lecture_1/intro_lecture_supplementary.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import pickle\n",
    "from copy import copy, deepcopy\n",
    "\n",
    "from time import time\n",
    "import math"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from pandas import DataFrame\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.autograd as autograd\n",
    "import torch.optim as optim\n",
    "import torch.distributions as distr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "USE_CUDA = torch.cuda.is_available()\n",
    "USE_CUDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "torch.cuda.set_device(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device('cuda' if USE_CUDA else 'cpu')\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from torchvision.datasets import MNIST, CIFAR10\n",
    "from torchvision import transforms\n",
    "from torch.utils.data import TensorDataset, DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "batch_size = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dataset_name = 'cifar10'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "kwargs = {'num_workers': 0, 'pin_memory': True} if USE_CUDA else {}\n",
    "\n",
    "if dataset_name == 'mnist':\n",
    "    dataset = MNIST\n",
    "elif dataset_name == 'cifar10':\n",
    "    dataset = CIFAR10\n",
    "else:\n",
    "    raise NotImplementedError\n",
    "\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    dataset(\n",
    "        './data', train=True, download=True, transform=transforms.ToTensor()\n",
    "    ),\n",
    "    batch_size=batch_size, shuffle=True, **kwargs\n",
    ")\n",
    "\n",
    "test_loader = torch.utils.data.DataLoader(\n",
    "    dataset(\n",
    "        './data', train=False, transform=transforms.ToTensor()\n",
    "    ),\n",
    "    batch_size=batch_size, shuffle=True, **kwargs\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/golikov/anaconda3/lib/python3.6/site-packages/sklearn/model_selection/_split.py:2026: FutureWarning: From version 0.21, test_size will always complement train_size unless both are specified.\n",
      "  FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "train_dataset_train, train_dataset_valid, train_labels_train, train_labels_valid = train_test_split(\n",
    "    np.array(train_loader.dataset.train_data), np.array(train_loader.dataset.train_labels), \n",
    "    stratify=np.array(train_loader.dataset.train_labels), \n",
    "    train_size=100 if dataset_name == 'mnist' else 4000 if dataset_name == 'cifar10' else None\n",
    ")\n",
    "\n",
    "train_loader_train = DataLoader(\n",
    "    TensorDataset(\n",
    "        torch.from_numpy(\n",
    "            train_dataset_train if dataset_name == 'mnist' else \\\n",
    "            train_dataset_train.transpose(0, 3, 1, 2) if dataset_name == 'cifar10' else None\n",
    "        ).float() / 255, \n",
    "        torch.Tensor(train_labels_train).long()\n",
    "    ),\n",
    "    batch_size=batch_size, shuffle=True, **kwargs\n",
    ")\n",
    "train_loader_valid = DataLoader(\n",
    "    TensorDataset(\n",
    "        torch.from_numpy(\n",
    "            train_dataset_valid if dataset_name == 'mnist' else \\\n",
    "            train_dataset_valid.transpose(0, 3, 1, 2) if dataset_name == 'cifar10' else None\n",
    "        ).float() / 255, \n",
    "        torch.Tensor(train_labels_valid).long()\n",
    "    ),\n",
    "    batch_size=batch_size, shuffle=True, **kwargs\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([3, 32, 32])"
      ]
     },
     "execution_count": 136,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_loader_train.dataset[0][0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4000, 46000, 10000)"
      ]
     },
     "execution_count": 137,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_size = train_loader_train.dataset.tensors[0].shape[0]\n",
    "valid_size = train_loader_valid.dataset.tensors[0].shape[0]\n",
    "test_size = test_loader.dataset.test_data.shape[0]\n",
    "train_size, valid_size, test_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "if dataset_name == 'mnist':\n",
    "    input_shape = [1, 28, 28]\n",
    "elif dataset_name == 'cifar10':\n",
    "    input_shape = [3, 32, 32]\n",
    "else:\n",
    "    raise NotImplementedError\n",
    "output_size = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class ReshapeLayer(nn.Module):\n",
    "    def __init__(self, *args):\n",
    "        super(ReshapeLayer, self).__init__()\n",
    "        self.shape = args\n",
    "\n",
    "    def forward(self, x):\n",
    "        return x.view(x.shape[0], *self.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class DeepConv(nn.Module):\n",
    "    def __init__(self, input_shape, output_size, init_num_filters=32):\n",
    "        super(type(self), self).__init__()\n",
    "        self.layers = nn.Sequential(\n",
    "            ReshapeLayer(*input_shape),\n",
    "            nn.Conv2d(input_shape[0], init_num_filters, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters, init_num_filters, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters, init_num_filters, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.MaxPool2d(2),\n",
    "            \n",
    "            nn.Dropout(p=0.5),\n",
    "            nn.Conv2d(init_num_filters, init_num_filters * 2, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters * 2, init_num_filters * 2, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters * 2, init_num_filters * 2, 3, padding=1),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.MaxPool2d(2),\n",
    "            \n",
    "            nn.Dropout(p=0.5),\n",
    "            nn.Conv2d(init_num_filters * 2, init_num_filters * 2, 3),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters * 2, init_num_filters * 2, 1),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.Conv2d(init_num_filters * 2, init_num_filters * 2, 1),\n",
    "            nn.BatchNorm2d(init_num_filters * 2),\n",
    "            nn.LeakyReLU(negative_slope=0.2),\n",
    "            nn.AvgPool2d((input_shape[1] // 2 // 2 - 2, input_shape[2] // 2 // 2 - 2)),\n",
    "            ReshapeLayer(-1),\n",
    "            nn.Linear(init_num_filters * 2, output_size)\n",
    "        )\n",
    "        \n",
    "    def forward(self, X):\n",
    "        return self.layers(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4\n",
      "2810\n",
      "epoch 1\n",
      "loss: 2.33    acc:10.07\n",
      "loss: 2.35    acc:11.40\n",
      "epoch 2\n",
      "loss: 2.33    acc:10.77\n",
      "loss: 2.34    acc:11.30\n",
      "epoch 3\n",
      "loss: 2.32    acc:11.97\n",
      "loss: 2.32    acc:12.90\n",
      "epoch 4\n",
      "loss: 2.30    acc:13.52\n",
      "loss: 2.32    acc:11.70\n",
      "epoch 5\n",
      "loss: 2.29    acc:14.15\n",
      "loss: 2.28    acc:13.40\n",
      "epoch 6\n",
      "loss: 2.28    acc:15.25\n",
      "loss: 2.29    acc:13.50\n",
      "epoch 7\n",
      "loss: 2.26    acc:15.90\n",
      "loss: 2.25    acc:12.90\n",
      "epoch 8\n",
      "loss: 2.24    acc:16.25\n",
      "loss: 2.25    acc:13.60\n",
      "epoch 9\n",
      "loss: 2.22    acc:16.45\n",
      "loss: 2.21    acc:17.70\n",
      "epoch 10\n",
      "loss: 2.21    acc:16.40\n",
      "loss: 2.22    acc:16.20\n",
      "epoch 11\n",
      "loss: 2.19    acc:16.65\n",
      "loss: 2.25    acc:13.20\n",
      "epoch 12\n",
      "loss: 2.17    acc:16.72\n",
      "loss: 2.22    acc:12.30\n",
      "epoch 13\n",
      "loss: 2.16    acc:17.27\n",
      "loss: 2.19    acc:15.60\n",
      "epoch 14\n",
      "loss: 2.13    acc:18.75\n",
      "loss: 2.18    acc:15.30\n",
      "epoch 15\n",
      "loss: 2.12    acc:19.02\n",
      "loss: 2.16    acc:18.30\n",
      "epoch 16\n",
      "loss: 2.10    acc:20.35\n",
      "loss: 2.16    acc:17.80\n",
      "epoch 17\n",
      "loss: 2.08    acc:20.35\n",
      "loss: 2.12    acc:18.20\n",
      "epoch 18\n",
      "loss: 2.07    acc:21.42\n",
      "loss: 2.15    acc:16.30\n",
      "epoch 19\n",
      "loss: 2.05    acc:21.82\n",
      "loss: 2.13    acc:17.80\n",
      "epoch 20\n",
      "loss: 2.03    acc:22.67\n",
      "loss: 2.11    acc:19.90\n",
      "epoch 21\n",
      "loss: 2.02    acc:22.35\n",
      "loss: 2.13    acc:18.40\n",
      "epoch 22\n",
      "loss: 2.00    acc:23.17\n",
      "loss: 2.06    acc:18.60\n",
      "epoch 23\n",
      "loss: 1.99    acc:23.25\n",
      "loss: 2.07    acc:17.70\n",
      "epoch 24\n",
      "loss: 1.97    acc:23.67\n",
      "loss: 2.09    acc:20.50\n",
      "epoch 25\n",
      "loss: 1.96    acc:23.82\n",
      "loss: 2.10    acc:17.80\n",
      "epoch 26\n",
      "loss: 1.95    acc:27.05\n",
      "loss: 2.07    acc:21.50\n",
      "epoch 27\n",
      "loss: 1.94    acc:26.87\n",
      "loss: 2.03    acc:21.30\n",
      "epoch 28\n",
      "loss: 1.92    acc:28.22\n",
      "loss: 2.04    acc:20.20\n",
      "epoch 29\n",
      "loss: 1.91    acc:28.77\n",
      "loss: 2.10    acc:18.10\n",
      "epoch 30\n",
      "loss: 1.90    acc:29.12\n",
      "loss: 2.06    acc:19.80\n",
      "epoch 31\n",
      "loss: 1.89    acc:29.05\n",
      "loss: 2.04    acc:19.70\n",
      "epoch 32\n",
      "loss: 1.87    acc:30.10\n",
      "loss: 2.09    acc:19.80\n",
      "epoch 33\n",
      "loss: 1.87    acc:30.77\n",
      "loss: 2.02    acc:22.50\n",
      "epoch 34\n",
      "loss: 1.85    acc:29.62\n",
      "loss: 2.06    acc:22.60\n",
      "epoch 35\n",
      "loss: 1.84    acc:30.70\n",
      "loss: 2.01    acc:23.00\n",
      "epoch 36\n",
      "loss: 1.84    acc:30.62\n",
      "loss: 1.96    acc:24.20\n",
      "epoch 37\n",
      "loss: 1.82    acc:30.75\n",
      "loss: 2.08    acc:20.20\n",
      "epoch 38\n",
      "loss: 1.82    acc:31.65\n",
      "loss: 2.07    acc:20.40\n",
      "epoch 39\n",
      "loss: 1.82    acc:30.30\n",
      "loss: 1.99    acc:23.30\n",
      "epoch 40\n",
      "loss: 1.80    acc:31.75\n",
      "loss: 2.00    acc:23.10\n",
      "epoch 41\n",
      "loss: 1.80    acc:31.52\n",
      "loss: 2.02    acc:22.30\n",
      "epoch 42\n",
      "loss: 1.79    acc:31.55\n",
      "loss: 1.97    acc:21.80\n",
      "epoch 43\n",
      "loss: 1.78    acc:32.82\n",
      "loss: 1.99    acc:22.30\n",
      "epoch 44\n",
      "loss: 1.78    acc:31.32\n",
      "loss: 2.09    acc:21.20\n",
      "epoch 45\n",
      "loss: 1.77    acc:33.12\n",
      "loss: 2.15    acc:18.30\n",
      "epoch 46\n",
      "loss: 1.76    acc:32.45\n",
      "loss: 2.04    acc:22.00\n",
      "epoch 47\n",
      "loss: 1.76    acc:32.72\n",
      "loss: 1.98    acc:24.90\n",
      "epoch 48\n",
      "loss: 1.75    acc:33.50\n",
      "loss: 2.00    acc:22.30\n",
      "epoch 49\n",
      "loss: 1.75    acc:33.35\n",
      "loss: 1.99    acc:22.60\n",
      "epoch 50\n",
      "loss: 1.75    acc:33.05\n",
      "loss: 1.90    acc:26.60\n",
      "epoch 51\n",
      "loss: 1.74    acc:33.60\n",
      "loss: 1.90    acc:25.50\n",
      "epoch 52\n",
      "loss: 1.74    acc:33.10\n",
      "loss: 1.96    acc:24.00\n",
      "epoch 53\n",
      "loss: 1.74    acc:32.80\n",
      "loss: 2.03    acc:23.70\n",
      "epoch 54\n",
      "loss: 1.72    acc:33.42\n",
      "loss: 1.97    acc:26.00\n",
      "epoch 55\n",
      "loss: 1.72    acc:34.27\n",
      "loss: 1.95    acc:25.10\n",
      "epoch 56\n",
      "loss: 1.72    acc:34.30\n",
      "loss: 1.90    acc:27.00\n",
      "epoch 57\n",
      "loss: 1.73    acc:33.35\n",
      "loss: 1.91    acc:24.70\n",
      "epoch 58\n",
      "loss: 1.71    acc:34.50\n",
      "loss: 1.92    acc:27.00\n",
      "epoch 59\n",
      "loss: 1.71    acc:34.75\n",
      "loss: 2.01    acc:22.50\n",
      "epoch 60\n",
      "loss: 1.70    acc:34.40\n",
      "loss: 1.94    acc:26.80\n",
      "epoch 61\n",
      "loss: 1.70    acc:34.12\n",
      "loss: 1.94    acc:27.30\n",
      "epoch 62\n",
      "loss: 1.70    acc:34.05\n",
      "loss: 1.97    acc:24.20\n",
      "epoch 63\n",
      "loss: 1.70    acc:34.85\n",
      "loss: 1.99    acc:24.40\n",
      "epoch 64\n",
      "loss: 1.70    acc:33.97\n",
      "loss: 1.93    acc:27.00\n",
      "epoch 65\n",
      "loss: 1.69    acc:34.82\n",
      "loss: 1.96    acc:25.90\n",
      "epoch 66\n",
      "loss: 1.68    acc:35.82\n",
      "loss: 1.99    acc:24.20\n",
      "epoch 67\n",
      "loss: 1.69    acc:34.75\n",
      "loss: 2.00    acc:26.10\n",
      "epoch 68\n",
      "loss: 1.69    acc:35.22\n",
      "loss: 1.84    acc:28.70\n",
      "epoch 69\n",
      "loss: 1.67    acc:35.47\n",
      "loss: 1.98    acc:25.30\n",
      "epoch 70\n",
      "loss: 1.68    acc:35.17\n",
      "loss: 1.99    acc:25.50\n",
      "epoch 71\n",
      "loss: 1.68    acc:36.15\n",
      "loss: 1.93    acc:27.50\n",
      "epoch 72\n",
      "loss: 1.67    acc:35.70\n",
      "loss: 1.98    acc:26.30\n",
      "epoch 73\n",
      "loss: 1.68    acc:36.02\n",
      "loss: 1.98    acc:25.20\n",
      "epoch 74\n",
      "loss: 1.68    acc:35.27\n",
      "loss: 1.88    acc:28.70\n",
      "epoch 75\n",
      "loss: 1.66    acc:35.85\n",
      "loss: 1.94    acc:26.40\n",
      "epoch 76\n",
      "loss: 1.66    acc:37.00\n",
      "loss: 1.90    acc:26.10\n",
      "epoch 77\n",
      "loss: 1.66    acc:36.97\n",
      "loss: 1.85    acc:26.90\n",
      "epoch 78\n",
      "loss: 1.67    acc:36.20\n",
      "loss: 1.87    acc:29.60\n",
      "epoch 79\n",
      "loss: 1.67    acc:35.52\n",
      "loss: 1.94    acc:27.10\n",
      "epoch 80\n",
      "loss: 1.66    acc:36.87\n",
      "loss: 1.94    acc:25.00\n",
      "epoch 81\n",
      "loss: 1.65    acc:37.15\n",
      "loss: 1.90    acc:28.00\n",
      "epoch 82\n",
      "loss: 1.66    acc:35.42\n",
      "loss: 1.85    acc:30.30\n",
      "epoch 83\n",
      "loss: 1.65    acc:37.15\n",
      "loss: 1.79    acc:31.00\n",
      "epoch 84\n",
      "loss: 1.64    acc:37.65\n",
      "loss: 1.86    acc:26.20\n",
      "epoch 85\n",
      "loss: 1.65    acc:36.85\n",
      "loss: 1.92    acc:28.40\n",
      "epoch 86\n",
      "loss: 1.65    acc:36.17\n",
      "loss: 2.05    acc:25.80\n",
      "epoch 87\n",
      "loss: 1.64    acc:36.92\n",
      "loss: 1.86    acc:30.10\n",
      "epoch 88\n",
      "loss: 1.64    acc:36.67\n",
      "loss: 1.90    acc:29.90\n",
      "epoch 89\n",
      "loss: 1.65    acc:37.07\n",
      "loss: 1.96    acc:25.30\n",
      "epoch 90\n",
      "loss: 1.64    acc:37.70\n",
      "loss: 1.86    acc:29.70\n",
      "epoch 91\n",
      "loss: 1.63    acc:37.32\n",
      "loss: 1.93    acc:27.80\n",
      "epoch 92\n",
      "loss: 1.63    acc:36.67\n",
      "loss: 1.98    acc:25.80\n",
      "epoch 93\n",
      "loss: 1.63    acc:37.15\n",
      "loss: 1.80    acc:32.30\n",
      "epoch 94\n",
      "loss: 1.63    acc:37.40\n",
      "loss: 1.95    acc:26.70\n",
      "epoch 95\n",
      "loss: 1.63    acc:37.52\n",
      "loss: 2.01    acc:26.20\n",
      "epoch 96\n",
      "loss: 1.63    acc:37.80\n",
      "loss: 1.84    acc:31.50\n",
      "epoch 97\n",
      "loss: 1.62    acc:37.97\n",
      "loss: 2.02    acc:25.10\n",
      "epoch 98\n",
      "loss: 1.62    acc:37.35\n",
      "loss: 1.93    acc:28.30\n",
      "epoch 99\n",
      "loss: 1.63    acc:37.17\n",
      "loss: 1.83    acc:30.90\n",
      "epoch 100\n",
      "loss: 1.63    acc:38.25\n",
      "loss: 1.87    acc:31.80\n",
      "epoch 101\n",
      "loss: 1.63    acc:37.47\n",
      "loss: 1.87    acc:29.10\n",
      "epoch 102\n",
      "loss: 1.61    acc:38.67\n",
      "loss: 2.02    acc:25.70\n",
      "epoch 103\n",
      "loss: 1.62    acc:38.67\n",
      "loss: 1.82    acc:32.60\n",
      "epoch 104\n",
      "loss: 1.61    acc:37.97\n",
      "loss: 1.89    acc:28.90\n",
      "epoch 105\n",
      "loss: 1.61    acc:38.57\n",
      "loss: 1.83    acc:28.70\n",
      "epoch 106\n",
      "loss: 1.62    acc:38.02\n",
      "loss: 2.02    acc:25.10\n",
      "epoch 107\n",
      "loss: 1.61    acc:39.60\n",
      "loss: 1.83    acc:29.60\n",
      "epoch 108\n",
      "loss: 1.61    acc:38.17\n",
      "loss: 1.95    acc:28.80\n",
      "epoch 109\n",
      "loss: 1.61    acc:38.95\n",
      "loss: 1.84    acc:30.70\n",
      "epoch 110\n",
      "loss: 1.62    acc:38.55\n",
      "loss: 1.83    acc:32.10\n",
      "epoch 111\n",
      "loss: 1.60    acc:39.05\n",
      "loss: 1.83    acc:30.10\n",
      "epoch 112\n",
      "loss: 1.60    acc:39.47\n",
      "loss: 1.84    acc:30.80\n",
      "epoch 113\n",
      "loss: 1.60    acc:39.15\n",
      "loss: 1.84    acc:31.50\n",
      "epoch 114\n",
      "loss: 1.59    acc:39.75\n",
      "loss: 1.83    acc:31.10\n",
      "epoch 115\n",
      "loss: 1.59    acc:39.62\n",
      "loss: 1.96    acc:25.60\n",
      "epoch 116\n",
      "loss: 1.59    acc:38.65\n",
      "loss: 1.76    acc:31.60\n",
      "epoch 117\n",
      "loss: 1.59    acc:39.10\n",
      "loss: 1.77    acc:33.00\n",
      "epoch 118\n",
      "loss: 1.60    acc:38.87\n",
      "loss: 1.84    acc:30.20\n",
      "epoch 119\n",
      "loss: 1.59    acc:40.67\n",
      "loss: 1.83    acc:30.00\n",
      "epoch 120\n",
      "loss: 1.58    acc:40.62\n",
      "loss: 1.77    acc:32.80\n",
      "epoch 121\n",
      "loss: 1.58    acc:40.60\n",
      "loss: 1.87    acc:27.20\n",
      "epoch 122\n",
      "loss: 1.58    acc:39.15\n",
      "loss: 1.87    acc:28.70\n",
      "epoch 123\n",
      "loss: 1.59    acc:39.45\n",
      "loss: 1.81    acc:31.90\n",
      "epoch 124\n",
      "loss: 1.57    acc:39.32\n",
      "loss: 1.80    acc:32.60\n",
      "epoch 125\n",
      "loss: 1.58    acc:40.05\n",
      "loss: 1.85    acc:30.30\n",
      "epoch 126\n",
      "loss: 1.58    acc:39.82\n",
      "loss: 1.82    acc:30.70\n",
      "epoch 127\n",
      "loss: 1.58    acc:39.45\n",
      "loss: 1.90    acc:28.00\n",
      "epoch 128\n",
      "loss: 1.57    acc:40.12\n",
      "loss: 1.80    acc:32.30\n",
      "epoch 129\n",
      "loss: 1.56    acc:40.22\n",
      "loss: 1.75    acc:34.30\n",
      "epoch 130\n",
      "loss: 1.57    acc:40.12\n",
      "loss: 1.71    acc:33.70\n",
      "epoch 131\n",
      "loss: 1.57    acc:40.40\n",
      "loss: 1.72    acc:36.00\n",
      "epoch 132\n",
      "loss: 1.57    acc:40.45\n",
      "loss: 1.78    acc:31.60\n",
      "epoch 133\n",
      "loss: 1.56    acc:39.72\n",
      "loss: 1.76    acc:31.80\n",
      "epoch 134\n",
      "loss: 1.55    acc:42.20\n",
      "loss: 1.75    acc:32.00\n",
      "epoch 135\n",
      "loss: 1.56    acc:40.45\n",
      "loss: 1.83    acc:32.00\n",
      "epoch 136\n",
      "loss: 1.57    acc:40.70\n",
      "loss: 1.71    acc:35.10\n",
      "epoch 137\n",
      "loss: 1.56    acc:40.97\n",
      "loss: 1.71    acc:31.80\n",
      "epoch 138\n",
      "loss: 1.55    acc:41.22\n",
      "loss: 1.77    acc:33.30\n",
      "epoch 139\n",
      "loss: 1.54    acc:41.42\n",
      "loss: 1.86    acc:31.70\n",
      "epoch 140\n",
      "loss: 1.55    acc:39.97\n",
      "loss: 1.80    acc:32.00\n",
      "epoch 141\n",
      "loss: 1.55    acc:41.22\n",
      "loss: 1.78    acc:31.80\n",
      "epoch 142\n",
      "loss: 1.54    acc:41.27\n",
      "loss: 1.69    acc:36.30\n",
      "epoch 143\n",
      "loss: 1.54    acc:41.37\n",
      "loss: 1.75    acc:33.60\n",
      "epoch 144\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.54    acc:42.32\n",
      "loss: 1.71    acc:34.40\n",
      "epoch 145\n",
      "loss: 1.55    acc:41.07\n",
      "loss: 1.83    acc:34.20\n",
      "epoch 146\n",
      "loss: 1.54    acc:42.50\n",
      "loss: 1.73    acc:33.50\n",
      "epoch 147\n",
      "loss: 1.55    acc:41.05\n",
      "loss: 1.73    acc:34.50\n",
      "epoch 148\n",
      "loss: 1.54    acc:41.50\n",
      "loss: 1.76    acc:33.40\n",
      "epoch 149\n",
      "loss: 1.53    acc:41.70\n",
      "loss: 1.82    acc:31.00\n",
      "epoch 150\n",
      "loss: 1.53    acc:42.35\n",
      "loss: 1.81    acc:33.20\n",
      "epoch 151\n",
      "loss: 1.53    acc:42.02\n",
      "loss: 1.79    acc:34.00\n",
      "epoch 152\n",
      "loss: 1.54    acc:41.40\n",
      "loss: 1.75    acc:34.10\n",
      "epoch 153\n",
      "loss: 1.53    acc:41.22\n",
      "loss: 1.71    acc:36.50\n",
      "epoch 154\n",
      "loss: 1.53    acc:41.70\n",
      "loss: 1.70    acc:35.20\n",
      "epoch 155\n",
      "loss: 1.53    acc:41.95\n",
      "loss: 1.78    acc:33.80\n",
      "epoch 156\n",
      "loss: 1.52    acc:42.02\n",
      "loss: 1.68    acc:35.90\n",
      "epoch 157\n",
      "loss: 1.52    acc:43.35\n",
      "loss: 1.73    acc:35.20\n",
      "epoch 158\n",
      "loss: 1.52    acc:43.12\n",
      "loss: 1.76    acc:34.60\n",
      "epoch 159\n",
      "loss: 1.53    acc:42.37\n",
      "loss: 1.82    acc:32.60\n",
      "epoch 160\n",
      "loss: 1.52    acc:43.27\n",
      "loss: 1.88    acc:32.50\n",
      "epoch 161\n",
      "loss: 1.52    acc:42.90\n",
      "loss: 1.76    acc:35.40\n",
      "epoch 162\n",
      "loss: 1.52    acc:42.85\n",
      "loss: 1.77    acc:34.40\n",
      "epoch 163\n",
      "loss: 1.51    acc:42.65\n",
      "loss: 1.76    acc:33.70\n",
      "epoch 164\n",
      "loss: 1.52    acc:42.77\n",
      "loss: 1.78    acc:35.70\n",
      "epoch 165\n",
      "loss: 1.53    acc:42.27\n",
      "loss: 1.76    acc:34.50\n",
      "epoch 166\n",
      "loss: 1.52    acc:41.72\n",
      "loss: 1.73    acc:34.30\n",
      "epoch 167\n",
      "loss: 1.51    acc:43.07\n",
      "loss: 1.84    acc:34.90\n",
      "epoch 168\n",
      "loss: 1.51    acc:42.82\n",
      "loss: 1.80    acc:33.10\n",
      "epoch 169\n",
      "loss: 1.52    acc:42.35\n",
      "loss: 1.76    acc:34.50\n",
      "epoch 170\n",
      "loss: 1.50    acc:42.77\n",
      "loss: 1.67    acc:38.30\n",
      "epoch 171\n",
      "loss: 1.51    acc:42.25\n",
      "loss: 1.77    acc:32.90\n",
      "epoch 172\n",
      "loss: 1.51    acc:43.45\n",
      "loss: 1.81    acc:33.90\n",
      "epoch 173\n",
      "loss: 1.51    acc:42.57\n",
      "loss: 1.74    acc:38.50\n",
      "epoch 174\n",
      "loss: 1.51    acc:43.22\n",
      "loss: 1.75    acc:36.40\n",
      "epoch 175\n",
      "loss: 1.51    acc:43.52\n",
      "loss: 1.77    acc:36.30\n",
      "epoch 176\n",
      "loss: 1.52    acc:42.80\n",
      "loss: 1.73    acc:36.00\n",
      "epoch 177\n",
      "loss: 1.51    acc:43.72\n",
      "loss: 1.69    acc:35.80\n",
      "epoch 178\n",
      "loss: 1.50    acc:44.45\n",
      "loss: 1.78    acc:38.00\n",
      "epoch 179\n",
      "loss: 1.50    acc:42.95\n",
      "loss: 1.69    acc:34.90\n",
      "epoch 180\n",
      "loss: 1.50    acc:42.85\n",
      "loss: 1.80    acc:34.70\n",
      "epoch 181\n",
      "loss: 1.50    acc:42.75\n",
      "loss: 1.75    acc:36.70\n",
      "epoch 182\n",
      "loss: 1.51    acc:42.05\n",
      "loss: 1.75    acc:37.30\n",
      "epoch 183\n",
      "loss: 1.51    acc:42.10\n",
      "loss: 1.76    acc:38.60\n",
      "epoch 184\n",
      "loss: 1.49    acc:43.50\n",
      "loss: 1.74    acc:35.50\n",
      "epoch 185\n",
      "loss: 1.49    acc:43.85\n",
      "loss: 1.75    acc:37.20\n",
      "epoch 186\n",
      "loss: 1.49    acc:43.75\n",
      "loss: 1.78    acc:37.30\n",
      "epoch 187\n",
      "loss: 1.49    acc:43.72\n",
      "loss: 1.77    acc:35.90\n",
      "epoch 188\n",
      "loss: 1.49    acc:43.85\n",
      "loss: 1.91    acc:33.10\n",
      "epoch 189\n",
      "loss: 1.49    acc:44.07\n",
      "loss: 1.71    acc:36.80\n",
      "epoch 190\n",
      "loss: 1.48    acc:45.40\n",
      "loss: 1.70    acc:36.90\n",
      "epoch 191\n",
      "loss: 1.49    acc:43.55\n",
      "loss: 1.81    acc:34.30\n",
      "epoch 192\n",
      "loss: 1.49    acc:44.02\n",
      "loss: 1.76    acc:35.90\n",
      "epoch 193\n",
      "loss: 1.50    acc:43.75\n",
      "loss: 1.69    acc:37.70\n",
      "epoch 194\n",
      "loss: 1.47    acc:44.92\n",
      "loss: 1.77    acc:36.10\n",
      "epoch 195\n",
      "loss: 1.48    acc:44.12\n",
      "loss: 1.64    acc:39.40\n",
      "epoch 196\n",
      "loss: 1.49    acc:43.22\n",
      "loss: 1.67    acc:38.50\n",
      "epoch 197\n",
      "loss: 1.48    acc:44.17\n",
      "loss: 1.78    acc:35.50\n",
      "epoch 198\n",
      "loss: 1.49    acc:43.55\n",
      "loss: 1.69    acc:36.80\n",
      "epoch 199\n",
      "loss: 1.48    acc:44.62\n",
      "loss: 1.74    acc:35.50\n",
      "epoch 200\n",
      "loss: 1.48    acc:42.75\n",
      "loss: 1.64    acc:40.30\n",
      "epoch 201\n",
      "loss: 1.48    acc:44.55\n",
      "loss: 1.74    acc:33.60\n",
      "epoch 202\n",
      "loss: 1.48    acc:44.10\n",
      "loss: 1.68    acc:36.60\n",
      "epoch 203\n",
      "loss: 1.48    acc:44.22\n",
      "loss: 1.72    acc:35.80\n",
      "epoch 204\n",
      "loss: 1.49    acc:44.02\n",
      "loss: 1.72    acc:37.90\n",
      "epoch 205\n",
      "loss: 1.48    acc:43.97\n",
      "loss: 1.71    acc:38.00\n",
      "epoch 206\n",
      "loss: 1.47    acc:44.12\n",
      "loss: 1.69    acc:38.00\n",
      "epoch 207\n",
      "loss: 1.48    acc:44.10\n",
      "loss: 1.73    acc:37.20\n",
      "epoch 208\n",
      "loss: 1.47    acc:45.00\n",
      "loss: 1.80    acc:35.30\n",
      "epoch 209\n",
      "loss: 1.48    acc:44.37\n",
      "loss: 1.65    acc:40.00\n",
      "epoch 210\n",
      "loss: 1.48    acc:44.95\n",
      "loss: 1.70    acc:38.10\n",
      "epoch 211\n",
      "loss: 1.47    acc:44.90\n",
      "loss: 1.72    acc:37.90\n",
      "epoch 212\n",
      "loss: 1.48    acc:42.95\n",
      "loss: 1.82    acc:36.10\n",
      "epoch 213\n",
      "loss: 1.47    acc:45.02\n",
      "loss: 1.66    acc:40.50\n",
      "epoch 214\n",
      "loss: 1.47    acc:44.05\n",
      "loss: 1.65    acc:40.30\n",
      "epoch 215\n",
      "loss: 1.47    acc:44.60\n",
      "loss: 1.82    acc:34.50\n",
      "epoch 216\n",
      "loss: 1.45    acc:44.17\n",
      "loss: 1.71    acc:37.30\n",
      "epoch 217\n",
      "loss: 1.47    acc:45.67\n",
      "loss: 1.77    acc:36.20\n",
      "epoch 218\n",
      "loss: 1.48    acc:44.55\n",
      "loss: 1.75    acc:36.80\n",
      "epoch 219\n",
      "loss: 1.47    acc:44.02\n",
      "loss: 1.71    acc:37.20\n",
      "epoch 220\n",
      "loss: 1.46    acc:45.55\n",
      "loss: 1.70    acc:39.50\n",
      "epoch 221\n",
      "loss: 1.46    acc:45.35\n",
      "loss: 1.73    acc:37.70\n",
      "epoch 222\n",
      "loss: 1.46    acc:44.95\n",
      "loss: 1.56    acc:41.60\n",
      "epoch 223\n",
      "loss: 1.46    acc:45.60\n",
      "loss: 1.67    acc:39.20\n",
      "epoch 224\n",
      "loss: 1.45    acc:45.22\n",
      "loss: 1.66    acc:38.40\n",
      "epoch 225\n",
      "loss: 1.46    acc:44.80\n",
      "loss: 1.67    acc:40.40\n",
      "epoch 226\n",
      "loss: 1.46    acc:44.75\n",
      "loss: 1.68    acc:37.80\n",
      "epoch 227\n",
      "loss: 1.47    acc:44.30\n",
      "loss: 1.68    acc:38.70\n",
      "epoch 228\n",
      "loss: 1.46    acc:44.22\n",
      "loss: 1.76    acc:36.40\n",
      "epoch 229\n",
      "loss: 1.46    acc:45.52\n",
      "loss: 1.62    acc:41.30\n",
      "epoch 230\n",
      "loss: 1.45    acc:45.75\n",
      "loss: 1.73    acc:37.20\n",
      "epoch 231\n",
      "loss: 1.45    acc:45.00\n",
      "loss: 1.75    acc:34.60\n",
      "epoch 232\n",
      "loss: 1.46    acc:44.90\n",
      "loss: 1.67    acc:38.90\n",
      "epoch 233\n",
      "loss: 1.47    acc:45.10\n",
      "loss: 1.63    acc:40.40\n",
      "epoch 234\n",
      "loss: 1.45    acc:44.27\n",
      "loss: 1.66    acc:41.30\n",
      "epoch 235\n",
      "loss: 1.46    acc:46.15\n",
      "loss: 1.73    acc:36.50\n",
      "epoch 236\n",
      "loss: 1.47    acc:44.52\n",
      "loss: 1.80    acc:35.80\n",
      "epoch 237\n",
      "loss: 1.46    acc:45.42\n",
      "loss: 1.65    acc:37.80\n",
      "epoch 238\n",
      "loss: 1.46    acc:45.40\n",
      "loss: 1.69    acc:39.40\n",
      "epoch 239\n",
      "loss: 1.46    acc:45.50\n",
      "loss: 1.55    acc:42.40\n",
      "epoch 240\n",
      "loss: 1.46    acc:45.30\n",
      "loss: 1.60    acc:42.90\n",
      "epoch 241\n",
      "loss: 1.46    acc:45.22\n",
      "loss: 1.57    acc:40.60\n",
      "epoch 242\n",
      "loss: 1.45    acc:46.00\n",
      "loss: 1.65    acc:39.20\n",
      "epoch 243\n",
      "loss: 1.45    acc:45.12\n",
      "loss: 1.78    acc:35.70\n",
      "epoch 244\n",
      "loss: 1.45    acc:45.95\n",
      "loss: 1.64    acc:40.10\n",
      "epoch 245\n",
      "loss: 1.45    acc:44.75\n",
      "loss: 1.62    acc:40.90\n",
      "epoch 246\n",
      "loss: 1.45    acc:44.97\n",
      "loss: 1.60    acc:40.40\n",
      "epoch 247\n",
      "loss: 1.44    acc:45.85\n",
      "loss: 1.66    acc:39.00\n",
      "epoch 248\n",
      "loss: 1.44    acc:46.62\n",
      "loss: 1.59    acc:39.80\n",
      "epoch 249\n",
      "loss: 1.44    acc:44.97\n",
      "loss: 1.70    acc:38.90\n",
      "epoch 250\n",
      "loss: 1.45    acc:45.45\n",
      "loss: 1.70    acc:40.30\n",
      "epoch 251\n",
      "loss: 1.46    acc:45.35\n",
      "loss: 1.67    acc:38.70\n",
      "epoch 252\n",
      "loss: 1.45    acc:46.60\n",
      "loss: 1.63    acc:39.60\n",
      "epoch 253\n",
      "loss: 1.44    acc:45.72\n",
      "loss: 1.69    acc:38.30\n",
      "epoch 254\n",
      "loss: 1.45    acc:46.25\n",
      "loss: 1.66    acc:37.20\n",
      "epoch 255\n",
      "loss: 1.45    acc:45.97\n",
      "loss: 1.58    acc:43.60\n",
      "epoch 256\n",
      "loss: 1.44    acc:45.85\n",
      "loss: 1.70    acc:39.40\n",
      "epoch 257\n",
      "loss: 1.45    acc:45.47\n",
      "loss: 1.55    acc:43.80\n",
      "epoch 258\n",
      "loss: 1.44    acc:45.05\n",
      "loss: 1.66    acc:39.60\n",
      "epoch 259\n",
      "loss: 1.44    acc:46.20\n",
      "loss: 1.61    acc:40.80\n",
      "epoch 260\n",
      "loss: 1.43    acc:46.42\n",
      "loss: 1.66    acc:39.70\n",
      "epoch 261\n",
      "loss: 1.45    acc:45.55\n",
      "loss: 1.78    acc:38.50\n",
      "epoch 262\n",
      "loss: 1.43    acc:45.57\n",
      "loss: 1.64    acc:40.10\n",
      "epoch 263\n",
      "loss: 1.44    acc:45.22\n",
      "loss: 1.60    acc:42.20\n",
      "epoch 264\n",
      "loss: 1.43    acc:46.30\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 265\n",
      "loss: 1.42    acc:46.17\n",
      "loss: 1.57    acc:41.00\n",
      "epoch 266\n",
      "loss: 1.44    acc:45.77\n",
      "loss: 1.63    acc:40.50\n",
      "epoch 267\n",
      "loss: 1.44    acc:46.00\n",
      "loss: 1.54    acc:44.60\n",
      "epoch 268\n",
      "loss: 1.43    acc:45.55\n",
      "loss: 1.59    acc:42.90\n",
      "epoch 269\n",
      "loss: 1.44    acc:46.50\n",
      "loss: 1.59    acc:40.90\n",
      "epoch 270\n",
      "loss: 1.42    acc:46.35\n",
      "loss: 1.63    acc:39.90\n",
      "epoch 271\n",
      "loss: 1.43    acc:45.77\n",
      "loss: 1.60    acc:44.40\n",
      "epoch 272\n",
      "loss: 1.44    acc:45.52\n",
      "loss: 1.60    acc:41.00\n",
      "epoch 273\n",
      "loss: 1.44    acc:46.17\n",
      "loss: 1.62    acc:42.10\n",
      "epoch 274\n",
      "loss: 1.44    acc:46.02\n",
      "loss: 1.68    acc:40.40\n",
      "epoch 275\n",
      "loss: 1.43    acc:46.22\n",
      "loss: 1.59    acc:43.00\n",
      "epoch 276\n",
      "loss: 1.44    acc:45.42\n",
      "loss: 1.51    acc:42.30\n",
      "epoch 277\n",
      "loss: 1.42    acc:46.70\n",
      "loss: 1.58    acc:42.40\n",
      "epoch 278\n",
      "loss: 1.44    acc:45.57\n",
      "loss: 1.71    acc:39.60\n",
      "epoch 279\n",
      "loss: 1.42    acc:46.37\n",
      "loss: 1.61    acc:42.10\n",
      "epoch 280\n",
      "loss: 1.43    acc:46.52\n",
      "loss: 1.61    acc:43.40\n",
      "epoch 281\n",
      "loss: 1.44    acc:46.27\n",
      "loss: 1.55    acc:45.30\n",
      "epoch 282\n",
      "loss: 1.44    acc:46.20\n",
      "loss: 1.69    acc:39.60\n",
      "epoch 283\n",
      "loss: 1.44    acc:45.42\n",
      "loss: 1.58    acc:42.10\n",
      "epoch 284\n",
      "loss: 1.42    acc:46.72\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 285\n",
      "loss: 1.42    acc:46.02\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.52    acc:45.20\n",
      "epoch 286\n",
      "loss: 1.43    acc:45.10\n",
      "loss: 1.76    acc:40.90\n",
      "epoch 287\n",
      "loss: 1.41    acc:47.27\n",
      "loss: 1.55    acc:43.20\n",
      "epoch 288\n",
      "loss: 1.42    acc:45.87\n",
      "loss: 1.64    acc:40.10\n",
      "epoch 289\n",
      "loss: 1.44    acc:45.32\n",
      "loss: 1.59    acc:41.00\n",
      "epoch 290\n",
      "loss: 1.43    acc:46.02\n",
      "loss: 1.64    acc:41.90\n",
      "epoch 291\n",
      "loss: 1.43    acc:45.35\n",
      "loss: 1.50    acc:46.40\n",
      "epoch 292\n",
      "loss: 1.42    acc:47.35\n",
      "loss: 1.64    acc:39.70\n",
      "epoch 293\n",
      "loss: 1.43    acc:46.00\n",
      "loss: 1.57    acc:41.70\n",
      "epoch 294\n",
      "loss: 1.42    acc:45.77\n",
      "loss: 1.60    acc:40.90\n",
      "epoch 295\n",
      "loss: 1.42    acc:46.50\n",
      "loss: 1.64    acc:43.10\n",
      "epoch 296\n",
      "loss: 1.43    acc:46.42\n",
      "loss: 1.64    acc:41.70\n",
      "epoch 297\n",
      "loss: 1.42    acc:46.02\n",
      "loss: 1.55    acc:44.70\n",
      "epoch 298\n",
      "loss: 1.41    acc:46.40\n",
      "loss: 1.64    acc:41.00\n",
      "epoch 299\n",
      "loss: 1.42    acc:47.05\n",
      "loss: 1.59    acc:43.90\n",
      "epoch 300\n",
      "loss: 1.42    acc:46.40\n",
      "loss: 1.62    acc:42.50\n",
      "epoch 301\n",
      "loss: 1.42    acc:46.60\n",
      "loss: 1.52    acc:43.40\n",
      "epoch 302\n",
      "loss: 1.43    acc:46.30\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 303\n",
      "loss: 1.41    acc:46.17\n",
      "loss: 1.65    acc:41.30\n",
      "epoch 304\n",
      "loss: 1.41    acc:46.60\n",
      "loss: 1.46    acc:46.50\n",
      "epoch 305\n",
      "loss: 1.41    acc:47.37\n",
      "loss: 1.59    acc:41.30\n",
      "epoch 306\n",
      "loss: 1.43    acc:46.47\n",
      "loss: 1.70    acc:39.20\n",
      "epoch 307\n",
      "loss: 1.43    acc:46.05\n",
      "loss: 1.64    acc:39.50\n",
      "epoch 308\n",
      "loss: 1.41    acc:47.37\n",
      "loss: 1.62    acc:42.60\n",
      "epoch 309\n",
      "loss: 1.40    acc:46.85\n",
      "loss: 1.63    acc:41.50\n",
      "epoch 310\n",
      "loss: 1.42    acc:46.47\n",
      "loss: 1.71    acc:40.70\n",
      "epoch 311\n",
      "loss: 1.43    acc:46.10\n",
      "loss: 1.62    acc:40.40\n",
      "epoch 312\n",
      "loss: 1.42    acc:46.65\n",
      "loss: 1.60    acc:43.60\n",
      "epoch 313\n",
      "loss: 1.41    acc:46.85\n",
      "loss: 1.64    acc:41.80\n",
      "epoch 314\n",
      "loss: 1.42    acc:46.57\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 315\n",
      "loss: 1.40    acc:47.65\n",
      "loss: 1.70    acc:38.20\n",
      "epoch 316\n",
      "loss: 1.41    acc:46.30\n",
      "loss: 1.69    acc:39.60\n",
      "epoch 317\n",
      "loss: 1.43    acc:46.67\n",
      "loss: 1.68    acc:42.10\n",
      "epoch 318\n",
      "loss: 1.41    acc:46.77\n",
      "loss: 1.59    acc:43.50\n",
      "epoch 319\n",
      "loss: 1.42    acc:46.82\n",
      "loss: 1.51    acc:44.20\n",
      "epoch 320\n",
      "loss: 1.41    acc:46.42\n",
      "loss: 1.66    acc:41.40\n",
      "epoch 321\n",
      "loss: 1.42    acc:46.65\n",
      "loss: 1.56    acc:43.00\n",
      "epoch 322\n",
      "loss: 1.41    acc:47.02\n",
      "loss: 1.65    acc:43.90\n",
      "epoch 323\n",
      "loss: 1.40    acc:47.32\n",
      "loss: 1.51    acc:44.40\n",
      "epoch 324\n",
      "loss: 1.42    acc:46.95\n",
      "loss: 1.57    acc:40.50\n",
      "epoch 325\n",
      "loss: 1.40    acc:48.12\n",
      "loss: 1.57    acc:42.20\n",
      "epoch 326\n",
      "loss: 1.43    acc:45.42\n",
      "loss: 1.53    acc:44.20\n",
      "epoch 327\n",
      "loss: 1.41    acc:47.40\n",
      "loss: 1.61    acc:41.00\n",
      "epoch 328\n",
      "loss: 1.40    acc:47.12\n",
      "loss: 1.66    acc:41.50\n",
      "epoch 329\n",
      "loss: 1.41    acc:47.32\n",
      "loss: 1.58    acc:43.40\n",
      "epoch 330\n",
      "loss: 1.40    acc:46.72\n",
      "loss: 1.66    acc:39.60\n",
      "epoch 331\n",
      "loss: 1.41    acc:46.95\n",
      "loss: 1.64    acc:40.70\n",
      "epoch 332\n",
      "loss: 1.41    acc:46.42\n",
      "loss: 1.67    acc:41.10\n",
      "epoch 333\n",
      "loss: 1.40    acc:47.50\n",
      "loss: 1.81    acc:37.70\n",
      "epoch 334\n",
      "loss: 1.40    acc:47.10\n",
      "loss: 1.59    acc:42.10\n",
      "epoch 335\n",
      "loss: 1.42    acc:46.95\n",
      "loss: 1.56    acc:42.50\n",
      "epoch 336\n",
      "loss: 1.40    acc:47.55\n",
      "loss: 1.56    acc:42.90\n",
      "epoch 337\n",
      "loss: 1.39    acc:48.67\n",
      "loss: 1.60    acc:40.00\n",
      "epoch 338\n",
      "loss: 1.42    acc:47.52\n",
      "loss: 1.61    acc:40.50\n",
      "epoch 339\n",
      "loss: 1.40    acc:47.55\n",
      "loss: 1.73    acc:38.80\n",
      "epoch 340\n",
      "loss: 1.40    acc:47.62\n",
      "loss: 1.56    acc:41.80\n",
      "epoch 341\n",
      "loss: 1.39    acc:46.90\n",
      "loss: 1.52    acc:46.30\n",
      "epoch 342\n",
      "loss: 1.41    acc:47.37\n",
      "loss: 1.59    acc:42.80\n",
      "epoch 343\n",
      "loss: 1.41    acc:47.07\n",
      "loss: 1.54    acc:45.30\n",
      "epoch 344\n",
      "loss: 1.41    acc:46.92\n",
      "loss: 1.55    acc:43.30\n",
      "epoch 345\n",
      "loss: 1.41    acc:46.22\n",
      "loss: 1.63    acc:41.50\n",
      "epoch 346\n",
      "loss: 1.40    acc:47.20\n",
      "loss: 1.63    acc:41.30\n",
      "epoch 347\n",
      "loss: 1.41    acc:47.50\n",
      "loss: 1.63    acc:42.60\n",
      "epoch 348\n",
      "loss: 1.41    acc:47.30\n",
      "loss: 1.55    acc:42.50\n",
      "epoch 349\n",
      "loss: 1.40    acc:46.32\n",
      "loss: 1.80    acc:36.80\n",
      "epoch 350\n",
      "loss: 1.40    acc:48.42\n",
      "loss: 1.57    acc:44.00\n",
      "epoch 351\n",
      "loss: 1.40    acc:47.57\n",
      "loss: 1.71    acc:41.50\n",
      "epoch 352\n",
      "loss: 1.37    acc:48.05\n",
      "loss: 1.59    acc:42.30\n",
      "epoch 353\n",
      "loss: 1.40    acc:47.82\n",
      "loss: 1.62    acc:42.00\n",
      "epoch 354\n",
      "loss: 1.40    acc:47.70\n",
      "loss: 1.63    acc:41.10\n",
      "epoch 355\n",
      "loss: 1.40    acc:47.97\n",
      "loss: 1.66    acc:41.50\n",
      "epoch 356\n",
      "loss: 1.40    acc:47.05\n",
      "loss: 1.59    acc:41.80\n",
      "epoch 357\n",
      "loss: 1.40    acc:46.95\n",
      "loss: 1.58    acc:45.90\n",
      "epoch 358\n",
      "loss: 1.40    acc:47.17\n",
      "loss: 1.53    acc:42.50\n",
      "epoch 359\n",
      "loss: 1.40    acc:46.85\n",
      "loss: 1.59    acc:43.60\n",
      "epoch 360\n",
      "loss: 1.39    acc:47.50\n",
      "loss: 1.61    acc:39.90\n",
      "epoch 361\n",
      "loss: 1.39    acc:48.00\n",
      "loss: 1.59    acc:43.30\n",
      "epoch 362\n",
      "loss: 1.40    acc:47.87\n",
      "loss: 1.69    acc:39.70\n",
      "epoch 363\n",
      "loss: 1.41    acc:46.60\n",
      "loss: 1.54    acc:47.00\n",
      "epoch 364\n",
      "loss: 1.39    acc:47.20\n",
      "loss: 1.60    acc:42.40\n",
      "epoch 365\n",
      "loss: 1.40    acc:46.77\n",
      "loss: 1.67    acc:41.00\n",
      "epoch 366\n",
      "loss: 1.40    acc:47.10\n",
      "loss: 1.60    acc:40.90\n",
      "epoch 367\n",
      "loss: 1.39    acc:48.35\n",
      "loss: 1.59    acc:43.10\n",
      "epoch 368\n",
      "loss: 1.41    acc:47.27\n",
      "loss: 1.72    acc:39.70\n",
      "epoch 369\n",
      "loss: 1.39    acc:48.12\n",
      "loss: 1.60    acc:45.00\n",
      "epoch 370\n",
      "loss: 1.39    acc:47.92\n",
      "loss: 1.74    acc:41.00\n",
      "epoch 371\n",
      "loss: 1.40    acc:47.00\n",
      "loss: 1.58    acc:41.30\n",
      "epoch 372\n",
      "loss: 1.40    acc:47.65\n",
      "loss: 1.52    acc:45.80\n",
      "epoch 373\n",
      "loss: 1.40    acc:47.77\n",
      "loss: 1.60    acc:42.50\n",
      "Epoch   372: reducing learning rate of group 0 to 1.0000e-04.\n",
      "epoch 374\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.57    acc:43.90\n",
      "epoch 375\n",
      "loss: 1.38    acc:48.00\n",
      "loss: 1.55    acc:44.00\n",
      "epoch 376\n",
      "loss: 1.38    acc:47.45\n",
      "loss: 1.46    acc:47.60\n",
      "epoch 377\n",
      "loss: 1.38    acc:48.57\n",
      "loss: 1.52    acc:45.40\n",
      "epoch 378\n",
      "loss: 1.39    acc:47.62\n",
      "loss: 1.66    acc:41.60\n",
      "epoch 379\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.63    acc:39.80\n",
      "epoch 380\n",
      "loss: 1.39    acc:46.82\n",
      "loss: 1.61    acc:41.80\n",
      "epoch 381\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.56    acc:46.50\n",
      "epoch 382\n",
      "loss: 1.38    acc:48.05\n",
      "loss: 1.58    acc:43.50\n",
      "epoch 383\n",
      "loss: 1.38    acc:47.52\n",
      "loss: 1.63    acc:41.60\n",
      "epoch 384\n",
      "loss: 1.38    acc:48.00\n",
      "loss: 1.52    acc:44.30\n",
      "epoch 385\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.58    acc:41.90\n",
      "epoch 386\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.58    acc:43.00\n",
      "epoch 387\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.55    acc:43.40\n",
      "epoch 388\n",
      "loss: 1.38    acc:47.35\n",
      "loss: 1.51    acc:46.70\n",
      "epoch 389\n",
      "loss: 1.39    acc:47.80\n",
      "loss: 1.49    acc:45.40\n",
      "epoch 390\n",
      "loss: 1.39    acc:47.67\n",
      "loss: 1.64    acc:42.30\n",
      "epoch 391\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.51    acc:45.00\n",
      "epoch 392\n",
      "loss: 1.38    acc:48.30\n",
      "loss: 1.56    acc:43.30\n",
      "epoch 393\n",
      "loss: 1.38    acc:49.97\n",
      "loss: 1.57    acc:42.40\n",
      "epoch 394\n",
      "loss: 1.38    acc:47.77\n",
      "loss: 1.57    acc:43.80\n",
      "Epoch   393: reducing learning rate of group 0 to 3.3333e-05.\n",
      "epoch 395\n",
      "loss: 1.37    acc:47.00\n",
      "loss: 1.57    acc:45.20\n",
      "epoch 396\n",
      "loss: 1.38    acc:49.32\n",
      "loss: 1.57    acc:43.90\n",
      "epoch 397\n",
      "loss: 1.38    acc:47.60\n",
      "loss: 1.52    acc:44.00\n",
      "epoch 398\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.54    acc:44.60\n",
      "epoch 399\n",
      "loss: 1.38    acc:47.85\n",
      "loss: 1.59    acc:43.00\n",
      "epoch 400\n",
      "loss: 1.37    acc:48.17\n",
      "loss: 1.53    acc:44.10\n",
      "epoch 401\n",
      "loss: 1.37    acc:48.65\n",
      "loss: 1.56    acc:44.50\n",
      "epoch 402\n",
      "loss: 1.37    acc:49.02\n",
      "loss: 1.59    acc:42.00\n",
      "epoch 403\n",
      "loss: 1.36    acc:48.37\n",
      "loss: 1.50    acc:45.70\n",
      "epoch 404\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.60    acc:41.60\n",
      "epoch 405\n",
      "loss: 1.37    acc:48.70\n",
      "loss: 1.56    acc:46.30\n",
      "epoch 406\n",
      "loss: 1.39    acc:48.15\n",
      "loss: 1.57    acc:43.10\n",
      "epoch 407\n",
      "loss: 1.39    acc:47.42\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 408\n",
      "loss: 1.39    acc:46.72\n",
      "loss: 1.67    acc:40.80\n",
      "epoch 409\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.63    acc:41.80\n",
      "epoch 410\n",
      "loss: 1.37    acc:47.75\n",
      "loss: 1.64    acc:40.10\n",
      "epoch 411\n",
      "loss: 1.38    acc:48.82\n",
      "loss: 1.61    acc:43.20\n",
      "epoch 412\n",
      "loss: 1.37    acc:49.52\n",
      "loss: 1.57    acc:41.60\n",
      "epoch 413\n",
      "loss: 1.38    acc:49.02\n",
      "loss: 1.54    acc:44.30\n",
      "epoch 414\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.60    acc:43.20\n",
      "epoch 415\n",
      "loss: 1.37    acc:49.05\n",
      "loss: 1.60    acc:42.90\n",
      "epoch 416\n",
      "loss: 1.38    acc:49.25\n",
      "loss: 1.56    acc:42.60\n",
      "epoch 417\n",
      "loss: 1.36    acc:48.57\n",
      "loss: 1.64    acc:42.80\n",
      "epoch 418\n",
      "loss: 1.38    acc:48.35\n",
      "loss: 1.55    acc:43.90\n",
      "epoch 419\n",
      "loss: 1.36    acc:48.72\n",
      "loss: 1.58    acc:43.20\n",
      "epoch 420\n",
      "loss: 1.37    acc:48.27\n",
      "loss: 1.56    acc:44.80\n",
      "epoch 421\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.56    acc:42.80\n",
      "epoch 422\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.56    acc:41.30\n",
      "epoch 423\n",
      "loss: 1.37    acc:49.02\n",
      "loss: 1.56    acc:45.70\n",
      "epoch 424\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.47    acc:46.40\n",
      "Epoch   423: reducing learning rate of group 0 to 1.1111e-05.\n",
      "epoch 425\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.37    acc:48.30\n",
      "loss: 1.56    acc:44.30\n",
      "epoch 426\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.52    acc:46.40\n",
      "epoch 427\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.52    acc:44.40\n",
      "epoch 428\n",
      "loss: 1.38    acc:47.62\n",
      "loss: 1.52    acc:44.50\n",
      "epoch 429\n",
      "loss: 1.36    acc:49.15\n",
      "loss: 1.64    acc:41.00\n",
      "epoch 430\n",
      "loss: 1.37    acc:47.75\n",
      "loss: 1.52    acc:43.90\n",
      "epoch 431\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.56    acc:45.10\n",
      "epoch 432\n",
      "loss: 1.38    acc:48.77\n",
      "loss: 1.54    acc:43.70\n",
      "epoch 433\n",
      "loss: 1.38    acc:48.52\n",
      "loss: 1.59    acc:42.80\n",
      "epoch 434\n",
      "loss: 1.37    acc:47.75\n",
      "loss: 1.50    acc:43.20\n",
      "epoch 435\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.54    acc:42.80\n",
      "epoch 436\n",
      "loss: 1.37    acc:48.55\n",
      "loss: 1.55    acc:46.40\n",
      "epoch 437\n",
      "loss: 1.38    acc:48.50\n",
      "loss: 1.57    acc:42.40\n",
      "epoch 438\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.56    acc:42.50\n",
      "epoch 439\n",
      "loss: 1.38    acc:48.85\n",
      "loss: 1.56    acc:45.60\n",
      "epoch 440\n",
      "loss: 1.36    acc:48.62\n",
      "loss: 1.62    acc:43.00\n",
      "epoch 441\n",
      "loss: 1.37    acc:48.12\n",
      "loss: 1.53    acc:45.40\n",
      "epoch 442\n",
      "loss: 1.37    acc:48.17\n",
      "loss: 1.58    acc:41.30\n",
      "epoch 443\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.61    acc:40.90\n",
      "epoch 444\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.57    acc:44.00\n",
      "epoch 445\n",
      "loss: 1.38    acc:48.35\n",
      "loss: 1.56    acc:43.70\n",
      "Epoch   444: reducing learning rate of group 0 to 3.7037e-06.\n",
      "epoch 446\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.56    acc:43.00\n",
      "epoch 447\n",
      "loss: 1.36    acc:48.52\n",
      "loss: 1.56    acc:43.20\n",
      "epoch 448\n",
      "loss: 1.37    acc:49.02\n",
      "loss: 1.64    acc:40.40\n",
      "epoch 449\n",
      "loss: 1.36    acc:48.17\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 450\n",
      "loss: 1.38    acc:47.85\n",
      "loss: 1.58    acc:41.20\n",
      "epoch 451\n",
      "loss: 1.38    acc:47.90\n",
      "loss: 1.56    acc:43.80\n",
      "epoch 452\n",
      "loss: 1.37    acc:48.67\n",
      "loss: 1.61    acc:42.70\n",
      "epoch 453\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.56    acc:42.90\n",
      "epoch 454\n",
      "loss: 1.37    acc:49.32\n",
      "loss: 1.59    acc:43.20\n",
      "epoch 455\n",
      "loss: 1.37    acc:48.82\n",
      "loss: 1.53    acc:44.70\n",
      "epoch 456\n",
      "loss: 1.37    acc:48.90\n",
      "loss: 1.62    acc:41.90\n",
      "epoch 457\n",
      "loss: 1.37    acc:49.45\n",
      "loss: 1.47    acc:47.90\n",
      "epoch 458\n",
      "loss: 1.37    acc:48.55\n",
      "loss: 1.55    acc:45.10\n",
      "epoch 459\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.55    acc:42.40\n",
      "epoch 460\n",
      "loss: 1.39    acc:47.55\n",
      "loss: 1.64    acc:41.10\n",
      "epoch 461\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.56    acc:44.20\n",
      "epoch 462\n",
      "loss: 1.38    acc:48.87\n",
      "loss: 1.52    acc:45.40\n",
      "epoch 463\n",
      "loss: 1.36    acc:48.47\n",
      "loss: 1.55    acc:44.60\n",
      "epoch 464\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.52    acc:46.60\n",
      "epoch 465\n",
      "loss: 1.39    acc:48.32\n",
      "loss: 1.58    acc:42.90\n",
      "epoch 466\n",
      "loss: 1.37    acc:48.82\n",
      "loss: 1.56    acc:46.80\n",
      "Epoch   465: reducing learning rate of group 0 to 1.2346e-06.\n",
      "epoch 467\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.59    acc:44.80\n",
      "epoch 468\n",
      "loss: 1.38    acc:49.32\n",
      "loss: 1.57    acc:45.90\n",
      "epoch 469\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.60    acc:42.70\n",
      "epoch 470\n",
      "loss: 1.37    acc:48.97\n",
      "loss: 1.58    acc:44.50\n",
      "epoch 471\n",
      "loss: 1.37    acc:47.65\n",
      "loss: 1.58    acc:44.00\n",
      "epoch 472\n",
      "loss: 1.37    acc:48.20\n",
      "loss: 1.61    acc:41.40\n",
      "epoch 473\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.61    acc:39.90\n",
      "epoch 474\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.63    acc:42.10\n",
      "epoch 475\n",
      "loss: 1.37    acc:49.30\n",
      "loss: 1.57    acc:45.00\n",
      "epoch 476\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.53    acc:43.20\n",
      "epoch 477\n",
      "loss: 1.37    acc:49.30\n",
      "loss: 1.61    acc:43.10\n",
      "epoch 478\n",
      "loss: 1.39    acc:48.75\n",
      "loss: 1.59    acc:43.70\n",
      "epoch 479\n",
      "loss: 1.36    acc:48.75\n",
      "loss: 1.58    acc:42.40\n",
      "epoch 480\n",
      "loss: 1.39    acc:48.10\n",
      "loss: 1.58    acc:43.20\n",
      "epoch 481\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.55    acc:43.40\n",
      "epoch 482\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.56    acc:44.10\n",
      "epoch 483\n",
      "loss: 1.38    acc:48.97\n",
      "loss: 1.60    acc:43.20\n",
      "epoch 484\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.61    acc:43.90\n",
      "epoch 485\n",
      "loss: 1.36    acc:49.27\n",
      "loss: 1.60    acc:44.30\n",
      "epoch 486\n",
      "loss: 1.37    acc:48.12\n",
      "loss: 1.56    acc:44.90\n",
      "epoch 487\n",
      "loss: 1.37    acc:48.05\n",
      "loss: 1.53    acc:45.60\n",
      "epoch 488\n",
      "loss: 1.37    acc:49.12\n",
      "loss: 1.54    acc:45.30\n",
      "epoch 489\n",
      "loss: 1.38    acc:48.70\n",
      "loss: 1.53    acc:45.50\n",
      "epoch 490\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.56    acc:44.40\n",
      "epoch 491\n",
      "loss: 1.37    acc:48.40\n",
      "loss: 1.60    acc:43.20\n",
      "epoch 492\n",
      "loss: 1.37    acc:48.50\n",
      "loss: 1.54    acc:44.80\n",
      "epoch 493\n",
      "loss: 1.38    acc:48.90\n",
      "loss: 1.61    acc:41.60\n",
      "epoch 494\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.55    acc:45.50\n",
      "epoch 495\n",
      "loss: 1.37    acc:47.87\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 496\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.50    acc:47.20\n",
