",
      "110/484, train_loss: 0.0081\n",
      "111/484, train_loss: 0.1451\n",
      "112/484, train_loss: 0.0023\n",
      "113/484, train_loss: 0.0060\n",
      "114/484, train_loss: 0.0116\n",
      "115/484, train_loss: 0.0056\n",
      "116/484, train_loss: 0.0025\n",
      "117/484, train_loss: 0.0067\n",
      "118/484, train_loss: 0.0040\n",
      "119/484, train_loss: 0.0060\n",
      "120/484, train_loss: 0.0024\n",
      "121/484, train_loss: 0.3311\n",
      "122/484, train_loss: 0.0323\n",
      "123/484, train_loss: 0.0027\n",
      "124/484, train_loss: 0.0050\n",
      "125/484, train_loss: 0.4613\n",
      "126/484, train_loss: 0.0199\n",
      "127/484, train_loss: 0.0050\n",
      "128/484, train_loss: 0.0030\n",
      "129/484, train_loss: 0.0029\n",
      "130/484, train_loss: 0.0096\n",
      "131/484, train_loss: 0.0179\n",
      "132/484, train_loss: 0.0022\n",
      "133/484, train_loss: 0.0072\n",
      "134/484, train_loss: 0.0017\n",
      "135/484, train_loss: 0.0020\n",
      "136/484, train_loss: 0.0078\n",
      "137/484, train_loss: 0.0040\n",
      "138/484, train_loss: 0.0045\n",
      "139/484, train_loss: 0.1559\n",
      "140/484, train_loss: 0.0037\n",
      "141/484, train_loss: 0.0018\n",
      "142/484, train_loss: 0.2143\n",
      "143/484, train_loss: 0.6876\n",
      "144/484, train_loss: 0.0191\n",
      "145/484, train_loss: 0.0086\n",
      "146/484, train_loss: 0.0044\n",
      "147/484, train_loss: 0.0040\n",
      "148/484, train_loss: 0.0030\n",
      "149/484, train_loss: 0.0022\n",
      "150/484, train_loss: 0.0107\n",
      "151/484, train_loss: 0.0033\n",
      "152/484, train_loss: 0.0016\n",
      "153/484, train_loss: 0.0034\n",
      "154/484, train_loss: 0.4701\n",
      "155/484, train_loss: 0.4070\n",
      "156/484, train_loss: 0.0026\n",
      "157/484, train_loss: 0.0032\n",
      "158/484, train_loss: 0.0078\n",
      "159/484, train_loss: 0.0025\n",
      "160/484, train_loss: 0.0075\n",
      "161/484, train_loss: 0.0359\n",
      "162/484, train_loss: 0.0020\n",
      "163/484, train_loss: 0.0013\n",
      "164/484, train_loss: 0.0024\n",
      "165/484, train_loss: 0.0029\n",
      "166/484, train_loss: 0.0091\n",
      "167/484, train_loss: 0.0083\n",
      "168/484, train_loss: 0.0092\n",
      "169/484, train_loss: 0.0031\n",
      "170/484, train_loss: 0.0024\n",
      "171/484, train_loss: 0.2647\n",
      "172/484, train_loss: 0.0009\n",
      "173/484, train_loss: 0.6003\n",
      "174/484, train_loss: 0.2898\n",
      "175/484, train_loss: 0.0107\n",
      "176/484, train_loss: 0.3756\n",
      "177/484, train_loss: 0.0011\n",
      "178/484, train_loss: 0.0043\n",
      "179/484, train_loss: 0.0117\n",
      "180/484, train_loss: 0.0014\n",
      "181/484, train_loss: 0.7666\n",
      "182/484, train_loss: 0.0117\n",
      "183/484, train_loss: 0.1537\n",
      "184/484, train_loss: 0.0028\n",
      "185/484, train_loss: 0.0060\n",
      "186/484, train_loss: 0.0021\n",
      "187/484, train_loss: 0.0012\n",
      "188/484, train_loss: 0.0010\n",
      "189/484, train_loss: 0.0021\n",
      "190/484, train_loss: 0.0054\n",
      "191/484, train_loss: 0.0026\n",
      "192/484, train_loss: 0.4142\n",
      "193/484, train_loss: 0.0032\n",
      "194/484, train_loss: 0.0038\n",
      "195/484, train_loss: 0.0011\n",
      "196/484, train_loss: 0.0041\n",
      "197/484, train_loss: 0.0065\n",
      "198/484, train_loss: 0.0039\n",
      "199/484, train_loss: 0.0047\n",
      "200/484, train_loss: 0.0036\n",
      "201/484, train_loss: 0.0028\n",
      "202/484, train_loss: 0.0103\n",
      "203/484, train_loss: 0.1781\n",
      "204/484, train_loss: 0.2124\n",
      "205/484, train_loss: 0.0174\n",
      "206/484, train_loss: 0.0027\n",
      "207/484, train_loss: 0.1327\n",
      "208/484, train_loss: 0.0026\n",
      "209/484, train_loss: 0.0019\n",
      "210/484, train_loss: 0.0028\n",
      "211/484, train_loss: 0.0020\n",
      "212/484, train_loss: 0.4686\n",
      "213/484, train_loss: 0.0769\n",
      "214/484, train_loss: 0.0032\n",
      "215/484, train_loss: 0.0017\n",
      "216/484, train_loss: 0.0016\n",
      "217/484, train_loss: 0.0083\n",
      "218/484, train_loss: 0.0042\n",
      "219/484, train_loss: 0.0196\n",
      "220/484, train_loss: 0.0049\n",
      "221/484, train_loss: 0.0022\n",
      "222/484, train_loss: 0.0020\n",
      "223/484, train_loss: 0.4370\n",
      "224/484, train_loss: 0.0012\n",
      "225/484, train_loss: 0.1275\n",
      "226/484, train_loss: 0.0153\n",
      "227/484, train_loss: 0.0030\n",
      "228/484, train_loss: 0.0034\n",
      "229/484, train_loss: 0.0046\n",
      "230/484, train_loss: 0.0025\n",
      "231/484, train_loss: 0.5434\n",
      "232/484, train_loss: 0.0027\n",
      "233/484, train_loss: 0.4172\n",
      "234/484, train_loss: 1.0823\n",
      "235/484, train_loss: 0.0038\n",
      "236/484, train_loss: 0.0012\n",
      "237/484, train_loss: 0.0404\n",
      "238/484, train_loss: 0.9116\n",
      "239/484, train_loss: 0.0057\n",
      "240/484, train_loss: 0.0012\n",
      "241/484, train_loss: 0.0081\n",
      "242/484, train_loss: 0.0184\n",
      "243/484, train_loss: 0.0026\n",
      "244/484, train_loss: 0.7294\n",
      "245/484, train_loss: 0.0010\n",
      "246/484, train_loss: 0.0023\n",
      "247/484, train_loss: 0.0016\n",
      "248/484, train_loss: 0.3740\n",
      "249/484, train_loss: 0.2812\n",
      "250/484, train_loss: 0.0179\n",
      "251/484, train_loss: 0.0182\n",
      "252/484, train_loss: 0.0012\n",
      "253/484, train_loss: 0.0053\n",
      "254/484, train_loss: 0.0009\n",
      "255/484, train_loss: 0.0016\n",
      "256/484, train_loss: 0.0131\n",
      "257/484, train_loss: 0.4581\n",
      "258/484, train_loss: 0.0075\n",
      "259/484, train_loss: 0.0037\n",
      "260/484, train_loss: 0.0059\n",
      "261/484, train_loss: 0.0038\n",
      "262/484, train_loss: 0.0032\n",
      "263/484, train_loss: 0.0020\n",
      "264/484, train_loss: 0.0008\n",
      "265/484, train_loss: 0.0107\n",
      "266/484, train_loss: 0.2899\n",
      "267/484, train_loss: 0.0047\n",
      "268/484, train_loss: 0.0106\n",
      "269/484, train_loss: 0.0037\n",
      "270/484, train_loss: 0.0017\n",
      "271/484, train_loss: 0.0031\n",
      "272/484, train_loss: 0.0022\n",
      "273/484, train_loss: 0.0011\n",
      "274/484, train_loss: 0.0048\n",
      "275/484, train_loss: 1.5882\n",
      "276/484, train_loss: 0.0903\n",
      "277/484, train_loss: 0.0025\n",
      "278/484, train_loss: 0.0037\n",
      "279/484, train_loss: 0.6446\n",
      "280/484, train_loss: 0.0017\n",
      "281/484, train_loss: 0.0145\n",
      "282/484, train_loss: 0.0012\n",
      "283/484, train_loss: 0.6735\n",
      "284/484, train_loss: 0.5589\n",
      "285/484, train_loss: 0.7504\n",
      "286/484, train_loss: 0.0072\n",
      "287/484, train_loss: 0.0045\n",
      "288/484, train_loss: 0.0023\n",
      "289/484, train_loss: 0.0035\n",
      "290/484, train_loss: 0.0042\n",
      "291/484, train_loss: 0.0068\n",
      "292/484, train_loss: 0.0012\n",
      "293/484, train_loss: 0.0017\n",
      "294/484, train_loss: 0.0016\n",
      "295/484, train_loss: 0.0018\n",
      "296/484, train_loss: 0.0032\n",
      "297/484, train_loss: 0.0020\n",
      "298/484, train_loss: 0.0330\n",
      "299/484, train_loss: 0.0040\n",
      "300/484, train_loss: 0.0017\n",
      "301/484, train_loss: 0.0035\n",
      "302/484, train_loss: 0.0023\n",
      "303/484, train_loss: 0.0031\n",
      "304/484, train_loss: 0.0015\n",
      "305/484, train_loss: 0.0010\n",
      "306/484, train_loss: 0.0012\n",
      "307/484, train_loss: 0.0058\n",
      "308/484, train_loss: 0.3355\n",
      "309/484, train_loss: 0.0022\n",
      "310/484, train_loss: 0.0218\n",
      "311/484, train_loss: 0.0014\n",
      "312/484, train_loss: 0.2725\n",
      "313/484, train_loss: 0.1434\n",
      "314/484, train_loss: 0.0031\n",
      "315/484, train_loss: 0.0037\n",
      "316/484, train_loss: 0.0260\n",
      "317/484, train_loss: 0.0017\n",
      "318/484, train_loss: 0.5512\n",
      "319/484, train_loss: 0.0620\n",
      "320/484, train_loss: 0.0042\n",
      "321/484, train_loss: 0.0085\n",
      "322/484, train_loss: 0.0038\n",
      "323/484, train_loss: 0.7069\n",
      "324/484, train_loss: 0.0010\n",
      "325/484, train_loss: 0.0014\n",
      "326/484, train_loss: 0.0045\n",
      "327/484, train_loss: 0.0020\n",
      "328/484, train_loss: 0.1048\n",
      "329/484, train_loss: 0.0143\n",
      "330/484, train_loss: 1.9049\n",
      "331/484, train_loss: 0.0021\n",
      "332/484, train_loss: 0.2554\n",
      "333/484, train_loss: 0.0016\n",
      "334/484, train_loss: 0.0107\n",
      "335/484, train_loss: 0.0036\n",
      "336/484, train_loss: 0.0022\n",
      "337/484, train_loss: 0.0058\n",
      "338/484, train_loss: 0.5830\n",
      "339/484, train_loss: 0.0020\n",
      "340/484, train_loss: 0.0025\n",
      "341/484, train_loss: 0.0014\n",
      "342/484, train_loss: 0.0012\n",
      "343/484, train_loss: 0.0030\n",
      "344/484, train_loss: 0.0019\n",
      "345/484, train_loss: 0.0031\n",
      "346/484, train_loss: 0.0041\n",
      "347/484, train_loss: 0.0052\n",
      "348/484, train_loss: 0.5186\n",
      "349/484, train_loss: 0.0021\n",
      "350/484, train_loss: 0.0024\n",
      "351/484, train_loss: 0.7821\n",
      "352/484, train_loss: 0.0021\n",
      "353/484, train_loss: 0.0030\n",
      "354/484, train_loss: 0.0022\n",
      "355/484, train_loss: 0.2534\n",
      "356/484, train_loss: 0.0053\n",
      "357/484, train_loss: 0.0024\n",
      "358/484, train_loss: 0.0023\n",
      "359/484, train_loss: 0.0047\n",
      "360/484, train_loss: 0.0023\n",
      "361/484, train_loss: 0.7841\n",
      "362/484, train_loss: 0.0079\n",
      "363/484, train_loss: 0.4229\n",
      "364/484, train_loss: 0.0053\n",
      "365/484, train_loss: 0.0097\n",
      "366/484, train_loss: 0.0041\n",
      "367/484, train_loss: 0.0157\n",
      "368/484, train_loss: 0.0022\n",
      "369/484, train_loss: 0.0061\n",
      "370/484, train_loss: 0.0278\n",
      "371/484, train_loss: 0.0052\n",
      "372/484, train_loss: 0.0042\n",
      "373/484, train_loss: 0.0100\n",
      "374/484, train_loss: 0.0061\n",
      "375/484, train_loss: 0.0042\n",
      "376/484, train_loss: 0.0042\n",
      "377/484, train_loss: 0.0023\n",
      "378/484, train_loss: 0.0077\n",
      "379/484, train_loss: 0.0045\n",
      "380/484, train_loss: 0.0063\n",
      "381/484, train_loss: 0.0012\n",
      "382/484, train_loss: 0.0021\n",
      "383/484, train_loss: 0.0016\n",
      "384/484, train_loss: 0.0029\n",
      "385/484, train_loss: 0.0178\n",
      "386/484, train_loss: 0.0027\n",
      "387/484, train_loss: 0.0026\n",
      "388/484, train_loss: 0.0031\n",
      "389/484, train_loss: 0.0042\n",
      "390/484, train_loss: 0.0030\n",
      "391/484, train_loss: 0.0028\n",
      "392/484, train_loss: 0.3918\n",
      "393/484, train_loss: 0.1636\n",
      "394/484, train_loss: 0.3224\n",
      "395/484, train_loss: 0.0017\n",
      "396/484, train_loss: 0.0070\n",
      "397/484, train_loss: 0.4694\n",
      "398/484, train_loss: 0.0018\n",
      "399/484, train_loss: 0.0013\n",
      "400/484, train_loss: 0.0027\n",
      "401/484, train_loss: 0.0026\n",
      "402/484, train_loss: 0.0488\n",
      "403/484, train_loss: 0.0049\n",
      "404/484, train_loss: 0.0047\n",
      "405/484, train_loss: 0.0064\n",
      "406/484, train_loss: 0.0098\n",
      "407/484, train_loss: 0.0017\n",
      "408/484, train_loss: 0.0036\n",
      "409/484, train_loss: 0.0042\n",
      "410/484, train_loss: 0.0057\n",
      "411/484, train_loss: 0.0071\n",
      "412/484, train_loss: 0.0128\n",
      "413/484, train_loss: 0.0028\n",
      "414/484, train_loss: 0.0095\n",
      "415/484, train_loss: 0.3413\n",
      "416/484, train_loss: 0.0139\n",
      "417/484, train_loss: 0.0052\n",
      "418/484, train_loss: 0.0016\n",
      "419/484, train_loss: 0.0634\n",
      "420/484, train_loss: 0.0053\n",
      "421/484, train_loss: 0.5235\n",
      "422/484, train_loss: 0.0030\n",
      "423/484, train_loss: 0.1807\n",
      "424/484, train_loss: 0.0028\n",
      "425/484, train_loss: 0.0154\n",
      "426/484, train_loss: 0.0023\n",
      "427/484, train_loss: 0.0064\n",
      "428/484, train_loss: 0.4164\n",
      "429/484, train_loss: 0.0020\n",
      "430/484, train_loss: 0.0052\n",
      "431/484, train_loss: 2.1478\n",
      "432/484, train_loss: 0.0012\n",
      "433/484, train_loss: 0.0064\n",
      "434/484, train_loss: 0.0028\n",
      "435/484, train_loss: 0.0096\n",
      "436/484, train_loss: 0.0043\n",
