DeepMatrixCapsules_on_Pavia.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DZU9nS-J714K"
   },
   "source": [
    "# **EM Routing on Pavia University**\n",
    "### **Parameters:**\n",
    "\n",
    "WindowSize: 19\n",
    "\n",
    "No. of PCA Components: 20\n",
    "\n",
    "No. of epochs: 50\n",
    "\n",
    "Batch size: 64\n",
    "\n",
    "Learning Rate: 3e-3\n",
    "\n",
    "EM Iterations: 2\n",
    "\n",
    "Architecture: [64, 32, 16, 16, 9]\n",
    "\n",
    "Percentage of training data used: 50% (21388 samples)\n",
    "\n",
    "Percentage of validation data: 10% (4277 samples)\n",
    "\n",
    "Percentage of testing data: 40% (17111 samples)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1lo7Iv6t8c7q"
   },
   "source": [
    "## **Imports & Initialization**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "CyNgS4dq353c"
   },
   "outputs": [],
   "source": [
    "from datetime import datetime\n",
    "import math\n",
    "from operator import truediv\n",
    "import os\n",
    "import pickle\n",
    "import random\n",
    "import time\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "import scipy.io as sio\n",
    "\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score\n",
    "from sklearn.metrics import classification_report, cohen_kappa_score\n",
    "\n",
    "import spectral\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from torchvision import datasets, transforms\n",
    "from torch.nn.modules.loss import _Loss\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QRXMOdRD70RL",
    "outputId": "351e0e30-c7c7-44a9-fdef-5db3ed78cf48"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:0\n"
     ]
    }
   ],
   "source": [
    "seed = 2\n",
    "\n",
    "# Making the model deterministic\n",
    "\n",
    "def set_seed(seed):\n",
    "    torch.manual_seed(seed)\n",
    "    torch.cuda.manual_seed_all(seed)\n",
    "    torch.backends.cudnn.deterministic = True\n",
    "    torch.backends.cudnn.benchmark = False\n",
    "    np.random.seed(seed)\n",
    "    random.seed(seed)\n",
    "    os.environ['PYTHONHASHSEED'] = str(seed)\n",
    "\n",
    "set_seed(seed)\n",
    "\n",
    "torch.cuda.empty_cache()\n",
    "\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z7hCepAWyzjQ"
   },
   "source": [
    "## **Parameters**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "dvmHgMALy6SQ"
   },
   "source": [
    "### Dataset Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ZPaayr7SVkp6"
   },
   "outputs": [],
   "source": [
    "test_ratio = 0.8\n",
    "val_and_test_ratio = 0.5\n",
    "windowSize = 19\n",
    "batch_size = 64\n",
    "no_of_PCA_components = 20"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Gx16jdZty2lL"
   },
   "source": [
    "### Model Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "TtR6i2Df4eIg"
   },
   "outputs": [],
   "source": [
    "# model architecture\n",
    "A, B, C, D = 64, 32, 16, 16\n",
    "\n",
    "# EM Iterations\n",
    "em_iters = 2\n",
    "\n",
    "# PCA Componenets\n",
    "pca_components = 20\n",
    "\n",
    "# Epochs\n",
    "epochs = 50\n",
    "\n",
    "# Learning rate\n",
    "lr = 3e-3\n",
    "\n",
    "# Weight Decay\n",
    "wd = 2e-7\n",
    "\n",
    "# Test Intervals\n",
    "test_intervals = 1\n",
    "\n",
    "# Log Interval\n",
    "log_interval = 10\n",
    "\n",
    "# No. of classes\n",
    "num_class = 9"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "QeXpt6io3znQ"
   },
   "source": [
    "## **Loading Data**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cowVRmhKVe3B"
   },
   "source": [
    "#### Defining Required Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "VhxVJKj8Vohj"
   },
   "outputs": [],
   "source": [
    "def loadData():\n",
    "    images = sio.loadmat('Data/Pavia/PaviaU.mat')['paviaU']\n",
    "    labels = sio.loadmat('Data/Pavia/PaviaU_gt.mat')['paviaU_gt']\n",
    "    return images, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "AkWk5WP6VqsP"
   },
   "outputs": [],
   "source": [
    "def applyPCA(X, n_components=no_of_PCA_components, seed=seed):\n",
    "    newX = np.reshape(X, (-1, X.shape[2]))\n",
    "    pca = PCA(n_components=n_components, whiten=True, random_state=seed)\n",
    "    newX = pca.fit_transform(newX)\n",
    "    newX = np.reshape(newX, (X.shape[0], X.shape[1], n_components))\n",
    "    return newX"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "lUMwqDrtVs0F"
   },
   "outputs": [],
   "source": [
    "def padWithZeros(X, margin=2):\n",
    "    newX = np.zeros((X.shape[0] + 2*margin, X.shape[1] + 2*margin, X.shape[2]))\n",
    "    x_offset = margin\n",
    "    y_offset = margin\n",
    "    newX [x_offset : X.shape[0] + x_offset, y_offset : X.shape[1] + y_offset, :] = X\n",
    "    return newX"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "fcj3-RsNVuvl"
   },
   "outputs": [],
   "source": [
    "def createImageCubes(X, y, windowSize=windowSize, removeZeroLabels=True):\n",
    "    margin = int((windowSize-1)/2)\n",
    "    zeroPaddedX = padWithZeros(X, margin=margin)\n",
    "    \n",
    "    patchesData = np.zeros((X.shape[0]*X.shape[1], windowSize, windowSize, X.shape[2]))\n",
    "    patchesLabels = np.zeros((X.shape[0] * X.shape[1]))\n",
    "    patchIndex = 0\n",
    "    \n",
    "    for r in range(margin, zeroPaddedX.shape[0] - margin):\n",
    "        for c in range(margin, zeroPaddedX.shape[1] - margin):\n",
    "            patch = zeroPaddedX[r - margin: r + margin + 1, c - margin:c + margin + 1]\n",
    "            patchesData[patchIndex, :, :, :] = patch\n",
    "            patchesLabels[patchIndex] = y[r-margin, c-margin]\n",
    "            patchIndex += 1\n",
    "\n",
    "    if removeZeroLabels:\n",
    "        patchesData = patchesData[patchesLabels > 0, :, :, :]\n",
    "        patchesLabels = patchesLabels[patchesLabels > 0]\n",
    "        patchesLabels -= 1\n",
    "    \n",
    "    return patchesData, patchesLabels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "_zPDW-7XVxXl"
   },
   "outputs": [],
   "source": [
    "def dataSplit(X, y, val_and_test_ratio=val_and_test_ratio, test_ratio=test_ratio, random_state=seed):\n",
    "    X_train, X_val_and_test, y_train, y_val_and_test = train_test_split(X, y, test_size=val_and_test_ratio, random_state=random_state, stratify=y)\n",
    "    X_val, X_test, y_val, y_test = train_test_split(X_val_and_test, y_val_and_test, test_size=test_ratio, random_state=random_state, stratify=y_val_and_test)\n",
    "    return X_train, y_train, X_val, y_val, X_test, y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "WyxuQV9OV0LS"
   },
   "outputs": [],
   "source": [
    "def convertToDataloader(X_train, y_train, X_val, y_val, X_test, y_test, batch_size=batch_size, K=no_of_PCA_components):\n",
    "\n",
    "    X_train = torch.Tensor(X_train)\n",
    "    y_train = torch.Tensor(y_train)\n",
    "    X_val = torch.Tensor(X_val)\n",
    "    y_val = torch.Tensor(y_val)\n",
    "    X_test = torch.Tensor(X_test)\n",
    "    y_test = torch.Tensor(y_test)\n",
    "\n",
    "    train_dataset = torch.utils.data.TensorDataset(X_train, y_train)\n",
    "    val_dataset = torch.utils.data.TensorDataset(X_val, y_val)\n",
    "    test_dataset = torch.utils.data.TensorDataset(X_test, y_test)\n",
    "\n",
    "    train_dataloader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
    "    val_dataloader = torch.utils.data.DataLoader(val_dataset, batch_size=batch_size)\n",
    "    test_dataloader = torch.utils.data.DataLoader(test_dataset, batch_size=batch_size)\n",
    "\n",
    "    return train_dataloader, val_dataloader, test_dataloader"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pQNIH-ClV26m"
   },
   "source": [
    "#### Loading Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "X3X-GTxKV2qx",
    "outputId": "35056aa9-2514-47fd-9667-4148b44df443"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((610, 340, 103), (610, 340))"
      ]
     },
     "execution_count": 15,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X, y = loadData()\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "AS6ZWMX5V6gI",
    "outputId": "c2dffd0d-2032-4609-a250-ebb940c5ec58"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(610, 340, 20)"
      ]
     },
     "execution_count": 16,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = applyPCA(X)\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4rglm_OIV9Nl",
    "outputId": "cd3681ce-35ff-4a0e-c92e-68c3dc7dd067"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((42776, 19, 19, 20), (42776,))"
      ]
     },
     "execution_count": 17,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X, y = createImageCubes(X, y)\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "I9Ez4Ls8V_If",
    "outputId": "cd6f5eb8-2f37-4bfd-a6c0-161b6732659f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((42776, 20, 19, 19), (42776,))"
      ]
     },
     "execution_count": 18,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = X.transpose(0, 3, 1, 2)\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "JVCEEFM6WAvs",
    "outputId": "262129c6-cf22-49f0-d668-b2a22befe259"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X Train:  (21388, 20, 19, 19)\n",
      "Y Train:  (21388,)\n",
      "X Val:  (4277, 20, 19, 19)\n",
      "Y Val:  (4277,)\n",
      "X Test:  (17111, 20, 19, 19)\n",
      "Y Test:  (17111,)\n"
     ]
    }
   ],
   "source": [
    "X_train, y_train, X_val, y_val, X_test, y_test = dataSplit(X, y)\n",
    "print(\"X Train: \", X_train.shape)\n",
    "print(\"Y Train: \", y_train.shape)\n",
    "print(\"X Val: \", X_val.shape)\n",
    "print(\"Y Val: \", y_val.shape)\n",
    "print(\"X Test: \", X_test.shape)\n",
    "print(\"Y Test: \", y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "a9E56ZwIWC6R"
   },
   "outputs": [],
   "source": [
    "train_loader, val_loader, test_loader = convertToDataloader(X_train, y_train, X_val, y_val, X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Mw6iBRJpWETi"
   },
   "outputs": [],
   "source": [
    "dataloaders = {'train': train_loader, 'valid': val_loader,'test': test_loader}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0D_5INAM4u33"
   },
   "source": [
    "## **Model**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CBmkAG3_43c-"
   },
   "source": [
    "#### Capsule Network Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Qp1hds7dy_9v"
   },
   "outputs": [],
   "source": [
    "class PrimaryCaps(nn.Module):\n",
    "    \n",
    "    def __init__(self, A=32, B=32, K=1, P=4, stride=1):\n",
    "        super(PrimaryCaps, self).__init__()\n",
    "        self.pose = nn.Conv2d(in_channels=A, out_channels=B*P*P,\n",
    "                            kernel_size=K, stride=stride, bias=True)\n",
    "        self.a = nn.Conv2d(in_channels=A, out_channels=B,\n",
    "                            kernel_size=K, stride=stride, bias=True)\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "\n",
    "    def forward(self, x):\n",
    "        p = self.pose(x)\n",
    "        a = self.a(x)\n",
    "        a = self.sigmoid(a)\n",
    "        out = torch.cat([p, a], dim=1)\n",
    "        out = out.permute(0, 2, 3, 1)\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Ey4FeNfp42l7"
   },
   "outputs": [],
   "source": [
    "class ConvCaps(nn.Module):\n",
    "\n",
    "    def __init__(self, B=32, C=32, K=3, P=4, stride=2, iters=3,\n",
    "                 coor_add=False, w_shared=False):\n",
    "        super(ConvCaps, self).__init__()\n",
    "        self.B = B\n",
    "        self.C = C\n",
    "        self.K = K\n",
    "        self.P = P\n",
    "        self.psize = P*P\n",
    "        self.stride = stride\n",
    "        self.iters = iters\n",
    "        self.coor_add = coor_add\n",
    "        self.w_shared = w_shared\n",
    "        self.eps = 1e-8\n",
    "        self._lambda = 1e-03\n",
    "        self.ln_2pi = torch.cuda.FloatTensor(1).fill_(math.log(2*math.pi))\n",
    "\n",
    "        self.beta_u = nn.Parameter(torch.zeros(C))\n",
    "        self.beta_a = nn.Parameter(torch.zeros(C))\n",
    "\n",
    "        self.weights = nn.Parameter(torch.randn(1, K*K*B, C, P, P))\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "        self.softmax = nn.Softmax(dim=2)\n",
    "\n",
    "    def m_step(self, a_in, r, v, eps, b, B, C, psize):\n",
    "\n",
    "        r = r * a_in\n",
    "        r = r / (r.sum(dim=2, keepdim=True) + eps)\n",
    "        r_sum = r.sum(dim=1, keepdim=True)\n",
    "        coeff = r / (r_sum + eps)\n",
    "        coeff = coeff.view(b, B, C, 1)\n",
    "\n",
    "        mu = torch.sum(coeff * v, dim=1, keepdim=True)\n",
    "        sigma_sq = torch.sum(coeff * (v - mu)**2, dim=1, keepdim=True) + eps\n",
    "\n",
    "        r_sum = r_sum.view(b, C, 1)\n",
    "        sigma_sq = sigma_sq.view(b, C, psize)\n",
    "        cost_h = (self.beta_u.view(C, 1) + torch.log(sigma_sq.sqrt())) * r_sum\n",
    "\n",
    "        a_out = self.sigmoid(self._lambda*(self.beta_a - cost_h.sum(dim=2)))\n",
    "        sigma_sq = sigma_sq.view(b, 1, C, psize)\n",
    "\n",
    "        return a_out, mu, sigma_sq\n",
    "\n",
    "    def e_step(self, mu, sigma_sq, a_out, v, eps, b, C):\n",
    "\n",
    "        ln_p_j_h = -1. * (v - mu)**2 / (2 * sigma_sq) \\\n",
    "                    - torch.log(sigma_sq.sqrt()) \\\n",
    "                    - 0.5*self.ln_2pi\n",
    "\n",
    "        ln_ap = ln_p_j_h.sum(dim=3) + torch.log(a_out.view(b, 1, C))\n",
    "        r = self.softmax(ln_ap)\n",
    "        return r\n",
    "\n",
    "    def caps_em_routing(self, v, a_in, C, eps):\n",
    "\n",
    "        b, B, c, psize = v.shape\n",
    "        assert c == C\n",
    "        assert (b, B, 1) == a_in.shape\n",
    "\n",
    "        r = torch.cuda.FloatTensor(b, B, C).fill_(1./C)\n",
    "        for iter_ in range(self.iters):\n",
    "            a_out, mu, sigma_sq = self.m_step(a_in, r, v, eps, b, B, C, psize)\n",
    "            if iter_ < self.iters - 1:\n",
    "                r = self.e_step(mu, sigma_sq, a_out, v, eps, b, C)\n",
    "\n",
    "        return mu, a_out\n",
    "\n",
    "    def add_pathes(self, x, B, K, psize, stride):\n",
    "\n",
    "        b, h, w, c = x.shape\n",
    "        assert h == w\n",
    "        assert c == B*(psize+1)\n",
    "        oh = ow = int(((h - K )/stride)+ 1)\n",
    "        idxs = [[(h_idx + k_idx) \\\n",
    "                for k_idx in range(0, K)] \\\n",
    "                for h_idx in range(0, h - K + 1, stride)]\n",
    "        x = x[:, idxs, :, :]\n",
    "        x = x[:, :, :, idxs, :]\n",
    "        x = x.permute(0, 1, 3, 2, 4, 5).contiguous()\n",
    "        return x, oh, ow\n",
    "\n",
    "    def transform_view(self, x, w, C, P, w_shared=False):\n",
    "\n",
    "        b, B, psize = x.shape\n",
    "        assert psize == P*P\n",
    "\n",
    "        x = x.view(b, B, 1, P, P)\n",
    "        if w_shared:\n",
    "            hw = int(B / w.size(1))\n",
    "            w = w.repeat(1, hw, 1, 1, 1)\n",
    "\n",
    "        w = w.repeat(b, 1, 1, 1, 1)\n",
    "        x = x.repeat(1, 1, C, 1, 1)\n",
    "        v = torch.matmul(x, w)\n",
    "        v = v.view(b, B, C, P*P)\n",
    "        return v\n",
    "\n",
    "    def add_coord(self, v, b, h, w, B, C, psize):\n",
    "\n",
    "        assert h == w\n",
    "        v = v.view(b, h, w, B, C, psize)\n",
    "        coor = torch.arange(h, dtype=torch.float32) / h\n",
    "        coor_h = torch.cuda.FloatTensor(1, h, 1, 1, 1, self.psize).fill_(0.)\n",
    "        coor_w = torch.cuda.FloatTensor(1, 1, w, 1, 1, self.psize).fill_(0.)\n",
    "        coor_h[0, :, 0, 0, 0, 0] = coor\n",
    "        coor_w[0, 0, :, 0, 0, 1] = coor\n",
    "        v = v + coor_h + coor_w\n",
    "        v = v.view(b, h*w*B, C, psize)\n",
    "        return v\n",
    "\n",
    "    def forward(self, x):\n",
    "        b, h, w, c = x.shape\n",
    "        if not self.w_shared:\n",
    "            x, oh, ow = self.add_pathes(x, self.B, self.K, self.psize, self.stride)\n",
    "\n",
    "            p_in = x[:, :, :, :, :, :self.B*self.psize].contiguous()\n",
    "            a_in = x[:, :, :, :, :, self.B*self.psize:].contiguous()\n",
    "            p_in = p_in.view(b*oh*ow, self.K*self.K*self.B, self.psize)\n",
    "            a_in = a_in.view(b*oh*ow, self.K*self.K*self.B, 1)\n",
    "            v = self.transform_view(p_in, self.weights, self.C, self.P)\n",
    "\n",
    "            p_out, a_out = self.caps_em_routing(v, a_in, self.C, self.eps)\n",
    "            p_out = p_out.view(b, oh, ow, self.C*self.psize)\n",
    "            a_out = a_out.view(b, oh, ow, self.C)\n",
    "            out = torch.cat([p_out, a_out], dim=3)\n",
    "        else:\n",
    "            assert c == self.B*(self.psize+1)\n",
    "            assert 1 == self.K\n",
    "            assert 1 == self.stride\n",
    "            p_in = x[:, :, :, :self.B*self.psize].contiguous()\n",
    "            p_in = p_in.view(b, h*w*self.B, self.psize)\n",
    "            a_in = x[:, :, :, self.B*self.psize:].contiguous()\n",
    "            a_in = a_in.view(b, h*w*self.B, 1)\n",
    "\n",
    "            v = self.transform_view(p_in, self.weights, self.C, self.P, self.w_shared)\n",
    "\n",
    "            if self.coor_add:\n",
    "                v = self.add_coord(v, b, h, w, self.B, self.C, self.psize)\n",
    "\n",
    "            _, out = self.caps_em_routing(v, a_in, self.C, self.eps)\n",
    "\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "GKbBGB1FzGPw"
   },
   "outputs": [],
   "source": [
    "class CapsNet(nn.Module):\n",
    "\n",
    "    def __init__(self, A=32, B=32, C=32, D=32, E=10, K=3, P=4, iters=3):\n",
    "        super(CapsNet, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(in_channels=pca_components, out_channels=A,\n",
    "                               kernel_size=5, stride=2, padding=2)\n",
    "        self.bn1 = nn.BatchNorm2d(num_features=A, eps=0.001,\n",
    "                                 momentum=0.1, affine=True)\n",
    "        self.relu1 = nn.ReLU(inplace=False)\n",
    "        self.primary_caps = PrimaryCaps(A, B, 1, P, stride=1)\n",
    "        self.conv_caps1 = ConvCaps(B, C, K, P, stride=2, iters=iters)\n",
    "        self.conv_caps2 = ConvCaps(C, D, K, P, stride=1, iters=iters)\n",
    "        self.class_caps = ConvCaps(D, E, 1, P, stride=1, iters=iters,\n",
    "                                        coor_add=True, w_shared=True)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.conv1(x)\n",
    "        x = self.bn1(x)\n",
    "        x = self.relu1(x)\n",
    "        x = self.primary_caps(x)\n",
    "        x = self.conv_caps1(x)\n",
    "        x = self.conv_caps2(x)\n",
    "        x = self.class_caps(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4ypxKKRG5hvO"
   },
   "source": [
    "#### Loss Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "4J7NoS985UHc"
   },
   "outputs": [],
   "source": [
    "class SpreadLoss(_Loss):\n",
    "\n",
    "    def __init__(self, m_min=0.2, m_max=0.9, num_class=9):\n",
    "        super(SpreadLoss, self).__init__()\n",
    "        self.m_min = m_min\n",
    "        self.m_max = m_max\n",
    "        self.num_class = num_class\n",
    "\n",
    "    def forward(self, x, target, r):\n",
    "        b, E = x.shape\n",
    "        assert E == self.num_class\n",
    "        margin = self.m_min + (self.m_max - self.m_min)*r\n",
    "\n",
    "        at = torch.cuda.FloatTensor(b).fill_(0)\n",
    "        for i, a in enumerate(target):\n",
    "            lb = int(a.item())\n",
    "            at[i] = x[i][lb]\n",
    "        at = at.view(b, 1).repeat(1, E)\n",
    "\n",
    "        zeros = x.new_zeros(x.shape)\n",
    "        loss = torch.max(margin - (at - x), zeros)\n",
    "        loss = loss**2\n",
    "        loss = loss.sum() / b - margin**2\n",
    "\n",
    "        return loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "05AZSkZ16Jxp"
   },
   "source": [
    "## **Training Setup**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "suwWgu2z6rg8"
   },
   "source": [
    "#### Data Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "kIgNfsLi5jlU"
   },
   "outputs": [],
   "source": [
    "train_loader = dataloaders['train']\n",
    "valid_loader = dataloaders['valid']\n",
    "test_loader = dataloaders['test']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aq1YU_XD7Cco"
   },
   "source": [
    "#### Accuracy Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "w7v8XF9M6uQR"
   },
   "outputs": [],
   "source": [
    "def accuracy(output, target):\n",
    "    y_pred = output.argmax(dim=1).float()\n",
    "    res = accuracy_score(target.to('cpu'), y_pred.to('cpu')) * 100\n",
    "    return res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hvXyaGUH7HLo"
   },
   "source": [
    "#### Learning Rate Decay function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EW1wGZrS7EQA"
   },
   "outputs": [],
   "source": [
    "def exp_lr_decay(optimizer, global_step, init_lr = 3e-3, decay_steps = 20000,\n",
    "                                        decay_rate = 0.96, lr_clip = 3e-3 ,staircase=False):\n",
    "        \n",
    "    if staircase:\n",
    "        lr = (init_lr * decay_rate**(global_step // decay_steps)) \n",
    "    else:\n",
    "        lr = (init_lr * decay_rate**(global_step / decay_steps)) \n",
    "    \n",
    "    for param_group in optimizer.param_groups:\n",
    "        param_group['lr'] = lr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Rkhe68NX7P8p"
   },
   "source": [
    "#### Average Meter Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "WdmfQJTt7KG8"
   },
   "outputs": [],
   "source": [
    "class AverageMeter(object):\n",
    "\n",
    "    def __init__(self):\n",
    "        self.reset()\n",
    "\n",
    "    def reset(self):\n",
    "        self.val = 0\n",
    "        self.avg = 0\n",
    "        self.sum = 0\n",
    "        self.count = 0\n",
    "\n",
    "    def update(self, val, n=1):\n",
    "        self.val = val\n",
    "        self.sum += val * n\n",
    "        self.count += n\n",
    "        self.avg = self.sum / self.count"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6zWNMYc87U-4"
   },
   "source": [
    "#### Train Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Gvsb63dX7R_X"
   },
   "outputs": [],
   "source": [
    "def train(train_loader, model, criterion, optimizer, epoch, device):\n",
    "    batch_time = AverageMeter()\n",
    "    data_time = AverageMeter()\n",
    "\n",
    "    model.train()\n",
    "    train_len = len(train_loader)\n",
    "    epoch_acc = 0\n",
    "    end = time.time()\n",
    "\n",
    "    for batch_idx, (data, target) in enumerate(train_loader):\n",
    "        data_time.update(time.time() - end)\n",
    "\n",
    "        data, target = data.to(device), target.to(device)\n",
    "        optimizer.zero_grad()\n",
    "        output = model(data)\n",
    "        r = (1.*batch_idx + (epoch-1)*train_len) / (epochs*train_len)\n",
    "        loss = criterion(output, target, r)\n",
    "        acc = accuracy(output, target)\n",
    "        \n",
    "        global_step = (batch_idx+1) + (epoch - 1) * len(train_loader) \n",
    "        exp_lr_decay(optimizer=optimizer, global_step=global_step)\n",
    "        \n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        batch_time.update(time.time() - end)\n",
    "        end = time.time()\n",
    "\n",
    "        epoch_acc += acc\n",
    "        if batch_idx % log_interval == 0:\n",
    "            print('Train Epoch: {}\\t[{}/{} ({:.0f}%)]\\t'\n",
    "                  'Loss: {:.6f}\\tAccuracy: {:.6f}\\t'\n",
    "                  'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n",
    "                  'Data {data_time.val:.3f} ({data_time.avg:.3f})'.format(\n",
    "                  epoch, batch_idx * len(data), len(train_loader.dataset),\n",
    "                  100. * batch_idx / len(train_loader),\n",
    "                  loss.item(), acc,\n",
    "                  batch_time=batch_time, data_time=data_time))\n",
    "            \n",
    "    return epoch_acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "iZzi5Vn27d-X"
   },
   "source": [
    "#### Test Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "49eSAz-I7dTP"
   },
   "outputs": [],
   "source": [
    "def test(test_loader, model, criterion, device, test=False):\n",
    "    model.eval()\n",
    "    test_loss = 0\n",
    "    acc = 0\n",
    "    test_len = len(test_loader)\n",
    "    with torch.no_grad():\n",
    "        for data, target in test_loader:\n",
    "            data, target = data.to(device), target.to(device)\n",
    "            output = model(data)\n",
    "            test_loss += criterion(output, target, r=1).item()\n",
    "            acc += accuracy(output, target)\n",
    "\n",
    "    test_loss /= test_len\n",
    "    acc /= test_len\n",
    "    if test:\n",
    "        print('\\nTest set: Average loss: {:.6f}, Accuracy: {:.6f} \\n'.format(\n",
    "        test_loss, acc))\n",
    "    else:\n",
    "        print('\\nVal set: Average loss: {:.6f}, Accuracy: {:.6f} \\n'.format(\n",
    "            test_loss, acc))\n",
    "\n",
    "    return test_loss, acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5hRWF2apsMzc"
   },
   "source": [
    "## **Training Process**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SGwX3eimsZmA"
   },
   "source": [
    "#### Model Initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "8_YYc6YxsXZV",
    "outputId": "50876aeb-4b7a-45c3-9bb5-09f6c7331783"
   },
   "outputs": [],
   "source": [
    "model = CapsNet(A=A, B=B, C=C, D=D, E=num_class,\n",
    "                    iters=em_iters).to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rZwvqJ_4ZZaI"
   },
   "source": [
    "#### Model Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "mk54JHDaZY6l",
    "outputId": "1161a92d-8afa-41b1-eaa6-dbf6669a7c91"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "conv1.weight : 64x20x5x5 = 32000\n",
      "conv1.bias : 64\n",
      "bn1.weight : 64\n",
      "bn1.bias : 64\n",
      "primary_caps.pose.weight : 512x64x1x1 = 32768\n",
      "primary_caps.pose.bias : 512\n",
      "primary_caps.a.weight : 32x64x1x1 = 2048\n",
      "primary_caps.a.bias : 32\n",
      "conv_caps1.beta_u : 16\n",
      "conv_caps1.beta_a : 16\n",
      "conv_caps1.weights : 1x288x16x4x4 = 73728\n",
      "conv_caps2.beta_u : 16\n",
      "conv_caps2.beta_a : 16\n",
      "conv_caps2.weights : 1x144x16x4x4 = 36864\n",
      "class_caps.beta_u : 9\n",
      "class_caps.beta_a : 9\n",