      "epoch 497\n",
      "loss: 1.37    acc:47.80\n",
      "loss: 1.57    acc:44.00\n",
      "epoch 498\n",
      "loss: 1.37    acc:49.17\n",
      "loss: 1.55    acc:42.70\n",
      "epoch 499\n",
      "loss: 1.36    acc:48.60\n",
      "loss: 1.60    acc:42.20\n",
      "epoch 500\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.55    acc:45.70\n",
      "epoch 501\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.60    acc:42.50\n",
      "epoch 502\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.56    acc:41.80\n",
      "epoch 503\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.54    acc:44.20\n",
      "epoch 504\n",
      "loss: 1.36    acc:49.45\n",
      "loss: 1.54    acc:44.60\n",
      "epoch 505\n",
      "loss: 1.39    acc:48.20\n",
      "loss: 1.55    acc:46.10\n",
      "epoch 506\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.64    acc:40.00\n",
      "Epoch   505: reducing learning rate of group 0 to 4.1152e-07.\n",
      "epoch 507\n",
      "loss: 1.36    acc:48.87\n",
      "loss: 1.55    acc:44.70\n",
      "epoch 508\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.64    acc:41.50\n",
      "epoch 509\n",
      "loss: 1.39    acc:47.80\n",
      "loss: 1.59    acc:44.50\n",
      "epoch 510\n",
      "loss: 1.37    acc:48.90\n",
      "loss: 1.58    acc:42.50\n",
      "epoch 511\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.64    acc:42.10\n",
      "epoch 512\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.55    acc:43.40\n",
      "epoch 513\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.64    acc:42.30\n",
      "epoch 514\n",
      "loss: 1.38    acc:48.62\n",
      "loss: 1.63    acc:39.80\n",
      "epoch 515\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.56    acc:42.90\n",
      "epoch 516\n",
      "loss: 1.36    acc:49.92\n",
      "loss: 1.58    acc:42.80\n",
      "epoch 517\n",
      "loss: 1.37    acc:49.17\n",
      "loss: 1.63    acc:42.30\n",
      "epoch 518\n",
      "loss: 1.36    acc:48.92\n",
      "loss: 1.55    acc:43.90\n",
      "epoch 519\n",
      "loss: 1.38    acc:48.35\n",
      "loss: 1.54    acc:43.90\n",
      "epoch 520\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.50    acc:44.30\n",
      "epoch 521\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.56    acc:42.90\n",
      "epoch 522\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.51    acc:44.50\n",
      "epoch 523\n",
      "loss: 1.36    acc:48.80\n",
      "loss: 1.53    acc:46.50\n",
      "epoch 524\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.55    acc:43.40\n",
      "epoch 525\n",
      "loss: 1.38    acc:47.92\n",
      "loss: 1.54    acc:45.40\n",
      "epoch 526\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.54    acc:43.50\n",
      "epoch 527\n",
      "loss: 1.37    acc:48.27\n",
      "loss: 1.52    acc:45.20\n",
      "Epoch   526: reducing learning rate of group 0 to 1.3717e-07.\n",
      "epoch 528\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.55    acc:42.30\n",
      "epoch 529\n",
      "loss: 1.37    acc:48.12\n",
      "loss: 1.52    acc:46.20\n",
      "epoch 530\n",
      "loss: 1.36    acc:49.15\n",
      "loss: 1.64    acc:41.70\n",
      "epoch 531\n",
      "loss: 1.38    acc:49.67\n",
      "loss: 1.61    acc:43.10\n",
      "epoch 532\n",
      "loss: 1.36    acc:48.40\n",
      "loss: 1.60    acc:44.50\n",
      "epoch 533\n",
      "loss: 1.38    acc:48.97\n",
      "loss: 1.60    acc:41.50\n",
      "epoch 534\n",
      "loss: 1.37    acc:48.55\n",
      "loss: 1.56    acc:45.80\n",
      "epoch 535\n",
      "loss: 1.37    acc:48.55\n",
      "loss: 1.60    acc:45.20\n",
      "epoch 536\n",
      "loss: 1.38    acc:48.07\n",
      "loss: 1.52    acc:44.80\n",
      "epoch 537\n",
      "loss: 1.38    acc:48.90\n",
      "loss: 1.58    acc:41.70\n",
      "epoch 538\n",
      "loss: 1.38    acc:49.02\n",
      "loss: 1.58    acc:43.40\n",
      "epoch 539\n",
      "loss: 1.37    acc:47.87\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 540\n",
      "loss: 1.36    acc:49.02\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 541\n",
      "loss: 1.36    acc:49.32\n",
      "loss: 1.55    acc:43.40\n",
      "epoch 542\n",
      "loss: 1.38    acc:47.95\n",
      "loss: 1.57    acc:41.10\n",
      "epoch 543\n",
      "loss: 1.38    acc:48.55\n",
      "loss: 1.54    acc:45.80\n",
      "epoch 544\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.49    acc:46.10\n",
      "epoch 545\n",
      "loss: 1.37    acc:49.42\n",
      "loss: 1.59    acc:43.90\n",
      "epoch 546\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.52    acc:44.80\n",
      "epoch 547\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.56    acc:44.60\n",
      "epoch 548\n",
      "loss: 1.37    acc:48.22\n",
      "loss: 1.61    acc:42.10\n",
      "Epoch   547: reducing learning rate of group 0 to 4.5725e-08.\n",
      "epoch 549\n",
      "loss: 1.39    acc:48.25\n",
      "loss: 1.58    acc:46.10\n",
      "epoch 550\n",
      "loss: 1.36    acc:49.07\n",
      "loss: 1.56    acc:44.30\n",
      "epoch 551\n",
      "loss: 1.36    acc:49.35\n",
      "loss: 1.57    acc:43.60\n",
      "epoch 552\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.54    acc:44.30\n",
      "epoch 553\n",
      "loss: 1.36    acc:49.72\n",
      "loss: 1.59    acc:43.90\n",
      "epoch 554\n",
      "loss: 1.39    acc:47.80\n",
      "loss: 1.58    acc:41.60\n",
      "epoch 555\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.59    acc:41.70\n",
      "epoch 556\n",
      "loss: 1.38    acc:47.90\n",
      "loss: 1.54    acc:47.20\n",
      "epoch 557\n",
      "loss: 1.36    acc:49.12\n",
      "loss: 1.56    acc:45.90\n",
      "epoch 558\n",
      "loss: 1.38    acc:47.60\n",
      "loss: 1.63    acc:42.40\n",
      "epoch 559\n",
      "loss: 1.38    acc:48.02\n",
      "loss: 1.55    acc:44.00\n",
      "epoch 560\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.55    acc:43.80\n",
      "epoch 561\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.37    acc:48.62\n",
      "loss: 1.52    acc:43.80\n",
      "epoch 562\n",
      "loss: 1.37    acc:48.50\n",
      "loss: 1.52    acc:43.70\n",
      "epoch 563\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.50    acc:45.80\n",
      "epoch 564\n",
      "loss: 1.37    acc:48.82\n",
      "loss: 1.56    acc:43.90\n",
      "epoch 565\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.51    acc:46.50\n",
      "epoch 566\n",
      "loss: 1.36    acc:48.47\n",
      "loss: 1.56    acc:43.80\n",
      "epoch 567\n",
      "loss: 1.37    acc:48.57\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 568\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.59    acc:41.60\n",
      "epoch 569\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.55    acc:43.90\n",
      "Epoch   568: reducing learning rate of group 0 to 1.5242e-08.\n",
      "epoch 570\n",
      "loss: 1.37    acc:48.57\n",
      "loss: 1.55    acc:44.20\n",
      "epoch 571\n",
      "loss: 1.38    acc:47.90\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 572\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 573\n",
      "loss: 1.38    acc:49.22\n",
      "loss: 1.55    acc:45.80\n",
      "epoch 574\n",
      "loss: 1.37    acc:48.65\n",
      "loss: 1.58    acc:43.80\n",
      "epoch 575\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.61    acc:43.10\n",
      "epoch 576\n",
      "loss: 1.38    acc:49.17\n",
      "loss: 1.57    acc:40.40\n",
      "epoch 577\n",
      "loss: 1.36    acc:49.35\n",
      "loss: 1.55    acc:43.90\n",
      "epoch 578\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.50    acc:47.10\n",
      "epoch 579\n",
      "loss: 1.39    acc:47.60\n",
      "loss: 1.61    acc:41.70\n",
      "epoch 580\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.55    acc:42.90\n",
      "epoch 581\n",
      "loss: 1.38    acc:49.05\n",
      "loss: 1.61    acc:45.20\n",
      "epoch 582\n",
      "loss: 1.37    acc:47.55\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 583\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.65    acc:40.30\n",
      "epoch 584\n",
      "loss: 1.37    acc:48.52\n",
      "loss: 1.53    acc:45.30\n",
      "epoch 585\n",
      "loss: 1.37    acc:48.47\n",
      "loss: 1.54    acc:45.40\n",
      "epoch 586\n",
      "loss: 1.38    acc:49.12\n",
      "loss: 1.55    acc:44.00\n",
      "epoch 587\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.58    acc:41.90\n",
      "epoch 588\n",
      "loss: 1.38    acc:48.72\n",
      "loss: 1.57    acc:43.60\n",
      "epoch 589\n",
      "loss: 1.37    acc:47.72\n",
      "loss: 1.61    acc:42.10\n",
      "epoch 590\n",
      "loss: 1.36    acc:49.22\n",
      "loss: 1.57    acc:43.60\n",
      "Epoch   589: reducing learning rate of group 0 to 5.0805e-09.\n",
      "epoch 591\n",
      "loss: 1.38    acc:48.85\n",
      "loss: 1.54    acc:44.60\n",
      "epoch 592\n",
      "loss: 1.37    acc:47.97\n",
      "loss: 1.60    acc:42.30\n",
      "epoch 593\n",
      "loss: 1.36    acc:49.67\n",
      "loss: 1.56    acc:43.30\n",
      "epoch 594\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.59    acc:43.60\n",
      "epoch 595\n",
      "loss: 1.38    acc:49.10\n",
      "loss: 1.56    acc:43.80\n",
      "epoch 596\n",
      "loss: 1.36    acc:49.42\n",
      "loss: 1.53    acc:46.30\n",
      "epoch 597\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.53    acc:45.60\n",
      "epoch 598\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.54    acc:45.70\n",
      "epoch 599\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.57    acc:44.00\n",
      "epoch 600\n",
      "loss: 1.37    acc:49.47\n",
      "loss: 1.50    acc:44.20\n",
      "epoch 601\n",
      "loss: 1.38    acc:49.55\n",
      "loss: 1.52    acc:44.00\n",
      "epoch 602\n",
      "loss: 1.37    acc:47.57\n",
      "loss: 1.53    acc:45.50\n",
      "epoch 603\n",
      "loss: 1.38    acc:47.77\n",
      "loss: 1.59    acc:42.10\n",
      "epoch 604\n",
      "loss: 1.36    acc:49.02\n",
      "loss: 1.53    acc:44.90\n",
      "epoch 605\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.55    acc:43.80\n",
      "epoch 606\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.63    acc:41.60\n",
      "epoch 607\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.55    acc:44.90\n",
      "epoch 608\n",
      "loss: 1.38    acc:47.57\n",
      "loss: 1.60    acc:45.80\n",
      "epoch 609\n",
      "loss: 1.38    acc:48.07\n",
      "loss: 1.62    acc:42.20\n",
      "epoch 610\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.57    acc:42.50\n",
      "epoch 611\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.57    acc:44.10\n",
      "epoch 612\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.52    acc:43.80\n",
      "epoch 613\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.58    acc:44.60\n",
      "epoch 614\n",
      "loss: 1.36    acc:49.25\n",
      "loss: 1.68    acc:40.20\n",
      "epoch 615\n",
      "loss: 1.37    acc:48.95\n",
      "loss: 1.54    acc:43.90\n",
      "epoch 616\n",
      "loss: 1.36    acc:48.77\n",
      "loss: 1.53    acc:43.90\n",
      "epoch 617\n",
      "loss: 1.38    acc:47.92\n",
      "loss: 1.52    acc:45.80\n",
      "epoch 618\n",
      "loss: 1.37    acc:48.85\n",
      "loss: 1.54    acc:43.40\n",
      "epoch 619\n",
      "loss: 1.38    acc:49.62\n",
      "loss: 1.54    acc:44.30\n",
      "epoch 620\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.55    acc:44.50\n",
      "epoch 621\n",
      "loss: 1.37    acc:48.50\n",
      "loss: 1.62    acc:40.70\n",
      "epoch 622\n",
      "loss: 1.39    acc:47.82\n",
      "loss: 1.56    acc:43.40\n",
      "epoch 623\n",
      "loss: 1.37    acc:47.90\n",
      "loss: 1.59    acc:44.30\n",
      "epoch 624\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.57    acc:42.90\n",
      "epoch 625\n",
      "loss: 1.38    acc:47.57\n",
      "loss: 1.49    acc:46.10\n",
      "epoch 626\n",
      "loss: 1.39    acc:47.65\n",
      "loss: 1.54    acc:44.30\n",
      "epoch 627\n",
      "loss: 1.38    acc:48.97\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 628\n",
      "loss: 1.36    acc:48.52\n",
      "loss: 1.49    acc:44.40\n",
      "epoch 629\n",
      "loss: 1.37    acc:49.22\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 630\n",
      "loss: 1.37    acc:48.95\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 631\n",
      "loss: 1.37    acc:49.12\n",
      "loss: 1.68    acc:40.30\n",
      "epoch 632\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.53    acc:42.80\n",
      "epoch 633\n",
      "loss: 1.38    acc:47.67\n",
      "loss: 1.62    acc:43.90\n",
      "epoch 634\n",
      "loss: 1.38    acc:49.40\n",
      "loss: 1.56    acc:41.70\n",
      "epoch 635\n",
      "loss: 1.38    acc:47.65\n",
      "loss: 1.66    acc:41.30\n",
      "epoch 636\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.57    acc:45.20\n",
      "epoch 637\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.56    acc:44.00\n",
      "epoch 638\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.52    acc:43.90\n",
      "epoch 639\n",
      "loss: 1.37    acc:49.32\n",
      "loss: 1.62    acc:42.40\n",
      "epoch 640\n",
      "loss: 1.37    acc:48.20\n",
      "loss: 1.62    acc:41.70\n",
      "epoch 641\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.56    acc:42.20\n",
      "epoch 642\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.54    acc:45.50\n",
      "epoch 643\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.51    acc:44.60\n",
      "epoch 644\n",
      "loss: 1.37    acc:48.20\n",
      "loss: 1.63    acc:42.20\n",
      "epoch 645\n",
      "loss: 1.38    acc:47.72\n",
      "loss: 1.58    acc:42.20\n",
      "epoch 646\n",
      "loss: 1.38    acc:48.20\n",
      "loss: 1.56    acc:42.90\n",
      "epoch 647\n",
      "loss: 1.37    acc:49.50\n",
      "loss: 1.58    acc:42.00\n",
      "epoch 648\n",
      "loss: 1.36    acc:49.35\n",
      "loss: 1.59    acc:43.50\n",
      "epoch 649\n",
      "loss: 1.36    acc:48.97\n",
      "loss: 1.54    acc:42.90\n",
      "epoch 650\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.65    acc:41.20\n",
      "epoch 651\n",
      "loss: 1.38    acc:47.35\n",
      "loss: 1.56    acc:43.40\n",
      "epoch 652\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.58    acc:42.80\n",
      "epoch 653\n",
      "loss: 1.38    acc:48.40\n",
      "loss: 1.54    acc:44.90\n",
      "epoch 654\n",
      "loss: 1.37    acc:48.47\n",
      "loss: 1.54    acc:44.90\n",
      "epoch 655\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.64    acc:41.00\n",
      "epoch 656\n",
      "loss: 1.37    acc:49.12\n",
      "loss: 1.58    acc:42.30\n",
      "epoch 657\n",
      "loss: 1.38    acc:48.40\n",
      "loss: 1.59    acc:43.00\n",
      "epoch 658\n",
      "loss: 1.38    acc:47.57\n",
      "loss: 1.57    acc:43.10\n",
      "epoch 659\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.60    acc:43.70\n",
      "epoch 660\n",
      "loss: 1.36    acc:49.10\n",
      "loss: 1.62    acc:41.80\n",
      "epoch 661\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.55    acc:43.70\n",
      "epoch 662\n",
      "loss: 1.37    acc:49.02\n",
      "loss: 1.60    acc:41.90\n",
      "epoch 663\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.54    acc:44.10\n",
      "epoch 664\n",
      "loss: 1.37    acc:48.82\n",
      "loss: 1.58    acc:43.00\n",
      "epoch 665\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.58    acc:41.90\n",
      "epoch 666\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.60    acc:43.90\n",
      "epoch 667\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.58    acc:43.40\n",
      "epoch 668\n",
      "loss: 1.38    acc:47.87\n",
      "loss: 1.55    acc:44.30\n",
      "epoch 669\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.53    acc:45.00\n",
      "epoch 670\n",
      "loss: 1.38    acc:48.40\n",
      "loss: 1.55    acc:44.10\n",
      "epoch 671\n",
      "loss: 1.38    acc:47.65\n",
      "loss: 1.60    acc:43.70\n",
      "epoch 672\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.53    acc:44.40\n",
      "epoch 673\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.66    acc:39.60\n",
      "epoch 674\n",
      "loss: 1.37    acc:48.85\n",
      "loss: 1.57    acc:43.10\n",
      "epoch 675\n",
      "loss: 1.38    acc:47.72\n",
      "loss: 1.55    acc:45.10\n",
      "epoch 676\n",
      "loss: 1.38    acc:48.30\n",
      "loss: 1.67    acc:40.10\n",
      "epoch 677\n",
      "loss: 1.37    acc:48.52\n",
      "loss: 1.54    acc:42.80\n",
      "epoch 678\n",
      "loss: 1.38    acc:47.97\n",
      "loss: 1.62    acc:42.30\n",
      "epoch 679\n",
      "loss: 1.37    acc:48.20\n",
      "loss: 1.54    acc:45.20\n",
      "epoch 680\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.51    acc:44.10\n",
      "epoch 681\n",
      "loss: 1.38    acc:47.85\n",
      "loss: 1.63    acc:43.30\n",
      "epoch 682\n",
      "loss: 1.38    acc:47.47\n",
      "loss: 1.67    acc:41.80\n",
      "epoch 683\n",
      "loss: 1.36    acc:48.90\n",
      "loss: 1.68    acc:40.00\n",
      "epoch 684\n",
      "loss: 1.37    acc:47.65\n",
      "loss: 1.58    acc:43.40\n",
      "epoch 685\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.60    acc:42.40\n",
      "epoch 686\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.53    acc:46.10\n",
      "epoch 687\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.55    acc:43.90\n",
      "epoch 688\n",
      "loss: 1.38    acc:48.65\n",
      "loss: 1.61    acc:41.20\n",
      "epoch 689\n",
      "loss: 1.38    acc:48.57\n",
      "loss: 1.58    acc:43.30\n",
      "epoch 690\n",
      "loss: 1.36    acc:48.95\n",
      "loss: 1.61    acc:41.70\n",
      "epoch 691\n",
      "loss: 1.37    acc:49.30\n",
      "loss: 1.59    acc:41.70\n",
      "epoch 692\n",
      "loss: 1.38    acc:48.67\n",
      "loss: 1.53    acc:44.20\n",
      "epoch 693\n",
      "loss: 1.37    acc:47.90\n",
      "loss: 1.56    acc:45.00\n",
      "epoch 694\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.58    acc:43.30\n",
      "epoch 695\n",
      "loss: 1.38    acc:48.75\n",
      "loss: 1.54    acc:44.10\n",
      "epoch 696\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.58    acc:41.80\n",
      "epoch 697\n",
      "loss: 1.37    acc:48.27\n",
      "loss: 1.55    acc:43.90\n",
      "epoch 698\n",
      "loss: 1.36    acc:48.97\n",
      "loss: 1.58    acc:43.00\n",
      "epoch 699\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.60    acc:41.00\n",
      "epoch 700\n",
      "loss: 1.36    acc:48.62\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.62    acc:40.20\n",
      "epoch 701\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.69    acc:42.10\n",
      "epoch 702\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.60    acc:44.90\n",
      "epoch 703\n",
      "loss: 1.38    acc:48.32\n",
      "loss: 1.63    acc:39.70\n",
      "epoch 704\n",
      "loss: 1.38    acc:47.62\n",
      "loss: 1.63    acc:40.40\n",
      "epoch 705\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.52    acc:46.60\n",
      "epoch 706\n",
      "loss: 1.37    acc:47.62\n",
      "loss: 1.64    acc:42.10\n",
      "epoch 707\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.52    acc:46.10\n",
      "epoch 708\n",
      "loss: 1.37    acc:47.67\n",
      "loss: 1.58    acc:42.60\n",
      "epoch 709\n",
      "loss: 1.37    acc:47.02\n",
      "loss: 1.58    acc:41.30\n",
      "epoch 710\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.63    acc:42.60\n",
      "epoch 711\n",
      "loss: 1.36    acc:48.55\n",
      "loss: 1.53    acc:45.30\n",
      "epoch 712\n",
      "loss: 1.37    acc:48.85\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 713\n",
      "loss: 1.38    acc:48.67\n",
      "loss: 1.61    acc:41.40\n",
      "epoch 714\n",
      "loss: 1.37    acc:49.32\n",
      "loss: 1.63    acc:41.90\n",
      "epoch 715\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.53    acc:45.10\n",
      "epoch 716\n",
      "loss: 1.37    acc:48.52\n",
      "loss: 1.53    acc:43.30\n",
      "epoch 717\n",
      "loss: 1.38    acc:47.65\n",
      "loss: 1.50    acc:43.30\n",
      "epoch 718\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.62    acc:42.50\n",
      "epoch 719\n",
      "loss: 1.37    acc:49.05\n",
      "loss: 1.59    acc:41.00\n",
      "epoch 720\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.65    acc:42.40\n",
      "epoch 721\n",
      "loss: 1.37    acc:48.00\n",
      "loss: 1.57    acc:42.80\n",
      "epoch 722\n",
      "loss: 1.37    acc:49.50\n",
      "loss: 1.55    acc:43.60\n",
      "epoch 723\n",
      "loss: 1.36    acc:48.75\n",
      "loss: 1.65    acc:43.80\n",
      "epoch 724\n",
      "loss: 1.37    acc:48.35\n",
      "loss: 1.64    acc:43.90\n",
      "epoch 725\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.54    acc:45.00\n",
      "epoch 726\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 727\n",
      "loss: 1.38    acc:47.27\n",
      "loss: 1.53    acc:45.30\n",
      "epoch 728\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.60    acc:43.00\n",
      "epoch 729\n",
      "loss: 1.39    acc:47.60\n",
      "loss: 1.55    acc:45.50\n",
      "epoch 730\n",
      "loss: 1.37    acc:49.50\n",
      "loss: 1.55    acc:44.20\n",
      "epoch 731\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 732\n",
      "loss: 1.36    acc:48.65\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 733\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.48    acc:45.90\n",
      "epoch 734\n",
      "loss: 1.36    acc:48.30\n",
      "loss: 1.54    acc:44.00\n",
      "epoch 735\n",
      "loss: 1.37    acc:48.90\n",
      "loss: 1.52    acc:45.70\n",
      "epoch 736\n",
      "loss: 1.38    acc:47.92\n",
      "loss: 1.58    acc:43.80\n",
      "epoch 737\n",
      "loss: 1.38    acc:49.15\n",
      "loss: 1.57    acc:43.10\n",
      "epoch 738\n",
      "loss: 1.36    acc:49.00\n",
      "loss: 1.56    acc:44.80\n",
      "epoch 739\n",
      "loss: 1.36    acc:48.75\n",
      "loss: 1.60    acc:41.40\n",
      "epoch 740\n",
      "loss: 1.38    acc:49.45\n",
      "loss: 1.57    acc:41.50\n",
      "epoch 741\n",
      "loss: 1.36    acc:48.62\n",
      "loss: 1.58    acc:42.30\n",
      "epoch 742\n",
      "loss: 1.37    acc:48.10\n",
      "loss: 1.58    acc:41.70\n",
      "epoch 743\n",
      "loss: 1.38    acc:47.67\n",
      "loss: 1.58    acc:44.00\n",
      "epoch 744\n",
      "loss: 1.38    acc:48.07\n",
      "loss: 1.60    acc:43.60\n",
      "epoch 745\n",
      "loss: 1.37    acc:49.17\n",
      "loss: 1.53    acc:43.40\n",
      "epoch 746\n",
      "loss: 1.38    acc:49.02\n",
      "loss: 1.52    acc:45.80\n",
      "epoch 747\n",
      "loss: 1.37    acc:48.85\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 748\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.56    acc:44.80\n",
      "epoch 749\n",
      "loss: 1.38    acc:47.57\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 750\n",
      "loss: 1.38    acc:48.77\n",
      "loss: 1.60    acc:44.20\n",
      "epoch 751\n",
      "loss: 1.38    acc:47.87\n",
      "loss: 1.54    acc:44.70\n",
      "epoch 752\n",
      "loss: 1.37    acc:48.97\n",
      "loss: 1.58    acc:43.70\n",
      "epoch 753\n",
      "loss: 1.38    acc:48.52\n",
      "loss: 1.59    acc:41.70\n",
      "epoch 754\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.54    acc:43.90\n",
      "epoch 755\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.52    acc:46.40\n",
      "epoch 756\n",
      "loss: 1.36    acc:48.30\n",
      "loss: 1.55    acc:43.70\n",
      "epoch 757\n",
      "loss: 1.39    acc:48.42\n",
      "loss: 1.55    acc:42.00\n",
      "epoch 758\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.56    acc:43.20\n",
      "epoch 759\n",
      "loss: 1.38    acc:48.20\n",
      "loss: 1.61    acc:41.50\n",
      "epoch 760\n",
      "loss: 1.38    acc:48.72\n",
      "loss: 1.48    acc:46.60\n",
      "epoch 761\n",
      "loss: 1.38    acc:48.62\n",
      "loss: 1.59    acc:41.60\n",
      "epoch 762\n",
      "loss: 1.38    acc:47.92\n",
      "loss: 1.54    acc:44.10\n",
      "epoch 763\n",
      "loss: 1.37    acc:48.17\n",
      "loss: 1.57    acc:42.50\n",
      "epoch 764\n",
      "loss: 1.37    acc:48.67\n",
      "loss: 1.61    acc:41.30\n",
      "epoch 765\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 766\n",
      "loss: 1.36    acc:47.62\n",
      "loss: 1.49    acc:45.00\n",
      "epoch 767\n",
      "loss: 1.37    acc:48.52\n",
      "loss: 1.60    acc:41.50\n",
      "epoch 768\n",
      "loss: 1.36    acc:48.85\n",
      "loss: 1.60    acc:43.20\n",
      "epoch 769\n",
      "loss: 1.38    acc:47.77\n",
      "loss: 1.53    acc:43.90\n",
      "epoch 770\n",
      "loss: 1.37    acc:48.70\n",
      "loss: 1.59    acc:42.50\n",
      "epoch 771\n",
      "loss: 1.37    acc:48.22\n",
      "loss: 1.58    acc:43.10\n",
      "epoch 772\n",
      "loss: 1.37    acc:48.12\n",
      "loss: 1.53    acc:44.90\n",
      "epoch 773\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.57    acc:43.30\n",
      "epoch 774\n",
      "loss: 1.37    acc:49.05\n",
      "loss: 1.50    acc:47.10\n",
      "epoch 775\n",
      "loss: 1.37    acc:48.70\n",
      "loss: 1.58    acc:43.40\n",
      "epoch 776\n",
      "loss: 1.38    acc:47.95\n",
      "loss: 1.55    acc:44.10\n",
      "epoch 777\n",
      "loss: 1.37    acc:47.45\n",
      "loss: 1.63    acc:42.80\n",
      "epoch 778\n",
      "loss: 1.38    acc:48.52\n",
      "loss: 1.53    acc:43.50\n",
      "epoch 779\n",
      "loss: 1.38    acc:48.32\n",
      "loss: 1.51    acc:45.30\n",
      "epoch 780\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.56    acc:45.80\n",
      "epoch 781\n",
      "loss: 1.37    acc:48.87\n",
      "loss: 1.55    acc:44.60\n",
      "epoch 782\n",
      "loss: 1.37    acc:48.05\n",
      "loss: 1.54    acc:44.60\n",
      "epoch 783\n",
      "loss: 1.38    acc:46.87\n",
      "loss: 1.57    acc:41.50\n",
      "epoch 784\n",
      "loss: 1.37    acc:49.15\n",
      "loss: 1.59    acc:42.40\n",
      "epoch 785\n",
      "loss: 1.37    acc:47.12\n",
      "loss: 1.61    acc:41.90\n",
      "epoch 786\n",
      "loss: 1.37    acc:49.00\n",
      "loss: 1.52    acc:44.60\n",
      "epoch 787\n",
      "loss: 1.38    acc:47.47\n",
      "loss: 1.63    acc:43.20\n",
      "epoch 788\n",
      "loss: 1.37    acc:48.10\n",
      "loss: 1.58    acc:43.50\n",
      "epoch 789\n",
      "loss: 1.38    acc:48.82\n",
      "loss: 1.53    acc:43.70\n",
      "epoch 790\n",
      "loss: 1.36    acc:48.62\n",
      "loss: 1.60    acc:42.70\n",
      "epoch 791\n",
      "loss: 1.37    acc:47.60\n",
      "loss: 1.59    acc:43.10\n",
      "epoch 792\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.61    acc:42.80\n",
      "epoch 793\n",
      "loss: 1.37    acc:48.82\n",
      "loss: 1.57    acc:44.10\n",
      "epoch 794\n",
      "loss: 1.37    acc:49.40\n",
      "loss: 1.53    acc:46.00\n",
      "epoch 795\n",
      "loss: 1.37    acc:49.12\n",
      "loss: 1.56    acc:43.90\n",
      "epoch 796\n",
      "loss: 1.38    acc:48.67\n",
      "loss: 1.60    acc:42.70\n",
      "epoch 797\n",
      "loss: 1.38    acc:48.50\n",
      "loss: 1.56    acc:44.20\n",
      "epoch 798\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.61    acc:42.00\n",
      "epoch 799\n",
      "loss: 1.37    acc:48.87\n",
      "loss: 1.59    acc:40.10\n",
      "epoch 800\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.52    acc:45.50\n",
      "epoch 801\n",
      "loss: 1.38    acc:48.32\n",
      "loss: 1.61    acc:43.00\n",
      "epoch 802\n",
      "loss: 1.39    acc:47.67\n",
      "loss: 1.58    acc:42.50\n",
      "epoch 803\n",
      "loss: 1.36    acc:48.32\n",
      "loss: 1.57    acc:42.70\n",
      "epoch 804\n",
      "loss: 1.36    acc:49.02\n",
      "loss: 1.55    acc:46.20\n",
      "epoch 805\n",
      "loss: 1.37    acc:47.87\n",
      "loss: 1.59    acc:45.20\n",
      "epoch 806\n",
      "loss: 1.36    acc:49.25\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 807\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.56    acc:42.00\n",
      "epoch 808\n",
      "loss: 1.37    acc:49.12\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 809\n",
      "loss: 1.36    acc:49.40\n",
      "loss: 1.52    acc:44.10\n",
      "epoch 810\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.54    acc:44.80\n",
      "epoch 811\n",
      "loss: 1.37    acc:48.87\n",
      "loss: 1.58    acc:43.30\n",
      "epoch 812\n",
      "loss: 1.38    acc:48.17\n",
      "loss: 1.56    acc:43.30\n",
      "epoch 813\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.56    acc:42.20\n",
      "epoch 814\n",
      "loss: 1.38    acc:48.72\n",
      "loss: 1.59    acc:43.90\n",
      "epoch 815\n",
      "loss: 1.38    acc:48.82\n",
      "loss: 1.55    acc:44.60\n",
      "epoch 816\n",
      "loss: 1.37    acc:48.60\n",
      "loss: 1.58    acc:42.80\n",
      "epoch 817\n",
      "loss: 1.37    acc:48.65\n",
      "loss: 1.59    acc:41.70\n",
      "epoch 818\n",
      "loss: 1.37    acc:48.05\n",
      "loss: 1.61    acc:45.50\n",
      "epoch 819\n",
      "loss: 1.36    acc:49.35\n",
      "loss: 1.60    acc:44.10\n",
      "epoch 820\n",
      "loss: 1.36    acc:49.57\n",
      "loss: 1.52    acc:43.90\n",
      "epoch 821\n",
      "loss: 1.36    acc:49.42\n",
      "loss: 1.59    acc:44.10\n",
      "epoch 822\n",
      "loss: 1.37    acc:48.90\n",
      "loss: 1.56    acc:44.00\n",
      "epoch 823\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.60    acc:41.60\n",
      "epoch 824\n",
      "loss: 1.36    acc:48.55\n",
      "loss: 1.59    acc:41.60\n",
      "epoch 825\n",
      "loss: 1.37    acc:48.27\n",
      "loss: 1.54    acc:45.10\n",
      "epoch 826\n",
      "loss: 1.35    acc:50.37\n",
      "loss: 1.51    acc:47.30\n",
      "epoch 827\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.49    acc:45.00\n",
      "epoch 828\n",
      "loss: 1.38    acc:48.57\n",
      "loss: 1.56    acc:45.60\n",
      "epoch 829\n",
      "loss: 1.36    acc:48.67\n",
      "loss: 1.60    acc:42.40\n",
      "epoch 830\n",
      "loss: 1.37    acc:49.77\n",
      "loss: 1.53    acc:45.50\n",
      "epoch 831\n",
      "loss: 1.38    acc:48.02\n",
      "loss: 1.60    acc:42.40\n",
      "epoch 832\n",
      "loss: 1.37    acc:49.27\n",
      "loss: 1.54    acc:45.10\n",
      "epoch 833\n",
      "loss: 1.36    acc:49.15\n",
      "loss: 1.58    acc:43.80\n",
      "epoch 834\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.58    acc:43.50\n",
      "epoch 835\n",
      "loss: 1.37    acc:49.20\n",
      "loss: 1.57    acc:44.70\n",
      "epoch 836\n",
      "loss: 1.38    acc:47.65\n",
      "loss: 1.55    acc:44.50\n",
      "epoch 837\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.59    acc:43.10\n",
      "epoch 838\n",
      "loss: 1.38    acc:48.25\n",
      "loss: 1.58    acc:43.90\n",
      "epoch 839\n",
      "loss: 1.38    acc:48.52\n",
      "loss: 1.67    acc:41.40\n",
      "epoch 840\n",
      "loss: 1.37    acc:49.42\n",
      "loss: 1.65    acc:43.50\n",
      "epoch 841\n",
      "loss: 1.37    acc:49.00\n",
      "loss: 1.52    acc:44.40\n",
      "epoch 842\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.38    acc:47.35\n",
      "loss: 1.57    acc:43.40\n",
      "epoch 843\n",
      "loss: 1.38    acc:47.52\n",
      "loss: 1.52    acc:43.40\n",
      "epoch 844\n",
      "loss: 1.37    acc:47.87\n",
      "loss: 1.54    acc:45.60\n",
      "epoch 845\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.57    acc:45.50\n",
      "epoch 846\n",
      "loss: 1.38    acc:47.70\n",
      "loss: 1.62    acc:42.90\n",
      "epoch 847\n",
      "loss: 1.37    acc:48.85\n",
      "loss: 1.59    acc:43.30\n",
      "epoch 848\n",
      "loss: 1.38    acc:48.75\n",
      "loss: 1.60    acc:40.70\n",
      "epoch 849\n",
      "loss: 1.37    acc:49.30\n",
      "loss: 1.52    acc:45.20\n",
      "epoch 850\n",
      "loss: 1.37    acc:49.55\n",
      "loss: 1.49    acc:47.30\n",
      "epoch 851\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.55    acc:44.00\n",
      "epoch 852\n",
      "loss: 1.36    acc:49.30\n",
      "loss: 1.56    acc:42.40\n",
      "epoch 853\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.55    acc:44.70\n",
      "epoch 854\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.57    acc:42.70\n",
      "epoch 855\n",
      "loss: 1.38    acc:48.65\n",
      "loss: 1.50    acc:44.70\n",
      "epoch 856\n",
      "loss: 1.36    acc:48.62\n",
      "loss: 1.61    acc:42.40\n",
      "epoch 857\n",
      "loss: 1.37    acc:48.57\n",
      "loss: 1.51    acc:45.50\n",
      "epoch 858\n",
      "loss: 1.38    acc:48.12\n",
      "loss: 1.57    acc:45.30\n",
      "epoch 859\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.55    acc:44.00\n",
      "epoch 860\n",
      "loss: 1.37    acc:49.25\n",
      "loss: 1.49    acc:46.00\n",
      "epoch 861\n",
      "loss: 1.37    acc:48.50\n",
      "loss: 1.57    acc:42.20\n",
      "epoch 862\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.52    acc:44.10\n",
      "epoch 863\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.53    acc:46.20\n",
      "epoch 864\n",
      "loss: 1.38    acc:48.85\n",
      "loss: 1.63    acc:43.00\n",
      "epoch 865\n",
      "loss: 1.36    acc:48.67\n",
      "loss: 1.58    acc:45.10\n",
      "epoch 866\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.53    acc:44.50\n",
      "epoch 867\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.52    acc:45.20\n",
      "epoch 868\n",
      "loss: 1.37    acc:47.57\n",
      "loss: 1.57    acc:45.20\n",
      "epoch 869\n",
      "loss: 1.38    acc:48.17\n",
      "loss: 1.63    acc:41.50\n",
      "epoch 870\n",
      "loss: 1.37    acc:47.75\n",
      "loss: 1.61    acc:41.30\n",
      "epoch 871\n",
      "loss: 1.39    acc:47.15\n",
      "loss: 1.57    acc:43.20\n",
      "epoch 872\n",
      "loss: 1.37    acc:49.00\n",
      "loss: 1.66    acc:42.80\n",
      "epoch 873\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.60    acc:42.50\n",
      "epoch 874\n",
      "loss: 1.39    acc:48.20\n",
      "loss: 1.59    acc:43.10\n",
      "epoch 875\n",
      "loss: 1.39    acc:48.20\n",
      "loss: 1.61    acc:42.50\n",
      "epoch 876\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.52    acc:45.00\n",
      "epoch 877\n",
      "loss: 1.37    acc:49.25\n",
      "loss: 1.52    acc:43.40\n",
      "epoch 878\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.61    acc:41.70\n",
      "epoch 879\n",
      "loss: 1.36    acc:49.02\n",
      "loss: 1.60    acc:42.10\n",
      "epoch 880\n",
      "loss: 1.38    acc:47.75\n",
      "loss: 1.52    acc:46.00\n",
      "epoch 881\n",
      "loss: 1.38    acc:49.27\n",
      "loss: 1.57    acc:44.60\n",
      "epoch 882\n",
      "loss: 1.36    acc:48.52\n",
      "loss: 1.53    acc:44.40\n",
      "epoch 883\n",
      "loss: 1.38    acc:48.42\n",
      "loss: 1.64    acc:41.80\n",
      "epoch 884\n",
      "loss: 1.37    acc:48.15\n",
      "loss: 1.60    acc:41.20\n",
      "epoch 885\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.49    acc:46.50\n",
      "epoch 886\n",
      "loss: 1.38    acc:48.92\n",
      "loss: 1.53    acc:42.20\n",
      "epoch 887\n",
      "loss: 1.38    acc:48.40\n",
      "loss: 1.63    acc:42.30\n",
      "epoch 888\n",
      "loss: 1.37    acc:48.92\n",
      "loss: 1.59    acc:43.30\n",
      "epoch 889\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.59    acc:42.80\n",
      "epoch 890\n",
      "loss: 1.39    acc:48.02\n",
      "loss: 1.60    acc:43.50\n",
      "epoch 891\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.57    acc:44.70\n",
      "epoch 892\n",
      "loss: 1.37    acc:47.82\n",
      "loss: 1.58    acc:42.60\n",
      "epoch 893\n",
      "loss: 1.37    acc:48.45\n",
      "loss: 1.55    acc:43.50\n",
      "epoch 894\n",
      "loss: 1.38    acc:49.02\n",
      "loss: 1.61    acc:40.90\n",
      "epoch 895\n",
      "loss: 1.37    acc:47.85\n",
      "loss: 1.53    acc:44.50\n",
      "epoch 896\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.62    acc:42.90\n",
      "epoch 897\n",
      "loss: 1.36    acc:49.32\n",
      "loss: 1.58    acc:43.90\n",
      "epoch 898\n",
      "loss: 1.38    acc:48.17\n",
      "loss: 1.57    acc:44.00\n",
      "epoch 899\n",
      "loss: 1.37    acc:49.02\n",
      "loss: 1.59    acc:42.90\n",
      "epoch 900\n",
      "loss: 1.37    acc:48.50\n",
      "loss: 1.62    acc:42.00\n",
      "epoch 901\n",
      "loss: 1.37    acc:48.37\n",
      "loss: 1.68    acc:41.70\n",
      "epoch 902\n",
      "loss: 1.36    acc:49.22\n",
      "loss: 1.56    acc:43.50\n",
      "epoch 903\n",
      "loss: 1.36    acc:48.90\n",
      "loss: 1.49    acc:45.30\n",
      "epoch 904\n",
      "loss: 1.36    acc:48.85\n",
      "loss: 1.56    acc:43.90\n",
      "epoch 905\n",
      "loss: 1.35    acc:49.67\n",
      "loss: 1.61    acc:42.80\n",
      "epoch 906\n",
      "loss: 1.37    acc:49.20\n",
      "loss: 1.54    acc:44.80\n",
      "epoch 907\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.61    acc:42.10\n",
      "epoch 908\n",
      "loss: 1.37    acc:48.30\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 909\n",
      "loss: 1.36    acc:49.40\n",
      "loss: 1.58    acc:44.10\n",
      "epoch 910\n",
      "loss: 1.38    acc:48.32\n",
      "loss: 1.56    acc:41.90\n",
      "epoch 911\n",
      "loss: 1.37    acc:49.25\n",
      "loss: 1.63    acc:41.20\n",
      "epoch 912\n",
      "loss: 1.38    acc:48.17\n",
      "loss: 1.65    acc:41.80\n",
      "epoch 913\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.54    acc:44.00\n",
      "epoch 914\n",
      "loss: 1.37    acc:49.67\n",
      "loss: 1.56    acc:45.80\n",
      "epoch 915\n",
      "loss: 1.36    acc:48.70\n",
      "loss: 1.54    acc:44.80\n",
      "epoch 916\n",
      "loss: 1.37    acc:48.32\n",
      "loss: 1.57    acc:44.20\n",
      "epoch 917\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.57    acc:42.60\n",
      "epoch 918\n",
      "loss: 1.37    acc:49.40\n",
      "loss: 1.53    acc:45.30\n",
      "epoch 919\n",
      "loss: 1.37    acc:48.87\n",
      "loss: 1.59    acc:42.40\n",
      "epoch 920\n",
      "loss: 1.37    acc:48.70\n",
      "loss: 1.59    acc:42.70\n",
      "epoch 921\n",
      "loss: 1.37    acc:48.80\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 922\n",
      "loss: 1.38    acc:47.85\n",
      "loss: 1.59    acc:42.60\n",
      "epoch 923\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.56    acc:43.60\n",
      "epoch 924\n",
      "loss: 1.37    acc:48.22\n",
      "loss: 1.56    acc:43.40\n",
      "epoch 925\n",
      "loss: 1.37    acc:48.62\n",
      "loss: 1.59    acc:42.30\n",
      "epoch 926\n",
      "loss: 1.37    acc:49.30\n",
      "loss: 1.57    acc:45.20\n",
      "epoch 927\n",
      "loss: 1.38    acc:49.30\n",
      "loss: 1.49    acc:46.70\n",
      "epoch 928\n",
      "loss: 1.37    acc:47.95\n",
      "loss: 1.55    acc:45.10\n",
      "epoch 929\n",
      "loss: 1.37    acc:48.07\n",
      "loss: 1.57    acc:42.30\n",
      "epoch 930\n",
      "loss: 1.37    acc:48.10\n",
      "loss: 1.62    acc:42.80\n",
      "epoch 931\n",
      "loss: 1.37    acc:48.95\n",
      "loss: 1.53    acc:43.00\n",
      "epoch 932\n",
      "loss: 1.37    acc:47.60\n",
      "loss: 1.60    acc:46.30\n",
      "epoch 933\n",
      "loss: 1.38    acc:47.45\n",
      "loss: 1.56    acc:43.50\n",
      "epoch 934\n",
      "loss: 1.38    acc:48.32\n",
      "loss: 1.56    acc:43.80\n",
      "epoch 935\n",
      "loss: 1.38    acc:47.95\n",
      "loss: 1.58    acc:41.70\n",
      "epoch 936\n",
      "loss: 1.37    acc:49.10\n",
      "loss: 1.56    acc:43.10\n",
      "epoch 937\n",
      "loss: 1.38    acc:49.52\n",
      "loss: 1.48    acc:43.30\n",
      "epoch 938\n",
      "loss: 1.36    acc:48.87\n",
      "loss: 1.57    acc:43.60\n",
      "epoch 939\n",
      "loss: 1.38    acc:48.77\n",
      "loss: 1.61    acc:41.60\n",
      "epoch 940\n",
      "loss: 1.38    acc:48.55\n",
      "loss: 1.51    acc:44.00\n",
      "epoch 941\n",
      "loss: 1.39    acc:48.10\n",
      "loss: 1.51    acc:47.00\n",
      "epoch 942\n",
      "loss: 1.38    acc:48.47\n",
      "loss: 1.58    acc:42.90\n",
      "epoch 943\n",
      "loss: 1.39    acc:47.15\n",
      "loss: 1.58    acc:44.10\n",
      "epoch 944\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.52    acc:45.60\n",
      "epoch 945\n",
      "loss: 1.38    acc:48.22\n",
      "loss: 1.59    acc:42.00\n",
      "epoch 946\n",
      "loss: 1.37    acc:48.52\n",
      "loss: 1.55    acc:45.30\n",
      "epoch 947\n",
      "loss: 1.37    acc:48.70\n",
      "loss: 1.52    acc:44.50\n",
      "epoch 948\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.59    acc:41.90\n",
      "epoch 949\n",
      "loss: 1.38    acc:48.45\n",
      "loss: 1.61    acc:44.90\n",
      "epoch 950\n",
      "loss: 1.38    acc:47.10\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 951\n",
      "loss: 1.38    acc:48.85\n",
      "loss: 1.59    acc:43.50\n",
      "epoch 952\n",
      "loss: 1.38    acc:48.30\n",
      "loss: 1.55    acc:43.50\n",
      "epoch 953\n",
      "loss: 1.37    acc:49.57\n",
      "loss: 1.59    acc:42.80\n",
      "epoch 954\n",
      "loss: 1.37    acc:48.72\n",
      "loss: 1.60    acc:42.10\n",
      "epoch 955\n",
      "loss: 1.38    acc:48.17\n",
      "loss: 1.55    acc:45.30\n",
      "epoch 956\n",
      "loss: 1.38    acc:48.80\n",
      "loss: 1.54    acc:45.50\n",
      "epoch 957\n",
      "loss: 1.37    acc:48.57\n",
      "loss: 1.57    acc:45.30\n",
      "epoch 958\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.65    acc:40.70\n",
      "epoch 959\n",
      "loss: 1.38    acc:48.37\n",
      "loss: 1.59    acc:42.80\n",
      "epoch 960\n",
      "loss: 1.37    acc:48.12\n",
      "loss: 1.57    acc:43.60\n",
      "epoch 961\n",
      "loss: 1.37    acc:49.20\n",
      "loss: 1.61    acc:43.10\n",
      "epoch 962\n",
      "loss: 1.37    acc:49.25\n",
      "loss: 1.56    acc:43.70\n",
      "epoch 963\n",
      "loss: 1.38    acc:48.30\n",
      "loss: 1.56    acc:44.40\n",
      "epoch 964\n",
      "loss: 1.37    acc:48.42\n",
      "loss: 1.63    acc:43.00\n",
      "epoch 965\n",
      "loss: 1.38    acc:48.87\n",
      "loss: 1.65    acc:41.60\n",
      "epoch 966\n",
      "loss: 1.37    acc:49.60\n",
      "loss: 1.54    acc:42.60\n",
      "epoch 967\n",
      "loss: 1.37    acc:49.35\n",
      "loss: 1.58    acc:42.50\n",
      "epoch 968\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.54    acc:43.80\n",
      "epoch 969\n",
      "loss: 1.36    acc:48.05\n",
      "loss: 1.54    acc:45.40\n",
      "epoch 970\n",
      "loss: 1.36    acc:48.70\n",
      "loss: 1.59    acc:41.50\n",
      "epoch 971\n",
      "loss: 1.37    acc:48.27\n",
      "loss: 1.58    acc:43.50\n",
      "epoch 972\n",
      "loss: 1.37    acc:49.20\n",
      "loss: 1.57    acc:44.50\n",
      "epoch 973\n",
      "loss: 1.38    acc:48.10\n",
      "loss: 1.58    acc:42.20\n",
      "epoch 974\n",
      "loss: 1.39    acc:47.47\n",
      "loss: 1.60    acc:42.70\n",
      "epoch 975\n",
      "loss: 1.39    acc:48.02\n",
      "loss: 1.58    acc:43.80\n",
      "epoch 976\n",
      "loss: 1.37    acc:47.92\n",
      "loss: 1.53    acc:42.20\n",
      "epoch 977\n",
      "loss: 1.38    acc:49.30\n",
      "loss: 1.56    acc:46.00\n",
      "epoch 978\n",
      "loss: 1.37    acc:49.87\n",
      "loss: 1.51    acc:45.70\n",
      "epoch 979\n",
      "loss: 1.37    acc:48.47\n",
      "loss: 1.57    acc:44.90\n",
      "epoch 980\n",
      "loss: 1.39    acc:48.07\n",
      "loss: 1.58    acc:43.10\n",
      "epoch 981\n",
      "loss: 1.37    acc:49.32\n",
      "loss: 1.62    acc:42.70\n",
      "epoch 982\n",
      "loss: 1.37    acc:48.75\n",
      "loss: 1.59    acc:42.20\n",
      "epoch 983\n",
      "loss: 1.37    acc:47.75\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.57    acc:43.80\n",
      "epoch 984\n",
      "loss: 1.37    acc:48.77\n",
      "loss: 1.62    acc:43.80\n",
      "epoch 985\n",
      "loss: 1.39    acc:48.97\n",
      "loss: 1.60    acc:40.80\n",
      "epoch 986\n",
      "loss: 1.37    acc:48.17\n",
      "loss: 1.62    acc:42.50\n",
      "epoch 987\n",
      "loss: 1.37    acc:49.17\n",
      "loss: 1.57    acc:44.40\n",
      "epoch 988\n",
      "loss: 1.37    acc:49.07\n",
      "loss: 1.51    acc:46.00\n",
      "epoch 989\n",
      "loss: 1.36    acc:49.20\n",
      "loss: 1.57    acc:43.70\n",
      "epoch 990\n",
      "loss: 1.37    acc:48.55\n",
      "loss: 1.60    acc:42.90\n",
      "epoch 991\n",
      "loss: 1.39    acc:48.00\n",
      "loss: 1.60    acc:43.10\n",
      "epoch 992\n",
      "loss: 1.37    acc:48.65\n",
      "loss: 1.55    acc:44.60\n",
      "epoch 993\n",
      "loss: 1.39    acc:47.80\n",
      "loss: 1.60    acc:41.50\n",
      "epoch 994\n",
      "loss: 1.36    acc:49.70\n",
      "loss: 1.59    acc:42.60\n",
      "epoch 995\n",
      "loss: 1.38    acc:48.27\n",
      "loss: 1.66    acc:41.60\n",
      "epoch 996\n",
      "loss: 1.37    acc:49.42\n",
      "loss: 1.55    acc:44.80\n",
      "epoch 997\n",
      "loss: 1.36    acc:49.50\n",
      "loss: 1.54    acc:43.80\n",
      "epoch 998\n",
      "loss: 1.37    acc:48.65\n",
      "loss: 1.54    acc:42.80\n",
      "epoch 999\n",
      "loss: 1.36    acc:48.90\n",
      "loss: 1.60    acc:44.30\n",
      "epoch 1000\n",
      "loss: 1.38    acc:48.05\n",
      "loss: 1.52    acc:43.60\n",
      "16\n",
      "40394\n",
      "epoch 1\n",
      "loss: 2.26    acc:16.12\n",
      "loss: 2.27    acc:10.30\n",
      "epoch 2\n",
      "loss: 2.17    acc:20.65\n",
      "loss: 2.13    acc:23.80\n",
      "epoch 3\n",
      "loss: 2.09    acc:24.62\n",
      "loss: 2.06    acc:26.10\n",
      "epoch 4\n",
      "loss: 2.01    acc:26.27\n",
      "loss: 1.96    acc:27.20\n",
      "epoch 5\n",
      "loss: 1.95    acc:27.85\n",
      "loss: 1.95    acc:30.20\n",
      "epoch 6\n",
      "loss: 1.90    acc:28.85\n",
      "loss: 1.89    acc:29.80\n",
      "epoch 7\n",
      "loss: 1.84    acc:32.60\n",
      "loss: 1.84    acc:30.40\n",
      "epoch 8\n",
      "loss: 1.80    acc:33.85\n",
      "loss: 1.75    acc:37.20\n",
      "epoch 9\n",
      "loss: 1.75    acc:35.30\n",
      "loss: 1.69    acc:36.90\n",
      "epoch 10\n",
      "loss: 1.71    acc:36.85\n",
      "loss: 1.66    acc:39.40\n",
      "epoch 11\n",
      "loss: 1.68    acc:38.32\n",
      "loss: 1.69    acc:35.30\n",
      "epoch 12\n",
      "loss: 1.65    acc:39.02\n",
      "loss: 1.61    acc:38.70\n",
      "epoch 13\n",
      "loss: 1.63    acc:39.65\n",
      "loss: 1.61    acc:38.00\n",
      "epoch 14\n",
      "loss: 1.60    acc:41.02\n",
      "loss: 1.63    acc:37.70\n",
      "epoch 15\n",
      "loss: 1.57    acc:42.20\n",
      "loss: 1.64    acc:39.00\n",
      "epoch 16\n",
      "loss: 1.55    acc:43.62\n",
      "loss: 1.53    acc:41.40\n",
      "epoch 17\n",
      "loss: 1.53    acc:43.40\n",
      "loss: 1.54    acc:44.20\n",
      "epoch 18\n",
      "loss: 1.52    acc:44.42\n",
      "loss: 1.54    acc:43.00\n",
      "epoch 19\n",
      "loss: 1.49    acc:45.92\n",
      "loss: 1.55    acc:39.40\n",
      "epoch 20\n",
      "loss: 1.48    acc:46.30\n",
      "loss: 1.51    acc:43.20\n",
      "epoch 21\n",
      "loss: 1.47    acc:46.45\n",
      "loss: 1.50    acc:42.60\n",
      "epoch 22\n",
      "loss: 1.44    acc:48.00\n",
      "loss: 1.45    acc:46.20\n",
      "epoch 23\n",
      "loss: 1.43    acc:48.42\n",
      "loss: 1.42    acc:47.10\n",
      "epoch 24\n",
      "loss: 1.41    acc:49.37\n",
      "loss: 1.50    acc:44.40\n",
      "epoch 25\n",
      "loss: 1.39    acc:49.87\n",
      "loss: 1.44    acc:47.00\n",
      "epoch 26\n",
      "loss: 1.38    acc:50.22\n",
      "loss: 1.41    acc:47.70\n",
      "epoch 27\n",
      "loss: 1.37    acc:50.82\n",
      "loss: 1.40    acc:50.60\n",
      "epoch 28\n",
      "loss: 1.35    acc:51.17\n",
      "loss: 1.38    acc:50.40\n",
      "epoch 29\n",
      "loss: 1.33    acc:51.87\n",
      "loss: 1.36    acc:48.90\n",
      "epoch 30\n",
      "loss: 1.33    acc:52.17\n",
      "loss: 1.44    acc:45.60\n",
      "epoch 31\n",
      "loss: 1.30    acc:53.20\n",
      "loss: 1.39    acc:49.80\n",
      "epoch 32\n",
      "loss: 1.29    acc:53.77\n",
      "loss: 1.42    acc:47.00\n",
      "epoch 33\n",
      "loss: 1.27    acc:53.92\n",
      "loss: 1.35    acc:50.00\n",
      "epoch 34\n",
      "loss: 1.28    acc:54.10\n",
      "loss: 1.29    acc:53.10\n",
      "epoch 35\n",
      "loss: 1.25    acc:55.57\n",
      "loss: 1.35    acc:51.30\n",
      "epoch 36\n",
      "loss: 1.24    acc:55.67\n",
      "loss: 1.34    acc:50.00\n",
      "epoch 37\n",
      "loss: 1.23    acc:55.82\n",
      "loss: 1.32    acc:48.80\n",
      "epoch 38\n",
      "loss: 1.21    acc:56.90\n",
      "loss: 1.28    acc:52.00\n",
      "epoch 39\n",
      "loss: 1.21    acc:56.77\n",
      "loss: 1.27    acc:51.30\n",
      "epoch 40\n",
      "loss: 1.20    acc:56.82\n",
      "loss: 1.32    acc:50.10\n",
      "epoch 41\n",
      "loss: 1.18    acc:57.65\n",
      "loss: 1.28    acc:55.80\n",
      "epoch 42\n",
      "loss: 1.19    acc:57.30\n",
      "loss: 1.20    acc:56.90\n",
      "epoch 43\n",
      "loss: 1.17    acc:58.65\n",
      "loss: 1.28    acc:52.40\n",
      "epoch 44\n",
      "loss: 1.16    acc:58.25\n",
      "loss: 1.25    acc:55.60\n",
      "epoch 45\n",
      "loss: 1.14    acc:58.92\n",
      "loss: 1.36    acc:49.10\n",
      "epoch 46\n",
      "loss: 1.14    acc:58.97\n",
      "loss: 1.23    acc:53.80\n",
      "epoch 47\n",
      "loss: 1.12    acc:60.00\n",
      "loss: 1.35    acc:50.20\n",
      "epoch 48\n",
      "loss: 1.12    acc:59.32\n",
      "loss: 1.41    acc:49.00\n",
      "epoch 49\n",
      "loss: 1.11    acc:60.40\n",
      "loss: 1.27    acc:55.60\n",
      "epoch 50\n",
      "loss: 1.11    acc:60.52\n",
      "loss: 1.16    acc:58.40\n",
      "epoch 51\n",
      "loss: 1.09    acc:61.55\n",
      "loss: 1.25    acc:54.20\n",
      "epoch 52\n",
      "loss: 1.09    acc:60.52\n",
      "loss: 1.29    acc:52.90\n",
      "epoch 53\n",
      "loss: 1.08    acc:61.30\n",
      "loss: 1.14    acc:59.20\n",
      "epoch 54\n",
      "loss: 1.07    acc:61.12\n",
      "loss: 1.15    acc:56.40\n",
      "epoch 55\n",
      "loss: 1.06    acc:62.42\n",
      "loss: 1.29    acc:54.30\n",
      "epoch 56\n",
      "loss: 1.07    acc:61.27\n",
      "loss: 1.30    acc:54.30\n",
      "epoch 57\n",
      "loss: 1.06    acc:62.00\n",
      "loss: 1.24    acc:55.10\n",
      "epoch 58\n",
      "loss: 1.04    acc:63.20\n",
      "loss: 1.23    acc:54.70\n",
      "epoch 59\n",
      "loss: 1.05    acc:62.40\n",
      "loss: 1.35    acc:50.80\n",
      "epoch 60\n",
      "loss: 1.03    acc:63.02\n",
      "loss: 1.21    acc:57.40\n",
      "epoch 61\n",
      "loss: 1.03    acc:63.22\n",
      "loss: 1.26    acc:53.90\n",
      "epoch 62\n",
      "loss: 1.03    acc:63.15\n",
      "loss: 1.28    acc:53.60\n",
      "epoch 63\n",