      "437/484, train_loss: 0.0029\n",
      "438/484, train_loss: 0.0035\n",
      "439/484, train_loss: 0.0130\n",
      "440/484, train_loss: 0.2603\n",
      "441/484, train_loss: 0.0018\n",
      "442/484, train_loss: 0.0022\n",
      "443/484, train_loss: 0.0046\n",
      "444/484, train_loss: 0.0411\n",
      "445/484, train_loss: 0.2288\n",
      "446/484, train_loss: 0.0031\n",
      "447/484, train_loss: 0.0063\n",
      "448/484, train_loss: 0.0015\n",
      "449/484, train_loss: 0.0027\n",
      "450/484, train_loss: 0.0046\n",
      "451/484, train_loss: 0.0057\n",
      "452/484, train_loss: 0.0026\n",
      "453/484, train_loss: 0.0020\n",
      "454/484, train_loss: 0.5173\n",
      "455/484, train_loss: 0.0029\n",
      "456/484, train_loss: 0.0016\n",
      "457/484, train_loss: 0.0019\n",
      "458/484, train_loss: 0.0040\n",
      "459/484, train_loss: 0.0075\n",
      "460/484, train_loss: 1.3043\n",
      "461/484, train_loss: 0.0013\n",
      "462/484, train_loss: 0.0049\n",
      "463/484, train_loss: 0.4326\n",
      "464/484, train_loss: 0.0058\n",
      "465/484, train_loss: 0.0061\n",
      "466/484, train_loss: 0.0027\n",
      "467/484, train_loss: 0.5194\n",
      "468/484, train_loss: 0.2754\n",
      "469/484, train_loss: 2.3946\n",
      "470/484, train_loss: 0.0056\n",
      "471/484, train_loss: 0.0049\n",
      "472/484, train_loss: 0.0600\n",
      "473/484, train_loss: 0.4306\n",
      "474/484, train_loss: 1.7717\n",
      "475/484, train_loss: 0.0023\n",
      "476/484, train_loss: 0.0196\n",
      "477/484, train_loss: 0.0195\n",
      "478/484, train_loss: 0.0021\n",
      "479/484, train_loss: 0.0026\n",
      "480/484, train_loss: 0.0431\n",
      "481/484, train_loss: 0.0027\n",
      "482/484, train_loss: 0.0152\n",
      "483/484, train_loss: 0.0063\n",
      "484/484, train_loss: 1.4283\n",
      "epoch 16 average loss: 0.1039\n",
      "Current epoch: 16 current accuracy: 0.7121 \n",
      "Best accuracy: 0.8669 at epoch 14\n",
      "----------\n",
      "epoch 17/20\n",
      "1/484, train_loss: 0.8319\n",
      "2/484, train_loss: 0.0055\n",
      "3/484, train_loss: 0.0064\n",
      "4/484, train_loss: 0.0047\n",
      "5/484, train_loss: 0.0036\n",
      "6/484, train_loss: 0.0044\n",
      "7/484, train_loss: 0.0085\n",
      "8/484, train_loss: 0.0036\n",
      "9/484, train_loss: 0.7867\n",
      "10/484, train_loss: 0.2275\n",
      "11/484, train_loss: 0.0018\n",
      "12/484, train_loss: 0.0209\n",
      "13/484, train_loss: 0.0069\n",
      "14/484, train_loss: 0.1954\n",
      "15/484, train_loss: 0.0070\n",
      "16/484, train_loss: 0.0025\n",
      "17/484, train_loss: 0.0032\n",
      "18/484, train_loss: 0.0130\n",
      "19/484, train_loss: 0.7401\n",
      "20/484, train_loss: 2.3259\n",
      "21/484, train_loss: 0.0070\n",
      "22/484, train_loss: 0.0307\n",
      "23/484, train_loss: 0.1177\n",
      "24/484, train_loss: 0.8390\n",
      "25/484, train_loss: 0.0814\n",
      "26/484, train_loss: 0.1877\n",
      "27/484, train_loss: 0.0271\n",
      "28/484, train_loss: 1.5118\n",
      "29/484, train_loss: 0.6649\n",
      "30/484, train_loss: 3.0051\n",
      "31/484, train_loss: 0.0384\n",
      "32/484, train_loss: 0.0051\n",
      "33/484, train_loss: 0.0260\n",
      "34/484, train_loss: 0.5290\n",
      "35/484, train_loss: 0.0058\n",
      "36/484, train_loss: 0.4052\n",
      "37/484, train_loss: 0.0027\n",
      "38/484, train_loss: 0.0056\n",
      "39/484, train_loss: 0.0579\n",
      "40/484, train_loss: 0.2675\n",
      "41/484, train_loss: 0.0077\n",
      "42/484, train_loss: 0.0268\n",
      "43/484, train_loss: 0.1893\n",
      "44/484, train_loss: 1.0146\n",
      "45/484, train_loss: 0.0100\n",
      "46/484, train_loss: 0.2923\n",
      "47/484, train_loss: 0.3270\n",
      "48/484, train_loss: 2.2860\n",
      "49/484, train_loss: 0.0054\n",
      "50/484, train_loss: 0.0527\n",
      "51/484, train_loss: 0.0054\n",
      "52/484, train_loss: 0.0067\n",
      "53/484, train_loss: 0.0023\n",
      "54/484, train_loss: 0.0126\n",
      "55/484, train_loss: 0.0095\n",
      "56/484, train_loss: 0.2305\n",
      "57/484, train_loss: 0.0232\n",
      "58/484, train_loss: 0.0138\n",
      "59/484, train_loss: 0.0057\n",
      "60/484, train_loss: 0.0333\n",
      "61/484, train_loss: 0.0711\n",
      "62/484, train_loss: 0.1179\n",
      "63/484, train_loss: 0.0321\n",
      "64/484, train_loss: 0.0380\n",
      "65/484, train_loss: 0.5015\n",
      "66/484, train_loss: 0.1698\n",
      "67/484, train_loss: 0.0129\n",
      "68/484, train_loss: 0.0172\n",
      "69/484, train_loss: 0.0948\n",
      "70/484, train_loss: 0.0623\n",
      "71/484, train_loss: 0.0041\n",
      "72/484, train_loss: 0.0078\n",
      "73/484, train_loss: 0.0114\n",
      "74/484, train_loss: 0.0171\n",
      "75/484, train_loss: 0.0100\n",
      "76/484, train_loss: 0.0189\n",
      "77/484, train_loss: 0.1847\n",
      "78/484, train_loss: 0.0047\n",
      "79/484, train_loss: 0.0050\n",
      "80/484, train_loss: 0.0028\n",
      "81/484, train_loss: 0.2673\n",
      "82/484, train_loss: 0.0086\n",
      "83/484, train_loss: 1.2492\n",
      "84/484, train_loss: 0.0041\n",
      "85/484, train_loss: 1.1037\n",
      "86/484, train_loss: 0.0148\n",
      "87/484, train_loss: 0.0800\n",
      "88/484, train_loss: 0.0472\n",
      "89/484, train_loss: 0.9977\n",
      "90/484, train_loss: 4.4922\n",
      "91/484, train_loss: 2.8968\n",
      "92/484, train_loss: 0.6839\n",
      "93/484, train_loss: 0.0201\n",
      "94/484, train_loss: 1.6652\n",
      "95/484, train_loss: 0.0068\n",
      "96/484, train_loss: 0.2447\n",
      "97/484, train_loss: 0.0051\n",
      "98/484, train_loss: 0.0108\n",
      "99/484, train_loss: 0.0321\n",
      "100/484, train_loss: 0.0278\n",
      "101/484, train_loss: 0.0431\n",
      "102/484, train_loss: 2.3493\n",
      "103/484, train_loss: 0.0813\n",
      "104/484, train_loss: 0.1635\n",
      "105/484, train_loss: 0.0113\n",
      "106/484, train_loss: 1.1770\n",
      "107/484, train_loss: 0.0190\n",
      "108/484, train_loss: 0.2435\n",
      "109/484, train_loss: 0.0101\n",
      "110/484, train_loss: 0.0446\n",
      "111/484, train_loss: 0.0205\n",
      "112/484, train_loss: 1.1864\n",
      "113/484, train_loss: 1.4964\n",
      "114/484, train_loss: 0.1364\n",
      "115/484, train_loss: 0.0176\n",
      "116/484, train_loss: 0.2717\n",
      "117/484, train_loss: 0.0527\n",
      "118/484, train_loss: 0.0544\n",
      "119/484, train_loss: 0.0080\n",
      "120/484, train_loss: 1.0981\n",
      "121/484, train_loss: 0.0171\n",
      "122/484, train_loss: 0.0656\n",
      "123/484, train_loss: 0.0050\n",
      "124/484, train_loss: 0.0817\n",
      "125/484, train_loss: 0.2491\n",
      "126/484, train_loss: 0.6913\n",
      "127/484, train_loss: 0.0246\n",
      "128/484, train_loss: 0.0370\n",
      "129/484, train_loss: 0.0231\n",
      "130/484, train_loss: 0.1102\n",
      "131/484, train_loss: 0.1805\n",
      "132/484, train_loss: 0.3191\n",
      "133/484, train_loss: 0.0210\n",
      "134/484, train_loss: 0.0355\n",
      "135/484, train_loss: 0.4066\n",
      "136/484, train_loss: 0.0293\n",
      "137/484, train_loss: 0.0128\n",
      "138/484, train_loss: 0.2151\n",
      "139/484, train_loss: 0.0219\n",
      "140/484, train_loss: 0.6748\n",
      "141/484, train_loss: 0.0059\n",
      "142/484, train_loss: 0.2159\n",
      "143/484, train_loss: 0.0196\n",
      "144/484, train_loss: 1.4011\n",
      "145/484, train_loss: 0.0837\n",
      "146/484, train_loss: 0.0849\n",
      "147/484, train_loss: 0.1589\n",
      "148/484, train_loss: 0.1766\n",
      "149/484, train_loss: 0.0704\n",
      "150/484, train_loss: 0.9261\n",
      "151/484, train_loss: 0.2546\n",
      "152/484, train_loss: 0.3084\n",
      "153/484, train_loss: 0.0047\n",
      "154/484, train_loss: 0.0628\n",
      "155/484, train_loss: 1.0150\n",
      "156/484, train_loss: 0.0285\n",
      "157/484, train_loss: 0.0211\n",
      "158/484, train_loss: 0.8486\n",
      "159/484, train_loss: 0.0412\n",
      "160/484, train_loss: 0.0092\n",
      "161/484, train_loss: 0.0233\n",
      "162/484, train_loss: 0.0046\n",
      "163/484, train_loss: 0.8760\n",
      "164/484, train_loss: 0.3333\n",
      "165/484, train_loss: 0.2793\n",
      "166/484, train_loss: 0.5776\n",
      "167/484, train_loss: 0.0539\n",
      "168/484, train_loss: 0.5280\n",
      "169/484, train_loss: 0.3850\n",
      "170/484, train_loss: 0.2944\n",
      "171/484, train_loss: 0.8062\n",
      "172/484, train_loss: 0.0067\n",
      "173/484, train_loss: 0.0469\n",
      "174/484, train_loss: 0.0250\n",
      "175/484, train_loss: 0.0145\n",
      "176/484, train_loss: 0.0072\n",
      "177/484, train_loss: 0.0097\n",
      "178/484, train_loss: 0.0371\n",
      "179/484, train_loss: 0.3792\n",
      "180/484, train_loss: 0.0905\n",
      "181/484, train_loss: 0.0106\n",
      "182/484, train_loss: 0.0405\n",
      "183/484, train_loss: 0.7904\n",
      "184/484, train_loss: 0.0065\n",
      "185/484, train_loss: 0.0901\n",
      "186/484, train_loss: 0.0146\n",
      "187/484, train_loss: 0.0202\n",
      "188/484, train_loss: 0.2940\n",
      "189/484, train_loss: 0.0315\n",
      "190/484, train_loss: 0.0140\n",
      "191/484, train_loss: 0.0105\n",
      "192/484, train_loss: 0.4090\n",
      "193/484, train_loss: 0.0213\n",
      "194/484, train_loss: 0.5974\n",
      "195/484, train_loss: 0.0236\n",
      "196/484, train_loss: 0.0227\n",
      "197/484, train_loss: 0.0094\n",
      "198/484, train_loss: 0.2007\n",
      "199/484, train_loss: 0.0084\n",
      "200/484, train_loss: 0.0062\n",
      "201/484, train_loss: 0.2017\n",
      "202/484, train_loss: 0.0182\n",
      "203/484, train_loss: 0.1539\n",
      "204/484, train_loss: 0.0076\n",
      "205/484, train_loss: 0.0049\n",
      "206/484, train_loss: 0.2264\n",
      "207/484, train_loss: 0.0031\n",
      "208/484, train_loss: 0.0404\n",
      "209/484, train_loss: 0.0230\n",
      "210/484, train_loss: 0.0055\n",
      "211/484, train_loss: 0.2389\n",
      "212/484, train_loss: 0.0189\n",
      "213/484, train_loss: 0.0135\n",
      "214/484, train_loss: 0.0793\n",
      "215/484, train_loss: 0.0480\n",
      "216/484, train_loss: 0.0242\n",
      "217/484, train_loss: 0.0043\n",
      "218/484, train_loss: 0.3530\n",
      "219/484, train_loss: 0.1490\n",
      "220/484, train_loss: 0.0137\n",
      "221/484, train_loss: 0.0072\n",
      "222/484, train_loss: 0.0032\n",
      "223/484, train_loss: 0.1053\n",
      "224/484, train_loss: 0.0120\n",
      "225/484, train_loss: 0.0215\n",
      "226/484, train_loss: 0.0952\n",
      "227/484, train_loss: 0.6348\n",
      "228/484, train_loss: 0.5022\n",
      "229/484, train_loss: 0.8451\n",
      "230/484, train_loss: 0.1756\n",
      "231/484, train_loss: 0.0068\n",
      "232/484, train_loss: 0.0078\n",
      "233/484, train_loss: 0.0100\n",
      "234/484, train_loss: 0.0151\n",
      "235/484, train_loss: 0.0097\n",
      "236/484, train_loss: 0.0633\n",
      "237/484, train_loss: 0.0415\n",
      "238/484, train_loss: 0.3001\n",
      "239/484, train_loss: 0.6503\n",
      "240/484, train_loss: 0.0701\n",
      "241/484, train_loss: 0.0063\n",
      "242/484, train_loss: 0.0219\n",
      "243/484, train_loss: 0.0755\n",
      "244/484, train_loss: 0.0206\n",
      "245/484, train_loss: 0.0143\n",
      "246/484, train_loss: 0.0598\n",
      "247/484, train_loss: 0.0228\n",
      "248/484, train_loss: 0.1050\n",
      "249/484, train_loss: 0.0372\n",
      "250/484, train_loss: 0.0187\n",
      "251/484, train_loss: 0.0120\n",
      "252/484, train_loss: 0.0283\n",
      "253/484, train_loss: 1.1823\n",
      "254/484, train_loss: 0.5359\n",
      "255/484, train_loss: 0.0166\n",
      "256/484, train_loss: 0.0040\n",
      "257/484, train_loss: 0.0083\n",
      "258/484, train_loss: 0.0129\n",
      "259/484, train_loss: 0.0034\n",
      "260/484, train_loss: 0.2001\n",
      "261/484, train_loss: 0.0505\n",
      "262/484, train_loss: 0.0713\n",
      "263/484, train_loss: 0.0196\n",
      "264/484, train_loss: 0.1880\n",
      "265/484, train_loss: 0.0082\n",
      "266/484, train_loss: 0.0074\n",
      "267/484, train_loss: 0.0047\n",
      "268/484, train_loss: 0.0074\n",
      "269/484, train_loss: 0.5410\n",
      "270/484, train_loss: 0.0058\n",
      "271/484, train_loss: 0.0497\n",
      "272/484, train_loss: 0.0070\n",
      "273/484, train_loss: 0.0142\n",
      "274/484, train_loss: 0.0340\n",
      "275/484, train_loss: 0.5761\n",
      "276/484, train_loss: 0.0145\n",
      "277/484, train_loss: 0.0046\n",