      "class_caps.weights : 1x16x9x4x4 = 2304\n",
      "\n",
      " Total Trainable Parameters:  180530\n"
     ]
    }
   ],
   "source": [
    "def model_summary(model):\n",
    "    total_param = 0\n",
    "    for name, param in model.named_parameters():\n",
    "        if param.requires_grad:\n",
    "            num_param = np.prod(param.size())\n",
    "            if param.dim() > 1:\n",
    "                print(name, ':', 'x'.join(str(x) for x in list(param.size())), '=', num_param)\n",
    "            else:\n",
    "                print(name, ':', num_param)\n",
    "            total_param += num_param\n",
    "    print(\"\\n Total Trainable Parameters: \", total_param)\n",
    "\n",
    "model_summary(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "R0JocqrGZmIN"
   },
   "source": [
    "#### Training Time!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "fEPfkoFhZva8"
   },
   "outputs": [],
   "source": [
    "criterion = SpreadLoss(num_class=num_class, m_min=0.2, m_max=0.5)\n",
    "optimizer = optim.Adam(model.parameters(), lr = lr, weight_decay=wd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ayyIRBgr7goB",
    "outputId": "2cc08ddb-00c5-406c-c5fd-b93921c966aa"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Val set: Average loss: 2.058935, Accuracy: 9.888940 \n",
      "\n",
      "Train Epoch: 1\t[0/21388 (0%)]\tLoss: 0.404170\tAccuracy: 1.562500\tTime 0.397 (0.397)\tData 0.005 (0.005)\n",
      "Train Epoch: 1\t[640/21388 (3%)]\tLoss: 0.162319\tAccuracy: 46.875000\tTime 0.347 (0.354)\tData 0.001 (0.003)\n",
      "Train Epoch: 1\t[1280/21388 (6%)]\tLoss: 0.106031\tAccuracy: 68.750000\tTime 0.348 (0.351)\tData 0.002 (0.002)\n",
      "Train Epoch: 1\t[1920/21388 (9%)]\tLoss: 0.074427\tAccuracy: 70.312500\tTime 0.350 (0.350)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[2560/21388 (12%)]\tLoss: 0.044155\tAccuracy: 79.687500\tTime 0.349 (0.350)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[3200/21388 (15%)]\tLoss: 0.021808\tAccuracy: 92.187500\tTime 0.350 (0.350)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[3840/21388 (18%)]\tLoss: 0.013735\tAccuracy: 93.750000\tTime 0.352 (0.350)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[4480/21388 (21%)]\tLoss: 0.018326\tAccuracy: 92.187500\tTime 0.355 (0.350)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[5120/21388 (24%)]\tLoss: 0.015397\tAccuracy: 92.187500\tTime 0.352 (0.351)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[5760/21388 (27%)]\tLoss: 0.008214\tAccuracy: 95.312500\tTime 0.354 (0.351)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[6400/21388 (30%)]\tLoss: 0.011634\tAccuracy: 95.312500\tTime 0.356 (0.351)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[7040/21388 (33%)]\tLoss: 0.005276\tAccuracy: 98.437500\tTime 0.355 (0.352)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[7680/21388 (36%)]\tLoss: 0.008108\tAccuracy: 98.437500\tTime 0.356 (0.352)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[8320/21388 (39%)]\tLoss: 0.006304\tAccuracy: 98.437500\tTime 0.357 (0.352)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[8960/21388 (42%)]\tLoss: 0.002776\tAccuracy: 100.000000\tTime 0.356 (0.353)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[9600/21388 (45%)]\tLoss: 0.004384\tAccuracy: 98.437500\tTime 0.358 (0.353)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[10240/21388 (48%)]\tLoss: 0.007712\tAccuracy: 96.875000\tTime 0.357 (0.353)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[10880/21388 (51%)]\tLoss: 0.005817\tAccuracy: 95.312500\tTime 0.359 (0.354)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[11520/21388 (54%)]\tLoss: 0.002108\tAccuracy: 98.437500\tTime 0.366 (0.354)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[12160/21388 (57%)]\tLoss: 0.005253\tAccuracy: 96.875000\tTime 0.363 (0.354)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[12800/21388 (60%)]\tLoss: 0.001348\tAccuracy: 100.000000\tTime 0.361 (0.355)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[13440/21388 (63%)]\tLoss: 0.001827\tAccuracy: 100.000000\tTime 0.361 (0.355)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[14080/21388 (66%)]\tLoss: 0.017784\tAccuracy: 93.750000\tTime 0.359 (0.355)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[14720/21388 (69%)]\tLoss: 0.002395\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[15360/21388 (72%)]\tLoss: 0.002720\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[16000/21388 (75%)]\tLoss: 0.002138\tAccuracy: 98.437500\tTime 0.358 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[16640/21388 (78%)]\tLoss: 0.005979\tAccuracy: 98.437500\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[17280/21388 (81%)]\tLoss: 0.001545\tAccuracy: 98.437500\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[17920/21388 (84%)]\tLoss: 0.002223\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[18560/21388 (87%)]\tLoss: 0.002288\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[19200/21388 (90%)]\tLoss: 0.003852\tAccuracy: 98.437500\tTime 0.354 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[19840/21388 (93%)]\tLoss: 0.003131\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[20480/21388 (96%)]\tLoss: 0.001854\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 1\t[21120/21388 (99%)]\tLoss: 0.006719\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.296216, Accuracy: 99.178928 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 2\t[0/21388 (0%)]\tLoss: 0.002328\tAccuracy: 98.437500\tTime 0.355 (0.355)\tData 0.002 (0.002)\n",
      "Train Epoch: 2\t[640/21388 (3%)]\tLoss: 0.001516\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[1280/21388 (6%)]\tLoss: 0.000560\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[1920/21388 (9%)]\tLoss: 0.003782\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[2560/21388 (12%)]\tLoss: 0.001588\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[3200/21388 (15%)]\tLoss: 0.000586\tAccuracy: 100.000000\tTime 0.364 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[3840/21388 (18%)]\tLoss: 0.001610\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[4480/21388 (21%)]\tLoss: 0.003379\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[5120/21388 (24%)]\tLoss: 0.006376\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[5760/21388 (27%)]\tLoss: 0.000557\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[6400/21388 (30%)]\tLoss: 0.002827\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[7040/21388 (33%)]\tLoss: 0.001279\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 2\t[7680/21388 (36%)]\tLoss: 0.005235\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[8320/21388 (39%)]\tLoss: 0.000221\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[8960/21388 (42%)]\tLoss: 0.002738\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[9600/21388 (45%)]\tLoss: 0.000397\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[10240/21388 (48%)]\tLoss: 0.000699\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[10880/21388 (51%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[11520/21388 (54%)]\tLoss: 0.004485\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[12160/21388 (57%)]\tLoss: 0.001177\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[12800/21388 (60%)]\tLoss: 0.002388\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[13440/21388 (63%)]\tLoss: 0.006068\tAccuracy: 95.312500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[14080/21388 (66%)]\tLoss: 0.001360\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[14720/21388 (69%)]\tLoss: 0.001759\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 2\t[15360/21388 (72%)]\tLoss: 0.001739\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[16000/21388 (75%)]\tLoss: 0.000342\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[16640/21388 (78%)]\tLoss: 0.000820\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[17280/21388 (81%)]\tLoss: 0.003077\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[17920/21388 (84%)]\tLoss: 0.004845\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[18560/21388 (87%)]\tLoss: 0.000583\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[19200/21388 (90%)]\tLoss: 0.006437\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[19840/21388 (93%)]\tLoss: 0.005298\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[20480/21388 (96%)]\tLoss: 0.004510\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 2\t[21120/21388 (99%)]\tLoss: 0.001007\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.252826, Accuracy: 99.580224 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 3\t[0/21388 (0%)]\tLoss: 0.000545\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.002 (0.002)\n",
      "Train Epoch: 3\t[640/21388 (3%)]\tLoss: 0.000682\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[1280/21388 (6%)]\tLoss: 0.000219\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[1920/21388 (9%)]\tLoss: 0.001979\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[2560/21388 (12%)]\tLoss: 0.000368\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[3200/21388 (15%)]\tLoss: 0.000479\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[3840/21388 (18%)]\tLoss: 0.000407\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[4480/21388 (21%)]\tLoss: 0.001098\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[5120/21388 (24%)]\tLoss: 0.000424\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[5760/21388 (27%)]\tLoss: 0.001784\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[6400/21388 (30%)]\tLoss: 0.002479\tAccuracy: 98.437500\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[7040/21388 (33%)]\tLoss: 0.000428\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[7680/21388 (36%)]\tLoss: 0.003039\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[8320/21388 (39%)]\tLoss: 0.008143\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[8960/21388 (42%)]\tLoss: 0.000447\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[9600/21388 (45%)]\tLoss: 0.000588\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[10240/21388 (48%)]\tLoss: 0.000233\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[10880/21388 (51%)]\tLoss: 0.001158\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[11520/21388 (54%)]\tLoss: 0.001196\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[12160/21388 (57%)]\tLoss: 0.007545\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[12800/21388 (60%)]\tLoss: 0.004161\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[13440/21388 (63%)]\tLoss: 0.001281\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 3\t[14080/21388 (66%)]\tLoss: 0.005374\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[14720/21388 (69%)]\tLoss: 0.002917\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[15360/21388 (72%)]\tLoss: 0.001528\tAccuracy: 98.437500\tTime 0.359 (0.357)\tData 0.003 (0.001)\n",
      "Train Epoch: 3\t[16000/21388 (75%)]\tLoss: 0.000100\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[16640/21388 (78%)]\tLoss: 0.000294\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[17280/21388 (81%)]\tLoss: 0.001172\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[17920/21388 (84%)]\tLoss: 0.000436\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[18560/21388 (87%)]\tLoss: 0.002043\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[19200/21388 (90%)]\tLoss: 0.000680\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[19840/21388 (93%)]\tLoss: 0.000144\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[20480/21388 (96%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 3\t[21120/21388 (99%)]\tLoss: 0.000099\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.210140, Accuracy: 99.482100 \n",
      "\n",
      "Train Epoch: 4\t[0/21388 (0%)]\tLoss: 0.002233\tAccuracy: 98.437500\tTime 0.359 (0.359)\tData 0.004 (0.004)\n",
      "Train Epoch: 4\t[640/21388 (3%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[1280/21388 (6%)]\tLoss: 0.000081\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[1920/21388 (9%)]\tLoss: 0.000278\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[2560/21388 (12%)]\tLoss: 0.001258\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[3200/21388 (15%)]\tLoss: 0.000138\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[3840/21388 (18%)]\tLoss: 0.000180\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[4480/21388 (21%)]\tLoss: 0.000181\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[5120/21388 (24%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[5760/21388 (27%)]\tLoss: 0.002131\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[6400/21388 (30%)]\tLoss: 0.000456\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[7040/21388 (33%)]\tLoss: 0.002424\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[7680/21388 (36%)]\tLoss: 0.000920\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[8320/21388 (39%)]\tLoss: 0.001729\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[8960/21388 (42%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 4\t[9600/21388 (45%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[10240/21388 (48%)]\tLoss: 0.001220\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[10880/21388 (51%)]\tLoss: 0.000208\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[11520/21388 (54%)]\tLoss: 0.000199\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 4\t[12160/21388 (57%)]\tLoss: 0.002049\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[12800/21388 (60%)]\tLoss: 0.000021\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[13440/21388 (63%)]\tLoss: 0.000333\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[14080/21388 (66%)]\tLoss: 0.000175\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[14720/21388 (69%)]\tLoss: 0.001169\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[15360/21388 (72%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[16000/21388 (75%)]\tLoss: 0.000101\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[16640/21388 (78%)]\tLoss: 0.007135\tAccuracy: 96.875000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[17280/21388 (81%)]\tLoss: 0.001098\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[17920/21388 (84%)]\tLoss: 0.000298\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[18560/21388 (87%)]\tLoss: 0.000812\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[19200/21388 (90%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[19840/21388 (93%)]\tLoss: 0.001161\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[20480/21388 (96%)]\tLoss: 0.000041\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 4\t[21120/21388 (99%)]\tLoss: 0.000157\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.183535, Accuracy: 99.860075 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 5\t[0/21388 (0%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.003 (0.003)\n",
      "Train Epoch: 5\t[640/21388 (3%)]\tLoss: 0.001198\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 5\t[1280/21388 (6%)]\tLoss: 0.000200\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[1920/21388 (9%)]\tLoss: 0.002377\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 5\t[2560/21388 (12%)]\tLoss: 0.000362\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 5\t[3200/21388 (15%)]\tLoss: 0.000793\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[3840/21388 (18%)]\tLoss: 0.000637\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[4480/21388 (21%)]\tLoss: 0.000283\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 5\t[5120/21388 (24%)]\tLoss: 0.000129\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 5\t[5760/21388 (27%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[6400/21388 (30%)]\tLoss: 0.001413\tAccuracy: 98.437500\tTime 0.360 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[7040/21388 (33%)]\tLoss: 0.000201\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[7680/21388 (36%)]\tLoss: 0.000174\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[8320/21388 (39%)]\tLoss: 0.000567\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[8960/21388 (42%)]\tLoss: 0.000212\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[9600/21388 (45%)]\tLoss: 0.002294\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[10240/21388 (48%)]\tLoss: 0.000270\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[10880/21388 (51%)]\tLoss: 0.000636\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[11520/21388 (54%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[12160/21388 (57%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[12800/21388 (60%)]\tLoss: 0.000097\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[13440/21388 (63%)]\tLoss: 0.000376\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[14080/21388 (66%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[14720/21388 (69%)]\tLoss: 0.000172\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[15360/21388 (72%)]\tLoss: 0.000023\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[16000/21388 (75%)]\tLoss: 0.000976\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[16640/21388 (78%)]\tLoss: 0.000630\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[17280/21388 (81%)]\tLoss: 0.000278\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[17920/21388 (84%)]\tLoss: 0.000849\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[18560/21388 (87%)]\tLoss: 0.000340\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[19200/21388 (90%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[19840/21388 (93%)]\tLoss: 0.000106\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[20480/21388 (96%)]\tLoss: 0.001095\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 5\t[21120/21388 (99%)]\tLoss: 0.000140\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.178627, Accuracy: 99.720149 \n",
      "\n",
      "Train Epoch: 6\t[0/21388 (0%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 6\t[640/21388 (3%)]\tLoss: 0.000418\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[1280/21388 (6%)]\tLoss: 0.002443\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[1920/21388 (9%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[2560/21388 (12%)]\tLoss: 0.000884\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[3200/21388 (15%)]\tLoss: 0.000855\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[3840/21388 (18%)]\tLoss: 0.000255\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[4480/21388 (21%)]\tLoss: 0.000377\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[5120/21388 (24%)]\tLoss: 0.002413\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 6\t[5760/21388 (27%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[6400/21388 (30%)]\tLoss: 0.000030\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[7040/21388 (33%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[7680/21388 (36%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[8320/21388 (39%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[8960/21388 (42%)]\tLoss: 0.000066\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[9600/21388 (45%)]\tLoss: 0.001774\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 6\t[10240/21388 (48%)]\tLoss: 0.000085\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[10880/21388 (51%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[11520/21388 (54%)]\tLoss: 0.000664\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[12160/21388 (57%)]\tLoss: 0.000021\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[12800/21388 (60%)]\tLoss: 0.000631\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[13440/21388 (63%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[14080/21388 (66%)]\tLoss: 0.000214\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[14720/21388 (69%)]\tLoss: 0.001573\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[15360/21388 (72%)]\tLoss: 0.002692\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[16000/21388 (75%)]\tLoss: 0.000466\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[16640/21388 (78%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[17280/21388 (81%)]\tLoss: 0.001200\tAccuracy: 100.000000\tTime 0.363 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 6\t[17920/21388 (84%)]\tLoss: 0.000133\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[18560/21388 (87%)]\tLoss: 0.001689\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[19200/21388 (90%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[19840/21388 (93%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[20480/21388 (96%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 6\t[21120/21388 (99%)]\tLoss: 0.000468\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.159782, Accuracy: 99.766791 \n",
      "\n",
      "Train Epoch: 7\t[0/21388 (0%)]\tLoss: 0.000520\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 7\t[640/21388 (3%)]\tLoss: 0.000174\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[1280/21388 (6%)]\tLoss: 0.000403\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[1920/21388 (9%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[2560/21388 (12%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[3200/21388 (15%)]\tLoss: 0.000036\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[3840/21388 (18%)]\tLoss: 0.000374\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 7\t[4480/21388 (21%)]\tLoss: 0.000165\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[5120/21388 (24%)]\tLoss: 0.003599\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[5760/21388 (27%)]\tLoss: 0.001090\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[6400/21388 (30%)]\tLoss: 0.000585\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[7040/21388 (33%)]\tLoss: 0.003203\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[7680/21388 (36%)]\tLoss: 0.000183\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[8320/21388 (39%)]\tLoss: 0.000694\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[8960/21388 (42%)]\tLoss: 0.003625\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[9600/21388 (45%)]\tLoss: 0.000187\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 7\t[10240/21388 (48%)]\tLoss: 0.000430\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[10880/21388 (51%)]\tLoss: 0.000677\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[11520/21388 (54%)]\tLoss: 0.000341\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[12160/21388 (57%)]\tLoss: 0.000857\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[12800/21388 (60%)]\tLoss: 0.000023\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[13440/21388 (63%)]\tLoss: 0.001249\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[14080/21388 (66%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[14720/21388 (69%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[15360/21388 (72%)]\tLoss: 0.000061\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[16000/21388 (75%)]\tLoss: 0.000619\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[16640/21388 (78%)]\tLoss: 0.003253\tAccuracy: 96.875000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[17280/21388 (81%)]\tLoss: 0.001641\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[17920/21388 (84%)]\tLoss: 0.000135\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[18560/21388 (87%)]\tLoss: 0.000178\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[19200/21388 (90%)]\tLoss: 0.001731\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[19840/21388 (93%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[20480/21388 (96%)]\tLoss: 0.000230\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 7\t[21120/21388 (99%)]\tLoss: 0.001549\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.148558, Accuracy: 99.766791 \n",
      "\n",
      "Train Epoch: 8\t[0/21388 (0%)]\tLoss: 0.000490\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 8\t[640/21388 (3%)]\tLoss: 0.000919\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 8\t[1280/21388 (6%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 8\t[1920/21388 (9%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 8\t[2560/21388 (12%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 8\t[3200/21388 (15%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[3840/21388 (18%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[4480/21388 (21%)]\tLoss: 0.000107\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[5120/21388 (24%)]\tLoss: 0.000200\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[5760/21388 (27%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[6400/21388 (30%)]\tLoss: 0.000058\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[7040/21388 (33%)]\tLoss: 0.000631\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[7680/21388 (36%)]\tLoss: 0.000948\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[8320/21388 (39%)]\tLoss: 0.000103\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[8960/21388 (42%)]\tLoss: 0.006939\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[9600/21388 (45%)]\tLoss: 0.000673\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[10240/21388 (48%)]\tLoss: 0.004904\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[10880/21388 (51%)]\tLoss: 0.000157\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[11520/21388 (54%)]\tLoss: 0.002630\tAccuracy: 98.437500\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[12160/21388 (57%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[12800/21388 (60%)]\tLoss: 0.002361\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[13440/21388 (63%)]\tLoss: 0.000713\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[14080/21388 (66%)]\tLoss: 0.000829\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[14720/21388 (69%)]\tLoss: 0.000482\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[15360/21388 (72%)]\tLoss: 0.001968\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[16000/21388 (75%)]\tLoss: 0.000261\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[16640/21388 (78%)]\tLoss: 0.000653\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[17280/21388 (81%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[17920/21388 (84%)]\tLoss: 0.000022\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[18560/21388 (87%)]\tLoss: 0.000112\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[19200/21388 (90%)]\tLoss: 0.000551\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[19840/21388 (93%)]\tLoss: 0.004152\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[20480/21388 (96%)]\tLoss: 0.000143\tAccuracy: 100.000000\tTime 0.361 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 8\t[21120/21388 (99%)]\tLoss: 0.000059\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.129313, Accuracy: 99.743470 \n",
      "\n",
      "Train Epoch: 9\t[0/21388 (0%)]\tLoss: 0.005520\tAccuracy: 98.437500\tTime 0.359 (0.359)\tData 0.004 (0.004)\n",