      "loss: 1.04    acc:62.60\n",
      "loss: 1.27    acc:54.30\n",
      "epoch 64\n",
      "loss: 1.01    acc:63.97\n",
      "loss: 1.11    acc:58.40\n",
      "epoch 65\n",
      "loss: 1.01    acc:63.62\n",
      "loss: 1.25    acc:54.40\n",
      "epoch 66\n",
      "loss: 1.00    acc:63.92\n",
      "loss: 1.23    acc:55.60\n",
      "epoch 67\n",
      "loss: 1.01    acc:63.90\n",
      "loss: 1.16    acc:58.10\n",
      "epoch 68\n",
      "loss: 1.00    acc:63.75\n",
      "loss: 1.15    acc:58.50\n",
      "epoch 69\n",
      "loss: 0.98    acc:64.65\n",
      "loss: 1.15    acc:57.50\n",
      "epoch 70\n",
      "loss: 0.99    acc:64.72\n",
      "loss: 1.20    acc:55.60\n",
      "epoch 71\n",
      "loss: 0.98    acc:64.57\n",
      "loss: 1.12    acc:60.30\n",
      "epoch 72\n",
      "loss: 0.97    acc:64.60\n",
      "loss: 1.17    acc:59.00\n",
      "epoch 73\n",
      "loss: 0.95    acc:66.35\n",
      "loss: 1.20    acc:56.20\n",
      "epoch 74\n",
      "loss: 0.96    acc:65.37\n",
      "loss: 1.21    acc:56.60\n",
      "epoch 75\n",
      "loss: 0.97    acc:64.62\n",
      "loss: 1.09    acc:60.80\n",
      "epoch 76\n",
      "loss: 0.95    acc:66.22\n",
      "loss: 1.18    acc:59.10\n",
      "epoch 77\n",
      "loss: 0.95    acc:65.75\n",
      "loss: 1.22    acc:57.00\n",
      "epoch 78\n",
      "loss: 0.94    acc:66.40\n",
      "loss: 1.11    acc:60.50\n",
      "epoch 79\n",
      "loss: 0.94    acc:66.70\n",
      "loss: 1.25    acc:56.00\n",
      "epoch 80\n",
      "loss: 0.94    acc:65.95\n",
      "loss: 1.08    acc:62.00\n",
      "epoch 81\n",
      "loss: 0.94    acc:66.92\n",
      "loss: 1.20    acc:58.20\n",
      "epoch 82\n",
      "loss: 0.92    acc:66.45\n",
      "loss: 1.09    acc:62.40\n",
      "epoch 83\n",
      "loss: 0.90    acc:68.02\n",
      "loss: 1.10    acc:60.10\n",
      "epoch 84\n",
      "loss: 0.90    acc:67.27\n",
      "loss: 1.14    acc:60.70\n",
      "epoch 85\n",
      "loss: 0.90    acc:67.70\n",
      "loss: 1.18    acc:59.00\n",
      "epoch 86\n",
      "loss: 0.89    acc:68.12\n",
      "loss: 1.07    acc:61.90\n",
      "epoch 87\n",
      "loss: 0.91    acc:66.92\n",
      "loss: 1.23    acc:59.10\n",
      "epoch 88\n",
      "loss: 0.88    acc:68.72\n",
      "loss: 1.11    acc:61.00\n",
      "epoch 89\n",
      "loss: 0.89    acc:67.47\n",
      "loss: 1.09    acc:61.00\n",
      "epoch 90\n",
      "loss: 0.90    acc:67.60\n",
      "loss: 1.22    acc:56.60\n",
      "epoch 91\n",
      "loss: 0.88    acc:69.12\n",
      "loss: 1.19    acc:56.80\n",
      "epoch 92\n",
      "loss: 0.87    acc:68.70\n",
      "loss: 1.16    acc:57.30\n",
      "epoch 93\n",
      "loss: 0.86    acc:69.32\n",
      "loss: 1.10    acc:59.80\n",
      "epoch 94\n",
      "loss: 0.88    acc:68.30\n",
      "loss: 1.20    acc:58.70\n",
      "epoch 95\n",
      "loss: 0.87    acc:69.40\n",
      "loss: 1.22    acc:59.40\n",
      "epoch 96\n",
      "loss: 0.86    acc:68.95\n",
      "loss: 1.20    acc:59.50\n",
      "epoch 97\n",
      "loss: 0.86    acc:68.60\n",
      "loss: 1.15    acc:59.60\n",
      "epoch 98\n",
      "loss: 0.85    acc:68.90\n",
      "loss: 1.20    acc:57.90\n",
      "epoch 99\n",
      "loss: 0.85    acc:69.30\n",
      "loss: 1.31    acc:56.10\n",
      "epoch 100\n",
      "loss: 0.85    acc:69.45\n",
      "loss: 1.06    acc:62.70\n",
      "epoch 101\n",
      "loss: 0.84    acc:70.30\n",
      "loss: 1.19    acc:59.40\n",
      "epoch 102\n",
      "loss: 0.82    acc:70.47\n",
      "loss: 1.09    acc:61.70\n",
      "epoch 103\n",
      "loss: 0.84    acc:69.47\n",
      "loss: 1.10    acc:61.80\n",
      "epoch 104\n",
      "loss: 0.84    acc:69.25\n",
      "loss: 1.14    acc:58.60\n",
      "epoch 105\n",
      "loss: 0.84    acc:69.80\n",
      "loss: 1.02    acc:65.80\n",
      "epoch 106\n",
      "loss: 0.83    acc:70.47\n",
      "loss: 1.20    acc:58.50\n",
      "epoch 107\n",
      "loss: 0.82    acc:70.55\n",
      "loss: 1.17    acc:60.20\n",
      "epoch 108\n",
      "loss: 0.82    acc:70.00\n",
      "loss: 1.15    acc:61.70\n",
      "epoch 109\n",
      "loss: 0.82    acc:70.67\n",
      "loss: 1.12    acc:61.20\n",
      "epoch 110\n",
      "loss: 0.83    acc:70.22\n",
      "loss: 1.13    acc:60.00\n",
      "epoch 111\n",
      "loss: 0.80    acc:71.42\n",
      "loss: 1.06    acc:62.60\n",
      "epoch 112\n",
      "loss: 0.79    acc:71.32\n",
      "loss: 1.10    acc:61.10\n",
      "epoch 113\n",
      "loss: 0.79    acc:71.15\n",
      "loss: 1.28    acc:57.00\n",
      "epoch 114\n",
      "loss: 0.81    acc:71.00\n",
      "loss: 1.10    acc:64.20\n",
      "epoch 115\n",
      "loss: 0.81    acc:70.97\n",
      "loss: 1.16    acc:59.70\n",
      "epoch 116\n",
      "loss: 0.79    acc:71.90\n",
      "loss: 1.03    acc:63.70\n",
      "epoch 117\n",
      "loss: 0.78    acc:72.42\n",
      "loss: 1.14    acc:60.00\n",
      "epoch 118\n",
      "loss: 0.77    acc:72.42\n",
      "loss: 1.09    acc:61.10\n",
      "epoch 119\n",
      "loss: 0.78    acc:71.70\n",
      "loss: 1.07    acc:62.20\n",
      "epoch 120\n",
      "loss: 0.78    acc:71.32\n",
      "loss: 1.11    acc:59.30\n",
      "epoch 121\n",
      "loss: 0.77    acc:71.45\n",
      "loss: 1.07    acc:62.40\n",
      "epoch 122\n",
      "loss: 0.78    acc:72.22\n",
      "loss: 1.11    acc:60.90\n",
      "epoch 123\n",
      "loss: 0.76    acc:72.05\n",
      "loss: 1.14    acc:60.30\n",
      "epoch 124\n",
      "loss: 0.76    acc:72.90\n",
      "loss: 1.01    acc:65.20\n",
      "epoch 125\n",
      "loss: 0.76    acc:72.40\n",
      "loss: 1.19    acc:59.70\n",
      "epoch 126\n",
      "loss: 0.76    acc:73.25\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.06    acc:65.20\n",
      "epoch 127\n",
      "loss: 0.75    acc:73.27\n",
      "loss: 1.24    acc:60.20\n",
      "epoch 128\n",
      "loss: 0.76    acc:73.07\n",
      "loss: 1.21    acc:61.40\n",
      "epoch 129\n",
      "loss: 0.75    acc:73.30\n",
      "loss: 1.12    acc:60.00\n",
      "epoch 130\n",
      "loss: 0.74    acc:74.42\n",
      "loss: 1.14    acc:61.00\n",
      "epoch 131\n",
      "loss: 0.74    acc:72.90\n",
      "loss: 1.08    acc:62.50\n",
      "epoch 132\n",
      "loss: 0.73    acc:73.85\n",
      "loss: 1.18    acc:59.90\n",
      "epoch 133\n",
      "loss: 0.74    acc:73.50\n",
      "loss: 1.19    acc:61.00\n",
      "epoch 134\n",
      "loss: 0.73    acc:74.45\n",
      "loss: 1.06    acc:62.10\n",
      "epoch 135\n",
      "loss: 0.74    acc:73.62\n",
      "loss: 1.09    acc:60.40\n",
      "epoch 136\n",
      "loss: 0.73    acc:73.47\n",
      "loss: 1.08    acc:63.60\n",
      "epoch 137\n",
      "loss: 0.73    acc:74.25\n",
      "loss: 1.15    acc:60.40\n",
      "epoch 138\n",
      "loss: 0.72    acc:74.72\n",
      "loss: 1.13    acc:59.90\n",
      "epoch 139\n",
      "loss: 0.70    acc:74.97\n",
      "loss: 1.10    acc:61.40\n",
      "epoch 140\n",
      "loss: 0.72    acc:73.65\n",
      "loss: 1.16    acc:62.70\n",
      "epoch 141\n",
      "loss: 0.72    acc:74.22\n",
      "loss: 1.18    acc:61.80\n",
      "epoch 142\n",
      "loss: 0.71    acc:74.55\n",
      "loss: 1.24    acc:60.00\n",
      "epoch 143\n",
      "loss: 0.70    acc:74.92\n",
      "loss: 1.06    acc:63.40\n",
      "epoch 144\n",
      "loss: 0.69    acc:76.12\n",
      "loss: 1.11    acc:62.50\n",
      "epoch 145\n",
      "loss: 0.71    acc:74.10\n",
      "loss: 1.13    acc:62.90\n",
      "epoch 146\n",
      "loss: 0.72    acc:74.60\n",
      "loss: 1.17    acc:60.90\n",
      "epoch 147\n",
      "loss: 0.71    acc:74.52\n",
      "loss: 1.20    acc:59.70\n",
      "epoch 148\n",
      "loss: 0.68    acc:75.20\n",
      "loss: 1.10    acc:63.20\n",
      "epoch 149\n",
      "loss: 0.69    acc:75.20\n",
      "loss: 1.03    acc:64.80\n",
      "epoch 150\n",
      "loss: 0.70    acc:74.40\n",
      "loss: 1.12    acc:64.60\n",
      "epoch 151\n",
      "loss: 0.68    acc:75.95\n",
      "loss: 1.17    acc:59.80\n",
      "epoch 152\n",
      "loss: 0.68    acc:75.67\n",
      "loss: 1.15    acc:62.80\n",
      "epoch 153\n",
      "loss: 0.69    acc:75.57\n",
      "loss: 1.13    acc:62.40\n",
      "epoch 154\n",
      "loss: 0.69    acc:75.05\n",
      "loss: 1.20    acc:59.90\n",
      "epoch 155\n",
      "loss: 0.69    acc:74.60\n",
      "loss: 1.05    acc:65.00\n",
      "epoch 156\n",
      "loss: 0.68    acc:75.25\n",
      "loss: 1.23    acc:60.80\n",
      "epoch 157\n",
      "loss: 0.68    acc:75.65\n",
      "loss: 1.10    acc:63.20\n",
      "epoch 158\n",
      "loss: 0.68    acc:76.27\n",
      "loss: 1.16    acc:62.70\n",
      "epoch 159\n",
      "loss: 0.66    acc:76.10\n",
      "loss: 1.25    acc:60.90\n",
      "epoch 160\n",
      "loss: 0.68    acc:76.17\n",
      "loss: 1.14    acc:61.70\n",
      "epoch 161\n",
      "loss: 0.66    acc:76.15\n",
      "loss: 1.05    acc:65.70\n",
      "epoch 162\n",
      "loss: 0.67    acc:75.10\n",
      "loss: 1.21    acc:60.90\n",
      "epoch 163\n",
      "loss: 0.66    acc:77.30\n",
      "loss: 1.16    acc:62.90\n",
      "epoch 164\n",
      "loss: 0.65    acc:76.77\n",
      "loss: 1.12    acc:63.10\n",
      "epoch 165\n",
      "loss: 0.66    acc:76.65\n",
      "loss: 1.21    acc:60.40\n",
      "epoch 166\n",
      "loss: 0.67    acc:75.95\n",
      "loss: 1.11    acc:62.10\n",
      "epoch 167\n",
      "loss: 0.65    acc:76.35\n",
      "loss: 1.11    acc:65.00\n",
      "epoch 168\n",
      "loss: 0.66    acc:76.15\n",
      "loss: 1.12    acc:63.50\n",
      "epoch 169\n",
      "loss: 0.64    acc:77.32\n",
      "loss: 1.08    acc:61.70\n",
      "epoch 170\n",
      "loss: 0.65    acc:76.50\n",
      "loss: 1.18    acc:61.50\n",
      "epoch 171\n",
      "loss: 0.66    acc:76.15\n",
      "loss: 1.08    acc:62.60\n",
      "epoch 172\n",
      "loss: 0.64    acc:77.12\n",
      "loss: 1.05    acc:65.00\n",
      "epoch 173\n",
      "loss: 0.63    acc:77.60\n",
      "loss: 1.20    acc:61.80\n",
      "epoch 174\n",
      "loss: 0.63    acc:77.20\n",
      "loss: 1.11    acc:64.30\n",
      "epoch 175\n",
      "loss: 0.64    acc:77.55\n",
      "loss: 1.12    acc:64.10\n",
      "epoch 176\n",
      "loss: 0.65    acc:76.62\n",
      "loss: 1.02    acc:66.80\n",
      "epoch 177\n",
      "loss: 0.63    acc:78.00\n",
      "loss: 1.11    acc:63.80\n",
      "epoch 178\n",
      "loss: 0.63    acc:77.80\n",
      "loss: 1.21    acc:62.70\n",
      "epoch 179\n",
      "loss: 0.62    acc:78.32\n",
      "loss: 1.04    acc:64.70\n",
      "epoch 180\n",
      "loss: 0.62    acc:77.67\n",
      "loss: 0.95    acc:67.60\n",
      "epoch 181\n",
      "loss: 0.64    acc:75.87\n",
      "loss: 1.16    acc:61.90\n",
      "epoch 182\n",
      "loss: 0.61    acc:78.32\n",
      "loss: 1.06    acc:64.40\n",
      "epoch 183\n",
      "loss: 0.61    acc:78.25\n",
      "loss: 1.08    acc:62.90\n",
      "epoch 184\n",
      "loss: 0.60    acc:77.85\n",
      "loss: 1.17    acc:62.30\n",
      "epoch 185\n",
      "loss: 0.62    acc:77.97\n",
      "loss: 1.13    acc:63.80\n",
      "epoch 186\n",
      "loss: 0.62    acc:78.00\n",
      "loss: 1.32    acc:59.60\n",
      "epoch 187\n",
      "loss: 0.61    acc:77.85\n",
      "loss: 1.16    acc:62.20\n",
      "epoch 188\n",
      "loss: 0.62    acc:77.97\n",
      "loss: 1.09    acc:63.70\n",
      "epoch 189\n",
      "loss: 0.61    acc:78.45\n",
      "loss: 1.04    acc:64.80\n",
      "epoch 190\n",
      "loss: 0.61    acc:77.72\n",
      "loss: 1.05    acc:64.40\n",
      "epoch 191\n",
      "loss: 0.60    acc:78.17\n",
      "loss: 1.22    acc:60.20\n",
      "epoch 192\n",
      "loss: 0.60    acc:77.85\n",
      "loss: 1.24    acc:61.20\n",
      "epoch 193\n",
      "loss: 0.62    acc:77.60\n",
      "loss: 1.03    acc:66.10\n",
      "epoch 194\n",
      "loss: 0.60    acc:77.95\n",
      "loss: 1.11    acc:64.90\n",
      "epoch 195\n",
      "loss: 0.59    acc:79.00\n",
      "loss: 1.09    acc:64.30\n",
      "epoch 196\n",
      "loss: 0.60    acc:78.57\n",
      "loss: 1.22    acc:61.10\n",
      "epoch 197\n",
      "loss: 0.59    acc:78.87\n",
      "loss: 1.21    acc:62.70\n",
      "epoch 198\n",
      "loss: 0.60    acc:78.12\n",
      "loss: 1.15    acc:63.60\n",
      "epoch 199\n",
      "loss: 0.58    acc:78.70\n",
      "loss: 1.22    acc:61.40\n",
      "epoch 200\n",
      "loss: 0.57    acc:79.25\n",
      "loss: 1.00    acc:66.20\n",
      "epoch 201\n",
      "loss: 0.58    acc:79.25\n",
      "loss: 1.20    acc:60.30\n",
      "epoch 202\n",
      "loss: 0.59    acc:78.95\n",
      "loss: 1.09    acc:62.60\n",
      "epoch 203\n",
      "loss: 0.60    acc:77.97\n",
      "loss: 1.08    acc:63.70\n",
      "epoch 204\n",
      "loss: 0.58    acc:79.25\n",
      "loss: 1.10    acc:63.70\n",
      "epoch 205\n",
      "loss: 0.58    acc:79.00\n",
      "loss: 1.17    acc:63.60\n",
      "epoch 206\n",
      "loss: 0.58    acc:78.77\n",
      "loss: 1.13    acc:63.90\n",
      "epoch 207\n",
      "loss: 0.58    acc:78.97\n",
      "loss: 1.10    acc:65.80\n",
      "epoch 208\n",
      "loss: 0.58    acc:78.52\n",
      "loss: 1.11    acc:64.60\n",
      "epoch 209\n",
      "loss: 0.59    acc:78.72\n",
      "loss: 1.19    acc:62.90\n",
      "epoch 210\n",
      "loss: 0.58    acc:78.57\n",
      "loss: 1.04    acc:66.70\n",
      "epoch 211\n",
      "loss: 0.57    acc:79.40\n",
      "loss: 1.10    acc:65.80\n",
      "epoch 212\n",
      "loss: 0.57    acc:78.42\n",
      "loss: 1.05    acc:66.10\n",
      "epoch 213\n",
      "loss: 0.59    acc:78.75\n",
      "loss: 1.11    acc:64.30\n",
      "epoch 214\n",
      "loss: 0.56    acc:79.52\n",
      "loss: 1.07    acc:65.00\n",
      "epoch 215\n",
      "loss: 0.57    acc:78.90\n",
      "loss: 1.07    acc:66.20\n",
      "epoch 216\n",
      "loss: 0.57    acc:79.05\n",
      "loss: 1.12    acc:64.50\n",
      "epoch 217\n",
      "loss: 0.57    acc:79.55\n",
      "loss: 1.20    acc:62.70\n",
      "epoch 218\n",
      "loss: 0.56    acc:79.85\n",
      "loss: 1.13    acc:63.80\n",
      "epoch 219\n",
      "loss: 0.56    acc:79.60\n",
      "loss: 1.11    acc:65.60\n",
      "epoch 220\n",
      "loss: 0.56    acc:79.42\n",
      "loss: 1.18    acc:63.70\n",
      "epoch 221\n",
      "loss: 0.57    acc:79.25\n",
      "loss: 1.23    acc:61.40\n",
      "epoch 222\n",
      "loss: 0.56    acc:79.77\n",
      "loss: 1.22    acc:62.80\n",
      "epoch 223\n",
      "loss: 0.55    acc:80.40\n",
      "loss: 1.16    acc:64.60\n",
      "epoch 224\n",
      "loss: 0.57    acc:79.62\n",
      "loss: 1.12    acc:63.70\n",
      "epoch 225\n",
      "loss: 0.55    acc:79.65\n",
      "loss: 1.15    acc:63.00\n",
      "epoch 226\n",
      "loss: 0.55    acc:78.77\n",
      "loss: 1.16    acc:64.20\n",
      "epoch 227\n",
      "loss: 0.55    acc:79.92\n",
      "loss: 1.03    acc:65.80\n",
      "epoch 228\n",
      "loss: 0.56    acc:79.67\n",
      "loss: 1.16    acc:62.40\n",
      "epoch 229\n",
      "loss: 0.55    acc:80.22\n",
      "loss: 1.22    acc:62.20\n",
      "epoch 230\n",
      "loss: 0.55    acc:79.72\n",
      "loss: 1.15    acc:63.90\n",
      "epoch 231\n",
      "loss: 0.54    acc:80.25\n",
      "loss: 1.08    acc:65.10\n",
      "epoch 232\n",
      "loss: 0.56    acc:79.77\n",
      "loss: 1.31    acc:63.00\n",
      "epoch 233\n",
      "loss: 0.55    acc:79.67\n",
      "loss: 1.27    acc:60.30\n",
      "epoch 234\n",
      "loss: 0.53    acc:81.12\n",
      "loss: 1.07    acc:66.40\n",
      "epoch 235\n",
      "loss: 0.55    acc:80.07\n",
      "loss: 1.19    acc:62.10\n",
      "epoch 236\n",
      "loss: 0.55    acc:80.67\n",
      "loss: 1.21    acc:61.80\n",
      "epoch 237\n",
      "loss: 0.52    acc:82.05\n",
      "loss: 1.11    acc:64.30\n",
      "epoch 238\n",
      "loss: 0.54    acc:79.90\n",
      "loss: 1.17    acc:64.40\n",
      "epoch 239\n",
      "loss: 0.55    acc:79.82\n",
      "loss: 1.18    acc:62.40\n",
      "epoch 240\n",
      "loss: 0.52    acc:81.30\n",
      "loss: 1.10    acc:65.30\n",
      "epoch 241\n",
      "loss: 0.51    acc:81.62\n",
      "loss: 1.23    acc:62.40\n",
      "epoch 242\n",
      "loss: 0.53    acc:79.95\n",
      "loss: 1.21    acc:64.10\n",
      "epoch 243\n",
      "loss: 0.53    acc:80.52\n",
      "loss: 1.12    acc:66.60\n",
      "epoch 244\n",
      "loss: 0.52    acc:81.65\n",
      "loss: 1.12    acc:64.40\n",
      "epoch 245\n",
      "loss: 0.50    acc:82.12\n",
      "loss: 1.22    acc:63.90\n",
      "epoch 246\n",
      "loss: 0.50    acc:81.75\n",
      "loss: 1.11    acc:66.00\n",
      "epoch 247\n",
      "loss: 0.50    acc:82.47\n",
      "loss: 1.15    acc:65.50\n",
      "epoch 248\n",
      "loss: 0.52    acc:81.25\n",
      "loss: 1.24    acc:61.90\n",
      "epoch 249\n",
      "loss: 0.54    acc:80.82\n",
      "loss: 1.06    acc:65.40\n",
      "epoch 250\n",
      "loss: 0.51    acc:81.62\n",
      "loss: 1.15    acc:66.50\n",
      "epoch 251\n",
      "loss: 0.53    acc:81.00\n",
      "loss: 1.16    acc:65.60\n",
      "epoch 252\n",
      "loss: 0.51    acc:81.20\n",
      "loss: 1.22    acc:62.50\n",
      "epoch 253\n",
      "loss: 0.51    acc:80.97\n",
      "loss: 1.14    acc:66.10\n",
      "epoch 254\n",
      "loss: 0.50    acc:82.25\n",
      "loss: 1.09    acc:66.00\n",
      "epoch 255\n",
      "loss: 0.50    acc:81.47\n",
      "loss: 1.16    acc:64.10\n",
      "epoch 256\n",
      "loss: 0.51    acc:81.32\n",
      "loss: 1.18    acc:65.50\n",
      "epoch 257\n",
      "loss: 0.52    acc:81.52\n",
      "loss: 1.29    acc:64.30\n",
      "epoch 258\n",
      "loss: 0.52    acc:80.10\n",
      "loss: 1.28    acc:61.50\n",
      "epoch 259\n",
      "loss: 0.50    acc:81.70\n",
      "loss: 1.24    acc:63.30\n",
      "epoch 260\n",
      "loss: 0.52    acc:81.27\n",
      "loss: 1.04    acc:67.10\n",
      "epoch 261\n",
      "loss: 0.51    acc:81.17\n",
      "loss: 1.07    acc:65.70\n",
      "epoch 262\n",
      "loss: 0.49    acc:81.62\n",
      "loss: 1.32    acc:61.10\n",
      "epoch 263\n",
      "loss: 0.51    acc:81.37\n",
      "loss: 1.26    acc:61.90\n",
      "epoch 264\n",
      "loss: 0.50    acc:81.95\n",
      "loss: 1.32    acc:61.10\n",
      "epoch 265\n",
      "loss: 0.50    acc:81.80\n",
      "loss: 1.23    acc:64.30\n",
      "epoch 266\n",
      "loss: 0.52    acc:81.25\n",
      "loss: 1.07    acc:65.20\n",
      "epoch 267\n",
      "loss: 0.50    acc:81.70\n",
      "loss: 1.07    acc:66.00\n",
      "epoch 268\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.50    acc:82.42\n",
      "loss: 1.23    acc:64.60\n",
      "epoch 269\n",
      "loss: 0.50    acc:81.47\n",
      "loss: 1.10    acc:66.20\n",
      "epoch 270\n",
      "loss: 0.49    acc:82.27\n",
      "loss: 1.20    acc:63.70\n",
      "epoch 271\n",
      "loss: 0.49    acc:82.32\n",
      "loss: 1.09    acc:64.80\n",
      "epoch 272\n",
      "loss: 0.49    acc:81.80\n",
      "loss: 1.01    acc:69.40\n",
      "epoch 273\n",
      "loss: 0.49    acc:81.82\n",
      "loss: 1.10    acc:65.90\n",
      "epoch 274\n",
      "loss: 0.49    acc:81.45\n",
      "loss: 1.18    acc:64.20\n",
      "epoch 275\n",
      "loss: 0.49    acc:82.45\n",
      "loss: 1.19    acc:63.80\n",
      "epoch 276\n",
      "loss: 0.48    acc:82.67\n",
      "loss: 1.16    acc:64.80\n",
      "epoch 277\n",
      "loss: 0.48    acc:82.77\n",
      "loss: 1.26    acc:63.20\n",
      "epoch 278\n",
      "loss: 0.49    acc:81.95\n",
      "loss: 1.13    acc:65.20\n",
      "epoch 279\n",
      "loss: 0.48    acc:82.77\n",
      "loss: 1.08    acc:67.40\n",
      "epoch 280\n",
      "loss: 0.48    acc:82.70\n",
      "loss: 1.22    acc:64.30\n",
      "epoch 281\n",
      "loss: 0.48    acc:82.87\n",
      "loss: 1.15    acc:63.10\n",
      "epoch 282\n",
      "loss: 0.47    acc:82.92\n",
      "loss: 1.17    acc:64.90\n",
      "epoch 283\n",
      "loss: 0.49    acc:81.82\n",
      "loss: 1.29    acc:63.30\n",
      "epoch 284\n",
      "loss: 0.47    acc:82.62\n",
      "loss: 1.22    acc:65.20\n",
      "epoch 285\n",
      "loss: 0.48    acc:82.55\n",
      "loss: 1.21    acc:63.50\n",
      "epoch 286\n",
      "loss: 0.50    acc:81.25\n",
      "loss: 1.08    acc:67.00\n",
      "epoch 287\n",
      "loss: 0.48    acc:82.57\n",
      "loss: 1.16    acc:64.40\n",
      "epoch 288\n",
      "loss: 0.48    acc:83.15\n",
      "loss: 1.21    acc:65.00\n",
      "epoch 289\n",
      "loss: 0.47    acc:82.92\n",
      "loss: 1.16    acc:63.50\n",
      "epoch 290\n",
      "loss: 0.48    acc:81.90\n",
      "loss: 1.07    acc:67.10\n",
      "epoch 291\n",
      "loss: 0.46    acc:82.82\n",
      "loss: 1.05    acc:67.50\n",
      "epoch 292\n",
      "loss: 0.47    acc:82.37\n",
      "loss: 1.06    acc:67.90\n",
      "epoch 293\n",
      "loss: 0.46    acc:83.02\n",
      "loss: 1.24    acc:62.70\n",
      "epoch 294\n",
      "loss: 0.48    acc:82.62\n",
      "loss: 1.12    acc:65.90\n",
      "epoch 295\n",
      "loss: 0.47    acc:82.60\n",
      "loss: 1.20    acc:64.80\n",
      "epoch 296\n",
      "loss: 0.45    acc:84.10\n",
      "loss: 1.15    acc:64.40\n",
      "epoch 297\n",
      "loss: 0.47    acc:83.00\n",
      "loss: 1.24    acc:66.40\n",
      "epoch 298\n",
      "loss: 0.46    acc:83.17\n",
      "loss: 1.27    acc:63.90\n",
      "epoch 299\n",
      "loss: 0.44    acc:84.10\n",
      "loss: 1.19    acc:64.50\n",
      "epoch 300\n",
      "loss: 0.45    acc:83.65\n",
      "loss: 1.26    acc:60.40\n",
      "epoch 301\n",
      "loss: 0.45    acc:83.55\n",
      "loss: 1.09    acc:65.70\n",
      "epoch 302\n",
      "loss: 0.47    acc:82.90\n",
      "loss: 1.04    acc:66.40\n",
      "epoch 303\n",
      "loss: 0.45    acc:83.42\n",
      "loss: 1.14    acc:66.70\n",
      "epoch 304\n",
      "loss: 0.44    acc:84.07\n",
      "loss: 1.17    acc:66.00\n",
      "epoch 305\n",
      "loss: 0.46    acc:82.97\n",
      "loss: 1.18    acc:64.80\n",
      "epoch 306\n",
      "loss: 0.47    acc:83.07\n",
      "loss: 1.30    acc:62.80\n",
      "epoch 307\n",
      "loss: 0.45    acc:83.72\n",
      "loss: 1.31    acc:60.30\n",
      "epoch 308\n",
      "loss: 0.45    acc:83.55\n",
      "loss: 1.14    acc:63.30\n",
      "epoch 309\n",
      "loss: 0.47    acc:82.90\n",
      "loss: 1.14    acc:67.80\n",
      "epoch 310\n",
      "loss: 0.45    acc:83.97\n",
      "loss: 1.11    acc:65.70\n",
      "epoch 311\n",
      "loss: 0.46    acc:82.95\n",
      "loss: 1.22    acc:66.00\n",
      "epoch 312\n",
      "loss: 0.44    acc:83.62\n",
      "loss: 1.20    acc:64.60\n",
      "epoch 313\n",
      "loss: 0.45    acc:83.52\n",
      "loss: 1.22    acc:65.20\n",
      "epoch 314\n",
      "loss: 0.46    acc:83.02\n",
      "loss: 1.03    acc:69.00\n",
      "epoch 315\n",
      "loss: 0.44    acc:83.72\n",
      "loss: 1.15    acc:65.50\n",
      "epoch 316\n",
      "loss: 0.44    acc:84.02\n",
      "loss: 1.17    acc:66.80\n",
      "epoch 317\n",
      "loss: 0.46    acc:83.67\n",
      "loss: 1.33    acc:63.70\n",
      "epoch 318\n",
      "loss: 0.46    acc:83.45\n",
      "loss: 1.25    acc:65.30\n",
      "epoch 319\n",
      "loss: 0.44    acc:84.12\n",
      "loss: 1.37    acc:60.60\n",
      "epoch 320\n",
      "loss: 0.44    acc:83.82\n",
      "loss: 1.19    acc:65.50\n",
      "epoch 321\n",
      "loss: 0.45    acc:83.15\n",
      "loss: 1.18    acc:64.40\n",
      "epoch 322\n",
      "loss: 0.44    acc:83.40\n",
      "loss: 1.16    acc:68.00\n",
      "epoch 323\n",
      "loss: 0.48    acc:82.05\n",
      "loss: 1.19    acc:65.50\n",
      "epoch 324\n",
      "loss: 0.47    acc:82.80\n",
      "loss: 1.13    acc:67.10\n",
      "epoch 325\n",
      "loss: 0.44    acc:83.87\n",
      "loss: 1.17    acc:65.90\n",
      "epoch 326\n",
      "loss: 0.45    acc:84.30\n",
      "loss: 1.14    acc:66.70\n",
      "epoch 327\n",
      "loss: 0.45    acc:83.40\n",
      "loss: 1.22    acc:64.10\n",
      "epoch 328\n",
      "loss: 0.45    acc:84.25\n",
      "loss: 1.14    acc:67.50\n",
      "epoch 329\n",
      "loss: 0.46    acc:84.15\n",
      "loss: 1.18    acc:66.10\n",
      "epoch 330\n",
      "loss: 0.45    acc:84.37\n",
      "loss: 1.15    acc:66.80\n",
      "epoch 331\n",
      "loss: 0.41    acc:85.02\n",
      "loss: 1.30    acc:63.50\n",
      "epoch 332\n",
      "loss: 0.43    acc:84.50\n",
      "loss: 1.16    acc:65.10\n",
      "epoch 333\n",
      "loss: 0.43    acc:84.42\n",
      "loss: 1.31    acc:62.80\n",
      "epoch 334\n",
      "loss: 0.43    acc:84.12\n",
      "loss: 1.11    acc:66.20\n",
      "epoch 335\n",
      "loss: 0.41    acc:85.07\n",
      "loss: 1.13    acc:65.20\n",
      "epoch 336\n",
      "loss: 0.43    acc:84.70\n",
      "loss: 1.21    acc:67.30\n",
      "epoch 337\n",
      "loss: 0.45    acc:83.60\n",
      "loss: 1.18    acc:65.70\n",
      "epoch 338\n",
      "loss: 0.43    acc:84.12\n",
      "loss: 1.15    acc:65.80\n",
      "epoch 339\n",
      "loss: 0.46    acc:83.72\n",
      "loss: 1.30    acc:62.50\n",
      "epoch 340\n",
      "loss: 0.42    acc:85.30\n",
      "loss: 1.12    acc:68.20\n",
      "epoch 341\n",
      "loss: 0.42    acc:84.22\n",
      "loss: 1.24    acc:66.30\n",
      "epoch 342\n",
      "loss: 0.45    acc:83.05\n",
      "loss: 1.23    acc:64.90\n",
      "epoch 343\n",
      "loss: 0.46    acc:83.67\n",
      "loss: 1.16    acc:67.90\n",
      "epoch 344\n",
      "loss: 0.43    acc:84.30\n",
      "loss: 1.15    acc:65.80\n",
      "epoch 345\n",
      "loss: 0.42    acc:85.00\n",
      "loss: 1.22    acc:64.40\n",
      "epoch 346\n",
      "loss: 0.43    acc:84.57\n",
      "loss: 1.24    acc:63.20\n",
      "epoch 347\n",
      "loss: 0.43    acc:84.77\n",
      "loss: 1.18    acc:63.70\n",
      "epoch 348\n",
      "loss: 0.42    acc:84.97\n",
      "loss: 1.31    acc:63.50\n",
      "epoch 349\n",
      "loss: 0.43    acc:83.95\n",
      "loss: 1.29    acc:64.60\n",
      "epoch 350\n",
      "loss: 0.41    acc:85.00\n",
      "loss: 1.21    acc:65.60\n",
      "epoch 351\n",
      "loss: 0.42    acc:85.10\n",
      "loss: 1.20    acc:66.40\n",
      "epoch 352\n",
      "loss: 0.42    acc:84.22\n",
      "loss: 1.36    acc:62.80\n",
      "epoch 353\n",
      "loss: 0.41    acc:85.17\n",
      "loss: 1.11    acc:68.20\n",
      "epoch 354\n",
      "loss: 0.43    acc:84.47\n",
      "loss: 1.18    acc:66.10\n",
      "epoch 355\n",
      "loss: 0.41    acc:85.47\n",
      "loss: 1.26    acc:65.90\n",
      "epoch 356\n",
      "loss: 0.38    acc:87.00\n",
      "loss: 1.26    acc:65.70\n",
      "epoch 357\n",
      "loss: 0.43    acc:85.00\n",
      "loss: 1.18    acc:66.50\n",
      "epoch 358\n",
      "loss: 0.41    acc:85.42\n",
      "loss: 1.15    acc:65.20\n",
      "epoch 359\n",
      "loss: 0.41    acc:84.97\n",
      "loss: 1.21    acc:64.00\n",
      "epoch 360\n",
      "loss: 0.42    acc:84.52\n",
      "loss: 1.24    acc:64.60\n",
      "epoch 361\n",
      "loss: 0.42    acc:85.17\n",
      "loss: 1.18    acc:65.90\n",
      "epoch 362\n",
      "loss: 0.41    acc:85.72\n",
      "loss: 1.19    acc:65.90\n",
      "epoch 363\n",
      "loss: 0.39    acc:86.22\n",
      "loss: 1.31    acc:62.00\n",
      "epoch 364\n",
      "loss: 0.40    acc:85.67\n",
      "loss: 1.12    acc:68.60\n",
      "epoch 365\n",
      "loss: 0.41    acc:85.32\n",
      "loss: 1.30    acc:64.40\n",
      "epoch 366\n",
      "loss: 0.41    acc:85.67\n",
      "loss: 1.20    acc:65.20\n",
      "epoch 367\n",
      "loss: 0.40    acc:86.27\n",
      "loss: 1.37    acc:64.90\n",
      "epoch 368\n",
      "loss: 0.41    acc:84.85\n",
      "loss: 1.27    acc:62.60\n",
      "epoch 369\n",
      "loss: 0.41    acc:85.50\n",
      "loss: 1.27    acc:64.20\n",
      "epoch 370\n",
      "loss: 0.42    acc:84.92\n",
      "loss: 1.20    acc:65.80\n",
      "epoch 371\n",
      "loss: 0.42    acc:84.75\n",
      "loss: 1.43    acc:60.90\n",
      "epoch 372\n",
      "loss: 0.41    acc:84.77\n",
      "loss: 1.25    acc:65.10\n",
      "epoch 373\n",
      "loss: 0.38    acc:86.07\n",
      "loss: 1.19    acc:65.00\n",
      "epoch 374\n",
      "loss: 0.39    acc:86.17\n",
      "loss: 1.34    acc:62.90\n",
      "epoch 375\n",
      "loss: 0.42    acc:85.22\n",
      "loss: 1.23    acc:65.40\n",
      "epoch 376\n",
      "loss: 0.40    acc:85.52\n",
      "loss: 1.17    acc:66.40\n",
      "epoch 377\n",
      "loss: 0.42    acc:84.00\n",
      "loss: 1.27    acc:63.60\n",
      "epoch 378\n",
      "loss: 0.41    acc:84.85\n",
      "loss: 1.30    acc:63.50\n",
      "epoch 379\n",
      "loss: 0.38    acc:85.92\n",
      "loss: 1.17    acc:66.20\n",
      "epoch 380\n",
      "loss: 0.40    acc:85.32\n",
      "loss: 1.13    acc:66.20\n",
      "epoch 381\n",
      "loss: 0.41    acc:85.25\n",
      "loss: 1.30    acc:65.20\n",
      "epoch 382\n",
      "loss: 0.42    acc:84.72\n",
      "loss: 1.24    acc:64.90\n",
      "epoch 383\n",
      "loss: 0.38    acc:86.15\n",
      "loss: 1.26    acc:63.40\n",
      "epoch 384\n",
      "loss: 0.40    acc:85.17\n",
      "loss: 1.21    acc:65.90\n",
      "epoch 385\n",
      "loss: 0.41    acc:85.00\n",
      "loss: 1.10    acc:68.50\n",
      "epoch 386\n",
      "loss: 0.39    acc:86.10\n",
      "loss: 1.29    acc:66.20\n",
      "epoch 387\n",
      "loss: 0.42    acc:85.07\n",
      "loss: 1.22    acc:66.70\n",
      "epoch 388\n",
      "loss: 0.40    acc:85.55\n",
      "loss: 1.22    acc:64.90\n",
      "epoch 389\n",
      "loss: 0.40    acc:85.55\n",
      "loss: 1.30    acc:64.60\n",
      "epoch 390\n",
      "loss: 0.40    acc:85.70\n",
      "loss: 1.25    acc:65.70\n",
      "epoch 391\n",
      "loss: 0.39    acc:86.45\n",
      "loss: 1.18    acc:67.00\n",
      "epoch 392\n",
      "loss: 0.40    acc:85.62\n",
      "loss: 1.24    acc:67.10\n",
      "epoch 393\n",
      "loss: 0.39    acc:85.62\n",
      "loss: 1.30    acc:66.70\n",
      "epoch 394\n",
      "loss: 0.41    acc:84.92\n",
      "loss: 1.30    acc:65.10\n",
      "epoch 395\n",
      "loss: 0.40    acc:85.35\n",
      "loss: 1.28    acc:65.30\n",
      "epoch 396\n",
      "loss: 0.39    acc:85.40\n",
      "loss: 1.33    acc:62.10\n",
      "epoch 397\n",
      "loss: 0.37    acc:87.17\n",
      "loss: 1.16    acc:66.60\n",
      "epoch 398\n",
      "loss: 0.38    acc:86.87\n",
      "loss: 1.23    acc:67.50\n",
      "epoch 399\n",
      "loss: 0.41    acc:84.75\n",
      "loss: 1.18    acc:65.60\n",
      "epoch 400\n",
      "loss: 0.38    acc:86.20\n",
      "loss: 1.31    acc:63.80\n",
      "epoch 401\n",
      "loss: 0.39    acc:85.82\n",
      "loss: 1.39    acc:63.00\n",
      "epoch 402\n",
      "loss: 0.39    acc:85.55\n",
      "loss: 1.28    acc:63.70\n",
      "epoch 403\n",
      "loss: 0.37    acc:86.47\n",
      "loss: 1.16    acc:66.10\n",
      "epoch 404\n",
      "loss: 0.38    acc:86.50\n",
      "loss: 1.30    acc:63.60\n",
      "epoch 405\n",
      "loss: 0.38    acc:86.65\n",
      "loss: 1.27    acc:66.00\n",
      "epoch 406\n",
      "loss: 0.38    acc:86.25\n",
      "loss: 1.28    acc:65.60\n",
      "epoch 407\n",
      "loss: 0.36    acc:87.20\n",
      "loss: 1.28    acc:66.80\n",
      "epoch 408\n",
      "loss: 0.39    acc:85.67\n",
      "loss: 1.37    acc:64.30\n",
      "epoch 409\n",
      "loss: 0.38    acc:86.37\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.20    acc:66.30\n",
      "epoch 410\n",
      "loss: 0.38    acc:85.67\n",
      "loss: 1.12    acc:67.60\n",
      "epoch 411\n",
      "loss: 0.39    acc:85.62\n",
      "loss: 1.23    acc:66.70\n",
      "epoch 412\n",
      "loss: 0.39    acc:85.62\n",
      "loss: 1.32    acc:64.10\n",
      "epoch 413\n",
      "loss: 0.37    acc:87.02\n",
      "loss: 1.33    acc:65.00\n",
      "epoch 414\n",
      "loss: 0.40    acc:84.80\n",
      "loss: 1.21    acc:66.40\n",
      "epoch 415\n",
      "loss: 0.38    acc:86.37\n",
      "loss: 1.34    acc:64.20\n",
      "epoch 416\n",
      "loss: 0.38    acc:86.50\n",
      "loss: 1.39    acc:63.50\n",
      "epoch 417\n",
      "loss: 0.38    acc:86.07\n",
      "loss: 1.31    acc:64.50\n",
      "epoch 418\n",
      "loss: 0.38    acc:85.82\n",
      "loss: 1.31    acc:63.80\n",
      "epoch 419\n",
      "loss: 0.38    acc:85.80\n",
      "loss: 1.28    acc:64.50\n",
      "epoch 420\n",
      "loss: 0.37    acc:86.35\n",
      "loss: 1.31    acc:66.20\n",
      "epoch 421\n",
      "loss: 0.41    acc:85.97\n",
      "loss: 1.33    acc:62.60\n",
      "epoch 422\n",
      "loss: 0.39    acc:85.87\n",
      "loss: 1.26    acc:65.10\n",
      "epoch 423\n",
      "loss: 0.39    acc:85.65\n",
      "loss: 1.18    acc:67.40\n",
      "epoch 424\n",
      "loss: 0.37    acc:86.80\n",
      "loss: 1.26    acc:63.70\n",
      "epoch 425\n",
      "loss: 0.39    acc:86.02\n",
      "loss: 1.33    acc:64.00\n",
      "epoch 426\n",
      "loss: 0.37    acc:86.55\n",
      "loss: 1.31    acc:66.70\n",
      "epoch 427\n",
      "loss: 0.37    acc:85.72\n",
      "loss: 1.20    acc:67.20\n",
      "epoch 428\n",
      "loss: 0.35    acc:88.05\n",
      "loss: 1.23    acc:67.40\n",
      "epoch 429\n",
      "loss: 0.35    acc:86.97\n",
      "loss: 1.28    acc:65.10\n",
      "epoch 430\n",
      "loss: 0.35    acc:87.37\n",
      "loss: 1.30    acc:66.00\n",
      "epoch 431\n",
      "loss: 0.37    acc:86.65\n",
      "loss: 1.16    acc:67.30\n",
      "epoch 432\n",
      "loss: 0.39    acc:86.22\n",
      "loss: 1.35    acc:64.60\n",
      "epoch 433\n",
      "loss: 0.38    acc:86.20\n",
      "loss: 1.23    acc:66.90\n",
      "epoch 434\n",
      "loss: 0.36    acc:87.30\n",
      "loss: 1.19    acc:67.20\n",
      "epoch 435\n",
      "loss: 0.36    acc:86.92\n",
      "loss: 1.32    acc:62.70\n",
      "epoch 436\n",
      "loss: 0.35    acc:87.17\n",
      "loss: 1.28    acc:66.60\n",
      "epoch 437\n",
      "loss: 0.36    acc:87.37\n",
      "loss: 1.25    acc:64.10\n",
      "epoch 438\n",
      "loss: 0.35    acc:87.55\n",
      "loss: 1.33    acc:65.20\n",
      "epoch 439\n",
      "loss: 0.37    acc:86.82\n",
      "loss: 1.15    acc:68.60\n",
      "epoch 440\n",
      "loss: 0.36    acc:87.32\n",
      "loss: 1.26    acc:66.10\n",
      "epoch 441\n",
      "loss: 0.37    acc:87.37\n",
      "loss: 1.41    acc:62.30\n",
      "epoch 442\n",
      "loss: 0.37    acc:86.52\n",
      "loss: 1.31    acc:66.10\n",
      "epoch 443\n",
      "loss: 0.37    acc:86.20\n",
      "loss: 1.40    acc:63.60\n",
      "epoch 444\n",
      "loss: 0.36    acc:87.62\n",
      "loss: 1.17    acc:66.80\n",
      "epoch 445\n",
      "loss: 0.35    acc:87.60\n",
      "loss: 1.33    acc:63.60\n",
      "epoch 446\n",
      "loss: 0.34    acc:87.77\n",
      "loss: 1.21    acc:66.80\n",
      "epoch 447\n",
      "loss: 0.34    acc:87.95\n",
      "loss: 1.31    acc:66.30\n",
      "epoch 448\n",
      "loss: 0.34    acc:87.50\n",
      "loss: 1.20    acc:66.60\n",
      "epoch 449\n",
      "loss: 0.37    acc:86.42\n",
      "loss: 1.33    acc:66.10\n",
      "epoch 450\n",
      "loss: 0.37    acc:87.17\n",
      "loss: 1.23    acc:67.10\n",
      "epoch 451\n",
      "loss: 0.34    acc:87.60\n",
      "loss: 1.43    acc:64.50\n",
      "epoch 452\n",
      "loss: 0.36    acc:86.85\n",
      "loss: 1.26    acc:65.30\n",
      "epoch 453\n",
      "loss: 0.35    acc:87.37\n",
      "loss: 1.27    acc:63.90\n",
      "epoch 454\n",
      "loss: 0.34    acc:87.52\n",
      "loss: 1.30    acc:64.60\n",
      "epoch 455\n",
      "loss: 0.36    acc:86.62\n",
      "loss: 1.29    acc:65.70\n",
      "epoch 456\n",
      "loss: 0.37    acc:86.90\n",
      "loss: 1.30    acc:66.40\n",
      "epoch 457\n",
      "loss: 0.36    acc:86.90\n",
      "loss: 1.30    acc:66.50\n",
      "epoch 458\n",
      "loss: 0.36    acc:86.65\n",
      "loss: 1.34    acc:65.60\n",
      "epoch 459\n",
      "loss: 0.36    acc:86.77\n",
      "loss: 1.20    acc:66.40\n",
      "epoch 460\n",
      "loss: 0.36    acc:87.27\n",
      "loss: 1.37    acc:64.10\n",
      "epoch 461\n",
      "loss: 0.35    acc:87.45\n",
      "loss: 1.26    acc:65.20\n",
      "epoch 462\n",
      "loss: 0.35    acc:87.02\n",
      "loss: 1.25    acc:66.60\n",
      "epoch 463\n",
      "loss: 0.34    acc:88.15\n",
      "loss: 1.37    acc:64.90\n",
      "epoch 464\n",
      "loss: 0.37    acc:87.00\n",
      "loss: 1.28    acc:66.10\n",
      "epoch 465\n",
      "loss: 0.35    acc:87.60\n",
      "loss: 1.27    acc:65.60\n",
      "epoch 466\n",
      "loss: 0.35    acc:87.05\n",
      "loss: 1.31    acc:66.20\n",
      "epoch 467\n",
      "loss: 0.38    acc:86.40\n",
      "loss: 1.33    acc:65.30\n",
      "epoch 468\n",
      "loss: 0.36    acc:86.97\n",
      "loss: 1.30    acc:65.10\n",
      "epoch 469\n",
      "loss: 0.36    acc:87.00\n",
      "loss: 1.32    acc:62.60\n",
      "epoch 470\n",
      "loss: 0.35    acc:87.52\n",
      "loss: 1.34    acc:67.10\n",
      "epoch 471\n",
      "loss: 0.34    acc:88.05\n",
      "loss: 1.40    acc:63.90\n",
      "epoch 472\n",
      "loss: 0.35    acc:87.35\n",
      "loss: 1.22    acc:65.50\n",
      "epoch 473\n",
      "loss: 0.35    acc:87.97\n",
      "loss: 1.37    acc:64.00\n",
      "epoch 474\n",
      "loss: 0.36    acc:86.45\n",
      "loss: 1.21    acc:67.80\n",
      "epoch 475\n",
      "loss: 0.35    acc:87.75\n",
      "loss: 1.32    acc:65.50\n",
      "epoch 476\n",
      "loss: 0.35    acc:87.35\n",
      "loss: 1.19    acc:67.70\n",
      "epoch 477\n",
      "loss: 0.35    acc:86.80\n",
      "loss: 1.35    acc:64.10\n",
      "epoch 478\n",
      "loss: 0.36    acc:86.47\n",
      "loss: 1.34    acc:63.90\n",
      "epoch 479\n",
      "loss: 0.35    acc:87.27\n",
      "loss: 1.25    acc:64.90\n",
      "epoch 480\n",
      "loss: 0.35    acc:87.22\n",
      "loss: 1.28    acc:67.30\n",
      "epoch 481\n",
      "loss: 0.33    acc:88.00\n",
      "loss: 1.15    acc:66.90\n",
      "epoch 482\n",
      "loss: 0.36    acc:87.62\n",
      "loss: 1.33    acc:64.50\n",
      "epoch 483\n",
      "loss: 0.33    acc:88.07\n",
      "loss: 1.31    acc:64.70\n",
      "epoch 484\n",
      "loss: 0.34    acc:87.75\n",
      "loss: 1.31    acc:66.90\n",
      "epoch 485\n",
      "loss: 0.33    acc:88.50\n",
      "loss: 1.27    acc:64.60\n",
      "epoch 486\n",
      "loss: 0.33    acc:88.40\n",
      "loss: 1.16    acc:66.70\n",
      "epoch 487\n",
      "loss: 0.34    acc:88.00\n",
      "loss: 1.24    acc:66.30\n",
      "epoch 488\n",
      "loss: 0.33    acc:88.22\n",
      "loss: 1.25    acc:66.10\n",
      "epoch 489\n",
      "loss: 0.35    acc:87.27\n",
      "loss: 1.37    acc:65.00\n",
      "epoch 490\n",
      "loss: 0.35    acc:87.20\n",
      "loss: 1.23    acc:67.10\n",
      "epoch 491\n",
      "loss: 0.33    acc:87.87\n",
      "loss: 1.16    acc:68.40\n",
      "epoch 492\n",
      "loss: 0.33    acc:88.42\n",
      "loss: 1.30    acc:65.70\n",
      "epoch 493\n",
      "loss: 0.31    acc:88.95\n",
      "loss: 1.29    acc:65.50\n",
      "epoch 494\n",
      "loss: 0.33    acc:88.07\n",
      "loss: 1.32    acc:66.40\n",
      "epoch 495\n",
      "loss: 0.35    acc:86.82\n",
      "loss: 1.27    acc:67.10\n",
      "epoch 496\n",
      "loss: 0.33    acc:88.17\n",
      "loss: 1.18    acc:67.30\n",
      "epoch 497\n",
      "loss: 0.35    acc:86.85\n",
      "loss: 1.43    acc:65.00\n",
      "epoch 498\n",
      "loss: 0.34    acc:87.72\n",
      "loss: 1.34    acc:65.30\n",
      "epoch 499\n",
      "loss: 0.32    acc:88.25\n",
      "loss: 1.24    acc:66.40\n",
      "epoch 500\n",
      "loss: 0.33    acc:88.07\n",
      "loss: 1.33    acc:65.60\n",
      "epoch 501\n",
      "loss: 0.36    acc:86.87\n",
      "loss: 1.40    acc:65.60\n",
      "epoch 502\n",
      "loss: 0.34    acc:87.37\n",
      "loss: 1.34    acc:65.40\n",
      "epoch 503\n",
      "loss: 0.35    acc:87.05\n",
      "loss: 1.35    acc:64.50\n",
      "epoch 504\n",
      "loss: 0.34    acc:88.10\n",
      "loss: 1.35    acc:65.50\n",
      "epoch 505\n",
      "loss: 0.34    acc:88.15\n",
      "loss: 1.31    acc:65.60\n",
      "epoch 506\n",
      "loss: 0.33    acc:88.32\n",
      "loss: 1.22    acc:67.20\n",
      "epoch 507\n",
      "loss: 0.35    acc:87.15\n",
      "loss: 1.48    acc:63.60\n",
      "epoch 508\n",
      "loss: 0.33    acc:87.45\n",
      "loss: 1.24    acc:67.40\n",
      "epoch 509\n",
      "loss: 0.32    acc:88.42\n",
      "loss: 1.39    acc:65.50\n",
      "epoch 510\n",
      "loss: 0.33    acc:88.20\n",
      "loss: 1.24    acc:65.70\n",
      "epoch 511\n",
      "loss: 0.35    acc:87.45\n",
      "loss: 1.31    acc:64.60\n",
      "epoch 512\n",
      "loss: 0.34    acc:87.97\n",
      "loss: 1.33    acc:67.50\n",
      "epoch 513\n",
      "loss: 0.35    acc:88.15\n",
      "loss: 1.23    acc:69.10\n",
      "epoch 514\n",
      "loss: 0.31    acc:88.75\n",
      "loss: 1.28    acc:67.00\n",
      "Epoch   513: reducing learning rate of group 0 to 1.0000e-04.\n",
      "epoch 515\n",
      "loss: 0.31    acc:88.15\n",
      "loss: 1.23    acc:66.20\n",
      "epoch 516\n",
      "loss: 0.29    acc:89.45\n",
      "loss: 1.20    acc:66.60\n",
      "epoch 517\n",
      "loss: 0.27    acc:90.50\n",
      "loss: 1.28    acc:66.40\n",
      "epoch 518\n",
      "loss: 0.29    acc:89.87\n",
      "loss: 1.32    acc:64.90\n",
      "epoch 519\n",
      "loss: 0.28    acc:90.37\n",
      "loss: 1.30    acc:65.10\n",
      "epoch 520\n",
      "loss: 0.28    acc:90.20\n",
      "loss: 1.29    acc:67.00\n",
      "epoch 521\n",
      "loss: 0.28    acc:90.02\n",
      "loss: 1.24    acc:65.90\n",
      "epoch 522\n",
      "loss: 0.28    acc:90.80\n",
      "loss: 1.26    acc:65.80\n",
      "epoch 523\n",
      "loss: 0.27    acc:90.62\n",
      "loss: 1.24    acc:67.40\n",
      "epoch 524\n",
      "loss: 0.27    acc:90.07\n",
      "loss: 1.24    acc:67.00\n",
      "epoch 525\n",
      "loss: 0.29    acc:89.42\n",
      "loss: 1.27    acc:66.60\n",
      "epoch 526\n",
      "loss: 0.27    acc:90.92\n",
      "loss: 1.32    acc:67.20\n",
      "epoch 527\n",
      "loss: 0.26    acc:91.05\n",
      "loss: 1.09    acc:68.90\n",
      "epoch 528\n",
      "loss: 0.28    acc:90.12\n",
      "loss: 1.24    acc:68.10\n",
      "epoch 529\n",
      "loss: 0.26    acc:90.52\n",
      "loss: 1.22    acc:68.10\n",
      "epoch 530\n",
      "loss: 0.27    acc:90.60\n",
      "loss: 1.31    acc:66.40\n",
      "epoch 531\n",
      "loss: 0.28    acc:90.10\n",
      "loss: 1.24    acc:66.70\n",
      "epoch 532\n",
      "loss: 0.27    acc:91.17\n",
      "loss: 1.21    acc:68.90\n",
      "epoch 533\n",
      "loss: 0.27    acc:90.77\n",
      "loss: 1.18    acc:69.40\n",
      "epoch 534\n",
      "loss: 0.28    acc:90.12\n",
      "loss: 1.29    acc:66.00\n",
      "epoch 535\n",
      "loss: 0.26    acc:91.52\n",
      "loss: 1.27    acc:65.30\n",
      "epoch 536\n",
      "loss: 0.27    acc:90.17\n",
      "loss: 1.30    acc:66.90\n",
      "epoch 537\n",
      "loss: 0.27    acc:90.67\n",
      "loss: 1.23    acc:67.70\n",
      "epoch 538\n",
      "loss: 0.27    acc:90.62\n",
      "loss: 1.24    acc:68.80\n",
      "epoch 539\n",
      "loss: 0.28    acc:90.42\n",
      "loss: 1.25    acc:67.00\n",
      "epoch 540\n",
      "loss: 0.28    acc:90.30\n",
      "loss: 1.32    acc:67.00\n",
      "epoch 541\n",
      "loss: 0.28    acc:89.62\n",
      "loss: 1.29    acc:66.30\n",
      "epoch 542\n",
      "loss: 0.28    acc:89.57\n",
      "loss: 1.19    acc:68.40\n",
      "epoch 543\n",
      "loss: 0.27    acc:90.67\n",
      "loss: 1.43    acc:64.50\n",
      "epoch 544\n",
      "loss: 0.26    acc:91.12\n",
      "loss: 1.24    acc:67.30\n",
      "epoch 545\n",
      "loss: 0.24    acc:91.42\n",
      "loss: 1.33    acc:67.40\n",
      "epoch 546\n",
      "loss: 0.28    acc:90.07\n",
      "loss: 1.18    acc:68.90\n",
      "epoch 547\n",
      "loss: 0.26    acc:90.95\n",
      "loss: 1.36    acc:66.30\n",
      "epoch 548\n",
      "loss: 0.26    acc:90.67\n",
      "loss: 1.24    acc:67.70\n",
      "epoch 549\n",
      "loss: 0.27    acc:90.32\n",
      "loss: 1.24    acc:68.00\n",
      "epoch 550\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.27    acc:90.20\n",
      "loss: 1.32    acc:66.60\n",
      "epoch 551\n",
      "loss: 0.26    acc:90.45\n",
      "loss: 1.36    acc:64.90\n",
      "epoch 552\n",
      "loss: 0.27    acc:90.75\n",
      "loss: 1.24    acc:68.10\n",
      "epoch 553\n",
      "loss: 0.26    acc:90.50\n",
      "loss: 1.44    acc:63.00\n",
      "epoch 554\n",
      "loss: 0.26    acc:91.10\n",
      "loss: 1.30    acc:66.60\n",
      "epoch 555\n",
      "loss: 0.28    acc:89.90\n",
      "loss: 1.30    acc:65.10\n",
      "epoch 556\n",
      "loss: 0.27    acc:90.87\n",
      "loss: 1.25    acc:68.30\n",
      "epoch 557\n",
      "loss: 0.27    acc:90.40\n",
      "loss: 1.36    acc:67.50\n",
      "epoch 558\n",
      "loss: 0.28    acc:90.22\n",
      "loss: 1.28    acc:66.40\n",
      "epoch 559\n",
      "loss: 0.27    acc:89.92\n",
      "loss: 1.35    acc:66.10\n",
      "epoch 560\n",
      "loss: 0.26    acc:90.90\n",
      "loss: 1.15    acc:69.20\n",
      "epoch 561\n",
      "loss: 0.26    acc:90.92\n",
      "loss: 1.34    acc:66.70\n",
      "epoch 562\n",
      "loss: 0.26    acc:90.42\n",
      "loss: 1.30    acc:66.40\n",
      "epoch 563\n",
      "loss: 0.26    acc:90.87\n",
      "loss: 1.28    acc:66.00\n",
      "epoch 564\n",
      "loss: 0.28    acc:90.05\n",
      "loss: 1.28    acc:67.10\n",
      "epoch 565\n",
      "loss: 0.28    acc:89.85\n",
      "loss: 1.40    acc:65.90\n",
      "epoch 566\n",
      "loss: 0.29    acc:89.47\n",
      "loss: 1.27    acc:66.80\n",
      "Epoch   565: reducing learning rate of group 0 to 3.3333e-05.\n",
      "epoch 567\n",
      "loss: 0.26    acc:90.85\n",
      "loss: 1.38    acc:64.90\n",
      "epoch 568\n",
      "loss: 0.25    acc:91.20\n",
      "loss: 1.40    acc:65.00\n",
      "epoch 569\n",
      "loss: 0.26    acc:90.30\n",
      "loss: 1.23    acc:68.60\n",
      "epoch 570\n",
      "loss: 0.25    acc:90.97\n",
      "loss: 1.31    acc:65.70\n",
      "epoch 571\n",
      "loss: 0.24    acc:91.37\n",
      "loss: 1.32    acc:67.60\n",
      "epoch 572\n",
      "loss: 0.24    acc:90.95\n",
      "loss: 1.36    acc:65.50\n",
      "epoch 573\n",
      "loss: 0.24    acc:92.12\n",
      "loss: 1.30    acc:67.40\n",
      "epoch 574\n",
      "loss: 0.25    acc:91.67\n",
      "loss: 1.24    acc:67.20\n",
      "epoch 575\n",
      "loss: 0.24    acc:90.97\n",
      "loss: 1.26    acc:66.90\n",
      "epoch 576\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.33    acc:65.40\n",
      "epoch 577\n",
      "loss: 0.25    acc:91.37\n",
      "loss: 1.35    acc:66.10\n",
      "epoch 578\n",
      "loss: 0.25    acc:90.50\n",
      "loss: 1.32    acc:68.30\n",
      "epoch 579\n",
      "loss: 0.25    acc:90.92\n",
      "loss: 1.32    acc:66.00\n",
      "epoch 580\n",
      "loss: 0.25    acc:91.55\n",
      "loss: 1.22    acc:68.30\n",
      "epoch 581\n",
      "loss: 0.23    acc:92.02\n",
      "loss: 1.27    acc:66.60\n",
      "epoch 582\n",
      "loss: 0.26    acc:90.65\n",
      "loss: 1.30    acc:67.00\n",
      "epoch 583\n",
      "loss: 0.24    acc:91.17\n",
      "loss: 1.34    acc:68.10\n",
      "epoch 584\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.33    acc:68.00\n",
      "epoch 585\n",
      "loss: 0.25    acc:91.15\n",
      "loss: 1.22    acc:69.10\n",
      "epoch 586\n",
      "loss: 0.26    acc:90.67\n",
      "loss: 1.29    acc:66.70\n",
      "epoch 587\n",
      "loss: 0.23    acc:91.62\n",
      "loss: 1.30    acc:65.50\n",
      "epoch 588\n",
      "loss: 0.25    acc:91.50\n",
      "loss: 1.30    acc:68.00\n",
      "epoch 589\n",
      "loss: 0.25    acc:91.37\n",
      "loss: 1.30    acc:65.30\n",
      "epoch 590\n",
      "loss: 0.24    acc:91.40\n",
      "loss: 1.26    acc:66.00\n",
      "epoch 591\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.36    acc:67.20\n",
      "epoch 592\n",
      "loss: 0.25    acc:91.75\n",
      "loss: 1.23    acc:69.50\n",
      "epoch 593\n",
      "loss: 0.25    acc:91.50\n",
      "loss: 1.32    acc:65.90\n",
      "epoch 594\n",
      "loss: 0.24    acc:91.87\n",
      "loss: 1.31    acc:66.10\n",
      "epoch 595\n",
      "loss: 0.24    acc:91.95\n",
      "loss: 1.32    acc:66.70\n",
      "epoch 596\n",
      "loss: 0.25    acc:91.07\n",
      "loss: 1.35    acc:66.00\n",
      "epoch 597\n",
      "loss: 0.24    acc:91.40\n",
      "loss: 1.23    acc:68.20\n",
      "epoch 598\n",
      "loss: 0.24    acc:91.70\n",
      "loss: 1.29    acc:65.70\n",
      "epoch 599\n",
      "loss: 0.24    acc:91.40\n",
      "loss: 1.16    acc:66.60\n",
      "epoch 600\n",
      "loss: 0.25    acc:91.20\n",
      "loss: 1.24    acc:69.60\n",
      "epoch 601\n",
      "loss: 0.25    acc:90.77\n",
      "loss: 1.21    acc:68.80\n",
      "epoch 602\n",
      "loss: 0.26    acc:90.82\n",
      "loss: 1.39    acc:67.60\n",
      "epoch 603\n",
      "loss: 0.25    acc:90.92\n",
      "loss: 1.21    acc:69.30\n",
      "epoch 604\n",
      "loss: 0.25    acc:90.70\n",
      "loss: 1.28    acc:68.40\n",
      "epoch 605\n",
      "loss: 0.26    acc:90.82\n",
      "loss: 1.40    acc:65.00\n",
      "Epoch   604: reducing learning rate of group 0 to 1.1111e-05.\n",
      "epoch 606\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.32    acc:66.30\n",
      "epoch 607\n",
      "loss: 0.25    acc:91.25\n",
      "loss: 1.30    acc:67.20\n",
      "epoch 608\n",
      "loss: 0.24    acc:91.92\n",
      "loss: 1.19    acc:68.80\n",
      "epoch 609\n",
      "loss: 0.23    acc:91.45\n",
      "loss: 1.24    acc:67.20\n",
      "epoch 610\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.35    acc:65.70\n",
      "epoch 611\n",
      "loss: 0.24    acc:92.07\n",
      "loss: 1.29    acc:67.30\n",
      "epoch 612\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.29    acc:68.20\n",
      "epoch 613\n",
      "loss: 0.25    acc:91.00\n",
      "loss: 1.22    acc:68.20\n",
      "epoch 614\n",
      "loss: 0.26    acc:91.15\n",
      "loss: 1.28    acc:67.10\n",
      "epoch 615\n",
      "loss: 0.23    acc:92.00\n",
      "loss: 1.32    acc:67.20\n",
      "epoch 616\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.16    acc:69.70\n",
      "epoch 617\n",
      "loss: 0.25    acc:91.37\n",
      "loss: 1.14    acc:70.30\n",
      "epoch 618\n",
      "loss: 0.23    acc:92.02\n",
      "loss: 1.28    acc:66.30\n",
      "epoch 619\n",
      "loss: 0.25    acc:91.15\n",
      "loss: 1.21    acc:69.70\n",
      "epoch 620\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.34    acc:65.10\n",
      "epoch 621\n",
      "loss: 0.23    acc:92.45\n",
      "loss: 1.22    acc:67.60\n",
      "epoch 622\n",
      "loss: 0.24    acc:91.87\n",
      "loss: 1.32    acc:67.00\n",
      "epoch 623\n",
      "loss: 0.24    acc:90.85\n",
      "loss: 1.38    acc:65.50\n",
      "epoch 624\n",
      "loss: 0.24    acc:91.90\n",
      "loss: 1.21    acc:71.30\n",
      "epoch 625\n",
      "loss: 0.25    acc:91.55\n",
      "loss: 1.35    acc:65.90\n",
      "epoch 626\n",
      "loss: 0.24    acc:91.82\n",
      "loss: 1.19    acc:67.10\n",
      "epoch 627\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.22    acc:69.00\n",
      "epoch 628\n",
      "loss: 0.24    acc:91.97\n",
      "loss: 1.39    acc:66.60\n",
      "epoch 629\n",
      "loss: 0.24    acc:92.02\n",
      "loss: 1.31    acc:67.20\n",
      "epoch 630\n",
      "loss: 0.24    acc:91.92\n",