      "278/484, train_loss: 0.0047\n",
      "279/484, train_loss: 0.0038\n",
      "280/484, train_loss: 0.0077\n",
      "281/484, train_loss: 0.0209\n",
      "282/484, train_loss: 0.4715\n",
      "283/484, train_loss: 0.0112\n",
      "284/484, train_loss: 0.0033\n",
      "285/484, train_loss: 0.0064\n",
      "286/484, train_loss: 0.0022\n",
      "287/484, train_loss: 0.0030\n",
      "288/484, train_loss: 0.0191\n",
      "289/484, train_loss: 0.0105\n",
      "290/484, train_loss: 0.0125\n",
      "291/484, train_loss: 0.5341\n",
      "292/484, train_loss: 0.0087\n",
      "293/484, train_loss: 0.0137\n",
      "294/484, train_loss: 0.0030\n",
      "295/484, train_loss: 0.0123\n",
      "296/484, train_loss: 0.0141\n",
      "297/484, train_loss: 0.0080\n",
      "298/484, train_loss: 0.4209\n",
      "299/484, train_loss: 0.0055\n",
      "300/484, train_loss: 0.6349\n",
      "301/484, train_loss: 0.7420\n",
      "302/484, train_loss: 0.0045\n",
      "303/484, train_loss: 0.4646\n",
      "304/484, train_loss: 0.0069\n",
      "305/484, train_loss: 0.0203\n",
      "306/484, train_loss: 0.0036\n",
      "307/484, train_loss: 0.2895\n",
      "308/484, train_loss: 0.0156\n",
      "309/484, train_loss: 0.6659\n",
      "310/484, train_loss: 0.0060\n",
      "311/484, train_loss: 0.1126\n",
      "312/484, train_loss: 0.0091\n",
      "313/484, train_loss: 0.0232\n",
      "314/484, train_loss: 0.5071\n",
      "315/484, train_loss: 0.8437\n",
      "316/484, train_loss: 0.0420\n",
      "317/484, train_loss: 0.0033\n",
      "318/484, train_loss: 0.0053\n",
      "319/484, train_loss: 0.0400\n",
      "320/484, train_loss: 0.4873\n",
      "321/484, train_loss: 0.5177\n",
      "322/484, train_loss: 0.0051\n",
      "323/484, train_loss: 0.0038\n",
      "324/484, train_loss: 0.0042\n",
      "325/484, train_loss: 0.0064\n",
      "326/484, train_loss: 0.0041\n",
      "327/484, train_loss: 0.0276\n",
      "328/484, train_loss: 0.6257\n",
      "329/484, train_loss: 0.0044\n",
      "330/484, train_loss: 0.0073\n",
      "331/484, train_loss: 0.0133\n",
      "332/484, train_loss: 0.0161\n",
      "333/484, train_loss: 0.1010\n",
      "334/484, train_loss: 0.0094\n",
      "335/484, train_loss: 0.7842\n",
      "336/484, train_loss: 0.0131\n",
      "337/484, train_loss: 0.1674\n",
      "338/484, train_loss: 0.0312\n",
      "339/484, train_loss: 0.0051\n",
      "340/484, train_loss: 0.2969\n",
      "341/484, train_loss: 0.0046\n",
      "342/484, train_loss: 0.0028\n",
      "343/484, train_loss: 0.4146\n",
      "344/484, train_loss: 0.0055\n",
      "345/484, train_loss: 0.4108\n",
      "346/484, train_loss: 0.0145\n",
      "347/484, train_loss: 0.4491\n",
      "348/484, train_loss: 0.0038\n",
      "349/484, train_loss: 0.0111\n",
      "350/484, train_loss: 0.0044\n",
      "351/484, train_loss: 0.0112\n",
      "352/484, train_loss: 0.0023\n",
      "353/484, train_loss: 0.1689\n",
      "354/484, train_loss: 0.0163\n",
      "355/484, train_loss: 0.0045\n",
      "356/484, train_loss: 0.3016\n",
      "357/484, train_loss: 0.0950\n",
      "358/484, train_loss: 0.0200\n",
      "359/484, train_loss: 0.3518\n",
      "360/484, train_loss: 0.0053\n",
      "361/484, train_loss: 0.0956\n",
      "362/484, train_loss: 0.0028\n",
      "363/484, train_loss: 0.0075\n",
      "364/484, train_loss: 0.0045\n",
      "365/484, train_loss: 0.0024\n",
      "366/484, train_loss: 0.0214\n",
      "367/484, train_loss: 0.0060\n",
      "368/484, train_loss: 0.8113\n",
      "369/484, train_loss: 0.0049\n",
      "370/484, train_loss: 0.0941\n",
      "371/484, train_loss: 0.0082\n",
      "372/484, train_loss: 1.6394\n",
      "373/484, train_loss: 0.3851\n",
      "374/484, train_loss: 0.0670\n",
      "375/484, train_loss: 0.0084\n",
      "376/484, train_loss: 0.0139\n",
      "377/484, train_loss: 0.0082\n",
      "378/484, train_loss: 0.0145\n",
      "379/484, train_loss: 0.0050\n",
      "380/484, train_loss: 0.0038\n",
      "381/484, train_loss: 0.0166\n",
      "382/484, train_loss: 1.0732\n",
      "383/484, train_loss: 0.0089\n",
      "384/484, train_loss: 0.0136\n",
      "385/484, train_loss: 0.0256\n",
      "386/484, train_loss: 0.0145\n",
      "387/484, train_loss: 0.0027\n",
      "388/484, train_loss: 0.0021\n",
      "389/484, train_loss: 0.0057\n",
      "390/484, train_loss: 0.0031\n",
      "391/484, train_loss: 0.0353\n",
      "392/484, train_loss: 0.0026\n",
      "393/484, train_loss: 0.6371\n",
      "394/484, train_loss: 0.0033\n",
      "395/484, train_loss: 0.0052\n",
      "396/484, train_loss: 0.0031\n",
      "397/484, train_loss: 0.0155\n",
      "398/484, train_loss: 0.0097\n",
      "399/484, train_loss: 0.0059\n",
      "400/484, train_loss: 0.1003\n",
      "401/484, train_loss: 0.0074\n",
      "402/484, train_loss: 0.0032\n",
      "403/484, train_loss: 0.0228\n",
      "404/484, train_loss: 0.0151\n",
      "405/484, train_loss: 0.0025\n",
      "406/484, train_loss: 0.0014\n",
      "407/484, train_loss: 0.0087\n",
      "408/484, train_loss: 0.0033\n",
      "409/484, train_loss: 0.1642\n",
      "410/484, train_loss: 0.0032\n",
      "411/484, train_loss: 0.0060\n",
      "412/484, train_loss: 0.0044\n",
      "413/484, train_loss: 0.0041\n",
      "414/484, train_loss: 0.0061\n",
      "415/484, train_loss: 0.0026\n",
      "416/484, train_loss: 0.0013\n",
      "417/484, train_loss: 0.1813\n",
      "418/484, train_loss: 0.0022\n",
      "419/484, train_loss: 0.0043\n",
      "420/484, train_loss: 0.0112\n",
      "421/484, train_loss: 0.0076\n",
      "422/484, train_loss: 0.2194\n",
      "423/484, train_loss: 0.0072\n",
      "424/484, train_loss: 0.0014\n",
      "425/484, train_loss: 0.0027\n",
      "426/484, train_loss: 0.0028\n",
      "427/484, train_loss: 0.0045\n",
      "428/484, train_loss: 0.0072\n",
      "429/484, train_loss: 0.0062\n",
      "430/484, train_loss: 0.0053\n",
      "431/484, train_loss: 0.0022\n",
      "432/484, train_loss: 0.0212\n",
      "433/484, train_loss: 0.0159\n",
      "434/484, train_loss: 0.0088\n",
      "435/484, train_loss: 0.0060\n",
      "436/484, train_loss: 0.0564\n",
      "437/484, train_loss: 0.0033\n",
      "438/484, train_loss: 0.0169\n",
      "439/484, train_loss: 0.0243\n",
      "440/484, train_loss: 0.3416\n",
      "441/484, train_loss: 0.0030\n",
      "442/484, train_loss: 0.4433\n",
      "443/484, train_loss: 1.2164\n",
      "444/484, train_loss: 0.0021\n",
      "445/484, train_loss: 0.0075\n",
      "446/484, train_loss: 0.0063\n",
      "447/484, train_loss: 0.0110\n",
      "448/484, train_loss: 1.1639\n",
      "449/484, train_loss: 0.0029\n",
      "450/484, train_loss: 0.0044\n",
      "451/484, train_loss: 0.0092\n",
      "452/484, train_loss: 0.0207\n",
      "453/484, train_loss: 0.0024\n",
      "454/484, train_loss: 0.0017\n",
      "455/484, train_loss: 0.0094\n",
      "456/484, train_loss: 0.7379\n",
      "457/484, train_loss: 0.0039\n",
      "458/484, train_loss: 0.0193\n",
      "459/484, train_loss: 0.0036\n",
      "460/484, train_loss: 0.0025\n",
      "461/484, train_loss: 0.0369\n",
      "462/484, train_loss: 0.0080\n",
      "463/484, train_loss: 0.0045\n",
      "464/484, train_loss: 0.0184\n",
      "465/484, train_loss: 0.0100\n",
      "466/484, train_loss: 0.3198\n",
      "467/484, train_loss: 0.5096\n",
      "468/484, train_loss: 0.0075\n",
      "469/484, train_loss: 0.0025\n",
      "470/484, train_loss: 0.0049\n",
      "471/484, train_loss: 0.0026\n",
      "472/484, train_loss: 0.0096\n",
      "473/484, train_loss: 0.0063\n",
      "474/484, train_loss: 0.0030\n",
      "475/484, train_loss: 0.0151\n",
      "476/484, train_loss: 0.0034\n",
      "477/484, train_loss: 0.0209\n",
      "478/484, train_loss: 0.0032\n",
      "479/484, train_loss: 0.0050\n",
      "480/484, train_loss: 0.0025\n",
      "481/484, train_loss: 0.6745\n",
      "482/484, train_loss: 0.0016\n",
      "483/484, train_loss: 0.0024\n",
      "484/484, train_loss: 0.0015\n",
      "epoch 17 average loss: 0.1861\n",
      "----------\n",
      "epoch 18/20\n",
      "1/484, train_loss: 0.0022\n",
      "2/484, train_loss: 0.0030\n",
      "3/484, train_loss: 0.0040\n",
      "4/484, train_loss: 0.0035\n",
      "5/484, train_loss: 0.0029\n",
      "6/484, train_loss: 0.0215\n",
      "7/484, train_loss: 0.0024\n",
      "8/484, train_loss: 0.0066\n",
      "9/484, train_loss: 0.2421\n",
      "10/484, train_loss: 0.0052\n",
      "11/484, train_loss: 0.0050\n",
      "12/484, train_loss: 0.0134\n",
      "13/484, train_loss: 0.0024\n",
      "14/484, train_loss: 0.0042\n",
      "15/484, train_loss: 0.1673\n",
      "16/484, train_loss: 0.5240\n",
      "17/484, train_loss: 0.0037\n",
      "18/484, train_loss: 0.0022\n",
      "19/484, train_loss: 0.0121\n",
      "20/484, train_loss: 0.6268\n",
      "21/484, train_loss: 0.0111\n",
      "22/484, train_loss: 0.0086\n",
      "23/484, train_loss: 0.0106\n",
      "24/484, train_loss: 0.0117\n",
      "25/484, train_loss: 0.1408\n",
      "26/484, train_loss: 0.0691\n",
      "27/484, train_loss: 0.0046\n",
      "28/484, train_loss: 0.0065\n",
      "29/484, train_loss: 0.0065\n",
      "30/484, train_loss: 0.0024\n",
      "31/484, train_loss: 0.0215\n",
      "32/484, train_loss: 0.0075\n",
      "33/484, train_loss: 0.0041\n",
      "34/484, train_loss: 0.0022\n",
      "35/484, train_loss: 0.0032\n",
      "36/484, train_loss: 0.1931\n",
      "37/484, train_loss: 0.0036\n",
      "38/484, train_loss: 0.0118\n",
      "39/484, train_loss: 0.0040\n",
      "40/484, train_loss: 0.0063\n",
      "41/484, train_loss: 0.0023\n",
      "42/484, train_loss: 0.2839\n",
      "43/484, train_loss: 1.4963\n",
      "44/484, train_loss: 0.0095\n",
      "45/484, train_loss: 0.0039\n",
      "46/484, train_loss: 0.0064\n",
      "47/484, train_loss: 0.0032\n",
      "48/484, train_loss: 0.0027\n",
      "49/484, train_loss: 0.0026\n",
      "50/484, train_loss: 0.0021\n",
      "51/484, train_loss: 0.0039\n",
      "52/484, train_loss: 0.0086\n",
      "53/484, train_loss: 0.0018\n",
      "54/484, train_loss: 0.0047\n",
      "55/484, train_loss: 0.0023\n",
      "56/484, train_loss: 0.0024\n",
      "57/484, train_loss: 0.2917\n",
      "58/484, train_loss: 0.0028\n",
      "59/484, train_loss: 0.6382\n",
      "60/484, train_loss: 0.0024\n",
      "61/484, train_loss: 0.0022\n",
      "62/484, train_loss: 0.0078\n",
      "63/484, train_loss: 0.1468\n",
      "64/484, train_loss: 0.0019\n",
      "65/484, train_loss: 0.0021\n",
      "66/484, train_loss: 0.2012\n",
      "67/484, train_loss: 0.0200\n",
      "68/484, train_loss: 0.0024\n",
      "69/484, train_loss: 0.1313\n",
      "70/484, train_loss: 0.0024\n",
      "71/484, train_loss: 0.8155\n",
      "72/484, train_loss: 0.0047\n",
      "73/484, train_loss: 0.0017\n",
      "74/484, train_loss: 0.2992\n",
      "75/484, train_loss: 0.0055\n",
      "76/484, train_loss: 0.0079\n",
      "77/484, train_loss: 0.5673\n",
      "78/484, train_loss: 0.0030\n",
      "79/484, train_loss: 0.0676\n",
      "80/484, train_loss: 0.0027\n",
      "81/484, train_loss: 0.0029\n",
      "82/484, train_loss: 0.0081\n",
      "83/484, train_loss: 0.0015\n",
      "84/484, train_loss: 0.0099\n",
      "85/484, train_loss: 0.0016\n",
      "86/484, train_loss: 0.2291\n",
      "87/484, train_loss: 0.0028\n",
      "88/484, train_loss: 0.0056\n",
      "89/484, train_loss: 0.0160\n",
      "90/484, train_loss: 0.0056\n",
      "91/484, train_loss: 0.0122\n",
      "92/484, train_loss: 0.0069\n",
      "93/484, train_loss: 0.4793\n",
      "94/484, train_loss: 0.0027\n",
      "95/484, train_loss: 0.3984\n",
      "96/484, train_loss: 0.0238\n",
      "97/484, train_loss: 0.0031\n",
      "98/484, train_loss: 0.0050\n",
      "99/484, train_loss: 0.0027\n",
      "100/484, train_loss: 0.0037\n",
      "101/484, train_loss: 0.0042\n",
      "102/484, train_loss: 0.0043\n",
      "103/484, train_loss: 0.0018\n",
      "104/484, train_loss: 0.0163\n",
      "105/484, train_loss: 0.0046\n",
      "106/484, train_loss: 0.0017\n",
      "107/484, train_loss: 0.0023\n",
      "108/484, train_loss: 0.0016\n",
      "109/484, train_loss: 0.0226\n",
      "110/484, train_loss: 0.0106\n",
      "111/484, train_loss: 0.0054\n",
      "112/484, train_loss: 0.3018\n",
      "113/484, train_loss: 0.0016\n",
      "114/484, train_loss: 0.0019\n",
      "115/484, train_loss: 0.0039\n",
      "116/484, train_loss: 0.0013\n",
      "117/484, train_loss: 0.0042\n",
      "118/484, train_loss: 0.0020\n",
      "119/484, train_loss: 0.0046\n",
      "120/484, train_loss: 0.0039\n",
      "121/484, train_loss: 0.0033\n",