      "Train Epoch: 9\t[640/21388 (3%)]\tLoss: 0.000028\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[1280/21388 (6%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[1920/21388 (9%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[2560/21388 (12%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[3200/21388 (15%)]\tLoss: 0.000099\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[3840/21388 (18%)]\tLoss: 0.001650\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[4480/21388 (21%)]\tLoss: 0.000051\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[5120/21388 (24%)]\tLoss: 0.000199\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[6400/21388 (30%)]\tLoss: 0.000070\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[7040/21388 (33%)]\tLoss: 0.000063\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[7680/21388 (36%)]\tLoss: 0.000217\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[8320/21388 (39%)]\tLoss: 0.000652\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 9\t[8960/21388 (42%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.361 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[9600/21388 (45%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[10240/21388 (48%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[11520/21388 (54%)]\tLoss: 0.000860\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[12160/21388 (57%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[12800/21388 (60%)]\tLoss: 0.000286\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[13440/21388 (63%)]\tLoss: 0.000502\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[14080/21388 (66%)]\tLoss: 0.000102\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[14720/21388 (69%)]\tLoss: 0.000599\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[15360/21388 (72%)]\tLoss: 0.000711\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[16000/21388 (75%)]\tLoss: 0.001133\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[16640/21388 (78%)]\tLoss: 0.000755\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[17280/21388 (81%)]\tLoss: 0.005280\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[17920/21388 (84%)]\tLoss: 0.000069\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[18560/21388 (87%)]\tLoss: 0.004167\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[19200/21388 (90%)]\tLoss: 0.000078\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[19840/21388 (93%)]\tLoss: 0.000126\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[20480/21388 (96%)]\tLoss: 0.000370\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 9\t[21120/21388 (99%)]\tLoss: 0.000062\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.128088, Accuracy: 99.626866 \n",
      "\n",
      "Train Epoch: 10\t[0/21388 (0%)]\tLoss: 0.001406\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 10\t[640/21388 (3%)]\tLoss: 0.000491\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 10\t[1280/21388 (6%)]\tLoss: 0.001420\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 10\t[1920/21388 (9%)]\tLoss: 0.002530\tAccuracy: 98.437500\tTime 0.358 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 10\t[2560/21388 (12%)]\tLoss: 0.000582\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[3200/21388 (15%)]\tLoss: 0.002665\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[3840/21388 (18%)]\tLoss: 0.009550\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[4480/21388 (21%)]\tLoss: 0.002039\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[5120/21388 (24%)]\tLoss: 0.000250\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[5760/21388 (27%)]\tLoss: 0.000049\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[6400/21388 (30%)]\tLoss: 0.002837\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[7040/21388 (33%)]\tLoss: 0.000164\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[7680/21388 (36%)]\tLoss: 0.000104\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[8320/21388 (39%)]\tLoss: 0.000835\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 10\t[8960/21388 (42%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[9600/21388 (45%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[10240/21388 (48%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[10880/21388 (51%)]\tLoss: 0.001959\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[11520/21388 (54%)]\tLoss: 0.000216\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[12160/21388 (57%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[12800/21388 (60%)]\tLoss: 0.001872\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[13440/21388 (63%)]\tLoss: 0.000207\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[14080/21388 (66%)]\tLoss: 0.001179\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[14720/21388 (69%)]\tLoss: 0.001095\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[15360/21388 (72%)]\tLoss: 0.002060\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[16000/21388 (75%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[16640/21388 (78%)]\tLoss: 0.000385\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[17280/21388 (81%)]\tLoss: 0.007266\tAccuracy: 96.875000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[17920/21388 (84%)]\tLoss: 0.002119\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[18560/21388 (87%)]\tLoss: 0.000417\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[19200/21388 (90%)]\tLoss: 0.001633\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[19840/21388 (93%)]\tLoss: 0.004982\tAccuracy: 98.437500\tTime 0.362 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[20480/21388 (96%)]\tLoss: 0.002038\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 10\t[21120/21388 (99%)]\tLoss: 0.001642\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.116178, Accuracy: 99.836754 \n",
      "\n",
      "Train Epoch: 11\t[0/21388 (0%)]\tLoss: 0.000492\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.003 (0.003)\n",
      "Train Epoch: 11\t[640/21388 (3%)]\tLoss: 0.000290\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 11\t[1280/21388 (6%)]\tLoss: 0.000138\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[1920/21388 (9%)]\tLoss: 0.011483\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[2560/21388 (12%)]\tLoss: 0.000280\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[3200/21388 (15%)]\tLoss: 0.000987\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[3840/21388 (18%)]\tLoss: 0.002326\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[4480/21388 (21%)]\tLoss: 0.000170\tAccuracy: 100.000000\tTime 0.360 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[5120/21388 (24%)]\tLoss: 0.000648\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[5760/21388 (27%)]\tLoss: 0.000053\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[6400/21388 (30%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.361 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[7040/21388 (33%)]\tLoss: 0.003718\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[7680/21388 (36%)]\tLoss: 0.000175\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[8320/21388 (39%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[8960/21388 (42%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[9600/21388 (45%)]\tLoss: 0.000135\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[10240/21388 (48%)]\tLoss: 0.002748\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[10880/21388 (51%)]\tLoss: 0.000246\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[11520/21388 (54%)]\tLoss: 0.000932\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[12160/21388 (57%)]\tLoss: 0.001202\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[12800/21388 (60%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[13440/21388 (63%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[14080/21388 (66%)]\tLoss: 0.001861\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[14720/21388 (69%)]\tLoss: 0.000143\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[15360/21388 (72%)]\tLoss: 0.000986\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[16000/21388 (75%)]\tLoss: 0.000050\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[16640/21388 (78%)]\tLoss: 0.008099\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[17280/21388 (81%)]\tLoss: 0.000317\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[17920/21388 (84%)]\tLoss: 0.000082\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[18560/21388 (87%)]\tLoss: 0.000124\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[19200/21388 (90%)]\tLoss: 0.000021\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 11\t[19840/21388 (93%)]\tLoss: 0.001841\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[20480/21388 (96%)]\tLoss: 0.000221\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 11\t[21120/21388 (99%)]\tLoss: 0.000125\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.110471, Accuracy: 99.813433 \n",
      "\n",
      "Train Epoch: 12\t[0/21388 (0%)]\tLoss: 0.000834\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.004 (0.004)\n",
      "Train Epoch: 12\t[640/21388 (3%)]\tLoss: 0.006277\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[1280/21388 (6%)]\tLoss: 0.000682\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.003 (0.002)\n",
      "Train Epoch: 12\t[1920/21388 (9%)]\tLoss: 0.001374\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[2560/21388 (12%)]\tLoss: 0.000134\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[3200/21388 (15%)]\tLoss: 0.001120\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[3840/21388 (18%)]\tLoss: 0.000427\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[4480/21388 (21%)]\tLoss: 0.000105\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 12\t[5120/21388 (24%)]\tLoss: 0.000079\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[5760/21388 (27%)]\tLoss: 0.000114\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[6400/21388 (30%)]\tLoss: 0.000310\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[7040/21388 (33%)]\tLoss: 0.000050\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[7680/21388 (36%)]\tLoss: 0.001964\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[8320/21388 (39%)]\tLoss: 0.000184\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[8960/21388 (42%)]\tLoss: 0.001282\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[9600/21388 (45%)]\tLoss: 0.000937\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 12\t[10240/21388 (48%)]\tLoss: 0.002460\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 12\t[10880/21388 (51%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[11520/21388 (54%)]\tLoss: 0.000086\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[12160/21388 (57%)]\tLoss: 0.000055\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[12800/21388 (60%)]\tLoss: 0.001264\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[13440/21388 (63%)]\tLoss: 0.002591\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[14080/21388 (66%)]\tLoss: 0.000396\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[14720/21388 (69%)]\tLoss: 0.001975\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[15360/21388 (72%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[16000/21388 (75%)]\tLoss: 0.000357\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[16640/21388 (78%)]\tLoss: 0.003298\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[17280/21388 (81%)]\tLoss: 0.000500\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[17920/21388 (84%)]\tLoss: 0.000774\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[18560/21388 (87%)]\tLoss: 0.000438\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[19200/21388 (90%)]\tLoss: 0.000498\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[19840/21388 (93%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 12\t[20480/21388 (96%)]\tLoss: 0.000986\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 12\t[21120/21388 (99%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.104974, Accuracy: 99.738630 \n",
      "\n",
      "Train Epoch: 13\t[0/21388 (0%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 13\t[640/21388 (3%)]\tLoss: 0.000448\tAccuracy: 100.000000\tTime 0.361 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 13\t[1280/21388 (6%)]\tLoss: 0.000692\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 13\t[1920/21388 (9%)]\tLoss: 0.000073\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[2560/21388 (12%)]\tLoss: 0.001410\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 13\t[3200/21388 (15%)]\tLoss: 0.000184\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[3840/21388 (18%)]\tLoss: 0.000395\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[4480/21388 (21%)]\tLoss: 0.008147\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[5120/21388 (24%)]\tLoss: 0.011696\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[5760/21388 (27%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[6400/21388 (30%)]\tLoss: 0.000425\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[7040/21388 (33%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[7680/21388 (36%)]\tLoss: 0.000036\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[8320/21388 (39%)]\tLoss: 0.000794\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[8960/21388 (42%)]\tLoss: 0.000761\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[9600/21388 (45%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[10240/21388 (48%)]\tLoss: 0.000546\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[10880/21388 (51%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[11520/21388 (54%)]\tLoss: 0.000657\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[12160/21388 (57%)]\tLoss: 0.005184\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[12800/21388 (60%)]\tLoss: 0.000448\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[13440/21388 (63%)]\tLoss: 0.000267\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[14720/21388 (69%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[15360/21388 (72%)]\tLoss: 0.000253\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[16000/21388 (75%)]\tLoss: 0.000582\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[16640/21388 (78%)]\tLoss: 0.000115\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[17280/21388 (81%)]\tLoss: 0.000524\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[17920/21388 (84%)]\tLoss: 0.003270\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[18560/21388 (87%)]\tLoss: 0.001344\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[19200/21388 (90%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[19840/21388 (93%)]\tLoss: 0.004300\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[20480/21388 (96%)]\tLoss: 0.000095\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 13\t[21120/21388 (99%)]\tLoss: 0.000432\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.092580, Accuracy: 99.738630 \n",
      "\n",
      "Train Epoch: 14\t[0/21388 (0%)]\tLoss: 0.000153\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.003 (0.003)\n",
      "Train Epoch: 14\t[640/21388 (3%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 14\t[1280/21388 (6%)]\tLoss: 0.000696\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 14\t[1920/21388 (9%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[2560/21388 (12%)]\tLoss: 0.000308\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 14\t[3200/21388 (15%)]\tLoss: 0.001776\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[3840/21388 (18%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[4480/21388 (21%)]\tLoss: 0.000631\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[5120/21388 (24%)]\tLoss: 0.000798\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[5760/21388 (27%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[6400/21388 (30%)]\tLoss: 0.000189\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[7040/21388 (33%)]\tLoss: 0.004411\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[7680/21388 (36%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[8320/21388 (39%)]\tLoss: 0.000145\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[8960/21388 (42%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[9600/21388 (45%)]\tLoss: 0.000194\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[10240/21388 (48%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[11520/21388 (54%)]\tLoss: 0.001249\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[12160/21388 (57%)]\tLoss: 0.000161\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[12800/21388 (60%)]\tLoss: 0.000216\tAccuracy: 100.000000\tTime 0.360 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[13440/21388 (63%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[14080/21388 (66%)]\tLoss: 0.023133\tAccuracy: 96.875000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[14720/21388 (69%)]\tLoss: 0.000962\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[15360/21388 (72%)]\tLoss: 0.000286\tAccuracy: 100.000000\tTime 0.361 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[16000/21388 (75%)]\tLoss: 0.000397\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[16640/21388 (78%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[17280/21388 (81%)]\tLoss: 0.001803\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[17920/21388 (84%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[18560/21388 (87%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 14\t[19200/21388 (90%)]\tLoss: 0.000483\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[19840/21388 (93%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[20480/21388 (96%)]\tLoss: 0.001122\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 14\t[21120/21388 (99%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.088983, Accuracy: 99.906716 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 15\t[0/21388 (0%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.002 (0.002)\n",
      "Train Epoch: 15\t[640/21388 (3%)]\tLoss: 0.000328\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 15\t[1280/21388 (6%)]\tLoss: 0.000258\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 15\t[1920/21388 (9%)]\tLoss: 0.000533\tAccuracy: 100.000000\tTime 0.360 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 15\t[2560/21388 (12%)]\tLoss: 0.000813\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[3200/21388 (15%)]\tLoss: 0.004790\tAccuracy: 96.875000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[3840/21388 (18%)]\tLoss: 0.001263\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[4480/21388 (21%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[5120/21388 (24%)]\tLoss: 0.002137\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[5760/21388 (27%)]\tLoss: 0.000350\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[6400/21388 (30%)]\tLoss: 0.005819\tAccuracy: 98.437500\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[7040/21388 (33%)]\tLoss: 0.000369\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[7680/21388 (36%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[8320/21388 (39%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[8960/21388 (42%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[9600/21388 (45%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[10240/21388 (48%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[10880/21388 (51%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[11520/21388 (54%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[12160/21388 (57%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[12800/21388 (60%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[13440/21388 (63%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[14720/21388 (69%)]\tLoss: 0.005179\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[15360/21388 (72%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[16000/21388 (75%)]\tLoss: 0.000131\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[16640/21388 (78%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[17280/21388 (81%)]\tLoss: 0.000051\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[17920/21388 (84%)]\tLoss: 0.000684\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[18560/21388 (87%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[19200/21388 (90%)]\tLoss: 0.000190\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[19840/21388 (93%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 15\t[20480/21388 (96%)]\tLoss: 0.000764\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 15\t[21120/21388 (99%)]\tLoss: 0.000100\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.082676, Accuracy: 99.785272 \n",
      "\n",
      "Train Epoch: 16\t[0/21388 (0%)]\tLoss: 0.000524\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 16\t[640/21388 (3%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[1280/21388 (6%)]\tLoss: 0.000330\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[1920/21388 (9%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[2560/21388 (12%)]\tLoss: 0.000336\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[3200/21388 (15%)]\tLoss: 0.000324\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[3840/21388 (18%)]\tLoss: 0.001315\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[4480/21388 (21%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[5120/21388 (24%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[5760/21388 (27%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[6400/21388 (30%)]\tLoss: 0.000321\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[7040/21388 (33%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[7680/21388 (36%)]\tLoss: 0.000875\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[8320/21388 (39%)]\tLoss: 0.000645\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 16\t[8960/21388 (42%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[9600/21388 (45%)]\tLoss: 0.000056\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[10240/21388 (48%)]\tLoss: 0.000063\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[10880/21388 (51%)]\tLoss: 0.000437\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[11520/21388 (54%)]\tLoss: 0.000956\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[12160/21388 (57%)]\tLoss: 0.001204\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[12800/21388 (60%)]\tLoss: 0.000115\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 16\t[13440/21388 (63%)]\tLoss: 0.001505\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[14720/21388 (69%)]\tLoss: 0.000506\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[15360/21388 (72%)]\tLoss: 0.000054\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[16000/21388 (75%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[16640/21388 (78%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[17280/21388 (81%)]\tLoss: 0.000865\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[17920/21388 (84%)]\tLoss: 0.000184\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[18560/21388 (87%)]\tLoss: 0.000045\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[19200/21388 (90%)]\tLoss: 0.005741\tAccuracy: 96.875000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[19840/21388 (93%)]\tLoss: 0.000133\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[20480/21388 (96%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 16\t[21120/21388 (99%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.082587, Accuracy: 99.626866 \n",
      "\n",
      "Train Epoch: 17\t[0/21388 (0%)]\tLoss: 0.000438\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.004 (0.004)\n",
      "Train Epoch: 17\t[640/21388 (3%)]\tLoss: 0.000036\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[1280/21388 (6%)]\tLoss: 0.000570\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 17\t[1920/21388 (9%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[2560/21388 (12%)]\tLoss: 0.001145\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[3200/21388 (15%)]\tLoss: 0.000217\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[3840/21388 (18%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[4480/21388 (21%)]\tLoss: 0.000078\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[5120/21388 (24%)]\tLoss: 0.004219\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[5760/21388 (27%)]\tLoss: 0.000225\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[6400/21388 (30%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[7040/21388 (33%)]\tLoss: 0.000183\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[7680/21388 (36%)]\tLoss: 0.001950\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[8320/21388 (39%)]\tLoss: 0.000098\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[8960/21388 (42%)]\tLoss: 0.000235\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 17\t[9600/21388 (45%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[10240/21388 (48%)]\tLoss: 0.000162\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[11520/21388 (54%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[12160/21388 (57%)]\tLoss: 0.000089\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[12800/21388 (60%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[13440/21388 (63%)]\tLoss: 0.007293\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[14080/21388 (66%)]\tLoss: 0.001896\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[14720/21388 (69%)]\tLoss: 0.000130\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[15360/21388 (72%)]\tLoss: 0.000184\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[16000/21388 (75%)]\tLoss: 0.001696\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[16640/21388 (78%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[17280/21388 (81%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[17920/21388 (84%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 17\t[18560/21388 (87%)]\tLoss: 0.000207\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[19200/21388 (90%)]\tLoss: 0.000859\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[19840/21388 (93%)]\tLoss: 0.000308\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 17\t[20480/21388 (96%)]\tLoss: 0.007996\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 17\t[21120/21388 (99%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.002 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.069247, Accuracy: 99.813433 \n",
      "\n",