      "loss: 1.18    acc:68.90\n",
      "epoch 631\n",
      "loss: 0.24    acc:91.27\n",
      "loss: 1.20    acc:67.60\n",
      "epoch 632\n",
      "loss: 0.25    acc:91.02\n",
      "loss: 1.41    acc:65.40\n",
      "epoch 633\n",
      "loss: 0.25    acc:91.35\n",
      "loss: 1.19    acc:67.90\n",
      "epoch 634\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.37    acc:68.60\n",
      "epoch 635\n",
      "loss: 0.23    acc:92.27\n",
      "loss: 1.26    acc:68.00\n",
      "epoch 636\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.26    acc:68.40\n",
      "epoch 637\n",
      "loss: 0.24    acc:91.70\n",
      "loss: 1.36    acc:64.70\n",
      "epoch 638\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.37    acc:66.70\n",
      "epoch 639\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.39    acc:64.90\n",
      "epoch 640\n",
      "loss: 0.25    acc:91.47\n",
      "loss: 1.23    acc:70.00\n",
      "epoch 641\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.37    acc:66.00\n",
      "epoch 642\n",
      "loss: 0.23    acc:91.72\n",
      "loss: 1.29    acc:67.10\n",
      "epoch 643\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.27    acc:67.50\n",
      "epoch 644\n",
      "loss: 0.24    acc:91.67\n",
      "loss: 1.26    acc:68.60\n",
      "epoch 645\n",
      "loss: 0.25    acc:91.50\n",
      "loss: 1.26    acc:67.50\n",
      "epoch 646\n",
      "loss: 0.24    acc:91.12\n",
      "loss: 1.24    acc:67.70\n",
      "epoch 647\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.21    acc:70.40\n",
      "epoch 648\n",
      "loss: 0.25    acc:91.70\n",
      "loss: 1.24    acc:67.30\n",
      "epoch 649\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.31    acc:67.50\n",
      "epoch 650\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.21    acc:69.50\n",
      "epoch 651\n",
      "loss: 0.25    acc:91.50\n",
      "loss: 1.30    acc:67.50\n",
      "epoch 652\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.23    acc:69.50\n",
      "epoch 653\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.24    acc:68.50\n",
      "epoch 654\n",
      "loss: 0.25    acc:91.12\n",
      "loss: 1.33    acc:67.60\n",
      "epoch 655\n",
      "loss: 0.25    acc:91.62\n",
      "loss: 1.26    acc:66.90\n",
      "epoch 656\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.27    acc:67.80\n",
      "epoch 657\n",
      "loss: 0.23    acc:91.70\n",
      "loss: 1.32    acc:68.20\n",
      "epoch 658\n",
      "loss: 0.25    acc:90.97\n",
      "loss: 1.33    acc:66.60\n",
      "epoch 659\n",
      "loss: 0.24    acc:91.70\n",
      "loss: 1.26    acc:66.10\n",
      "epoch 660\n",
      "loss: 0.22    acc:91.80\n",
      "loss: 1.27    acc:67.50\n",
      "epoch 661\n",
      "loss: 0.23    acc:91.42\n",
      "loss: 1.30    acc:66.50\n",
      "epoch 662\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.14    acc:67.50\n",
      "epoch 663\n",
      "loss: 0.25    acc:91.60\n",
      "loss: 1.26    acc:70.00\n",
      "epoch 664\n",
      "loss: 0.25    acc:91.35\n",
      "loss: 1.25    acc:69.30\n",
      "epoch 665\n",
      "loss: 0.25    acc:90.92\n",
      "loss: 1.35    acc:67.20\n",
      "epoch 666\n",
      "loss: 0.25    acc:91.50\n",
      "loss: 1.34    acc:67.00\n",
      "epoch 667\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.31    acc:66.20\n",
      "epoch 668\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.25    acc:67.60\n",
      "epoch 669\n",
      "loss: 0.23    acc:92.45\n",
      "loss: 1.32    acc:65.00\n",
      "epoch 670\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.19    acc:68.70\n",
      "epoch 671\n",
      "loss: 0.24    acc:91.12\n",
      "loss: 1.41    acc:65.50\n",
      "epoch 672\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.18    acc:69.80\n",
      "epoch 673\n",
      "loss: 0.23    acc:92.22\n",
      "loss: 1.19    acc:69.90\n",
      "epoch 674\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.15    acc:69.40\n",
      "epoch 675\n",
      "loss: 0.22    acc:92.32\n",
      "loss: 1.32    acc:67.50\n",
      "epoch 676\n",
      "loss: 0.24    acc:92.02\n",
      "loss: 1.17    acc:69.10\n",
      "epoch 677\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.37    acc:65.60\n",
      "epoch 678\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.26    acc:67.20\n",
      "epoch 679\n",
      "loss: 0.25    acc:91.60\n",
      "loss: 1.24    acc:68.80\n",
      "epoch 680\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.39    acc:65.50\n",
      "epoch 681\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.29    acc:66.60\n",
      "Epoch   680: reducing learning rate of group 0 to 3.7037e-06.\n",
      "epoch 682\n",
      "loss: 0.24    acc:91.80\n",
      "loss: 1.36    acc:66.60\n",
      "epoch 683\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.24    acc:69.60\n",
      "epoch 684\n",
      "loss: 0.23    acc:91.75\n",
      "loss: 1.27    acc:66.10\n",
      "epoch 685\n",
      "loss: 0.24    acc:91.17\n",
      "loss: 1.24    acc:68.50\n",
      "epoch 686\n",
      "loss: 0.23    acc:91.70\n",
      "loss: 1.24    acc:67.50\n",
      "epoch 687\n",
      "loss: 0.24    acc:91.05\n",
      "loss: 1.23    acc:67.70\n",
      "epoch 688\n",
      "loss: 0.22    acc:92.27\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.24    acc:67.90\n",
      "epoch 689\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.33    acc:68.60\n",
      "epoch 690\n",
      "loss: 0.24    acc:91.77\n",
      "loss: 1.32    acc:67.10\n",
      "epoch 691\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.26    acc:66.80\n",
      "epoch 692\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.28    acc:65.30\n",
      "epoch 693\n",
      "loss: 0.24    acc:91.42\n",
      "loss: 1.34    acc:66.90\n",
      "epoch 694\n",
      "loss: 0.23    acc:92.47\n",
      "loss: 1.24    acc:69.60\n",
      "epoch 695\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.28    acc:69.30\n",
      "epoch 696\n",
      "loss: 0.24    acc:91.05\n",
      "loss: 1.29    acc:68.10\n",
      "epoch 697\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.20    acc:67.90\n",
      "epoch 698\n",
      "loss: 0.24    acc:91.17\n",
      "loss: 1.26    acc:67.90\n",
      "epoch 699\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.24    acc:68.70\n",
      "epoch 700\n",
      "loss: 0.22    acc:92.47\n",
      "loss: 1.25    acc:69.30\n",
      "epoch 701\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.34    acc:66.30\n",
      "epoch 702\n",
      "loss: 0.24    acc:92.30\n",
      "loss: 1.20    acc:68.90\n",
      "epoch 703\n",
      "loss: 0.21    acc:92.95\n",
      "loss: 1.26    acc:69.60\n",
      "epoch 704\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.26    acc:69.00\n",
      "epoch 705\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.21    acc:69.20\n",
      "epoch 706\n",
      "loss: 0.23    acc:91.57\n",
      "loss: 1.28    acc:67.20\n",
      "epoch 707\n",
      "loss: 0.22    acc:92.27\n",
      "loss: 1.36    acc:66.20\n",
      "epoch 708\n",
      "loss: 0.23    acc:91.27\n",
      "loss: 1.19    acc:69.70\n",
      "epoch 709\n",
      "loss: 0.23    acc:92.20\n",
      "loss: 1.21    acc:69.40\n",
      "epoch 710\n",
      "loss: 0.23    acc:92.40\n",
      "loss: 1.26    acc:68.10\n",
      "epoch 711\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.32    acc:66.80\n",
      "epoch 712\n",
      "loss: 0.23    acc:91.60\n",
      "loss: 1.26    acc:69.00\n",
      "epoch 713\n",
      "loss: 0.23    acc:92.35\n",
      "loss: 1.24    acc:69.30\n",
      "epoch 714\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.29    acc:66.80\n",
      "epoch 715\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.29    acc:68.90\n",
      "epoch 716\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.28    acc:67.50\n",
      "epoch 717\n",
      "loss: 0.23    acc:92.22\n",
      "loss: 1.22    acc:69.40\n",
      "epoch 718\n",
      "loss: 0.22    acc:92.72\n",
      "loss: 1.26    acc:68.50\n",
      "epoch 719\n",
      "loss: 0.22    acc:92.20\n",
      "loss: 1.40    acc:65.80\n",
      "epoch 720\n",
      "loss: 0.24    acc:91.97\n",
      "loss: 1.20    acc:68.20\n",
      "epoch 721\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.42    acc:65.50\n",
      "epoch 722\n",
      "loss: 0.22    acc:92.67\n",
      "loss: 1.38    acc:66.20\n",
      "epoch 723\n",
      "loss: 0.25    acc:91.07\n",
      "loss: 1.32    acc:67.10\n",
      "epoch 724\n",
      "loss: 0.23    acc:91.70\n",
      "loss: 1.31    acc:66.40\n",
      "Epoch   723: reducing learning rate of group 0 to 1.2346e-06.\n",
      "epoch 725\n",
      "loss: 0.22    acc:92.20\n",
      "loss: 1.19    acc:69.40\n",
      "epoch 726\n",
      "loss: 0.23    acc:92.00\n",
      "loss: 1.35    acc:64.70\n",
      "epoch 727\n",
      "loss: 0.23    acc:92.25\n",
      "loss: 1.30    acc:66.80\n",
      "epoch 728\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.16    acc:71.60\n",
      "epoch 729\n",
      "loss: 0.23    acc:91.72\n",
      "loss: 1.38    acc:65.10\n",
      "epoch 730\n",
      "loss: 0.23    acc:92.47\n",
      "loss: 1.30    acc:68.00\n",
      "epoch 731\n",
      "loss: 0.22    acc:92.27\n",
      "loss: 1.21    acc:68.40\n",
      "epoch 732\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.17    acc:69.50\n",
      "epoch 733\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.27    acc:68.20\n",
      "epoch 734\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.33    acc:66.80\n",
      "epoch 735\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.20    acc:68.60\n",
      "epoch 736\n",
      "loss: 0.22    acc:92.55\n",
      "loss: 1.33    acc:65.70\n",
      "epoch 737\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.25    acc:67.30\n",
      "epoch 738\n",
      "loss: 0.23    acc:91.65\n",
      "loss: 1.34    acc:67.60\n",
      "epoch 739\n",
      "loss: 0.24    acc:91.25\n",
      "loss: 1.36    acc:67.00\n",
      "epoch 740\n",
      "loss: 0.22    acc:93.00\n",
      "loss: 1.29    acc:67.00\n",
      "epoch 741\n",
      "loss: 0.23    acc:92.47\n",
      "loss: 1.27    acc:67.30\n",
      "epoch 742\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.31    acc:67.80\n",
      "epoch 743\n",
      "loss: 0.22    acc:92.45\n",
      "loss: 1.26    acc:68.70\n",
      "epoch 744\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.29    acc:69.40\n",
      "epoch 745\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.38    acc:66.10\n",
      "Epoch   744: reducing learning rate of group 0 to 4.1152e-07.\n",
      "epoch 746\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.10    acc:71.40\n",
      "epoch 747\n",
      "loss: 0.24    acc:91.57\n",
      "loss: 1.22    acc:68.00\n",
      "epoch 748\n",
      "loss: 0.24    acc:92.07\n",
      "loss: 1.29    acc:67.10\n",
      "epoch 749\n",
      "loss: 0.24    acc:92.02\n",
      "loss: 1.30    acc:66.90\n",
      "epoch 750\n",
      "loss: 0.21    acc:93.05\n",
      "loss: 1.20    acc:69.90\n",
      "epoch 751\n",
      "loss: 0.24    acc:91.85\n",
      "loss: 1.18    acc:69.70\n",
      "epoch 752\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.24    acc:68.20\n",
      "epoch 753\n",
      "loss: 0.22    acc:92.25\n",
      "loss: 1.33    acc:67.90\n",
      "epoch 754\n",
      "loss: 0.22    acc:92.32\n",
      "loss: 1.40    acc:65.30\n",
      "epoch 755\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.31    acc:65.80\n",
      "epoch 756\n",
      "loss: 0.23    acc:91.90\n",
      "loss: 1.15    acc:70.20\n",
      "epoch 757\n",
      "loss: 0.22    acc:92.65\n",
      "loss: 1.29    acc:67.00\n",
      "epoch 758\n",
      "loss: 0.23    acc:91.87\n",
      "loss: 1.25    acc:70.00\n",
      "epoch 759\n",
      "loss: 0.22    acc:92.07\n",
      "loss: 1.21    acc:67.30\n",
      "epoch 760\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.26    acc:68.20\n",
      "epoch 761\n",
      "loss: 0.23    acc:91.82\n",
      "loss: 1.35    acc:67.00\n",
      "epoch 762\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.16    acc:68.80\n",
      "epoch 763\n",
      "loss: 0.22    acc:92.42\n",
      "loss: 1.30    acc:68.50\n",
      "epoch 764\n",
      "loss: 0.22    acc:92.22\n",
      "loss: 1.31    acc:68.80\n",
      "epoch 765\n",
      "loss: 0.23    acc:92.25\n",
      "loss: 1.22    acc:70.20\n",
      "epoch 766\n",
      "loss: 0.23    acc:91.47\n",
      "loss: 1.31    acc:65.60\n",
      "epoch 767\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.40    acc:65.90\n",
      "epoch 768\n",
      "loss: 0.24    acc:91.62\n",
      "loss: 1.41    acc:64.50\n",
      "epoch 769\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.44    acc:65.60\n",
      "epoch 770\n",
      "loss: 0.22    acc:92.22\n",
      "loss: 1.18    acc:69.90\n",
      "epoch 771\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.21    acc:67.10\n",
      "Epoch   770: reducing learning rate of group 0 to 1.3717e-07.\n",
      "epoch 772\n",
      "loss: 0.23    acc:91.82\n",
      "loss: 1.28    acc:66.70\n",
      "epoch 773\n",
      "loss: 0.22    acc:92.92\n",
      "loss: 1.33    acc:66.50\n",
      "epoch 774\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.31    acc:67.10\n",
      "epoch 775\n",
      "loss: 0.23    acc:91.87\n",
      "loss: 1.32    acc:65.60\n",
      "epoch 776\n",
      "loss: 0.23    acc:91.57\n",
      "loss: 1.28    acc:69.00\n",
      "epoch 777\n",
      "loss: 0.23    acc:92.17\n",
      "loss: 1.28    acc:67.90\n",
      "epoch 778\n",
      "loss: 0.24    acc:91.70\n",
      "loss: 1.40    acc:65.50\n",
      "epoch 779\n",
      "loss: 0.23    acc:91.80\n",
      "loss: 1.37    acc:66.70\n",
      "epoch 780\n",
      "loss: 0.23    acc:91.95\n",
      "loss: 1.33    acc:66.80\n",
      "epoch 781\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.20    acc:69.30\n",
      "epoch 782\n",
      "loss: 0.24    acc:91.25\n",
      "loss: 1.27    acc:68.30\n",
      "epoch 783\n",
      "loss: 0.24    acc:91.57\n",
      "loss: 1.29    acc:67.90\n",
      "epoch 784\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.38    acc:67.20\n",
      "epoch 785\n",
      "loss: 0.25    acc:90.92\n",
      "loss: 1.29    acc:68.30\n",
      "epoch 786\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.36    acc:65.70\n",
      "epoch 787\n",
      "loss: 0.23    acc:92.15\n",
      "loss: 1.42    acc:66.20\n",
      "epoch 788\n",
      "loss: 0.23    acc:91.42\n",
      "loss: 1.29    acc:66.60\n",
      "epoch 789\n",
      "loss: 0.23    acc:91.77\n",
      "loss: 1.23    acc:68.50\n",
      "epoch 790\n",
      "loss: 0.24    acc:91.12\n",
      "loss: 1.27    acc:66.90\n",
      "epoch 791\n",
      "loss: 0.23    acc:91.45\n",
      "loss: 1.34    acc:67.10\n",
      "epoch 792\n",
      "loss: 0.23    acc:91.67\n",
      "loss: 1.33    acc:66.60\n",
      "Epoch   791: reducing learning rate of group 0 to 4.5725e-08.\n",
      "epoch 793\n",
      "loss: 0.24    acc:91.35\n",
      "loss: 1.20    acc:67.90\n",
      "epoch 794\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.35    acc:67.00\n",
      "epoch 795\n",
      "loss: 0.23    acc:91.52\n",
      "loss: 1.28    acc:67.60\n",
      "epoch 796\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.28    acc:68.20\n",
      "epoch 797\n",
      "loss: 0.25    acc:91.72\n",
      "loss: 1.25    acc:68.70\n",
      "epoch 798\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.40    acc:66.80\n",
      "epoch 799\n",
      "loss: 0.23    acc:91.80\n",
      "loss: 1.27    acc:71.00\n",
      "epoch 800\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.28    acc:68.30\n",
      "epoch 801\n",
      "loss: 0.22    acc:92.22\n",
      "loss: 1.32    acc:67.20\n",
      "epoch 802\n",
      "loss: 0.24    acc:91.37\n",
      "loss: 1.22    acc:69.70\n",
      "epoch 803\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.31    acc:68.80\n",
      "epoch 804\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.17    acc:69.90\n",
      "epoch 805\n",
      "loss: 0.23    acc:92.27\n",
      "loss: 1.26    acc:68.00\n",
      "epoch 806\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.34    acc:67.40\n",
      "epoch 807\n",
      "loss: 0.23    acc:91.90\n",
      "loss: 1.31    acc:66.50\n",
      "epoch 808\n",
      "loss: 0.23    acc:91.55\n",
      "loss: 1.22    acc:69.20\n",
      "epoch 809\n",
      "loss: 0.25    acc:91.75\n",
      "loss: 1.39    acc:65.90\n",
      "epoch 810\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.17    acc:70.30\n",
      "epoch 811\n",
      "loss: 0.25    acc:90.65\n",
      "loss: 1.31    acc:67.80\n",
      "epoch 812\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.28    acc:67.40\n",
      "epoch 813\n",
      "loss: 0.23    acc:91.90\n",
      "loss: 1.29    acc:67.70\n",
      "Epoch   812: reducing learning rate of group 0 to 1.5242e-08.\n",
      "epoch 814\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.31    acc:69.70\n",
      "epoch 815\n",
      "loss: 0.23    acc:92.02\n",
      "loss: 1.22    acc:68.30\n",
      "epoch 816\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.29    acc:67.60\n",
      "epoch 817\n",
      "loss: 0.23    acc:92.72\n",
      "loss: 1.18    acc:68.60\n",
      "epoch 818\n",
      "loss: 0.23    acc:92.07\n",
      "loss: 1.29    acc:69.40\n",
      "epoch 819\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.20    acc:70.10\n",
      "epoch 820\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.38    acc:66.50\n",
      "epoch 821\n",
      "loss: 0.25    acc:91.75\n",
      "loss: 1.36    acc:65.10\n",
      "epoch 822\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.25    acc:68.00\n",
      "epoch 823\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.23    acc:68.00\n",
      "epoch 824\n",
      "loss: 0.24    acc:91.42\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.30    acc:66.40\n",
      "epoch 825\n",
      "loss: 0.25    acc:90.95\n",
      "loss: 1.27    acc:67.70\n",
      "epoch 826\n",
      "loss: 0.25    acc:91.62\n",
      "loss: 1.44    acc:65.60\n",
      "epoch 827\n",
      "loss: 0.24    acc:91.95\n",
      "loss: 1.33    acc:67.40\n",
      "epoch 828\n",
      "loss: 0.24    acc:91.67\n",
      "loss: 1.22    acc:68.90\n",
      "epoch 829\n",
      "loss: 0.23    acc:92.35\n",
      "loss: 1.30    acc:67.80\n",
      "epoch 830\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.28    acc:66.80\n",
      "epoch 831\n",
      "loss: 0.24    acc:91.57\n",
      "loss: 1.33    acc:66.70\n",
      "epoch 832\n",
      "loss: 0.23    acc:92.32\n",
      "loss: 1.26    acc:68.70\n",
      "epoch 833\n",
      "loss: 0.22    acc:91.95\n",
      "loss: 1.27    acc:68.30\n",
      "epoch 834\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.33    acc:66.40\n",
      "Epoch   833: reducing learning rate of group 0 to 5.0805e-09.\n",
      "epoch 835\n",
      "loss: 0.25    acc:91.40\n",
      "loss: 1.33    acc:65.80\n",
      "epoch 836\n",
      "loss: 0.22    acc:92.60\n",
      "loss: 1.31    acc:66.80\n",
      "epoch 837\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.20    acc:71.60\n",
      "epoch 838\n",
      "loss: 0.23    acc:92.02\n",
      "loss: 1.28    acc:67.30\n",
      "epoch 839\n",
      "loss: 0.23    acc:92.70\n",
      "loss: 1.21    acc:68.40\n",
      "epoch 840\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.36    acc:67.00\n",
      "epoch 841\n",
      "loss: 0.23    acc:91.72\n",
      "loss: 1.39    acc:66.50\n",
      "epoch 842\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.19    acc:68.80\n",
      "epoch 843\n",
      "loss: 0.23    acc:91.65\n",
      "loss: 1.26    acc:68.30\n",
      "epoch 844\n",
      "loss: 0.24    acc:91.85\n",
      "loss: 1.37    acc:67.20\n",
      "epoch 845\n",
      "loss: 0.22    acc:92.62\n",
      "loss: 1.35    acc:67.60\n",
      "epoch 846\n",
      "loss: 0.24    acc:90.85\n",
      "loss: 1.36    acc:65.80\n",
      "epoch 847\n",
      "loss: 0.24    acc:91.30\n",
      "loss: 1.30    acc:68.00\n",
      "epoch 848\n",
      "loss: 0.23    acc:92.00\n",
      "loss: 1.25    acc:67.00\n",
      "epoch 849\n",
      "loss: 0.23    acc:92.20\n",
      "loss: 1.28    acc:67.80\n",
      "epoch 850\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.35    acc:65.30\n",
      "epoch 851\n",
      "loss: 0.22    acc:92.72\n",
      "loss: 1.33    acc:66.30\n",
      "epoch 852\n",
      "loss: 0.23    acc:91.95\n",
      "loss: 1.25    acc:67.60\n",
      "epoch 853\n",
      "loss: 0.22    acc:92.15\n",
      "loss: 1.26    acc:67.80\n",
      "epoch 854\n",
      "loss: 0.23    acc:91.67\n",
      "loss: 1.26    acc:70.10\n",
      "epoch 855\n",
      "loss: 0.22    acc:92.65\n",
      "loss: 1.22    acc:69.60\n",
      "epoch 856\n",
      "loss: 0.23    acc:92.12\n",
      "loss: 1.29    acc:67.50\n",
      "epoch 857\n",
      "loss: 0.24    acc:91.25\n",
      "loss: 1.29    acc:66.10\n",
      "epoch 858\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.32    acc:65.50\n",
      "epoch 859\n",
      "loss: 0.22    acc:92.37\n",
      "loss: 1.26    acc:69.30\n",
      "epoch 860\n",
      "loss: 0.24    acc:91.12\n",
      "loss: 1.30    acc:68.20\n",
      "epoch 861\n",
      "loss: 0.25    acc:91.25\n",
      "loss: 1.27    acc:68.30\n",
      "epoch 862\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.36    acc:65.80\n",
      "epoch 863\n",
      "loss: 0.22    acc:92.32\n",
      "loss: 1.25    acc:68.50\n",
      "epoch 864\n",
      "loss: 0.25    acc:91.30\n",
      "loss: 1.30    acc:66.40\n",
      "epoch 865\n",
      "loss: 0.24    acc:92.02\n",
      "loss: 1.22    acc:68.40\n",
      "epoch 866\n",
      "loss: 0.23    acc:91.75\n",
      "loss: 1.26    acc:68.90\n",
      "epoch 867\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.30    acc:66.20\n",
      "epoch 868\n",
      "loss: 0.23    acc:92.82\n",
      "loss: 1.29    acc:67.40\n",
      "epoch 869\n",
      "loss: 0.22    acc:92.22\n",
      "loss: 1.34    acc:68.30\n",
      "epoch 870\n",
      "loss: 0.23    acc:91.75\n",
      "loss: 1.20    acc:67.70\n",
      "epoch 871\n",
      "loss: 0.23    acc:91.87\n",
      "loss: 1.32    acc:68.40\n",
      "epoch 872\n",
      "loss: 0.22    acc:92.10\n",
      "loss: 1.24    acc:67.70\n",
      "epoch 873\n",
      "loss: 0.22    acc:92.40\n",
      "loss: 1.37    acc:66.90\n",
      "epoch 874\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.24    acc:67.70\n",
      "epoch 875\n",
      "loss: 0.24    acc:91.87\n",
      "loss: 1.20    acc:68.40\n",
      "epoch 876\n",
      "loss: 0.23    acc:91.87\n",
      "loss: 1.24    acc:69.00\n",
      "epoch 877\n",
      "loss: 0.24    acc:91.47\n",
      "loss: 1.29    acc:68.60\n",
      "epoch 878\n",
      "loss: 0.23    acc:91.57\n",
      "loss: 1.36    acc:66.10\n",
      "epoch 879\n",
      "loss: 0.24    acc:91.37\n",
      "loss: 1.31    acc:66.70\n",
      "epoch 880\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.24    acc:68.20\n",
      "epoch 881\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.35    acc:65.60\n",
      "epoch 882\n",
      "loss: 0.24    acc:91.42\n",
      "loss: 1.19    acc:69.40\n",
      "epoch 883\n",
      "loss: 0.22    acc:92.85\n",
      "loss: 1.26    acc:67.80\n",
      "epoch 884\n",
      "loss: 0.25    acc:91.32\n",
      "loss: 1.06    acc:69.30\n",
      "epoch 885\n",
      "loss: 0.23    acc:91.95\n",
      "loss: 1.25    acc:67.30\n",
      "epoch 886\n",
      "loss: 0.24    acc:91.95\n",
      "loss: 1.33    acc:66.10\n",
      "epoch 887\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.29    acc:67.30\n",
      "epoch 888\n",
      "loss: 0.23    acc:92.07\n",
      "loss: 1.24    acc:70.00\n",
      "epoch 889\n",
      "loss: 0.25    acc:91.00\n",
      "loss: 1.28    acc:68.60\n",
      "epoch 890\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.33    acc:66.70\n",
      "epoch 891\n",
      "loss: 0.23    acc:92.02\n",
      "loss: 1.30    acc:68.60\n",
      "epoch 892\n",
      "loss: 0.24    acc:91.37\n",
      "loss: 1.44    acc:66.10\n",
      "epoch 893\n",
      "loss: 0.24    acc:91.30\n",
      "loss: 1.17    acc:70.20\n",
      "epoch 894\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.32    acc:68.60\n",
      "epoch 895\n",
      "loss: 0.22    acc:92.27\n",
      "loss: 1.29    acc:68.20\n",
      "epoch 896\n",
      "loss: 0.23    acc:92.20\n",
      "loss: 1.25    acc:68.30\n",
      "epoch 897\n",
      "loss: 0.24    acc:92.00\n",
      "loss: 1.22    acc:67.00\n",
      "epoch 898\n",
      "loss: 0.24    acc:91.82\n",
      "loss: 1.21    acc:69.00\n",
      "epoch 899\n",
      "loss: 0.24    acc:91.65\n",
      "loss: 1.31    acc:65.80\n",
      "epoch 900\n",
      "loss: 0.23    acc:92.37\n",
      "loss: 1.35    acc:67.10\n",
      "epoch 901\n",
      "loss: 0.23    acc:91.52\n",
      "loss: 1.21    acc:69.60\n",
      "epoch 902\n",
      "loss: 0.23    acc:91.82\n",
      "loss: 1.27    acc:67.00\n",
      "epoch 903\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.24    acc:68.20\n",
      "epoch 904\n",
      "loss: 0.22    acc:92.50\n",
      "loss: 1.27    acc:67.70\n",
      "epoch 905\n",
      "loss: 0.23    acc:91.55\n",
      "loss: 1.27    acc:67.20\n",
      "epoch 906\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.20    acc:69.90\n",
      "epoch 907\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.24    acc:70.80\n",
      "epoch 908\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.29    acc:69.80\n",
      "epoch 909\n",
      "loss: 0.23    acc:92.12\n",
      "loss: 1.25    acc:68.70\n",
      "epoch 910\n",
      "loss: 0.24    acc:91.35\n",
      "loss: 1.39    acc:66.10\n",
      "epoch 911\n",
      "loss: 0.24    acc:91.87\n",
      "loss: 1.33    acc:67.40\n",
      "epoch 912\n",
      "loss: 0.23    acc:91.67\n",
      "loss: 1.32    acc:68.50\n",
      "epoch 913\n",
      "loss: 0.25    acc:92.07\n",
      "loss: 1.43    acc:64.80\n",
      "epoch 914\n",
      "loss: 0.22    acc:92.52\n",
      "loss: 1.24    acc:70.20\n",
      "epoch 915\n",
      "loss: 0.22    acc:92.37\n",
      "loss: 1.04    acc:70.70\n",
      "epoch 916\n",
      "loss: 0.24    acc:91.50\n",
      "loss: 1.21    acc:69.50\n",
      "epoch 917\n",
      "loss: 0.24    acc:91.80\n",
      "loss: 1.24    acc:68.60\n",
      "epoch 918\n",
      "loss: 0.24    acc:91.90\n",
      "loss: 1.36    acc:65.50\n",
      "epoch 919\n",
      "loss: 0.24    acc:91.92\n",
      "loss: 1.32    acc:67.40\n",
      "epoch 920\n",
      "loss: 0.23    acc:91.52\n",
      "loss: 1.28    acc:66.30\n",
      "epoch 921\n",
      "loss: 0.24    acc:91.57\n",
      "loss: 1.35    acc:66.00\n",
      "epoch 922\n",
      "loss: 0.23    acc:92.42\n",
      "loss: 1.22    acc:67.00\n",
      "epoch 923\n",
      "loss: 0.23    acc:92.17\n",
      "loss: 1.31    acc:67.30\n",
      "epoch 924\n",
      "loss: 0.25    acc:90.97\n",
      "loss: 1.27    acc:67.90\n",
      "epoch 925\n",
      "loss: 0.23    acc:92.42\n",
      "loss: 1.12    acc:71.50\n",
      "epoch 926\n",
      "loss: 0.23    acc:91.72\n",
      "loss: 1.36    acc:66.40\n",
      "epoch 927\n",
      "loss: 0.23    acc:91.52\n",
      "loss: 1.18    acc:67.70\n",
      "epoch 928\n",
      "loss: 0.24    acc:91.82\n",
      "loss: 1.30    acc:65.70\n",
      "epoch 929\n",
      "loss: 0.22    acc:92.02\n",
      "loss: 1.31    acc:68.20\n",
      "epoch 930\n",
      "loss: 0.23    acc:91.60\n",
      "loss: 1.27    acc:68.70\n",
      "epoch 931\n",
      "loss: 0.23    acc:91.65\n",
      "loss: 1.19    acc:69.00\n",
      "epoch 932\n",
      "loss: 0.23    acc:92.15\n",
      "loss: 1.33    acc:69.80\n",
      "epoch 933\n",
      "loss: 0.24    acc:91.32\n",
      "loss: 1.35    acc:65.30\n",
      "epoch 934\n",
      "loss: 0.24    acc:91.75\n",
      "loss: 1.29    acc:67.00\n",
      "epoch 935\n",
      "loss: 0.24    acc:92.12\n",
      "loss: 1.36    acc:64.70\n",
      "epoch 936\n",
      "loss: 0.26    acc:90.97\n",
      "loss: 1.46    acc:65.80\n",
      "epoch 937\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.23    acc:68.70\n",
      "epoch 938\n",
      "loss: 0.24    acc:91.22\n",
      "loss: 1.22    acc:67.80\n",
      "epoch 939\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.35    acc:65.70\n",
      "epoch 940\n",
      "loss: 0.24    acc:91.37\n",
      "loss: 1.40    acc:66.20\n",
      "epoch 941\n",
      "loss: 0.22    acc:92.40\n",
      "loss: 1.10    acc:72.10\n",
      "epoch 942\n",
      "loss: 0.23    acc:91.92\n",
      "loss: 1.22    acc:69.00\n",
      "epoch 943\n",
      "loss: 0.25    acc:91.07\n",
      "loss: 1.33    acc:66.00\n",
      "epoch 944\n",
      "loss: 0.23    acc:92.12\n",
      "loss: 1.31    acc:69.10\n",
      "epoch 945\n",
      "loss: 0.24    acc:91.70\n",
      "loss: 1.23    acc:68.20\n",
      "epoch 946\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.33    acc:67.10\n",
      "epoch 947\n",
      "loss: 0.24    acc:91.82\n",
      "loss: 1.33    acc:66.50\n",
      "epoch 948\n",
      "loss: 0.24    acc:91.72\n",
      "loss: 1.33    acc:66.70\n",
      "epoch 949\n",
      "loss: 0.24    acc:91.40\n",
      "loss: 1.25    acc:70.90\n",
      "epoch 950\n",
      "loss: 0.25    acc:91.15\n",
      "loss: 1.29    acc:67.10\n",
      "epoch 951\n",
      "loss: 0.24    acc:91.97\n",
      "loss: 1.22    acc:69.10\n",
      "epoch 952\n",
      "loss: 0.23    acc:92.50\n",
      "loss: 1.13    acc:68.20\n",
      "epoch 953\n",
      "loss: 0.25    acc:91.10\n",
      "loss: 1.34    acc:66.20\n",
      "epoch 954\n",
      "loss: 0.24    acc:91.87\n",
      "loss: 1.32    acc:66.90\n",
      "epoch 955\n",
      "loss: 0.23    acc:91.82\n",
      "loss: 1.27    acc:67.70\n",
      "epoch 956\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.26    acc:68.70\n",
      "epoch 957\n",
      "loss: 0.24    acc:91.60\n",
      "loss: 1.37    acc:65.70\n",
      "epoch 958\n",
      "loss: 0.23    acc:92.20\n",
      "loss: 1.37    acc:66.00\n",
      "epoch 959\n",
      "loss: 0.22    acc:92.35\n",
      "loss: 1.22    acc:69.50\n",
      "epoch 960\n",
      "loss: 0.22    acc:92.07\n",
      "loss: 1.21    acc:68.50\n",
      "epoch 961\n",
      "loss: 0.23    acc:92.22\n",
      "loss: 1.21    acc:67.40\n",
      "epoch 962\n",
      "loss: 0.23    acc:91.82\n",
      "loss: 1.32    acc:68.10\n",
      "epoch 963\n",
      "loss: 0.23    acc:91.77\n",
      "loss: 1.26    acc:69.90\n",
      "epoch 964\n",
      "loss: 0.22    acc:92.20\n",
      "loss: 1.27    acc:66.80\n",
      "epoch 965\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.23    acc:92.00\n",
      "loss: 1.07    acc:72.30\n",
      "epoch 966\n",
      "loss: 0.22    acc:91.95\n",
      "loss: 1.13    acc:70.50\n",
      "epoch 967\n",
      "loss: 0.24    acc:91.52\n",
      "loss: 1.28    acc:67.10\n",
      "epoch 968\n",
      "loss: 0.23    acc:91.75\n",
      "loss: 1.26    acc:68.30\n",
      "epoch 969\n",
      "loss: 0.23    acc:92.45\n",
      "loss: 1.19    acc:68.80\n",
      "epoch 970\n",
      "loss: 0.24    acc:90.97\n",
      "loss: 1.24    acc:68.20\n",
      "epoch 971\n",
      "loss: 0.23    acc:92.05\n",
      "loss: 1.19    acc:68.20\n",
      "epoch 972\n",
      "loss: 0.23    acc:91.95\n",
      "loss: 1.22    acc:68.00\n",
      "epoch 973\n",
      "loss: 0.23    acc:91.60\n",
      "loss: 1.31    acc:66.70\n",
      "epoch 974\n",
      "loss: 0.24    acc:91.85\n",
      "loss: 1.20    acc:67.50\n",
      "epoch 975\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.25    acc:67.90\n",
      "epoch 976\n",
      "loss: 0.25    acc:90.95\n",
      "loss: 1.21    acc:69.80\n",
      "epoch 977\n",
      "loss: 0.23    acc:92.37\n",
      "loss: 1.25    acc:70.30\n",
      "epoch 978\n",
      "loss: 0.24    acc:91.95\n",
      "loss: 1.30    acc:67.50\n",
      "epoch 979\n",
      "loss: 0.23    acc:91.62\n",
      "loss: 1.30    acc:68.90\n",
      "epoch 980\n",
      "loss: 0.23    acc:91.65\n",
      "loss: 1.28    acc:68.50\n",
      "epoch 981\n",
      "loss: 0.25    acc:90.92\n",
      "loss: 1.20    acc:67.30\n",
      "epoch 982\n",
      "loss: 0.24    acc:91.62\n",
      "loss: 1.25    acc:66.40\n",
      "epoch 983\n",
      "loss: 0.23    acc:92.10\n",
      "loss: 1.27    acc:68.10\n",
      "epoch 984\n",
      "loss: 0.23    acc:92.17\n",
      "loss: 1.19    acc:71.20\n",
      "epoch 985\n",
      "loss: 0.23    acc:91.95\n",
      "loss: 1.28    acc:69.20\n",
      "epoch 986\n",
      "loss: 0.24    acc:91.97\n",
      "loss: 1.26    acc:69.30\n",
      "epoch 987\n",
      "loss: 0.23    acc:92.25\n",
      "loss: 1.35    acc:67.00\n",
      "epoch 988\n",
      "loss: 0.23    acc:92.07\n",
      "loss: 1.29    acc:69.60\n",
      "epoch 989\n",
      "loss: 0.23    acc:91.72\n",
      "loss: 1.19    acc:68.80\n",
      "epoch 990\n",
      "loss: 0.23    acc:91.97\n",
      "loss: 1.39    acc:66.00\n",
      "epoch 991\n",
      "loss: 0.24    acc:91.55\n",
      "loss: 1.32    acc:66.30\n",
      "epoch 992\n",
      "loss: 0.23    acc:91.87\n",
      "loss: 1.24    acc:67.60\n",
      "epoch 993\n",
      "loss: 0.24    acc:90.92\n",
      "loss: 1.35    acc:67.40\n",
      "epoch 994\n",
      "loss: 0.22    acc:92.10\n",
      "loss: 1.22    acc:69.20\n",
      "epoch 995\n",
      "loss: 0.23    acc:91.50\n",
      "loss: 1.27    acc:69.50\n",
      "epoch 996\n",
      "loss: 0.22    acc:92.40\n",
      "loss: 1.27    acc:68.90\n",
      "epoch 997\n",
      "loss: 0.22    acc:92.40\n",
      "loss: 1.26    acc:68.30\n",
      "epoch 998\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.18    acc:69.40\n",
      "epoch 999\n",
      "loss: 0.23    acc:91.85\n",
      "loss: 1.29    acc:67.50\n",
      "epoch 1000\n",
      "loss: 0.24    acc:91.45\n",
      "loss: 1.22    acc:68.20\n",
      "64\n",
      "628490\n",
      "epoch 1\n",
      "loss: 2.07    acc:24.52\n",
      "loss: 2.20    acc:13.60\n",
      "epoch 2\n",
      "loss: 1.82    acc:35.42\n",
      "loss: 1.77    acc:32.20\n",
      "epoch 3\n",
      "loss: 1.68    acc:41.60\n",
      "loss: 1.68    acc:35.40\n",
      "epoch 4\n",
      "loss: 1.58    acc:43.85\n",
      "loss: 1.62    acc:40.20\n",
      "epoch 5\n",
      "loss: 1.51    acc:46.35\n",
      "loss: 1.48    acc:46.70\n",
      "epoch 6\n",
      "loss: 1.44    acc:49.97\n",
      "loss: 1.64    acc:36.60\n",
      "epoch 7\n",
      "loss: 1.38    acc:51.17\n",
      "loss: 1.52    acc:46.10\n",
      "epoch 8\n",
      "loss: 1.32    acc:54.20\n",
      "loss: 1.45    acc:46.70\n",
      "epoch 9\n",
      "loss: 1.26    acc:55.87\n",
      "loss: 1.34    acc:51.00\n",
      "epoch 10\n",
      "loss: 1.23    acc:56.70\n",
      "loss: 1.40    acc:49.80\n",
      "epoch 11\n",
      "loss: 1.17    acc:59.95\n",
      "loss: 1.34    acc:52.70\n",
      "epoch 12\n",
      "loss: 1.12    acc:61.17\n",
      "loss: 1.37    acc:49.60\n",
      "epoch 13\n",
      "loss: 1.09    acc:61.47\n",
      "loss: 1.21    acc:56.40\n",
      "epoch 14\n",
      "loss: 1.07    acc:63.00\n",
      "loss: 1.48    acc:50.00\n",
      "epoch 15\n",
      "loss: 1.02    acc:64.50\n",
      "loss: 1.12    acc:58.90\n",
      "epoch 16\n",
      "loss: 0.98    acc:65.80\n",
      "loss: 1.20    acc:59.30\n",
      "epoch 17\n",
      "loss: 0.96    acc:66.35\n",
      "loss: 1.25    acc:54.10\n",
      "epoch 18\n",
      "loss: 0.95    acc:66.17\n",
      "loss: 1.16    acc:59.00\n",
      "epoch 19\n",
      "loss: 0.93    acc:68.10\n",
      "loss: 1.18    acc:59.50\n",
      "epoch 20\n",
      "loss: 0.89    acc:68.85\n",
      "loss: 1.14    acc:60.60\n",
      "epoch 21\n",
      "loss: 0.84    acc:70.97\n",
      "loss: 1.05    acc:63.20\n",
      "epoch 22\n",
      "loss: 0.82    acc:72.25\n",
      "loss: 1.21    acc:57.90\n",
      "epoch 23\n",
      "loss: 0.81    acc:71.50\n",
      "loss: 1.20    acc:60.70\n",
      "epoch 24\n",
      "loss: 0.79    acc:72.72\n",
      "loss: 1.20    acc:58.40\n",
      "epoch 25\n",
      "loss: 0.77    acc:72.85\n",
      "loss: 1.06    acc:64.00\n",
      "epoch 26\n",
      "loss: 0.75    acc:73.95\n",
      "loss: 1.24    acc:58.30\n",
      "epoch 27\n",
      "loss: 0.72    acc:75.50\n",
      "loss: 1.19    acc:61.50\n",
      "epoch 28\n",
      "loss: 0.71    acc:75.35\n",
      "loss: 1.06    acc:62.60\n",
      "epoch 29\n",
      "loss: 0.70    acc:75.75\n",
      "loss: 1.04    acc:64.90\n",
      "epoch 30\n",
      "loss: 0.66    acc:77.20\n",
      "loss: 0.97    acc:67.30\n",
      "epoch 31\n",
      "loss: 0.63    acc:78.90\n",
      "loss: 1.17    acc:61.60\n",
      "epoch 32\n",
      "loss: 0.63    acc:77.90\n",
      "loss: 1.35    acc:57.90\n",
      "epoch 33\n",
      "loss: 0.61    acc:78.67\n",
      "loss: 1.13    acc:62.70\n",
      "epoch 34\n",
      "loss: 0.59    acc:80.07\n",
      "loss: 1.22    acc:61.60\n",
      "epoch 35\n",
      "loss: 0.58    acc:79.80\n",
      "loss: 1.12    acc:64.20\n",
      "epoch 36\n",
      "loss: 0.55    acc:81.62\n",
      "loss: 1.02    acc:67.00\n",
      "epoch 37\n",
      "loss: 0.54    acc:81.82\n",
      "loss: 1.27    acc:60.40\n",
      "epoch 38\n",
      "loss: 0.56    acc:79.95\n",
      "loss: 0.97    acc:67.10\n",
      "epoch 39\n",
      "loss: 0.49    acc:82.90\n",
      "loss: 1.12    acc:63.00\n",
      "epoch 40\n",
      "loss: 0.48    acc:83.30\n",
      "loss: 0.99    acc:68.90\n",
      "epoch 41\n",
      "loss: 0.47    acc:84.35\n",
      "loss: 1.22    acc:63.40\n",
      "epoch 42\n",
      "loss: 0.47    acc:83.77\n",
      "loss: 1.17    acc:62.40\n",
      "epoch 43\n",
      "loss: 0.43    acc:85.40\n",
      "loss: 0.93    acc:70.70\n",
      "epoch 44\n",
      "loss: 0.43    acc:86.05\n",
      "loss: 1.02    acc:69.00\n",
      "epoch 45\n",
      "loss: 0.41    acc:86.37\n",
      "loss: 1.13    acc:66.70\n",
      "epoch 46\n",
      "loss: 0.40    acc:86.65\n",
      "loss: 1.14    acc:66.00\n",
      "epoch 47\n",
      "loss: 0.37    acc:87.90\n",
      "loss: 1.10    acc:67.40\n",
      "epoch 48\n",
      "loss: 0.39    acc:87.02\n",
      "loss: 1.28    acc:62.00\n",
      "epoch 49\n",
      "loss: 0.36    acc:88.30\n",
      "loss: 1.03    acc:68.70\n",
      "epoch 50\n",
      "loss: 0.36    acc:87.65\n",
      "loss: 1.24    acc:64.30\n",
      "epoch 51\n",
      "loss: 0.34    acc:88.77\n",
      "loss: 1.13    acc:66.20\n",
      "epoch 52\n",
      "loss: 0.33    acc:89.17\n",
      "loss: 1.14    acc:65.00\n",
      "epoch 53\n",
      "loss: 0.31    acc:89.17\n",
      "loss: 1.05    acc:69.00\n",
      "epoch 54\n",
      "loss: 0.32    acc:89.20\n",
      "loss: 1.18    acc:63.30\n",
      "epoch 55\n",
      "loss: 0.31    acc:89.20\n",
      "loss: 1.01    acc:68.90\n",
      "epoch 56\n",
      "loss: 0.29    acc:90.82\n",
      "loss: 1.19    acc:66.70\n",
      "epoch 57\n",
      "loss: 0.29    acc:90.77\n",
      "loss: 1.09    acc:66.80\n",
      "epoch 58\n",
      "loss: 0.27    acc:90.90\n",
      "loss: 1.07    acc:66.80\n",
      "epoch 59\n",
      "loss: 0.27    acc:91.55\n",
      "loss: 1.27    acc:64.90\n",
      "epoch 60\n",
      "loss: 0.24    acc:91.85\n",
      "loss: 1.14    acc:68.30\n",
      "epoch 61\n",
      "loss: 0.23    acc:92.85\n",
      "loss: 1.30    acc:66.60\n",
      "epoch 62\n",
      "loss: 0.24    acc:92.45\n",
      "loss: 1.10    acc:67.40\n",
      "epoch 63\n",
      "loss: 0.24    acc:92.02\n",
      "loss: 1.14    acc:67.00\n",
      "epoch 64\n",
      "loss: 0.22    acc:93.05\n",
      "loss: 0.96    acc:71.30\n",
      "epoch 65\n",
      "loss: 0.23    acc:92.55\n",
      "loss: 1.21    acc:66.10\n",
      "epoch 66\n",
      "loss: 0.23    acc:92.42\n",
      "loss: 1.39    acc:62.20\n",
      "epoch 67\n",
      "loss: 0.20    acc:93.90\n",
      "loss: 1.21    acc:68.10\n",
      "epoch 68\n",
      "loss: 0.22    acc:92.92\n",
      "loss: 1.34    acc:67.30\n",
      "epoch 69\n",
      "loss: 0.19    acc:94.32\n",
      "loss: 1.01    acc:71.90\n",
      "epoch 70\n",
      "loss: 0.17    acc:94.45\n",
      "loss: 1.16    acc:69.30\n",
      "epoch 71\n",
      "loss: 0.17    acc:94.55\n",
      "loss: 1.12    acc:68.30\n",
      "epoch 72\n",
      "loss: 0.19    acc:94.00\n",
      "loss: 1.11    acc:68.60\n",
      "epoch 73\n",
      "loss: 0.19    acc:93.57\n",
      "loss: 1.02    acc:71.00\n",
      "epoch 74\n",
      "loss: 0.16    acc:95.25\n",
      "loss: 1.31    acc:65.80\n",
      "epoch 75\n",
      "loss: 0.14    acc:96.00\n",
      "loss: 1.30    acc:67.80\n",
      "epoch 76\n",
      "loss: 0.15    acc:95.72\n",
      "loss: 1.25    acc:69.50\n",
      "epoch 77\n",
      "loss: 0.14    acc:95.57\n",
      "loss: 1.16    acc:70.30\n",
      "epoch 78\n",
      "loss: 0.16    acc:94.62\n",
      "loss: 1.34    acc:67.80\n",
      "epoch 79\n",
      "loss: 0.17    acc:94.32\n",
      "loss: 1.27    acc:67.20\n",
      "epoch 80\n",
      "loss: 0.16    acc:94.67\n",
      "loss: 1.21    acc:68.20\n",
      "epoch 81\n",
      "loss: 0.16    acc:94.75\n",
      "loss: 1.04    acc:71.90\n",
      "epoch 82\n",
      "loss: 0.16    acc:94.62\n",
      "loss: 1.11    acc:68.10\n",
      "epoch 83\n",
      "loss: 0.17    acc:94.30\n",
      "loss: 1.14    acc:70.40\n",
      "epoch 84\n",
      "loss: 0.13    acc:95.85\n",
      "loss: 1.18    acc:67.80\n",
      "epoch 85\n",
      "loss: 0.14    acc:95.65\n",
      "loss: 1.24    acc:69.00\n",
      "epoch 86\n",
      "loss: 0.13    acc:96.07\n",
      "loss: 1.30    acc:68.60\n",
      "epoch 87\n",
      "loss: 0.12    acc:96.67\n",
      "loss: 1.12    acc:69.40\n",
      "epoch 88\n",
      "loss: 0.14    acc:95.72\n",
      "loss: 1.23    acc:69.20\n",
      "epoch 89\n",
      "loss: 0.13    acc:95.52\n",
      "loss: 1.21    acc:69.60\n",
      "epoch 90\n",
      "loss: 0.12    acc:96.12\n",
      "loss: 1.37    acc:68.90\n",
      "epoch 91\n",
      "loss: 0.12    acc:96.47\n",
      "loss: 1.22    acc:68.30\n",
      "epoch 92\n",
      "loss: 0.12    acc:96.32\n",
      "loss: 1.10    acc:70.60\n",
      "epoch 93\n",
      "loss: 0.11    acc:96.55\n",
      "loss: 1.48    acc:65.30\n",
      "epoch 94\n",
      "loss: 0.11    acc:96.37\n",
      "loss: 1.15    acc:70.40\n",
      "epoch 95\n",
      "loss: 0.10    acc:97.17\n",
      "loss: 1.21    acc:71.90\n",
      "epoch 96\n",
      "loss: 0.10    acc:97.07\n",
      "loss: 1.31    acc:67.20\n",
      "epoch 97\n",
      "loss: 0.10    acc:96.75\n",
      "loss: 1.20    acc:70.50\n",
      "epoch 98\n",
      "loss: 0.10    acc:96.80\n",
      "loss: 1.18    acc:71.50\n",
      "epoch 99\n",
      "loss: 0.11    acc:96.65\n",
      "loss: 1.22    acc:68.50\n",
      "epoch 100\n",
      "loss: 0.09    acc:97.40\n",
      "loss: 1.33    acc:69.10\n",
      "epoch 101\n",
      "loss: 0.08    acc:98.07\n",
      "loss: 1.10    acc:72.80\n",
      "epoch 102\n",
      "loss: 0.08    acc:97.85\n",
      "loss: 1.35    acc:66.50\n",
      "epoch 103\n",
      "loss: 0.09    acc:97.45\n",
      "loss: 1.29    acc:70.00\n",
      "epoch 104\n",
      "loss: 0.07    acc:97.75\n",
      "loss: 1.51    acc:66.90\n",
      "epoch 105\n",
      "loss: 0.11    acc:96.45\n",
      "loss: 1.20    acc:70.70\n",
      "epoch 106\n",
      "loss: 0.08    acc:97.50\n",
      "loss: 1.23    acc:70.10\n",
      "epoch 107\n",
      "loss: 0.09    acc:97.15\n",
      "loss: 1.27    acc:70.60\n",
      "epoch 108\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.08    acc:97.80\n",
      "loss: 1.23    acc:69.60\n",
      "epoch 109\n",
      "loss: 0.08    acc:97.72\n",
      "loss: 1.66    acc:65.60\n",
      "epoch 110\n",
      "loss: 0.08    acc:97.50\n",
      "loss: 1.36    acc:70.80\n",
      "epoch 111\n",
      "loss: 0.09    acc:96.72\n",
      "loss: 1.36    acc:70.40\n",
      "epoch 112\n",
      "loss: 0.10    acc:96.57\n",
      "loss: 1.11    acc:73.50\n",
      "epoch 113\n",
      "loss: 0.10    acc:97.15\n",
      "loss: 1.30    acc:69.90\n",
      "epoch 114\n",
      "loss: 0.08    acc:97.52\n",
      "loss: 1.26    acc:69.80\n",
      "epoch 115\n",
      "loss: 0.09    acc:97.05\n",
      "loss: 1.45    acc:68.20\n",
      "epoch 116\n",
      "loss: 0.08    acc:97.92\n",
      "loss: 1.35    acc:69.10\n",
      "epoch 117\n",
      "loss: 0.09    acc:97.02\n",
      "loss: 1.22    acc:70.00\n",
      "epoch 118\n",
      "loss: 0.08    acc:97.62\n",
      "loss: 1.35    acc:69.10\n",
      "epoch 119\n",
      "loss: 0.07    acc:97.95\n",
      "loss: 1.23    acc:70.00\n",
      "epoch 120\n",
      "loss: 0.07    acc:97.77\n",
      "loss: 1.20    acc:72.70\n",
      "epoch 121\n",
      "loss: 0.07    acc:97.97\n",
      "loss: 1.24    acc:71.00\n",
      "epoch 122\n",
      "loss: 0.06    acc:98.22\n",
      "loss: 1.39    acc:68.80\n",
      "epoch 123\n",
      "loss: 0.06    acc:98.32\n",
      "loss: 1.36    acc:70.20\n",
      "epoch 124\n",
      "loss: 0.07    acc:98.05\n",
      "loss: 1.38    acc:70.40\n",
      "epoch 125\n",
      "loss: 0.08    acc:97.70\n",
      "loss: 1.41    acc:68.90\n",
      "epoch 126\n",
      "loss: 0.07    acc:97.92\n",
      "loss: 1.04    acc:74.50\n",
      "epoch 127\n",
      "loss: 0.07    acc:97.50\n",
      "loss: 1.47    acc:68.40\n",
      "epoch 128\n",
      "loss: 0.08    acc:97.37\n",
      "loss: 1.26    acc:72.80\n",
      "epoch 129\n",
      "loss: 0.07    acc:98.07\n",
      "loss: 1.29    acc:70.00\n",
      "epoch 130\n",
      "loss: 0.06    acc:98.30\n",
      "loss: 1.35    acc:69.10\n",
      "epoch 131\n",
      "loss: 0.05    acc:98.42\n",
      "loss: 1.39    acc:70.00\n",
      "epoch 132\n",
      "loss: 0.05    acc:98.35\n",
      "loss: 1.36    acc:68.90\n",
      "epoch 133\n",
      "loss: 0.06    acc:98.12\n",
      "loss: 1.23    acc:72.40\n",
      "epoch 134\n",
      "loss: 0.07    acc:97.77\n",
      "loss: 1.35    acc:70.40\n",
      "epoch 135\n",
      "loss: 0.06    acc:97.97\n",
      "loss: 1.36    acc:69.50\n",
      "epoch 136\n",
      "loss: 0.06    acc:98.05\n",
      "loss: 1.41    acc:70.00\n",
      "epoch 137\n",
      "loss: 0.09    acc:97.25\n",
      "loss: 1.33    acc:70.90\n",
      "epoch 138\n",
      "loss: 0.08    acc:97.72\n",
      "loss: 1.55    acc:67.40\n",
      "epoch 139\n",
      "loss: 0.08    acc:97.35\n",
      "loss: 1.37    acc:71.10\n",
      "epoch 140\n",
      "loss: 0.07    acc:97.97\n",
      "loss: 1.34    acc:70.60\n",
      "epoch 141\n",
      "loss: 0.06    acc:98.45\n",
      "loss: 1.52    acc:67.80\n",
      "epoch 142\n",
      "loss: 0.07    acc:97.75\n",
      "loss: 1.20    acc:74.70\n",
      "epoch 143\n",
      "loss: 0.06    acc:98.20\n",
      "loss: 1.50    acc:69.60\n",
      "epoch 144\n",
      "loss: 0.07    acc:97.52\n",
      "loss: 1.59    acc:66.40\n",
      "epoch 145\n",
      "loss: 0.10    acc:96.85\n",
      "loss: 1.53    acc:69.20\n",
      "epoch 146\n",
      "loss: 0.06    acc:97.87\n",
      "loss: 1.24    acc:71.70\n",
      "epoch 147\n",
      "loss: 0.07    acc:97.70\n",
      "loss: 1.40    acc:71.10\n",
      "epoch 148\n",
      "loss: 0.06    acc:98.30\n",
      "loss: 1.24    acc:71.90\n",
      "epoch 149\n",
      "loss: 0.05    acc:98.40\n",
      "loss: 1.27    acc:71.80\n",
      "epoch 150\n",
      "loss: 0.05    acc:98.77\n",
      "loss: 1.34    acc:69.90\n",
      "epoch 151\n",
      "loss: 0.05    acc:98.52\n",
      "loss: 1.48    acc:68.90\n",
      "epoch 152\n",
      "loss: 0.05    acc:98.30\n",
      "loss: 1.43    acc:69.60\n",
      "epoch 153\n",
      "loss: 0.05    acc:98.37\n",
      "loss: 1.58    acc:66.70\n",
      "epoch 154\n",
      "loss: 0.05    acc:98.32\n",
      "loss: 1.55    acc:68.90\n",
      "epoch 155\n",
      "loss: 0.04    acc:98.77\n",
      "loss: 1.16    acc:74.80\n",
      "epoch 156\n",
      "loss: 0.06    acc:98.27\n",
      "loss: 1.48    acc:68.70\n",
      "epoch 157\n",
      "loss: 0.05    acc:98.35\n",
      "loss: 1.58    acc:68.20\n",
      "epoch 158\n",
      "loss: 0.03    acc:99.20\n",
      "loss: 1.31    acc:73.40\n",
      "epoch 159\n",
      "loss: 0.04    acc:98.57\n",
      "loss: 1.38    acc:71.30\n",
      "epoch 160\n",
      "loss: 0.06    acc:98.00\n",
      "loss: 1.63    acc:67.40\n",
      "epoch 161\n",
      "loss: 0.06    acc:98.30\n",
      "loss: 1.47    acc:69.00\n",
      "epoch 162\n",
      "loss: 0.05    acc:98.30\n",
      "loss: 1.37    acc:70.80\n",
      "epoch 163\n",
      "loss: 0.05    acc:98.22\n",
      "loss: 1.60    acc:69.70\n",
      "epoch 164\n",
      "loss: 0.06    acc:98.27\n",
      "loss: 1.34    acc:72.40\n",
      "epoch 165\n",
      "loss: 0.05    acc:98.55\n",
      "loss: 1.17    acc:72.80\n",
      "epoch 166\n",
      "loss: 0.05    acc:98.30\n",
      "loss: 1.30    acc:73.40\n",
      "epoch 167\n",
      "loss: 0.06    acc:98.12\n",
      "loss: 1.42    acc:71.70\n",
      "epoch 168\n",
      "loss: 0.07    acc:97.80\n",
      "loss: 1.47    acc:69.30\n",
      "epoch 169\n",
      "loss: 0.06    acc:98.07\n",
      "loss: 1.13    acc:73.70\n",
      "epoch 170\n",
      "loss: 0.05    acc:98.55\n",
      "loss: 1.39    acc:73.30\n",
      "epoch 171\n",
      "loss: 0.06    acc:98.20\n",
      "loss: 1.56    acc:70.20\n",
      "epoch 172\n",
      "loss: 0.06    acc:98.05\n",
      "loss: 1.27    acc:71.40\n",
      "epoch 173\n",
      "loss: 0.05    acc:98.47\n",
      "loss: 1.40    acc:71.10\n",
      "epoch 174\n",
      "loss: 0.05    acc:98.15\n",
      "loss: 1.25    acc:72.90\n",
      "epoch 175\n",
      "loss: 0.05    acc:98.57\n",
      "loss: 1.35    acc:71.80\n",
      "epoch 176\n",
      "loss: 0.04    acc:98.90\n",
      "loss: 1.41    acc:71.10\n",
      "epoch 177\n",
      "loss: 0.04    acc:98.77\n",
      "loss: 1.35    acc:72.20\n",
      "epoch 178\n",
      "loss: 0.05    acc:98.22\n",
      "loss: 1.49    acc:71.60\n",
      "epoch 179\n",
      "loss: 0.06    acc:98.27\n",
      "loss: 1.57    acc:67.80\n",
      "Epoch   178: reducing learning rate of group 0 to 1.0000e-04.\n",
      "epoch 180\n",
      "loss: 0.03    acc:99.35\n",
      "loss: 1.38    acc:71.10\n",
      "epoch 181\n",
      "loss: 0.01    acc:99.65\n",
      "loss: 1.37    acc:72.40\n",
      "epoch 182\n",
      "loss: 0.02    acc:99.62\n",
      "loss: 1.39    acc:71.60\n",
      "epoch 183\n",
      "loss: 0.02    acc:99.50\n",
      "loss: 1.19    acc:72.80\n",
      "epoch 184\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.25    acc:72.60\n",
      "epoch 185\n",
      "loss: 0.01    acc:99.67\n",
      "loss: 1.35    acc:72.90\n",
      "epoch 186\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.23    acc:75.00\n",
      "epoch 187\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.17    acc:74.80\n",
      "epoch 188\n",
      "loss: 0.01    acc:99.62\n",
      "loss: 1.23    acc:74.30\n",
      "epoch 189\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.36    acc:71.60\n",
      "epoch 190\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.15    acc:73.70\n",
      "epoch 191\n",
      "loss: 0.01    acc:99.72\n",
      "loss: 1.15    acc:75.40\n",
      "epoch 192\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.24    acc:75.10\n",
      "epoch 193\n",
      "loss: 0.01    acc:99.67\n",
      "loss: 1.35    acc:72.10\n",