      "122/484, train_loss: 0.0017\n",
      "123/484, train_loss: 0.0034\n",
      "124/484, train_loss: 0.0022\n",
      "125/484, train_loss: 0.0035\n",
      "126/484, train_loss: 0.0013\n",
      "127/484, train_loss: 0.0030\n",
      "128/484, train_loss: 0.4632\n",
      "129/484, train_loss: 0.0029\n",
      "130/484, train_loss: 0.0015\n",
      "131/484, train_loss: 0.0870\n",
      "132/484, train_loss: 0.0018\n",
      "133/484, train_loss: 0.0024\n",
      "134/484, train_loss: 0.0035\n",
      "135/484, train_loss: 0.0030\n",
      "136/484, train_loss: 0.0057\n",
      "137/484, train_loss: 0.0061\n",
      "138/484, train_loss: 0.0033\n",
      "139/484, train_loss: 0.7763\n",
      "140/484, train_loss: 0.0033\n",
      "141/484, train_loss: 0.0025\n",
      "142/484, train_loss: 0.0064\n",
      "143/484, train_loss: 0.0109\n",
      "144/484, train_loss: 0.0051\n",
      "145/484, train_loss: 0.2684\n",
      "146/484, train_loss: 0.0015\n",
      "147/484, train_loss: 0.0546\n",
      "148/484, train_loss: 0.0048\n",
      "149/484, train_loss: 0.0015\n",
      "150/484, train_loss: 0.2938\n",
      "151/484, train_loss: 0.0199\n",
      "152/484, train_loss: 0.0032\n",
      "153/484, train_loss: 0.0013\n",
      "154/484, train_loss: 0.6012\n",
      "155/484, train_loss: 0.0020\n",
      "156/484, train_loss: 0.0015\n",
      "157/484, train_loss: 0.0023\n",
      "158/484, train_loss: 0.0030\n",
      "159/484, train_loss: 0.0025\n",
      "160/484, train_loss: 0.0013\n",
      "161/484, train_loss: 0.0026\n",
      "162/484, train_loss: 0.6622\n",
      "163/484, train_loss: 0.0023\n",
      "164/484, train_loss: 0.0080\n",
      "165/484, train_loss: 0.0050\n",
      "166/484, train_loss: 0.0019\n",
      "167/484, train_loss: 0.0028\n",
      "168/484, train_loss: 0.0183\n",
      "169/484, train_loss: 0.0035\n",
      "170/484, train_loss: 0.0034\n",
      "171/484, train_loss: 0.0060\n",
      "172/484, train_loss: 0.0121\n",
      "173/484, train_loss: 0.1130\n",
      "174/484, train_loss: 0.0105\n",
      "175/484, train_loss: 0.0019\n",
      "176/484, train_loss: 0.0013\n",
      "177/484, train_loss: 0.0016\n",
      "178/484, train_loss: 0.0044\n",
      "179/484, train_loss: 0.1230\n",
      "180/484, train_loss: 0.0013\n",
      "181/484, train_loss: 0.0011\n",
      "182/484, train_loss: 0.0026\n",
      "183/484, train_loss: 0.0166\n",
      "184/484, train_loss: 0.1230\n",
      "185/484, train_loss: 0.0171\n",
      "186/484, train_loss: 0.0058\n",
      "187/484, train_loss: 0.0018\n",
      "188/484, train_loss: 0.0051\n",
      "189/484, train_loss: 0.0051\n",
      "190/484, train_loss: 0.0034\n",
      "191/484, train_loss: 1.0731\n",
      "192/484, train_loss: 0.1886\n",
      "193/484, train_loss: 0.0020\n",
      "194/484, train_loss: 0.0024\n",
      "195/484, train_loss: 0.1182\n",
      "196/484, train_loss: 0.1023\n",
      "197/484, train_loss: 0.1775\n",
      "198/484, train_loss: 0.0015\n",
      "199/484, train_loss: 0.0013\n",
      "200/484, train_loss: 0.0038\n",
      "201/484, train_loss: 0.1081\n",
      "202/484, train_loss: 0.0022\n",
      "203/484, train_loss: 0.0120\n",
      "204/484, train_loss: 0.0015\n",
      "205/484, train_loss: 0.0034\n",
      "206/484, train_loss: 0.0014\n",
      "207/484, train_loss: 0.0030\n",
      "208/484, train_loss: 0.0794\n",
      "209/484, train_loss: 0.0427\n",
      "210/484, train_loss: 1.0921\n",
      "211/484, train_loss: 0.0025\n",
      "212/484, train_loss: 0.0893\n",
      "213/484, train_loss: 0.0053\n",
      "214/484, train_loss: 0.0019\n",
      "215/484, train_loss: 0.0016\n",
      "216/484, train_loss: 0.0988\n",
      "217/484, train_loss: 0.0059\n",
      "218/484, train_loss: 0.0051\n",
      "219/484, train_loss: 0.0090\n",
      "220/484, train_loss: 0.0061\n",
      "221/484, train_loss: 0.0013\n",
      "222/484, train_loss: 0.0022\n",
      "223/484, train_loss: 0.0126\n",
      "224/484, train_loss: 0.0013\n",
      "225/484, train_loss: 0.0038\n",
      "226/484, train_loss: 0.0024\n",
      "227/484, train_loss: 0.0030\n",
      "228/484, train_loss: 0.8028\n",
      "229/484, train_loss: 0.0072\n",
      "230/484, train_loss: 1.7673\n",
      "231/484, train_loss: 0.0012\n",
      "232/484, train_loss: 0.0023\n",
      "233/484, train_loss: 0.0311\n",
      "234/484, train_loss: 0.0024\n",
      "235/484, train_loss: 0.9949\n",
      "236/484, train_loss: 0.3266\n",
      "237/484, train_loss: 0.0022\n",
      "238/484, train_loss: 0.0077\n",
      "239/484, train_loss: 0.0130\n",
      "240/484, train_loss: 0.0046\n",
      "241/484, train_loss: 0.0024\n",
      "242/484, train_loss: 0.7879\n",
      "243/484, train_loss: 0.0026\n",
      "244/484, train_loss: 0.0031\n",
      "245/484, train_loss: 0.0017\n",
      "246/484, train_loss: 0.0022\n",
      "247/484, train_loss: 0.0017\n",
      "248/484, train_loss: 0.0027\n",
      "249/484, train_loss: 0.1665\n",
      "250/484, train_loss: 0.0045\n",
      "251/484, train_loss: 0.0014\n",
      "252/484, train_loss: 0.0176\n",
      "253/484, train_loss: 0.0034\n",
      "254/484, train_loss: 0.0068\n",
      "255/484, train_loss: 0.0043\n",
      "256/484, train_loss: 0.0019\n",
      "257/484, train_loss: 0.0016\n",
      "258/484, train_loss: 0.0041\n",
      "259/484, train_loss: 0.0095\n",
      "260/484, train_loss: 0.0029\n",
      "261/484, train_loss: 0.0055\n",
      "262/484, train_loss: 0.0012\n",
      "263/484, train_loss: 0.0070\n",
      "264/484, train_loss: 0.4971\n",
      "265/484, train_loss: 0.0059\n",
      "266/484, train_loss: 0.0032\n",
      "267/484, train_loss: 0.0010\n",
      "268/484, train_loss: 0.0070\n",
      "269/484, train_loss: 0.0062\n",
      "270/484, train_loss: 0.0025\n",
      "271/484, train_loss: 0.0229\n",
      "272/484, train_loss: 0.0023\n",
      "273/484, train_loss: 0.0026\n",
      "274/484, train_loss: 0.0060\n",
      "275/484, train_loss: 0.0026\n",
      "276/484, train_loss: 0.0027\n",
      "277/484, train_loss: 0.0066\n",
      "278/484, train_loss: 0.0037\n",
      "279/484, train_loss: 0.0051\n",
      "280/484, train_loss: 0.2735\n",
      "281/484, train_loss: 0.5188\n",
      "282/484, train_loss: 0.0015\n",
      "283/484, train_loss: 0.0013\n",
      "284/484, train_loss: 0.0078\n",
      "285/484, train_loss: 0.4263\n",
      "286/484, train_loss: 0.0038\n",
      "287/484, train_loss: 0.0081\n",
      "288/484, train_loss: 0.0023\n",
      "289/484, train_loss: 0.0014\n",
      "290/484, train_loss: 0.0015\n",
      "291/484, train_loss: 0.0022\n",
      "292/484, train_loss: 0.0122\n",
      "293/484, train_loss: 0.9120\n",
      "294/484, train_loss: 0.0072\n",
      "295/484, train_loss: 0.0021\n",
      "296/484, train_loss: 0.0102\n",
      "297/484, train_loss: 0.0019\n",
      "298/484, train_loss: 0.0197\n",
      "299/484, train_loss: 0.6054\n",
      "300/484, train_loss: 0.0040\n",
      "301/484, train_loss: 0.0010\n",
      "302/484, train_loss: 0.0067\n",
      "303/484, train_loss: 0.0020\n",
      "304/484, train_loss: 0.0008\n",
      "305/484, train_loss: 0.0018\n",
      "306/484, train_loss: 0.0022\n",
      "307/484, train_loss: 0.4935\n",
      "308/484, train_loss: 0.0095\n",
      "309/484, train_loss: 0.0079\n",
      "310/484, train_loss: 0.0158\n",
      "311/484, train_loss: 0.0048\n",
      "312/484, train_loss: 0.0023\n",
      "313/484, train_loss: 0.0083\n",
      "314/484, train_loss: 0.0039\n",
      "315/484, train_loss: 0.0833\n",
      "316/484, train_loss: 0.0014\n",
      "317/484, train_loss: 0.0031\n",
      "318/484, train_loss: 0.0061\n",
      "319/484, train_loss: 0.0459\n",
      "320/484, train_loss: 0.0078\n",
      "321/484, train_loss: 0.0010\n",
      "322/484, train_loss: 0.0026\n",
      "323/484, train_loss: 0.7709\n",
      "324/484, train_loss: 0.1160\n",
      "325/484, train_loss: 0.0018\n",
      "326/484, train_loss: 0.0065\n",
      "327/484, train_loss: 0.0096\n",
      "328/484, train_loss: 0.4583\n",
      "329/484, train_loss: 0.0146\n",
      "330/484, train_loss: 0.0217\n",
      "331/484, train_loss: 0.0009\n",
      "332/484, train_loss: 0.0088\n",
      "333/484, train_loss: 0.0013\n",
      "334/484, train_loss: 0.0058\n",
      "335/484, train_loss: 0.0030\n",
      "336/484, train_loss: 0.0077\n",
      "337/484, train_loss: 0.0055\n",
      "338/484, train_loss: 0.0106\n",
      "339/484, train_loss: 0.0022\n",
      "340/484, train_loss: 0.0047\n",
      "341/484, train_loss: 0.4188\n",
      "342/484, train_loss: 0.0030\n",
      "343/484, train_loss: 0.0131\n",
      "344/484, train_loss: 0.0032\n",
      "345/484, train_loss: 0.0065\n",
      "346/484, train_loss: 0.0015\n",
      "347/484, train_loss: 0.0048\n",
      "348/484, train_loss: 0.0100\n",
      "349/484, train_loss: 0.0059\n",
      "350/484, train_loss: 0.1267\n",
      "351/484, train_loss: 0.0015\n",
      "352/484, train_loss: 0.0019\n",
      "353/484, train_loss: 0.0071\n",
      "354/484, train_loss: 0.0120\n",
      "355/484, train_loss: 0.3917\n",
      "356/484, train_loss: 0.0036\n",
      "357/484, train_loss: 0.0077\n",
      "358/484, train_loss: 0.0112\n",
      "359/484, train_loss: 0.0035\n",
      "360/484, train_loss: 0.0080\n",
      "361/484, train_loss: 0.0025\n",
      "362/484, train_loss: 0.0116\n",
      "363/484, train_loss: 0.0032\n",
      "364/484, train_loss: 0.0020\n",
      "365/484, train_loss: 0.0017\n",
      "366/484, train_loss: 0.8246\n",
      "367/484, train_loss: 0.0082\n",
      "368/484, train_loss: 0.0289\n",
      "369/484, train_loss: 0.0025\n",
      "370/484, train_loss: 0.4885\n",
      "371/484, train_loss: 0.4779\n",
      "372/484, train_loss: 0.1337\n",
      "373/484, train_loss: 0.0024\n",
      "374/484, train_loss: 0.0041\n",
      "375/484, train_loss: 0.1252\n",
      "376/484, train_loss: 0.0032\n",
      "377/484, train_loss: 0.0119\n",
      "378/484, train_loss: 0.9433\n",
      "379/484, train_loss: 0.0037\n",
      "380/484, train_loss: 0.0043\n",
      "381/484, train_loss: 0.0034\n",
      "382/484, train_loss: 0.0013\n",
      "383/484, train_loss: 0.0034\n",
      "384/484, train_loss: 0.0072\n",
      "385/484, train_loss: 0.0020\n",
      "386/484, train_loss: 0.0033\n",
      "387/484, train_loss: 0.5613\n",
      "388/484, train_loss: 0.0029\n",
      "389/484, train_loss: 0.0014\n",
      "390/484, train_loss: 0.3879\n",
      "391/484, train_loss: 0.0028\n",
      "392/484, train_loss: 0.0026\n",
      "393/484, train_loss: 0.0011\n",
      "394/484, train_loss: 0.0023\n",
      "395/484, train_loss: 0.5057\n",
      "396/484, train_loss: 0.7050\n",
      "397/484, train_loss: 0.0037\n",
      "398/484, train_loss: 0.0068\n",
      "399/484, train_loss: 0.0030\n",
      "400/484, train_loss: 0.0010\n",
      "401/484, train_loss: 0.1150\n",
      "402/484, train_loss: 0.9011\n",
      "403/484, train_loss: 0.3409\n",
      "404/484, train_loss: 0.0132\n",
      "405/484, train_loss: 0.0010\n",
      "406/484, train_loss: 0.0022\n",
      "407/484, train_loss: 0.6211\n",
      "408/484, train_loss: 0.0119\n",
      "409/484, train_loss: 1.0856\n",
      "410/484, train_loss: 0.0031\n",
      "411/484, train_loss: 0.0016\n",
      "412/484, train_loss: 0.0066\n",
      "413/484, train_loss: 0.0076\n",
      "414/484, train_loss: 0.0023\n",
      "415/484, train_loss: 0.1849\n",
      "416/484, train_loss: 0.0397\n",
      "417/484, train_loss: 0.0046\n",
      "418/484, train_loss: 0.0049\n",
      "419/484, train_loss: 1.1112\n",
      "420/484, train_loss: 0.0021\n",
      "421/484, train_loss: 0.2091\n",
      "422/484, train_loss: 0.0067\n",
      "423/484, train_loss: 0.7192\n",
      "424/484, train_loss: 0.0346\n",
      "425/484, train_loss: 0.0028\n",
      "426/484, train_loss: 0.0029\n",
      "427/484, train_loss: 0.0315\n",
      "428/484, train_loss: 0.0032\n",
      "429/484, train_loss: 0.0106\n",
      "430/484, train_loss: 0.0068\n",
      "431/484, train_loss: 0.0026\n",
      "432/484, train_loss: 0.0086\n",
      "433/484, train_loss: 0.0016\n",
      "434/484, train_loss: 0.0185\n",
      "435/484, train_loss: 0.0030\n",
      "436/484, train_loss: 0.1564\n",
      "437/484, train_loss: 0.0062\n",
      "438/484, train_loss: 1.1566\n",
      "439/484, train_loss: 0.0110\n",
      "440/484, train_loss: 0.0024\n",
      "441/484, train_loss: 0.6539\n",
      "442/484, train_loss: 0.0017\n",
      "443/484, train_loss: 0.0027\n",
      "444/484, train_loss: 0.0049\n",
      "445/484, train_loss: 0.0094\n",
      "446/484, train_loss: 0.0032\n",
      "447/484, train_loss: 0.0160\n",
      "448/484, train_loss: 0.8007\n",