      "Train Epoch: 18\t[0/21388 (0%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 18\t[640/21388 (3%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[1280/21388 (6%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[1920/21388 (9%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[2560/21388 (12%)]\tLoss: 0.000120\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[3200/21388 (15%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[3840/21388 (18%)]\tLoss: 0.000300\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[4480/21388 (21%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[5120/21388 (24%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[6400/21388 (30%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 18\t[7040/21388 (33%)]\tLoss: 0.000118\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[7680/21388 (36%)]\tLoss: 0.000131\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[8320/21388 (39%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[8960/21388 (42%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[9600/21388 (45%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[10240/21388 (48%)]\tLoss: 0.001227\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[11520/21388 (54%)]\tLoss: 0.000026\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[12160/21388 (57%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.362 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[12800/21388 (60%)]\tLoss: 0.000102\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[13440/21388 (63%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[14080/21388 (66%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[14720/21388 (69%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[15360/21388 (72%)]\tLoss: 0.000054\tAccuracy: 100.000000\tTime 0.360 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[16000/21388 (75%)]\tLoss: 0.000030\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[16640/21388 (78%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[17280/21388 (81%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[17920/21388 (84%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[18560/21388 (87%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[19200/21388 (90%)]\tLoss: 0.000648\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 18\t[19840/21388 (93%)]\tLoss: 0.000425\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[20480/21388 (96%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 18\t[21120/21388 (99%)]\tLoss: 0.000454\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.063852, Accuracy: 99.930037 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 19\t[0/21388 (0%)]\tLoss: 0.002394\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.003 (0.003)\n",
      "Train Epoch: 19\t[640/21388 (3%)]\tLoss: 0.000258\tAccuracy: 100.000000\tTime 0.364 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[1280/21388 (6%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[1920/21388 (9%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[2560/21388 (12%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[3200/21388 (15%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[3840/21388 (18%)]\tLoss: 0.000259\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[4480/21388 (21%)]\tLoss: 0.000111\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[5120/21388 (24%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[6400/21388 (30%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[7040/21388 (33%)]\tLoss: 0.000153\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[7680/21388 (36%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 19\t[8320/21388 (39%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[8960/21388 (42%)]\tLoss: 0.000219\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[9600/21388 (45%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 19\t[10240/21388 (48%)]\tLoss: 0.000167\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 19\t[10880/21388 (51%)]\tLoss: 0.000527\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 19\t[11520/21388 (54%)]\tLoss: 0.009359\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 19\t[12160/21388 (57%)]\tLoss: 0.001535\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 19\t[12800/21388 (60%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 19\t[13440/21388 (63%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[14720/21388 (69%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[15360/21388 (72%)]\tLoss: 0.000074\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[16000/21388 (75%)]\tLoss: 0.000298\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[16640/21388 (78%)]\tLoss: 0.000886\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[17280/21388 (81%)]\tLoss: 0.000056\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[17920/21388 (84%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[18560/21388 (87%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[19200/21388 (90%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[19840/21388 (93%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[20480/21388 (96%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 19\t[21120/21388 (99%)]\tLoss: 0.000292\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.061360, Accuracy: 99.883396 \n",
      "\n",
      "Train Epoch: 20\t[0/21388 (0%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 20\t[640/21388 (3%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 20\t[1280/21388 (6%)]\tLoss: 0.001534\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 20\t[1920/21388 (9%)]\tLoss: 0.000183\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.002 (0.002)\n",
      "Train Epoch: 20\t[2560/21388 (12%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 20\t[3200/21388 (15%)]\tLoss: 0.001441\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[3840/21388 (18%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[4480/21388 (21%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[5120/21388 (24%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[5760/21388 (27%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[6400/21388 (30%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[7040/21388 (33%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[7680/21388 (36%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[8320/21388 (39%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[8960/21388 (42%)]\tLoss: 0.000234\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[9600/21388 (45%)]\tLoss: 0.001132\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[10240/21388 (48%)]\tLoss: 0.000038\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[10880/21388 (51%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[11520/21388 (54%)]\tLoss: 0.000051\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[12160/21388 (57%)]\tLoss: 0.000931\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[12800/21388 (60%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[13440/21388 (63%)]\tLoss: 0.000145\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[14080/21388 (66%)]\tLoss: 0.000836\tAccuracy: 100.000000\tTime 0.360 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[14720/21388 (69%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[15360/21388 (72%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[16000/21388 (75%)]\tLoss: 0.000351\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[16640/21388 (78%)]\tLoss: 0.003234\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[17280/21388 (81%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[17920/21388 (84%)]\tLoss: 0.000104\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 20\t[18560/21388 (87%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[19200/21388 (90%)]\tLoss: 0.000053\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[19840/21388 (93%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[20480/21388 (96%)]\tLoss: 0.000222\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 20\t[21120/21388 (99%)]\tLoss: 0.000577\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.062952, Accuracy: 99.813433 \n",
      "\n",
      "Train Epoch: 21\t[0/21388 (0%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 21\t[640/21388 (3%)]\tLoss: 0.002211\tAccuracy: 98.437500\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[1280/21388 (6%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[1920/21388 (9%)]\tLoss: 0.001025\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 21\t[2560/21388 (12%)]\tLoss: 0.000508\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[3200/21388 (15%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[3840/21388 (18%)]\tLoss: 0.001859\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[4480/21388 (21%)]\tLoss: 0.009980\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[5120/21388 (24%)]\tLoss: 0.000126\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[5760/21388 (27%)]\tLoss: 0.000049\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[6400/21388 (30%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[7040/21388 (33%)]\tLoss: 0.000490\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[7680/21388 (36%)]\tLoss: 0.000064\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[8320/21388 (39%)]\tLoss: 0.000088\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[8960/21388 (42%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[9600/21388 (45%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[10240/21388 (48%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[10880/21388 (51%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[11520/21388 (54%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[12160/21388 (57%)]\tLoss: 0.000433\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[12800/21388 (60%)]\tLoss: 0.000088\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[13440/21388 (63%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[14080/21388 (66%)]\tLoss: 0.000391\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[14720/21388 (69%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[15360/21388 (72%)]\tLoss: 0.000104\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[16000/21388 (75%)]\tLoss: 0.000030\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[16640/21388 (78%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[17280/21388 (81%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[17920/21388 (84%)]\tLoss: 0.000309\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[18560/21388 (87%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[19200/21388 (90%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[19840/21388 (93%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[20480/21388 (96%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 21\t[21120/21388 (99%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.054497, Accuracy: 99.883396 \n",
      "\n",
      "Train Epoch: 22\t[0/21388 (0%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 22\t[640/21388 (3%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[1280/21388 (6%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[1920/21388 (9%)]\tLoss: 0.000494\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 22\t[2560/21388 (12%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[3200/21388 (15%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[3840/21388 (18%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[4480/21388 (21%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[5120/21388 (24%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 22\t[5760/21388 (27%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[6400/21388 (30%)]\tLoss: 0.000088\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[7040/21388 (33%)]\tLoss: 0.000022\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[7680/21388 (36%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[8320/21388 (39%)]\tLoss: 0.001010\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[8960/21388 (42%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[9600/21388 (45%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.359 (0.356)\tData 0.002 (0.001)\n",
      "Train Epoch: 22\t[10240/21388 (48%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[10880/21388 (51%)]\tLoss: 0.000179\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[11520/21388 (54%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[12160/21388 (57%)]\tLoss: 0.000771\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[12800/21388 (60%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[13440/21388 (63%)]\tLoss: 0.000026\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[14080/21388 (66%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[14720/21388 (69%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[15360/21388 (72%)]\tLoss: 0.000058\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[16000/21388 (75%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[16640/21388 (78%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[17280/21388 (81%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[17920/21388 (84%)]\tLoss: 0.004749\tAccuracy: 98.437500\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[18560/21388 (87%)]\tLoss: 0.002396\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[19200/21388 (90%)]\tLoss: 0.001216\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[19840/21388 (93%)]\tLoss: 0.000060\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[20480/21388 (96%)]\tLoss: 0.000721\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 22\t[21120/21388 (99%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.058391, Accuracy: 99.766791 \n",
      "\n",
      "Train Epoch: 23\t[0/21388 (0%)]\tLoss: 0.000162\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 23\t[640/21388 (3%)]\tLoss: 0.000343\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 23\t[1280/21388 (6%)]\tLoss: 0.000562\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 23\t[1920/21388 (9%)]\tLoss: 0.002051\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[2560/21388 (12%)]\tLoss: 0.004170\tAccuracy: 98.437500\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[3200/21388 (15%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[3840/21388 (18%)]\tLoss: 0.000053\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[4480/21388 (21%)]\tLoss: 0.001151\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[5120/21388 (24%)]\tLoss: 0.000153\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[5760/21388 (27%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[6400/21388 (30%)]\tLoss: 0.000383\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[7040/21388 (33%)]\tLoss: 0.000196\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[7680/21388 (36%)]\tLoss: 0.000021\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[8320/21388 (39%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[8960/21388 (42%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[9600/21388 (45%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[10240/21388 (48%)]\tLoss: 0.000056\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[10880/21388 (51%)]\tLoss: 0.000304\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[11520/21388 (54%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[12160/21388 (57%)]\tLoss: 0.007082\tAccuracy: 98.437500\tTime 0.354 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[12800/21388 (60%)]\tLoss: 0.000032\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[13440/21388 (63%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[14720/21388 (69%)]\tLoss: 0.000714\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[15360/21388 (72%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[16000/21388 (75%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[16640/21388 (78%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 23\t[17280/21388 (81%)]\tLoss: 0.000427\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[17920/21388 (84%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[18560/21388 (87%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[19200/21388 (90%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[19840/21388 (93%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[20480/21388 (96%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 23\t[21120/21388 (99%)]\tLoss: 0.004844\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.051206, Accuracy: 99.720149 \n",
      "\n",
      "Train Epoch: 24\t[0/21388 (0%)]\tLoss: 0.000485\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 24\t[640/21388 (3%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[1280/21388 (6%)]\tLoss: 0.000257\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[1920/21388 (9%)]\tLoss: 0.004081\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 24\t[2560/21388 (12%)]\tLoss: 0.001120\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[3200/21388 (15%)]\tLoss: 0.000087\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[3840/21388 (18%)]\tLoss: 0.000045\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[4480/21388 (21%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[5120/21388 (24%)]\tLoss: 0.000147\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[6400/21388 (30%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[7040/21388 (33%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[7680/21388 (36%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[8320/21388 (39%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[8960/21388 (42%)]\tLoss: 0.000057\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[9600/21388 (45%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[10240/21388 (48%)]\tLoss: 0.000107\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[10880/21388 (51%)]\tLoss: 0.000408\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[11520/21388 (54%)]\tLoss: 0.000067\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 24\t[12160/21388 (57%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 24\t[12800/21388 (60%)]\tLoss: 0.000135\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[13440/21388 (63%)]\tLoss: 0.000601\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 24\t[14080/21388 (66%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[14720/21388 (69%)]\tLoss: 0.000109\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[15360/21388 (72%)]\tLoss: 0.000134\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[16000/21388 (75%)]\tLoss: 0.000030\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[16640/21388 (78%)]\tLoss: 0.000164\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[17280/21388 (81%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[17920/21388 (84%)]\tLoss: 0.000190\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[18560/21388 (87%)]\tLoss: 0.010111\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[19200/21388 (90%)]\tLoss: 0.000112\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[19840/21388 (93%)]\tLoss: 0.000529\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[20480/21388 (96%)]\tLoss: 0.000067\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 24\t[21120/21388 (99%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.044935, Accuracy: 99.860075 \n",
      "\n",
      "Train Epoch: 25\t[0/21388 (0%)]\tLoss: 0.002097\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 25\t[640/21388 (3%)]\tLoss: 0.000033\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 25\t[1280/21388 (6%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 25\t[1920/21388 (9%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.354 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 25\t[2560/21388 (12%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 25\t[3200/21388 (15%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 25\t[3840/21388 (18%)]\tLoss: 0.000059\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[4480/21388 (21%)]\tLoss: 0.000976\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[5120/21388 (24%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[5760/21388 (27%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[6400/21388 (30%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[7040/21388 (33%)]\tLoss: 0.003121\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[7680/21388 (36%)]\tLoss: 0.001202\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[8320/21388 (39%)]\tLoss: 0.000284\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[8960/21388 (42%)]\tLoss: 0.000241\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[9600/21388 (45%)]\tLoss: 0.000105\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[10240/21388 (48%)]\tLoss: 0.001160\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[10880/21388 (51%)]\tLoss: 0.000145\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[11520/21388 (54%)]\tLoss: 0.000056\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[12160/21388 (57%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[12800/21388 (60%)]\tLoss: 0.000118\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[13440/21388 (63%)]\tLoss: 0.000908\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[14080/21388 (66%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[14720/21388 (69%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[15360/21388 (72%)]\tLoss: 0.000215\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[16000/21388 (75%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[16640/21388 (78%)]\tLoss: 0.000120\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[17280/21388 (81%)]\tLoss: 0.000773\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[17920/21388 (84%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[18560/21388 (87%)]\tLoss: 0.000026\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[19200/21388 (90%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[19840/21388 (93%)]\tLoss: 0.000083\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[20480/21388 (96%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 25\t[21120/21388 (99%)]\tLoss: 0.000291\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.042712, Accuracy: 99.878555 \n",
      "\n",
      "Train Epoch: 26\t[0/21388 (0%)]\tLoss: 0.000092\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 26\t[640/21388 (3%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[1280/21388 (6%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[1920/21388 (9%)]\tLoss: 0.000082\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[2560/21388 (12%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[3200/21388 (15%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[3840/21388 (18%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[4480/21388 (21%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[5120/21388 (24%)]\tLoss: 0.000049\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[5760/21388 (27%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[6400/21388 (30%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[7040/21388 (33%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[7680/21388 (36%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[8320/21388 (39%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[8960/21388 (42%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[9600/21388 (45%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[10240/21388 (48%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[11520/21388 (54%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[12160/21388 (57%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[12800/21388 (60%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[13440/21388 (63%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 26\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[14720/21388 (69%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[15360/21388 (72%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[16000/21388 (75%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[16640/21388 (78%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[17280/21388 (81%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[17920/21388 (84%)]\tLoss: 0.000100\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[18560/21388 (87%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[19200/21388 (90%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[19840/21388 (93%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[20480/21388 (96%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 26\t[21120/21388 (99%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.040701, Accuracy: 99.831914 \n",
      "\n",
      "Train Epoch: 27\t[0/21388 (0%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 27\t[640/21388 (3%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[1280/21388 (6%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[1920/21388 (9%)]\tLoss: 0.000048\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[2560/21388 (12%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[3200/21388 (15%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[3840/21388 (18%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[4480/21388 (21%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[5120/21388 (24%)]\tLoss: 0.000362\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[5760/21388 (27%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[6400/21388 (30%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[7040/21388 (33%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[7680/21388 (36%)]\tLoss: 0.000123\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[8320/21388 (39%)]\tLoss: 0.000540\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 27\t[8960/21388 (42%)]\tLoss: 0.000081\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[9600/21388 (45%)]\tLoss: 0.001790\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[10240/21388 (48%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[10880/21388 (51%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[11520/21388 (54%)]\tLoss: 0.000604\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[12160/21388 (57%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[12800/21388 (60%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[13440/21388 (63%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[14720/21388 (69%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[15360/21388 (72%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[16000/21388 (75%)]\tLoss: 0.000575\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[16640/21388 (78%)]\tLoss: 0.000053\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[17280/21388 (81%)]\tLoss: 0.000155\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[17920/21388 (84%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 27\t[18560/21388 (87%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[19200/21388 (90%)]\tLoss: 0.000450\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[19840/21388 (93%)]\tLoss: 0.005507\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[20480/21388 (96%)]\tLoss: 0.000323\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 27\t[21120/21388 (99%)]\tLoss: 0.003904\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.045002, Accuracy: 99.743470 \n",
      "\n",