      "epoch 194\n",
      "loss: 0.01    acc:99.67\n",
      "loss: 1.46    acc:72.30\n",
      "epoch 195\n",
      "loss: 0.01    acc:99.72\n",
      "loss: 1.25    acc:72.70\n",
      "epoch 196\n",
      "loss: 0.01    acc:99.82\n",
      "loss: 1.32    acc:72.10\n",
      "epoch 197\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.22    acc:72.70\n",
      "epoch 198\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.21    acc:74.10\n",
      "epoch 199\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.20    acc:72.80\n",
      "epoch 200\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.29    acc:75.00\n",
      "epoch 201\n",
      "loss: 0.01    acc:99.65\n",
      "loss: 1.35    acc:71.20\n",
      "epoch 202\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.36    acc:71.60\n",
      "epoch 203\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.30    acc:72.30\n",
      "epoch 204\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.23    acc:74.50\n",
      "epoch 205\n",
      "loss: 0.01    acc:99.75\n",
      "loss: 1.22    acc:74.70\n",
      "epoch 206\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.40    acc:71.80\n",
      "epoch 207\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.30    acc:72.60\n",
      "epoch 208\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.30    acc:72.40\n",
      "epoch 209\n",
      "loss: 0.01    acc:99.95\n",
      "loss: 1.51    acc:71.10\n",
      "epoch 210\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.32    acc:70.70\n",
      "epoch 211\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.35    acc:72.90\n",
      "epoch 212\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.23    acc:74.30\n",
      "epoch 213\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.33    acc:73.40\n",
      "epoch 214\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.29    acc:73.70\n",
      "epoch 215\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.23    acc:73.80\n",
      "epoch 216\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.52    acc:72.60\n",
      "epoch 217\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.27    acc:74.30\n",
      "epoch 218\n",
      "loss: 0.01    acc:99.70\n",
      "loss: 1.48    acc:72.10\n",
      "epoch 219\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.39    acc:73.00\n",
      "epoch 220\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.37    acc:71.80\n",
      "epoch 221\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.56    acc:71.50\n",
      "epoch 222\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.48    acc:72.00\n",
      "epoch 223\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.27    acc:72.20\n",
      "epoch 224\n",
      "loss: 0.01    acc:99.82\n",
      "loss: 1.52    acc:70.80\n",
      "epoch 225\n",
      "loss: 0.01    acc:99.50\n",
      "loss: 1.25    acc:74.10\n",
      "epoch 226\n",
      "loss: 0.01    acc:99.70\n",
      "loss: 1.44    acc:71.70\n",
      "epoch 227\n",
      "loss: 0.01    acc:99.62\n",
      "loss: 1.26    acc:73.80\n",
      "epoch 228\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.25    acc:73.70\n",
      "epoch 229\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.23    acc:74.80\n",
      "epoch 230\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.40    acc:72.40\n",
      "Epoch   229: reducing learning rate of group 0 to 3.3333e-05.\n",
      "epoch 231\n",
      "loss: 0.01    acc:99.95\n",
      "loss: 1.18    acc:74.90\n",
      "epoch 232\n",
      "loss: 0.01    acc:99.95\n",
      "loss: 1.42    acc:71.70\n",
      "epoch 233\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.21    acc:74.20\n",
      "epoch 234\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.35    acc:72.40\n",
      "epoch 235\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.29    acc:73.50\n",
      "epoch 236\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.19    acc:73.60\n",
      "epoch 237\n",
      "loss: 0.01    acc:99.95\n",
      "loss: 1.23    acc:74.20\n",
      "epoch 238\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.41    acc:75.20\n",
      "epoch 239\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.38    acc:71.10\n",
      "epoch 240\n",
      "loss: 0.01    acc:99.82\n",
      "loss: 1.25    acc:73.10\n",
      "epoch 241\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:73.20\n",
      "epoch 242\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.22    acc:74.20\n",
      "epoch 243\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.36    acc:72.50\n",
      "epoch 244\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.28    acc:74.40\n",
      "epoch 245\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:73.70\n",
      "epoch 246\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.34    acc:72.50\n",
      "epoch 247\n",
      "loss: 0.00    acc:99.95\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.28    acc:74.30\n",
      "epoch 248\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.80\n",
      "epoch 249\n",
      "loss: 0.01    acc:99.95\n",
      "loss: 1.31    acc:73.60\n",
      "epoch 250\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.28    acc:74.50\n",
      "epoch 251\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.27    acc:74.10\n",
      "epoch 252\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:75.20\n",
      "epoch 253\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.25    acc:73.60\n",
      "epoch 254\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:72.10\n",
      "epoch 255\n",
      "loss: 0.01    acc:99.82\n",
      "loss: 1.23    acc:74.20\n",
      "epoch 256\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.18    acc:76.60\n",
      "epoch 257\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:73.50\n",
      "epoch 258\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:72.90\n",
      "epoch 259\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.43    acc:71.20\n",
      "epoch 260\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.32    acc:73.30\n",
      "epoch 261\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.20    acc:75.60\n",
      "epoch 262\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.24    acc:75.80\n",
      "epoch 263\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:74.70\n",
      "epoch 264\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.40\n",
      "epoch 265\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:75.00\n",
      "epoch 266\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.80\n",
      "epoch 267\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.16    acc:76.30\n",
      "epoch 268\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.28    acc:73.90\n",
      "epoch 269\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:72.70\n",
      "epoch 270\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.18    acc:75.10\n",
      "epoch 271\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.23    acc:75.50\n",
      "epoch 272\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.33    acc:72.20\n",
      "epoch 273\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.46    acc:73.30\n",
      "epoch 274\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.20\n",
      "epoch 275\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:73.60\n",
      "epoch 276\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.36    acc:75.00\n",
      "epoch 277\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:73.70\n",
      "epoch 278\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.45    acc:72.40\n",
      "epoch 279\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.22    acc:76.00\n",
      "epoch 280\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.37    acc:74.50\n",
      "epoch 281\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.25    acc:73.60\n",
      "epoch 282\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.38    acc:73.50\n",
      "epoch 283\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:73.70\n",
      "epoch 284\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.38    acc:73.50\n",
      "epoch 285\n",
      "loss: 0.01    acc:99.92\n",
      "loss: 1.27    acc:75.10\n",
      "epoch 286\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.26    acc:74.10\n",
      "epoch 287\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.48    acc:70.60\n",
      "epoch 288\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.33    acc:73.60\n",
      "Epoch   287: reducing learning rate of group 0 to 1.1111e-05.\n",
      "epoch 289\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:72.90\n",
      "epoch 290\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:74.00\n",
      "epoch 291\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:73.30\n",
      "epoch 292\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:73.60\n",
      "epoch 293\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:74.20\n",
      "epoch 294\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:73.90\n",
      "epoch 295\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:72.40\n",
      "epoch 296\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.80\n",
      "epoch 297\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.00\n",
      "epoch 298\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 299\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.28    acc:74.90\n",
      "epoch 300\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:76.20\n",
      "epoch 301\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:73.30\n",
      "epoch 302\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:75.20\n",
      "epoch 303\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:75.30\n",
      "epoch 304\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.20\n",
      "epoch 305\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:74.30\n",
      "epoch 306\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.40\n",
      "epoch 307\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:75.00\n",
      "epoch 308\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:73.20\n",
      "epoch 309\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.24    acc:75.20\n",
      "epoch 310\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:73.60\n",
      "epoch 311\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:74.00\n",
      "epoch 312\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.47    acc:72.20\n",
      "epoch 313\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:74.00\n",
      "epoch 314\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.34    acc:74.20\n",
      "epoch 315\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.30    acc:73.50\n",
      "epoch 316\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.41    acc:72.60\n",
      "epoch 317\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:74.40\n",
      "epoch 318\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.00\n",
      "epoch 319\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.21    acc:74.30\n",
      "epoch 320\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:72.90\n",
      "epoch 321\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:73.00\n",
      "epoch 322\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.20    acc:74.40\n",
      "Epoch   321: reducing learning rate of group 0 to 3.7037e-06.\n",
      "epoch 323\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.21    acc:76.30\n",
      "epoch 324\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:73.80\n",
      "epoch 325\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:73.40\n",
      "epoch 326\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:74.00\n",
      "epoch 327\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:74.20\n",
      "epoch 328\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.17    acc:74.10\n",
      "epoch 329\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.42    acc:73.60\n",
      "epoch 330\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 331\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:75.00\n",
      "epoch 332\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.00\n",
      "epoch 333\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:71.90\n",
      "epoch 334\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.26    acc:73.50\n",
      "epoch 335\n",
      "loss: 0.01    acc:99.87\n",
      "loss: 1.22    acc:75.10\n",
      "epoch 336\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:74.00\n",
      "epoch 337\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:74.50\n",
      "epoch 338\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.16    acc:75.30\n",
      "epoch 339\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.30\n",
      "epoch 340\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.90\n",
      "epoch 341\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:73.70\n",
      "epoch 342\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.10\n",
      "epoch 343\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:73.20\n",
      "epoch 344\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.40\n",
      "epoch 345\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.57    acc:71.50\n",
      "epoch 346\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:75.40\n",
      "epoch 347\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.52    acc:70.90\n",
      "epoch 348\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.60\n",
      "epoch 349\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:73.40\n",
      "epoch 350\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.44    acc:70.90\n",
      "epoch 351\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.44    acc:73.00\n",
      "epoch 352\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.70\n",
      "epoch 353\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:74.80\n",
      "epoch 354\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:72.20\n",
      "epoch 355\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:73.10\n",
      "epoch 356\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:73.00\n",
      "epoch 357\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:73.60\n",
      "epoch 358\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:73.80\n",
      "Epoch   357: reducing learning rate of group 0 to 1.2346e-06.\n",
      "epoch 359\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:73.70\n",
      "epoch 360\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.38    acc:72.70\n",
      "epoch 361\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:72.70\n",
      "epoch 362\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.13    acc:76.50\n",
      "epoch 363\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.40    acc:73.50\n",
      "epoch 364\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.20\n",
      "epoch 365\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:72.00\n",
      "epoch 366\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:72.30\n",
      "epoch 367\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:71.30\n",
      "epoch 368\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.10\n",
      "epoch 369\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:76.80\n",
      "epoch 370\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.44    acc:74.10\n",
      "epoch 371\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:77.50\n",
      "epoch 372\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.20    acc:76.20\n",
      "epoch 373\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.28    acc:75.90\n",
      "epoch 374\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:73.90\n",
      "epoch 375\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.20    acc:75.80\n",
      "epoch 376\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:73.30\n",
      "epoch 377\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.20\n",
      "epoch 378\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:74.40\n",
      "epoch 379\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.20\n",
      "epoch 380\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.10\n",
      "epoch 381\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:73.60\n",
      "epoch 382\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.40\n",
      "epoch 383\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:75.60\n",
      "epoch 384\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.30\n",
      "epoch 385\n",
      "loss: 0.00    acc:100.00\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 1.33    acc:73.40\n",
      "epoch 386\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.30\n",
      "epoch 387\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.35    acc:73.50\n",
      "epoch 388\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:73.10\n",
      "epoch 389\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.28    acc:72.80\n",
      "Epoch   388: reducing learning rate of group 0 to 4.1152e-07.\n",
      "epoch 390\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:74.70\n",
      "epoch 391\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:75.20\n",
      "epoch 392\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:74.40\n",
      "epoch 393\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.19    acc:76.10\n",
      "epoch 394\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.00\n",
      "epoch 395\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.10\n",
      "epoch 396\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.34    acc:73.50\n",
      "epoch 397\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.11    acc:76.50\n",
      "epoch 398\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:72.40\n",
      "epoch 399\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.40    acc:72.80\n",
      "epoch 400\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.00\n",
      "epoch 401\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:72.00\n",
      "epoch 402\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:73.20\n",
      "epoch 403\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:72.40\n",
      "epoch 404\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:73.20\n",
      "epoch 405\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.33    acc:73.20\n",
      "epoch 406\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.29    acc:74.40\n",
      "epoch 407\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:74.50\n",
      "epoch 408\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:72.90\n",
      "epoch 409\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:73.10\n",
      "epoch 410\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:74.40\n",
      "epoch 411\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:75.00\n",
      "epoch 412\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:74.50\n",
      "epoch 413\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:72.70\n",
      "Epoch   412: reducing learning rate of group 0 to 1.3717e-07.\n",
      "epoch 414\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:72.80\n",
      "epoch 415\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:74.70\n",
      "epoch 416\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.14    acc:77.20\n",
      "epoch 417\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:74.10\n",
      "epoch 418\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.48    acc:71.20\n",
      "epoch 419\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:72.20\n",
      "epoch 420\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:74.50\n",
      "epoch 421\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:72.50\n",
      "epoch 422\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:74.20\n",
      "epoch 423\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:73.70\n",
      "epoch 424\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:75.10\n",
      "epoch 425\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.20\n",
      "epoch 426\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.33    acc:74.60\n",
      "epoch 427\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.47    acc:70.90\n",
      "epoch 428\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:73.80\n",
      "epoch 429\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:73.20\n",
      "epoch 430\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:73.60\n",
      "epoch 431\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.10\n",
      "epoch 432\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.00\n",
      "epoch 433\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.22    acc:74.50\n",
      "epoch 434\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.20    acc:77.20\n",
      "Epoch   433: reducing learning rate of group 0 to 4.5725e-08.\n",
      "epoch 435\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.40    acc:73.20\n",
      "epoch 436\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.51    acc:71.10\n",
      "epoch 437\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.30\n",
      "epoch 438\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.20\n",
      "epoch 439\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:73.90\n",
      "epoch 440\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.39    acc:73.50\n",
      "epoch 441\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:73.40\n",
      "epoch 442\n",
      "loss: 0.00    acc:99.85\n",
      "loss: 1.43    acc:73.40\n",
      "epoch 443\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.90\n",
      "epoch 444\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.19    acc:74.80\n",
      "epoch 445\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.44    acc:72.70\n",
      "epoch 446\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:74.20\n",
      "epoch 447\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.00\n",
      "epoch 448\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.20\n",
      "epoch 449\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.30\n",
      "epoch 450\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:74.30\n",
      "epoch 451\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.19    acc:76.90\n",
      "epoch 452\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:71.80\n",
      "epoch 453\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.70\n",
      "epoch 454\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.60\n",
      "epoch 455\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:74.30\n",
      "Epoch   454: reducing learning rate of group 0 to 1.5242e-08.\n",
      "epoch 456\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:72.20\n",
      "epoch 457\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:72.90\n",
      "epoch 458\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.35    acc:72.40\n",
      "epoch 459\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:75.10\n",
      "epoch 460\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.49    acc:71.70\n",
      "epoch 461\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:72.70\n",
      "epoch 462\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.16    acc:76.30\n",
      "epoch 463\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.16    acc:74.60\n",
      "epoch 464\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.20\n",
      "epoch 465\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.15    acc:75.30\n",
      "epoch 466\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:71.80\n",
      "epoch 467\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.43    acc:74.70\n",
      "epoch 468\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:74.30\n",
      "epoch 469\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.15    acc:78.00\n",
      "epoch 470\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:72.70\n",
      "epoch 471\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:74.00\n",
      "epoch 472\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.27    acc:74.30\n",
      "epoch 473\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.70\n",
      "epoch 474\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:74.80\n",
      "epoch 475\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.18    acc:74.90\n",
      "epoch 476\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:73.80\n",
      "Epoch   475: reducing learning rate of group 0 to 5.0805e-09.\n",
      "epoch 477\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.90\n",
      "epoch 478\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.40\n",
      "epoch 479\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:72.20\n",
      "epoch 480\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.35    acc:71.50\n",
      "epoch 481\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.46    acc:73.80\n",
      "epoch 482\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:73.20\n",
      "epoch 483\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:73.50\n",
      "epoch 484\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:71.70\n",
      "epoch 485\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:73.50\n",
      "epoch 486\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.25    acc:74.30\n",
      "epoch 487\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:73.40\n",
      "epoch 488\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:73.70\n",
      "epoch 489\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:73.30\n",
      "epoch 490\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:72.50\n",
      "epoch 491\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:74.00\n",
      "epoch 492\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:74.20\n",
      "epoch 493\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.40\n",
      "epoch 494\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:75.50\n",
      "epoch 495\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:72.60\n",
      "epoch 496\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:70.20\n",
      "epoch 497\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.00\n",
      "epoch 498\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.00\n",
      "epoch 499\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.16    acc:74.70\n",
      "epoch 500\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:74.90\n",
      "epoch 501\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:74.60\n",
      "epoch 502\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:73.10\n",
      "epoch 503\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:74.40\n",
      "epoch 504\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.20    acc:75.20\n",
      "epoch 505\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:74.00\n",
      "epoch 506\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.10    acc:77.20\n",
      "epoch 507\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.10\n",
      "epoch 508\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.29    acc:74.30\n",
      "epoch 509\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.10\n",
      "epoch 510\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:73.90\n",
      "epoch 511\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:77.20\n",
      "epoch 512\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:73.90\n",
      "epoch 513\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:73.20\n",
      "epoch 514\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.42    acc:73.40\n",
      "epoch 515\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:73.00\n",
      "epoch 516\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.15    acc:75.10\n",
      "epoch 517\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.44    acc:72.10\n",
      "epoch 518\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:74.20\n",
      "epoch 519\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.46    acc:73.00\n",
      "epoch 520\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.18    acc:76.70\n",
      "epoch 521\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:75.00\n",
      "epoch 522\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:73.90\n",
      "epoch 523\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.58    acc:70.20\n",
      "epoch 524\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:72.90\n",
      "epoch 525\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:73.40\n",
      "epoch 526\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.70\n",
      "epoch 527\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.29    acc:75.30\n",
      "epoch 528\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.21    acc:75.40\n",
      "epoch 529\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:74.60\n",
      "epoch 530\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.10\n",
      "epoch 531\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 532\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.20\n",
      "epoch 533\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.10\n",
      "epoch 534\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:73.60\n",
      "epoch 535\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:74.50\n",
      "epoch 536\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.28    acc:73.90\n",
      "epoch 537\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:72.50\n",
      "epoch 538\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:73.20\n",
      "epoch 539\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.30\n",
      "epoch 540\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:72.00\n",
      "epoch 541\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:73.80\n",
      "epoch 542\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.46    acc:71.70\n",
      "epoch 543\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:73.50\n",
      "epoch 544\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.41    acc:72.90\n",
      "epoch 545\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:75.20\n",
      "epoch 546\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:74.60\n",
      "epoch 547\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:74.30\n",
      "epoch 548\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:72.80\n",
      "epoch 549\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.70\n",
      "epoch 550\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.40\n",
      "epoch 551\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.30\n",
      "epoch 552\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:74.20\n",
      "epoch 553\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.37    acc:73.70\n",
      "epoch 554\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.15    acc:76.20\n",
      "epoch 555\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:74.20\n",
      "epoch 556\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.47    acc:72.10\n",
      "epoch 557\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:74.30\n",
      "epoch 558\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:73.10\n",
      "epoch 559\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:72.50\n",
      "epoch 560\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:73.60\n",
      "epoch 561\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:74.10\n",
      "epoch 562\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.46    acc:74.00\n",
      "epoch 563\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:72.80\n",
      "epoch 564\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:71.80\n",
      "epoch 565\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.20    acc:75.40\n",
      "epoch 566\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.37    acc:75.70\n",
      "epoch 567\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.26    acc:75.60\n",
      "epoch 568\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:73.10\n",
      "epoch 569\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:74.30\n",
      "epoch 570\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.90\n",
      "epoch 571\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:75.30\n",
      "epoch 572\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.36    acc:73.60\n",
      "epoch 573\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.10\n",
      "epoch 574\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:72.30\n",
      "epoch 575\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.40    acc:73.60\n",
      "epoch 576\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.47    acc:72.80\n",
      "epoch 577\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.34    acc:72.50\n",
      "epoch 578\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.70\n",
      "epoch 579\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:75.20\n",
      "epoch 580\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:73.80\n",
      "epoch 581\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 582\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:75.30\n",
      "epoch 583\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:73.80\n",
      "epoch 584\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.47    acc:72.80\n",
      "epoch 585\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.50\n",
      "epoch 586\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:74.60\n",
      "epoch 587\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:72.30\n",
      "epoch 588\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.20\n",
      "epoch 589\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:74.30\n",
      "epoch 590\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:74.40\n",
      "epoch 591\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:74.60\n",
      "epoch 592\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.20\n",
      "epoch 593\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.00\n",
      "epoch 594\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:73.70\n",
      "epoch 595\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:73.70\n",
      "epoch 596\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.60\n",
      "epoch 597\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.10\n",
      "epoch 598\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.60\n",
      "epoch 599\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:71.70\n",
      "epoch 600\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:72.50\n",
      "epoch 601\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:74.00\n",
      "epoch 602\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:73.80\n",
      "epoch 603\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:72.70\n",
      "epoch 604\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:73.40\n",
      "epoch 605\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.49    acc:71.50\n",
      "epoch 606\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:74.30\n",
      "epoch 607\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:73.70\n",
      "epoch 608\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.30\n",
      "epoch 609\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:76.40\n",
      "epoch 610\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.60\n",
      "epoch 611\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:73.20\n",
      "epoch 612\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.00\n",
      "epoch 613\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:74.30\n",
      "epoch 614\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:75.00\n",
      "epoch 615\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:72.40\n",
      "epoch 616\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:72.90\n",
      "epoch 617\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:75.90\n",
      "epoch 618\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.42    acc:72.60\n",
      "epoch 619\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.10\n",
      "epoch 620\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.10\n",
      "epoch 621\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.42    acc:71.70\n",
      "epoch 622\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.19    acc:76.40\n",
      "epoch 623\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:74.20\n",
      "epoch 624\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.51    acc:72.10\n",
      "epoch 625\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.52    acc:71.00\n",
      "epoch 626\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:73.60\n",
      "epoch 627\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:74.60\n",
      "epoch 628\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:73.50\n",
      "epoch 629\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:73.90\n",
      "epoch 630\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:74.80\n",
      "epoch 631\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.52    acc:71.50\n",
      "epoch 632\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:72.80\n",
      "epoch 633\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.10\n",
      "epoch 634\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.41    acc:72.40\n",
      "epoch 635\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:76.40\n",
      "epoch 636\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:74.00\n",
      "epoch 637\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:74.50\n",
      "epoch 638\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.29    acc:74.90\n",
      "epoch 639\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:71.60\n",
      "epoch 640\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:75.80\n",
      "epoch 641\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.49    acc:71.70\n",
      "epoch 642\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.47    acc:71.40\n",
      "epoch 643\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:73.50\n",
      "epoch 644\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:75.40\n",
      "epoch 645\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:74.40\n",
      "epoch 646\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.18    acc:75.90\n",
      "epoch 647\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:74.40\n",
      "epoch 648\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:74.70\n",
      "epoch 649\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.24    acc:75.20\n",
      "epoch 650\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:72.00\n",
      "epoch 651\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:74.70\n",
      "epoch 652\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.39    acc:72.70\n",
      "epoch 653\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:75.50\n",
      "epoch 654\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.47    acc:73.20\n",
      "epoch 655\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:73.80\n",
      "epoch 656\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:75.70\n",
      "epoch 657\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:74.80\n",
      "epoch 658\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:73.70\n",
      "epoch 659\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:74.80\n",
      "epoch 660\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.70\n",
      "epoch 661\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:72.70\n",
      "epoch 662\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:75.10\n",
      "epoch 663\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.52    acc:71.90\n",
      "epoch 664\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:73.30\n",
      "epoch 665\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.24    acc:75.00\n",
      "epoch 666\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.80\n",
      "epoch 667\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.41    acc:72.70\n",
      "epoch 668\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:73.40\n",
      "epoch 669\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.27    acc:73.40\n",
      "epoch 670\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:74.60\n",
      "epoch 671\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.20    acc:75.90\n",
      "epoch 672\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.60\n",
      "epoch 673\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.00\n",
      "epoch 674\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:74.80\n",
      "epoch 675\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:72.50\n",
      "epoch 676\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:74.50\n",
      "epoch 677\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.47    acc:71.50\n",
      "epoch 678\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.40\n",
      "epoch 679\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.40\n",
      "epoch 680\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:75.30\n",
      "epoch 681\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:73.90\n",
      "epoch 682\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:72.70\n",
      "epoch 683\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:71.80\n",
      "epoch 684\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.70\n",
      "epoch 685\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.70\n",
      "epoch 686\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:75.30\n",
      "epoch 687\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:73.50\n",
      "epoch 688\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:73.60\n",
      "epoch 689\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.40\n",
      "epoch 690\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:75.30\n",
      "epoch 691\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:71.90\n",
      "epoch 692\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.40    acc:73.50\n",
      "epoch 693\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:75.60\n",
      "epoch 694\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.20\n",
      "epoch 695\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.60\n",
      "epoch 696\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.60\n",
      "epoch 697\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:76.10\n",
      "epoch 698\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:72.10\n",
      "epoch 699\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.30\n",
      "epoch 700\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:73.80\n",
      "epoch 701\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.40    acc:70.10\n",
      "epoch 702\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:75.30\n",
      "epoch 703\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.10\n",
      "epoch 704\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:73.60\n",
      "epoch 705\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.25    acc:74.00\n",
      "epoch 706\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:73.40\n",
      "epoch 707\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.41    acc:72.70\n",
      "epoch 708\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:76.30\n",
      "epoch 709\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.44    acc:73.10\n",
      "epoch 710\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:74.00\n",
      "epoch 711\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:72.70\n",
      "epoch 712\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.40\n",
      "epoch 713\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:73.90\n",
      "epoch 714\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.42    acc:72.40\n",
      "epoch 715\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.30\n",
      "epoch 716\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.50    acc:72.20\n",
      "epoch 717\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:74.90\n",
      "epoch 718\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.41    acc:73.00\n",
      "epoch 719\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.37    acc:72.10\n",
      "epoch 720\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.60\n",
      "epoch 721\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.42    acc:72.40\n",
      "epoch 722\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.43    acc:71.60\n",
      "epoch 723\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:74.50\n",
      "epoch 724\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.20\n",
      "epoch 725\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.36    acc:72.20\n",
      "epoch 726\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.50    acc:73.20\n",
      "epoch 727\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.60\n",
      "epoch 728\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:74.00\n",
      "epoch 729\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.33    acc:71.90\n",
      "epoch 730\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:74.00\n",
      "epoch 731\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.24    acc:73.80\n",
      "epoch 732\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.26    acc:74.20\n",
      "epoch 733\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.49    acc:70.70\n",
      "epoch 734\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.20\n",
      "epoch 735\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.30\n",
      "epoch 736\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.50\n",
      "epoch 737\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:73.30\n",
      "epoch 738\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:73.80\n",
      "epoch 739\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:74.00\n",
      "epoch 740\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:72.70\n",
      "epoch 741\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:74.30\n",
      "epoch 742\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:75.00\n",
      "epoch 743\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.40    acc:73.60\n",
      "epoch 744\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.15    acc:76.70\n",
      "epoch 745\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.90\n",
      "epoch 746\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:74.50\n",
      "epoch 747\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.52    acc:70.20\n",
      "epoch 748\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:72.80\n",
      "epoch 749\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.70\n",
      "epoch 750\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:73.10\n",
      "epoch 751\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:74.50\n",
      "epoch 752\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.49    acc:70.20\n",
      "epoch 753\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.36    acc:74.00\n",
      "epoch 754\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.60\n",
      "epoch 755\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:74.90\n",
      "epoch 756\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.24    acc:75.30\n",
      "epoch 757\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:71.40\n",
      "epoch 758\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.00\n",
      "epoch 759\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.17    acc:73.70\n",
      "epoch 760\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.70\n",
      "epoch 761\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:73.40\n",
      "epoch 762\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:73.90\n",
      "epoch 763\n",
      "loss: 0.00    acc:99.85\n",
      "loss: 1.17    acc:75.60\n",
      "epoch 764\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:73.30\n",
      "epoch 765\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:75.60\n",
      "epoch 766\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:73.00\n",
      "epoch 767\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.36    acc:74.00\n",
      "epoch 768\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:73.50\n",
      "epoch 769\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.35    acc:74.30\n",
      "epoch 770\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:75.10\n",
      "epoch 771\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:75.00\n",
      "epoch 772\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:73.50\n",
      "epoch 773\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.23    acc:74.20\n",
      "epoch 774\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:74.50\n",
      "epoch 775\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:73.80\n",
      "epoch 776\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:72.90\n",
      "epoch 777\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:74.90\n",
      "epoch 778\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.80\n",
      "epoch 779\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:72.10\n",
      "epoch 780\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:72.10\n",
      "epoch 781\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:72.90\n",
      "epoch 782\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.60\n",
      "epoch 783\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.47    acc:73.70\n",
      "epoch 784\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:73.70\n",
      "epoch 785\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.41    acc:71.60\n",
      "epoch 786\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.70\n",
      "epoch 787\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.90\n",
      "epoch 788\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.24    acc:73.60\n",
      "epoch 789\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:71.90\n",
      "epoch 790\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:74.10\n",
      "epoch 791\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:74.90\n",
      "epoch 792\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:73.20\n",
      "epoch 793\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.80\n",
      "epoch 794\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:73.60\n",
      "epoch 795\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:75.80\n",
      "epoch 796\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:75.90\n",
      "epoch 797\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.47    acc:71.70\n",
      "epoch 798\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:73.50\n",
      "epoch 799\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:72.90\n",
      "epoch 800\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.19    acc:75.30\n",
      "epoch 801\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.28    acc:75.00\n",
      "epoch 802\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.10\n",
      "epoch 803\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:99.97\n",