      "449/484, train_loss: 0.4312\n",
      "450/484, train_loss: 0.0013\n",
      "451/484, train_loss: 0.2591\n",
      "452/484, train_loss: 0.0028\n",
      "453/484, train_loss: 0.1844\n",
      "454/484, train_loss: 0.0076\n",
      "455/484, train_loss: 0.0012\n",
      "456/484, train_loss: 0.0069\n",
      "457/484, train_loss: 0.2623\n",
      "458/484, train_loss: 0.1222\n",
      "459/484, train_loss: 0.0045\n",
      "460/484, train_loss: 0.0018\n",
      "461/484, train_loss: 0.0010\n",
      "462/484, train_loss: 0.0137\n",
      "463/484, train_loss: 0.0113\n",
      "464/484, train_loss: 0.0340\n",
      "465/484, train_loss: 0.0109\n",
      "466/484, train_loss: 0.0100\n",
      "467/484, train_loss: 0.0064\n",
      "468/484, train_loss: 0.0010\n",
      "469/484, train_loss: 0.0020\n",
      "470/484, train_loss: 0.0051\n",
      "471/484, train_loss: 0.0024\n",
      "472/484, train_loss: 0.7546\n",
      "473/484, train_loss: 0.0064\n",
      "474/484, train_loss: 0.0194\n",
      "475/484, train_loss: 0.0017\n",
      "476/484, train_loss: 0.0137\n",
      "477/484, train_loss: 0.0015\n",
      "478/484, train_loss: 0.0032\n",
      "479/484, train_loss: 0.1443\n",
      "480/484, train_loss: 0.6423\n",
      "481/484, train_loss: 0.0033\n",
      "482/484, train_loss: 0.0024\n",
      "483/484, train_loss: 0.0022\n",
      "484/484, train_loss: 0.0019\n",
      "epoch 18 average loss: 0.0913\n",
      "Current epoch: 18 current accuracy: 0.7709 \n",
      "Best accuracy: 0.8669 at epoch 14\n",
      "----------\n",
      "epoch 19/20\n",
      "1/484, train_loss: 0.0133\n",
      "2/484, train_loss: 0.0059\n",
      "3/484, train_loss: 0.0610\n",
      "4/484, train_loss: 0.0019\n",
      "5/484, train_loss: 0.0123\n",
      "6/484, train_loss: 0.2175\n",
      "7/484, train_loss: 0.4493\n",
      "8/484, train_loss: 0.0026\n",
      "9/484, train_loss: 0.0161\n",
      "10/484, train_loss: 0.0006\n",
      "11/484, train_loss: 0.3318\n",
      "12/484, train_loss: 1.0573\n",
      "13/484, train_loss: 0.1410\n",
      "14/484, train_loss: 0.0042\n",
      "15/484, train_loss: 0.0023\n",
      "16/484, train_loss: 0.0140\n",
      "17/484, train_loss: 0.0715\n",
      "18/484, train_loss: 0.0025\n",
      "19/484, train_loss: 0.0024\n",
      "20/484, train_loss: 0.0154\n",
      "21/484, train_loss: 0.0010\n",
      "22/484, train_loss: 0.7917\n",
      "23/484, train_loss: 0.4550\n",
      "24/484, train_loss: 0.0023\n",
      "25/484, train_loss: 0.0480\n",
      "26/484, train_loss: 0.0053\n",
      "27/484, train_loss: 0.9866\n",
      "28/484, train_loss: 0.0421\n",
      "29/484, train_loss: 0.0040\n",
      "30/484, train_loss: 0.6332\n",
      "31/484, train_loss: 0.0022\n",
      "32/484, train_loss: 0.0086\n",
      "33/484, train_loss: 0.0102\n",
      "34/484, train_loss: 0.5689\n",
      "35/484, train_loss: 0.6337\n",
      "36/484, train_loss: 0.0087\n",
      "37/484, train_loss: 0.0146\n",
      "38/484, train_loss: 0.0850\n",
      "39/484, train_loss: 0.3099\n",
      "40/484, train_loss: 0.0059\n",
      "41/484, train_loss: 0.0035\n",
      "42/484, train_loss: 0.0010\n",
      "43/484, train_loss: 0.0046\n",
      "44/484, train_loss: 0.0033\n",
      "45/484, train_loss: 0.0234\n",
      "46/484, train_loss: 0.5278\n",
      "47/484, train_loss: 0.2077\n",
      "48/484, train_loss: 0.0170\n",
      "49/484, train_loss: 0.1084\n",
      "50/484, train_loss: 0.0038\n",
      "51/484, train_loss: 0.0027\n",
      "52/484, train_loss: 0.0526\n",
      "53/484, train_loss: 0.0084\n",
      "54/484, train_loss: 0.0640\n",
      "55/484, train_loss: 0.1317\n",
      "56/484, train_loss: 0.0095\n",
      "57/484, train_loss: 0.0017\n",
      "58/484, train_loss: 0.0057\n",
      "59/484, train_loss: 0.0036\n",
      "60/484, train_loss: 0.3574\n",
      "61/484, train_loss: 0.0029\n",
      "62/484, train_loss: 0.5600\n",
      "63/484, train_loss: 0.0189\n",
      "64/484, train_loss: 0.0196\n",
      "65/484, train_loss: 0.4133\n",
      "66/484, train_loss: 0.3416\n",
      "67/484, train_loss: 0.0026\n",
      "68/484, train_loss: 0.0179\n",
      "69/484, train_loss: 0.0030\n",
      "70/484, train_loss: 0.0014\n",
      "71/484, train_loss: 0.2634\n",
      "72/484, train_loss: 0.0041\n",
      "73/484, train_loss: 0.0469\n",
      "74/484, train_loss: 0.0021\n",
      "75/484, train_loss: 0.0095\n",
      "76/484, train_loss: 0.0013\n",
      "77/484, train_loss: 0.0013\n",
      "78/484, train_loss: 0.0020\n",
      "79/484, train_loss: 0.0030\n",
      "80/484, train_loss: 0.0065\n",
      "81/484, train_loss: 1.1135\n",
      "82/484, train_loss: 0.0019\n",
      "83/484, train_loss: 0.0050\n",
      "84/484, train_loss: 0.0056\n",
      "85/484, train_loss: 0.0010\n",
      "86/484, train_loss: 0.1339\n",
      "87/484, train_loss: 0.0018\n",
      "88/484, train_loss: 0.0021\n",
      "89/484, train_loss: 0.0150\n",
      "90/484, train_loss: 0.1368\n",
      "91/484, train_loss: 0.3351\n",
      "92/484, train_loss: 0.0029\n",
      "93/484, train_loss: 0.0057\n",
      "94/484, train_loss: 0.0125\n",
      "95/484, train_loss: 0.4661\n",
      "96/484, train_loss: 1.5563\n",
      "97/484, train_loss: 0.0013\n",
      "98/484, train_loss: 0.0015\n",
      "99/484, train_loss: 0.0050\n",
      "100/484, train_loss: 0.0039\n",
      "101/484, train_loss: 0.0016\n",
      "102/484, train_loss: 0.0027\n",
      "103/484, train_loss: 0.0022\n",
      "104/484, train_loss: 0.0082\n",
      "105/484, train_loss: 0.0909\n",
      "106/484, train_loss: 0.0030\n",
      "107/484, train_loss: 0.4799\n",
      "108/484, train_loss: 0.0022\n",
      "109/484, train_loss: 0.0143\n",
      "110/484, train_loss: 0.5945\n",
      "111/484, train_loss: 0.0020\n",
      "112/484, train_loss: 0.0120\n",
      "113/484, train_loss: 0.0015\n",
      "114/484, train_loss: 0.0015\n",
      "115/484, train_loss: 0.0058\n",
      "116/484, train_loss: 0.3752\n",
      "117/484, train_loss: 0.4072\n",
      "118/484, train_loss: 0.0024\n",
      "119/484, train_loss: 0.0050\n",
      "120/484, train_loss: 0.0273\n",
      "121/484, train_loss: 0.0013\n",
      "122/484, train_loss: 0.0034\n",
      "123/484, train_loss: 0.0009\n",
      "124/484, train_loss: 0.0058\n",
      "125/484, train_loss: 0.0016\n",
      "126/484, train_loss: 0.0022\n",
      "127/484, train_loss: 0.0126\n",
      "128/484, train_loss: 0.0054\n",
      "129/484, train_loss: 0.2485\n",
      "130/484, train_loss: 0.0051\n",
      "131/484, train_loss: 0.0033\n",
      "132/484, train_loss: 0.0114\n",
      "133/484, train_loss: 0.4551\n",
      "134/484, train_loss: 0.0693\n",
      "135/484, train_loss: 0.7985\n",
      "136/484, train_loss: 0.1929\n",
      "137/484, train_loss: 0.0071\n",
      "138/484, train_loss: 0.0057\n",
      "139/484, train_loss: 0.0029\n",
      "140/484, train_loss: 0.0027\n",
      "141/484, train_loss: 0.0047\n",
      "142/484, train_loss: 0.0011\n",
      "143/484, train_loss: 0.0041\n",
      "144/484, train_loss: 0.0058\n",
      "145/484, train_loss: 0.0871\n",
      "146/484, train_loss: 0.0229\n",
      "147/484, train_loss: 0.0075\n",
      "148/484, train_loss: 0.0016\n",
      "149/484, train_loss: 0.0096\n",
      "150/484, train_loss: 0.0632\n",
      "151/484, train_loss: 0.0117\n",
      "152/484, train_loss: 0.0122\n",
      "153/484, train_loss: 0.5709\n",
      "154/484, train_loss: 2.0978\n",
      "155/484, train_loss: 0.0107\n",
      "156/484, train_loss: 0.0223\n",
      "157/484, train_loss: 0.0034\n",
      "158/484, train_loss: 0.0176\n",
      "159/484, train_loss: 0.0380\n",
      "160/484, train_loss: 0.0080\n",
      "161/484, train_loss: 0.0070\n",
      "162/484, train_loss: 0.0076\n",
      "163/484, train_loss: 0.0058\n",
      "164/484, train_loss: 0.0019\n",
      "165/484, train_loss: 0.0066\n",
      "166/484, train_loss: 0.4582\n",
      "167/484, train_loss: 0.0509\n",
      "168/484, train_loss: 0.0437\n",
      "169/484, train_loss: 0.0110\n",
      "170/484, train_loss: 0.0064\n",
      "171/484, train_loss: 0.0029\n",
      "172/484, train_loss: 0.4157\n",
      "173/484, train_loss: 0.0043\n",
      "174/484, train_loss: 0.0462\n",
      "175/484, train_loss: 0.0049\n",
      "176/484, train_loss: 0.0016\n",
      "177/484, train_loss: 0.0020\n",
      "178/484, train_loss: 0.0213\n",
      "179/484, train_loss: 0.0054\n",
      "180/484, train_loss: 0.0181\n",
      "181/484, train_loss: 0.0032\n",
      "182/484, train_loss: 0.0056\n",
      "183/484, train_loss: 1.9020\n",
      "184/484, train_loss: 0.5395\n",
      "185/484, train_loss: 0.0058\n",
      "186/484, train_loss: 0.0072\n",
      "187/484, train_loss: 0.0130\n",
      "188/484, train_loss: 0.0038\n",
      "189/484, train_loss: 0.0067\n",
      "190/484, train_loss: 0.0014\n",
      "191/484, train_loss: 0.0017\n",
      "192/484, train_loss: 0.0066\n",
      "193/484, train_loss: 1.3223\n",
      "194/484, train_loss: 0.0101\n",
      "195/484, train_loss: 0.0100\n",
      "196/484, train_loss: 0.0872\n",
      "197/484, train_loss: 0.0053\n",
      "198/484, train_loss: 0.9071\n",
      "199/484, train_loss: 0.0276\n",
      "200/484, train_loss: 0.0110\n",
      "201/484, train_loss: 0.0041\n",
      "202/484, train_loss: 0.0018\n",
      "203/484, train_loss: 0.0129\n",
      "204/484, train_loss: 0.3428\n",
      "205/484, train_loss: 0.0591\n",
      "206/484, train_loss: 0.0020\n",
      "207/484, train_loss: 0.0062\n",
      "208/484, train_loss: 0.0038\n",
      "209/484, train_loss: 0.3240\n",
      "210/484, train_loss: 0.0063\n",
      "211/484, train_loss: 0.0022\n",
      "212/484, train_loss: 0.0011\n",
      "213/484, train_loss: 0.0027\n",
      "214/484, train_loss: 0.0012\n",
      "215/484, train_loss: 0.0057\n",
      "216/484, train_loss: 0.0077\n",
      "217/484, train_loss: 0.0012\n",
      "218/484, train_loss: 0.0019\n",
      "219/484, train_loss: 0.2950\n",
      "220/484, train_loss: 2.1252\n",
      "221/484, train_loss: 0.0037\n",
      "222/484, train_loss: 0.0041\n",
      "223/484, train_loss: 0.0073\n",
      "224/484, train_loss: 0.0155\n",
      "225/484, train_loss: 0.0266\n",
      "226/484, train_loss: 0.0095\n",
      "227/484, train_loss: 0.0085\n",
      "228/484, train_loss: 0.0043\n",
      "229/484, train_loss: 0.0062\n",
      "230/484, train_loss: 0.5980\n",
      "231/484, train_loss: 0.7625\n",
      "232/484, train_loss: 0.0021\n",
      "233/484, train_loss: 0.0187\n",
      "234/484, train_loss: 0.7199\n",
      "235/484, train_loss: 0.3149\n",
      "236/484, train_loss: 0.0021\n",
      "237/484, train_loss: 0.0038\n",
      "238/484, train_loss: 0.0202\n",
      "239/484, train_loss: 0.1444\n",
      "240/484, train_loss: 0.0029\n",
      "241/484, train_loss: 0.0019\n",
      "242/484, train_loss: 0.0033\n",
      "243/484, train_loss: 0.0450\n",
      "244/484, train_loss: 0.0027\n",
      "245/484, train_loss: 0.0037\n",
      "246/484, train_loss: 0.0168\n",
      "247/484, train_loss: 0.0838\n",
      "248/484, train_loss: 0.0131\n",
      "249/484, train_loss: 0.0026\n",
      "250/484, train_loss: 0.0081\n",
      "251/484, train_loss: 0.4760\n",
      "252/484, train_loss: 0.0016\n",
      "253/484, train_loss: 0.0044\n",
      "254/484, train_loss: 0.0166\n",
      "255/484, train_loss: 0.0084\n",
      "256/484, train_loss: 1.3418\n",
      "257/484, train_loss: 0.0109\n",
      "258/484, train_loss: 0.0213\n",
      "259/484, train_loss: 0.0509\n",
      "260/484, train_loss: 0.0065\n",
      "261/484, train_loss: 0.7314\n",
      "262/484, train_loss: 0.0045\n",
      "263/484, train_loss: 0.5042\n",
      "264/484, train_loss: 0.0146\n",
      "265/484, train_loss: 0.1383\n",
      "266/484, train_loss: 0.0165\n",
      "267/484, train_loss: 0.0023\n",
      "268/484, train_loss: 0.0110\n",
      "269/484, train_loss: 0.0089\n",
      "270/484, train_loss: 0.4554\n",
      "271/484, train_loss: 0.6247\n",
      "272/484, train_loss: 0.0026\n",
      "273/484, train_loss: 0.9633\n",
      "274/484, train_loss: 0.0017\n",
      "275/484, train_loss: 0.0102\n",
      "276/484, train_loss: 0.0073\n",
      "277/484, train_loss: 0.0130\n",
      "278/484, train_loss: 0.7792\n",
      "279/484, train_loss: 0.0016\n",
      "280/484, train_loss: 0.0044\n",
      "281/484, train_loss: 0.5326\n",
      "282/484, train_loss: 0.0065\n",
      "283/484, train_loss: 0.3978\n",
      "284/484, train_loss: 0.0272\n",
      "285/484, train_loss: 1.2798\n",
      "286/484, train_loss: 0.0118\n",
      "287/484, train_loss: 0.1806\n",
      "288/484, train_loss: 0.0027\n",
      "289/484, train_loss: 0.0038\n",