      "Train Epoch: 28\t[0/21388 (0%)]\tLoss: 0.000609\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 28\t[640/21388 (3%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[1280/21388 (6%)]\tLoss: 0.000095\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[1920/21388 (9%)]\tLoss: 0.000102\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[2560/21388 (12%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[3200/21388 (15%)]\tLoss: 0.000077\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[3840/21388 (18%)]\tLoss: 0.000033\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[4480/21388 (21%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.362 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[5120/21388 (24%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[6400/21388 (30%)]\tLoss: 0.000429\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[7040/21388 (33%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[7680/21388 (36%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[8320/21388 (39%)]\tLoss: 0.001919\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[8960/21388 (42%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[9600/21388 (45%)]\tLoss: 0.011551\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[10240/21388 (48%)]\tLoss: 0.000139\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 28\t[10880/21388 (51%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[11520/21388 (54%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[12160/21388 (57%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[12800/21388 (60%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[13440/21388 (63%)]\tLoss: 0.000121\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[14080/21388 (66%)]\tLoss: 0.000111\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[14720/21388 (69%)]\tLoss: 0.000115\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[15360/21388 (72%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[16000/21388 (75%)]\tLoss: 0.000760\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[16640/21388 (78%)]\tLoss: 0.003198\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[17280/21388 (81%)]\tLoss: 0.000037\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[17920/21388 (84%)]\tLoss: 0.000048\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[18560/21388 (87%)]\tLoss: 0.001265\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[19200/21388 (90%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[19840/21388 (93%)]\tLoss: 0.009762\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[20480/21388 (96%)]\tLoss: 0.000909\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 28\t[21120/21388 (99%)]\tLoss: 0.000921\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.037388, Accuracy: 99.883396 \n",
      "\n",
      "Train Epoch: 29\t[0/21388 (0%)]\tLoss: 0.001194\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 29\t[640/21388 (3%)]\tLoss: 0.001868\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 29\t[1280/21388 (6%)]\tLoss: 0.001003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 29\t[1920/21388 (9%)]\tLoss: 0.007219\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 29\t[2560/21388 (12%)]\tLoss: 0.009875\tAccuracy: 96.875000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 29\t[3200/21388 (15%)]\tLoss: 0.002673\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[3840/21388 (18%)]\tLoss: 0.002879\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[4480/21388 (21%)]\tLoss: 0.007149\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[5120/21388 (24%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[5760/21388 (27%)]\tLoss: 0.007249\tAccuracy: 98.437500\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[6400/21388 (30%)]\tLoss: 0.011233\tAccuracy: 98.437500\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[7040/21388 (33%)]\tLoss: 0.001748\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[7680/21388 (36%)]\tLoss: 0.007118\tAccuracy: 96.875000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[8320/21388 (39%)]\tLoss: 0.014653\tAccuracy: 96.875000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[8960/21388 (42%)]\tLoss: 0.000032\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[9600/21388 (45%)]\tLoss: 0.000770\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[10240/21388 (48%)]\tLoss: 0.000127\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[10880/21388 (51%)]\tLoss: 0.000098\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[11520/21388 (54%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[12160/21388 (57%)]\tLoss: 0.003982\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[12800/21388 (60%)]\tLoss: 0.000722\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[13440/21388 (63%)]\tLoss: 0.000220\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[14080/21388 (66%)]\tLoss: 0.000736\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[14720/21388 (69%)]\tLoss: 0.000177\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[15360/21388 (72%)]\tLoss: 0.000106\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 29\t[16000/21388 (75%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[16640/21388 (78%)]\tLoss: 0.000038\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[17280/21388 (81%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[17920/21388 (84%)]\tLoss: 0.000094\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[18560/21388 (87%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[19200/21388 (90%)]\tLoss: 0.000107\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[19840/21388 (93%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[20480/21388 (96%)]\tLoss: 0.000106\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 29\t[21120/21388 (99%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.030152, Accuracy: 99.953358 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 30\t[0/21388 (0%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.003 (0.003)\n",
      "Train Epoch: 30\t[640/21388 (3%)]\tLoss: 0.000246\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[1280/21388 (6%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[1920/21388 (9%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[2560/21388 (12%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[3200/21388 (15%)]\tLoss: 0.000065\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[3840/21388 (18%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[4480/21388 (21%)]\tLoss: 0.000255\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 30\t[5120/21388 (24%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[5760/21388 (27%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[6400/21388 (30%)]\tLoss: 0.000284\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[7040/21388 (33%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[7680/21388 (36%)]\tLoss: 0.000347\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[8320/21388 (39%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.358 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[8960/21388 (42%)]\tLoss: 0.000462\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[9600/21388 (45%)]\tLoss: 0.001185\tAccuracy: 100.000000\tTime 0.356 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[10240/21388 (48%)]\tLoss: 0.000044\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[10880/21388 (51%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[11520/21388 (54%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[12160/21388 (57%)]\tLoss: 0.002866\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[12800/21388 (60%)]\tLoss: 0.000283\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[13440/21388 (63%)]\tLoss: 0.001047\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[14080/21388 (66%)]\tLoss: 0.000209\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[14720/21388 (69%)]\tLoss: 0.000140\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[15360/21388 (72%)]\tLoss: 0.004029\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[16000/21388 (75%)]\tLoss: 0.000377\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[16640/21388 (78%)]\tLoss: 0.000689\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[17280/21388 (81%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[17920/21388 (84%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[18560/21388 (87%)]\tLoss: 0.000126\tAccuracy: 100.000000\tTime 0.365 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[19200/21388 (90%)]\tLoss: 0.001477\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[19840/21388 (93%)]\tLoss: 0.000067\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[20480/21388 (96%)]\tLoss: 0.000148\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 30\t[21120/21388 (99%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.027911, Accuracy: 99.883396 \n",
      "\n",
      "Train Epoch: 31\t[0/21388 (0%)]\tLoss: 0.000023\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 31\t[640/21388 (3%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[1280/21388 (6%)]\tLoss: 0.000062\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[1920/21388 (9%)]\tLoss: 0.000127\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[2560/21388 (12%)]\tLoss: 0.000454\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[3200/21388 (15%)]\tLoss: 0.000886\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[3840/21388 (18%)]\tLoss: 0.000085\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[4480/21388 (21%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[5120/21388 (24%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 31\t[5760/21388 (27%)]\tLoss: 0.000143\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[6400/21388 (30%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.001)\n",
      "Train Epoch: 31\t[7040/21388 (33%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[7680/21388 (36%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[8320/21388 (39%)]\tLoss: 0.000049\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[8960/21388 (42%)]\tLoss: 0.000061\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[9600/21388 (45%)]\tLoss: 0.000020\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[10240/21388 (48%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[10880/21388 (51%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[11520/21388 (54%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[12160/21388 (57%)]\tLoss: 0.000059\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[12800/21388 (60%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[13440/21388 (63%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[14080/21388 (66%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 31\t[14720/21388 (69%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[15360/21388 (72%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[16000/21388 (75%)]\tLoss: 0.000424\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[16640/21388 (78%)]\tLoss: 0.000192\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[17280/21388 (81%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[17920/21388 (84%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[18560/21388 (87%)]\tLoss: 0.000089\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[19200/21388 (90%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[19840/21388 (93%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[20480/21388 (96%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 31\t[21120/21388 (99%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.024627, Accuracy: 99.976679 \n",
      "\n",
      "-------------------- HIGHEST ACCURACY ---------------------------\n",
      "Train Epoch: 32\t[0/21388 (0%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.003 (0.003)\n",
      "Train Epoch: 32\t[640/21388 (3%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[1280/21388 (6%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.356)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[1920/21388 (9%)]\tLoss: 0.000026\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[2560/21388 (12%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[3200/21388 (15%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.357 (0.356)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[3840/21388 (18%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[4480/21388 (21%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[5120/21388 (24%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[5760/21388 (27%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[6400/21388 (30%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[7040/21388 (33%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[7680/21388 (36%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[8320/21388 (39%)]\tLoss: 0.000028\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 32\t[8960/21388 (42%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[9600/21388 (45%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[10240/21388 (48%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[10880/21388 (51%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[11520/21388 (54%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[12160/21388 (57%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[12800/21388 (60%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[13440/21388 (63%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[14080/21388 (66%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 32\t[14720/21388 (69%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[15360/21388 (72%)]\tLoss: 0.000023\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[16000/21388 (75%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[16640/21388 (78%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[17280/21388 (81%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.361 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[17920/21388 (84%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[18560/21388 (87%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[19200/21388 (90%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[19840/21388 (93%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[20480/21388 (96%)]\tLoss: 0.000144\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 32\t[21120/21388 (99%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.023023, Accuracy: 99.976679 \n",
      "\n",
      "Train Epoch: 33\t[0/21388 (0%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 33\t[640/21388 (3%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[1280/21388 (6%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[1920/21388 (9%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[2560/21388 (12%)]\tLoss: 0.000041\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[3200/21388 (15%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[3840/21388 (18%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[4480/21388 (21%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[5120/21388 (24%)]\tLoss: 0.000138\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[5760/21388 (27%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[6400/21388 (30%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[7040/21388 (33%)]\tLoss: 0.000173\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[7680/21388 (36%)]\tLoss: 0.000167\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[8320/21388 (39%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[8960/21388 (42%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[9600/21388 (45%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[10240/21388 (48%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[10880/21388 (51%)]\tLoss: 0.000028\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[11520/21388 (54%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[12160/21388 (57%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[12800/21388 (60%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[13440/21388 (63%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[14080/21388 (66%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[14720/21388 (69%)]\tLoss: 0.000695\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[15360/21388 (72%)]\tLoss: -0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[16000/21388 (75%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.362 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[16640/21388 (78%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[17280/21388 (81%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[17920/21388 (84%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[18560/21388 (87%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[19200/21388 (90%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[19840/21388 (93%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[20480/21388 (96%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 33\t[21120/21388 (99%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.022206, Accuracy: 99.953358 \n",
      "\n",
      "Train Epoch: 34\t[0/21388 (0%)]\tLoss: 0.000064\tAccuracy: 100.000000\tTime 0.364 (0.364)\tData 0.004 (0.004)\n",
      "Train Epoch: 34\t[640/21388 (3%)]\tLoss: 0.000041\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[1280/21388 (6%)]\tLoss: 0.000058\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[1920/21388 (9%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[2560/21388 (12%)]\tLoss: 0.000683\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 34\t[3200/21388 (15%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[3840/21388 (18%)]\tLoss: 0.000218\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[4480/21388 (21%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[5120/21388 (24%)]\tLoss: 0.000137\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[5760/21388 (27%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[6400/21388 (30%)]\tLoss: 0.000077\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[7040/21388 (33%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[7680/21388 (36%)]\tLoss: 0.000734\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[8320/21388 (39%)]\tLoss: 0.000022\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[8960/21388 (42%)]\tLoss: 0.000172\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[9600/21388 (45%)]\tLoss: 0.000035\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[10240/21388 (48%)]\tLoss: 0.000253\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[10880/21388 (51%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.354 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[11520/21388 (54%)]\tLoss: 0.000243\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[12160/21388 (57%)]\tLoss: 0.000019\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[12800/21388 (60%)]\tLoss: 0.000081\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[13440/21388 (63%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[14080/21388 (66%)]\tLoss: 0.000324\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[14720/21388 (69%)]\tLoss: 0.004197\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 34\t[15360/21388 (72%)]\tLoss: 0.000030\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[16000/21388 (75%)]\tLoss: 0.000219\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[16640/21388 (78%)]\tLoss: 0.000017\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[17280/21388 (81%)]\tLoss: 0.001032\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[17920/21388 (84%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.360 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[18560/21388 (87%)]\tLoss: 0.003153\tAccuracy: 98.437500\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[19200/21388 (90%)]\tLoss: 0.000222\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[19840/21388 (93%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[20480/21388 (96%)]\tLoss: 0.000119\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 34\t[21120/21388 (99%)]\tLoss: 0.000419\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.023731, Accuracy: 99.813433 \n",
      "\n",
      "Train Epoch: 35\t[0/21388 (0%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 35\t[640/21388 (3%)]\tLoss: 0.000445\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[1280/21388 (6%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[1920/21388 (9%)]\tLoss: 0.000032\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[2560/21388 (12%)]\tLoss: 0.000267\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[3200/21388 (15%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[3840/21388 (18%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[4480/21388 (21%)]\tLoss: 0.000015\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[5120/21388 (24%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[5760/21388 (27%)]\tLoss: 0.000519\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[6400/21388 (30%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[7040/21388 (33%)]\tLoss: 0.000035\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[7680/21388 (36%)]\tLoss: 0.000734\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[8320/21388 (39%)]\tLoss: 0.000089\tAccuracy: 100.000000\tTime 0.362 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 35\t[8960/21388 (42%)]\tLoss: 0.000089\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[9600/21388 (45%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[10240/21388 (48%)]\tLoss: 0.000104\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[10880/21388 (51%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.366 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[11520/21388 (54%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[12160/21388 (57%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[12800/21388 (60%)]\tLoss: 0.000195\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[13440/21388 (63%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[14080/21388 (66%)]\tLoss: 0.000099\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[14720/21388 (69%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 35\t[15360/21388 (72%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[16000/21388 (75%)]\tLoss: 0.000011\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[16640/21388 (78%)]\tLoss: 0.000032\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[17280/21388 (81%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[17920/21388 (84%)]\tLoss: 0.000007\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[18560/21388 (87%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[19200/21388 (90%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[19840/21388 (93%)]\tLoss: 0.000009\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[20480/21388 (96%)]\tLoss: 0.000432\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 35\t[21120/21388 (99%)]\tLoss: 0.000417\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.019222, Accuracy: 99.906716 \n",
      "\n",
      "Train Epoch: 36\t[0/21388 (0%)]\tLoss: 0.000111\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.003 (0.003)\n",
      "Train Epoch: 36\t[640/21388 (3%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[1280/21388 (6%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[1920/21388 (9%)]\tLoss: 0.000126\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 36\t[2560/21388 (12%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 36\t[3200/21388 (15%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 36\t[3840/21388 (18%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 36\t[4480/21388 (21%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 36\t[5120/21388 (24%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[5760/21388 (27%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[6400/21388 (30%)]\tLoss: 0.000000\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 36\t[7040/21388 (33%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[7680/21388 (36%)]\tLoss: 0.000184\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[8320/21388 (39%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[8960/21388 (42%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[9600/21388 (45%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[10240/21388 (48%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[10880/21388 (51%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[11520/21388 (54%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[12160/21388 (57%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.362 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[12800/21388 (60%)]\tLoss: 0.000037\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[13440/21388 (63%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[14080/21388 (66%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[14720/21388 (69%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[15360/21388 (72%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[16000/21388 (75%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[16640/21388 (78%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.361 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[17280/21388 (81%)]\tLoss: 0.000028\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.002 (0.002)\n",
      "Train Epoch: 36\t[17920/21388 (84%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[18560/21388 (87%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[19200/21388 (90%)]\tLoss: 0.000025\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[19840/21388 (93%)]\tLoss: 0.000047\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[20480/21388 (96%)]\tLoss: 0.000027\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 36\t[21120/21388 (99%)]\tLoss: 0.000014\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.018552, Accuracy: 99.855234 \n",
      "\n",
      "Train Epoch: 37\t[0/21388 (0%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.004 (0.004)\n",