      "loss: 1.38    acc:72.50\n",
      "epoch 804\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:74.60\n",
      "epoch 805\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:73.50\n",
      "epoch 806\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.41    acc:73.00\n",
      "epoch 807\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.32    acc:74.00\n",
      "epoch 808\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.30    acc:74.20\n",
      "epoch 809\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.20    acc:74.00\n",
      "epoch 810\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.60\n",
      "epoch 811\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.30\n",
      "epoch 812\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:72.60\n",
      "epoch 813\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:72.80\n",
      "epoch 814\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:74.30\n",
      "epoch 815\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.50\n",
      "epoch 816\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.49    acc:72.80\n",
      "epoch 817\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:73.90\n",
      "epoch 818\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:71.90\n",
      "epoch 819\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.26    acc:75.40\n",
      "epoch 820\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.40    acc:74.00\n",
      "epoch 821\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.54    acc:70.00\n",
      "epoch 822\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:72.60\n",
      "epoch 823\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.12    acc:76.10\n",
      "epoch 824\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.20\n",
      "epoch 825\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:73.70\n",
      "epoch 826\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.10\n",
      "epoch 827\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.40    acc:72.80\n",
      "epoch 828\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.27    acc:74.50\n",
      "epoch 829\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.60\n",
      "epoch 830\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:75.80\n",
      "epoch 831\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.43    acc:74.10\n",
      "epoch 832\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:73.80\n",
      "epoch 833\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.20    acc:75.30\n",
      "epoch 834\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.39    acc:71.80\n",
      "epoch 835\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:75.30\n",
      "epoch 836\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.16    acc:75.30\n",
      "epoch 837\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.48    acc:72.40\n",
      "epoch 838\n",
      "loss: 0.00    acc:99.82\n",
      "loss: 1.23    acc:72.40\n",
      "epoch 839\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.22    acc:74.90\n",
      "epoch 840\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.45    acc:72.90\n",
      "epoch 841\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.10\n",
      "epoch 842\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:74.60\n",
      "epoch 843\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:72.80\n",
      "epoch 844\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:73.30\n",
      "epoch 845\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:75.40\n",
      "epoch 846\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:72.70\n",
      "epoch 847\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.24    acc:76.50\n",
      "epoch 848\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.41    acc:72.60\n",
      "epoch 849\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.90\n",
      "epoch 850\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.10\n",
      "epoch 851\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:74.80\n",
      "epoch 852\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.16    acc:76.50\n",
      "epoch 853\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.50    acc:71.00\n",
      "epoch 854\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:74.40\n",
      "epoch 855\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:75.10\n",
      "epoch 856\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.26    acc:74.30\n",
      "epoch 857\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.47    acc:72.30\n",
      "epoch 858\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.80\n",
      "epoch 859\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.20\n",
      "epoch 860\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.38    acc:73.20\n",
      "epoch 861\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.48    acc:71.60\n",
      "epoch 862\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.30\n",
      "epoch 863\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.45    acc:73.90\n",
      "epoch 864\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.26    acc:74.60\n",
      "epoch 865\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:73.60\n",
      "epoch 866\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:75.00\n",
      "epoch 867\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.90\n",
      "epoch 868\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:75.60\n",
      "epoch 869\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.70\n",
      "epoch 870\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:73.40\n",
      "epoch 871\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.40\n",
      "epoch 872\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:74.50\n",
      "epoch 873\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.10\n",
      "epoch 874\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:73.30\n",
      "epoch 875\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.50\n",
      "epoch 876\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.80\n",
      "epoch 877\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.30    acc:74.60\n",
      "epoch 878\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:73.80\n",
      "epoch 879\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:74.40\n",
      "epoch 880\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:74.60\n",
      "epoch 881\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.43    acc:72.20\n",
      "epoch 882\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.21    acc:74.70\n",
      "epoch 883\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.50    acc:71.50\n",
      "epoch 884\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:72.90\n",
      "epoch 885\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:73.40\n",
      "epoch 886\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:74.20\n",
      "epoch 887\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:74.70\n",
      "epoch 888\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.70\n",
      "epoch 889\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:74.30\n",
      "epoch 890\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:74.60\n",
      "epoch 891\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:75.60\n",
      "epoch 892\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:74.50\n",
      "epoch 893\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.14    acc:76.60\n",
      "epoch 894\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:73.20\n",
      "epoch 895\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:73.00\n",
      "epoch 896\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.39    acc:72.90\n",
      "epoch 897\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:74.30\n",
      "epoch 898\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:73.30\n",
      "epoch 899\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.90\n",
      "epoch 900\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:74.10\n",
      "epoch 901\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.34    acc:72.90\n",
      "epoch 902\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:74.10\n",
      "epoch 903\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.37    acc:73.80\n",
      "epoch 904\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:73.50\n",
      "epoch 905\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.34    acc:75.50\n",
      "epoch 906\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.20    acc:76.60\n",
      "epoch 907\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.51    acc:72.60\n",
      "epoch 908\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:75.70\n",
      "epoch 909\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.00\n",
      "epoch 910\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.30    acc:73.40\n",
      "epoch 911\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.36    acc:75.60\n",
      "epoch 912\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:72.90\n",
      "epoch 913\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.33    acc:75.90\n",
      "epoch 914\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.00\n",
      "epoch 915\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.39    acc:73.30\n",
      "epoch 916\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:72.80\n",
      "epoch 917\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:74.40\n",
      "epoch 918\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.24    acc:74.20\n",
      "epoch 919\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:74.70\n",
      "epoch 920\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.43    acc:72.20\n",
      "epoch 921\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:75.50\n",
      "epoch 922\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.21    acc:75.10\n",
      "epoch 923\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:74.30\n",
      "epoch 924\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.26    acc:75.60\n",
      "epoch 925\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:74.60\n",
      "epoch 926\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.39    acc:74.40\n",
      "epoch 927\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:73.20\n",
      "epoch 928\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.28    acc:74.40\n",
      "epoch 929\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:75.50\n",
      "epoch 930\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.20    acc:75.60\n",
      "epoch 931\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.46    acc:72.70\n",
      "epoch 932\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:74.30\n",
      "epoch 933\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.42    acc:71.80\n",
      "epoch 934\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:72.90\n",
      "epoch 935\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.21    acc:75.40\n",
      "epoch 936\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.50\n",
      "epoch 937\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:74.30\n",
      "epoch 938\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:73.30\n",
      "epoch 939\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.35    acc:72.30\n",
      "epoch 940\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.37    acc:74.40\n",
      "epoch 941\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.49    acc:73.40\n",
      "epoch 942\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:74.70\n",
      "epoch 943\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:71.80\n",
      "epoch 944\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:74.70\n",
      "epoch 945\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:73.50\n",
      "epoch 946\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.19    acc:74.50\n",
      "epoch 947\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:73.70\n",
      "epoch 948\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.28    acc:75.30\n",
      "epoch 949\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:73.10\n",
      "epoch 950\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.29    acc:75.40\n",
      "epoch 951\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:73.70\n",
      "epoch 952\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.40    acc:72.70\n",
      "epoch 953\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.29    acc:73.90\n",
      "epoch 954\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:72.80\n",
      "epoch 955\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.31    acc:74.60\n",
      "epoch 956\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:74.80\n",
      "epoch 957\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.00\n",
      "epoch 958\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.39    acc:73.60\n",
      "epoch 959\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.36    acc:74.20\n",
      "epoch 960\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:74.70\n",
      "epoch 961\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:72.50\n",
      "epoch 962\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.22    acc:75.40\n",
      "epoch 963\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:72.60\n",
      "epoch 964\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.20\n",
      "epoch 965\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.49    acc:71.20\n",
      "epoch 966\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:73.80\n",
      "epoch 967\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.23    acc:74.70\n",
      "epoch 968\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:75.00\n",
      "epoch 969\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.16    acc:76.20\n",
      "epoch 970\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.47    acc:71.80\n",
      "epoch 971\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:73.80\n",
      "epoch 972\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:73.10\n",
      "epoch 973\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:73.60\n",
      "epoch 974\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.00\n",
      "epoch 975\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.15    acc:75.50\n",
      "epoch 976\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.45    acc:72.00\n",
      "epoch 977\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:76.40\n",
      "epoch 978\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:72.90\n",
      "epoch 979\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:72.80\n",
      "epoch 980\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.38    acc:73.20\n",
      "epoch 981\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:74.00\n",
      "epoch 982\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.40    acc:73.60\n",
      "epoch 983\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.28    acc:73.90\n",
      "epoch 984\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.47    acc:72.20\n",
      "epoch 985\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:72.20\n",
      "epoch 986\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:73.40\n",
      "epoch 987\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.15    acc:78.10\n",
      "epoch 988\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.37    acc:74.00\n",
      "epoch 989\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.33    acc:73.20\n",
      "epoch 990\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.00\n",
      "epoch 991\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.35    acc:74.50\n",
      "epoch 992\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:73.70\n",
      "epoch 993\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:74.30\n",
      "epoch 994\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.46    acc:70.70\n",
      "epoch 995\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.31    acc:74.50\n",
      "epoch 996\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:72.80\n",
      "epoch 997\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.30    acc:75.00\n",
      "epoch 998\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.16    acc:77.70\n",
      "epoch 999\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:73.50\n",
      "epoch 1000\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.20    acc:74.60\n",
      "256\n",
      "9985034\n",
      "epoch 1\n",
      "loss: 1.86    acc:30.92\n",
      "loss: 2.01    acc:24.10\n",
      "epoch 2\n",
      "loss: 1.58    acc:43.20\n",
      "loss: 1.60    acc:40.40\n",
      "epoch 3\n",
      "loss: 1.46    acc:47.52\n",
      "loss: 1.61    acc:41.90\n",
      "epoch 4\n",
      "loss: 1.32    acc:53.20\n",
      "loss: 1.83    acc:40.10\n",
      "epoch 5\n",
      "loss: 1.26    acc:55.67\n",
      "loss: 1.42    acc:48.20\n",
      "epoch 6\n",
      "loss: 1.17    acc:58.32\n",
      "loss: 1.37    acc:53.10\n",
      "epoch 7\n",
      "loss: 1.10    acc:61.37\n",
      "loss: 1.35    acc:51.40\n",
      "epoch 8\n",
      "loss: 1.10    acc:61.62\n",
      "loss: 1.33    acc:54.30\n",
      "epoch 9\n",
      "loss: 1.01    acc:63.87\n",
      "loss: 1.41    acc:53.10\n",
      "epoch 10\n",
      "loss: 0.97    acc:65.60\n",
      "loss: 1.36    acc:53.00\n",
      "epoch 11\n",
      "loss: 0.92    acc:67.52\n",
      "loss: 1.18    acc:57.40\n",
      "epoch 12\n",
      "loss: 0.89    acc:67.72\n",
      "loss: 1.45    acc:49.40\n",
      "epoch 13\n",
      "loss: 0.85    acc:70.20\n",
      "loss: 1.06    acc:61.70\n",
      "epoch 14\n",
      "loss: 0.80    acc:71.50\n",
      "loss: 1.33    acc:56.80\n",
      "epoch 15\n",
      "loss: 0.77    acc:72.57\n",
      "loss: 1.27    acc:56.70\n",
      "epoch 16\n",
      "loss: 0.75    acc:73.12\n",
      "loss: 1.17    acc:59.60\n",
      "epoch 17\n",
      "loss: 0.71    acc:75.07\n",
      "loss: 1.12    acc:60.60\n",
      "epoch 18\n",
      "loss: 0.67    acc:76.77\n",
      "loss: 1.12    acc:62.20\n",
      "epoch 19\n",
      "loss: 0.63    acc:77.80\n",
      "loss: 1.07    acc:64.50\n",
      "epoch 20\n",
      "loss: 0.58    acc:80.12\n",
      "loss: 1.05    acc:64.50\n",
      "epoch 21\n",
      "loss: 0.58    acc:79.92\n",
      "loss: 1.09    acc:63.00\n",
      "epoch 22\n",
      "loss: 0.52    acc:81.87\n",
      "loss: 0.92    acc:68.10\n",
      "epoch 23\n",
      "loss: 0.53    acc:81.52\n",
      "loss: 0.93    acc:69.30\n",
      "epoch 24\n",
      "loss: 0.48    acc:83.67\n",
      "loss: 0.98    acc:67.90\n",
      "epoch 25\n",
      "loss: 0.45    acc:84.10\n",
      "loss: 1.06    acc:66.40\n",
      "epoch 26\n",
      "loss: 0.43    acc:85.15\n",
      "loss: 1.04    acc:65.60\n",
      "epoch 27\n",
      "loss: 0.38    acc:87.12\n",
      "loss: 0.97    acc:65.90\n",
      "epoch 28\n",
      "loss: 0.37    acc:88.05\n",
      "loss: 0.87    acc:71.00\n",
      "epoch 29\n",
      "loss: 0.33    acc:89.22\n",
      "loss: 1.59    acc:56.60\n",
      "epoch 30\n",
      "loss: 0.31    acc:90.25\n",
      "loss: 1.25    acc:64.80\n",
      "epoch 31\n",
      "loss: 0.28    acc:90.85\n",
      "loss: 1.30    acc:62.50\n",
      "epoch 32\n",
      "loss: 0.27    acc:91.50\n",
      "loss: 1.04    acc:69.40\n",
      "epoch 33\n",
      "loss: 0.27    acc:91.07\n",
      "loss: 1.46    acc:60.50\n",
      "epoch 34\n",
      "loss: 0.26    acc:91.37\n",
      "loss: 1.24    acc:63.30\n",
      "epoch 35\n",
      "loss: 0.22    acc:92.77\n",
      "loss: 1.06    acc:68.70\n",
      "epoch 36\n",
      "loss: 0.20    acc:93.60\n",
      "loss: 1.09    acc:68.20\n",
      "epoch 37\n",
      "loss: 0.16    acc:95.15\n",
      "loss: 1.28    acc:65.70\n",
      "epoch 38\n",
      "loss: 0.16    acc:95.47\n",
      "loss: 1.15    acc:67.70\n",
      "epoch 39\n",
      "loss: 0.14    acc:95.80\n",
      "loss: 1.15    acc:70.40\n",
      "epoch 40\n",
      "loss: 0.14    acc:95.70\n",
      "loss: 1.08    acc:68.30\n",
      "epoch 41\n",
      "loss: 0.12    acc:96.72\n",
      "loss: 1.02    acc:71.70\n",
      "epoch 42\n",
      "loss: 0.13    acc:96.07\n",
      "loss: 1.05    acc:69.70\n",
      "epoch 43\n",
      "loss: 0.13    acc:96.32\n",
      "loss: 0.99    acc:69.80\n",
      "epoch 44\n",
      "loss: 0.10    acc:97.20\n",
      "loss: 0.92    acc:74.00\n",
      "epoch 45\n",
      "loss: 0.10    acc:97.12\n",
      "loss: 0.92    acc:73.20\n",
      "epoch 46\n",
      "loss: 0.10    acc:97.17\n",
      "loss: 1.21    acc:68.70\n",
      "epoch 47\n",
      "loss: 0.09    acc:97.82\n",
      "loss: 1.06    acc:72.00\n",
      "epoch 48\n",
      "loss: 0.09    acc:97.55\n",
      "loss: 1.03    acc:72.80\n",
      "epoch 49\n",
      "loss: 0.08    acc:97.97\n",
      "loss: 1.32    acc:67.70\n",
      "epoch 50\n",
      "loss: 0.08    acc:97.67\n",
      "loss: 1.04    acc:72.10\n",
      "epoch 51\n",
      "loss: 0.08    acc:97.57\n",
      "loss: 1.18    acc:69.30\n",
      "epoch 52\n",
      "loss: 0.10    acc:97.10\n",
      "loss: 1.47    acc:65.00\n",
      "epoch 53\n",
      "loss: 0.09    acc:97.22\n",
      "loss: 1.22    acc:69.50\n",
      "epoch 54\n",
      "loss: 0.08    acc:97.47\n",
      "loss: 1.19    acc:70.70\n",
      "epoch 55\n",
      "loss: 0.08    acc:97.62\n",
      "loss: 1.05    acc:72.40\n",
      "epoch 56\n",
      "loss: 0.06    acc:98.40\n",
      "loss: 1.12    acc:71.00\n",
      "epoch 57\n",
      "loss: 0.05    acc:98.57\n",
      "loss: 1.24    acc:69.50\n",
      "epoch 58\n",
      "loss: 0.05    acc:98.60\n",
      "loss: 1.21    acc:70.80\n",
      "epoch 59\n",
      "loss: 0.05    acc:98.82\n",
      "loss: 1.14    acc:71.80\n",
      "epoch 60\n",
      "loss: 0.05    acc:98.92\n",
      "loss: 1.26    acc:71.00\n",
      "epoch 61\n",
      "loss: 0.04    acc:98.95\n",
      "loss: 1.09    acc:73.30\n",
      "epoch 62\n",
      "loss: 0.04    acc:98.92\n",
      "loss: 1.14    acc:72.30\n",
      "epoch 63\n",
      "loss: 0.05    acc:98.47\n",
      "loss: 1.69    acc:66.70\n",
      "epoch 64\n",
      "loss: 0.07    acc:97.82\n",
      "loss: 1.36    acc:69.60\n",
      "epoch 65\n",
      "loss: 0.06    acc:98.00\n",
      "loss: 1.27    acc:71.00\n",
      "epoch 66\n",
      "loss: 0.05    acc:98.27\n",
      "loss: 1.30    acc:71.60\n",
      "epoch 67\n",
      "loss: 0.06    acc:98.52\n",
      "loss: 1.18    acc:72.90\n",
      "epoch 68\n",
      "loss: 0.05    acc:98.37\n",
      "loss: 1.11    acc:73.10\n",
      "epoch 69\n",
      "loss: 0.05    acc:98.35\n",
      "loss: 1.52    acc:67.10\n",
      "epoch 70\n",
      "loss: 0.06    acc:98.35\n",
      "loss: 1.53    acc:69.10\n",
      "epoch 71\n",
      "loss: 0.05    acc:98.45\n",
      "loss: 1.17    acc:73.70\n",
      "epoch 72\n",
      "loss: 0.04    acc:98.87\n",
      "loss: 1.11    acc:72.70\n",
      "epoch 73\n",
      "loss: 0.03    acc:99.37\n",
      "loss: 1.07    acc:73.90\n",
      "epoch 74\n",
      "loss: 0.02    acc:99.62\n",
      "loss: 1.11    acc:74.20\n",
      "epoch 75\n",
      "loss: 0.03    acc:99.27\n",
      "loss: 1.23    acc:72.50\n",
      "epoch 76\n",
      "loss: 0.04    acc:98.70\n",
      "loss: 1.22    acc:70.80\n",
      "epoch 77\n",
      "loss: 0.07    acc:97.72\n",
      "loss: 1.52    acc:68.50\n",
      "epoch 78\n",
      "loss: 0.09    acc:97.02\n",
      "loss: 1.49    acc:66.20\n",
      "epoch 79\n",
      "loss: 0.07    acc:97.50\n",
      "loss: 1.29    acc:69.40\n",
      "epoch 80\n",
      "loss: 0.05    acc:98.42\n",
      "loss: 1.17    acc:71.90\n",
      "epoch 81\n",
      "loss: 0.04    acc:98.67\n",
      "loss: 1.37    acc:70.00\n",
      "epoch 82\n",
      "loss: 0.05    acc:98.25\n",
      "loss: 1.21    acc:73.00\n",
      "epoch 83\n",
      "loss: 0.05    acc:98.52\n",
      "loss: 1.06    acc:74.00\n",
      "epoch 84\n",
      "loss: 0.04    acc:98.95\n",
      "loss: 1.36    acc:68.80\n",
      "epoch 85\n",
      "loss: 0.03    acc:99.42\n",
      "loss: 1.25    acc:71.80\n",
      "epoch 86\n",
      "loss: 0.02    acc:99.42\n",
      "loss: 1.05    acc:77.60\n",
      "epoch 87\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.02    acc:99.52\n",
      "loss: 1.20    acc:72.70\n",
      "epoch 88\n",
      "loss: 0.02    acc:99.47\n",
      "loss: 1.08    acc:75.10\n",
      "epoch 89\n",
      "loss: 0.02    acc:99.47\n",
      "loss: 1.40    acc:71.00\n",
      "epoch 90\n",
      "loss: 0.02    acc:99.60\n",
      "loss: 1.12    acc:74.30\n",
      "epoch 91\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.24    acc:72.60\n",
      "epoch 92\n",
      "loss: 0.02    acc:99.57\n",
      "loss: 1.24    acc:72.80\n",
      "epoch 93\n",
      "loss: 0.03    acc:99.05\n",
      "loss: 1.22    acc:72.20\n",
      "epoch 94\n",
      "loss: 0.04    acc:98.60\n",
      "loss: 1.29    acc:72.70\n",
      "epoch 95\n",
      "loss: 0.04    acc:98.87\n",
      "loss: 1.29    acc:72.30\n",
      "epoch 96\n",
      "loss: 0.03    acc:99.12\n",
      "loss: 1.42    acc:70.30\n",
      "epoch 97\n",
      "loss: 0.05    acc:98.25\n",
      "loss: 1.55    acc:68.90\n",
      "epoch 98\n",
      "loss: 0.05    acc:98.55\n",
      "loss: 1.62    acc:69.10\n",
      "epoch 99\n",
      "loss: 0.07    acc:97.42\n",
      "loss: 1.40    acc:69.30\n",
      "epoch 100\n",
      "loss: 0.07    acc:97.57\n",
      "loss: 1.45    acc:68.70\n",
      "epoch 101\n",
      "loss: 0.08    acc:97.30\n",
      "loss: 1.43    acc:68.40\n",
      "epoch 102\n",
      "loss: 0.06    acc:98.00\n",
      "loss: 1.39    acc:72.10\n",
      "epoch 103\n",
      "loss: 0.03    acc:99.00\n",
      "loss: 1.36    acc:71.70\n",
      "epoch 104\n",
      "loss: 0.03    acc:98.95\n",
      "loss: 1.31    acc:73.40\n",
      "epoch 105\n",
      "loss: 0.02    acc:99.55\n",
      "loss: 1.24    acc:72.50\n",
      "epoch 106\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.08    acc:74.60\n",
      "epoch 107\n",
      "loss: 0.02    acc:99.40\n",
      "loss: 1.38    acc:72.20\n",
      "epoch 108\n",
      "loss: 0.01    acc:99.70\n",
      "loss: 1.34    acc:71.60\n",
      "epoch 109\n",
      "loss: 0.01    acc:99.77\n",
      "loss: 1.21    acc:72.70\n",
      "epoch 110\n",
      "loss: 0.02    acc:99.55\n",
      "loss: 1.38    acc:71.90\n",
      "epoch 111\n",
      "loss: 0.02    acc:99.25\n",
      "loss: 1.31    acc:72.70\n",
      "epoch 112\n",
      "loss: 0.02    acc:99.62\n",
      "loss: 1.35    acc:70.80\n",
      "epoch 113\n",
      "loss: 0.01    acc:99.62\n",
      "loss: 1.34    acc:72.50\n",
      "epoch 114\n",
      "loss: 0.01    acc:99.70\n",
      "loss: 1.17    acc:74.50\n",
      "epoch 115\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.38    acc:74.20\n",
      "epoch 116\n",
      "loss: 0.03    acc:99.10\n",
      "loss: 1.42    acc:72.40\n",
      "epoch 117\n",
      "loss: 0.04    acc:98.82\n",
      "loss: 1.50    acc:70.90\n",
      "epoch 118\n",
      "loss: 0.04    acc:98.52\n",
      "loss: 1.45    acc:70.30\n",
      "epoch 119\n",
      "loss: 0.06    acc:98.12\n",
      "loss: 1.46    acc:70.10\n",
      "epoch 120\n",
      "loss: 0.05    acc:98.47\n",
      "loss: 1.47    acc:72.60\n",
      "epoch 121\n",
      "loss: 0.03    acc:99.12\n",
      "loss: 1.29    acc:72.10\n",
      "epoch 122\n",
      "loss: 0.02    acc:99.25\n",
      "loss: 1.22    acc:74.60\n",
      "epoch 123\n",
      "loss: 0.02    acc:99.57\n",
      "loss: 0.97    acc:77.00\n",
      "epoch 124\n",
      "loss: 0.01    acc:99.75\n",
      "loss: 1.35    acc:74.00\n",
      "epoch 125\n",
      "loss: 0.02    acc:99.37\n",
      "loss: 1.31    acc:73.10\n",
      "epoch 126\n",
      "loss: 0.04    acc:98.92\n",
      "loss: 1.45    acc:71.90\n",
      "epoch 127\n",
      "loss: 0.02    acc:99.30\n",
      "loss: 1.46    acc:71.20\n",
      "epoch 128\n",
      "loss: 0.02    acc:99.52\n",
      "loss: 1.22    acc:73.70\n",
      "epoch 129\n",
      "loss: 0.02    acc:99.52\n",
      "loss: 1.24    acc:75.20\n",
      "epoch 130\n",
      "loss: 0.02    acc:99.12\n",
      "loss: 1.23    acc:74.10\n",
      "epoch 131\n",
      "loss: 0.02    acc:99.12\n",
      "loss: 1.21    acc:74.60\n",
      "epoch 132\n",
      "loss: 0.02    acc:99.30\n",
      "loss: 1.38    acc:72.20\n",
      "epoch 133\n",
      "loss: 0.03    acc:99.20\n",
      "loss: 1.22    acc:74.50\n",
      "epoch 134\n",
      "loss: 0.02    acc:99.37\n",
      "loss: 1.35    acc:73.50\n",
      "epoch 135\n",
      "loss: 0.02    acc:99.52\n",
      "loss: 1.40    acc:73.70\n",
      "epoch 136\n",
      "loss: 0.02    acc:99.50\n",
      "loss: 1.42    acc:72.40\n",
      "Epoch   135: reducing learning rate of group 0 to 1.0000e-04.\n",
      "epoch 137\n",
      "loss: 0.01    acc:99.55\n",
      "loss: 1.16    acc:76.10\n",
      "epoch 138\n",
      "loss: 0.01    acc:99.85\n",
      "loss: 1.22    acc:74.10\n",
      "epoch 139\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:76.50\n",
      "epoch 140\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.13    acc:76.70\n",
      "epoch 141\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.10\n",
      "epoch 142\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.25    acc:75.00\n",
      "epoch 143\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.18    acc:76.70\n",
      "epoch 144\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.16    acc:76.30\n",
      "epoch 145\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.90\n",
      "epoch 146\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.13    acc:76.40\n",
      "epoch 147\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.05    acc:77.10\n",
      "epoch 148\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.11    acc:76.60\n",
      "epoch 149\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:75.90\n",
      "epoch 150\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.23    acc:75.20\n",
      "epoch 151\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.23    acc:74.40\n",
      "epoch 152\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:75.20\n",
      "epoch 153\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:74.50\n",
      "epoch 154\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.27    acc:74.90\n",
      "epoch 155\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:75.60\n",
      "epoch 156\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.20\n",
      "epoch 157\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:75.60\n",
      "epoch 158\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.05    acc:78.20\n",
      "epoch 159\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.00\n",
      "epoch 160\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.11    acc:77.30\n",
      "epoch 161\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.00    acc:78.50\n",
      "epoch 162\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.00\n",
      "epoch 163\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.80\n",
      "epoch 164\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.04    acc:78.40\n",
      "epoch 165\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.80\n",
      "epoch 166\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:76.90\n",
      "epoch 167\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.20\n",
      "epoch 168\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:73.20\n",
      "epoch 169\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.10\n",
      "epoch 170\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.80\n",
      "epoch 171\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.70\n",
      "epoch 172\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.60\n",
      "epoch 173\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:75.50\n",
      "epoch 174\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.41    acc:73.10\n",
      "epoch 175\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.80\n",
      "epoch 176\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:76.10\n",
      "epoch 177\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.60\n",
      "epoch 178\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:75.90\n",
      "epoch 179\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.02    acc:80.10\n",
      "epoch 180\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:76.70\n",
      "epoch 181\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.20\n",
      "epoch 182\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.40\n",
      "epoch 183\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.10\n",
      "epoch 184\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.07    acc:75.70\n",
      "epoch 185\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.90\n",
      "epoch 186\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.10\n",
      "epoch 187\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.80\n",
      "epoch 188\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.00\n",
      "epoch 189\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:79.80\n",
      "epoch 190\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.60\n",
      "epoch 191\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:75.80\n",
      "epoch 192\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.05    acc:78.70\n",
      "epoch 193\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.13    acc:76.00\n",
      "epoch 194\n",
      "loss: 0.01    acc:99.75\n",
      "loss: 1.24    acc:74.20\n",
      "epoch 195\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.21    acc:76.60\n",
      "epoch 196\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.25    acc:75.30\n",
      "epoch 197\n",
      "loss: 0.00    acc:99.85\n",
      "loss: 1.40    acc:74.10\n",
      "epoch 198\n",
      "loss: 0.01    acc:99.70\n",
      "loss: 1.32    acc:73.40\n",
      "epoch 199\n",
      "loss: 0.01    acc:99.90\n",
      "loss: 1.21    acc:75.50\n",
      "epoch 200\n",
      "loss: 0.00    acc:99.90\n",
      "loss: 1.39    acc:74.10\n",
      "epoch 201\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.25    acc:74.50\n",
      "epoch 202\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:74.10\n",
      "epoch 203\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.26    acc:75.40\n",
      "epoch 204\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.26    acc:74.10\n",
      "epoch 205\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.00\n",
      "epoch 206\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:75.10\n",
      "epoch 207\n",
      "loss: 0.00    acc:99.82\n",
      "loss: 1.43    acc:72.10\n",
      "epoch 208\n",
      "loss: 0.01    acc:99.80\n",
      "loss: 1.31    acc:74.90\n",
      "epoch 209\n",
      "loss: 0.00    acc:99.87\n",
      "loss: 1.21    acc:76.20\n",
      "epoch 210\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.25    acc:73.90\n",
      "epoch 211\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.37    acc:73.70\n",
      "epoch 212\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.13    acc:76.60\n",
      "Epoch   211: reducing learning rate of group 0 to 3.3333e-05.\n",
      "epoch 213\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 214\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.60\n",
      "epoch 215\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:76.10\n",
      "epoch 216\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:77.00\n",
      "epoch 217\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.50\n",
      "epoch 218\n",
      "loss: 0.00    acc:99.92\n",
      "loss: 1.32    acc:75.10\n",
      "epoch 219\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.10\n",
      "epoch 220\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:74.10\n",
      "epoch 221\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.00\n",
      "epoch 222\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.08    acc:79.90\n",
      "epoch 223\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:74.60\n",
      "epoch 224\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.70\n",
      "epoch 225\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.90\n",
      "epoch 226\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.20\n",
      "epoch 227\n",
      "loss: 0.00    acc:99.95\n",
      "loss: 1.32    acc:76.60\n",
      "epoch 228\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.70\n",
      "epoch 229\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.30\n",
      "epoch 230\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:76.60\n",
      "epoch 231\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.00\n",
      "epoch 232\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:75.60\n",
      "epoch 233\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:75.80\n",
      "Epoch   232: reducing learning rate of group 0 to 1.1111e-05.\n",
      "epoch 234\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.50\n",
      "epoch 235\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.70\n",
      "epoch 236\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.40\n",
      "epoch 237\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.30\n",
      "epoch 238\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:77.60\n",
      "epoch 239\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:77.30\n",
      "epoch 240\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:74.40\n",
      "epoch 241\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.60\n",
      "epoch 242\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.90\n",
      "epoch 243\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.18    acc:76.80\n",
      "epoch 244\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:75.30\n",
      "epoch 245\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.10\n",
      "epoch 246\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.02    acc:77.70\n",
      "epoch 247\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:74.80\n",
      "epoch 248\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.40\n",
      "epoch 249\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:77.40\n",
      "epoch 250\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:73.20\n",
      "epoch 251\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.20\n",
      "epoch 252\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.20\n",
      "epoch 253\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.20\n",
      "epoch 254\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:76.60\n",
      "epoch 255\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.70\n",
      "epoch 256\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.06    acc:78.20\n",
      "epoch 257\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.60\n",
      "epoch 258\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.30\n",
      "epoch 259\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:74.90\n",
      "epoch 260\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.10\n",
      "epoch 261\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.80\n",
      "epoch 262\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:76.60\n",
      "epoch 263\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.40\n",
      "epoch 264\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.80\n",
      "epoch 265\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.50\n",
      "epoch 266\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.16    acc:77.50\n",
      "epoch 267\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:77.40\n",
      "epoch 268\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.00\n",
      "epoch 269\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.05    acc:80.00\n",
      "epoch 270\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.40\n",
      "epoch 271\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.80\n",
      "epoch 272\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.80\n",
      "epoch 273\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.40\n",
      "epoch 274\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.90\n",
      "epoch 275\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.40\n",
      "epoch 276\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:76.40\n",
      "epoch 277\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.10\n",
      "epoch 278\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.90\n",
      "epoch 279\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:77.40\n",
      "Epoch   278: reducing learning rate of group 0 to 3.7037e-06.\n",
      "epoch 280\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.10\n",
      "epoch 281\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.20\n",
      "epoch 282\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:78.40\n",
      "epoch 283\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.00\n",
      "epoch 284\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.90\n",
      "epoch 285\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.80\n",
      "epoch 286\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.30\n",
      "epoch 287\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:75.50\n",
      "epoch 288\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.90\n",
      "epoch 289\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.50\n",
      "epoch 290\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.90\n",
      "epoch 291\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.60\n",
      "epoch 292\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.90\n",
      "epoch 293\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:79.20\n",
      "epoch 294\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:78.70\n",
      "epoch 295\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:76.90\n",
      "epoch 296\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.80\n",
      "epoch 297\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:77.20\n",
      "epoch 298\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.30\n",
      "epoch 299\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.17    acc:77.80\n",
      "epoch 300\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:73.40\n",
      "epoch 301\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.40\n",
      "epoch 302\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.60\n",
      "epoch 303\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.40\n",
      "epoch 304\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.30\n",
      "epoch 305\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.80\n",
      "epoch 306\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.10\n",
      "epoch 307\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:77.60\n",
      "epoch 308\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:76.50\n",
      "epoch 309\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.60\n",
      "epoch 310\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.90\n",
      "epoch 311\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.10\n",
      "epoch 312\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.00\n",
      "epoch 313\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.10\n",
      "epoch 314\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:79.30\n",
      "epoch 315\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.40\n",
      "epoch 316\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.60\n",
      "epoch 317\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.50\n",
      "epoch 318\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.90\n",
      "epoch 319\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:77.00\n",
      "epoch 320\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.80\n",
      "epoch 321\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.00\n",
      "epoch 322\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.30\n",
      "epoch 323\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.00\n",
      "epoch 324\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.30\n",
      "epoch 325\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.00\n",
      "epoch 326\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:78.00\n",
      "epoch 327\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.90\n",
      "epoch 328\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.40\n",
      "epoch 329\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.20\n",
      "epoch 330\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.30\n",
      "epoch 331\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.30\n",
      "epoch 332\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.41    acc:75.70\n",
      "epoch 333\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.40\n",
      "epoch 334\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.20\n",
      "epoch 335\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.70\n",
      "epoch 336\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.80\n",
      "epoch 337\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.00\n",
      "epoch 338\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.30\n",
      "epoch 339\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.25    acc:76.60\n",
      "epoch 340\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.80\n",
      "epoch 341\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.80\n",
      "epoch 342\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.90\n",
      "epoch 343\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.60\n",
      "epoch 344\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.90\n",
      "epoch 345\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.50\n",
      "epoch 346\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.06    acc:78.40\n",
      "epoch 347\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.40\n",
      "epoch 348\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:76.50\n",
      "epoch 349\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:78.20\n",
      "epoch 350\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:78.50\n",
      "epoch 351\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.50\n",
      "epoch 352\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.13    acc:78.00\n",
      "epoch 353\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:76.70\n",
      "epoch 354\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:78.00\n",
      "epoch 355\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:74.80\n",
      "epoch 356\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:73.90\n",
      "epoch 357\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.20\n",
      "epoch 358\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.90\n",
      "epoch 359\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.90\n",
      "epoch 360\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.00\n",
      "epoch 361\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.00\n",
      "epoch 362\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.60\n",
      "epoch 363\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:75.70\n",
      "epoch 364\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.06    acc:78.40\n",
      "epoch 365\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:78.50\n",
      "epoch 366\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.30\n",
      "epoch 367\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:77.90\n",
      "epoch 368\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:79.00\n",
      "epoch 369\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.20\n",
      "epoch 370\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.50\n",
      "epoch 371\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.30\n",
      "epoch 372\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:74.70\n",
      "epoch 373\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.90\n",
      "epoch 374\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.30\n",
      "epoch 375\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.60\n",
      "epoch 376\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.40\n",
      "epoch 377\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.70\n",
      "epoch 378\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.10\n",
      "epoch 379\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.70\n",
      "epoch 380\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.20\n",
      "epoch 381\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.60\n",
      "Epoch   380: reducing learning rate of group 0 to 1.2346e-06.\n",
      "epoch 382\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.50\n",
      "epoch 383\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.30\n",
      "epoch 384\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.50\n",
      "epoch 385\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.80\n",
      "epoch 386\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.70\n",
      "epoch 387\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.70\n",
      "epoch 388\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.40\n",
      "epoch 389\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.90\n",
      "epoch 390\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:79.70\n",
      "epoch 391\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.30\n",
      "epoch 392\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.50\n",
      "epoch 393\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.50\n",
      "epoch 394\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.70\n",
      "epoch 395\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:78.10\n",
      "epoch 396\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.10\n",
      "epoch 397\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.50\n",
      "epoch 398\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.60\n",
      "epoch 399\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.70\n",
      "epoch 400\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.90\n",
      "epoch 401\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:78.80\n",
      "epoch 402\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.60\n",
      "epoch 403\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.60\n",
      "epoch 404\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.90\n",
      "epoch 405\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.80\n",
      "epoch 406\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:78.70\n",
      "epoch 407\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:74.30\n",
      "epoch 408\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.90\n",
      "epoch 409\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.30\n",
      "epoch 410\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.80\n",
      "epoch 411\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:79.20\n",
      "epoch 412\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.10\n",
      "Epoch   411: reducing learning rate of group 0 to 4.1152e-07.\n",
      "epoch 413\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.30\n",