      "290/484, train_loss: 0.1433\n",
      "291/484, train_loss: 0.3274\n",
      "292/484, train_loss: 0.0153\n",
      "293/484, train_loss: 0.0036\n",
      "294/484, train_loss: 0.0085\n",
      "295/484, train_loss: 0.0049\n",
      "296/484, train_loss: 0.0185\n",
      "297/484, train_loss: 0.4786\n",
      "298/484, train_loss: 0.0035\n",
      "299/484, train_loss: 0.1207\n",
      "300/484, train_loss: 0.0491\n",
      "301/484, train_loss: 0.0041\n",
      "302/484, train_loss: 0.0045\n",
      "303/484, train_loss: 0.0054\n",
      "304/484, train_loss: 0.0054\n",
      "305/484, train_loss: 0.0150\n",
      "306/484, train_loss: 0.1602\n",
      "307/484, train_loss: 0.0946\n",
      "308/484, train_loss: 0.5091\n",
      "309/484, train_loss: 0.0055\n",
      "310/484, train_loss: 0.0138\n",
      "311/484, train_loss: 0.0078\n",
      "312/484, train_loss: 0.0052\n",
      "313/484, train_loss: 0.0214\n",
      "314/484, train_loss: 0.0047\n",
      "315/484, train_loss: 0.0042\n",
      "316/484, train_loss: 0.0180\n",
      "317/484, train_loss: 0.3696\n",
      "318/484, train_loss: 0.0050\n",
      "319/484, train_loss: 0.0065\n",
      "320/484, train_loss: 0.0319\n",
      "321/484, train_loss: 0.0124\n",
      "322/484, train_loss: 0.0049\n",
      "323/484, train_loss: 0.0153\n",
      "324/484, train_loss: 0.0112\n",
      "325/484, train_loss: 0.0064\n",
      "326/484, train_loss: 0.6531\n",
      "327/484, train_loss: 0.3295\n",
      "328/484, train_loss: 0.0033\n",
      "329/484, train_loss: 0.0123\n",
      "330/484, train_loss: 0.0029\n",
      "331/484, train_loss: 0.0158\n",
      "332/484, train_loss: 0.0201\n",
      "333/484, train_loss: 0.0028\n",
      "334/484, train_loss: 0.0072\n",
      "335/484, train_loss: 0.0236\n",
      "336/484, train_loss: 0.2114\n",
      "337/484, train_loss: 0.0045\n",
      "338/484, train_loss: 0.0042\n",
      "339/484, train_loss: 0.0047\n",
      "340/484, train_loss: 0.0121\n",
      "341/484, train_loss: 0.0062\n",
      "342/484, train_loss: 0.0357\n",
      "343/484, train_loss: 0.0027\n",
      "344/484, train_loss: 0.8749\n",
      "345/484, train_loss: 0.0042\n",
      "346/484, train_loss: 0.0136\n",
      "347/484, train_loss: 0.0025\n",
      "348/484, train_loss: 0.0034\n",
      "349/484, train_loss: 0.0030\n",
      "350/484, train_loss: 0.0130\n",
      "351/484, train_loss: 0.1999\n",
      "352/484, train_loss: 1.4897\n",
      "353/484, train_loss: 0.0035\n",
      "354/484, train_loss: 0.0078\n",
      "355/484, train_loss: 0.0080\n",
      "356/484, train_loss: 0.0042\n",
      "357/484, train_loss: 0.0119\n",
      "358/484, train_loss: 0.0131\n",
      "359/484, train_loss: 0.2562\n",
      "360/484, train_loss: 0.0274\n",
      "361/484, train_loss: 0.3042\n",
      "362/484, train_loss: 0.0100\n",
      "363/484, train_loss: 0.0068\n",
      "364/484, train_loss: 0.3630\n",
      "365/484, train_loss: 0.7482\n",
      "366/484, train_loss: 0.0086\n",
      "367/484, train_loss: 0.0054\n",
      "368/484, train_loss: 0.0025\n",
      "369/484, train_loss: 0.0122\n",
      "370/484, train_loss: 0.0027\n",
      "371/484, train_loss: 1.2312\n",
      "372/484, train_loss: 0.7148\n",
      "373/484, train_loss: 0.0126\n",
      "374/484, train_loss: 0.0365\n",
      "375/484, train_loss: 0.0147\n",
      "376/484, train_loss: 0.2463\n",
      "377/484, train_loss: 0.0045\n",
      "378/484, train_loss: 0.0167\n",
      "379/484, train_loss: 0.2956\n",
      "380/484, train_loss: 0.0015\n",
      "381/484, train_loss: 0.0038\n",
      "382/484, train_loss: 0.0042\n",
      "383/484, train_loss: 0.1466\n",
      "384/484, train_loss: 0.0047\n",
      "385/484, train_loss: 0.0026\n",
      "386/484, train_loss: 1.2222\n",
      "387/484, train_loss: 0.0099\n",
      "388/484, train_loss: 0.6360\n",
      "389/484, train_loss: 0.0030\n",
      "390/484, train_loss: 0.0033\n",
      "391/484, train_loss: 0.0038\n",
      "392/484, train_loss: 0.0291\n",
      "393/484, train_loss: 0.0034\n",
      "394/484, train_loss: 0.0349\n",
      "395/484, train_loss: 0.0123\n",
      "396/484, train_loss: 0.4975\n",
      "397/484, train_loss: 0.0883\n",
      "398/484, train_loss: 0.0804\n",
      "399/484, train_loss: 0.0044\n",
      "400/484, train_loss: 0.0092\n",
      "401/484, train_loss: 0.0078\n",
      "402/484, train_loss: 0.0669\n",
      "403/484, train_loss: 0.0020\n",
      "404/484, train_loss: 0.0179\n",
      "405/484, train_loss: 0.0091\n",
      "406/484, train_loss: 0.0033\n",
      "407/484, train_loss: 0.0126\n",
      "408/484, train_loss: 0.1374\n",
      "409/484, train_loss: 0.0058\n",
      "410/484, train_loss: 0.0078\n",
      "411/484, train_loss: 0.0020\n",
      "412/484, train_loss: 0.0019\n",
      "413/484, train_loss: 0.0106\n",
      "414/484, train_loss: 0.0028\n",
      "415/484, train_loss: 0.4974\n",
      "416/484, train_loss: 0.0042\n",
      "417/484, train_loss: 0.0025\n",
      "418/484, train_loss: 0.0200\n",
      "419/484, train_loss: 0.0014\n",
      "420/484, train_loss: 0.0014\n",
      "421/484, train_loss: 0.8918\n",
      "422/484, train_loss: 0.0058\n",
      "423/484, train_loss: 0.0011\n",
      "424/484, train_loss: 0.0135\n",
      "425/484, train_loss: 0.3182\n",
      "426/484, train_loss: 0.0013\n",
      "427/484, train_loss: 0.0054\n",
      "428/484, train_loss: 0.0034\n",
      "429/484, train_loss: 2.7324\n",
      "430/484, train_loss: 0.0117\n",
      "431/484, train_loss: 2.8668\n",
      "432/484, train_loss: 4.0972\n",
      "433/484, train_loss: 0.0072\n",
      "434/484, train_loss: 0.2864\n",
      "435/484, train_loss: 0.0020\n",
      "436/484, train_loss: 0.0048\n",
      "437/484, train_loss: 0.0036\n",
      "438/484, train_loss: 0.7773\n",
      "439/484, train_loss: 0.0133\n",
      "440/484, train_loss: 0.0174\n",
      "441/484, train_loss: 0.1056\n",
      "442/484, train_loss: 2.9923\n",
      "443/484, train_loss: 0.2428\n",
      "444/484, train_loss: 0.2214\n",
      "445/484, train_loss: 2.5370\n",
      "446/484, train_loss: 0.2979\n",
      "447/484, train_loss: 0.0065\n",
      "448/484, train_loss: 0.0056\n",
      "449/484, train_loss: 0.2216\n",
      "450/484, train_loss: 0.0046\n",
      "451/484, train_loss: 0.0204\n",
      "452/484, train_loss: 0.0421\n",
      "453/484, train_loss: 0.4632\n",
      "454/484, train_loss: 0.0290\n",
      "455/484, train_loss: 0.0611\n",
      "456/484, train_loss: 0.1842\n",
      "457/484, train_loss: 0.0085\n",
      "458/484, train_loss: 0.0505\n",
      "459/484, train_loss: 0.0206\n",
      "460/484, train_loss: 1.1104\n",
      "461/484, train_loss: 0.3502\n",
      "462/484, train_loss: 0.0011\n",
      "463/484, train_loss: 0.2180\n",
      "464/484, train_loss: 0.0031\n",
      "465/484, train_loss: 0.0101\n",
      "466/484, train_loss: 1.1361\n",
      "467/484, train_loss: 0.0068\n",
      "468/484, train_loss: 0.3409\n",
      "469/484, train_loss: 0.3067\n",
      "470/484, train_loss: 0.0057\n",
      "471/484, train_loss: 0.9297\n",
      "472/484, train_loss: 0.0109\n",
      "473/484, train_loss: 0.6477\n",
      "474/484, train_loss: 0.0494\n",
      "475/484, train_loss: 0.0695\n",
      "476/484, train_loss: 0.0329\n",
      "477/484, train_loss: 0.4196\n",
      "478/484, train_loss: 0.1396\n",
      "479/484, train_loss: 0.0395\n",
      "480/484, train_loss: 0.0041\n",
      "481/484, train_loss: 0.0025\n",
      "482/484, train_loss: 0.3072\n",
      "483/484, train_loss: 0.0507\n",
      "484/484, train_loss: 0.0622\n",
      "epoch 19 average loss: 0.1694\n",
      "----------\n",
      "epoch 20/20\n",
      "1/484, train_loss: 0.0079\n",
      "2/484, train_loss: 0.4675\n",
      "3/484, train_loss: 0.0041\n",
      "4/484, train_loss: 0.0018\n",
      "5/484, train_loss: 0.0119\n",
      "6/484, train_loss: 0.2523\n",
      "7/484, train_loss: 0.0037\n",
      "8/484, train_loss: 0.0054\n",
      "9/484, train_loss: 0.0120\n",
      "10/484, train_loss: 0.0033\n",
      "11/484, train_loss: 0.5638\n",
      "12/484, train_loss: 0.0037\n",
      "13/484, train_loss: 0.0067\n",
      "14/484, train_loss: 0.0094\n",
      "15/484, train_loss: 0.0428\n",
      "16/484, train_loss: 0.0049\n",
      "17/484, train_loss: 0.3861\n",
      "18/484, train_loss: 0.0101\n",
      "19/484, train_loss: 1.3018\n",
      "20/484, train_loss: 0.0613\n",
      "21/484, train_loss: 0.0506\n",
      "22/484, train_loss: 0.4443\n",
      "23/484, train_loss: 0.2659\n",
      "24/484, train_loss: 0.0035\n",
      "25/484, train_loss: 0.0020\n",
      "26/484, train_loss: 0.0067\n",
      "27/484, train_loss: 0.0076\n",
      "28/484, train_loss: 0.0260\n",
      "29/484, train_loss: 0.4891\n",
      "30/484, train_loss: 0.0145\n",
      "31/484, train_loss: 0.0103\n",
      "32/484, train_loss: 0.3414\n",
      "33/484, train_loss: 0.0032\n",
      "34/484, train_loss: 0.0108\n",
      "35/484, train_loss: 0.0310\n",
      "36/484, train_loss: 0.0062\n",
      "37/484, train_loss: 0.0078\n",
      "38/484, train_loss: 0.0234\n",
      "39/484, train_loss: 0.5985\n",
      "40/484, train_loss: 0.4369\n",
      "41/484, train_loss: 0.3204\n",
      "42/484, train_loss: 0.0053\n",
      "43/484, train_loss: 0.1680\n",
      "44/484, train_loss: 0.0033\n",
      "45/484, train_loss: 0.0114\n",
      "46/484, train_loss: 0.0021\n",
      "47/484, train_loss: 0.0070\n",
      "48/484, train_loss: 0.0082\n",
      "49/484, train_loss: 0.0041\n",
      "50/484, train_loss: 0.0090\n",
      "51/484, train_loss: 0.8196\n",
      "52/484, train_loss: 0.3145\n",
      "53/484, train_loss: 0.0181\n",
      "54/484, train_loss: 0.0031\n",
      "55/484, train_loss: 0.3807\n",
      "56/484, train_loss: 0.0111\n",
      "57/484, train_loss: 0.5225\n",
      "58/484, train_loss: 0.6219\n",
      "59/484, train_loss: 0.0040\n",
      "60/484, train_loss: 0.0030\n",
      "61/484, train_loss: 0.4520\n",
      "62/484, train_loss: 0.0029\n",
      "63/484, train_loss: 0.0039\n",
      "64/484, train_loss: 0.0233\n",
      "65/484, train_loss: 2.1207\n",
      "66/484, train_loss: 0.0382\n",
      "67/484, train_loss: 0.0064\n",
      "68/484, train_loss: 0.0022\n",
      "69/484, train_loss: 0.1090\n",
      "70/484, train_loss: 0.0059\n",
      "71/484, train_loss: 0.2346\n",
      "72/484, train_loss: 0.0050\n",
      "73/484, train_loss: 0.3089\n",
      "74/484, train_loss: 0.0070\n",
      "75/484, train_loss: 0.0034\n",
      "76/484, train_loss: 0.5632\n",
      "77/484, train_loss: 0.0262\n",
      "78/484, train_loss: 0.0062\n",
      "79/484, train_loss: 0.0096\n",
      "80/484, train_loss: 0.0027\n",
      "81/484, train_loss: 0.0043\n",
      "82/484, train_loss: 0.3127\n",
      "83/484, train_loss: 0.0102\n",
      "84/484, train_loss: 0.5783\n",
      "85/484, train_loss: 0.0086\n",
      "86/484, train_loss: 0.0143\n",
      "87/484, train_loss: 0.0364\n",
      "88/484, train_loss: 0.0068\n",
      "89/484, train_loss: 0.4091\n",
      "90/484, train_loss: 0.0087\n",
      "91/484, train_loss: 0.0595\n",
      "92/484, train_loss: 0.0130\n",
      "93/484, train_loss: 0.0159\n",
      "94/484, train_loss: 0.0099\n",
      "95/484, train_loss: 0.1282\n",
      "96/484, train_loss: 0.0093\n",
      "97/484, train_loss: 0.5203\n",
      "98/484, train_loss: 0.0309\n",
      "99/484, train_loss: 0.0146\n",
      "100/484, train_loss: 0.0058\n",
      "101/484, train_loss: 1.4224\n",
      "102/484, train_loss: 0.0023\n",
      "103/484, train_loss: 0.0127\n",
      "104/484, train_loss: 0.0314\n",
      "105/484, train_loss: 0.0049\n",
      "106/484, train_loss: 0.9152\n",
      "107/484, train_loss: 0.0039\n",
      "108/484, train_loss: 0.5206\n",
      "109/484, train_loss: 0.0180\n",
      "110/484, train_loss: 0.0072\n",
      "111/484, train_loss: 0.0018\n",
      "112/484, train_loss: 0.0126\n",
      "113/484, train_loss: 0.1597\n",
      "114/484, train_loss: 0.0497\n",
      "115/484, train_loss: 0.6396\n",
      "116/484, train_loss: 1.4265\n",
      "117/484, train_loss: 0.0026\n",
      "118/484, train_loss: 0.0017\n",
      "119/484, train_loss: 0.0159\n",
      "120/484, train_loss: 0.0109\n",
      "121/484, train_loss: 0.0057\n",
      "122/484, train_loss: 0.0200\n",
      "123/484, train_loss: 0.0143\n",
      "124/484, train_loss: 0.0074\n",
      "125/484, train_loss: 0.0036\n",
      "126/484, train_loss: 0.0034\n",
      "127/484, train_loss: 0.0039\n",
      "128/484, train_loss: 0.0150\n",
      "129/484, train_loss: 0.8481\n",
      "130/484, train_loss: 0.0059\n",
      "131/484, train_loss: 0.0128\n",
      "132/484, train_loss: 0.0070\n",
      "133/484, train_loss: 0.0027\n",