      "Train Epoch: 37\t[640/21388 (3%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[1280/21388 (6%)]\tLoss: 0.000010\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[1920/21388 (9%)]\tLoss: 0.000041\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[2560/21388 (12%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[3200/21388 (15%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[3840/21388 (18%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[4480/21388 (21%)]\tLoss: 0.000002\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[5120/21388 (24%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[5760/21388 (27%)]\tLoss: 0.000008\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[6400/21388 (30%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.359 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[7040/21388 (33%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[7680/21388 (36%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[8320/21388 (39%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[8960/21388 (42%)]\tLoss: 0.000013\tAccuracy: 100.000000\tTime 0.355 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[9600/21388 (45%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[10240/21388 (48%)]\tLoss: 0.000003\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[10880/21388 (51%)]\tLoss: 0.000005\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[11520/21388 (54%)]\tLoss: 0.000004\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[12160/21388 (57%)]\tLoss: 0.000001\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[12800/21388 (60%)]\tLoss: 0.000018\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[13440/21388 (63%)]\tLoss: 0.000012\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[14080/21388 (66%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[14720/21388 (69%)]\tLoss: 0.000034\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[15360/21388 (72%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[16000/21388 (75%)]\tLoss: 0.000245\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[16640/21388 (78%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 37\t[17280/21388 (81%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[17920/21388 (84%)]\tLoss: 0.000076\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[18560/21388 (87%)]\tLoss: 0.000006\tAccuracy: 100.000000\tTime 0.356 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[19200/21388 (90%)]\tLoss: 0.000016\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[19840/21388 (93%)]\tLoss: 0.000029\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[20480/21388 (96%)]\tLoss: 0.000075\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "Train Epoch: 37\t[21120/21388 (99%)]\tLoss: 0.000790\tAccuracy: 100.000000\tTime 0.357 (0.357)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.019814, Accuracy: 99.813433 \n",
      "\n",
      "Train Epoch: 38\t[0/21388 (0%)]\tLoss: 0.000067\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.004 (0.004)\n",
      "Train Epoch: 38\t[640/21388 (3%)]\tLoss: 0.000242\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[1280/21388 (6%)]\tLoss: 0.000078\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[1920/21388 (9%)]\tLoss: 0.000024\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[2560/21388 (12%)]\tLoss: 0.000140\tAccuracy: 100.000000\tTime 0.358 (0.357)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[3200/21388 (15%)]\tLoss: 0.009105\tAccuracy: 98.437500\tTime 0.361 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[3840/21388 (18%)]\tLoss: 0.003771\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[4480/21388 (21%)]\tLoss: 0.073459\tAccuracy: 95.312500\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[5120/21388 (24%)]\tLoss: 0.003289\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[5760/21388 (27%)]\tLoss: 0.001765\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[6400/21388 (30%)]\tLoss: 0.007123\tAccuracy: 98.437500\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[7040/21388 (33%)]\tLoss: 0.001760\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[7680/21388 (36%)]\tLoss: 0.004830\tAccuracy: 100.000000\tTime 0.360 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[8320/21388 (39%)]\tLoss: 0.001379\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[8960/21388 (42%)]\tLoss: 0.000332\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[9600/21388 (45%)]\tLoss: 0.000638\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[10240/21388 (48%)]\tLoss: 0.000552\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[10880/21388 (51%)]\tLoss: 0.000086\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[11520/21388 (54%)]\tLoss: 0.000879\tAccuracy: 100.000000\tTime 0.360 (0.358)\tData 0.002 (0.002)\n",
      "Train Epoch: 38\t[12160/21388 (57%)]\tLoss: 0.001134\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[12800/21388 (60%)]\tLoss: 0.000091\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[13440/21388 (63%)]\tLoss: 0.002405\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[14080/21388 (66%)]\tLoss: 0.000587\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[14720/21388 (69%)]\tLoss: 0.000048\tAccuracy: 100.000000\tTime 0.360 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[15360/21388 (72%)]\tLoss: 0.000261\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[16000/21388 (75%)]\tLoss: 0.000280\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[16640/21388 (78%)]\tLoss: 0.000061\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[17280/21388 (81%)]\tLoss: 0.000433\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 38\t[17920/21388 (84%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[18560/21388 (87%)]\tLoss: 0.000189\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[19200/21388 (90%)]\tLoss: 0.000509\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[19840/21388 (93%)]\tLoss: 0.000220\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[20480/21388 (96%)]\tLoss: 0.000039\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 38\t[21120/21388 (99%)]\tLoss: 0.000834\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.017349, Accuracy: 99.836754 \n",
      "\n",
      "Train Epoch: 39\t[0/21388 (0%)]\tLoss: 0.000231\tAccuracy: 100.000000\tTime 0.364 (0.364)\tData 0.004 (0.004)\n",
      "Train Epoch: 39\t[640/21388 (3%)]\tLoss: 0.011517\tAccuracy: 98.437500\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[1280/21388 (6%)]\tLoss: 0.000806\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[1920/21388 (9%)]\tLoss: 0.002309\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[2560/21388 (12%)]\tLoss: 0.000595\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.002 (0.002)\n",
      "Train Epoch: 39\t[3200/21388 (15%)]\tLoss: 0.000525\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[3840/21388 (18%)]\tLoss: 0.000364\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[4480/21388 (21%)]\tLoss: 0.000120\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[5120/21388 (24%)]\tLoss: 0.005528\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[5760/21388 (27%)]\tLoss: 0.000146\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[6400/21388 (30%)]\tLoss: 0.000057\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[7040/21388 (33%)]\tLoss: 0.000286\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[7680/21388 (36%)]\tLoss: 0.000999\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[8320/21388 (39%)]\tLoss: 0.000476\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[8960/21388 (42%)]\tLoss: 0.000108\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[9600/21388 (45%)]\tLoss: 0.000536\tAccuracy: 100.000000\tTime 0.356 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[10240/21388 (48%)]\tLoss: 0.000366\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[10880/21388 (51%)]\tLoss: 0.000222\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[11520/21388 (54%)]\tLoss: 0.000236\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[12160/21388 (57%)]\tLoss: 0.002312\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[12800/21388 (60%)]\tLoss: 0.000142\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[13440/21388 (63%)]\tLoss: 0.000098\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[14080/21388 (66%)]\tLoss: 0.000284\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[14720/21388 (69%)]\tLoss: 0.000165\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.002 (0.001)\n",
      "Train Epoch: 39\t[15360/21388 (72%)]\tLoss: 0.000263\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[16000/21388 (75%)]\tLoss: 0.000102\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[16640/21388 (78%)]\tLoss: 0.000146\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 39\t[17280/21388 (81%)]\tLoss: 0.000054\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[17920/21388 (84%)]\tLoss: 0.000160\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[18560/21388 (87%)]\tLoss: 0.000711\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[19200/21388 (90%)]\tLoss: 0.000245\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[19840/21388 (93%)]\tLoss: 0.000553\tAccuracy: 100.000000\tTime 0.362 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[20480/21388 (96%)]\tLoss: 0.000537\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 39\t[21120/21388 (99%)]\tLoss: 0.000050\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.002 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.014792, Accuracy: 99.906716 \n",
      "\n",
      "Train Epoch: 40\t[0/21388 (0%)]\tLoss: 0.000523\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 40\t[640/21388 (3%)]\tLoss: 0.000123\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[1280/21388 (6%)]\tLoss: 0.000088\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[1920/21388 (9%)]\tLoss: 0.000231\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[2560/21388 (12%)]\tLoss: 0.000121\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[3200/21388 (15%)]\tLoss: 0.000053\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[3840/21388 (18%)]\tLoss: 0.001210\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.002)\n",
      "Train Epoch: 40\t[4480/21388 (21%)]\tLoss: 0.000067\tAccuracy: 100.000000\tTime 0.356 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[5120/21388 (24%)]\tLoss: 0.000133\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[5760/21388 (27%)]\tLoss: 0.000032\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[6400/21388 (30%)]\tLoss: 0.000340\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[7040/21388 (33%)]\tLoss: 0.000132\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[7680/21388 (36%)]\tLoss: 0.000090\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[8320/21388 (39%)]\tLoss: 0.000059\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[8960/21388 (42%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[9600/21388 (45%)]\tLoss: 0.000042\tAccuracy: 100.000000\tTime 0.360 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[10240/21388 (48%)]\tLoss: 0.000043\tAccuracy: 100.000000\tTime 0.357 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[10880/21388 (51%)]\tLoss: 0.000050\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[11520/21388 (54%)]\tLoss: 0.000433\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[12160/21388 (57%)]\tLoss: 0.000667\tAccuracy: 100.000000\tTime 0.360 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[12800/21388 (60%)]\tLoss: 0.000052\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[13440/21388 (63%)]\tLoss: 0.000065\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[14080/21388 (66%)]\tLoss: 0.001393\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[14720/21388 (69%)]\tLoss: 0.001160\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.002 (0.001)\n",
      "Train Epoch: 40\t[15360/21388 (72%)]\tLoss: 0.000264\tAccuracy: 100.000000\tTime 0.358 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[16000/21388 (75%)]\tLoss: 0.000367\tAccuracy: 100.000000\tTime 0.359 (0.358)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[16640/21388 (78%)]\tLoss: 0.000231\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[17280/21388 (81%)]\tLoss: 0.000046\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[17920/21388 (84%)]\tLoss: 0.000057\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[18560/21388 (87%)]\tLoss: 0.000240\tAccuracy: 100.000000\tTime 0.361 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[19200/21388 (90%)]\tLoss: 0.000091\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[19840/21388 (93%)]\tLoss: 0.000395\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[20480/21388 (96%)]\tLoss: 0.000844\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "Train Epoch: 40\t[21120/21388 (99%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.013201, Accuracy: 99.906716 \n",
      "\n",
      "Train Epoch: 41\t[0/21388 (0%)]\tLoss: 0.000240\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.004 (0.004)\n",
      "Train Epoch: 41\t[640/21388 (3%)]\tLoss: 0.000090\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[1280/21388 (6%)]\tLoss: 0.000158\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[1920/21388 (9%)]\tLoss: 0.000077\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[2560/21388 (12%)]\tLoss: 0.000097\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[3200/21388 (15%)]\tLoss: 0.000040\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[3840/21388 (18%)]\tLoss: 0.000206\tAccuracy: 100.000000\tTime 0.363 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[4480/21388 (21%)]\tLoss: 0.000061\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[5120/21388 (24%)]\tLoss: 0.000228\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[5760/21388 (27%)]\tLoss: 0.000069\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[6400/21388 (30%)]\tLoss: 0.000115\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[7040/21388 (33%)]\tLoss: 0.000068\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[7680/21388 (36%)]\tLoss: 0.000160\tAccuracy: 100.000000\tTime 0.364 (0.359)\tData 0.002 (0.002)\n",
      "Train Epoch: 41\t[8320/21388 (39%)]\tLoss: 0.000108\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[8960/21388 (42%)]\tLoss: 0.000322\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[9600/21388 (45%)]\tLoss: 0.000062\tAccuracy: 100.000000\tTime 0.357 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[10240/21388 (48%)]\tLoss: 0.000134\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[10880/21388 (51%)]\tLoss: 0.000175\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[11520/21388 (54%)]\tLoss: 0.000107\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[12160/21388 (57%)]\tLoss: 0.000105\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[12800/21388 (60%)]\tLoss: 0.000135\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[13440/21388 (63%)]\tLoss: 0.000798\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[14080/21388 (66%)]\tLoss: 0.000250\tAccuracy: 100.000000\tTime 0.361 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[14720/21388 (69%)]\tLoss: 0.000071\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[15360/21388 (72%)]\tLoss: 0.000558\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[16000/21388 (75%)]\tLoss: 0.000093\tAccuracy: 100.000000\tTime 0.361 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[16640/21388 (78%)]\tLoss: 0.000071\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[17280/21388 (81%)]\tLoss: 0.000238\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[17920/21388 (84%)]\tLoss: 0.000125\tAccuracy: 100.000000\tTime 0.359 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[18560/21388 (87%)]\tLoss: 0.000263\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[19200/21388 (90%)]\tLoss: 0.000261\tAccuracy: 100.000000\tTime 0.360 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[19840/21388 (93%)]\tLoss: 0.000088\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[20480/21388 (96%)]\tLoss: 0.000153\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "Train Epoch: 41\t[21120/21388 (99%)]\tLoss: 0.000063\tAccuracy: 100.000000\tTime 0.358 (0.359)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.011091, Accuracy: 99.976679 \n",
      "\n",
      "Train Epoch: 42\t[0/21388 (0%)]\tLoss: 0.000096\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.004 (0.004)\n",
      "Train Epoch: 42\t[640/21388 (3%)]\tLoss: 0.007858\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[1280/21388 (6%)]\tLoss: 0.000428\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[1920/21388 (9%)]\tLoss: 0.000093\tAccuracy: 100.000000\tTime 0.362 (0.360)\tData 0.002 (0.002)\n",
      "Train Epoch: 42\t[2560/21388 (12%)]\tLoss: 0.000222\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[3200/21388 (15%)]\tLoss: 0.000111\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[3840/21388 (18%)]\tLoss: 0.000234\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[4480/21388 (21%)]\tLoss: 0.000122\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[5120/21388 (24%)]\tLoss: 0.000201\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[5760/21388 (27%)]\tLoss: 0.000093\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[6400/21388 (30%)]\tLoss: 0.000129\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[7040/21388 (33%)]\tLoss: 0.000105\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[7680/21388 (36%)]\tLoss: 0.000128\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[8320/21388 (39%)]\tLoss: 0.000096\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[8960/21388 (42%)]\tLoss: 0.000167\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[9600/21388 (45%)]\tLoss: 0.000142\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[10240/21388 (48%)]\tLoss: 0.000209\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[10880/21388 (51%)]\tLoss: 0.000255\tAccuracy: 100.000000\tTime 0.364 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[11520/21388 (54%)]\tLoss: 0.000099\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[12160/21388 (57%)]\tLoss: 0.000181\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[12800/21388 (60%)]\tLoss: 0.000552\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[13440/21388 (63%)]\tLoss: 0.000626\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[14080/21388 (66%)]\tLoss: 0.000108\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[14720/21388 (69%)]\tLoss: 0.000061\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[15360/21388 (72%)]\tLoss: 0.000355\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[16000/21388 (75%)]\tLoss: 0.000116\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[16640/21388 (78%)]\tLoss: 0.000515\tAccuracy: 100.000000\tTime 0.363 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[17280/21388 (81%)]\tLoss: 0.000057\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[17920/21388 (84%)]\tLoss: 0.000098\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[18560/21388 (87%)]\tLoss: 0.000127\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[19200/21388 (90%)]\tLoss: 0.000153\tAccuracy: 100.000000\tTime 0.364 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[19840/21388 (93%)]\tLoss: 0.000118\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[20480/21388 (96%)]\tLoss: 0.000074\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 42\t[21120/21388 (99%)]\tLoss: 0.000251\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.009385, Accuracy: 100.000000 \n",
      "\n",
      "Train Epoch: 43\t[0/21388 (0%)]\tLoss: 0.000158\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.004 (0.004)\n",
      "Train Epoch: 43\t[640/21388 (3%)]\tLoss: 0.000166\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[1280/21388 (6%)]\tLoss: 0.000074\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[1920/21388 (9%)]\tLoss: 0.000124\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[2560/21388 (12%)]\tLoss: 0.000203\tAccuracy: 100.000000\tTime 0.357 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[3200/21388 (15%)]\tLoss: 0.000312\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[3840/21388 (18%)]\tLoss: 0.000074\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[4480/21388 (21%)]\tLoss: 0.000147\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[5120/21388 (24%)]\tLoss: 0.000125\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[5760/21388 (27%)]\tLoss: 0.000135\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[6400/21388 (30%)]\tLoss: 0.000122\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[7040/21388 (33%)]\tLoss: 0.000074\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[7680/21388 (36%)]\tLoss: 0.000164\tAccuracy: 100.000000\tTime 0.366 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[8320/21388 (39%)]\tLoss: 0.000305\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[8960/21388 (42%)]\tLoss: 0.000187\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[9600/21388 (45%)]\tLoss: 0.000118\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[10240/21388 (48%)]\tLoss: 0.000072\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[10880/21388 (51%)]\tLoss: 0.000128\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[11520/21388 (54%)]\tLoss: 0.000178\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[12160/21388 (57%)]\tLoss: 0.000196\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[12800/21388 (60%)]\tLoss: 0.000080\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[13440/21388 (63%)]\tLoss: 0.000174\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[14080/21388 (66%)]\tLoss: 0.000155\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[14720/21388 (69%)]\tLoss: 0.004016\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[15360/21388 (72%)]\tLoss: 0.000824\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[16000/21388 (75%)]\tLoss: 0.000320\tAccuracy: 100.000000\tTime 0.362 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[16640/21388 (78%)]\tLoss: 0.001288\tAccuracy: 100.000000\tTime 0.361 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[17280/21388 (81%)]\tLoss: 0.000901\tAccuracy: 100.000000\tTime 0.359 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[17920/21388 (84%)]\tLoss: 0.004833\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[18560/21388 (87%)]\tLoss: 0.000280\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[19200/21388 (90%)]\tLoss: 0.000344\tAccuracy: 100.000000\tTime 0.362 (0.360)\tData 0.001 (0.002)\n",
      "Train Epoch: 43\t[19840/21388 (93%)]\tLoss: 0.000443\tAccuracy: 100.000000\tTime 0.360 (0.360)\tData 0.001 (0.001)\n",
      "Train Epoch: 43\t[20480/21388 (96%)]\tLoss: 0.000176\tAccuracy: 100.000000\tTime 0.358 (0.360)\tData 0.001 (0.001)\n",
      "Train Epoch: 43\t[21120/21388 (99%)]\tLoss: 0.004665\tAccuracy: 98.437500\tTime 0.360 (0.360)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.009441, Accuracy: 99.953358 \n",
      "\n",
      "Train Epoch: 44\t[0/21388 (0%)]\tLoss: 0.000341\tAccuracy: 100.000000\tTime 0.364 (0.364)\tData 0.004 (0.004)\n",
      "Train Epoch: 44\t[640/21388 (3%)]\tLoss: 0.000613\tAccuracy: 100.000000\tTime 0.365 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[1280/21388 (6%)]\tLoss: 0.033733\tAccuracy: 96.875000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[1920/21388 (9%)]\tLoss: 0.001052\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[2560/21388 (12%)]\tLoss: 0.003303\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[3200/21388 (15%)]\tLoss: 0.000504\tAccuracy: 100.000000\tTime 0.360 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[3840/21388 (18%)]\tLoss: 0.000476\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[4480/21388 (21%)]\tLoss: 0.000553\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[5120/21388 (24%)]\tLoss: 0.001447\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[5760/21388 (27%)]\tLoss: 0.002289\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[6400/21388 (30%)]\tLoss: 0.007741\tAccuracy: 98.437500\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[7040/21388 (33%)]\tLoss: 0.004570\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[7680/21388 (36%)]\tLoss: 0.002773\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[8320/21388 (39%)]\tLoss: 0.097451\tAccuracy: 92.187500\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[8960/21388 (42%)]\tLoss: 0.005563\tAccuracy: 98.437500\tTime 0.361 (0.363)\tData 0.002 (0.002)\n",
      "Train Epoch: 44\t[9600/21388 (45%)]\tLoss: 0.010884\tAccuracy: 98.437500\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[10240/21388 (48%)]\tLoss: 0.001376\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[10880/21388 (51%)]\tLoss: 0.002722\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[11520/21388 (54%)]\tLoss: 0.000397\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[12160/21388 (57%)]\tLoss: 0.000583\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[12800/21388 (60%)]\tLoss: 0.006327\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[13440/21388 (63%)]\tLoss: 0.000666\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[14080/21388 (66%)]\tLoss: 0.005263\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[14720/21388 (69%)]\tLoss: 0.000349\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.002 (0.002)\n",
      "Train Epoch: 44\t[15360/21388 (72%)]\tLoss: 0.000358\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[16000/21388 (75%)]\tLoss: 0.000787\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[16640/21388 (78%)]\tLoss: 0.000272\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[17280/21388 (81%)]\tLoss: 0.000402\tAccuracy: 100.000000\tTime 0.360 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[17920/21388 (84%)]\tLoss: 0.000577\tAccuracy: 100.000000\tTime 0.366 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[18560/21388 (87%)]\tLoss: 0.000434\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[19200/21388 (90%)]\tLoss: 0.000397\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[19840/21388 (93%)]\tLoss: 0.000286\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[20480/21388 (96%)]\tLoss: 0.000601\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 44\t[21120/21388 (99%)]\tLoss: 0.000309\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.007473, Accuracy: 99.953358 \n",
      "\n",
      "Train Epoch: 45\t[0/21388 (0%)]\tLoss: 0.000422\tAccuracy: 100.000000\tTime 0.365 (0.365)\tData 0.004 (0.004)\n",