      "epoch 414\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.90\n",
      "epoch 415\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:77.50\n",
      "epoch 416\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.90\n",
      "epoch 417\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.70\n",
      "epoch 418\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.00    acc:79.70\n",
      "epoch 419\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:78.60\n",
      "epoch 420\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:78.20\n",
      "epoch 421\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.13    acc:77.70\n",
      "epoch 422\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.60\n",
      "epoch 423\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.20\n",
      "epoch 424\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.80\n",
      "epoch 425\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:78.30\n",
      "epoch 426\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.30\n",
      "epoch 427\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.80\n",
      "epoch 428\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.20\n",
      "epoch 429\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.15    acc:77.50\n",
      "epoch 430\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.50\n",
      "epoch 431\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:76.30\n",
      "epoch 432\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.00\n",
      "epoch 433\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.20\n",
      "epoch 434\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.30\n",
      "epoch 435\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.30\n",
      "epoch 436\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:79.60\n",
      "epoch 437\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.40\n",
      "Epoch   436: reducing learning rate of group 0 to 1.3717e-07.\n",
      "epoch 438\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.50\n",
      "epoch 439\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.60\n",
      "epoch 440\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.29    acc:75.30\n",
      "epoch 441\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.20\n",
      "epoch 442\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.04    acc:78.20\n",
      "epoch 443\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:76.80\n",
      "epoch 444\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.10\n",
      "epoch 445\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.60\n",
      "epoch 446\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.00\n",
      "epoch 447\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:78.10\n",
      "epoch 448\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.00\n",
      "epoch 449\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:76.50\n",
      "epoch 450\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:79.00\n",
      "epoch 451\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:77.80\n",
      "epoch 452\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:78.20\n",
      "epoch 453\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:76.40\n",
      "epoch 454\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.40\n",
      "epoch 455\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.40\n",
      "epoch 456\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.80\n",
      "epoch 457\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.20\n",
      "epoch 458\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.50\n",
      "epoch 459\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.55    acc:72.80\n",
      "epoch 460\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.90\n",
      "epoch 461\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:78.10\n",
      "epoch 462\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:76.30\n",
      "epoch 463\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.30\n",
      "epoch 464\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.40\n",
      "epoch 465\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:75.00\n",
      "epoch 466\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.41    acc:75.60\n",
      "epoch 467\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:78.00\n",
      "epoch 468\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.90\n",
      "epoch 469\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:77.10\n",
      "epoch 470\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.80\n",
      "epoch 471\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.70\n",
      "epoch 472\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:76.80\n",
      "epoch 473\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.80\n",
      "epoch 474\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.60\n",
      "epoch 475\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:77.20\n",
      "epoch 476\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:78.50\n",
      "epoch 477\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.60\n",
      "epoch 478\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.90\n",
      "epoch 479\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:76.90\n",
      "Epoch   478: reducing learning rate of group 0 to 4.5725e-08.\n",
      "epoch 480\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:79.20\n",
      "epoch 481\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:76.00\n",
      "epoch 482\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.10\n",
      "epoch 483\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:76.90\n",
      "epoch 484\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.60\n",
      "epoch 485\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.90\n",
      "epoch 486\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.90\n",
      "epoch 487\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:78.20\n",
      "epoch 488\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.70\n",
      "epoch 489\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.30\n",
      "epoch 490\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:75.80\n",
      "epoch 491\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.00\n",
      "epoch 492\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.10\n",
      "epoch 493\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.40\n",
      "epoch 494\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:76.10\n",
      "epoch 495\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.10\n",
      "epoch 496\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.90\n",
      "epoch 497\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.40\n",
      "epoch 498\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:75.20\n",
      "epoch 499\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.50\n",
      "epoch 500\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:76.20\n",
      "Epoch   499: reducing learning rate of group 0 to 1.5242e-08.\n",
      "epoch 501\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:78.90\n",
      "epoch 502\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.30\n",
      "epoch 503\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.00\n",
      "epoch 504\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 505\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.80\n",
      "epoch 506\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.90\n",
      "epoch 507\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.10\n",
      "epoch 508\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.00\n",
      "epoch 509\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.40\n",
      "epoch 510\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:79.50\n",
      "epoch 511\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.00\n",
      "epoch 512\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:79.00\n",
      "epoch 513\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.20\n",
      "epoch 514\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.70\n",
      "epoch 515\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.50\n",
      "epoch 516\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:76.00\n",
      "epoch 517\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.20\n",
      "epoch 518\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.50\n",
      "epoch 519\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.00\n",
      "epoch 520\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.20\n",
      "epoch 521\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.10\n",
      "Epoch   520: reducing learning rate of group 0 to 5.0805e-09.\n",
      "epoch 522\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.70\n",
      "epoch 523\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.70\n",
      "epoch 524\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.00\n",
      "epoch 525\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.80\n",
      "epoch 526\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.80\n",
      "epoch 527\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.10\n",
      "epoch 528\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.70\n",
      "epoch 529\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.00\n",
      "epoch 530\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:79.40\n",
      "epoch 531\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.80\n",
      "epoch 532\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.60\n",
      "epoch 533\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.09    acc:78.50\n",
      "epoch 534\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.80\n",
      "epoch 535\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.90\n",
      "epoch 536\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.50\n",
      "epoch 537\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:75.70\n",
      "epoch 538\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.50\n",
      "epoch 539\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.10\n",
      "epoch 540\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.80\n",
      "epoch 541\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.60\n",
      "epoch 542\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.00\n",
      "epoch 543\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.70\n",
      "epoch 544\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.40\n",
      "epoch 545\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.90\n",
      "epoch 546\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.30\n",
      "epoch 547\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.60\n",
      "epoch 548\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:76.70\n",
      "epoch 549\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.20\n",
      "epoch 550\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:75.00\n",
      "epoch 551\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.90\n",
      "epoch 552\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.06    acc:78.80\n",
      "epoch 553\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.70\n",
      "epoch 554\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.50\n",
      "epoch 555\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.20\n",
      "epoch 556\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.20    acc:77.00\n",
      "epoch 557\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:78.50\n",
      "epoch 558\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:78.00\n",
      "epoch 559\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.90\n",
      "epoch 560\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.90\n",
      "epoch 561\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:76.20\n",
      "epoch 562\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.50\n",
      "epoch 563\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:76.10\n",
      "epoch 564\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:75.70\n",
      "epoch 565\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.80\n",
      "epoch 566\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.90\n",
      "epoch 567\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.30\n",
      "epoch 568\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.30    acc:76.00\n",
      "epoch 569\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.90\n",
      "epoch 570\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:79.50\n",
      "epoch 571\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.37    acc:75.20\n",
      "epoch 572\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.50\n",
      "epoch 573\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.40\n",
      "epoch 574\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.50\n",
      "epoch 575\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.80\n",
      "epoch 576\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.10\n",
      "epoch 577\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.10\n",
      "epoch 578\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:79.00\n",
      "epoch 579\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:79.30\n",
      "epoch 580\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.00\n",
      "epoch 581\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:75.40\n",
      "epoch 582\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:79.80\n",
      "epoch 583\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:78.40\n",
      "epoch 584\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.40\n",
      "epoch 585\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.20\n",
      "epoch 586\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.60\n",
      "epoch 587\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.60\n",
      "epoch 588\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.50\n",
      "epoch 589\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.00\n",
      "epoch 590\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.90\n",
      "epoch 591\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.00\n",
      "epoch 592\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:74.60\n",
      "epoch 593\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.40\n",
      "epoch 594\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:74.10\n",
      "epoch 595\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:77.30\n",
      "epoch 596\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.70\n",
      "epoch 597\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.50\n",
      "epoch 598\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.20\n",
      "epoch 599\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.12    acc:78.80\n",
      "epoch 600\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.90\n",
      "epoch 601\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.90\n",
      "epoch 602\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.40\n",
      "epoch 603\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:74.40\n",
      "epoch 604\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.10\n",
      "epoch 605\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:77.20\n",
      "epoch 606\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:74.60\n",
      "epoch 607\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.80\n",
      "epoch 608\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:74.50\n",
      "epoch 609\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.60\n",
      "epoch 610\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.50\n",
      "epoch 611\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.00\n",
      "epoch 612\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.70\n",
      "epoch 613\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.00\n",
      "epoch 614\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:76.10\n",
      "epoch 615\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.30\n",
      "epoch 616\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.50\n",
      "epoch 617\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.90\n",
      "epoch 618\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.60\n",
      "epoch 619\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.30\n",
      "epoch 620\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:79.10\n",
      "epoch 621\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:78.30\n",
      "epoch 622\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:75.60\n",
      "epoch 623\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.18    acc:77.80\n",
      "epoch 624\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:74.90\n",
      "epoch 625\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:75.60\n",
      "epoch 626\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.80\n",
      "epoch 627\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.90\n",
      "epoch 628\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.30\n",
      "epoch 629\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:78.50\n",
      "epoch 630\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.10\n",
      "epoch 631\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:77.50\n",
      "epoch 632\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.80\n",
      "epoch 633\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.20\n",
      "epoch 634\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.30\n",
      "epoch 635\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:75.40\n",
      "epoch 636\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 637\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.70\n",
      "epoch 638\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.01    acc:78.70\n",
      "epoch 639\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.60\n",
      "epoch 640\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.00\n",
      "epoch 641\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.60\n",
      "epoch 642\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:78.40\n",
      "epoch 643\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:74.90\n",
      "epoch 644\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:78.50\n",
      "epoch 645\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.50\n",
      "epoch 646\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.90\n",
      "epoch 647\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.60\n",
      "epoch 648\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:78.70\n",
      "epoch 649\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.70\n",
      "epoch 650\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.30\n",
      "epoch 651\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.50\n",
      "epoch 652\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.80\n",
      "epoch 653\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.90\n",
      "epoch 654\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.20\n",
      "epoch 655\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.50\n",
      "epoch 656\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:79.10\n",
      "epoch 657\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.20\n",
      "epoch 658\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.50\n",
      "epoch 659\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.10\n",
      "epoch 660\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:75.20\n",
      "epoch 661\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.40\n",
      "epoch 662\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:78.20\n",
      "epoch 663\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:75.90\n",
      "epoch 664\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.20\n",
      "epoch 665\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.30\n",
      "epoch 666\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.80\n",
      "epoch 667\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:79.60\n",
      "epoch 668\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.07    acc:79.20\n",
      "epoch 669\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.90\n",
      "epoch 670\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.80\n",
      "epoch 671\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.20\n",
      "epoch 672\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.10\n",
      "epoch 673\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.40\n",
      "epoch 674\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.50\n",
      "epoch 675\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.80\n",
      "epoch 676\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.80\n",
      "epoch 677\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.70\n",
      "epoch 678\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.60\n",
      "epoch 679\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.90\n",
      "epoch 680\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.90\n",
      "epoch 681\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.40\n",
      "epoch 682\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:74.70\n",
      "epoch 683\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.80\n",
      "epoch 684\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.44    acc:75.70\n",
      "epoch 685\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.90\n",
      "epoch 686\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.60\n",
      "epoch 687\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.60\n",
      "epoch 688\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.06    acc:80.80\n",
      "epoch 689\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.10\n",
      "epoch 690\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.80\n",
      "epoch 691\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.40\n",
      "epoch 692\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.70\n",
      "epoch 693\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.04    acc:81.30\n",
      "epoch 694\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.18    acc:77.70\n",
      "epoch 695\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.00\n",
      "epoch 696\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.70\n",
      "epoch 697\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.00\n",
      "epoch 698\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.30\n",
      "epoch 699\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:78.20\n",
      "epoch 700\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.20\n",
      "epoch 701\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.10\n",
      "epoch 702\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.90\n",
      "epoch 703\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:75.30\n",
      "epoch 704\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:78.30\n",
      "epoch 705\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.50\n",
      "epoch 706\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.20\n",
      "epoch 707\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.50\n",
      "epoch 708\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.90\n",
      "epoch 709\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.60\n",
      "epoch 710\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.20\n",
      "epoch 711\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.60\n",
      "epoch 712\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:77.20\n",
      "epoch 713\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.20\n",
      "epoch 714\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.80\n",
      "epoch 715\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.70\n",
      "epoch 716\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.00\n",
      "epoch 717\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.90\n",
      "epoch 718\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:78.60\n",
      "epoch 719\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.60\n",
      "epoch 720\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:78.30\n",
      "epoch 721\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.20\n",
      "epoch 722\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.10\n",
      "epoch 723\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.70\n",
      "epoch 724\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.02    acc:78.50\n",
      "epoch 725\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.20\n",
      "epoch 726\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.20\n",
      "epoch 727\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.60\n",
      "epoch 728\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.10\n",
      "epoch 729\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.70\n",
      "epoch 730\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.50\n",
      "epoch 731\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.80\n",
      "epoch 732\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.40\n",
      "epoch 733\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.70\n",
      "epoch 734\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.10\n",
      "epoch 735\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.90\n",
      "epoch 736\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.00\n",
      "epoch 737\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:75.00\n",
      "epoch 738\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.30\n",
      "epoch 739\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:78.70\n",
      "epoch 740\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.60\n",
      "epoch 741\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.50\n",
      "epoch 742\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:78.50\n",
      "epoch 743\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.40\n",
      "epoch 744\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.46    acc:76.20\n",
      "epoch 745\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.90\n",
      "epoch 746\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:74.40\n",
      "epoch 747\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.00\n",
      "epoch 748\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:75.70\n",
      "epoch 749\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.80\n",
      "epoch 750\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.80\n",
      "epoch 751\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.17    acc:77.20\n",
      "epoch 752\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:77.00\n",
      "epoch 753\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.50\n",
      "epoch 754\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:77.00\n",
      "epoch 755\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.30\n",
      "epoch 756\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.00\n",
      "epoch 757\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.30\n",
      "epoch 758\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:74.00\n",
      "epoch 759\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:78.80\n",
      "epoch 760\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.90\n",
      "epoch 761\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:76.90\n",
      "epoch 762\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.40\n",
      "epoch 763\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:77.40\n",
      "epoch 764\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:75.80\n",
      "epoch 765\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:75.70\n",
      "epoch 766\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.48    acc:73.30\n",
      "epoch 767\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.80\n",
      "epoch 768\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.10\n",
      "epoch 769\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.80\n",
      "epoch 770\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.60\n",
      "epoch 771\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:78.00\n",
      "epoch 772\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.80\n",
      "epoch 773\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.00\n",
      "epoch 774\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.20\n",
      "epoch 775\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.10\n",
      "epoch 776\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.30\n",
      "epoch 777\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:77.50\n",
      "epoch 778\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.70\n",
      "epoch 779\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.40\n",
      "epoch 780\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.80\n",
      "epoch 781\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.30\n",
      "epoch 782\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:76.60\n",
      "epoch 783\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:76.70\n",
      "epoch 784\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:74.50\n",
      "epoch 785\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:75.70\n",
      "epoch 786\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:78.60\n",
      "epoch 787\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.00\n",
      "epoch 788\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.30\n",
      "epoch 789\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:77.30\n",
      "epoch 790\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:73.80\n",
      "epoch 791\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:76.80\n",
      "epoch 792\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.10\n",
      "epoch 793\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.12    acc:78.40\n",
      "epoch 794\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:77.80\n",
      "epoch 795\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.10\n",
      "epoch 796\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.32    acc:76.90\n",
      "epoch 797\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.20\n",
      "epoch 798\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.10\n",
      "epoch 799\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.20\n",
      "epoch 800\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:79.30\n",
      "epoch 801\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.10\n",
      "epoch 802\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.20\n",
      "epoch 803\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.50\n",
      "epoch 804\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 805\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.20\n",
      "epoch 806\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:77.10\n",
      "epoch 807\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.30\n",
      "epoch 808\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.00\n",
      "epoch 809\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:74.90\n",
      "epoch 810\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.10\n",
      "epoch 811\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:77.50\n",
      "epoch 812\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.70\n",
      "epoch 813\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:77.40\n",
      "epoch 814\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:75.60\n",
      "epoch 815\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.20\n",
      "epoch 816\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:75.10\n",
      "epoch 817\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.00\n",
      "epoch 818\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.50\n",
      "epoch 819\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.50\n",
      "epoch 820\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:78.00\n",
      "epoch 821\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:75.40\n",
      "epoch 822\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.70\n",
      "epoch 823\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:77.90\n",
      "epoch 824\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:77.60\n",
      "epoch 825\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.80\n",
      "epoch 826\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:78.50\n",
      "epoch 827\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.60\n",
      "epoch 828\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.40\n",
      "epoch 829\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.30\n",
      "epoch 830\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.70\n",
      "epoch 831\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.40\n",
      "epoch 832\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.07    acc:79.00\n",
      "epoch 833\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:77.10\n",
      "epoch 834\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.10\n",
      "epoch 835\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:77.80\n",
      "epoch 836\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 837\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.80\n",
      "epoch 838\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.40\n",
      "epoch 839\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.20\n",
      "epoch 840\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:76.40\n",
      "epoch 841\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.90\n",
      "epoch 842\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.22    acc:76.10\n",
      "epoch 843\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.45    acc:75.90\n",
      "epoch 844\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:79.20\n",
      "epoch 845\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.70\n",
      "epoch 846\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.38    acc:75.20\n",
      "epoch 847\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:77.90\n",
      "epoch 848\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.10\n",
      "epoch 849\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.27    acc:77.00\n",
      "epoch 850\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:75.30\n",
      "epoch 851\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.10\n",
      "epoch 852\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.80\n",
      "epoch 853\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 0.99    acc:81.80\n",
      "epoch 854\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.20\n",
      "epoch 855\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:78.80\n",
      "epoch 856\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.09    acc:77.60\n",
      "epoch 857\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:77.10\n",
      "epoch 858\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:78.40\n",
      "epoch 859\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:79.10\n",
      "epoch 860\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.40\n",
      "epoch 861\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:75.10\n",
      "epoch 862\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.40\n",
      "epoch 863\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:79.40\n",
      "epoch 864\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:74.30\n",
      "epoch 865\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.33    acc:76.20\n",
      "epoch 866\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.10    acc:77.30\n",
      "epoch 867\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.90\n",
      "epoch 868\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:77.20\n",
      "epoch 869\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.60\n",
      "epoch 870\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:76.70\n",
      "epoch 871\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:75.40\n",
      "epoch 872\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:78.30\n",
      "epoch 873\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.30\n",
      "epoch 874\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.00\n",
      "epoch 875\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.60\n",
      "epoch 876\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:79.00\n",
      "epoch 877\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:78.30\n",
      "epoch 878\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:75.00\n",
      "epoch 879\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.90\n",
      "epoch 880\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.12    acc:79.40\n",
      "epoch 881\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.00\n",
      "epoch 882\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 883\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.26    acc:76.50\n",
      "epoch 884\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.07    acc:78.70\n",
      "epoch 885\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.60\n",
      "epoch 886\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:78.20\n",
      "epoch 887\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.70\n",
      "epoch 888\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:76.70\n",
      "epoch 889\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.10\n",
      "epoch 890\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:76.90\n",
      "epoch 891\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.20\n",
      "epoch 892\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:75.70\n",
      "epoch 893\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:75.60\n",
      "epoch 894\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.00\n",
      "epoch 895\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.80\n",
      "epoch 896\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.40\n",
      "epoch 897\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.90\n",
      "epoch 898\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.60\n",
      "epoch 899\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.50\n",
      "epoch 900\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:78.10\n",
      "epoch 901\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.70\n",
      "epoch 902\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:78.40\n",
      "epoch 903\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:77.00\n",
      "epoch 904\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:74.80\n",
      "epoch 905\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.43    acc:73.70\n",
      "epoch 906\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:75.20\n",
      "epoch 907\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.80\n",
      "epoch 908\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.60\n",
      "epoch 909\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.40\n",
      "epoch 910\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.50\n",
      "epoch 911\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.60\n",
      "epoch 912\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:75.90\n",
      "epoch 913\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:77.50\n",
      "epoch 914\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.36    acc:75.30\n",
      "epoch 915\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.60\n",
      "epoch 916\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:80.20\n",
      "epoch 917\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.34    acc:76.70\n",
      "epoch 918\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:76.80\n",
      "epoch 919\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.20\n",
      "epoch 920\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.13    acc:79.50\n",
      "epoch 921\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.30\n",
      "epoch 922\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:77.20\n",
      "epoch 923\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.00\n",
      "epoch 924\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:79.40\n",
      "epoch 925\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.90\n",
      "epoch 926\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.34    acc:76.80\n",
      "epoch 927\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.20\n",
      "epoch 928\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:76.70\n",
      "epoch 929\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:75.80\n",
      "epoch 930\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:77.10\n",
      "epoch 931\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.40\n",
      "epoch 932\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:76.80\n",
      "epoch 933\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.20\n",
      "epoch 934\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.90\n",
      "epoch 935\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.22    acc:77.90\n",
      "epoch 936\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.80\n",
      "epoch 937\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.16    acc:77.70\n",
      "epoch 938\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.07    acc:79.10\n",
      "epoch 939\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:76.10\n",
      "epoch 940\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:78.80\n",
      "epoch 941\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:78.20\n",
      "epoch 942\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.20\n",
      "epoch 943\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.90\n",
      "epoch 944\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:79.10\n",
      "epoch 945\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:76.90\n",
      "epoch 946\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:77.50\n",
      "epoch 947\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.70\n",
      "epoch 948\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.90\n",
      "epoch 949\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.17    acc:78.30\n",
      "epoch 950\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.40\n",
      "epoch 951\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:77.20\n",
      "epoch 952\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:78.90\n",
      "epoch 953\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.39    acc:74.90\n",
      "epoch 954\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.80\n",
      "epoch 955\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.24    acc:77.90\n",
      "epoch 956\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.20\n",
      "epoch 957\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:79.60\n",
      "epoch 958\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.31    acc:73.60\n",
      "epoch 959\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.24    acc:78.90\n",
      "epoch 960\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.40\n",
      "epoch 961\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:80.40\n",
      "epoch 962\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:77.30\n",
      "epoch 963\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.28    acc:77.00\n",
      "epoch 964\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:76.50\n",
      "epoch 965\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:77.10\n",
      "epoch 966\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.21    acc:77.00\n",
      "epoch 967\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.40\n",
      "epoch 968\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.50\n",
      "epoch 969\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:77.40\n",
      "epoch 970\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.15    acc:77.60\n",
      "epoch 971\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.80\n",
      "epoch 972\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.42    acc:75.60\n",
      "epoch 973\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.30\n",
      "epoch 974\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.31    acc:75.40\n",
      "epoch 975\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.40    acc:75.80\n",
      "epoch 976\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.14    acc:77.60\n",
      "epoch 977\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.11    acc:79.20\n",
      "epoch 978\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.29    acc:75.70\n",
      "epoch 979\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.37    acc:76.10\n",
      "epoch 980\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:76.60\n",
      "epoch 981\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:76.80\n",
      "epoch 982\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:76.00\n",
      "epoch 983\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:75.20\n",
      "epoch 984\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.08    acc:78.80\n",
      "epoch 985\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.23    acc:77.00\n",
      "epoch 986\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:78.40\n",
      "epoch 987\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.30    acc:75.30\n",
      "epoch 988\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:76.20\n",
      "epoch 989\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.50\n",
      "epoch 990\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:77.30\n",
      "epoch 991\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.20    acc:77.20\n",
      "epoch 992\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.28    acc:76.50\n",
      "epoch 993\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.32    acc:76.70\n",
      "epoch 994\n",
      "loss: 0.00    acc:99.97\n",
      "loss: 1.23    acc:78.50\n",
      "epoch 995\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.18    acc:78.90\n",
      "epoch 996\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.12    acc:78.70\n",
      "epoch 997\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.25    acc:76.90\n",
      "epoch 998\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.19    acc:77.60\n",
      "epoch 999\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.27    acc:75.60\n",
      "epoch 1000\n",
      "loss: 0.00    acc:100.00\n",
      "loss: 1.35    acc:76.50\n"
     ]
    }
   ],
   "source": [
    "num_epochs = 1000\n",
    "min_num_train_batches = 10\n",
    "max_num_test_batches = 10\n",
    "init_lr = 3e-4\n",
    "\n",
    "test_accs = {}\n",
    "train_accs = {}\n",
    "param_counts = {}\n",
    "num_filters_alts = 2 ** np.arange(2, 9, 2)\n",
    "\n",
    "for num_filters in num_filters_alts:\n",
    "    print(num_filters)\n",
    "    test_accs[num_filters] = []\n",
    "    train_accs[num_filters] = []\n",
    "    model = DeepConv(input_shape, output_size, init_num_filters=num_filters).to(device)\n",
    "    optimizer = optim.Adam(model.parameters(), lr=init_lr)\n",
    "    scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, factor=1/3, patience=20, verbose=True)\n",
    "    loss_function = nn.CrossEntropyLoss()\n",
    "\n",
    "    param_count = 0\n",
    "    for param in model.parameters():\n",
    "        param_count += param.view(-1).shape[0]\n",
    "    param_counts[num_filters] = param_count\n",
    "    print(param_count)\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        print('epoch', epoch+1)\n",
    "\n",
    "        cum_batch_size = 0\n",
    "        cum_loss = 0\n",
    "        cum_acc = 0\n",
    "\n",
    "        model.train()\n",
    "        i = 0\n",
    "        while i < min_num_train_batches:\n",
    "            for X, y in train_loader_train:\n",
    "                batch_size = X.shape[0]\n",
    "                cum_batch_size += batch_size\n",
    "\n",
    "                X, y = X.to(device), y.to(device)\n",
    "\n",
    "                logits = model(X)\n",
    "\n",
    "                loss = loss_function(logits, y)\n",
    "                cum_loss += loss.item() * batch_size\n",
    "\n",
    "                acc = torch.mean((torch.max(logits, dim=-1)[1] == y).float())\n",
    "                cum_acc += acc.item() * batch_size\n",
    "\n",
    "                optimizer.zero_grad()\n",
    "                loss.backward()\n",
    "                optimizer.step()\n",
    "                \n",
    "                i += 1\n",
    "\n",
    "        mean_train_loss = cum_loss / cum_batch_size\n",
    "        mean_train_acc = cum_acc / cum_batch_size\n",
    "\n",
    "        train_accs[num_filters].append(mean_train_acc)\n",
    "\n",
    "        print('loss: {:.2f}    acc:{:.2f}'.format(mean_train_loss, mean_train_acc*100))\n",
    "\n",
    "        cum_batch_size = 0\n",
    "        cum_loss = 0\n",
    "        cum_acc = 0\n",
    "\n",
    "        model.eval()\n",
    "        with torch.no_grad():\n",
    "            for i, (X, y) in enumerate(test_loader):\n",
    "                if i >= max_num_test_batches:\n",
    "                    break\n",
    "                batch_size = X.shape[0]\n",
    "                cum_batch_size += batch_size\n",
    "\n",
    "                X, y = X.to(device), y.to(device)\n",
    "\n",
    "                logits = model(X)\n",
    "\n",
    "                loss = loss_function(logits, y)\n",
    "                cum_loss += loss.item() * batch_size\n",
    "\n",
    "                acc = torch.mean((torch.max(logits, dim=-1)[1] == y).float())\n",
    "                cum_acc += acc.item() * batch_size\n",
    "\n",
    "        mean_test_loss = cum_loss / cum_batch_size\n",
    "        mean_test_acc = cum_acc / cum_batch_size\n",
    "\n",
    "        test_accs[num_filters].append(mean_test_acc)\n",
    "\n",
    "        print('loss: {:.2f}    acc:{:.2f}'.format(mean_test_loss, mean_test_acc*100))\n",
    "        \n",
    "        scheduler.step(mean_train_loss)\n",
    "        #if optimizer.param_groups[0]['lr'] < init_lr / 100:\n",
    "        #    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'CIFAR10, m=4000; test accuracy')"
      ]
     },
     "execution_count": 151,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAswAAAHwCAYAAACygTlRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdb4c31d7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "\n",
    "for num_filters in num_filters_alts:\n",
    "    curve = DataFrame(np.array(test_accs[num_filters])).ewm(alpha=0.1).mean()\n",
    "    plt.plot(curve)\n",
    "plt.ylim(0.4, 0.8)\n",
    "plt.legend(['{} params'.format(param_counts[num_filters]) for num_filters in num_filters_alts])\n",
    "plt.xlabel('epoch')\n",
    "plt.title('CIFAR10, m=4000; test accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'CIFAR10, m=4000; train accuracy')"