      "134/484, train_loss: 0.0029\n",
      "135/484, train_loss: 0.0124\n",
      "136/484, train_loss: 0.0141\n",
      "137/484, train_loss: 0.2348\n",
      "138/484, train_loss: 0.0026\n",
      "139/484, train_loss: 0.0023\n",
      "140/484, train_loss: 0.4870\n",
      "141/484, train_loss: 0.0052\n",
      "142/484, train_loss: 0.0047\n",
      "143/484, train_loss: 0.3656\n",
      "144/484, train_loss: 0.0126\n",
      "145/484, train_loss: 0.0052\n",
      "146/484, train_loss: 0.0030\n",
      "147/484, train_loss: 0.0042\n",
      "148/484, train_loss: 0.0017\n",
      "149/484, train_loss: 0.0093\n",
      "150/484, train_loss: 0.1447\n",
      "151/484, train_loss: 0.0035\n",
      "152/484, train_loss: 0.1519\n",
      "153/484, train_loss: 0.0094\n",
      "154/484, train_loss: 0.0054\n",
      "155/484, train_loss: 0.0037\n",
      "156/484, train_loss: 0.6736\n",
      "157/484, train_loss: 1.0294\n",
      "158/484, train_loss: 0.0029\n",
      "159/484, train_loss: 0.8358\n",
      "160/484, train_loss: 0.0179\n",
      "161/484, train_loss: 1.1490\n",
      "162/484, train_loss: 0.0035\n",
      "163/484, train_loss: 0.0027\n",
      "164/484, train_loss: 0.4037\n",
      "165/484, train_loss: 0.0043\n",
      "166/484, train_loss: 0.0019\n",
      "167/484, train_loss: 0.0026\n",
      "168/484, train_loss: 0.0074\n",
      "169/484, train_loss: 0.0063\n",
      "170/484, train_loss: 0.1268\n",
      "171/484, train_loss: 0.0154\n",
      "172/484, train_loss: 0.0034\n",
      "173/484, train_loss: 0.2667\n",
      "174/484, train_loss: 0.0049\n",
      "175/484, train_loss: 0.0069\n",
      "176/484, train_loss: 0.0042\n",
      "177/484, train_loss: 0.0046\n",
      "178/484, train_loss: 0.0045\n",
      "179/484, train_loss: 0.0036\n",
      "180/484, train_loss: 0.0031\n",
      "181/484, train_loss: 0.0112\n",
      "182/484, train_loss: 0.0121\n",
      "183/484, train_loss: 0.0032\n",
      "184/484, train_loss: 0.0031\n",
      "185/484, train_loss: 0.0437\n",
      "186/484, train_loss: 0.0029\n",
      "187/484, train_loss: 0.7539\n",
      "188/484, train_loss: 0.0030\n",
      "189/484, train_loss: 0.0049\n",
      "190/484, train_loss: 0.0112\n",
      "191/484, train_loss: 0.0049\n",
      "192/484, train_loss: 0.0024\n",
      "193/484, train_loss: 0.0088\n",
      "194/484, train_loss: 0.0015\n",
      "195/484, train_loss: 0.6021\n",
      "196/484, train_loss: 0.0131\n",
      "197/484, train_loss: 0.0073\n",
      "198/484, train_loss: 0.0029\n",
      "199/484, train_loss: 0.0386\n",
      "200/484, train_loss: 0.0322\n",
      "201/484, train_loss: 0.0053\n",
      "202/484, train_loss: 0.0062\n",
      "203/484, train_loss: 0.0017\n",
      "204/484, train_loss: 0.0265\n",
      "205/484, train_loss: 0.0058\n",
      "206/484, train_loss: 0.0049\n",
      "207/484, train_loss: 1.0187\n",
      "208/484, train_loss: 0.0081\n",
      "209/484, train_loss: 0.0034\n",
      "210/484, train_loss: 0.0101\n",
      "211/484, train_loss: 0.0088\n",
      "212/484, train_loss: 0.2581\n",
      "213/484, train_loss: 0.5631\n",
      "214/484, train_loss: 0.6385\n",
      "215/484, train_loss: 0.0058\n",
      "216/484, train_loss: 0.0038\n",
      "217/484, train_loss: 0.1789\n",
      "218/484, train_loss: 0.0159\n",
      "219/484, train_loss: 0.2664\n",
      "220/484, train_loss: 0.0155\n",
      "221/484, train_loss: 0.0046\n",
      "222/484, train_loss: 0.0037\n",
      "223/484, train_loss: 0.0063\n",
      "224/484, train_loss: 0.0061\n",
      "225/484, train_loss: 0.6742\n",
      "226/484, train_loss: 0.0027\n",
      "227/484, train_loss: 0.0094\n",
      "228/484, train_loss: 0.0038\n",
      "229/484, train_loss: 0.3736\n",
      "230/484, train_loss: 0.1293\n",
      "231/484, train_loss: 0.0133\n",
      "232/484, train_loss: 0.0470\n",
      "233/484, train_loss: 1.1523\n",
      "234/484, train_loss: 0.0056\n",
      "235/484, train_loss: 0.0033\n",
      "236/484, train_loss: 0.7660\n",
      "237/484, train_loss: 0.4980\n",
      "238/484, train_loss: 0.4265\n",
      "239/484, train_loss: 0.1960\n",
      "240/484, train_loss: 0.8681\n",
      "241/484, train_loss: 0.3031\n",
      "242/484, train_loss: 0.0025\n",
      "243/484, train_loss: 0.0051\n",
      "244/484, train_loss: 0.0058\n",
      "245/484, train_loss: 0.0028\n",
      "246/484, train_loss: 3.4302\n",
      "247/484, train_loss: 0.0043\n",
      "248/484, train_loss: 0.0205\n",
      "249/484, train_loss: 0.0157\n",
      "250/484, train_loss: 0.0320\n",
      "251/484, train_loss: 0.0581\n",
      "252/484, train_loss: 0.0427\n",
      "253/484, train_loss: 1.0407\n",
      "254/484, train_loss: 0.0036\n",
      "255/484, train_loss: 0.0116\n",
      "256/484, train_loss: 0.0100\n",
      "257/484, train_loss: 0.3141\n",
      "258/484, train_loss: 1.6115\n",
      "259/484, train_loss: 0.0142\n",
      "260/484, train_loss: 0.0042\n",
      "261/484, train_loss: 0.6437\n",
      "262/484, train_loss: 0.0215\n",
      "263/484, train_loss: 0.0114\n",
      "264/484, train_loss: 0.0850\n",
      "265/484, train_loss: 0.8539\n",
      "266/484, train_loss: 0.0404\n",
      "267/484, train_loss: 0.6470\n",
      "268/484, train_loss: 0.0064\n",
      "269/484, train_loss: 0.0059\n",
      "270/484, train_loss: 0.0069\n",
      "271/484, train_loss: 0.0083\n",
      "272/484, train_loss: 0.4648\n",
      "273/484, train_loss: 0.0094\n",
      "274/484, train_loss: 0.0104\n",
      "275/484, train_loss: 0.0080\n",
      "276/484, train_loss: 0.0053\n",
      "277/484, train_loss: 0.0167\n",
      "278/484, train_loss: 0.3262\n",
      "279/484, train_loss: 0.0140\n",
      "280/484, train_loss: 0.0334\n",
      "281/484, train_loss: 0.0161\n",
      "282/484, train_loss: 0.0028\n",
      "283/484, train_loss: 0.0094\n",
      "284/484, train_loss: 0.0061\n",
      "285/484, train_loss: 0.0073\n",
      "286/484, train_loss: 0.0031\n",
      "287/484, train_loss: 0.0211\n",
      "288/484, train_loss: 0.0105\n",
      "289/484, train_loss: 0.0152\n",
      "290/484, train_loss: 0.0073\n",
      "291/484, train_loss: 0.0138\n",
      "292/484, train_loss: 0.1667\n",
      "293/484, train_loss: 0.0083\n",
      "294/484, train_loss: 0.0048\n",
      "295/484, train_loss: 0.0156\n",
      "296/484, train_loss: 0.0142\n",
      "297/484, train_loss: 0.4885\n",
      "298/484, train_loss: 0.0040\n",
      "299/484, train_loss: 0.0024\n",
      "300/484, train_loss: 0.0561\n",
      "301/484, train_loss: 0.0149\n",
      "302/484, train_loss: 0.0208\n",
      "303/484, train_loss: 0.1471\n",
      "304/484, train_loss: 2.9756\n",
      "305/484, train_loss: 0.0214\n",
      "306/484, train_loss: 0.0029\n",
      "307/484, train_loss: 0.0029\n",
      "308/484, train_loss: 0.2329\n",
      "309/484, train_loss: 0.0038\n",
      "310/484, train_loss: 0.0089\n",
      "311/484, train_loss: 0.0554\n",
      "312/484, train_loss: 0.5109\n",
      "313/484, train_loss: 0.0103\n",
      "314/484, train_loss: 0.0110\n",
      "315/484, train_loss: 0.3092\n",
      "316/484, train_loss: 0.0036\n",
      "317/484, train_loss: 0.0054\n",
      "318/484, train_loss: 0.0040\n",
      "319/484, train_loss: 1.0076\n",
      "320/484, train_loss: 0.0098\n",
      "321/484, train_loss: 0.0286\n",
      "322/484, train_loss: 0.0074\n",
      "323/484, train_loss: 0.0947\n",
      "324/484, train_loss: 0.0118\n",
      "325/484, train_loss: 0.1741\n",
      "326/484, train_loss: 0.0039\n",
      "327/484, train_loss: 0.0141\n",
      "328/484, train_loss: 0.0031\n",
      "329/484, train_loss: 0.0104\n",
      "330/484, train_loss: 0.0110\n",
      "331/484, train_loss: 0.0051\n",
      "332/484, train_loss: 0.0911\n",
      "333/484, train_loss: 0.0121\n",
      "334/484, train_loss: 1.7593\n",
      "335/484, train_loss: 0.0055\n",
      "336/484, train_loss: 0.0121\n",
      "337/484, train_loss: 0.0669\n",
      "338/484, train_loss: 0.0249\n",
      "339/484, train_loss: 0.1202\n",
      "340/484, train_loss: 0.0075\n",
      "341/484, train_loss: 0.0369\n",
      "342/484, train_loss: 0.0160\n",
      "343/484, train_loss: 0.6156\n",
      "344/484, train_loss: 0.0087\n",
      "345/484, train_loss: 0.0076\n",
      "346/484, train_loss: 0.6278\n",
      "347/484, train_loss: 0.0082\n",
      "348/484, train_loss: 0.0082\n",
      "349/484, train_loss: 0.0714\n",
      "350/484, train_loss: 0.1192\n",
      "351/484, train_loss: 0.1053\n",
      "352/484, train_loss: 0.0234\n",
      "353/484, train_loss: 0.2667\n",
      "354/484, train_loss: 0.0216\n",
      "355/484, train_loss: 0.0087\n",
      "356/484, train_loss: 0.0343\n",
      "357/484, train_loss: 0.0105\n",
      "358/484, train_loss: 1.0049\n",
      "359/484, train_loss: 0.0027\n",
      "360/484, train_loss: 0.0067\n",
      "361/484, train_loss: 0.0075\n",
      "362/484, train_loss: 0.0201\n",
      "363/484, train_loss: 0.7386\n",
      "364/484, train_loss: 0.0071\n",
      "365/484, train_loss: 0.0058\n",
      "366/484, train_loss: 0.3857\n",
      "367/484, train_loss: 0.0171\n",
      "368/484, train_loss: 0.0137\n",
      "369/484, train_loss: 0.0036\n",
      "370/484, train_loss: 0.0044\n",
      "371/484, train_loss: 0.0027\n",
      "372/484, train_loss: 0.0758\n",
      "373/484, train_loss: 0.0072\n",
      "374/484, train_loss: 0.0142\n",
      "375/484, train_loss: 0.3305\n",
      "376/484, train_loss: 0.0088\n",
      "377/484, train_loss: 0.0033\n",
      "378/484, train_loss: 0.0041\n",
      "379/484, train_loss: 0.0076\n",
      "380/484, train_loss: 0.1680\n",
      "381/484, train_loss: 0.0037\n",
      "382/484, train_loss: 0.4026\n",
      "383/484, train_loss: 0.0092\n",
      "384/484, train_loss: 0.0054\n",
      "385/484, train_loss: 2.0600\n",
      "386/484, train_loss: 0.0585\n",
      "387/484, train_loss: 1.1493\n",
      "388/484, train_loss: 0.0061\n",
      "389/484, train_loss: 0.0055\n",
      "390/484, train_loss: 0.1554\n",
      "391/484, train_loss: 0.0142\n",
      "392/484, train_loss: 0.0327\n",
      "393/484, train_loss: 0.0050\n",
      "394/484, train_loss: 0.0043\n",
      "395/484, train_loss: 0.0035\n",
      "396/484, train_loss: 0.1223\n",
      "397/484, train_loss: 0.0053\n",
      "398/484, train_loss: 0.0045\n",
      "399/484, train_loss: 0.0087\n",
      "400/484, train_loss: 0.0389\n",
      "401/484, train_loss: 0.3990\n",
      "402/484, train_loss: 0.3972\n",
      "403/484, train_loss: 0.0065\n",
      "404/484, train_loss: 0.0069\n",
      "405/484, train_loss: 0.0068\n",
      "406/484, train_loss: 0.0106\n",
      "407/484, train_loss: 0.0054\n",
      "408/484, train_loss: 0.9528\n",
      "409/484, train_loss: 0.0066\n",
      "410/484, train_loss: 0.0119\n",
      "411/484, train_loss: 0.1448\n",
      "412/484, train_loss: 0.0122\n",
      "413/484, train_loss: 0.0047\n",
      "414/484, train_loss: 0.1175\n",
      "415/484, train_loss: 0.7969\n",
      "416/484, train_loss: 0.0357\n",
      "417/484, train_loss: 0.0373\n",
      "418/484, train_loss: 0.0081\n",
      "419/484, train_loss: 0.0222\n",
      "420/484, train_loss: 0.0084\n",
      "421/484, train_loss: 0.0301\n",
      "422/484, train_loss: 0.1628\n",
      "423/484, train_loss: 0.0082\n",
      "424/484, train_loss: 0.4114\n",
      "425/484, train_loss: 0.0134\n",
      "426/484, train_loss: 0.0038\n",
      "427/484, train_loss: 0.0120\n",
      "428/484, train_loss: 0.6536\n",
      "429/484, train_loss: 0.0041\n",
      "430/484, train_loss: 0.0044\n",
      "431/484, train_loss: 0.0316\n",
      "432/484, train_loss: 0.0062\n",
      "433/484, train_loss: 0.2733\n",
      "434/484, train_loss: 0.0105\n",
      "435/484, train_loss: 0.0037\n",
      "436/484, train_loss: 0.0026\n",
      "437/484, train_loss: 0.0092\n",
      "438/484, train_loss: 0.1461\n",
      "439/484, train_loss: 0.0051\n",
      "440/484, train_loss: 0.2787\n",
      "441/484, train_loss: 0.0621\n",
      "442/484, train_loss: 0.0034\n",
      "443/484, train_loss: 0.0177\n",
      "444/484, train_loss: 0.0036\n",
      "445/484, train_loss: 0.0044\n",
      "446/484, train_loss: 0.0027\n",
      "447/484, train_loss: 1.0464\n",
      "448/484, train_loss: 0.7615\n",
      "449/484, train_loss: 0.0166\n",
      "450/484, train_loss: 0.0090\n",
      "451/484, train_loss: 0.0036\n",
      "452/484, train_loss: 0.0049\n",
      "453/484, train_loss: 0.0039\n",
      "454/484, train_loss: 0.5053\n",
      "455/484, train_loss: 0.0057\n",
      "456/484, train_loss: 0.0136\n",
      "457/484, train_loss: 0.0081\n",
      "458/484, train_loss: 0.0092\n",
      "459/484, train_loss: 0.0084\n",
      "460/484, train_loss: 0.0948\n",