      "Train Epoch: 45\t[640/21388 (3%)]\tLoss: 0.001063\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[1280/21388 (6%)]\tLoss: 0.000269\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[1920/21388 (9%)]\tLoss: 0.000587\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[2560/21388 (12%)]\tLoss: 0.019103\tAccuracy: 98.437500\tTime 0.360 (0.361)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[3200/21388 (15%)]\tLoss: 0.000432\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[3840/21388 (18%)]\tLoss: 0.000823\tAccuracy: 100.000000\tTime 0.361 (0.361)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[4480/21388 (21%)]\tLoss: 0.000686\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[5120/21388 (24%)]\tLoss: 0.002642\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[5760/21388 (27%)]\tLoss: 0.000460\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[6400/21388 (30%)]\tLoss: 0.000622\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[7040/21388 (33%)]\tLoss: 0.000398\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[7680/21388 (36%)]\tLoss: 0.000385\tAccuracy: 100.000000\tTime 0.359 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[8320/21388 (39%)]\tLoss: 0.000672\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 45\t[8960/21388 (42%)]\tLoss: 0.000273\tAccuracy: 100.000000\tTime 0.366 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 45\t[9600/21388 (45%)]\tLoss: 0.000379\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[10240/21388 (48%)]\tLoss: 0.000985\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[10880/21388 (51%)]\tLoss: 0.000327\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[11520/21388 (54%)]\tLoss: 0.000225\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[12160/21388 (57%)]\tLoss: 0.000226\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[12800/21388 (60%)]\tLoss: 0.000530\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 45\t[13440/21388 (63%)]\tLoss: 0.000333\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[14080/21388 (66%)]\tLoss: 0.001039\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[14720/21388 (69%)]\tLoss: 0.000401\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[15360/21388 (72%)]\tLoss: 0.000415\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[16000/21388 (75%)]\tLoss: 0.000404\tAccuracy: 100.000000\tTime 0.365 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[16640/21388 (78%)]\tLoss: 0.000344\tAccuracy: 100.000000\tTime 0.359 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[17280/21388 (81%)]\tLoss: 0.000286\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[17920/21388 (84%)]\tLoss: 0.000870\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[18560/21388 (87%)]\tLoss: 0.000392\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[19200/21388 (90%)]\tLoss: 0.000640\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[19840/21388 (93%)]\tLoss: 0.000350\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[20480/21388 (96%)]\tLoss: 0.000484\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 45\t[21120/21388 (99%)]\tLoss: 0.000531\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.006154, Accuracy: 99.930037 \n",
      "\n",
      "Train Epoch: 46\t[0/21388 (0%)]\tLoss: 0.000358\tAccuracy: 100.000000\tTime 0.367 (0.367)\tData 0.004 (0.004)\n",
      "Train Epoch: 46\t[640/21388 (3%)]\tLoss: 0.000363\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[1280/21388 (6%)]\tLoss: 0.000527\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[1920/21388 (9%)]\tLoss: 0.000269\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[2560/21388 (12%)]\tLoss: 0.000478\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[3200/21388 (15%)]\tLoss: 0.000282\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[3840/21388 (18%)]\tLoss: 0.000177\tAccuracy: 100.000000\tTime 0.367 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[4480/21388 (21%)]\tLoss: 0.000386\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[5120/21388 (24%)]\tLoss: 0.000391\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[5760/21388 (27%)]\tLoss: 0.000226\tAccuracy: 100.000000\tTime 0.360 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[6400/21388 (30%)]\tLoss: 0.000360\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[7040/21388 (33%)]\tLoss: 0.000293\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[7680/21388 (36%)]\tLoss: 0.000318\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[8320/21388 (39%)]\tLoss: 0.000340\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[8960/21388 (42%)]\tLoss: 0.000619\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[9600/21388 (45%)]\tLoss: 0.000454\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[10240/21388 (48%)]\tLoss: 0.000498\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[10880/21388 (51%)]\tLoss: 0.000314\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[11520/21388 (54%)]\tLoss: 0.000318\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[12160/21388 (57%)]\tLoss: 0.000308\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[12800/21388 (60%)]\tLoss: 0.000391\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[13440/21388 (63%)]\tLoss: 0.000551\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[14080/21388 (66%)]\tLoss: 0.000978\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[14720/21388 (69%)]\tLoss: 0.000427\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 46\t[15360/21388 (72%)]\tLoss: 0.000390\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[16000/21388 (75%)]\tLoss: 0.000556\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[16640/21388 (78%)]\tLoss: 0.000559\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[17280/21388 (81%)]\tLoss: 0.000445\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[17920/21388 (84%)]\tLoss: 0.000519\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[18560/21388 (87%)]\tLoss: 0.001105\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[19200/21388 (90%)]\tLoss: 0.000544\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[19840/21388 (93%)]\tLoss: 0.000346\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[20480/21388 (96%)]\tLoss: 0.000320\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 46\t[21120/21388 (99%)]\tLoss: 0.000476\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.005043, Accuracy: 99.883396 \n",
      "\n",
      "Train Epoch: 47\t[0/21388 (0%)]\tLoss: 0.000350\tAccuracy: 100.000000\tTime 0.369 (0.369)\tData 0.008 (0.008)\n",
      "Train Epoch: 47\t[640/21388 (3%)]\tLoss: 0.000467\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[1280/21388 (6%)]\tLoss: 0.000390\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[1920/21388 (9%)]\tLoss: 0.000683\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[2560/21388 (12%)]\tLoss: 0.000406\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[3200/21388 (15%)]\tLoss: 0.000429\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[3840/21388 (18%)]\tLoss: 0.000340\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[4480/21388 (21%)]\tLoss: 0.000559\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[5120/21388 (24%)]\tLoss: 0.000304\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 47\t[5760/21388 (27%)]\tLoss: 0.000524\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[6400/21388 (30%)]\tLoss: 0.000408\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[7040/21388 (33%)]\tLoss: 0.000601\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[7680/21388 (36%)]\tLoss: 0.000507\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[8320/21388 (39%)]\tLoss: 0.000484\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[8960/21388 (42%)]\tLoss: 0.000472\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[9600/21388 (45%)]\tLoss: 0.000342\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[10240/21388 (48%)]\tLoss: 0.000659\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[10880/21388 (51%)]\tLoss: 0.000505\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[11520/21388 (54%)]\tLoss: 0.000394\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[12160/21388 (57%)]\tLoss: 0.000720\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[12800/21388 (60%)]\tLoss: 0.000405\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 47\t[13440/21388 (63%)]\tLoss: 0.000534\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[14080/21388 (66%)]\tLoss: 0.000571\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[14720/21388 (69%)]\tLoss: 0.000438\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 47\t[15360/21388 (72%)]\tLoss: 0.000597\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[16000/21388 (75%)]\tLoss: 0.001096\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[16640/21388 (78%)]\tLoss: 0.000653\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 47\t[17280/21388 (81%)]\tLoss: 0.001063\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 47\t[17920/21388 (84%)]\tLoss: 0.000826\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[18560/21388 (87%)]\tLoss: 0.000579\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[19200/21388 (90%)]\tLoss: 0.001670\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[19840/21388 (93%)]\tLoss: 0.000757\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[20480/21388 (96%)]\tLoss: 0.000868\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 47\t[21120/21388 (99%)]\tLoss: 0.000703\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.004154, Accuracy: 99.906716 \n",
      "\n",
      "Train Epoch: 48\t[0/21388 (0%)]\tLoss: 0.000484\tAccuracy: 100.000000\tTime 0.368 (0.368)\tData 0.003 (0.003)\n",
      "Train Epoch: 48\t[640/21388 (3%)]\tLoss: 0.000626\tAccuracy: 100.000000\tTime 0.362 (0.364)\tData 0.001 (0.003)\n",
      "Train Epoch: 48\t[1280/21388 (6%)]\tLoss: 0.000622\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[1920/21388 (9%)]\tLoss: 0.000813\tAccuracy: 100.000000\tTime 0.360 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[2560/21388 (12%)]\tLoss: 0.000741\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[3200/21388 (15%)]\tLoss: 0.000587\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[3840/21388 (18%)]\tLoss: 0.000578\tAccuracy: 100.000000\tTime 0.362 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[4480/21388 (21%)]\tLoss: 0.000583\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[5120/21388 (24%)]\tLoss: 0.000734\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[5760/21388 (27%)]\tLoss: 0.000733\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[6400/21388 (30%)]\tLoss: 0.001097\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[7040/21388 (33%)]\tLoss: 0.000619\tAccuracy: 100.000000\tTime 0.365 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[7680/21388 (36%)]\tLoss: 0.000656\tAccuracy: 100.000000\tTime 0.359 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[8320/21388 (39%)]\tLoss: 0.000828\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[8960/21388 (42%)]\tLoss: 0.000715\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[9600/21388 (45%)]\tLoss: 0.000467\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[10240/21388 (48%)]\tLoss: 0.000511\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[10880/21388 (51%)]\tLoss: 0.000783\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[11520/21388 (54%)]\tLoss: 0.000585\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[12160/21388 (57%)]\tLoss: 0.000690\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[12800/21388 (60%)]\tLoss: 0.000860\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[13440/21388 (63%)]\tLoss: 0.000541\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[14080/21388 (66%)]\tLoss: 0.000723\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[14720/21388 (69%)]\tLoss: 0.000726\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[15360/21388 (72%)]\tLoss: 0.000676\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[16000/21388 (75%)]\tLoss: 0.001073\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[16640/21388 (78%)]\tLoss: 0.001057\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 48\t[17280/21388 (81%)]\tLoss: 0.000584\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[17920/21388 (84%)]\tLoss: 0.000558\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[18560/21388 (87%)]\tLoss: 0.000836\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.002 (0.002)\n",
      "Train Epoch: 48\t[19200/21388 (90%)]\tLoss: 0.000947\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[19840/21388 (93%)]\tLoss: 0.000841\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[20480/21388 (96%)]\tLoss: 0.000796\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 48\t[21120/21388 (99%)]\tLoss: 0.000638\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.002938, Accuracy: 99.953358 \n",
      "\n",
      "Train Epoch: 49\t[0/21388 (0%)]\tLoss: 0.000745\tAccuracy: 100.000000\tTime 0.365 (0.365)\tData 0.004 (0.004)\n",
      "Train Epoch: 49\t[640/21388 (3%)]\tLoss: 0.000664\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[1280/21388 (6%)]\tLoss: 0.000540\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[1920/21388 (9%)]\tLoss: 0.000675\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[2560/21388 (12%)]\tLoss: 0.000696\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[3200/21388 (15%)]\tLoss: 0.000492\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[3840/21388 (18%)]\tLoss: 0.000760\tAccuracy: 100.000000\tTime 0.358 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[4480/21388 (21%)]\tLoss: 0.000776\tAccuracy: 100.000000\tTime 0.358 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[5120/21388 (24%)]\tLoss: 0.000406\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[5760/21388 (27%)]\tLoss: 0.000628\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[6400/21388 (30%)]\tLoss: 0.000549\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[7040/21388 (33%)]\tLoss: 0.001181\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[7680/21388 (36%)]\tLoss: 0.000786\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[8320/21388 (39%)]\tLoss: 0.000540\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[8960/21388 (42%)]\tLoss: 0.000573\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[9600/21388 (45%)]\tLoss: 0.000735\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[10240/21388 (48%)]\tLoss: 0.000544\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[10880/21388 (51%)]\tLoss: 0.000577\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[11520/21388 (54%)]\tLoss: 0.000963\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[12160/21388 (57%)]\tLoss: 0.000761\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[12800/21388 (60%)]\tLoss: 0.000734\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[13440/21388 (63%)]\tLoss: 0.000869\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[14080/21388 (66%)]\tLoss: 0.000712\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[14720/21388 (69%)]\tLoss: 0.000831\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[15360/21388 (72%)]\tLoss: 0.000964\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[16000/21388 (75%)]\tLoss: 0.000840\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[16640/21388 (78%)]\tLoss: 0.000757\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[17280/21388 (81%)]\tLoss: 0.000690\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[17920/21388 (84%)]\tLoss: 0.000860\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[18560/21388 (87%)]\tLoss: 0.005961\tAccuracy: 100.000000\tTime 0.359 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[19200/21388 (90%)]\tLoss: 0.000954\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[19840/21388 (93%)]\tLoss: 0.003011\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[20480/21388 (96%)]\tLoss: 0.000917\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 49\t[21120/21388 (99%)]\tLoss: 0.000888\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "\n",
      "Val set: Average loss: 0.002259, Accuracy: 99.930037 \n",
      "\n",
      "Train Epoch: 50\t[0/21388 (0%)]\tLoss: 0.001053\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.004 (0.004)\n",
      "Train Epoch: 50\t[640/21388 (3%)]\tLoss: 0.000878\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[1280/21388 (6%)]\tLoss: 0.000720\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[1920/21388 (9%)]\tLoss: 0.000868\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[2560/21388 (12%)]\tLoss: 0.000863\tAccuracy: 100.000000\tTime 0.360 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[3200/21388 (15%)]\tLoss: 0.000841\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[3840/21388 (18%)]\tLoss: 0.000842\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[4480/21388 (21%)]\tLoss: 0.000986\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[5120/21388 (24%)]\tLoss: 0.000755\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[5760/21388 (27%)]\tLoss: 0.000983\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[6400/21388 (30%)]\tLoss: 0.001022\tAccuracy: 100.000000\tTime 0.366 (0.362)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[7040/21388 (33%)]\tLoss: 0.001093\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[7680/21388 (36%)]\tLoss: 0.001082\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[8320/21388 (39%)]\tLoss: 0.001185\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[8960/21388 (42%)]\tLoss: 0.001059\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[9600/21388 (45%)]\tLoss: 0.001126\tAccuracy: 100.000000\tTime 0.361 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[10240/21388 (48%)]\tLoss: 0.001047\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[10880/21388 (51%)]\tLoss: 0.000740\tAccuracy: 100.000000\tTime 0.363 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[11520/21388 (54%)]\tLoss: 0.000942\tAccuracy: 100.000000\tTime 0.364 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[12160/21388 (57%)]\tLoss: 0.001109\tAccuracy: 100.000000\tTime 0.362 (0.362)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[12800/21388 (60%)]\tLoss: 0.001085\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[13440/21388 (63%)]\tLoss: 0.000713\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[14080/21388 (66%)]\tLoss: 0.000874\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[14720/21388 (69%)]\tLoss: 0.001910\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[15360/21388 (72%)]\tLoss: 0.001109\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[16000/21388 (75%)]\tLoss: 0.000857\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[16640/21388 (78%)]\tLoss: 0.000921\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[17280/21388 (81%)]\tLoss: 0.001147\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.001)\n",
      "Train Epoch: 50\t[17920/21388 (84%)]\tLoss: 0.001190\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[18560/21388 (87%)]\tLoss: 0.000848\tAccuracy: 100.000000\tTime 0.365 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[19200/21388 (90%)]\tLoss: 0.001359\tAccuracy: 100.000000\tTime 0.363 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[19840/21388 (93%)]\tLoss: 0.000954\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[20480/21388 (96%)]\tLoss: 0.001121\tAccuracy: 100.000000\tTime 0.361 (0.363)\tData 0.001 (0.002)\n",
      "Train Epoch: 50\t[21120/21388 (99%)]\tLoss: 0.001033\tAccuracy: 100.000000\tTime 0.364 (0.363)\tData 0.001 (0.001)\n",
      "\n",
      "Val set: Average loss: 0.001736, Accuracy: 99.953358 \n",
      "\n",
      "\n",
      "Val set: Average loss: 0.001736, Accuracy: 99.953358 \n",
      "\n",
      "\n",
      "\n",
      "Best validation accuracy: 100.000000\n",
      "\n",
      " Total time taken in seconds:  6354.947791814804\n"
     ]
    }
   ],
   "source": [
    "# Trainging Loop\n",
    "\n",
    "training_start_time = time.time()\n",
    "\n",
    "best_acc = test(valid_loader, model, criterion, device)[1]\n",
    "\n",
    "for epoch in range(1, epochs + 1):\n",
    "    acc = train(train_loader, model, criterion, optimizer, epoch, device)\n",
    "    acc /= len(train_loader)\n",
    "\n",
    "    current_acc = test(valid_loader, model, criterion, device)[1]\n",
    "    \n",
    "    if current_acc > best_acc:\n",
    "        best_acc = current_acc\n",
    "        print(\"-------------------- HIGHEST ACCURACY ---------------------------\")\n",
    "        torch.save(model.state_dict(), './model.pth')\n",
    "\n",
    "training_stop_time = time.time()\n",
    "\n",
    "best_acc = max(best_acc, test(valid_loader, model, criterion, device)[0])\n",
    "print('\\n\\nBest validation accuracy: {:.6f}'.format(best_acc))\n",
    "\n",
    "print('\\n Total time taken in seconds: ', training_stop_time - training_start_time)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1-PusJd6f_FQ"
   },
   "source": [
    "#### Testing time!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "-BoOtctOAApZ",
    "outputId": "e61604ad-3213-4ed1-bf02-798489b78168"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Test set: Average loss: 0.024737, Accuracy: 99.991126 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "model = CapsNet(A=A, B=B, C=C, D=D, E=num_class, iters=em_iters).to(device)\n",
    "model.load_state_dict(torch.load('./model.pth'))\n",
    "test(test_loader, model, criterion, device, test=True);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "B93ieiCMZ5eB"
   },
   "source": [
    "## **Inference and Visualization**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "q-xUh6iustlT"
   },
   "source": [
    "#### Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "0IG8C_OQZ8H6"
   },
   "outputs": [],
   "source": [
    "def AA_andEachClassAccuracy(confusion_matrix):\n",
    "    counter = confusion_matrix.shape[0]\n",
    "    list_diag = np.diag(confusion_matrix)\n",
    "    list_raw_sum = np.sum(confusion_matrix, axis=1)\n",
    "    each_acc = np.nan_to_num(truediv(list_diag, list_raw_sum))\n",
    "    average_acc = np.mean(each_acc)\n",
    "    return each_acc, average_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "krZ2Z-9bZ-mA",
    "outputId": "13a237da-6dae-4d02-ba48-de8a9405bdb2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Test set: Average loss: 0.024737, Accuracy: 99.991126 \n",
      "\n",
      "------------------------ CONFUSION MATRIX -------------------------\n",
      "[[2652    0    0    0    0    0    0    1    0]\n",
      " [   0 7459    0    0    0    0    0    0    0]\n",
      " [   0    0  840    0    0    0    0    0    0]\n",
      " [   0    0    0 1226    0    0    0    0    0]\n",
      " [   0    0    0    0  538    0    0    0    0]\n",
      " [   0    0    0    0    0 2011    0    0    0]\n",
      " [   0    0    0    0    0    0  532    0    0]\n",
      " [   1    0    0    0    0    0    0 1472    0]\n",
      " [   0    0    0    0    0    0    0    0  379]]\n",
      "\n",
      "\n",
      "---------------------- CLASSIFICATION REPORT ----------------------\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00      2653\n",
      "           1       1.00      1.00      1.00      7459\n",
      "           2       1.00      1.00      1.00       840\n",
      "           3       1.00      1.00      1.00      1226\n",
      "           4       1.00      1.00      1.00       538\n",
      "           5       1.00      1.00      1.00      2011\n",
      "           6       1.00      1.00      1.00       532\n",
      "           7       1.00      1.00      1.00      1473\n",
      "           8       1.00      1.00      1.00       379\n",
      "\n",
      "    accuracy                           1.00     17111\n",
      "   macro avg       1.00      1.00      1.00     17111\n",
      "weighted avg       1.00      1.00      1.00     17111\n",
      "\n",
      "\n",
      "\n",
      "------------------------ OVERALL ACCURACY -------------------------\n",
      "0.9999112644965227\n",
      "\n",
      "\n",
      "------------------------ AVERAGE ACCURACY -------------------------\n",
      "0.9998976969310671\n",
      "\n",
      "\n",
      "--------------------------- KAPPA SCORE ---------------------------\n",
      "0.9998305268809668\n",
      "\n",
      "\n",
      "-------------------------- TEST ACCURACY --------------------------\n",
      "\n",
      "Test set: Average loss: 0.024737, Accuracy: 99.991126 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "def report(model=model, test_loader=test_loader, criterion=criterion, device=device):\n",
    "\n",
    "    # Getting prediction vectors from test loader\n",
    "    y_pred = []\n",
    "    y_test = []\n",
    "    model.eval()\n",
    "    test_loss = 0\n",
    "    acc = 0\n",
    "    test_len = len(test_loader)\n",
    "    with torch.no_grad():\n",
    "        for data, target in test_loader:\n",
    "            output = model(data.to(device))\n",
    "            y_pred.extend(output.argmax(dim=1).float().tolist())\n",
    "            y_test.extend(target.tolist())\n",
    "\n",
    "    target_labels = ['0', '1', '2', '3', '4', '5', '6', '7', \n",
    "            '8']\n",
    "\n",
    "    classificationReport = classification_report(y_test, y_pred, target_names=target_labels)\n",
    "    confusionMatrix = confusion_matrix(y_test, y_pred)\n",
    "    oa = accuracy_score(y_test, y_pred)\n",
    "    each_acc, aa = AA_andEachClassAccuracy(confusionMatrix)\n",
    "    kappa = cohen_kappa_score(y_test, y_pred)\n",
    "    test_loss, acc = test(test_loader, model, criterion, device, test=True)\n",
    "\n",
    "    print(\"------------------------ CONFUSION MATRIX -------------------------\")\n",
    "    print(confusionMatrix)\n",
    "\n",
    "    print(\"\\n\\n---------------------- CLASSIFICATION REPORT ----------------------\")\n",
    "    print(classificationReport)\n",
    "\n",
    "    print('\\n\\n------------------------ OVERALL ACCURACY -------------------------')\n",
    "    print(oa)\n",
    "\n",
    "    print('\\n\\n------------------------ AVERAGE ACCURACY -------------------------')\n",
    "    print(aa)\n",
    "\n",
    "    print('\\n\\n--------------------------- KAPPA SCORE ---------------------------')\n",
    "    print(kappa)\n",
    "\n",
    "    print('\\n\\n-------------------------- TEST ACCURACY --------------------------')\n",
    "    print('\\nTest set: Average loss: {:.6f}, Accuracy: {:.6f} \\n'.format(\n",
    "        test_loss, acc))\n",
    "\n",
    "report()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mfyI8o-raJm3"
   },
   "source": [
    "#### Visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zAsXUh8TaaRR"
   },
   "outputs": [],
   "source": [
    "X,y = loadData()\n",
    "X = applyPCA(X)\n",
    "X = padWithZeros(X, windowSize//2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "emvkETrFaslj"
   },
   "outputs": [],
   "source": [
    "def Patch(data,height_index,width_index):\n",
    "    height_slice = slice(height_index, height_index + windowSize)\n",
    "    width_slice = slice(width_index, width_index + windowSize)\n",
    "    patch = data[height_slice, width_slice, :]\n",
    "    return patch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "aB1sYuXwaupH"
   },
   "outputs": [],
   "source": [
    "height,width = y.shape[0],y.shape[1]\n",
    "outputs = np.zeros((height,width))\n",
    "model.eval()\n",
    "with torch.no_grad():\n",
    "  for i in range(height):\n",
    "      for j in range(width):\n",
    "          target = int(y[i,j])\n",
    "          if target == 0 :\n",
    "              continue\n",
    "          else :\n",
    "              image_patch=Patch(X,i,j)\n",
    "              X_test_image = image_patch.transpose(2,0,1)\n",
    "              X_test_image = X_test_image.reshape(1,X_test_image.shape[0],X_test_image.shape[1], X_test_image.shape[2])\n",
    "              X_test_image = torch.Tensor(X_test_image).to(device)\n",