      "461/484, train_loss: 0.0180\n",
      "462/484, train_loss: 0.0067\n",
      "463/484, train_loss: 0.0069\n",
      "464/484, train_loss: 0.0046\n",
      "465/484, train_loss: 0.0050\n",
      "466/484, train_loss: 0.0061\n",
      "467/484, train_loss: 0.0730\n",
      "468/484, train_loss: 0.0067\n",
      "469/484, train_loss: 0.0630\n",
      "470/484, train_loss: 0.0401\n",
      "471/484, train_loss: 0.0165\n",
      "472/484, train_loss: 0.8286\n",
      "473/484, train_loss: 0.0080\n",
      "474/484, train_loss: 0.0106\n",
      "475/484, train_loss: 0.0063\n",
      "476/484, train_loss: 0.0134\n",
      "477/484, train_loss: 0.0097\n",
      "478/484, train_loss: 0.0146\n",
      "479/484, train_loss: 0.0027\n",
      "480/484, train_loss: 0.0056\n",
      "481/484, train_loss: 0.0051\n",
      "482/484, train_loss: 0.0143\n",
      "483/484, train_loss: 0.0326\n",
      "484/484, train_loss: 0.0134\n",
      "epoch 20 average loss: 0.1529\n",
      "Current epoch: 20 current accuracy: 0.8328 \n",
      "Best accuracy: 0.8669 at epoch 14\n",
      "Training completed, best_metric: 0.8669 at epoch: 14\n"
     ]
    }
   ],
   "source": [
    "# Create DenseNet121, CrossEntropyLoss and Adam optimizer\n",
    "loss_function = torch.nn.CrossEntropyLoss()\n",
    "\n",
    "optimizer = torch.optim.Adam(model.parameters(), 1e-4)\n",
    "\n",
    "# start a typical PyTorch training\n",
    "val_interval = 2\n",
    "best_metric = -1\n",
    "best_metric_epoch = -1\n",
    "epoch_loss_values = []\n",
    "metric_values = []\n",
    "writer = SummaryWriter()\n",
    "max_epochs = 20\n",
    "\n",
    "for epoch in range(max_epochs):\n",
    "    print(\"-\" * 10)\n",
    "    print(f\"epoch {epoch + 1}/{max_epochs}\")\n",
    "    model.train()\n",
    "    epoch_loss = 0\n",
    "    step = 0\n",
    "\n",
    "    for batch_data in train_loader:\n",
    "        step += 1\n",
    "        inputs, labels = batch_data[0].to(device), batch_data[1].to(device)\n",
    "        optimizer.zero_grad()\n",
    "        outputs = model(inputs)\n",
    "        loss = loss_function(outputs, labels)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        epoch_loss += loss.item()\n",
    "        epoch_len = len(train_ds) // train_loader.batch_size\n",
    "        print(f\"{step}/{epoch_len}, train_loss: {loss.item():.4f}\")\n",
    "        writer.add_scalar(\"train_loss\", loss.item(), epoch_len * epoch + step)\n",
    "\n",
    "    epoch_loss /= step\n",
    "    epoch_loss_values.append(epoch_loss)\n",
    "    print(f\"epoch {epoch + 1} average loss: {epoch_loss:.4f}\")\n",
    "\n",
    "    if (epoch + 1) % val_interval == 0:\n",
    "        model.eval()\n",
    "\n",
    "        num_correct = 0.0\n",
    "        metric_count = 0\n",
    "        for val_data in val_loader:\n",
    "            val_images, val_labels = val_data[0].to(device), val_data[1].to(device)\n",
    "            with torch.no_grad():\n",
    "                val_outputs = model(val_images)\n",
    "                value = torch.eq(val_outputs.argmax(dim=1), val_labels)\n",
    "                metric_count += len(value)\n",
    "                num_correct += value.sum().item()\n",
    "\n",
    "        metric = num_correct / metric_count\n",
    "        metric_values.append(metric)\n",
    "\n",
    "        if metric > best_metric:\n",
    "            best_metric = metric\n",
    "            best_metric_epoch = epoch + 1\n",
    "            torch.save(model.state_dict(), \"best_metric_model_classification3d_array.pth\")\n",
    "            print(\"saved new best metric model\")\n",
    "\n",
    "        print(f\"Current epoch: {epoch+1} current accuracy: {metric:.4f} \")\n",
    "        print(f\"Best accuracy: {best_metric:.4f} at epoch {best_metric_epoch}\")\n",
    "        writer.add_scalar(\"val_accuracy\", metric, epoch + 1)\n",
    "\n",
    "print(f\"Training completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}\")\n",
    "writer.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_ds = ImageDataset(image_files=test_paths, labels=test_labels, transform=val_transforms)\n",
    "test_loader = DataLoader(test_ds, batch_size=2, num_workers=2, pin_memory=pin_memory)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DenseNet121(\n",
       "  (features): Sequential(\n",
       "    (conv0): Conv3d(1, 64, kernel_size=(7, 7, 7), stride=(2, 2, 2), padding=(3, 3, 3), bias=False)\n",
       "    (norm0): BatchNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (relu0): ReLU(inplace=True)\n",
       "    (pool0): MaxPool3d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "    (denseblock1): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(96, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(128, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(160, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(192, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(224, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (transition1): _Transition(\n",
       "      (norm): BatchNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv): Conv3d(256, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "      (pool): AvgPool3d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock2): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(128, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(160, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(192, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(224, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(256, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(288, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(320, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(352, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(384, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(416, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(448, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(480, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (transition2): _Transition(\n",
       "      (norm): BatchNorm3d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv): Conv3d(512, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "      (pool): AvgPool3d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock3): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(256, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(288, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(320, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(352, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(384, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(416, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(448, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(480, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(512, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(544, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(576, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(608, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer13): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(640, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer14): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(672, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer15): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(704, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer16): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(736, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer17): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(768, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer18): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(800, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer19): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(832, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer20): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(864, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer21): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(896, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer22): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(928, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer23): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(960, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer24): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(992, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (transition3): _Transition(\n",
       "      (norm): BatchNorm3d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv): Conv3d(1024, 512, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "      (pool): AvgPool3d(kernel_size=2, stride=2, padding=0)\n",
       "    )\n",
       "    (denseblock4): _DenseBlock(\n",
       "      (denselayer1): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(512, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer2): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(544, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer3): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(576, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer4): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(608, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer5): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(640, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer6): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(672, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer7): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(704, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer8): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(736, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer9): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(768, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer10): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(800, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer11): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(832, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer12): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(864, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer13): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(896, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(896, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer14): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(928, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(928, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer15): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(960, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "      (denselayer16): _DenseLayer(\n",
       "        (layers): Sequential(\n",
       "          (norm1): BatchNorm3d(992, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu1): ReLU(inplace=True)\n",
       "          (conv1): Conv3d(992, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
       "          (norm2): BatchNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (relu2): ReLU(inplace=True)\n",
       "          (conv2): Conv3d(128, 32, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (norm5): BatchNorm3d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  )\n",
       "  (class_layers): Sequential(\n",
       "    (relu): ReLU(inplace=True)\n",
       "    (pool): AdaptiveAvgPool3d(output_size=1)\n",
       "    (flatten): Flatten(start_dim=1, end_dim=-1)\n",
       "    (out): Linear(in_features=1024, out_features=3, bias=True)\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.load_state_dict(torch.load(\"best_metric_model_classification3d_array.pth\"))\n",
    "model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = []\n",
    "ground_truth = []\n",
    "\n",
    "for test_data in test_loader:\n",
    "    test_images, test_labels = test_data[0].to(device), test_data[1].to(device)\n",
    "    with torch.no_grad():\n",
    "        test_outputs = model(test_images)\n",
    "        # Perform any post-processing on the test_outputs if needed\n",
    "        predictions.extend(test_outputs.cpu().numpy())  # Assuming you want to store the predictions\n",
    "\n",
    "    ground_truth.extend(test_labels.cpu().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 0.8576\n"
     ]
    }
   ],
   "source": [
    "# Convert the list of predictions and ground truth to numpy arrays if not already done\n",
    "predictions = np.array(predictions)\n",
    "ground_truth = np.array(ground_truth)\n",
    "\n",
    "accuracy = accuracy_score(ground_truth, predictions.argmax(axis=1))\n",
    "print(f\"Test Accuracy: {accuracy:.4f}\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mri-classifier",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}