    "              pred = model(X_test_image)\n",
    "              pred = pred.cpu().numpy()\n",
    "              pred = np.argmax(pred, axis=-1)\n",
    "              outputs[i][j] = pred[0]+1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5Hk9UvESa4Be"
   },
   "source": [
    "##### Ground Truth"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 432
    },
    "id": "oaSqDBy_a5Ep",
    "outputId": "40229b4a-c0cd-402e-eb9b-d5207a5d1673"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAGfCAYAAAB7vL8nAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO19bcw1x1neddfOByU0jkNqWbbVJMUqyg9K8r7KcyKiqIKGkrRqaBVVQW2xquxjiY+KilbFCKmC/ipULRApSshzSJsgRJLSolgRNAQnFeqPc8L7ku+kxm9SKttyYj7ilLYSYLj7Yz52ds/Mzszu7O7Mnrned89zzu7szOzsXHPfc889M8TMqKioOA/8hbUzUFFRsRwq4SsqzgiV8BUVZ4RK+IqKM0IlfEXFGaESvqLijDAL4Ynou4joUSK6RUQPzZFGRUVFPCj1ODwR3QbgdwC8HsATAH4LwPcw8+eTJlRRURGNOST8qwHcYuYvMfOfAHgfgDfNkE5FRUUkbp8hznsAPG78fgLAxdANRHQW7n7XjO83I+4JDTs7rnWfwcRN/VGRA5iZbOfnIHwQiOhBAA+ulf4auGF8t76NHjgwHFRzGRR4JBgAMW4Y9YiIAaZ5061IijkI/ySA+4zf98pzHTDzuwC8CzgfCU9ouRlMZg8Ykng6hdRgHTVXshePOfrwvwXgfiJ6GRE9F8BbADw8QzpFwuTHUCsX2gJScMh4MATJ1aHTrGQvFsklPDM/S0Q/CODDAG4D8G5m/lzqdEqGKeltCJXXQsumWbgnNBAGy5hNJYxBoEr2IpF8WG5UJs5EpTdhPnCfO+EKuiDkPNzjrlQPy1BFJnAZ7aqn3UqYyp1ZW8he5JXs20ElfIYI4ZZUtNPykKGt8cwEIqHQ22VFRYmohF8RefGINd872mC1zW0KlfAZYLx6nkixt0RD1BrsKraDSviVMZZSDKTTtalroIOMu9J9e1jN066ihW+YznVPaka24+up3IIqckOV8BkjpBGYrtSLGLQKX03ym0YlfMGQBvXom5RxrjvWXlX4c0AlfOmIZSm1jFdDb0PGOVYfZ+catU1UwmeOIZ7FS2R7bK54WvdaCMNedHoVuaESPhOMV6fDaKhn6fUmwfiG3lhb66vKvwVUwmeEWEJFd98tM96G0iQZrnrabQeV8JnBxi0Xscn4TJbYSeJVsm8JlfAZYg6CncybDxXble2bQiX85mE3sVc1/TxRPe0yRd/7Ltr3zZj1puPsLWJRcX6oEj5jTKKkQXYiNjzoaMZFMypyRyX8BtGf4ton+Kpk5+rDsyYq4TNH6KKX3XsypZTsZlTtYj1UwheAONJnSnaJSvd1cfaEz5seLcJIz50lpdUSVdlQrK6eszrOnvAlVcBB0hsniLg33dUC7t1UcRY4S8JvsZp3xtVZyPWSGrOKZXCWhN8iEcS4PYWvX1EXujhLnCXhS8aQWl85XOFDJXxFxRmhEr5AeC32WzRSVCRBJfwWYbQIlfsVJirhM4VvLbnQcXm9RFVFBSrhs4W5ltwYqI0q6kSZChNnS3gtPNUKrutmp4vAzIRY7CvZK0ycFeHZOIhHC8/FELLIZLCffW6NWsUq2PwCGGYlz53gGsYezZRARKvFM6q0r9gs4RW3iyF5HxHsNFfHsa2M03fIqTvHnS82qdKPIXvpBAjOfwFdmYr5sDkJ3ye7OalEb45qC4fySa+wpWepSItNEb6/uSL3rdTG7zFbNM+Fjp1h7sSom2ZtGM4Lm1HpfWS3+bB01nlbyYptahpT8xC7HFYl+/khC8Jfm3i/b9tktSmi9Rqtv0a7mfelGp5K9vMEMa+v2BKNNyO5yG5KeJ/6asZx0g2YGbb8T8nDkBYzmH5tATYFZrsYy0LCj4HykgtpKnxj0GrYeymym15+tvxPkfL9/PviqXPozwtFEd70lPMhljRLOaaEDhmq/I8hfuVvhQvFWOnHOtLkoq2O8g2Qw4ipn6Gq8eeLYiT8GN/3IMv3WDEaCFN1H2OpGGvdqEthVdiQNeFjVPgp8c/lera+ObSiootsCW/OaJvKx6H+MCm9OSGmSvVOXBMMiUHj8nUW3VkhO8LPJdFNF1p9Tn0mmpGWMu9Ljhp401AtWG0ZikdWRjufA81UDAnz1jknnmJjDYrWuGTyqYjum0kXxHYCmEnuL187/yUjCwl/DafOL+YxBf24bNFpso9IzOx6pMhnTvPWGejsVVdRPrIg/M1rp4Qh4xhb30xpGUSkiHSU+p6jVO8jdMHLIZU9ZPWdivyRl0rvqPhTqln4PPGwDnNK9V0lC6zgznty9vT5xfBdq3alWH2nYl1kRXhgwN9dESOAaLEaQci49BxEX1J9Hz8deIUWqWI2ZEd4F2IrrM2nfEydLVGihzyrM4xvplFF0ciiDx+CKLKfBI7fkCHlWDoAtYPzIlKdHP3xkIk1vKTaUbE4vIQnoncT0dNE9Fnj3J1E9BEiekz+fZE8T0T0NiK6RUSfJqJXpcjkGEt4J2jEhgxz+AEsP+WWRk+5q1zfNkIk/H8E8F29cw8BeISZ7wfwiPwNAG8AcL88HgTwjrEZMx1ZJkvYQMKZ6ntqqb4kSCZus2WEGzHT5aciH3gJz8y/CeAPe6ffBOA98vt7AHy3cf69LHAAcAcR3R2bqSHixYzNh9ZZs2FJRfSlpfoJBtIPGqason6TGNuHv4uZn5LfvwzgLvn9HgCPG+GekOdOQEQPEtENIrqB32vPu4gXS6JQ4pZufa+oiMFkox2LNbKi6cLM72Lm68x8HS8R51xEH2vs8nVjNyfV50BV7TeFsYT/ilLV5d+n5fknAdxnhLtXnhsFF4GmeN8B6b3ktivVK9u3hrGEfxjAA/L7AwA+aJz/Xmmt3wH4mqH6ByOltOwsXY00Ul3bETZLdIUtqixnDmYePAD8EoCnAPwpRJ/8rQBeDGGdfwzAbwC4U4YlAG8H8EUAnwFw3Rc/M+MawNw74DkY4NB/nXgn/gvNX34HF5rveow5XFzLaplqlZMch47W8HlPCkbBma+IRRHLVMeqx0vM2jSHAccmt36TCoDiPA0rtomsCB+LqcY7H/rTa0fFMeHelBCehkLjsxG/NgbngaIJPxemWt5VRwoj758DyvsOmGCw5NowlI5K+B5S9NVzIfkJ1DzgsS631WhfPCrhDUwhuynVx8axNqr03j6KmQ8/hJjFMQbvH3MvtjMWv4VnqBhG8YRXlXR0t7SqqSfIzf5QkQ5nqdL3PeXiI2i/bpkUVcXfHjZBeEb4WndJ/N63zHJ0H89ZrNViXyQ2QXgf2VMQfYuVm9Wn5eF85VS7QmViE4R3IcUEl7Poz3rmEdsubbo8NoyiCd8fCutcSyCBtijVTZjOOGNdb7deRltDsVZ6V7895dz0s5BiZP2qf6sizsVFuGIaiiS8jewpJricb4Ueu43mOZdZmShOpe8L9VSz2ebctbYE1O7PeaAICa8NZ9wbMkq1Ko7+qOgjSK3nWoalIFsJby7f4VrcMmrufLqsbR6d1UgQMC6/Fd/iM0CWEn5IxY5V37fk674ERHmNE9n6XsTfW7EMspLwpkQ/uTZyTH3OardFrUGXfW9lEW85siA7M0GsrrTF0ikfWRD+GtyqO5CfV5fZMG2uWg/MmfdBLaNGek/5ZLmqSIQsCH/zWm85aUowwWUm9BumTZLeAXNm4skzk9kLEJt3nlXhFIIs+/DATCRn4HDZ/txdhSXUWdiiVmA3eo0zM4GIEbpzb8X8yIbwKRaM9KfBIIPxTFfeqhgyRk+8brfDtKrPPTwW0jtXxjuWGalkzwdZqPS4uZwl/bDf6+/b0Di5089Y8nmG0lLSfZG1xCuCkQfhFwKBsLsyqunxMinjlyebsIqbKghlRHoAVbxnhrMiPACAulL+eDkQFghe+37xvr3STrRFXOlI8/eXyfHdPKf67hV5YXuEZ+DQAO4dtAgXV13Sh+y2lSXpAYAJJA0IrRWdcWjmTdZfHNVQlyM2RXiGsMLvmkZo6y4CGjVx1zRepmbrwssE6gxdsnhoJlzsh29NgUro8rApwpvQpLdcI4ghOSXl6XDplfKhqn1SsJDW5ta1UH9JL1Al/rIwlDGJRiA/Mg4tV1KxFIohvBp6UpXfBgKw23PXEu+qZD1GpLTfTY+H9Z7UHSiDnOH1o78SMiU6JNfrjIYckAXhlWutC8zAsQFwbECHSxwbJ40BwxK/a5rBiHf79uKuaUAeMR8i5af247UcNN35Oq6HBaJyPRtkQfhBsJC+u6ZrhRomZ1u7DpcDriJMHW3gcElJxPwU0hPQIbh2bTf76gbxc1GSQ/PR7YhULI38CU/Axb5bQfrkt0EReVDK9+LeNY1wvR2oj6F9+SlVWg+wGW5zrGasgtrJKTGp9Se5JwT3/g4FHNLNKuZH/oSHHOC5uOqePA4wUxJZkX5IylNPyu+aBry2w7xkvFDvGdQbY2MY6j3JKamWaLRzDhis+v8zPFqwtq4976p+vxayIvxgXew5zADAoRlSwdtK5ZPyfQzbCMKQxm2XRV52eyntZdyexLTa3M8D0+xmAK8VxJr+Jnyci0BWhAeG3vtpTRmSxoSeUjBghu8rDz5j31JDdAQxnu7yZrNCWv1OvPB6zjk5gXPN2AaRFeF9yt7F/lTKHwekPAE9o5wjHPFJvKn6mpM1haELvVaHTaIro590zmmH7OaRplP4Wrm+HLIg/E2EvXRCl/SH/b47GcZyw64nvV36wEk8noH5oPzOrKZ2jHdqGE+eI2Jt2ddg2d9fVa2vWBPEIY7kc2fi1Ozsgeyfkn++NUOM4SvLPl90+8NmyEND3RGAiys4Auu4Q3etnYdjFhckNS21y3T5n4YeJ1GOWsyZlHjCuriGC8z2Zj0LCR8PoaOG2Hv7fXm34D6V8oNj+CuC0brS2qGVdyHVAVFeG2NGJXs8CiV8JOh04Qsb+tM5UxnvUi20YVrfnSt+qrCsnJNok7Sog3vjcBaEF31/w912wBe/b7xb25tVj6X3re/KIKe88Xred6Q8d1bMf1rdaNmFPbaK8gkvdFtvsL6DjT1Q0Kmo6236gQEt8etVbRTR0braat9bI6E1/YbmbF+ql950ZLOI5VgwCHRshAo7VNsIuADjAMKuP/Bu4OIKwLGTgLcW91aZGsjrBEJIonOP4Pq51UXpUbemYmJaPlLlRfTXeUSEa5dGXihfwsv+edj0VvHiB41xBPDuqueWuy7UOLryT2mNcayH5oTUl88UYDRQhr8cjZI2jHkL4vHWf385oXjCE4yxdk/dPfG+c4QRi122gX2UWNJ4x+CeMc7su1PHq84Vgz6kH345iLNJ9Hs7FRtQ6QFodf0oe3mDmr3oFAvVfqhLT4JAxwZYWy00BbE2xlmgtXpHJFoXMJaPnmujiDnU+orpKF7CKxDkIhYJVVSlEQy575ph55LySlJRgMjqX9Vj9ipxZfhTnnibHLSrcGEzhGcIK3wQmVau4aNJHwM9eGGM1xsLapjOOazV/LJQkAkiG2yG8Gqs/diQd5SOpVddyMiu0p5dE2/6eVgbigRaogPdoTwp0RXRj426L33ue869acFymC6HQi8ImyG8AAUvOy3652EdgN0+vLquawOTNDYn0agPMkO1i2rsrsryxetK9VJynQ8KnTxjh5ooc7EPM0MxVH/WH6+yjIfUsUUn1XDnz0nCZHyKgCwapYDnToW0E2qqCTAErskz27DSSxAAjpDGhHBVnaUenLKqnUxqi0BH0pkkV1s0Kyec/q40ZPbjS0SxGc8CXpWeiO4joo8R0eeJ6HNE9EPy/J1E9BEiekz+fZE8T0T0NiK6RUSfJqJXzf0QHTB5jWLKcn1o4ioPBfpzx8Q6SrVRNjbD6m4udqEt7xYXwDXoUimaD0L68M8C+OfM/AoAOwA/QESvAPAQgEeY+X4Aj8jfAPAGAPfL40EA70ie6wEQWes5gK53WcdhJyRexHndzdWXZ6DTR1eq+cliF7NazMYjo6ycJbwqPTM/BeAp+f2PiOgLAO4B8CYAf0MGew+A/wbgR+T597IwDhyI6A4iulvGMxkhWjAxganbj1eyWXunyUsXrV7sizUiBzKko+Hp5jWuL69zoR1n1G+pgei4qLt1dO36ViDSSk9ELwXwSojpJXcZJP4ygLvk93sAPG7c9oQ814/rQSK6QUQ3ovIQGEh0udWWTSynkTqYFSF2Qi3783KrlepqeI2OjdRAWlGvXW09c+eXQKYKx/mBmYMOAC+AWH7u78vfz/Suf1X+/RCA1xrnHwFw3RM3hxwcEY7BfGigBtvdYRlB4VRYHBrl0hKW38B/oc/Wf0YcGpF/UZLWPBsl7Y9TBY7MT3ie54m7Hv337uBaINmfA+DDAH7YOPcogLvl97sBPCq//xyA77GFm0p4VWn84TQfguIUhA8Jy5pcUfmdgfCHBmGNj0n4gUat2zCwP94xldA45qjk9TDf50jCQ2hj7wXwM73z/xbAQ/L7QwB+Sn7/2wB+Td63A/DxgDSiKk1o2HByRhCZVdiI/Ib+m+PFw0P4DstbsseUc3x+7O+RAX15rvTP5XBxLaQP/20A/jGAbyeiT8rjjQD+DYDXE9FjAP6m/A0AvwrgSwBuAbgC8P0BaQRjjE95SKy7K0bIVFE2jGRhMSPYYh8YZRR0j1715VmlI6ltLm+tfO1X8rzTAw8Fef6VhhAr/X+Hm2ffYQnPAH5gYr6SYLcX02Av4G8oGCS2q/NCUYKjKmboqjjzwBytEB9iizfFMMukGoZlues0OXEWQ529Nzs25kvfh1x6OoRokZIbx8tgAgcPuQVmNRadOPsWe4PsSq9WhTDHhJoQqNRnKYwzx8YJLxAmuduwQ/WMIObIH/b79XeZ9UATB9yV5uYhdX6GmIdAxFBr2M9Fd9OjwZZnpT0VtRhPIdg84Vl7pgyHU0TeNY3fWUZGeYxwzQ3ty09tQ5TVRn3R0Zlqe3e4HoCYJis8D5dlWf9x17QhnAM2T3iFIGlMal16f9h2l5pwhoaSfizn2fg4cb+VObAmz2KX2rxYxsFzFyrCsanZcjYQAL4CjiAxZjBQqVXYwyVh57VXGZrDDESJj5a7lkG1x1yI3+6M6rsPznQz3t66ZJyFhCdDdfWGxfC+82bAiz37O/2W+H1Sfoxazz2ya/3dkZZW/VdCf66PNQBQDXeJcRaEVzhckr8Cyc48HS79QfVuNvPUyphYtTS3Ssa+xbv10QhKo28PWArsbxwrIhHqSz/ngSW8jxiMQzPoXtoeYzzvIl1RE7vbsorTiJ/Bwv225zbc8apzlIeKz/TCc4WtR36Hi2ub78NryA768VKsbDssOGL65+P68qFTZ0PjNMfSiRmHS7Fmn80YYTq/2JNgQE4x1uvXwxpwGchsVGGfAGtL98UkvKg2M/vXh0t5GTiplFd5waEZvI/Va5eCAEpyW15PzDOlPli+B52/eoSX3dlLeACxknt3FbBDTXS87R0cFjQmWu13MBRe28PMde8YJx54fbGuvBBTL36pUu1Hq5+kivZkODPCC8v6ASHDbvJ6yFZTZOxMG0VPPwytOiz82EQADA3lMQsPuDmWulWN32DJpS3Ws8VZWekBY9gtIKzyJQ8Jq1xaDoEr4qi8rGGFZuDUp16jS3ZBRBl2gXWt7WUXWqIVXqzdfx/Th9d9u7H9S9nn5qC0wsPq/nNEn5MFn9L35R1pmRb6k6OTbzPo9LSD87dAOudwuLhWnoRnqVpOiiLcq4OUlTogSSYSY/gR+VtEyksmgdF1zmFjOev+LXoNwOX70LbSM2tzxXgU1YdnoN3eWL35EZVRLP4o9qHzuduCgJ20Vvnmv5MMc7zEbH352BjZJLk+ScZcd3uMS+1K00kTbkKTeo46PjcJRUn4zgQQr29mWGxhk2oMK3xg0BhRNJuU75PdstCF/sbh9oo54XytfVfhtTNaKIoifEoQxHTQ0GWnmcLd5ndXwvGFc6qVJ7vSGM9C0jW3swRWQJQzZDMoTV6gG7RRnC3hAQh1PXRFnCjRIsKGNibqjpBJNVEkI0MrMuNm2556Svqz3wahWQc9Nr8kqlY/HkX14WcBQe6RPtznJgC85zB3V9nvFxtDxPXlfS638d3Y7v47rXR0z41ncRm9NmK2JbhioPv5aU0kZ4PzlvAAlEddUP+cySDxMJTH27GhYJbMWX+VlV7sHusyPsqzxMYKsua93Bm/J+KVjHs83CpWOFE+4VOM1ZCaOuuJKKILQJDz5RFoGDTuS67aA+1adSHhoOKXhetYPUcbPtsbkmFoGI6nW2zPFlkTPrhrPfndk1iYMmC+PJu7N3pjFdpDzNp3bTqeuKNJH5gHbqX8Sfz9pbIY4CQt7in63QnXNdv1Cjcy7sMv20lTY/O+deyFBCZQ68niiRhh/vj9NIJCJsZQI8LUG52UclYZFGZ4VUPloEccids8VHiRtYRfGsqrzsc2JeUOAf1zjrARxCKVEU0RmFWk1sTQkehLeuHZkuh4EVSyByNjCT9uqeJJegEJD7mDHJgajIeAC/anpzSCYxPnfaf68nPbpjrj2q7EpATl3KSpOVKaS54yx+Yk/GSpFyjlASXg/O41BLl5RYTFPhTjn9cwi/XnwRs+9orgWmtfmFgh3ZvqhBOOYgnvNBUFktAJEnvSUahoJYA4ID1ZKQ+XSDZMp9asjIXVAi5JTkDrkUdstH/ridEAt4eKQBRJeFH1BjzCmPzeYoOIsULJHr3Ham8O08XIZBupjcVpx9vLNJONyCC2dxKNARsKwFq7zEUgA6egEpA94W2SXKmbp9Wwr6IuVAXkOL5ftSftZx+at5MnJPv5GAgpzq1UV9pHI7JFhlGOdIB8Ka9KMt8c5oPsCd/aknpNuM12ZEotdXURzispHxw0Kl/KgJd2BKy7kSQxy66Mmcky0MlyFfODIF5j9kM/EwEd5lGteDtYOyJXLRFC/Eej1nyTfva7fR7mZZF35JAVLwbrARu9kzMHs926k72EV4iVbKxuUp3SkWmy/PT5kxEMT70AC96uabKRRplr7B0MCvJKdi+KIXwU2LDUT60BJIbTiNnLeLGjLONwqd1YnGF5dxW9L11FAIac8Cs2Sng1rNT5PQasre/C0DYYVIQ5XmLXNNJ/fqDWyX3pYobpNocJmpfzdtOMU3GCjD3tVobedQG4ECeAq4FaRAwYk2R2TSPWqXesmUck0zhegulqpF9heVD2VzEDDxjrUTnYnFbV3oltSngLlL94uFQh3bkVIwXkceiRPvMGdk0j1XtHnuQqtzEr45QLsda2GNNnY8bttIXAvF54E+LeIs6G8GR8RtzQ+X249Dn0nMY/ZJxrjYKB3nolgqEXm7eVHbH9fIJkUel+iqrSx2JoJI0AhpDYu6bRp48n68cZt5DsPRwvETBlpxiYqjt5jGhTn9p2v2rgt1Ga6VDMOHwOYIj17y72wyvU6zXvISW8qyNv3HFQ4b0bV2YO1Wsylsny3iKHT8csl+XzzzhXL7zix+FzgPKH961TR8pYJ+Eo+84dejmsAtXQNucsptAinOyAodaPeHSfg51zC70zRVXpoyGq2OESgyvjCOu+mFt/DJLaUgn1RpwbWP+ZmmXRWIyPxdo1sE23AIxWoJiCToIq4SNBgJbex2YwKChCchPEnHkRNnOwluet93K6qOeBMipIke/XuraJ8yT8VE8s7VUXElSQnobG54x4cXE1aORbCwy0+/Uaw2opya4nScXeFxhOrXbLMPfWOy+cJeEjFrUZgIrEHwtJrzpfip2hwwyMqQC0NFfScQmPgTEpeCfLGc440UO0G8JZ9uHVzLBJr1z20dFc+oeVqF0rz7uuHYm18g6XhAtM69OmQGqV3Qe9ZNd58nF2nCXhU0HsRAMEzctkApTfvKcyR21nPSPU8Fr+RgUB5W57nua4MFTCY4LjBwE7hljl1mdhpna/uV0I6WVjsmZPkwCA5bMtSfoJ+8B7pixpbKVRiK0fZ0/41uVz5KuXqv2xIfDeF4vZ0/QxXjQm61J+naQF38d3Z3yk192UDbBd+6wFrn129oRPY8BRsfj756ovH9Q/J/2xMlaQ8sDktm7w9hyKdSp6w4shIw/ZWumdIzQZdinVGHrYuvPtfnO5PYcLSrVfNM2J73nIat+5VspL6KGfbSJGyNrCWRK+Vbm67iraR3uWtzRxcF6NpoWIQaV9FTRDTrq7L5xomvJxut2ijIUyXI1Wn+Ahj5K9Sm/ab9RDztP/oiBj+8DdwJUw4Pkt64TdXobltpFYewhuEKv15WdOY97oR6G1KynYcxlP9wAJT0TPJ6KPE9GniOhzRPQT8vzLiOhIRLeI6P1E9Fx5/nny9y15/aVBOellXb/oXt9x1m4tTRtzblfJDotEzLpTHms5Vr0uVsnhRKXLm+epXpczQHeh5HLrPh+PqBfDzIOHjO4F8vtzABwB7AB8AMBb5Pl3Avg++f37AbxTfn8LgPcHpMHWw/xnu+a6b9LB0+Jm8KERjzacxsR0ZjpErlitWdG9JjPurTQJD1lDpj0TMBBHnu9h+nt0cC2m/AH8RQC/DTGf6/cB3C7PvwbAh+X3DwN4jfx+uwxHowgPyJfRfemM03NJD1cjE3zoBzuJV9fiDCrF6TOL7A01WGyEW4z0EwnJxrF6OTvKPnXeXMUZ1IcnotsA3ATwTQDeDuCLAJ5h5mdlkCcA3CO/3wPgcYhUnyWirwF4MQTxR6Pfl5+zxyvSYUxRYk8dbBhq8kb0Qg8M3ddI72wr2HS87OqGbR+ylxp1srMISJbd2OcOccbRz7R0v0UmvNQ23EFWemb+M2b+VgD3Ang1gG+emjARPUhEN4joRtgNy40Dt+U+LkGGmCHHJOSKaMLFCx3n0Sfu58mbZJ5kFGDo1XlMuPbKi+0yJsPExx4apgOWbcAAlQ9uW5mFCjVqWI6ZnwHwMQgV/g4iUhrCvQCelN+fBHAfAMjrLwTwB5a43sXM15n5enAGbMM0M72oKfultqYWAhOBpFQfFaPxfKFjrcFQdiGI6b7qUBC+AqcFzKBVhujmetcifmjtZQm0Dl/LFmSIlf4lRHSH/P51AF4P4AsQxH+zDPYAgA/K7w/L35DXP8qyo+7CtYgMmy0xA7NVPO1SvA0AACAASURBVDUEOKUCJHmdclhCkX36454+0YXa445EA2WS3rYA5RpSPoUE9kl5YAVvwoXhXcSSiL4FwHsA3AbRQHyAmf81Eb0cwPsA3AngEwD+ETP/MRE9H8AvAHglgD+EsOR/aTiN6wy+aa9Flg7jspNG1+jYpYV4ApZF6Vc1GNxR83dXLDWVbpxLE0Rkfdq7cFgmNgd2LOmTyaq11xm4YSeyzUJkq7Rz8bLwGsIsyB66W60ZHlBktxONtcFphozb8kZhDdZgHPJvoa8zGC7CZ+Vp1/esGqxIRt2dTQbPah2fFyLrosfv3RdP39OG312x7Ku7n5sW7PMCbTd+yntQFntXPK02lPh9m4JrxT2t8/KlV0scq96za9G03rm5vO86jY/VfJUvlA3CJPtg/qW1/nCpRhhaQ6OlG99NZ8HKO3Y5624cw9fmJPvai2fmodJfJ8ZN0xrXWoFP+E7dcOabmbe3zau2zGOgpHyIeg5ZD06vM+SyPuCLvdWHgFfoy8/phyFTkX8TpqKIv0A9cqn0WUj4azcBwxG962BD7THUOM7vDLLCUJSEcJ2SH70jpr0+XMqJR+a98jdLK72mEQOHBsDhEs1OrMft2iBo6XkAvq2rEqaUPDpecMzdmoUsJDwRBzd+fWYXJnWjweLD1Q9Xfe1TycudorLdryX+iaYkltVudg32h71cMN9VzrxAY+tAgr5c+WMwdmQt4YG20H2Fr/Yha2/Mo289Tx6EKBsyuukdbU3XfemZLYhMVmcdNfZOmsjyvkPTITvvrmD1GmqTWW3sOoWgVzaKeaHey+wJeZGNhAfCW1tlSW1PLNWXd8ecPE2PZHfhwhh6Mx1NVF++rxGogqdjcyLR3crTilK9j0RSPjQW/dyBSeq6urAmmvk4vFF1oljfDnOo8kxtagkdTUlP+HiyAwMb1Upmm8NsDIAOvf2yLtwrcbLxmQXZYTMyjowHIe/vxII8Od25UMQ4vEI0eYzZRilfgW6dqSsxh8MnSJfHx+Nc7bVn2Tb76U2zM1R8S5zqY42FLIOwRE+ccbpgZHnIpg+vQdAOFs4itemTqSdX6LhIdr+4c23Wtp7GSXd56zCkVf4SRwCQZNe9+ZPAYsgtfgvopeCtKxFwxaG1xhwLIBJ5Sngy+j1Al01DZZ5igklffedTS3bRAwOkVPk9sLvCDva9rFkKzRLqOGkBMM9b0fpdAWXhQ5aEF0RTot5YikI3tUZjoH6bX0frw10yS9PUiV3IFX0OjUBIY0Qg8A4gG9mV9pKt+m7HVLoHWeuVINJ3lIf8VHoFNfymKp75NpQnnnLG6dsnRlbUlixKjR2xOk0CEFPwdtQmdlesnWBEv9sdh2XYXgyzUb7qex6QFa9Q5Et4GGQ2jfiyvM1x+w7p9WSX0ETar1qyy/Hr1SBViljSs1EXCQwcA/akB6RmswWiT8t86Jh8W8PyGFuPQdaEP6GcQ109EfIh6uiAZrYm1zUiSW9Kd0DsgiNcYgurkSORqqHykp5Ntf5Uucwd+Y3D2xDQbfI54+gwAWPqYzCbySjQAac71529E15OkylduuPEuDo6HgxEMVAXlxgcDEU5jjcWBHsr9YbrOhZ729vQqiwsFyMwZ0ui4h8gvdACTGsjo6EjrnARPGbBPclVKpZ1xLHchDE3pkfRhAegSeUddutb8qk9Tb1wh8tBx7IICKcMmnPqI7cN38l015M0xcPxbh+cFdUv3YKUB083to4hfFfLXJf1xRO+I7x9ZWnGZlXrOe2c6oH0UoONgnCqr1pzia+yPSWpSCy19t2J85VqlFdZ4L6L4gkPIFzK+6NY+32MgL8KameZSamULeVT9eMBX4l3C6nPr77vxtLIfnpsEJS6NqFCrvEiXNmNewzSY+Wnw0HiAkn/gSkoYVNLG9rh2nRdqmHprqYdi+NEZmXaapZFeAN5Fmc4VH8v6jlIVqzDJXBsdCEwxLz41qh3GmuwdwIBSTe7mBlqk1UynRAWgJmarVSFgM2vlpal0qNVye1am2E8K6DSjs6lpa/YV8XbabBtqxDct1U2gPWrhhOtn9VMRlIzLfiSsKv36TYPiUdR02OHoFRyywgbjg2Bjo0k/fJ5i8XoLOrnJ+MUGSRV3R5REQ+XBFwKY2WQAWN+Dk1Cp5++QEb9Sdj673kKn+IIbwOz8CLd7w9AE+ZoskUQk/K3UWc618WOtmGkF4st5iXluy7UC6cdk6SZz8xQnErfh1rIgXdXIq6M567O2dYrgd5uouB21LnYhyqafePgemCLVpMz9PtYSa3fxrCcgVnG0wuFKgsBUgZ9qE0brCvWhmw7JT/XlvL1HcdjG8NyBlRrb1+p5UwgxYhaoNJc81yVD8MxASfMYL+qdtoa3s/4HSdGsYQHUIp2NwuU9D3I7aF2V2xdgokGhtnClDtahfGmCn/Grzk5NmG0syEPj+b5YHJbT55xBgZ2vVVwD5eEI+SkGZ/RXnURFlDtlQU+5/eWl909DmVLeBekIXq5LYlWgCG5fT4nQ96FenNGT1ktodozKGuj6xZQrNHOCi3WOx4om61ArVEt4BkZvSG7U+yu2ON7PM/kmhKNcrlL+c0Z7RS0YLKRXf1ev02bBdKLO4jsHMDSwyX1CvM0xVSVXLnEVqPcsiiS8LpOKtW9o8MP3HNmEEUkPo9NGKVa0rviTLSLLqvxlTLJnqenvB/FGO20KqlWqFG1RIoKvdaljfQUqPYugOG5AGlS0NocCaL7VHlXPLYcmsWuzzlqvrNhKMT1OQSlGYez78ObRD+9aNnwcKCTudZEhn4ueIZtsXTs0s14DC72bUsUmjft4GO9lkN5zwfzqXN7zs1MngHQ8afuP1U7scIyJj2rZEVYcy8nVOQ0BVXvKLvvFk6IYWrI3TWfJ5wHplqfuxFPIfs+vB4OUiT3zJLyDR/NN5bstyOIDIiPOStHzNLWYl28vorNALPcWXZ9DTBnmMVWQkllT3igR/oAphDgZv0CY8mrIjB5PQTX2z6aGcDhEg0d9T7xFcMoqYSKIDwQ7/jR1wr4RP8XUqzfLKsRgNjWWi15tPbLJww3Orsr1ltD9/X3YwNc0gEAcIULlFWV80DuUj57o91oeCwqJ5MWO+HZbhAsBGr2XH+W3MWV26+GIbaQbpqdCFfig6+MHCz2DOA6gBtbmx47iMiS7xj2e156pZJeqSpqYg3DMxQmCyEHLaVk9CvyEmXZT3OI8GVa6R3oEjf8vo713rTwZzR+Hw1Sqj0DIZttVI+3aPgs82uQ3Ydi+vB99B9UjQf7JpJYPUdNTscaCzJG68embBkADg0C58VWDKD1P+iW5VIkH5rrdHPg3iIJrwvbeGJlrArQWv3okX4T9GAIqxwQvo10hROqvtkMpOaox1xp2+SaR9YBKEmlN/SnzrBb//wA+mOmQa1xyWq9hnC/a3YN9nxRLXKJ4FPnc2xTyyF8f3NIMlw6YwgZ2Dh0opzdRW9uEA57YIdqfp8FA3ajpTzwgqt/MVZ626yTrkE9fcGalvuJFvv5J81UrAKPkXjuoTpX/MXPhzcdaTTPjUeapUDNSOUa7WObR+cEoA4mJFCxDjwd5zkb9zGNSTGEN9XrDm9CLBUT0N38YPySzUwBRsWA0YHaHpSJKe9tyCIfW/WLITyArvU8qATZOgTFvcOXpDWCSKRql2p3IEcMa2ZTrPbc+z61wS/HaKcQ7D0ndqTBUfwybyR9aozRj2cbp89tH7KKUKQ3zITMtR+TZFkSPhCK7M2uUSccAVtVfbDp7DvjzLZOXiV7qQjVFEOqTQjZx6IcK70H2hlHrufU7BrsD3vw7srdc+4b0gYkvQ5qhK9+5xUagSMwPkNbKqu+y0oPZg46ANwG4BMAPiR/vwxCYb4F4P0AnivPP0/+viWvvzQg7n63esTBjEPDDQ7c4MA4NMyiA+88WHfy5cF6wqz94MjwMx9rpl2Pae+NLb9Tvk8X12JU+h8C8AXj908C+Glm/iYAXwXwVnn+rQC+Ks//tAy3AEjvINs0O/Bu75W/HY89wG+Ft6j2a260WLWLPDC2Cih2KizyPgOl+70AHgHw7QA+JPP2+wBul9dfA+DD8vuHAbxGfr9dhqP5JbxsIQ9NVEvJgFVyD94XG74eGz+GNcn+wb1jjjxNlfA/A+BfAvhz+fvFAJ5h5mfl7ycA3CO/3wPgcYhUnwXwNRm+AyJ6kIhuENGNwDx4QQCwu4pqKXXYfpeHPTctPaNOvcqKDBFeF9Z+hV7CE9HfAfA0Mw/NuosGM7+Lma8z8/WU8Y5SjAgnDjZiMYjQ+yPCzoS10z9bjGyITb+MJd9dyDj8twH4u0T0RgDPB/CXAPwsgDuI6HYpxe8F8KQM/ySA+wA8QUS3A3ghgD9InvPEUHw3l7ge9H1vb/CHTZVBz2X2B6tIjcBVgkxSU++7ajOWeHdeCc/MP8rM9zLzSwG8BcBHmfkfAvgYgDfLYA8A+KD8/rD8DXn9o5zD2F8ATjR1j1HOFnbNB61kXwNxQ7Nrv6Mpjjc/AuCHiegWRB/95+X5nwfwYnn+hwE8NC2Ly0K0uNRhs4vENit/RUUfqlbEaAFzYTOON8nB6LjeDqrrvbCrN+MVs0Oo4H5FPJTssWG9cZU8PZbVx4LNgm2WnCt5PukLrI/8WtBtgYxPF1RzEFsz5nx3RRBeu7Qu2Ek+Ga4b6M/nRXWBHPO0GQTWwdh3YIafq5oXQfjVhGc/3aEGJ0MpX5EOHQWz4NdcBOG1ZI2aC58qcaBvmBtU7W3nU+epYlGo/rprW+yU6Ev51D3Z/AmvbGFqzvEK3gr9PrpL0A9mrbK+PEi2iSXRUy1h4secqeRDeFdTRhbD98Kk13wPWepqcSN9bUlSQxuJzW2MVlbjU73lfAgvoXlvNABOTzdgVdLHpD1L9557fysmQw+3SQ+66RPTT4+Q1zVX+5IP4btu7GFOLDQTkdzJpQEDODSY7IGfofvCmmCw2Elr1L3i01Tfp7xvFZ+2+RjzM/q2gCVHnPNb006tAQ/Pbqcq+Pw5Ok1Q+dAHOtALjzy04QgAX4EOl+DduGcQWhCJdfAKthqngCbr4RLYjfNKH7WpiSsznTqs3o9007HUGQK3KxZ3zncbghT+9vlIeIUCtmjuG/FCW+duOOm6Edu0y/4Osag8tr3Nzgli/cJGkP1ij7E1R+8TN7b1Vf1Qk9CS6GY7b7UH6Ib7tDKkNhVm5VqbogVbDDGutyo8oAMxxN6OFyFbOZvpJWgNiypnF1gvX4jdlXyiNR5KvhdWpJ1JWpksDXLTLcG1dmw5idaCl90FuW/ACwl/QvYR8jlBZSqd7KwKEEqor6sSarLP2Oik8sLLSsKPgvHy+WI/aq/E1pASd3N/5Cb4noNcPnt35b9xTCKh+UgW20Iw3vVhvw/XjmbFfFLdkpKGL7kiJHwUGDg0rVo3hezHBqPM/XpkMPRWFmRX6+X7tpbS1t2AnpxYVY3RLq82jPWJEgmD7Li4yoTswJJdiRRSPj8rvRdyaON4Cb3PxESyhwha1/1hRhU5+Ho0NscADDJb4tWX/BoAE+N4SUCjWr+rXNiQBszA8RK4kC+K9pt6vBiYlvtRWlrouvRzHrC6J/QOBh8adNaePzQjV/3kbnzR98tDylNvmEMDnWe1Zr588JOwiFkF13yGQyPLoy3YUWWTycGyarbPtX6ecjkY/rrn5NraZA8hPIO7ZDk0MudjCopbYsjKNKXgEUBOs/IKsrviiXgmg+yHRhZlJz5duMUd6h1NatQ3frBxWK+XSXjWJNEvP1T62QpIyXZdKNMqEwNaW7Cetz3PyXmVtcB0uSVEqyk4iC1CF0UYNt45x5TLGR7yxduvFUd4KRG1VM/55StqqRcgpTW8jVNImPblwiB6OBkGGoTMDk32whqp9crLTfq8CX/tNNMMqbIOSbCcDpPkQZpDBNmlpO6UR0xDweHhVy0/h21j8Xe4dllEHGwc3TqTM+FxzUoSLugFsKowPrLLZxoMY4Rl8OiGT6TBk7pC53Cod6frW0FlxcbROe/gWiaON9cZuIHsneiTQJW3f7itdR31B7dGwQAZY9dMdXtrE4x24ory/Sp1C/B+rWKHk0d24/DiJWwTDHR2thkKyKZ33X5camJsHuDdFY4NjfImTI2lc+BMj6FJLsa2KflElRyRl6fdytsvzwb5TGHONCwkOxNoirSRafLuCgTCbs84NNSKgoXBLDwjaSGNcliPYugFLqSnXOlE15IdwLWBcHkRHsDq+zWlhhDr4stArVIdMjCw2/Moz8E+xOScdnLmbs+G5rAczK4Fp3iwobTkX3cqqiUsU3UfQsjzZKfSbwEdNTLAPVb1JYNcaV0RnPy0VWjS3YVFqrtsu+nYSE1jVN8kJrnhiydz1bcHwrCEz8toZ6JEA55ZlDGz59Q7GCH9GIxj097nmxvepjdn+UqmH4XVkeVMl7mSG7QLaA2rXIPcGLiMdnkQ/joxbp7mo6gX1JcgIeHB7TPGz4KQlnxx48VVRHuhJO8shWsMLywwpc1Fdq016d8F1aUEcBF+9TF4ZsY1wBxObI9SxkNjxtbFg2k3iNi0ZEJ8aAzHpNhykvePSd9bDodGPNyM5W1WEnsYWUyR72VLR+bj8Go5T0teClDtGRFZZJ7UV1fCE4iU6r14cGxGTyseinbuoT+zhriG26KWHtsoXBI+Gyv90A6s6zdJw3BVKO79YIwnO0MM1+2aBrsrFtryaPue+DxehpdtSDgyPueAykOQzY23MdyWGtlY6RXfT14QsXRWWSFTEyAEjeyjE8uVrcc9B0MswXzYA9hdYWx/VOVJ9fsB5xocJ1i7+E2yD8JZkSqAXIx25pp2Bav2LViq3yMMatbophj3VI5aa74an2dIRyczc53yp14c67yCYLJXaGRttEPf6KAnzXQNeCUYX5RBTU12WT/PopBdeWJdxkb4zm8z7LLvYP2yK/dwcS2bPryJof788I3yWAss+sW7K5ZSM171TvcIkp6HSxwuhZedPU+Gu60nYdN9c85iVnFXiZ4e+an0Coyuah+g0utbSqwpDLEYZUNGgzEuno4lf++3gcQMHJgvKnUxV6KnQ96ON6516U3SRzm0oLBaIye2IIyg7ljafnrMbiwsyzlEH+m/qKKK+YxwNoQvS8qzlMgTjXuGdjCmwYgl/NJFW2QbvjKyJvx1Ir7pujhWymdeO5i1qzl2VzxqcQrxmIZ2MLLBECr9hG7E7BCLVNRx9XDkbaW/5rHUGlZ7zsACOvVg6YI6aRVeMW5xskT1lPxgQVfmofzar9lHDurheqeZW+nZcV4b7JkApnELZIy5Zw6wWATieAng4mr0vHcG43gJbX0fu5ciy8/jpSfgDHC6xYI7k148d1REIgvCX7uJwXUvtAVZHcEEloLhKGaKrAXR7HJ3e2MKchC1xGUY5vY8Kg4VE8lhu13T+IMviTFDshVByMK11tl/d8EkvaVuMOSecc0lcJTBVrLktX11ZT3HhL4y64aRRrvdqT4/YXdsxC6sV9J/4GJktlJBPVPl+3xYu/+uPO3YOBDaT7H8ZrTbFOWwqQGDJ+2iwoAuKOEtlyAOtcWWtATofHLtJ2/lcHJtbbIrwrcVc6QBShqeOnvQRRp6RqUbdLTEintp0910lXHP3HBSxWcNa+Z5LkPZRCNjPULee0GEj64QskJ3dmYNJBgbRzYVkWHsBDs+DqXtmHu1DcXXIXzqsmDoos6ijDd+uLiWxTi86XjD6A676zBDETB31k8L7SKbaXXOr9iNFJwbP5VWRtJxr93tOeihmNMPxesyPsN15dYE5+x40yd851pYXZWVlRHKEhfZgbahWbJiqjZrquVdkH2cI056wrctd1GzmzeAIgivcmLm1HZuCnR8IY+9RA01n3zChHdDycGF3Joq2nMvEeFFk2WoaRPjtDXOtQEZhovweYzDA906b2DcaLUdUWTHqbaRGgw5Nk8YPy7PIh41LXeKI04SsPKpkM8zISOqQ2p7X5vcoWgJrG2wY2bhWjujwYxbXkT90/clzgukMU3ZTcfHlWZnWJUnNqzy6lxwPMooNzEvMe9srvqyhcPFtSwk/BJQtgCWgifEmWucFFFlbrnCEGvKNST616MNWIIOenbcRClKssWw5SaoCEQkSCXRwSI6X/lvbVeyRbC2dGcW69LP3Vrb4g+RIMH54qG13jmRNGa9Hj0riTqpTIbzFP5Opkt2XdYjtLA5602ph1ObDhwn/10AnwHwSQA35Lk7AXwEwGPy74vkeQLwNgC3AHwawKtCx+GtFWDOQvFUsND0WWrn9vBsOLyMz6cip3pr055d5HdKnpZ8D753NHc9KfFIQfhv7J37KQAPye8PAfhJ+f2NAH4Ngvg7AMexhFcvdO7KZla4URXIJSFFjCLKEcRSr0fvMjNZokMTnTORjmbZj/2Xw3PkdsxB+EcB3C2/3w3gUfn95wB8jy1cDOF1ZZhQkULvV5Uu5p7BQxJ9ynZOJtHVq5pUFoyEGkKiSqnLatq/JO9sY4eLa6FGOwbw60R0k4gelOfuYuan5PcvA7hLfr8HwOPGvU/Icx0Q0YNEdIOIbvSv6TCBmbNlVhl92BMO3Br0pqSp4mOIoTZzs4fYSBjcGVMfb9wT8R2b1qmHRw7/DUQ/+r5UQ2vVeBeBQAl/j/z7lwF8CsDrADzTC/NV+fdDAF5rnH8EwPUxEn7MwRbJ0ZcA3AuXRkKIh1Gqcht3hDQ1pTonyBMbPvkJy/f0uUfEkfhfmne4ncPFtaD58Mz8pPz7NBH9CoBXA/gKEd3NzE8R0d0AnpbBnwRwn3H7vfLc7GC4nTQ6fvlGmClrLbDxzeoCTGZId0IsP4+X1HrJ7cfny0z/AvbhtkHYnqWNshNuRNTVaWZFeFV6Ivp6IvoG9R3AdwL4LICHATwggz0A4IPy+8MAvpcEdgC+Zqj+i0KNu2t1nXsTc2i8J5+ouIxjozxiSUx4OQHpRSZdMR0bka+pXnJKfHZTHxFP6E2hEx1UvJiP7FWtD0SAOv9yCDX+UwA+B+DH5PkXQ6jrjwH4DQB3yvME4O0AvggxlDeozqdU6U1VkXGq5jHs56MPkUKE5ZxPVHRWRa+zNPXZ2bDmJypL7/U4Y6L5fub6N/ndbuSYZKWf+0j6oAGVdVpBjhwm47avy0aDkSKvDG7H+hP2133lHJsWA7MTHpX0so7ZuZbVbLl8IaqQaXmPncbKgN6qedoU2DY+sDFDbs8yxoWmzYgHCk5PB18AdSYdwI7ZclksYjmE6dSYlrb4sNWguFwRIJfbBrCfTnZiwfYD9nr75xQI5nHmbfSa9SZnZEt4VZ0i7UJJM3C8BHAJ7FJYzKGeYUJVZICJQXJHWAC42I+wwg9AN0w4XZfgJJVMRWl/VKbCwNr9d1sfni39vcX6ZdwawNp+8ch42Na3gu7Lx8XXz9MyfXX3EW6ws73Puf8tVl8yPYoz2vUryDIvsLvo42gHGEn3jlGOzedqDXgheQKzdWnp1Q716kIrX+9dLvVvmTqT51Gk0Y6Rzu11CAzhfmpi3AaPrP8wkZhnTkrlZrD+JgINxW7madeIDSMm7Rsf8QT+FNTrys9g10k3zx7HInAZ7VaX7i4JD7Sq4BItNU9Mj7mdnNLerwShLT23SswWTQMjpHrsczDs6ZzGE6llrPhvibqT41GkhC8CDL03u00rELywyHJZE6lnGVNGOQBiG6gx+73Lv1G7bDPCLKQqXObSXad/plLeJeHPjvCiAkr1enJNaPdm30kfeFd6tsS6jYH8xQQ2mGrepok8mCMZxijRoC5RANfbXId1dWx5WRqV8F2cFeEV+Q6XNLE/LMTz4VKsKecbGFNFfJIci4+WkI5YAgWqT5o6K39IS6KzUY501zhDxp894bWkBdRsl1FxgMX4fOza7wy7as5StA61PX2KubTqIYINSflAvkdoAf78LIlzlPIuwm9+1VphwVBkJ4xd/x0MOTNu3NrvHVW9e8V/b+BJ02nGhE+lDy4R4uLIXtHFtgkvSTp1CWWGWMXmYs/jGwyVfE+jEsoGI4YbndT59JpJ+mRDmgWTtzY8LbZLeJYklUtEjZ30zjDWf59AGzVeApxyh7ST/QgMSPpUZGegeH/VynmBDfbh5XjX8RK8uxpPUm53YHVZ4COiAtCqw2zps7v6+Pa4xmcnclStvSfSOp+bVD23fnzWffhrCeJgAIcGYphMzRkdIZFUnx9gXFyJySmTugKN+EbGljcEmwrv6uN386ZH7KLy0R4ElpGEx6A2hiydMJm1Qasgm9lykUKnd690fGla/9jDfg/axcXateRPVOAZOKrhvytDvGh928iaNA0w06kzjgF1a4xDjSa5caNPUttKrfR93XVbWvJDJEAWKv11Ir6BEe+i55kGCKIDkXujs0EMYIQPvY5GD9sBGOz3M9oFKzoE41NvPVs68aS3//Iisg+Qozpv4lxU+6xVeoXoeiJdWhWUK+puH0f2wyVAaqbKGLJr456IZ3fFrbHQlXXdvPQqIMG44rrXMMixv9xinscWV+hgXO5kVyggi/Nh7Ykz5uQZxpjJDsbijRETOkRa5r32SS6hcZjTV8OfwT65hnVR+NNX/3xpqrBD4frXWOYxpjxK+Bdfx8o7ipk8w4hXuRgMn7daJzSTVruBEVZ4WWNUHHr66qi16lg+M52cDdE1zLfnCs0Yb5kP1YF1GoVg66p9MdNj2TgQ2poJFrPkjjMM96SxEqWh6UBKMHPqKg6NfIh4DcGMr58PVvIo9PknXHflC4jQVgr7N6pMCjqKkfCAyDHgaIEZ1taZYbrP9uJiubeasuKLDrYrBTdY9PeVRAfiV6/tRMfo7iHXzzfWtI1zcOqi7OfOzwzYsIgvbvLMIOkd6JNe/zb194urcVb4HtmjRgEskalptcqSzwxQT+82RggXRnjpF0t22AXHVlCMSm8ePEIdZYBbhZi7q8bEqKg9dVWrJN+vLQAACwpJREFU8Z6ugy9vDLuRUef7RDWboH6OyKdIK6yc9Psp9N+Y+lXKkb9Kz3C2uIwRQ0vcFex91Ts8TnORCx4tbhndIUSxvPRJIJEvi9vt8CCfLb2TEf6IOxF0H6Nc6a6wVSmfv0o/UDfHVFt9X08v1kQYeNNtGLHIxa5pIj15erFxu2uNr89vJ3c4CSdBMjjEEw8on+waG2S8i/DZuNYOFTphHOmVXc68V8TlHsJjlmQ/XgJHeXJ35SWBJSaR8PESh0u54izIu+tMO2/e1EZIf86KALIbQTeDsQKlRGTlaSeguiH2K/G3t4tEq/NO6z9YdAOO7SKSu6s2jjBwZxKPMvD1V8zgNtFu1qUbHRsXCGifwZt6+xkHofWcS8U/V+Sj0msMSxlva6w7ljal2LXhouin9yffxDrkMLorzmqpHjK+yMazkdA0+rcK7QPde1xDlOHZjuoxbKHfbsPW+vL59+FTwdMP1Y9L7fzzQ9OO0evxdUn2IOMXQy9VLe7tMzMu7603ffcplE1CDSuG2COCkgxsIjbXdzdwLoTPUKUXYP0xcN0GSTTXyyNStjfZV+41eLsr7syBJ61nu3NyuISW7LuJy2Cp+win6eorRteEQSqT0ckpssdYB7ZIdgBBk5C2gGwlfIiWmcTYwu3KNsDplFafTb3jQJNYTLjStukvQ/m0XTPJHpLlraryJrYk5YtW6V1qZ8xgVZjqqpr54TevVGu1vr3N6q8IMk8lCh+ZHyL7VifGjMU5ED4Plf7asDol1K3T/LfWaz/6xjr7PQQmoeqbVn0TzOi46zrJLm+epwIZIw/ekF2MIfu54BzU+jwk/HVi3AirgynUeFMzcKm2yoZg9nG7U2odhjlDtA97DrLxu0176FoaxOhF5yPdFbYi5fNW6SXhgeEC9/U7wxuDrg28S7LTGGwlFDLS5kq5s6LNyQ2MzrtKWQEdQ5be4GeEubcmXwr5e9pJDPV7xbUBC7z86++piwWmVJhW3VfWbpt13BWXQYohcnbG1ckuaFUY6kr4EAzJbd3IRAwVniPZge673CLy6MP34OxLUZfUrvro69f3+/MaActad9JlQ2D6bu1f75DazK+MTO4kEWqQtCWhruo++1Zr8QzYaluXJeEBvwGFwIMSyJTXcYkapDDyYPbpVd58HHKmrxnKOk6dafMBPPw043eRXbjpxvULzlW6nwOyJTzQkt5W98xK7ApjWvHD6i+dkK7DE6O/ESTUuRtGNRodtUCq2WNlrzUfbULRNoDK8xZbLIvs+vB9uNZfJ0uYEIk4hVjR7OmFpUijmQthjxufhql4nDu22pfPWsKbGFLxTdVaCzfzXuOIVvNTIsJo5otmMJYJSVSyd7G14iiG8ICb9H1pP2R7CzH6lYRUz1D77W5sqViKIjzgN+b1Lfm2wKaEzJH4jMB8cRqSVrK7sbVyyb4P70JwT3igz3xiUAuNcyaMycNYz7CN1ePZMd3ykgeKJHzMRoqdEENOPW2Q0NiSwUzTG7/5DCPIXg1z8fB1E0tCkYRX8M1GCyH2iT9M73efF6l5EuQOY4iXMRLdVNk3Um8rRqK4PnwfUTOc+sNkgbekJok5amADO76re2NQ++dpsJUyLFrCK5gqVyghTlzYPZ20uSWjmbyZl7EjeVV1r7CheAmvoPv1Y+5Fr4/Wi2QOznDvR9ejb3ycHXfbSvak2EJxbobwgHRljVHxzXsdP0yVmOUnmxcHU3Nf65BxIsHVF+odFelRerFuQqXvw7TiA9PUcSX9W089w++der/7iLUoDsDWF68GuGWxBXfbIAlPRHcQ0S8T0f8goi8Q0WuI6E4i+ggRPSb/vkiGJSJ6GxHdIqJPE9Gr5n2EgXxPUPM78Ti+236HXwxHNbzlhZJfRahK/7MA/iszfzOAvw7gCwAeAvAIM98P4BH5GwDeAOB+eTwI4B1JcxwJrd7Kw+zn5v7ilLpeyZ4XSl77zkt4InohgNcB+HkAYOY/YeZnALwJwHtksPcA+G75/U0A3ssCBwB3ENHdyXM+Ev1+br8BWLshMPNQiV6RGiES/mUAfg/AfyCiTxDRnoi+HsBdzPyUDPNlAHfJ7/cAeNy4/wl5Lkv0GwBXQzAn+iSvRrcyUOIrCiH87QBeBeAdzPxKAP8XrfoOAGL3+cjnJ6IHiegGEd3A78XcuQxOLN6Ww6UdxB5mehVloNR3FUL4JwA8wcxq8+RfhmgAvqJUdfn3aXn9SQD3GfffK891wMzvYubrzHwdLxmb/XXh0g7GHBXlocS+vJfwzPxlAI8T0V+Tp74DwOcBPAzgAXnuAQAflN8fBvC90lq/A/A1Q/WvqNgcSiJ96Dj8PwXwi0T0XABfAvBPIBqLDxDRWwH8LwD/QIb9VQBvBHALwP+TYSsqNgnl1s0oY3g+u40oKipKRG471uS9t1xFReEopT+fh4Qn+iMAj66YhW8E8Ps17Zr2RtL/K8xsNYXn4kv/KDNfXytxIrqxVvo17fNKe+30q0pfUXFGqISvqDgj5EL4d51x+jXt80p71fSzMNpVVFQsg1wkfEVFxQJYnfBE9F1E9KhcMOMh/x2T0/tdIvoMEX2SiG7Ic9bFPBKk9W4iepqIPmucW2zhEEf6P05ET8rn/yQRvdG49qMy/UeJ6G9NSPc+IvoYEX2eiD5HRD8kzy/y7APpL/HszyeijxPRp2TaPyHPv4yIjjKN90uvVRDR8+TvW/L6S6c8uxfMvNoB4DYAXwTwcgDPBfApAK+YOc3fBfCNvXM/BeAh+f0hAD+ZKK3XQUw0+qwvLQh35F+DcNjaATjOlP6PA/gXlrCvkOX/PIgp0V8EcNvIdO8G8Cr5/RsA/I6Mf5FnH0h/iWcnAC+Q358D4Cif6QMA3iLPvxPA98nv3w/gnfL7WwC8f876v7aEfzWAW8z8JWb+EwDvg1hAY2m4FvOYBGb+TQB/GJhW8oVDHOm78CYA72PmP2bm/wkxF+LVI9N9ipl/W37/I4gVku7BQs8+kL4LKZ+dmfn/yJ/PkQcD+HaImabA6bOrMvllAN9BRLN56a5N+DUWy2AAv05EN4noQXnOtZjHHMhh4ZAflKrzu43uyyzpSxX1lRCSbvFn76UPLPDsRHQbEX0SYsr4RyA0hmeY+VlL/Dptef1rAF48Nm0f1ib8GngtM78KYu29HyCi15kXWehWiwxdLJmWgXcA+KsAvhXAUwD+3VwJEdELAPxnAP+Mmf+3eW2JZ7ekv8izM/OfMfO3QqwF8WoA3zxHOmOwNuGDFstICWZ+Uv59GsCvQLwQ12Iec2DSwiFTwcxfkRXyzwFcoVVdk6ZPRM+BINsvMvN/kacXe3Zb+ks9uwKLtR8/BuA1EN0U5cpuxq/TltdfCOAPpqbtwtqE/y0A90sL5nMhjBYPz5UYEX09EX2D+g7gOwF8Fu7FPObAqguH9PrGfw/i+VX6b5FW45dBrDr88ZFpEMSip19g5n9vXFrk2V3pL/TsLyGiO+T3rwPweggbwscAvFkG6z+7KpM3A/io1H7mwZwWwUCr5hshrKhfBPBjM6f1cghr7KcAfE6lB9FnegTAYwB+A8CdidL7JQjV8U8h+m1vdaUFYd19uyyHzwC4PlP6vyDj/zREZbvbCP9jMv1HAbxhQrqvhVDXPw3gk/J441LPPpD+Es/+LQA+IdP4LIB/ZdS9j0MYBP8TgOfJ88+Xv2/J6y+fkwPV066i4oywtkpfUVGxICrhKyrOCJXwFRVnhEr4ioozQiV8RcUZoRK+ouKMUAlfUXFGqISvqDgj/H/SXNJXl3xBTQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ground_truth = spectral.imshow(classes = y,figsize =(7,7))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7c_v4cuGa7o3"
   },
   "source": [
    "##### Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 432
    },
    "id": "jKAdGZCMa80E",
    "outputId": "8346702b-717e-41a1-ee6b-f52f31cf24ab"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "predict_image = spectral.imshow(classes = outputs.astype(int),figsize =(7,7))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "machine_shape": "hm",
   "name": "[FINAL] PaviaSeed2.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}