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsYAAAHwCAYAAAClo6mYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdb4c2a6048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "\n",
    "for num_filters in num_filters_alts:\n",
    "    curve = DataFrame(np.array(train_accs[num_filters])).ewm(alpha=0.1).mean()\n",
    "    plt.plot(curve)\n",
    "plt.ylim(0.4, 1.0)\n",
    "plt.legend(['{} params'.format(param_counts[num_filters]) for num_filters in num_filters_alts])\n",
    "plt.xlabel('epoch')\n",
    "plt.title('CIFAR10, m=4000; train accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'CIFAR10, m=4000; risk difference')"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsYAAAHwCAYAAAClo6mYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzs3Xd0VcXi9vHvhDQSegktYESBAAk1\nQCDSq16FywW8VBHEilexgth4uf4EbICKoiiKqKAI0lE6SDPSSyghEEjoNZX0ef844dwQAgRMAsrz\nWSvLc/bMnj17n6hP5syebay1iIiIiIjc7lxudgdERERERG4FCsYiIiIiIigYi4iIiIgACsYiIiIi\nIoCCsYiIiIgIoGAsIiIiIgIoGIuI3PaMMf2NMYtyUe9bY8yIGzzGW8aYrzNfVzXGxGcpq2CMWWOM\niTPGjDEO3xhjzhtj1t3I8UREboSCsYhclTGmtzFmozEm3hhzzBizyBhzT2bZCGPMt1nqWmNMQmbd\neGPM+Wxt3Z1Z58Ns212z7RttjHnXGOOSpU4vY8x6Y0yiMWZpDv1sYIzZnFn+hzGmTt5fjRtjjPlv\n5vm1yrLN0xjztTEmNvO6Ppttnw7GmL2Z57PcGFMlt/teL2vtFGvtvX+mjes83gFrbZEsm54AjgLF\nrLVDgVZAS6CitbZZQfVLRETBWESuyBjzPDAOeBsoB1QBPgG6XGW3utbaIpk/JbKV9QfOAr2MMW45\n7Fs7MzC1Afpl1r/oDPAB8G4O/fQA5gBfASWBacDsKxyjQBljquO4XiezFf0X8MNxTdsDw40x7TL3\nKQf8BLwClAa2At/nZt8b6J/rjeyXx+4Awuz/njh1B3DQWpt4vQ3dIucjIn9RCsYikiNjTHFgJDDY\nWjvLWptgrU211s6z1r50A+0ZHGH3FcAA/7hSXWvtPmAdUC/LtsXW2hnAsRx2aeuoYj+y1iYDYwEP\nHKOO1+rXxdHqJ40xEZlf579pjKlmjNmQOSo77U+E7AnAi0Bqtu0PASOtteettTuBycDDmWXdgK2Z\n1/0CMAJoZIy5+1r7GmMKZU5BCL7C+Q4yxqw2xnxojDkLvJa5bWVmuUtm2UljTIwxZrsxplYO7RTL\nbGds5mebvbyqMea3zOv5K46Af7HsbmOMzXw9FeiDI9zHG2MeAyYCzTPfv55Zr7MxZlvmua0xxgRk\naS/aGPOSMWYHkJi5zdcY87Mx5pQx5qAxZnCW+m9lfqbfZvZvpzGmQZbyO4wxszP3PW2MGZ/t+u0x\nxpwzjm9PKud0nUXkr0nBWESupCngCfycR+21wjHq/AMwA0e4y5ExpiYQAuzPZdu1gW0X32SOPO7I\n3J5b7XEE8RDgVRwj4z1xjF7WBx7M7NudmeHsSj8PZjmPXkCstXZxtvMrC/hk7XPm64v9zX4+scBB\noPa19rXWpltrS1hrN1zlXJsBu4GywJhsZfcCwUA1HKPvPXGM8mftfxlgObDcWvtclpHerKYDG4Ay\nwGgcfxRdxlrbD8fvxNuZ3zJ8DjwN/Jb5/r/GmEbAJGAQjoA9GZhjjHHP0lTPzL4XN8YUAuYDfwCV\ncHy2Lxlj2map/09gKlACWAR8mHlursACHL97fkBl4MfMsu7ASzi+ASgL/M6lI/ki8henYCwiV1Ia\nOG2tTbvO/TZnCYlZ5xL3BxZYa2NwhIl/GGNKZ9t3uzEmAQgDlgCf5fKYRYCYbNtigKLX0e8x1to4\na+12HKHxF2ttpLX2HPArjnCMtfZgZvC80s/FEFUMx5SH567Q34t9zKm/Vzufa+2bG4ettZ9mhugL\n2cpSgWKAP4C1NsxaezxLeSVgFfCdtXZETo0bY6ri+CPjTWttsrV2BbDwOvqX3WPAJ9baPzL7PDlz\ne6MsdcZba6MzzycYx3zlt621Kdba/cCXOMLzRaustb9aa9NxBOSL3040xRHmh2Z+S3LBWrs2s+xx\nHAF+b+a/F28BjY0xlf7EuYnILUTBWESu5AxQ5gbmbDbIEhKfATDGeOOYHvBdZp01OKZE9Mq2bx0c\nAa83joDilctjxuMIc1kVA+Kuo98nsry+kMP7Ilyf/wKTrbWHcyi7uCJD1j5n7e/Vzuda++ZG1JUK\nMke3JwKfAieMMRONMVlDd2fADccI7pVUBM5kmyN86Dr6l90dwNCsI/NABRwh/aKobPWrZKv/MlA+\nS52sYT8R8M58XRmIzAzMOfVjQpY2TwMZgO+fODcRuYUoGIvIlawHknB85fxndcMRLD83xhzHEYrL\nk8N0CmtthrV2GrAReC2X7e8C6l58kznnNTBze57KnDsbf5Wff2dWbQs8Z4w5nnnOFYBZxpgXrbWn\ngFNZ+5z5+mJ/s59PUeBOYFcu9s2NnKY+/K/Q2nHW2gZAAFALeD5L8URgBbDAGHOlP1yOAaWNMYWz\nbKtyhbq5EQX8v2wj814XR+cvdjtb/fBs9Ytaax/I5bHuyJyOkVPZI9naLWyt/f3GT01EbiUKxiKS\no8wpD2/gGCH7pzHGyxjjZoy51xjzznU21x/HCGMgjq+s6wEtgKDM+cQ5GQU8kTmn9uJNZZ6AK+Bi\nHEuWXRzNXg4UMsYMNo4VKp7FMSVgVea+g4wxuZ2vfFUXlxq7ys8PmVVbZjvfEzjmyE7MLP8GeN0Y\nUyLz5raBwNeZZTOBepnX3RN4E9iYOSXgqvua/91MeM+NnJ8xpnHmjyuQAKQAWUdPLY7l1Q4AczP7\nl/0aRQDbgRHGGHdjTAuucrNlLnwODDbGNDIORYwxD2R+E5GT9UCKMeaFzN+TQsaYQGNMw1wcaz2O\nb0vezvydL2yMCcksmwi8evF3NvP6d/8T5yUitxgFYxG5ImvtBzhGC1/DMUoZhePGqNm5bcM41t9t\nBYyz1h7P8hMKLOXSJdmyHnsrjpDyYuamATimNHwEtM58PTGzbhKOG6IGAeeBvkAXa+3FlSAqA2sp\nQNbaM1nPF8dX7mettRenQryO43pG4Qj2o6y1SzP3PYHjZr93gHNAAxzTS7jWvjjONRbYeYNdL4Fj\nPu55IBLH6O/YbOdmgUdwLEH3c+YfI9n1xHEj41kcNzNOvcH+kDki+ySO6R3ngH04PuMr1U8D7gMa\nZ57DaRzz1bNPT7nSvvcDNXFc38NA98yyGTiWDJxhjInFEf473uBpicgtyOR8M7GIyN+HMWYZ8KR1\nLAP3t2aMeRi4y1r7+s3ui4jIX42CsYiIiIgIuZxKYYzpZByPJt1vjBmWQ/nDmQuhb838GZT3XRUR\nERERyT/XXIYp887cCTgWSI8G/jDGzLXWhmWr+oO19ul86KOIiIiISL7LzYhxY2B/5p3YKTieZtQl\nf7slIiIiIlKwchOMK3HpwunRXLqo+kXdjDHbjTE/6dnxIiIiIvJXk5snWpkctmW/Y28eMM1am2yM\neQKYArS5rCFjHsPxaE+8vb0b+vv7X2d3RURERESuz6ZNm05ba8teq15ugnE0jnUxL/IFjmatYK09\nk+XtJGBMTg1Zaz/HsVA7QUFBduPGjbk4vIiIiIjIjTPG5Oqx9LmZSvEHUM0Yc6cxxh3Hou1zsx2s\nQpa3nYHdue2oiIiIiMit4JojxtbaNGPM08CvQCFgsrV2lzFmJI5HlM4FnjHGdAbScDzl6OF87LOI\niIiISJ67aQ/40FQKERERESkIxphN1tqga9XLzRxjERERkb+F1NRUoqOjSUpKutldkXzg6emJr68v\nbm5uN7S/grGIiIjcNqKjoylatCh+fn4Yk9PCW/JXZa3lzJkzREdHc+edd95QG7l6JLSIiIjI30FS\nUhKlS5dWKP4bMsZQunTpP/VtgIKxiIiI3FYUiv++/uxnq2AsIiIiUkCioqJo3bo1NWvWpHbt2owf\nP95ZtnXrVoKDg6lXrx5BQUGEhoYCsGfPHpo2bYqHhwfvvffeJe398ssv1KhRg7vvvpvRo0cX6Ln8\nHSkYi4iIiBQQV1dX3n//fXbv3s2GDRuYMGECYWFhALz88su8+eabbN26lZEjR/Lyyy8DUKpUKT78\n8ENefPHFS9pKT09n8ODBLFq0iLCwMKZNm+ZsK6+kpaXlaXu3OgVjERERkQJSoUIFGjRoAEDRokWp\nWbMmR44cARzTAGJjYwGIiYmhYsWKAPj4+NCoUaPLVloIDQ3l7rvvpmrVqri7u9OzZ0/mzJlz2TFb\ntWrFkCFDaNasGQEBAc6R6NDQUJo1a0b9+vVp1qwZe/fuBeDrr7+mR48ePPDAA3To0IH4+Hjatm1L\ngwYNCAwMdB4jMjISf39/Bg0aREBAAH369GHp0qWEhIRQrVo153FWrVpFvXr1qFevHvXr1ycuLi6v\nL2ue0aoUIiIiclv6f/N2EXY0Nk/brFWxGG8+UDtXdSMjI9myZQtNmjQBYNy4cXTs2JEXX3yRjIwM\n1q1bd9X9jxw5QuXKlZ3vfX19+f3333Osm5CQwLp161i9ejUDBw5k586d+Pv7s3r1alxdXVm6dCnD\nhw9n5syZAKxfv57t27dTqlQp0tLS+PnnnylWrBinT58mODiYzp07A7B//35mzJjB559/TqNGjfj+\n++9Zs2YNc+fO5e2332b27Nm89957TJgwgZCQEOLj4/H09MzV9bkZFIxFREREClh8fDzdunVj3Lhx\nFCtWDIBPP/2UsWPH0q1bN3788UceeeQRli5desU2cnpI25VuPuvVqxcALVq0IDY2lvPnzxMXF0f/\n/v0JDw/HGENqaqqzfvv27SlVqpTzOMOHD2f16tW4uLhw5MgRTpw4AcCdd95JYGAgALVr16Zt27YY\nYwgMDCQyMhKAkJAQnn/+efr06cO//vUvfH19r/NqFRwFYxEREbkt5XZkN6+lpqbSrVs3Z1C8aMqU\nKc6b8Xr06MGgQYOu2o6vry9RUVHO99HR0c7pF9llD8zGGF5//XVat27Nzz//TGRkJK1atXKWe3t7\nO19/9913nDp1ik2bNuHm5oafn59zSTQPDw9nPRcXF+d7FxcX5/zkYcOG8Y9//IOFCxcSHBzM0qVL\n8ff3v+q53SyaYywiIiJSQKy1PPLII9SsWZPnn3/+krKKFSuyatUqAJYvX061atWu2lajRo0IDw/n\n4MGDpKSkMH36dOcUh+x++OEHANasWUPx4sUpXrw4MTExVKpUCXDMK76SmJgYfHx8cHNzY8WKFRw6\ndCi3pwtAREQEgYGBDB06lKCgIPbs2XNd+xckjRiLiIiIFJC1a9cydepUAgMDqVevHgBvv/029913\nH5MmTeLZZ58lLS0NT09PPv/8cwCOHz9OUFAQsbGxuLi4MG7cOMLCwihWrBgff/wxHTt2JD09nYED\nB1K7ds6j4CVLlqRZs2bExsYyefJkwLEKRv/+/fnggw9o06bNFfvcp08fHnjgAYKCgqhXr951j/aO\nGzeOFStWUKhQIWrVqsW99957XfsXJJPT/JSCEBQUZDdu3HhTji0iIiK3p927d1OzZs2b3Y0C1apV\nK9577z2CgoJudlcKRE6fsTFmk7X2mhdAUylERERERNBUChEREZG/tZUrV97sLvxlaMRYRERERAQF\nYxERERERQMFYRERERARQMBYRERERARSMRURERApceno69evX5/7773duO3jwIE2aNKFatWr8+9//\nJiUlBYCJEyc61z2+5557CAsLAyAlJYUBAwYQGBhI3bp1c7zJrnPnzgQEBBTIOf0dKBiLiIiIFLDx\n48dfttbu0KFDee655wgPD6dkyZJ8+eWXAPTu3ZsdO3awdetWXn75ZecT8yZNmgTAjh07WLJkCS+8\n8AIZGRnO9mbNmkWRIkXypf/p6en50u7NpmAsIiIiUoCio6NZsGABgwYNcm6z1rJ8+XK6d+8OQP/+\n/Zk9ezYAxYoVc9ZLSEjAGANAWFgYbdu2BcDHx4cSJUpw8eFp8fHxfPDBB7z22mtX7MfDDz/ME088\nQfPmzalevTrz588HIDIykubNm9OgQQMaNGjAunXrAMeyb61bt6Z3794EBgYC8M9//pOGDRtSu3Zt\n55P6AIoUKcLQoUNp2LAh7dq1IzQ0lFatWlG1alXmzp0LwK5du2jcuDH16tWjTp06hIeH/4mrmje0\njrGIiIjcnhYNg+M78rbN8oFw7+irVhkyZAjvvPMOcXFxzm1nzpyhRIkSuLo6opmvry9Hjhxxlk+Y\nMIEPPviAlJQUli9fDkDdunWZM2cOPXv2JCoqik2bNhEVFUXjxo15/fXXeeGFF/Dy8rpqXyIjI1m1\nahURERG0bt2a/fv34+Pjw5IlS/D09CQ8PJxevXo5A3doaCg7d+7kzjvvBGDy5MmUKlWKCxcu0KhR\nI7p160bp0qVJSEigVatWjBkzhq5du/Laa6+xZMkSwsLC6N+/P507d2bixIk8++yz9OnTh5SUlFti\nFFojxiIiIiIFZP78+fj4+NCwYcNLtltrL6t7cWQYYPDgwURERDBmzBjeeustAAYOHIivry9BQUEM\nGTKEZs2a4erqytatW9m/fz9du3a9Zn8efPBBXFxcqFatGlWrVmXPnj2kpqby6KOPEhgYSI8ePZxz\nmgEaN27sDMUAH374IXXr1iU4OJioqCjnqK+7uzudOnUCIDAwkJYtW+Lm5kZgYCCRkZEANG3alLff\nfpsxY8Zw6NAhChcunMurmH80YiwiIiK3p2uM7OaHtWvXMnfuXBYuXEhSUhKxsbH07duXqVOncv78\nedLS0nB1dSU6OpqKFStetn/Pnj158sknAXB1dWXs2LHOsmbNmlGtWjVWrVrFpk2b8PPzIy0tjZMn\nT9KqVascb87LGr4vvh87dizlypVj27ZtZGRk4Onp6Sz39vZ2vl65ciVLly5l/fr1eHl50apVK5KS\nkgBwc3Nztu3i4oKHh4fzdVpaGuCYO92kSRMWLFhAx44d+eKLL2jTps2NXNY8oxFjERERkQIyatQo\noqOjiYyMZPr06bRp04Zvv/0WYwytW7fmp59+AmDKlCl06dIF4JK5twsWLKBatWoAJCYmkpCQAMCS\nJUtwdXWlVq1aPPnkkxw9epTIyEjWrFlD9erVr/hY6BkzZpCRkUFERAQHDhygRo0axMTEUKFCBVxc\nXJg6deoVpzjExMRQsmRJvLy82LNnDxs2bLiua3HgwAGqVq3KM888Q+fOndm+fft17Z8fNGIsIiIi\ncgsYM2YMPXv25LXXXqN+/fo88sgjAHz88ccsXboUNzc3SpYsyZQpUwA4efIkHTt2xMXFhUqVKjF1\n6tTrPmaNGjVo2bIlJ06cYOLEiXh6evLUU0/RrVs3ZsyYQevWrS8ZJc6qU6dOTJw4kTp16lCjRg2C\ng4Ov69g//PAD3377LW5ubpQvX5433njjuvuf10xOc1oKQlBQkL04kVtERESkIOzevfuyZdJuVw8/\n/DD333+/cyWMv4ucPmNjzCZrbdC19tVUChERERERNJVCRERE5Lb09ddf3+wu3HI0YiwiIiIigoKx\niIiIiAigYCwiIiIiAigYi4iIiIgACsYiIiIiBer8+fN0794df39/atasyfr16wF46aWX8Pf3p06d\nOnTt2pXz588DkJqaSv/+/QkMDKRmzZqMGjXqkvbS09OpX78+999/v3Pb8uXLadCgAQEBAfTv39/5\ntDlrLc888wx33303derUYfPmzQV01n8NCsYiIiIiBejZZ5+lU6dO7Nmzh23btjnX3G3fvj07d+5k\n+/btVK9e3RmAZ8yYQXJyMjt27GDTpk189tlnREZGOtsbP378Jev2ZmRk0L9/f6ZPn87OnTu54447\nnA8FWbRoEeHh4YSHh/P55587Hy+dly6G8L8iBWMRERGRAhIbG8vq1audT7Vzd3enRIkSAHTo0AFX\nV8dKusHBwURHRwNgjCEhIYG0tDQuXLiAu7s7xYoVAyA6OpoFCxYwaNAg5zHOnDmDh4cH1atXBxyB\ne+bMmQDMmTOHhx56CGMMwcHBnD9/nmPHjl3Sx8jISPz9/enfvz916tShe/fuJCYmAjBy5EgaNWpE\nQEAAjz32GBcfFNeqVSuGDx9Oy5YtGT9+PPPmzaNJkybUr1+fdu3aceLECQBGjBhB//796dChA35+\nfsyaNYuXX36ZwMBAOnXqRGpqKgDDhg2jVq1a1KlThxdffDGPP4Ur0zrGIiIiclsaEzqGPWf35Gmb\n/qX8Gdp46BXLDxw4QNmyZRkwYADbtm2jYcOGjB8//rLHLk+ePJl///vfAHTv3p05c+ZQoUIFEhMT\nGTt2LKVKlQJgyJAhvPPOO8TFxTn3LVOmDKmpqWzcuJGgoCB++uknoqKiADhy5AiVK1d21vX19eXI\nkSNUqFDhkuPv3buXL7/8kpCQEAYOHMgnn3zCiy++yNNPP+18dHO/fv2YP38+DzzwAOCYIrJq1SoA\nzp07x4YNGzDG8MUXX/DOO+/w/vvvAxAREcGKFSsICwujadOmzJw5k3feeYeuXbuyYMECWrRowc8/\n/8yePXswxjinlBQEjRiLiIiIFJC0tDQ2b97Mk08+yZYtW/D29mb06NGX1Pm///s/XF1d6dOnDwCh\noaEUKlSIo0ePcvDgQd5//30OHDjA/Pnz8fHxoWHDhpfsb4xh+vTpPPfcczRu3JiiRYs6R6IvjvBm\nr59d5cqVCQkJAaBv376sWbMGgBUrVtCkSRMCAwNZvnw5u3btcu5zMciDYyS7Y8eOBAYG8u67715S\n795778XNzY3AwEDS09Pp1KkTAIGBgURGRlKsWDE8PT0ZNGgQs2bNwsvLK/cX+E/SiLGIiIjclq42\nsptffH198fX1pUmTJoBjNDhrMJ4yZQrz589n2bJlzsD6/fff06lTJ9zc3PDx8SEkJISNGzeyZcsW\n5s6dy8KFC0lKSiI2Npa+ffvy7bff0rRpU3777TcAFi9ezL59+5zHvzh6DI4AW7Fixcv6mT0sG2NI\nSkriqaeeYuPGjVSuXJkRI0aQlJTkrJN11Ps///kPzz//PJ07d2blypWMGDHCWebh4QGAi4sLbm5u\nzmO5uLiQlpaGq6sroaGhLFu2jOnTp/Pxxx+zfPny67/YN0AjxiIiIiIFpHz58lSuXJm9e/cCsGzZ\nMmrVqgXAL7/8wpgxY5g7d+4lo6RVqlRh+fLlWGtJSEhgw4YN+Pv7M2rUKKKjo4mMjGT69Om0adOG\nb7/9FoCTJ08CkJyczJgxY3jiiScA6Ny5M9988w3WWjZs2EDx4sUvm0YBcPjwYedqGdOmTeOee+5x\nhuAyZcoQHx/PTz/9dMXzjImJoVKlSgDOG/9yKz4+npiYGO677z7GjRvH1q1br2v/P0MjxiIiIiIF\n6KOPPqJPnz6kpKRQtWpVvvrqKwCefvppkpOTad++PeC4AW/ixIkMHjyYAQMGEBAQgLWWAQMGUKdO\nnase491332X+/PlkZGTw5JNP0qZNGwDuu+8+Fi5cyN13342Xl5fz2NnVrFmTKVOm8Pjjj1OtWjWe\nfPJJvLy8ePTRRwkMDMTPz49GjRpd8fgjRoygR48eVKpUieDgYA4ePJjr6xMXF0eXLl1ISkrCWsvY\nsWNzve+fZXKaa1IQgoKC7MaNG2/KsUVEROT2tHv37kuWNpPLRUZGcv/997Nz586b3ZUbktNnbIzZ\nZK0Nuta+mkohIiIiIoKCsYiIiIhk4efn95cdLf6zFIxFRERERFAwFhEREREBFIxFRERERAAFYxER\nERERQMFYREREpECNHz+egIAAateuzbhx45zbt23bRtOmTQkMDOSBBx4gNjYWgNTUVPr3709gYCA1\na9Zk1KhRzn38/PwIDAykXr16BAX9bzWys2fP0r59e6pVq0b79u05d+4cAHPmzKFOnTrO+hcf9XxR\nbGwslSpV4umnn87PS3DLUjAWERERKSA7d+5k0qRJhIaGsm3bNubPn094eDgAgwYNYvTo0ezYsYOu\nXbvy7rvvAjBjxgySk5PZsWMHmzZt4rPPPiMyMtLZ5ooVK9i6dStZnw8xevRo2rZtS3h4OG3btnU+\ndrpt27Zs27aNrVu3MnnyZAYNGnRJ/15//XVatmyZL+eelpaWL+3mJQVjERERkQKye/dugoOD8fLy\nwtXVlZYtW/Lzzz8DsHfvXlq0aAFA+/btmTlzJgDGGBISEkhLS+PChQu4u7tTrFixqx5nzpw59O/f\nH4D+/fsze/ZsAIoUKYIxBoCEhATna4BNmzZx4sQJOnTocMV2/fz8GDp0KI0bN6Zx48bs378fgHnz\n5tGkSRPq169Pu3btOHHiBOB4At5jjz1Ghw4deOihh4iMjKR58+Y0aNCABg0asG7dOgBWrlxJy5Yt\nefDBB6levTrDhg3ju+++o3HjxgQGBhIREQE4/kgICAigbt26zmuVl/RIaBEREbktHX/7bZJ378nT\nNj1q+lN++PArlgcEBPDqq69y5swZChcuzMKFC51TIAICApg7dy5dunRhxowZREVFAdC9e3fmzJlD\nhQoVSExMZOzYsZQqVQpwhOYOHTpgjOHxxx/nscceA+DEiRNUqFABgAoVKnDy5ElnH37++WdeeeUV\nTp48yYIFCwDIyMjghRdeYOrUqSxbtuyq51isWDFCQ0P55ptvGDJkCPPnz+eee+5hw4YNGGP44osv\neOedd3j//fcBR+Bes2YNhQsXJjExkSVLluDp6Ul4eDi9evVyjnRv27aN3bt3U6pUKapWrcqgQYMI\nDQ1l/PjxfPTRR4wbN46RI0fy66+/UqlSJc6fP3/dn8+1aMRYREREpIDUrFmToUOH0r59ezp16kTd\nunVxdXWMU06ePJkJEybQsGFD4uLicHd3ByA0NJRChQpx9OhRDh48yPvvv8+BAwcAWLt2LZs3b2bR\nokVMmDCB1atXX7MPXbt2Zc+ePcyePZvXX38dgE8++YT77ruPypUrX3P/Xr16Of+5fv16AKKjo+nY\nsSOBgYG8++677Nq1y1m/c+fOFC5cGHDMl3700UcJDAykR48ehIWFOes1atSIChUq4OHhwV133eUc\nuQ4MDHROHQkJCeHhhx9m0qRJpKenX7Ov10sjxiIiInJbutrIbn565JFHeOSRRwAYPnw4vr6+APj7\n+7N48WIA9u3b5xzN/f777+nUqRNubm74+PgQEhLCxo0bqVq1KhUrVgTAx8eHrl27EhoaSosWLShX\nrhzHjh2jQoUKHDt2DB8fn8tO2S4EAAAgAElEQVT60aJFCyIiIjh9+jTr16/nt99+45NPPiE+Pp6U\nlBSKFCninJucVdbpFxdf/+c//+H555+nc+fOrFy5khEjRjjreHt7O1+PHTuWcuXKsW3bNjIyMvD0\n9HSWeXh4OF+7uLg437u4uDjnJ0+cOJHff/+dBQsWUK9ePbZu3Urp0qVze+mvSSPGIiIiIgXo4rSG\nw4cPM2vWLOcI7MXtGRkZvPXWWzzxxBMAVKlSheXLl2OtJSEhgQ0bNuDv709CQgJxcXGAY77w4sWL\nCQgIAByjtFOmTAFgypQpdOnSBYD9+/djrQVg8+bNpKSkULp0ab777jsOHz5MZGQk7733Hg899FCO\noRjghx9+cP6zadOmAMTExFCpUiXn8a4kJiaGChUq4OLiwtSpU6971DciIoImTZowcuRIypQp45xu\nklc0YiwiIiJSgLp168aZM2dwc3NjwoQJlCxZEoBp06YxYcIEAP71r38xYMAAAAYPHsyAAQMICAjA\nWsuAAQOoU6cOBw4coGvXroBjxYfevXvTqVMnAIYNG8aDDz7Il19+SZUqVZgxYwYAM2fO5JtvvsHN\nzY3ChQvzww8/XDICnBvJyck0adKEjIwMpk2bBjhusuvRoweVKlUiODiYgwcP5rjvU089Rbdu3Zgx\nYwatW7e+ZDQ5N1566SXCw8Ox1tK2bVvq1q17Xftfi7n4V0NBCwoKslmXFRERERHJb7t376ZmzZo3\nuxt/WX5+fmzcuJEyZcrc7K5cUU6fsTFmk7U26Aq7OGkqhYiIiIgImkohIiIiIrmU9cEif0caMRYR\nERERQcFYREREbjM36/4qyX9/9rNVMBYREZHbhqenJ2fOnFE4/huy1nLmzJlL1ka+XppjLCIiIrcN\nX19foqOjOXXq1M3uiuQDT09P5wNTboSCsYiIiNw23NzcuPPOO292N+QWpakUIiIiIiIoGIuIiIiI\nAArGIiIiIiKAgrGIiIiICKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAigY\ni4iIiIgACsYiIiIiIoCCsYiIiIgIoGAsIiIiIgLkMhgbYzoZY/YaY/YbY4ZdpV53Y4w1xgTlXRdF\nRERERPLfNYOxMaYQMAG4F6gF9DLG1MqhXlHgGeD3vO6kiIiIiEh+y82IcWNgv7X2gLU2BZgOdMmh\n3n+Bd4CkPOyfiIiIiEiByE0wrgREZXkfnbnNyRhTH6hsrZ2fh30TERERESkwuQnGJodt1llojAsw\nFnjhmg0Z85gxZqMxZuOpU6dy30sRERERkXyWm2AcDVTO8t4XOJrlfVEgAFhpjIkEgoG5Od2AZ639\n3FobZK0NKlu27I33WkREREQkj+UmGP8BVDPG3GmMcQd6AnMvFlprY6y1Zay1ftZaP2AD0NlauzFf\neiwiIiIikg+uGYyttWnA08CvwG7gR2vtLmPMSGNM5/zuoIiIiIhIQXDNTSVr7UJgYbZtb1yhbqs/\n3y0RERERkYKlJ9+JiIiIiKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAigY\ni4iIiIgACsYiIiIiIoCCsYiIiIgIoGAsIiIiIgIoGIuIiIiIAArGIiIiIiKAgrGIiIiICKBgLCIi\nIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAigYi4iIiIgACsYiIiIiIoCCsYiIiIgI\noGAsIiIiIgIoGIuIiIiIAArGIiIiIiKAgrGIiIiICKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKx\niIiIiAigYCwiIiIiAigYi4iIiIgACsYiIiIiIoCCsYiIiIgIoGAsIiIiIgIoGIuIiIiIAArGIiIi\nIiKAgrGIiIiICKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAigYi4iIiIgA\nCsYiIiIiIoCCsYiIiIgIoGAsIiIiIgIoGIuIiIiIAArGIiIiIiKAgrGIiIiICKBgLCIiIiICKBiL\niIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAigYi4iIiIgACsYiIiIiIoCCsYiIiIgIoGAsIiIi\nIgIoGIuIiIiIAArGIiIiIiKAgrGIiIiICKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAig\nYCwiIiIiAigYi4iIiIgACsYiIiIiIoCCsYiIiIgIoGAsIiIiIgIoGIuIiIiIAArGIiIiIiKAgrGI\niIiICKBgLCIiIiICKBiLiIiIiAAKxiIiIiIigIKxiIiIiAigYCwiIiIiAuQyGBtjOhlj9hpj9htj\nhuVQ/oQxZocxZqsxZo0xplbed1VEREREJP9cMxgbYwoBE4B7gVpArxyC7/fW2kBrbT3gHeCDPO+p\niIiIiEg+ys2IcWNgv7X2gLU2BZgOdMlawVobm+WtN2DzrosiIiIiIvnPNRd1KgFRWd5HA02yVzLG\nDAaeB9yBNnnSOxERERGRApKbEWOTw7bLRoSttROstXcBQ4HXcmzImMeMMRuNMRtPnTp1fT0VERER\nEclHuQnG0UDlLO99gaNXqT8d+GdOBdbaz621QdbaoLJly+a+lyIiIiIi+Sw3wfgPoJox5k5jjDvQ\nE5ibtYIxplqWt/8AwvOuiyIiIiIi+e+ac4yttWnGmKeBX4FCwGRr7S5jzEhgo7V2LvC0MaYdkAqc\nA/rnZ6dFRERERPJabm6+w1q7EFiYbdsbWV4/m8f9EhEREREpUHrynYiIiIgICsYiIiIiIoCCsYiI\niIgIoGAsIiIiIgIoGIuIiIiIAArGIiIiIiKAgrGIiIiICKBgLCIiIiICKBhLAZgbMZdj8cec76Ni\nozgce/gm9khERETkcrl68p3IjVodvZpX17xKnTJ1OJt0ljeavsHjSx7HYgmpFMKnbT/FGHOzuyki\nIiKiEWPJXz+H/wzA9tPbiY6P5rElj2GxAKw9spZ95/bdzO6JiIiIOCkYS77ae27vZdt8vHxY0n0J\nAOuOrivoLomIiIjkSMH4NnQg5gDv/vEuqRmpedLekBVDmL1/Nq+ueZVpe6Y5tyekJhAVF+V837xS\ncwCaVmhKee/y3F3ibtYeXZsnfRARERH5szTH+Da08MBCvgn7Br/ifvSo3uOK9dIy0rBY3FzcAEhJ\nT2HfuX0ElAkA4MPNH3Ii8QTLDi9j2eFlgONGuwY+DahRqoZz25AGQwg/H86IpiM4FHuIykUrA9Cs\nYjOm7ZnGhbQLFHYtnJ+nLCIiInJNGjH+C7LW5rpuckQEhwc+Quyxw+w6swuAQ7GHAJi4bSJJaUn/\nq5uezN6zjqkP6Rnp/Gvuv3h40cPO8lGho+i1oBc7Tu1g4raJTNoxibkRcy875te7viY9I51xm8ZR\np0wdHqr9EKObj8bT1ZMapWrg5eYFQEjFEFIzUtl4fON1XwMRERGRvKZgfAvKSEzkxKjRpJ09+79t\nSUlEPzuE3f412d+6DRkpKblq69jrb5Cwbh2LRw2m5/yeHIs/xqHYQ5TwKMHJxJOsPbqW1IxUJm2f\nRNC3QXSf151FBxfR4acOHIw5yPbT28mwGaRmpDIvYh4AvRf2ZsLWCZcdq4RHCR4NfJT5B+bz8daP\nOXXhFP1q93OOOGfXoFwDPAp5sDJqJQdiDtzAlRIRERHJOwrGt4iU6GjSTp0C4Nz333N2yhTOfj3F\nWR77yy/E/forAGnHj5OwejVxy1eQHp9wxTaTwsK4sHkzALUX7ydkVwaPLXmM3Wd306pyKwAOxhxk\nfsR8PtzyoXO/l1e/zIW0C9QrWw+A1j+2ZuK2iSSnJ1PKsxQGQ0jFEIq4FWFB1wX0rdkXgBqlavB4\n3cepWrwqX+z4AoOhRaUWV+yfp6snjco34sd9P9JldhfiUuKuWHdl1MpL5ivnhwyb4RwxFxERkduP\ngvEtICUqioP/7Mr+Nm2JW7mS+FWrAUg+EOGskxQWhvH0pMJb/wUg+un/EP3UU5z/4Ycc28xISuLI\niy9RqHhx/Jb8yrkihq7rMjhy9iAAbSq3oeHZ4nh9v4hpm76guEdxvu70Nf6l/AHo4NeBF4JeAOBs\n0lk+3/45LXxbsOrfq/ij7x9MbD+Rdb3WUaVYFR6s8SBNyjfhrZC38CjkwciQkQA8VOsh57SJK3mm\n/jPO1ztO77isfN+5fTT7vhn/Wf4f/jHrH7y/8f0bvmkwITWBlPQrj7R/tu0zus/rzrJDy65ruoqI\niIj8PSgY3wJifp5NRnw8Lt7eHB06jKSwMAAS160n9eRJwBGMPf39KdG9O97Nmjn3TTmc8xPkEtat\nI+XAAcq/9V/eOvQZs4MNVU7DVzub8a5XP2pN+4Ohn52h/s9h3DPnIMMbDycgthj/b6UPnUo3p2/N\nvtxV4q5L2vzn3f8EwKOQB4DzwRx3Fr+TLzp+QXnv8gDULVuXJd2X8HzQ89c895qla7Kk+xIMhm2n\ntl1W/s4f7xCfGs8DVR8guEIwX+/62rk28vXIsBn0mNeD1j+25mj80cvK41Li+GGv44+MISuH8P2e\n76/7GCIiIvLXpmB8C0g5fBg3X18qjRtLRkwMGQkJlHyoHzY1ldOffEJGSgrJYbvxrFULALdKlZz7\nph45kmOb8StW4lKkCEVbtmTn6Z3sa3s33h3a477oN+54/SvOfvWVs+494W60K9ecw489jlmwnMHf\nx+CXUZKi7kUJ7RNKnTJ1CCoXRJvKbXJ9TuW9y+NicvfrVd67PDVL1+STrZ/wym+v8NXOr8iwGaRn\npLP91HZ61+zN283f5rP2n+FbxJdV0aty3Y+LNp/YTFRcFLEpsUwNm3pJmbWWYb8NIyY5xrlixmfb\nPrvq1A4RERH5+9FybTdRRkICsYuXkBwRgXuVKng1bozbHVVIPXSYom3akpGQQMycuXg3bkxGYiLe\nLRzrAJfs24e4JUtwrVAhx2BsMzKIX7kS75AQvo+YwYGYAwwMGEiZh1uSsHgJ7lWrUmncWNJOnSI+\n7iwMeZnIDveSfuYM7lWrcmHzZs5Nn07ZwYMp7FqYb+79BmNMroPujehwRwfCzoQx/8B8AM4nnycm\nOYYLaReoWaom4BihbuHbglnhs0hMTbzmNI3YlFiKuRcD4JfIX/As5Mk9le7hu93fUat0LR646wFn\n2ero1bzc6GX61erHztM76bOwD59t+4wXG72Yb+csIiIitxaNGN8kaadPc/ixxzn2yisk796N2x1V\nMC4ulOzVC+Pmhkf1apTq2xd74QJHnn8BFy8vvIODAfCsUYPqG9bj3bQpKQcPcmHnLme7yenJ7Fg3\nl7RTp9ha3ZXRoaMB8Cvmh1eDBlTfsJ6q8+fhWb06RUJCKNfhH3hUq0b6mTMU/9e/uGvhAjzr1CFh\n9W/ONgu5FMrXUAzQvXp3+tXqx08P/IRnIU8m75zMzPCZANQqXctZr4NfB5LSk5iya4pz+bmczIuY\nR8i0EBYcWEByejJLDi2huW9zRjUfRePyjRm+ZjhdZndh9v7Z/HfDfwkoHUBv/94ABJQJoG2VtkwJ\nm0KPeT34cPOH7D6zmwybka/XQERERG4uc7NuMgoKCrIbN96e69fajAwO9e3nXDECoNxrr1Gqbx9s\nRgZpJ07gVqECAFFPPkX8ihWUe/VVSvXre0k7cctXEP3UUwB4BQfj8nR/hs0bTK0oyz/XWR59thBx\nXo55wN/c+w31fern2J+0s2eJX72aYh064OLlxcnx4zkz8TNqbNmMi6dnflyCq9pwbAMrDq+g052d\nOJF4gk5+nZxlGTaD+2bdx5F4x0j5d/d9R52ydUhJT+GHvT9wZ/E7aViuIR1+6sD55POU9y5PRe+K\nbDm5hYntJ9KsYjNikmPou7AvkbGRgGPO9Hf3fUeNUjWcx/n92O8MWjzokn41Kt+I0c1H4+PlQ3pG\nOoVcCuX/xfgbmbJrCnvO7mFU81E3uyvyF3X6wmnKFC5zs7txS0pMTcS9kDuuLvoiWCQnxphN1tqg\na9ZTMC54F7ZtI/LfPSn/5hsU79aNpO3b8QwMxMXD47K6Ni2NtNOncS1XznmzW1Yn3n2Xs19Ovmz7\nnkrwxkOudLmrC13u7kJQuaAc989J7C+/cGTIc9w5ayaetWqRkZKCi7v79Z9oPvlyx5eM2zwOgG7V\nujGi2Qje2vCW8+a5i+71u5dFkYsAGNN8DPdVvc9Zlp6Rzvd7vuf3Y7/zevDrlPMud9lxouKiuG+W\nY587it3BycSTFHUvSobN4PSF0wA8EvAIwRWD8fHyoaRHSUp6lsyXc/6rO55wnPY/tQfg996/X3Ma\njEh2O0/vpNeCXjzX8Dn61bry+ui3o9SMVDr81IH6PvX5oNUHl5SlpKfw9a6v6e3fmyLuRW5SD/Ne\nTHIMb657k27VutHct/nN7o78BeQ2GGsqRQHKSEkhZsECzv/0EwBFO3bExd0dr6CgHEMxgHF1xa18\n+SuG2qLt2jnqeXlxuF8rvmvlwqliENW2JuNaj+PV4FdpVL5RrkMxgMddjtUokiMiODv1W/Y1akzi\nli3Xc6r56uHaD/N5+89pV6Uda46sYeymsfyw9wda+ba6pN6bzd4EIKRSyCWhGBzTQ/rV6sfHbT/O\nMRQDVC5amVX/XsX2h7Yzv+t8Xmr0EicTT3L6wmka+DQA4MudX/Lo4kfpMrsLrX9szZJDS/L+hP+C\nUtJTeGnVS8zc55gOk/W67Du372Z16y/pu93f8czyZwp0CcGo2Chm7JtBekZ6rupba1lzZM0NL6V4\n1b7ERZGYmsioUMc3DWM3jSVkWojzj9PrcSz+GDP3zWRM6JhLtu87t4+pYVM5l3QuT/pckEKPhTL6\n99GcvnCaJYeWsOfsHk5fOM3pC6cJPxfOggML+GjLR0zaMelmdzVPTdw2kWWHl/HZ9s9udlfkb0bf\nuRSgE2/9H+d//BEAz1q1cC1V6k+36VW/PtVDfyfd25MX5/6LmLtK0fOtT3moiC8lPEvcUJvud9wB\nhQoR/9tvJKxajU1O5tT7H3DHt1OvvXMBKORSiKYVm3LqwimWHl7K5J2T6VatG68Gv4q1ls6zO9Ov\nVj+83bxZ3mM5Rd2L3vCxSnn+7zO6v+r9jFzvWKP5iw5fsPXUVhYcWMCSQ0vw8fLhROIJRq4fSZWi\nVdh9djdNyjehnHc5hq8ZTvi5cL659xu83bz/9Pnfanac2oFfcT8upF3gzXVvEpMcg5uLG5tPbuZo\nwlG6Ve/G4sjFFHUvSlxKHMsOL6OeT72b3e188cfxP/Ar5kdZr7I33EZkTCSjQ0czMGAgjSs05tfI\nX9lycgvrjq4jpFJIHvY2Z59u+5RPtn4CQNnCZZ0PA7qaFVEreHbFszxV7ymerPtknvVlbsRcXl3z\nqvN9+zvaU9i1MHMj5rLxxMZLplnl5Pvd35Nu0+nt35vjicfpMbcHcamO1Wb61+7vXGLy/zb8H5tP\nbiY6LppXmrySZ/3PT1/v/JpZ+2dxMObgJdt7zOuBu4s7boXcSEj93wOgcvuApEnbJzE3Yi7jWo+7\nbMnOG3Ey8SR7zu6hhW8LouOiKedVDrdCNz7a/8fxP/hoy0fOVYMizkdoatstYFXUKqaETeE/9f9z\nxWmbfxUKxgUk7dQpzs+ciXFzw6amUvaFa6/xm1uFihXjs62fcCj2EMU9ihNQJuBPtWfc3Snati2x\nc+eBiwuF69cncdMmx5SOMrfO/L52VdrxKo7/aQ5vMtz51eov3X5x1vkzASW7wq6F+bTdp5TwKIFb\nITcalW9Eo/KNGNFsBACHYg/x8C8P031ed8Axd3lyx8ksOLAAcPyHI/vI9fVISE3gfPJ5dpzeQUXv\nigxdPZRXmrxCC98rP10wv80/MJ9XfnuF6iWrE5Mcw/nk81QvWZ3tp7ZTuWhltp/azne7v2Prqa28\nFPQSq4+sZsquKfSp2ccZSv6sqNgoxm0eR/fq3Qk7E8bhuMO81uQ1YlNiOZZwjIAyAYSfC+fDzR8S\ndjaMDnd04JHAR/B282bTiU2sjFpJC98Wf/o6bj6xmYG/DuSeSvfwabtPr2vfb3Z9w+/Hf6daiWr8\nduQ39p3bx/pj63m1yavsObsHgMk7J+d7MN5zdg+fbP2EUp6lOJt0lq92fkViaiL+pfypWqLqJXUz\nbAa/Rv6KMYYFEY7f8RWHV+RZMM6wGYzfPB6Au0vcTc8aPelRowfpNp0lh5aw5cSWqwbj4wnHnaPM\nk7ZPIik9iQtpF/Ao5EFyejIbT2ykTeU2bD21lS0nHd+ILTq4iKfqPUVxj+J5cg45nVNe3Mi88/RO\n3t/0vvO9RyEPPm77MY8ufhSAlIwUUjIufZjRphObSE1Pxa2QG7tO78LHyyfH/z7OiZjDodhDvLDy\nBd5t+S5H4o/k6o+j7Ky1GGP4aMtHzN4/m5HNRvLGujeoXLQyn7X/zLk05vWIjIlk4K8Dne99Cvtw\n8sJJlh1exrmkc/iX9qdu2brX3W5eSstI45uwb0hMTWRgwMDbYtpYanoqI9aP4PSF0yw9tNQZjBNS\nExj22zCi46IZXG8w7e5od5N7mjuaY1xAEkJDOfxQf6p8NZnCderg4p13I4cZNoMus7sQGRvJey3f\no6Nfxz/fZkoKcYsW4Va5CoVKFOfAP+6n1IAB+Lz04nVNy8hvocdCKelZkmolq93srgCOG8ze2/ie\n8337O9qz5NASCplCtKrcinGtx91Qu3vP7uW5lc9dNupTu3Rtpt8//brbi0+JZ1b4LDrf1fmGv1kA\neHTxo2w4tsH5fuq9U6nnU4/0jHTWHl3L4GWDAXB3cWdtr7XsObuHfov68VGbj27of7bZRZyPoN/C\nfs5RwIv61erH4sjFnEg8wfR/TGfYb8OcN1sCdL6rMxk2w7k8YGnP0vzS7Rc8XT1JTk/mfNJ5vNy8\nmLRjEvEp8bwe/PpVf++nhk3lnT/eAaCYezHW9FyT639PTiScoN1P//sfhrebN8ObDGfhwYWsPbIW\ncKzMEnYmjAltJ9xQgN94fCMjN4ykt39vevr3zLHOhbQLdJrZibNJZxnfejznks4xYv0IZ3mP6j2o\n71OfZhWbUbpwaWbsm+H8BgXAy9WLxLRE5v1zHn7F/S5p+3jCcT7Y+AFDGw+ldOHSuerzvIh5DF8z\n/LL7AwAeW/wYW09t5dN2n9KwXEMOxR5i84nNdK3WlV2nd/1/9s47PKqiC+O/Te+FVBJC6ITee5Pe\npXcQBAEBgYA0RSkqCIKCShUEpEmT3lvoHaUGSEhI771vtsz3x5ANS0ILoPiZ93nyZPfemblzZ295\n58x7zmG733ZqutRk+jkpJYtKjyI0NRRHc0eO9TxG061NyVJn4WjuSGR6JABf1v+SeVfm0dqzNfOb\nzs+vS8/ExYiLbLm/hTE1xlDOvly+ZX64/gMbfDcwsdZEBlUcBEgSdT/hPpUcKuldL5vvbSY2M5bx\nNcfnaScsNYwee3tgbmROvaL1aFuiLS2KyxjzW+9vJTI9km5lu3Ek6AgDKwzEP8mfW7G3+O7qdzR0\na4i5kTknQk5gqDDkSI8juFi6IITgZOhJFCgY7zOeUralCEwO1B1zUMVBnAw5SWXHyixouuCF13Zi\nViJDjwylunN1jgUfI1mZnKfMr21+pW7RugA6wp6DnAhAOZOIu3F3We+7nosRF0lU5spdZjSYwbIb\ny/RkNbu77M5j6dYKLQoUr/TuylBlsOLWChq5NaKOa52XntAsvr6YX+/8CkAxq2JMrTv1jTzrCor4\nzHjmX5lPp9KdqFe0ni5BF0BgciA3Y27StUzXPGOz6d4mfON98a7prZtAJWUlYWlimUffn2MgAWha\nrClLWy4F4FToKcaeHKsrd23gNb3j/90odL57x5C8bx8Rk6dQ6uBBTEuVfGPt3oi5wRfnvyA4JZhv\nm3xLp1Kd3ljbTyJi6jSS9+zBsEgRSu3ZTcqhQ1jUq49Z+fxfAv9VZGuyORB4ADcrN11UixrONfAq\n4sVO/52c6XPmlS0IKq2Kxr83JkOdQdsSbWlXoh2XIi+x7cE2jA2M2ddtH25Wbi9sJyErgQ2+G2ju\n0ZwVN1dwNvwsNZxrsL79+jxl51yaw9Hgo6xtt5ZStqXy7BdCkJKdQpsdbehYqiP+if5Ym1izrNUy\nvTKrb6/mp79+ooVHC35s8SPpqnTqb67P2BpjGVF1xCuNQ36YdHoSZ8POsrL1Sr48/yWl7UpjaWzJ\n3oC9ujL1XOtxOeoy7lbufP/e9xwMPMjGexvRCi1FzIowo/4MvE95M7r6aIZUGsKgg4PwS/TDxNAE\npUYJwLdNvsXCyIJStqXwtPHUe4lEpkXS5o82ev1a1nIZ5YuUx9rEGnMj82f2XwjBwmsLWe+7nmUt\nl+Fo7kgZuzIYGxoTnhZOuz+kRfRIjyOMOTGGFGUKHUt1pI5rnZd2OErLTqPjro4kZCXgZunG4R6H\n87wE78bfpe9+SZjH1xzPsMrDAPly3Ppgq96k4j2P98hSZ3Ep8hIVilRgXM1x3E+4T8eSHWm3sx29\ny/VmVPVRejKkcSfH4RPqw+hqoxlV/fkWZY1Ww96Avcy4MAMHMwcOdD+QR4K0++Fuvjz/JQArW69k\n5LGRgEwWFJUeBUBZ+7JEpUVxtu9ZNELDxFMT6Vy6M21LtOV27G0mn5msi24DcOuDW8y5PIe9AXs5\n1fsUJoYmPEh4wN34u3Qr2w0DDPSW6yPTIlnvux4XCxed9dbTxpO9XfdioDBAqVFyJOgILYu3xMTQ\nhObbmusIYp/yfXA0d+Rq1FWuRF3hq4Zf0a1sN13bVX6rAsC+rvtwtnBm4umJfFjpQ2IzY3UEZE3b\nNdRxrfPcscyBVmgZdHAQt+Ju5dm3vNVy0lXpTDqdG7N9buO57A3Yq5v0GigMMDYwRqlRUsauDPOa\nzONM2BncrdxpX7J9nutpzqU5bHkgJ+zGBsYMqjiI9b7r6Vm2J5bGljriuKnDJpwtnOmzvw8DKwyk\nr1dfglOCWfznYu7F3wTO0sEAACAASURBVKNdiXakqlI5HXqaDHUGZoZmfNngS2o618Qn1Iee5Xri\nn+jPyZCTBKUE8Wf0n7hZubGl0xbuxt8lPjMeR3NH+uzvA8DMBjNp7N6YiLQITI1MqeRQ6ZljljMx\nA7A3tae+W31sTWyJzYzF3syemQ1m5qmTrcnmvW3v0dCtITWca+jCpT4tMdIKLWqtmqj0KGIyYrga\ndZU2Jdq8EelKDh4mPuTPmD/5/f7vPEx6CIChwpAeZXvQtUxX4rPidaQ1J4a/EIKHSQ/J1mbT/0B/\ntEJL02JNWdJiCctvLmf5zeUMrzKccTWlz8OSG0s4GXKSh0kPKWlbkjJ2ZbgVe4uPq32Mi4ULx4KP\nsevhLtqWaMuRoCOsa7eOWi613tg5vioKifE7hrhVq4j9/gfKX7/2xqzFadlptN/Znix1Fs2LN2de\nk3lvLd5w1oMHPOoiU0KblCxJ9qNHmFWrSsmtW19Q87+Lzrs6E5QSxK73d5GqSuWDQx/QuVRnPq/3\n+St5hz9IeEDPfT35ot4X9PHqo9semRZJlz1dcDJ3YnWb1RS1KppvfY1WQ1xmHMtuLmOn/07ddgcz\nB+Kz4ulRtgeR6ZE4mDmQnJ1M2xJt9XSdMxrMoFe5XnptHgg8wLSz0wAZ8aNNiTYoUOSr8/MJ8aGm\nS03d8nS7P9pR0aFiHu/5V8XBwINMPTuVEVVHMLbGWNRaNYYKQ5KVyYw9ORYPaw/iMuO4GHkRgD1d\n91DKthRxmXE039YcgOn1ptPXqy9TTk/hcNBhupftzh/+f1DXtS4lbUvSsnhLRhzTJ/ADKwxkat2p\nuu9TTk/hUNAhHMwcWNhsIXOvzCUhM4F0VTpaoWVjh41UcKiQp/9KjZJvL3/LH/5/0LJ4Sxa9tygP\nwfjh2g8UtSpKP69+3Iq9xYCDA3T7VrdZTb2i9Z45PvsC9vEg4QGN3Bsx4tgIOpTswMFHB1nZaiUN\n3WVa+Ux1JiEpIcy8MJO78Xdp49mGb5t8i4mhfhSawORAUpQpLLi6QI9c7e26l5K2uRP9wYcG82fM\nnxgZGPFbu9+o6lSVO3F36H+gPwKBjYkNS1supapTVQ4EHqCyY2WOBh2lZ7me2JjasOj6Is6EnSE4\nJZiqTlX5ucXPegQ7B5nqTOZensvuh7ufef4g/QKeFR7wduxt+h+Uscu9a3ozrMowrkZdZeiRodRy\nqYUQgj9jckNqNnRryMrWuY5ePff25EHiAwCqOlWle5nuzLo4i9/a/UZNl5psvb+Vby5/g4uFC/28\n+rH4z8V8Ue8Lvrn8TZ6+VChSgZ9b/IyLpQsqrYqaG6Rzb05s9ZwU9QoUVHSoyLga43S/4csiNTuV\ny5GXaVqsKWqtmiNBR5hxYQY9yvYgIi2Ci5EXqehQEQ9rD76o9wX3Eu4x4tgIvmr4Fe1LtkcjNNTf\nXD9Pu183+pquZbqSpc7C+5Q3bTzbsPj6YuoVrcfYGmNxNHfEwthCJ60AmHdlHpvubaK6U3WSlEm6\niVdN55q6MXe3ctebuMxsMJPmHs2fu+Kw6d4m5l2ZR6virTgechxjA2OqO1fnatRVQBJDjch1KB1b\nY6wuWodKq8JIYYR/kj+RaZHsD9zPiZATDKwwkLV31+Y51spWK2ng1gC/RD9K2pYkW5NNg98bALCg\n6QLalWxHanYq31z6hkOPDnGw+0GKWRdj1a1VrLu7DoHQPSMg7/VVEBwJOkK2JptOpTrR9o+2utUQ\nQHf/P40c3496rvUwMjDifMR53b6c38PIwAi1Vg2AkYERNiY2NHZvzN6AvXhYe1C/aH06lerE+Yjz\n/HLrF732Paw9WNN2Da13tNY9b/8pFBLjdwxRX39D8r59lL9y+Y21ue7OOr6//j2/d/z9tXXFL4PU\nEyfIunuXVJ9TKO/dA6DMKR+MXd+MVvT/DQlZCZgYmOhI8NIbS1lxcwUNijZgWatlunijt2Nvcznq\nMsMqD+NI8BGW3VjGlo5bCE4JZvr56fgn+gN5SQhIR5SxJ8dS2aEyq9qsynep0NvHmxMhJwAp7Wjo\n1pDojGhaeLSg9/7eABS3Lk5sZiyZ6szcejW9ORJ0hHsJ99jQfgPl7MthbmROqiqVcSfHcT36Oj3L\n9eTzup+/kjPN9HPT8Qn14XSf088NuSWEYPKZyRgoDPiq4VeYGenH1B52ZBgxGTHs7rI7X0IuhNB5\n41saW3Kx30Xd+OwL2MeKmytkmnHrYmSps2i/sz1xmXEYKgy50O+CzrL/7eVv2Xx/M8MqDyM4JZjj\nIccZVHEQTuZONHJvRI+9PXRWFICbsTcZeDA35ngJmxLs67YvT/9yZDc9yvZgRoMZLzWpvRJ5hRux\nN1h3dx2p2aksabGEZh7N8pSbfXE2O/xk9JuRVUey8tZKjvU8xgeHPsDJ3ImNHTaSqc5k4KGBuuvr\nZV5a58LPMer4KKyNrTnT90yemLlHg47y6elPAfig4gcMrTyUjrs6YmJgwtS6U3WTqSctuwClbUvT\nukRrVtxcgbuVOxNqTaBV8VYvdKiKz4znvW3vYW1sTY9yPYhOjyZTncm16GukqdJeaFWNz4zH3sxe\nb+z/8PtDT0LyJMZUH0Nj98ZMPzedwORAStuWxt3anYm1JuJq6UqLbS0oYlaEpa2W0nV3V+zN7BFC\nkKhMxNXSlYPdDnIx8iKmhqasvbuW8+HnGVhhIBvvbQSkxTk8NRy1UOd7fFtTW35t86tezPXXwScn\nPiE4JZh0VToN3Bowp/Ecvf3R6dE4Wzjr7pvApEBGHR9FRHoEv7b5lWU3l3E9+joAjuaOepKGxc0X\n07J4y3yPK4Tgxz9/1FmOx9UYx9IbS3WkdW7juXQu3Rm1Vs3CawsJSw3j5xY/v1AKkZCVwPu7388j\n3Wjj2QaN0Oieg09jSp0pLL+xHE8bTwKTA8lQZwAwoMIAptWdRnxmPLMuzuJU6CmqOFYhLjOO2IxY\narrU5ErUlTztne1zVidRi0qPot0f7ajkWImRVUfq5GUgifq4muPwS/TjQOABJtWeRJ/yffI863KQ\nlJX0TOlbbEYsLbZLSc24GuP46a+fqOpUlbqudWno1pA6rnWISIug/4H+xGfFA3JFwsnCiZ57exKW\nFpanTZ/ePnTZ3YWU7BSauDehtWdrZlyYodtf2aEymztu1rs+ltxYQv2i9UlXpZOsTKZe0XrUL1qf\nquurAnIi1aV0l39EkllIjN8xhH7yCargEErt2/viws9BbEYsDuYOuugLjuaO/Nb+tzfUy5dH8v4D\nREyahFmVKpTcvu1vP/6/FdsebOPrS1/Tq1wvItIjqOJYhRU3VwAyJnNOtr82nm04FXoKQwNDMtWZ\nmBuZc6n/pXzJ0+rbq3VOSoe6H6KYdTFALtclZiXScntLytmXo69XX7qU7qIjG0IIlt5YiqeNp06C\nE54WzuTTk6lTtA4Ta00kQ5VBp12diM2MBaBZsWbciL1BsjKZKo5V2Nxx8yuPwYmQE3j7eOtZLkFK\nRtRatU56cCH8AiOPyyXypsWaMrfxXJ3VOTYjltY7WjO08lAdIc0Pt2JvMfn0ZDqW6vjccgDjT47n\nZOhJKhSpwLbOude0EIKw1DA8bDzyJIfJwcHuB/WciT499SknQ0/St3xfNt7bSKdSnbA0tsS7prdu\nojTi6AhJ7Ls+3+qZH65HX2fI4SH08+rH5/U+121Xa9XMvDBTT0pibWKNsYExp/ucZoffDmZfnM2Q\nSkMQQvCb7290KtWJyo6V6VO+z0slh7gWdQ1rE+tnkrN0VTqfnPiEDHUG9YrWY+2dtbrJ+9GgoxwP\nPq6LL54f/hz45ytNtFKyU1BpVHqWRK3QkqJMKbB+/quLX7Hdbzv7uu7D0MAQYwNjeu7rmYdwPf27\nnwk7o0d8epTtQU2Xmqy9s5aJtSbqyV8yVBlcj75OI/dGzL8ynz/8/9BJdwCm1Z3GmbAzdCvTjWLW\nxdj6YCuf1f3sjTpz5VhYASbUmsDQykNfUENanjVaDXZmdoSmhNLvYD/cLN1IU6VRx7WOblXqRXrS\nwORAuuzuAsDND25yPvw816KvMaTSkNeKB6/WqrkZe5PPzn5GJYdK+IT6sLfrXhzMHfjm0jd42njS\n0K2hbvXFydxJ93zLgaHCkF7lejGx9kQ9KVRUehR2pnb4Jfrp6rf2bI2ZoRk3Ym9gbmTOtk7b8kzo\n1t9dz4JrCzA3Msfe1J49XfeQoc5AgQJ7M3v8Ev3osbcHIPXcU+pMyXNeVyKvMOzoMH547wdae7bO\ns/9p/xZPG0+2ddqW7/XyV8xfOFs4427lDsjEOUqNEp8QHxQKBR7WHhS3Lk4J2xKMODqCi5EXWddu\nHVUcq1BrY64U4kWrVk9iwIEB3Iq7hbmROYe6H3ppX4M3iUJi/I4hsHNnjN3c8Vi5osBtnA07y+gT\no/mq4Vekq9KZf3X+M2+Stw0hBFEzZ5G0bRtlTp/G2MX5b+/DvxVfnv/yhUvAIJdXl7dazu242xQx\nK0JVp6r5lgtNCaXDLumcVM2pGo3dGxOaGqpHjrZ22qqXWvtV8KRs4kkMqzwM71rer9yeUqOkyZYm\nZKozeb/0+zRxb0L5IuWZcmYKSo2SbZ22YWJowoADA0jISqBNiTasu7uO9iXa810z6eA29/Jctj/Y\nzp6ueyhuU7xA5/U0fvrzJ1bdXpVHsvI0fEJ8GOcjSXYjt0Z4WHswvf50vTLpqnRCUkKISI/A2yd3\njIZWHsrQykNRaVW03dGWPl598n0Jvgw+OvIRKdkpeiR+l/8unUVnQ/sNjDs5jkRlIo3cG7Gi1QpU\nWhVfnv9SFyklJ0HOm8aqW6v46a+fAPLo2JUaJTv9d9K0WFPOhp3F3cqdqk5V2fNwD47mjq8VueVN\nQSu0hKSE6DkRhqaE4hPqw8XIi6Sr0hlUcVCeZ68Qgi57uvAo+RFVHKuwrOWylybnOc93GxMbFjRd\n8MpSiYIgMSuRplulM2dBHTvVWrXehOqH6z9gY2LDR1U+ek4tiVHHR1HGrgyf1v70lY/7MlBpVYSl\nhuVZaQPwT/RHK7S4WLjQZKucsMxoMIO7cXeZUGvCcyOTCCEYfHgwlR0r6+5fIQRqrTrfSZ1Kq6LJ\nliakq9J1Moun28uxqBYxK4JPbx+SlcnYmdoRlxnH0eCjugmMpbEl0+pOo0vpLhwLPoZA0NqzNQMO\nDCBNlcbadmu5FnWN2q6130iWyJiMGA4EHmBwpcEYKAzwT/TnYdJDErISGFBhwIsbeIy4zDhUGhXp\nqnTK2Jd57X4VBIXE+B2COjER/wYNcfL2xvHjkQVq48nIEzmo5lSNDe03/GNRIrL8/Hj0fhdcZ8/G\nvk/vf6QP/0bk6BgNFAY6fdm8JvO4EXODolZFaePZBp9QH7qU6YKNic1Lt+mf6K8LT/Uknrek+bKI\ny4zjatRVppyRL4FuZbrxRf0v8mhRXxZL/lryzMD8/b36E5oaytnws7ol1W8vf8s2v20c63kMU0NT\nWm5vSWvP1nmWfl8HKdkpnAs7l68z0dPlGv3eiCGVhrzwhZ4zaSlpW5IKRSrk0fgtb7Wcxu6NC9Tf\n5TeWs+LWCsbVGMfme5uZUHsCC64uAGB4leEMqDAA33hftvttZ3jV4TrrkBCCdXfX8TDpIVPqTHkr\nocky1Zl8fOxjyhcpj3dN7/9EyKocBKcEs/7ueoZXHf5KIQmFEBwJOkLTYk3/1vE6H36ey5GXGV19\n9DOX8P/fkePseHvw7bd2jK8ufkV4WjgrWq3I9/lyNeoqhx8dZpvfNqyMrUhTpTGt7jT2B+znTvwd\nANws3YhIjwCgfcn2HHokV16qOFbhdtztPP4PhdBHITF+hxA6egxpJ0/iuWkjFrUK5pF5Pvw8Hx//\nWG/bx9U+Zkz1Mc+o8fYhhCCgdRtMS5d+LUv4fw1CCC5GXKSyU2VCU0JRC/Ubi70ZkBSAq6UrN2Ju\nYGJogoe1xxuLFwxSwjDr4ixmNpiJs0XBVwmEEFyPvs6MCzN0jm4dS3VkwdUFOvI4qtooRlUbhUKh\nIDglmE679COubO6wmSpOVV7rfAqKhKwEbE1sX6iBFUKw3W8773m8h62pLStvrmRPwB5iMmIAuDLg\nynOjVjwPEWkRtP1DPzSjAgVbO23N19mvEIUoRP6IzYhFqVHqZGj/FKLSo2i9I+8K8OTak2nt2Rpb\nU1uyNdn0O9CPsLQwHM0dsTax5lHyI6o6VmV129UFfp78F1BIjN8RaFJT8atTF7NKlSixdQsKo1fP\nqRKaEsrgw4PRCA0Lmi5g4umJJCuT/1FikIPoefNJ3LSJ0sePF8opCvHaSMlOYeKpidRyrpUnrNfE\nUxP1Ukvf+uDWOxVT+1UQmBRISGrIa8c33RewD7VWzYHAA1yOukzbEm1Z2GzhiysWohCFeCfx818/\n4xvvi0qj4nKUdNa/0O+CXgbXDFUGgcmBFLcpTmRaJB8d/Yjvm32viwtdiPxRSIzfEWTdv8+jrt1w\nX7wIm3bPT1+aH1RaFb329iIgOUAXaxAgWZn81rIzvQqyQ0MJaNceEw8PSmzdgqHtP9+nQvx/QqVV\ncTz4OFPOTKGNZxu+f+/7F1f6j+Bc+DnW313P/KbzX8txqRCFKMS7ASEEX5z/gjqudehapus/3Z3/\nC7wsMS5MCf2WoQqTIVCM3Qu2RHM79jYByQF5sj+9C6QYwMTDg6Jff03k55+TvH8/RQa8vBj/n4Aq\nPBxjd/d/uhuFKACMDYxpX7I9dVzr6FlPCgGN3RsXWKtciEIU4t2DQqF4oz4UhXh5vJ1sEIXQITuH\nGBcrGBm7G38X4KUzHP0TsOveDdPy5UnavgOh1f7T3Xkmknbu4mHLVgT16UtQn76oExNfXKkQ7xwc\nzR3/0bSihShEIQpRiP9fFBLjt4jM27eJmTcfjI0xtCtYPM07cXdwtnDW5Sp/V+EwbCjK+/dJWJ83\nvfA/jcQtWwjs2o3Iz2W818ybN8m8eZPYxT/+wz0rRCEKUYj/E/y1EQ58ChoVXF0N5xb/0z0qRCEK\nhEIpxVtEzHcLwMCAol99VWAnoYdJDylv/2ayHL1N2HTqRMqRo8R8twDLunUxqyhj5gqNhjQfH4yL\nFcPMy+tv7VPCpk2kHj2GOjqa7KAgAGy7d8e0dGmU/v6kHDiAy+efYWBaaH0sRCHeSSSFgIUDmFi+\nnfZVmZAcBo5l3077/wQ0KtjnDZ4NoMbAF5d/HQgBCgWcWwTHZ8ltIZcgWoYXo3J3sHszccbfGqJu\nw45hULkHNJ4ARgULQVmI/x8UWozfEoRWS+bt2xQZNBC7bgUTzgshCE0NfWMJDN4mFAYGuM2dg6GN\nDbFLlgJyDMInTSLsk7E86tad+LXrECrV39Kf6PnfEf31N2Rcvkx2UBDW7dvhNGECRed8g8Owodh0\n7oQ2LY2006f/lv4UohCFeElkpcC2wTDLFhZXgQOT3t6xdo+GJbXh3j4IviCJ3r8d19bAjY3gM/fZ\nZWLuwaFpkJX87DIvwqMzMNseHp3NJcUgSXEN6SSO757c7Umh8u9dw5VfIP4hnJoL12SKah6egMxC\nqd1/FYXE+C0gfu06gvv1R2RlYVKmYBle1Fo1433Gk6nO1Es5+i7D0NYWu969STt1iuzQUEKGDiP1\n0GHs+/fH1MuLmPnzSTl8+K33QxkYSMK6dXrb7Hv3xnHkCJ3l3rJePQxsbAgfN560s2f5p6KzFKIQ\n/wlE3Ya1HSE5PP/9Wi2EXIbMJNjxIfg+kRky8NTb69ddmb6YrQNhbXu4uPTtHSs/qJVw9As4/wZk\nXVd/hfkl4dDjTIop4bCxJzx4Iv22EKDKkud6ebmUPmSlSCvz04i4AbF+zz7esRmAgN8exxcfeQYa\nTwT32tDmayhWB66tffzbXoLFlWFVC9nuyTmQkfD65/wiPDoLWwaARp3/fo0K7u6BKr3AviQEnYP4\nANjYHTbnk/0yNQqyM+Tnd/Wd8fCE/G2yUuTKQbTvP92jfx0KpRRvGEIIYubP1303LSAx3nRvEz6h\nPgD/GmIMYN+vLwkbNhDQvgOo1Th5e+MwYjhotTyoXYeIyVMwq1QJ01Kl3lofoufMxcDSktJHDhP5\n5QwM7e2wqKsf31FhZITDsGHELlpE6PARWLdrR7HFiwAp/9AkJGBgbY2B2X8zE1QhCvHGIARs6Abp\nsbB9MFQfAPaeULpFbpkzC6TFLgctZ0DUHTA0hltbpaXR7g0/ByNv6n8vVgdOfQuulaHUe6/WljIV\nDn8GWjXUHgYeTzlLp0bDxZ+h2TQwtcrd/vA4XPhZfi7bBpwLmJglPR4OTMy7/eExCPQB7zty/Dd0\ng4w4uc+zMdzeLv+q9IYeq3LradTwSzP5+eNz4PpUvPzsdDnZyUH9MVC0mvzLQYMxsH0IrH8/1zKd\nHpPb7pnvpHSh5Uwpx3jTEAKOfCb7mRAITuVAqwEDQ0nWk4KldVuZDF4dZZ2AE3Dzd/k59DLc3SV/\nI0NTqD0U1rQFUxto+4383exLQLeVYPRYjqdRgYHR2zmfHKgyITEo/2slPV6SeoAbv8vxNjaHdnkz\noupBq5XW8opdwerd9mf6O1BoMX7DUPr56z4b2tpiWvbVtGuhqaH8/NfP7PDbodv2byLGxkWL4vbt\nt5iVK4fDyJE4jByBwsAAhZGRTncc/unbWxrNuHqV9PPncfxkDEYODngsW4rbnDkoDPNmKHMcOQKL\n2jKkYerhw2Tdu0fyvn3cr1QZ/yZNCez8PqqYGNSJiQitlvTLV4hdshSh0eR7bCEESTt3kXX//ls7\nv0IU4l+FjATY+4kkZQBhV2G/tyRofkcleUkKhXM/6NerPwZ6rYWG4+T3/KzG0Xfly/9ZlrsnZQIa\nlSQN6fG5lspD08DUFpwrwnufQ5dlYOUiLYxJIS8+ryetqf7H4K8NklTtGqlfVq2EdR0kkfrWXZ53\nDgJOyv8KQ7i0/PnHfB78Hq/EtZkDDmVyJx1Fq0uyfuxLWNlEkmLPxlCxCwzZn1v/9jbZzxzrYqBP\n7r4VjeU+ZVrutrCrst0OC2HwfmiXj2yjYldo8QUEnYWoW9B0Mgw7Bq2/kr+va1WpTb655c3KFtLj\n5Bj7Hc4l7/EPpfX6ey9IeAQHJ8FP1eVKgaEplG4O5drI6/TMAjlJMjKTxP6vjZI0rmgE2WlyErRj\nKET8JYnzoSnyGgy5BF875q463Nkp21emvrlz06hhdStYVh9Cr0jr//2D8vcBCLvyxDjI7JokBuvX\nv7srt3wOgs/LMdk2SEpknmVh/4+g0GL8hpHmIx8oZc6cxsjBIV9C9jxMPTOV23HyZp5SZwpl7ctS\n0rbkG+/n24RNu7bYtGubZ3vRb74hbNxYlPfukR0Sgknx/LXT4Z9OIvX4cTAwwNDejlJ792JoZZVv\n2SeRHRpKyIiRGNjYYNez10v11WniRCI//5zsoCASf99C8q5dun2q0FAeNm2WW1ihACEw8SiG0t8f\n67btMLSzJTs4BMtGDUnaupWoWbMxdHKk7OnTKAwK552F+A8i9oEkCiaW0sKWEi4JaLcVcH4xpEVD\nWgxs7iWtxzc2yXrdV8HO4dJabPx4pcalEti4S5JTc1DuMVIiJWETWkgJAztPKN8eTB/Ht763T5KS\nIQfBzBb2jJZWNoG0EI65AuHXoN5IaPNNbrsDd8BPNeDGZijXFtxq5D2/kMuwpg0oDGDSQ7B0yHU2\nA0nMtFrIuf8v/CyJWQ4295LnauUMt7aDVyewdJQEsfl0sHZ59TEP9JGkvsEYaPgJpERIh7Luv0jZ\nxO3tspy5PXx4ILde/22wuTcYGEvpQKAPfLAXAnz02//mcVbTplOg2RS5X2EIVXvL8c0PCoUkw0Hn\nZbu1h4KNG3g8Xr3TaqS0YvfHsl9THhXc0qrKktIQt5pwZDpEPybE1m6QGiEnZxnxcttP1XPruVSG\n93+W103FblBqg7T4dv8Fjs+EP9dLaYhHPbi0FCp1h84/wpHPpRU9KQQuLoGw67nHPDpdEtQcfXWF\nLlC1V66jYg6yUiDiz1dbnbizI/daO/8juNeEE19B6ZYwaCeE/ymvy76b4dQ8QEDsPUmEjUwl+b2+\nFrr9AtWekIrc+UP+D7kIv3WGcu3lZKfI21vZfZdRmPnuDUIIQVDPXqBQUHLH9leun6xMpunWpmiF\njAV8sd9FrExeTAj/TVA+ekRg+w64TJ+ObdcuGFrLF5kqMhIDa2sURkY8qFMXhaEhhkXsUUdE4rlh\nPRZ1XhzHOW7FSmIXL8Zz8yYsatZ86T4JIfCr3wBtcjIG1taUPihfHNmhYaTs30fqiZOoo6Mxr1UL\n5cOHaJOlJcrQyRGFgSHq6Og8bRZf8yuWDRu+dB8KUYh/DVKjpPXQ8RkysVmPiZKRuSSug/dKq6yZ\nTW6Z8D9hVfPc78UbwocHJam0dNQnEEe/kORy0G6wdoXr66QW+dYWaR3NIZ2NvKH1bEmSfqwGaVEv\nPpe+m3OX0XOwtB7EPl71GX4S3Gvl7gs8Deu7IBk20HON1Kauag4uVaDReNj5kSTk1q6ybHKoJFdl\n20ipBk+8c83sYMQpOU7L6kuC2Hs9FK//4r7nIClU6ncrdZdW9qdx/yBs6Sc/d1okCeqTuLVNTkhy\nULyBtLAbGkPND2C3fmp23GtLaUKx2jDgJd5zWSlStvC0HAMg7Bqsbik/T/AF2wImXzo2U066clBr\nCNw/AL3WwbrHv2/JZpKUn1kgiV/v3yQJNniG8SotRhLJtnPl73R8FjT2BtsnknVlJcO8FzjHl2gi\nx/Ts9zDuLykjAljTThLRzj9KK3bYVei3Jfc+iXso9+dMCLWPrxFDY0nob23RP06fTZIkGxrDqPNy\n26l5j6855ATw6Bfyc8NxUgd+fZ2UAKky5JhYu8hVGM1ji/K0kGdPfP6FKMx89w8g9egxsu7exXXW\nrALV3xewD63Q8lXDr6jqVPX/jhQDmJQogZGrK9Fz5hC3fDllfE6iSUwkoG07RHY2CjMzUKlw/3Ex\nZl5ePGzREmVAfvRLSAAAIABJREFUwEsR49TjxzGrVvWVSDHIDEOmZcqQef06RQYNxMhJaqyMnJyw\nqFmDIkOGEPH557jNnUvilq0krFmDZePGZFy9mmsVegzLxo3JuHaN1OMnColxIf7/kB4P3z8OHzkt\nVJ/sgnzB50CdCfVG5U/y3GtCxx+kLrZkUxi8T27PT9/YfDrc3Cpf4rEPpAUM5FJ87/XSQvjggNSH\ntp4tLX1Pk+IyraX1MuCEtIL9tREcSkOp5nkOR60PH1vmfKVF88NDkhz7H5ORHsxsYfRFSVL8j+eS\niLrDwauD1KBe+xWMLSQpBml5fm+q/FOmyji/GpWUGxR5vCLYdbk8l829pTNYcpgcI2tXKT+pNUSf\npOdg22PiVLJp3n0g+/RF7LPDkHl1hPIdpEXTvRb4PLagN54A1fvLMbIoAihkv498JvdX65d/e0/D\nzCZ/UgySXA87Br+2lhbr6v2hwWi5Ly1W6rLda0npx9PISpGrEbe355JiMztoNw+q94NOi+UEq/0C\nqeuu3l+eY+WeMjzfswhxDqyc5fWVg44L8zk3W+j8E5xZCB+fBUMTuPCTJKOlW8rzPr9YyklAXn+1\nPpROfiEX5bZ943Pbu/MH1P5Qfv5jGETekPeKSyW4vw/iHsjJWFpMLjHu/KM8/tbHWWf7PzFZqT4g\nlxgf/UJekyZWso/l2sGhqaDOkvtrDIQKnaDOR3I1BuT95qHvn/NfQKHF+A0icuYsUg4dotzFC68s\noQCY4DMBv0Q/DnQ/8OLC/2Ik79tPzHffoY6NxWPVKjKuXyN+xUq9MmUvXsDQzg6/WrURGg1mFSpg\n1awplo2boM3IIGLqVEps3oSxmxsgLc4Pm7fA6dOJOA4fnt9hn4uouXNJ/mMnpY8fw8je/pnlNGnp\nxP/yCw5DP5SxkY2NUd6/j4GlFeHe3nisXEHS7t2kHjmKw8gROI8fjyoiAlVUNBY181mWLUQh/gXI\nvHMXkZ2Nxa0vc1/yFbtK0lJ3uNRvqtKlBS3HmazVLGg4Ps/kUQetRjrAOZR+sVVqzxhJZgHe+0we\np85Hsi7ox9EFue/ePinbmPJILpUbGr/aSd/eIcmJZ6PHGtILcnubb6DhWP0+1fkIOn4vP++fIEOm\nAZgXgeafQ6Vu0hL+IkT7wvIGud8tnSTx2tJPSkr6/S6jJhSvDyik5nVJbWlxH7z31c/xaaTFwM4R\n8rjt5z8mxE9Aq4Wv7BFaUHwZnSt5eR2osmDOE/KRL+Ol7GXJE5OAL+OkrtvaRU4sTKxkRI2c8GoA\n3VdDlZ5v1/HtZRH+pyTFadGwrpO8/qJuQdW+kujmRA5xLCcnQA3Hwt3HKyKD98p9i6tKS3ud4fJe\nWt1STqY+uSqvgaV1clc8rq/LJdizngrBF3NP3ms3NssJxu1tcoID8r776CQEn5Mh9gwM5bV+fKaU\nary/RF/CpEyViVscysjJx/Pw1yZ5f9QZJn+XdwAvazEuJMZvEEF9+6EwMsJz44YC1R94cCBmhmas\nbrv6Dffs3YM2Kwu/ho0wKeGJNjUNQxsbPH/fTMTUqdi0bYtNu3YABA/6QFpm84Hr7NlYNWuKJiGB\nsAkTUAWHUOrQQUxLvromW5OWjjY5CWP3Ai7lIeM2KwwM0KSlEzVrFin79+OxahWhj4m6w8cjQStw\nnjihYO0/vlfDvSdg1fw97LoWLD7260Kbno6B5VtKuFCIfJF64gSxP/2M+8IFr+zQW1AIjYaE39Zj\n7FGM8LHSCa587ygMmk2QmsWLS2TB0i1ynchAOjNNeiBlAY+hTU9HYWqKwqiAi5QPj8PGHvLz9Cjp\naf8kwq9L624ORpyWlsrUKPB8jZWbi0ulnhSkVKNkUyjzeOk/7Jp0hLLzgKFHpH4WIPKWdHQDaDVb\nLr+/Ci78LIl+++8kSRX5O/vqYfC+Z1uM3zDUNw8S8OF0XGfOwrZLPpbcgsD/2OOwakqpMc9IyL2+\nAGyKSWeyppOlFKLuCFlHaKBYXWg6KXeS9K5i18e5ES9ASlKGn8j9fnCyJK9Tg+Hk17lWcCsXOfHy\nmQsDdkDZVnK7Rg2Gj+8nVRYsqgj1R8kxeh6yUmTs5lvboMN3+WuctRqYUxTqjcjV4Gs1sLYDhF4C\nY0v4PFznd0PQORmhIydyTGIQ/FgdEFKL/tGx/Fc7/mYUEuO/GUII/GrXwbZLF1xnfFmgNtruaEtN\nl5p82+QFoVX+T5B68iQRk6egTU/Htls33L7N69msiowkaft2rFu3JnrefDIuX85TxtjdHVV4+DPb\n+CegTkzEv0lTUOf17vXcuEEXDeNlITQaQkeMJP38ed22CvfvvXY/XwUZ168T9fU3KB88wH3Ronwd\nLAvx5qFJS8e/YUNEdjYAHqtWYdWkcYHaUkVHE/nll9h164ZN+/b5llH6+2NSqhTpFy4QOnyE3j63\n+onYTlomM5qlxcLPtaQzm0NZNO7NyDi4EaNG/THu8iVGRaS1Me3sOcLGjcOsXDmK/7auYCEQhYBl\nDcC1Mspq04hZsBC7Pn2wbvFYCqFRw9cOueVnJLx4qfxlkRIpnayK18u7LzNJLk8/LVNIDpdEuaDW\nyxxHrYNT4MpKsHDMDbP2NCr3kM58b+p8X4DUU6cI+3gUJqVKUWr3LjA2LnBmVz2olXLyk7Mi4VpF\nammfJJNAZoK0ipsXUUGLLyUp/jcgNVo6hJZoJCVGFg65xBbg1nZUG0eQXWYIlhFyxUFU64/q/FZM\nrDRyPD4+9+z2n3bue1388p7sc042QGWqlGN41JfkeMRpGZ7v6mqpxS7TCgY+duI7PltanEdflDKZ\n7AxwqQjV+suJ5T+UabJQY/w3Q+nnjzY9HbOKBYtDKYQgNjMWJ/P/TgxB6xYt8Ny8idBRo7Bp3y7f\nMsZFi+I0Tlqriq9bS8rBg5hXqkRAp8460plDiot+Nftv6/uLYGRvj8vnn5Fy4CA27dphWrYsyXv3\nkrxzJ1GzZ1Ni+/Y8BEFotaSfP4+JhwcmJUro7Uvet0+PFANolcq3ls46/cIFknbtxrRsWey6dcXI\nyYnYJUtQPg5FF+7tTXKLFli3bIldj+65/TxwgOSdu7Af0B/rFi1IOXwYdVw8RQYOeCv9fJeReeMG\nRkXdMHZxLnAbQghivvsOkZ1NkWFDSfh1DanHjumIsVCpSDt/HqvGjZ9pjU2/cAHlwwDMKlYgYtpn\nqMLCSD9zFssmTTG0yrX8Zz14QPS8eWRcvKSbbAK4L16MkbUxYWM+JjXcDNuc+KlWTtBlCQRfINW4\nBWGfjAeKwLnDKH48hefGjZh5lSdu+XJEZiaZN2+SeuIEth075tPLF0ChgFHniV/3GzGdOgOQefs2\nVmfPSNmaoRF8dEI64hmZ6pFEoVaTeesW5tWrFyxSjE1R+ZcfzO3y315QJ7Ic5BCcljPksnW1vjLS\ngbk92BaXMXbLt5cRCP5m6UDWbRkVITswkPtVq2HZtAkeK1e+FDnOvHmTmO9/QBUejsfKFZiWKUN2\nWBixPyzCrk8fLN//OTdqRP9txKzZhlnxqdhUc4P0OFRBAQRtOQVA0bqJmDYpjvmzD/duwdpFWk6f\ngTifYGL3ugKH8WjvhtV3fxG7+Gfi97tQsl0MZnUbPLMu8Oavg3LtpD750BMW6BJNpBZ+cWUZj7pU\n89yQeEHnZPbIS8ulRMqzITiVl9bus9/LbTmxw6cG6a0ovWsotBi/IcT+vIS4Zcsoe/YMRo4voSV7\nCsnKZBpvacyUOlMYVHHQiyv8x5F2+jSqyCjiV61CFR5OGZ+TGBd9xsvrHULa2XNSWqFQYNu1q87C\nrU5MJGzsWDKvXQcjI2y7vI95tWrY9+4NQFC//miSkih18ABpp04RNmr0a1kOn4RQqaQO1MCA1CNH\n0GZkEjVzpvSdV6kwsLTEvGZN0s+exWnCBEw8PQn3lsvDBra2lD19CgMzM7RKJX516krLppER5tWq\nkXn9OiDDFxo75xLElIMHybx5C+dpU0m/cIH4X1ZhP3AANq1bv/b5vA4yb93C0MYmz8TkVaBJSiLL\nz4+QDwZjYGND8V9Xk3b6DI4fj3xlKUHa6dOEjvyYIsOG4jJ5MmFjx5Jx/U9KbNuGSTF34tetI2be\nfOz798N1xgy9utlBQSTt2UPC6l91qdgNixTBfkB/4n5egsfq1Vg1bgQJj4he8D0Ju3KXdS0quKOJ\neoSlSzYu71cGp3JELNtF8iNLDO3ssO3eHcfRozC0siLj+nVChkiHIatWLdGmp5N+Rlr9DKyt0aam\n4jx5EnG/rMK6ZUucJ3ijMDHB0DZ/XXHa6dMYWNtgXr2aHpHV3TtPIOceENnZZIeGooqIxLJ+PRTG\n0qqYfvkKUbNmkf3oEUW/+Rq7nlLrmLBxE0aOjphVroxJseeTWGXgI0KHD8eidm3c5s8jdtkylH7+\nuM37tkDWb21WFknbd2DXu9dbm9i+SWT5+mJcrBiGNjYIlYpHPXoitBqESoUqWMZ7dhj+Edbt2mFe\nqVKe+sl79pBy8BDFli8jcsYMkndIq6L9oEG4Tv+cMO8JpB4+jGXjxjLWfbQvTt5jUWXbEtBWGky8\nbt1EYWJCzKLFxK9cicJAILSSCJbav6/AibQA3fPqVSdNOYYMI0dHzCq8nEFMaDSkHjuOablymJbK\nlfxlh4UR0Er/2WfsWVw3vq5timA/eXHexDFPty8EaLUF8nHKg4RH0hG09deQGgkIqDlYTjo39pTJ\nYx4jNroepsqb2JQi15mv+RfQbLK0ZGcmwuUVUu9cpRdUfP/1+1cAFEop/mYEDx6CNiODktu3Fai+\nf6I/3fd2Z0HTBbQrmb/1tBB5oY6PR2Fq+lJxjt8VxCxerHM2LHvuLAbm5kTOnk3K3n0U+fBD0i9d\nQnlPyiQMLCwwr1GD9PPncflsGkUGD0arVOLfoCFGLi54btqIoY0N4ZMnY2BpiYmnJ5qkJCxq18ai\nRg2U/v6YenkRNXMWjmNGY1paX4enTkwkeMBATDw8sG7Thsjp0+UOIyPKnDiBNj2d8E8/JTswEPMa\nNfBYuQIUCoIHDMS0XFmS/9hJ0blzsevejYzr1wkeMBDnyZOJ/fFHRHY2ZtWqknXzFk4TJ+I4QpKa\nJ18CVi1bkuXrizoyEiO3opQ5fBiFyTO8598yUo4cJXz8eEy9vCi1excxCxciVCqcJ00i7cwZrJo2\n1RGupyGEIPXQIYyLexI6ciSa+Pg8ZUzLlkWrVGLVuDHWrVuBoSHm1arlS460WVnErVxJwrrfMLSy\noszJEyiMjcm8e5eQIR9iXrUq7j/+SFCPHmQHB2NgbU2Zkyd04Q8zb90i5KPhaFNSMHJywm3Bd2Rc\nv47Ve+9hWqIED+rWw657N5ynTEE5tyHBu9RYlzHBZc5ijLICUBz6FKxc9aI7ZCs8iAqtR/o56YRW\nZPBgXD6bRsT06aQeOkzpE8d1jqv3q1VHKJVYNW+ObbeuWLdqRfjET8m4cgVNQgJmlSpR8o8dec87\nO5sHVavpxqvYsqWYeHigzcjgQU2pUTQpXRq3uXMIGztOriiNH0fI0GG6NqzbtKHYTz+Sde8ej7r3\n0CUAMfbwoPjatYjMDAI7576Y7fr2wXX6dJL37MGycWOMXV31+hTzwyLif/kFAMuGDUm/cEG3z9DJ\nEZGtoug3X7/UpC7rwQNiFiwk/dw53BYuxLZTAaznfyO0mZk8qFETs0qVsO/fX/d8cP/pR6waNwYj\nIwJatUYdI5NJOI4eheOoUbr7RJ2YiH+DXI23ob095lWrYmBpQdrpM5Q6eJBHXbuiSdRP8FFs+TJS\n9u0n5eBBAGw6tMesUiViFizE6r33sOvQjLApcoXQ1MsLMy8vHEd9jImn5yudn8jOJqBTZxQGBhRf\nu+aljSvpl6+QeuwYiRs3YuTsTJmTJxAazQsnOpGzZpG0ZSvGxYtTev8+3bMufvVqYhZ+T5mTJ9Am\nRhHYQ3+FzaZzZ5zGjH7uhF2oVIQMH4E2LY3ia37F0MbmmWVfG6pM6TRoYokyIJDAfvLZ7tUngsQg\nR4ycHLH5bPM7p/suJMZ/M/ybt8Cybh3cnkgH/So4FXqKsSfHsqH9Bqo7V39xhUL8ayGEIH71amK/\n/4Fiy5eRtHUbaadOAVI3LLKzSd67l8gvcrXqRi4ulD5yWGehil6wgIRf12BWpQqOo0cRNmr0C49r\n1bw57gsXkH75MurYONLPnSX12PF8yxYZ/AEun332wvMI7NQZhaEhnps2kvj7FmJ/+IGyF84jsrJ0\nkTiCB31AdnAwJXZsx9jZmfg1a4n57jv9vjVrRtrp01i1bIn7oh8w+JvJcdKOHXrjXerAfgLf7wIa\nDbbdupG8axcmZUpTbNEijFxddQQ0B1m+vpKEPQHLRo0o8sEgUg4fQZuWhjY9HXViom7SA5JcOk/6\nVI9wp1+4oEf0XGfNwr5vbjD++LXriJk/HwNLS7RZWdh27kzy7t0AlLt6hcxbtwgd9hEAZhUrUuzb\nqRhHnYY4PxnuqccawucsJeXQ42xpCAzNFZTpGImB0RPvgx6/SuezWoPB7wg0Go8oVhfl/fvErfyF\njEuXKLlvL4+698CiTm2KLVqkqxr97TwSfvuNclev6Mbq6TH2uuebZ/k949o1ggcOwtTLi+ygIBQm\nJpQ5dpS006eJmDIVh+Ef4eTtjcLQkOQ9e4iYOi3vj2lgQNkzp4maM4e0kz6U2L6N7IAAIj77HENr\na0zLlcsjS3oStl26oDA3o+jjsJtBffqSeTM3fbShgwMKAwPUsbG6bU+mlAdpTQz+4AO0yck4jhuH\niacnj97Xd1RzHPsJjqNHo46OzkPGQV5TKUeP4jRu3FtNFiSEQHnvHmlnzqKOicZ56lQMTE3JeuDH\noyec6xSmpgilUq50zcv1g8m8dYv4tWtBrSH12DEUZmYUGTIYZ29vkvftJ2KyvjOY08SJ2LRpTWDn\n9zGwsECTnIx127aknTmD64wZxP70E+rISFnWezwKIyNiFn6vq58TIz7Lz4/MmzeJ+lKulFi3b6d3\nDT4JVXQMMQsWYNmoEUlbt2LXuzdxS5agjo9HKGW4PatWLfFYsiTf+k8i4+pVggd9kHeHkRElt297\npvU4R5tt7OGBKjQUt+/mY/u+nKCFjRtP1r17lDkmsyJm3rmLOjYGQ2trYhf/SMZjruQ8eRIOw4bp\ntSu0WpQPHhDzwyLSzz7WZxsbU3Lb1pe2ZL8K1AkJoNVi5OhI5t27RH/1de79YWgAGpmHwbxWLdQx\nMVg2bEjR2bPeeD8KgkJi/DdCm53Ng2rVcRwzBqdPxrxy/Ux1JtPPTed48HEu9b+EhbHFW+hlId4l\naDMyeFCnriQ3KSmAtLbk6KlBOkEZOjqSvHMnNh065LFmpBw8SPiUqTqttccvK1HHxmHVojmxi39E\nm5aGyFbqyK+RkxOG9vYo/R6nslUosGrWDIfhHxE1Zw5K33u4L16EkZMT5tWqvdSyf+qJE4SN99b1\nIcfa+iTSzp4ldPQYTDw8KLlju7RkpqdTYsvvKAMDUUdHY9W8OYmbNxP99TcojI0xr1EDtwULnqnP\n1WZkILRCTyMLUnf9sNl7OH7yCfZ9+7zUOQghCGjVWmrVu3cneedOTEqVkpZYCwu0qY9TuhoZ6c4z\nJ4mMNiMDTXIyGdeuETFZhmAyLl4ci9q1cRw5Io8FS2Rnc/+xRdS8WjUyb97EqGhRiv30I+ZVqqBO\nTCSoT19UoaG4zp6FSXFPLOvrO31plUoC2rdHHRGJ0/hx2PbooZ+hESlxsWnXDqfx4zHa3EaS4hy4\nVkGTqSF4ayzKJEnIi80ci7VjtExecPCxM9Nzgvtn3r5NUO8+0hprYCDJSv3ceMVCrUabkaFntcoJ\nqZiDUgf2661gCCGI+HSSLuRl1v0HhAwZgkmZ0pgU9yTzxg3Knj+nR6YTt2xB+TAAi5o1sKhbl+zQ\nUIL79dftt+vVi6JffwVAlp8fwX37oc3IwK5PHxxHjyZ8wgQ08fFkBz+RNvcxbDq0p8jQYQT16oXj\nqI9RmJlj3aoVpqVKoo6PJ+PqVazbtiVi6lTSz1+g7LmziIwMkvbsIWXvPjJv3NC1ZT9gAImbZIa/\n4mvXEDljJiYlS6BNTSPzr790ZE+oVKRfvkLEtGlo4qSznftPP2LdokWea1mrVJJ25gwW1avrYq8/\nDSEEIiODjBs3MHZxwcTTEyGE3sQz9fhxwj4Zq/tu7Fkc8ypVyfzzT1QREXKjkRHGzs4UnTsXi3p1\nn6knTjl8hHBvb7mC4XOS8PHeZN25Q4kd2zG0syPj6jUsG9THwNyc2J9+Im7ZckwrVqDEhg26qCWa\npCSSdu3GwMoSu549USgUxK1YScqBAxRbukQvY6oQgoQ1a4hZsBCMjHCe4I39gAEYmJkhhEChUKDN\nyCCoTx+U/g/z7bN127YYWFqSdvo05c7n79ymSU1FqNWIjAyCBg1CHSGJe4kdOwjqmRuKzHH0aJzG\njc1TX2g0BHboiMLEmBI7dhDY+X0UCgXF16/H2MUZ/xYtsKheHfcffshTN+v+fTJv3CBhw0bQail9\n6KDe+YcMHkLGlSsoTE1x/GQMRk5ORM+Zi4G5OW4LFmBZ783FIRYqFQGdOqEKDsFx3FjifvoZjI1x\nmTKFlCOHpRQQKa1J3LJV9+zMkToKIdDExxdIbvom8LLEWN44/8BfrVq1xP8LsgIChG95L5G0Z0+B\n6s+7PE9UXldZVF5X+Q33rBDvMuLXrRMP23cQETNmCnViYoHaSNzxh/At7yV8y3sJrUqVZ79WrRYp\nx4+L2GXLdOWiFy0S2RERQp2crCunSkgQmb6+BT6PnLaj5s3Pt0zqmbO6Mr7lvUTsipX5lgv9ZKyu\nTPyGjc88ZvCIEeJe5SpCFRf3zOPcr1NXJB89KuJ+XSOSDxx4Zlupp04J3/JeImHrVqHNzhZhEyYK\n3/JeImzSZJHw++/Cv1VrkXrqlEg5cVI8bN9B+Jb3EuHTpwt1aqoIeL+LuFetugj6YLDwrVhJZEdE\nCI1S+dzxyrz/QCiDgkTc2rV6Y5IdHS2iFy4UvpUqi/Tr15/bRlZgoEi7eElotVohhBDZ4eHiQf0G\nwre8l/Bv3kJkPXwoC0bdFWKmjRAXlwuREinEnZ3y+0wbof3CRoS3LS4yPnEQIjsjt/HbfwhxYPJz\njy+EvPaiF34v0i5dfmHZHKRdvCgy798XvhUribAJE/Su2Yzbd+T1ufB7IYQQWq1WBHTrphuf0PHe\nL2xfq9GIe5Wr6Oo8PY4Zt26LqPnfieyoaL3tCb//LiJmzBQpx48LVVyc3u9yv249oU5Keu44+Jb3\nEhl37ojwKVP16sYsXar77N+mjVCGhAghhAgZM0b/fli+QgghRNT87/S25/w9qFdfZPz1l95xQ8d7\n6/bHrVolIufMEalnzgitWi1ily8XQUOGiPt16+nK3KtaTQQNGCjuVakq4n/7TYR9OklkBQSKh+3a\nC9/yXiLg/S4icccOETToA+FbuYq4V6OmiJo7V0R9O++F4/4kkvbt1z+3lb/k/1up1SLL319oX3C/\nvAzUKSkiZOTHwre8l/Br0lQk7tolohcsEIHde+h+g8DuPeTvWaOm/N6zl25MY3/5RfiW9xLq1NQ8\nbWvS0sTDtu2Eb8VKur/EXbtE2mV53Yd9Okn+vi1aioAuXYVWrc7TRvKRI8K3vJdIPnRICCFE6unT\n4l71GiJowECR9fChfN6tW/fcc4xbI58XkbNni+iFC4VWrRYxS5YI3/JeImjIEJEdnXtNR86erRt/\ndXKy7jnxOsgKDNR7PvuW9xKBvXrr3iPqlBTxsENHkbR/v66OMjRM+FaoKKK//0FotVoROWeO8GvS\nVKhiY1+7PwUBcE28BD8ttBi/AeQ4yHj+vhmLGq+exKHzrs4EpQTRsnhLFjdf/OIKhSjEYwitluhv\nvkGbnoHb/HnPLZv14AHJu/fgOGZMHkvr60IVEUHCb79RZMiQfHV6QgjuV6io+17mlE++y8ea1FQy\nb94iauZMjN3csOvdG21GBgYWFqjCQjFydsHqf+zdd3iUVfbA8e87M+m9QkggEHqvUqSJiF1E17aK\nddVdy/5s6+rqWnbXdYsu69oVC2IvK6IuUgRUpHcIIYEQEtJ7TyaZ8v7+uCGTkEmYCek5n+fhmXnr\n3KHlzJ1zzzlnLkdnqkWHwVdfjfeYMQRdeon6WvXZZ6lNPtZ0gJrGiIRDjWa6dKuVzAcfonztWryG\nD2fgxx9h8FXf1pgTE/GI6e/09yn7iSco+bxpfqxHdDRD1jtPTXFGt1oxJyVRk5hE9uOPo3l6qiYa\nZ51F7PvLT3+DU9hKStT4jUbH4ptVv1dNEB5IUKviAQ6tUCvJJ9wAL02C6ffAhR1b5jDnz3+h+KOP\nCFx4GfbyCiLuv4/Kn38m77nnGy3U1K1WylavoXDpUiLuv4+AeU461Z2iau9eSr74gr6PPdbqetuW\n3DyyHnmEqm3biHzkEcJuvaXZc63FxRydNRuPqCgsGRn4zZqlUl+MRvznzKHg9dexFRYRcP6C+ln1\nik2bSL/jTvznzcOcmIjvlCn0feKPHJmqvh3wnzeP0JtvpjYtjZynnqp/Lc+4OGJefpmKDesbpRc0\n5D1uHOYDB9x6vydLr53MebWVloLR1Kr/J2wVFaRcdDHW/Pz6mfCOUrljB3n/+hfm/Y3fv9fQIcR+\n9BGVmzfjER1N5eYthN64uP7vR9natWT+3334TpvGgHffQTMY0G02Sr9aSdXu3ZR++WX9QtJTF7rq\nFguWrCyq9u4l+1GVfhayeDEhv7wOr8GD0XWd1Guvw1ZczODV39X/2zw1FejUb1BOZcnMJP3ue6hJ\nSgJUulvRe8sJuOACop9/rlE61qlpMD6TJjHgraX1/785U7ZuHX4zzm72zzz5vAVYMjIAlVtvyy+o\n/+asJen33Ev1nj30e+450m+/nZAbb6TPY39omxJ/bpJUig50MnduyIb19Z3YXFVkLuKcT89h8ajF\n/Hbib/HauyfeAAAgAElEQVQxdZviM0K4pXzjRszxhwj79Z2nzSHO+9cSCpcubfEcg68v9qoqAHyn\nT6dq2zYAgq68kqBLL8F77Fj1g23Pbsq/W03cN183ao5hTkzk+KIrABiy/nuXm7vYzWbSrr8Bc0IC\nff7wKD6TJpP5u4cIOGceff7gJOfVBanX30D1nj2A+hoy8qGHWnWfRhJXqXbB46+Dy19xfk5Juuqo\n1o45rM7odjsnbrq5PnfSe9QoNE9PbGVlDF7VNTp/6nY7NYmJeI0cedof4pm/e5iyb7/Fa8QIBn72\nqUs58tX79uE1dCgZ992vAraYGCzp6fR5/HGCFl1en5tdm5ZG6dffUPCK+jMMvu5aSld8hd/sWfT7\n+z8o++Zryv63itDbf0XO03/CmpND0KJF9Hn0EeyVlfWLMzPvux9LVhbB16m0Aq8hQyj58ks8Bwyg\n75NPtulX7raSEqz5+R3WjKYhS24eyXNValHAgvOoOZpM3z//Cb+pzb8/c1ISxy9XDZNCFi/Gf+4c\ndKu1fu2GISiIYVs2U5uSgmdsrNMFwvrJVKBVjlQHr6FD6lM4GlZFOenkojvP2FjiVn/nUrBoq6jg\nyIyzwWLBe9w4Bn7wfrPjSb3uuvoPCX5nn03IDdfjN3t2/d9PW3k5Bn9/alNTSbnoYnxnTCf23Xfr\n71H69deUrfoOY3Bw/TqGPk8+gc/YcVRu3kz4b3592vFWbt3KiVtvUxseHgzfvq3FAL09SWDcgQpe\nf538F/7D8P373C6/87+U//Hopkf55JJPGB3etNSNEL3RqTPMPpMmEXrrLVRs2EjpihUYw8IY9Nmn\nFC1fTs2xFCp/VrmB3uPGEfveMgw+jg+YltxckuefR/AVi4j6y1/q95/8QNuabonW4mKVY7pgQZvM\nfFTv30/WHx4j6PLLCbnu2mZLmbnlvYVQlAJ3bT59y+VOUPHzZnKeeoqACy+g6L3lYLUS+buHCLv9\n9s4emtv02lrK169vUhvaFSeDI2NQENEvveg0gNNtNizp6WQ99rj6AGU0ErfyqyZlyqrjD2FOOETQ\nwoVNSsnZKiowHzyI3wxHPVy9Lg+3pyn+5BP0mhpCbrzRpYWLem0t6Xfd3WhRpvfYsZgP1tXoNRgY\nmXDo9PepyxHPf+EFzPHx9fv95sym/+uvNxmLbrVSvm4dvtOm1TfEcUXBa69hrzYTfs/dLcYc9spK\nao4dI+P+++vzogMWnEfMSy9Rm55OysWXqPUUqan1ixBPTvBVbt1K+l13Y/DxwV5RgfeYMfR//TW3\n/2/SdZ3sxx7HXllJ4IUXEHjxxW5d35YkMO5AOX95htJvvmH4jqZd2U7n8Z8fZ1PGJn649gcMWsfO\n2gjRlRW89hr5/3kRzdOT2I8/qq+Raj58GFPfvvWlwRouCBn01Qq8R4xocq/cfz5H0Tvv4Dd7NlHP\nPINeYyb/xZfUgpvt29p11X+nsFng7wNg4mK4+LnOHs1pVe3ZQ+k33xD50O/aPM2nq9N1HfPBg3jE\nxJw2OKrcto2KH34k8LJLndYMFmfGVl5O1c5dZNytZor9583Dkp1N2K9+RdBll7p1L91qxZKTA7qO\nR79+bVNbuJWq4w9Rk3iYmqNHKXpvOYEXX0TZqu+anqhpBF1xBX5nn03Www/jOWgQA959B1N4eKeO\nv61I57sOZD2DVZZJRUmMCR8jQbEQpwj7zW8Iuf76JjMUp5Yg0jw86PvEkxR/8EGzhf4jH3wAzWik\n8K236r9mBZWn1+OCYoDsA2CpggGn6ZbVRfhOmnTaXMWeStM0fMaNc+lcv+nTG1X/EG3LGBBAwLnz\nCL3tNoreeYfwe+9p9QcQzWTCMyamjUfYOj5jRuMzZjS2igoqNm9uFBSPOJxQ/+1cwIIFlK5cScUP\nP+A9biyx777baWkPnUkC4zZgLcjHFBbm9nU2u43jpceZ0a97/PASoiNpmuby13b+s2aqLm7N3ctk\nIvKhB/EaMpisxx4Hm42wO+4g4rf3ttVwT688R3WAimz72qIAWMzgUff1+Ym6JhSxHbfwSYieIvKB\n+wm+6qpG3el6AqO/PwM//hhrTg7l6zdg8PFG0zT6v/0Wek0tlsxMyteuxVZUxIBlvTMoBgmM24Qt\nvwDv0aNOf+IpMioyqLXXEhcU1w6jEkKcKujyywm89NLO+VrwP+NVu9SnSuDUvM7aStj3EQxdAGVZ\nEDrYUUWiOboO6/8M/afB4a8h6Tu4dxf4hUHaVggZBAFNK38IIVqmeXj0uKD4JGNAgGp002BhpP9M\nNalgLSig+uABQm++Ge9hwzpriJ1OAuMzdDKPyP/cc09/8in25u0FYFhI7/0LKERH65SgeOsrKigG\nKM+GwH5gt6sAOT8RProWSho0mRh6AfzyYxUkB8U0DaRTf4ZlTtoJ73kPZtyrjo9a2PS4EEI0wxQe\nTvQpXUl7ox6YXNexatPS0Gtq8BrufnD75dEvGRg4kFFh7s82CyE6mc0K8V+q2d7m2G3wzX2w5jHH\nvtdnw/Gf4PkhsO1VVT2iJA3Q4Lw/QWgcHF0D656EF8aoR1AzxJm74cQ2FUg3NPXXED1ZtW5O2ww1\npTD8ojZ/y0II0dNJYHyGThbbdrYSviV23U58QTxzY+b2yHI5QvQo1SVQWdB434434Itb4YvbGu+3\nVINVlT4iez/sXqaeX/O+eqwqgA9+AVWFKmCuzFP7r14Gs+6Haz9Q21tfVo973lMzx1/dBUvPhXcu\nAL9w1bTjgUNw6b/h/Gdg4CxI3war/wA+IRB3Ttv+HgghRC8ggfEZMickgIcHXnHu5QnnV+VjsVvo\nH9C/nUYmhGgzX9wKzw2Gr+4Gc5maCd78ojp2ZLWaqbXbYfN/YMlIeCYS3pwH8f9V53j4wpDz4JFU\n8AwANJjzsOP+j2XBaNVggMgG3yBNvhXMpeqe+z9Ws8mzfwe3rIKgaJVmMeU2MHnCoLpqG/mH4aJ/\ngmfvKnsmhBBtQXKMz1Dlzp34jB3rtPNMSzIqVGvF6ADXum0JITpB1l7IiYdjG9T2vg+hphxCBkJF\nDly5FH78J6z6HQw9H3a+1eDaPepX+HC4d0fdTl81w+vpC8MvhvBhqvlGwyBW0+CSJSr3+MJ/qNzk\n45vUdcPOb36sg8+FX62D4NjTL9wTQgjhlATGZ8BWUala3N55h9vXZlZkAhDj3zXqHAohGihOUwHw\nm+c49l3/uUqN2PiM2g7oByMuVQvpPrhKBcWaEWLOgus+VOXZDn0J465rfO9xVzd4fo3z1z/rV47n\nV7zu2pg1Dfq3XVtfIYTojSQwPgOWjHSw2fAe7l5+MUBaWRoaGv38+7XDyIQQTpXnqtlUmwUMpqbV\nHkC1UX5xIng1qKF80XMqFWLgLEdgfPdWNfM7cBZc/yksXwjXLIeRdR2y/MKh75j2f09CCCHajATG\nZ8CarxbjmCLc73oXXxDPkJAheBrdS8EQQrRS0mr4+FoYew1k7oJBc+Cy/zQ97+cX1GNNKaDBY5mO\nVAdPX5WuUJ4DPsGOa+LmwqPp4B3Y7m9DCCFE+5HA+AxYC+sCYzfbQdt1OwfyD3DRICmnJES7s9aq\n6g87l6rtg5+px6IUmPFbCG/QRvrwN6oKxEmB/ZouYmsuXUGCYiGE6PakKsUZsBUWAmAMcy8wTi1N\npcJSwbiIce0xLCG6h/wjUFNx5veprVIzuKCqRZxq8wuOoNjZsVemwWc3Qe4h+HSx2j/rAYgYoR6F\nEEL0GjJjfAasBYVo3t4Y/NzrJ55QlAAgjT1E72UxwytnQVB/uO8AGM7gM/pXv4GElWDyAbsFLn8F\nxtcteEvbAhufhTFXweSbVdm0t+bDeU/D/k9hb11t4fxEdQ+AiTfC/KfUOUIIIXoVCYzPgLWgAFNY\nmNsNOhILE/EyehEX5F7tYyF6hNJMFbAClKbDjjfVQrWxV7l/r+T1KqCNnanq/ebGw3ePwNirVVWJ\ng5+rYHjhSyo/GODhFPANhcBo2PKSCph/el61agY1Dmm6I4QQvZIExmfAVliIMTzM7esOFx1maPBQ\nTAb57Re90FvzHUGohx+sfkQ9t9tUw4pd78Ci11TTipaUZsAHV6rn5/wBBs2GA5/Bl3eoYHv1o+rY\n8EscQTGAX92/2XHXOMqlnXU7/PB3OLQC+k1qm/cphBCi25HI7AzYysowhoa4dY2u6xwuOsyFAy9s\np1EJ0YUVJDuC4vOehgFnq8VuCSvh+6dAM0BZJvSfBtPubPleO+ryhq9ZroJicHR/OxkUA4y8zLWx\nnfOo+iWEEKLXcimxT9O0CzVNS9I0LVnTtCY/OTRNe1DTtARN0w5omrZe07TYth9q12MrL8MY4N5K\n9NSyVMpryxkR6n7tYyG6tcw98PJk9fzqZWph24BpsOhV+OXHagFdmWp8w5HvTn+/5O9h4GwYdblj\nX0Af6HvKotaGx4UQQogWnDYw1jTNCLwCXASMAn6padqpq8b2AlN0XR8HfAH8s60H2hXZy8oxBAa4\ndc3iVWrV++iw0e0xJCG6rh//4Xh+arrCoDlw12YYdhFEjlIVIpzRdUhcBek7VT7xkPOanrPwJbXY\n7ndH4f74xmkUQgghRAtcmTGeCiTrup6i63ot8AnQaApG1/WNuq5X1W1uA3p8n2Nd17FVVGD0dz0w\nrrJUUVZbxqTISVKRQvQO5jKoyFcL4Y5tgKl3wh0bIMTJl0p9RsP1n8CEG6AiV7VlriyErL0qIAa1\nWO6TX8I756ttZ4Fxvwlw1dvgHwnB/dvvvQkhhOhxXMkxjgbSG2xnANNaOP9XgAvfg3ZvutkMFotb\nM8a5VbkAXDXsKrcrWQjRLb01HwqOwEX/BFut6joXPbnla/pNUI+vzQSTF1QVqBJql7+sgmtQuch+\nkSqYFkIIIdqIK4GxswhOd3qipi0GpgBzmzl+J3AnwIABA1wcYtdkKy8HwBjgfmDc169vu4xJiC6n\n4Ih6/O73qqRa/7NOf03sTDjnMfjhWagth7hzVL3hybdC9j6YdBOMuBR0u5RVE0II0aZcCYwzgIbf\nR8YAWaeepGnaecDjwFxd12uc3UjX9TeBNwGmTJniNLjuLux1gbHBjcA4p1J15+rrK4Gx6AXs9sbb\nV77p2nWaBrMfhOLjMHKhqjjxnwnw1rnqeL+JMOyCth2rEEIIgWuB8U5gqKZpg4BM4Drg+oYnaJo2\nEXgDuFDX9bw2H2UXZCsrA8AY6HpViswKteI+0i+yXcYkRJdhqYaMnY7tG79SNYpdZfSAK153bF+9\nDNb/GSJHwrjr2myYQgghREOnDYx1XbdqmnYvsAYwAu/oun5I07Q/A7t0Xf8aeA7wBz6vy509oev6\nwnYcd6ezV1QAYPD3d+n8/Kp83jn4DuPCx+Fl9GrPoQnRuapL4M25UJyqtm/fADGnySs+nUGz4fZ1\nZzw0IYQQoiUuNfjQdX0VsOqUfU82eO5kaXjPZisuBlyfMU4qTqLWXssDkx9oz2EJ0fES/wefXA8P\nHoaSExD/pQqKp/1G1Rk+06BYCCGE6CDS+a6VquPj0by98XRxEWFGeQYAAwK796JDIeql74DD38CW\nF9X2gc9U9zqAuHlw0T+av1YIIYTogiQwbqXqPXvxGTsWzcPDpfPTy9PxNnoT4RPRziMTogPoOnx+\ni6NTHTiCYr9I1WRDCCGE6GZcagktGtN1nZojR/Ae7XoN1fTydGICYqR+segZcuNVULzwZZVDPL8u\nsypqPDyUJI01hBBCdEsyY9wKutmMXluLMSzU5WtSy1IZFDioHUclRAO6Dsnfw6C5YPJs+/sfWaMe\nh54PAX0gepKqNxw5GgzyeVsIIUT3JD/BWsFRqi3IpfNrbDWklaUxNGRoew5LCIe9H8CHV8HuZWd+\nL12H7P2N9x1Zo+oJB/RR25qmOtp5eJ/56wkhhBCdRALjVrCVlgJgdLEd9LGSY9h1O8NChrXnsIRw\n2P+JeixKaXrMYobcQ1Bb1fz1xzbCyntg7R8h/r/wxhw4+IU6Vl2iahQPPb/txy2EEEJ0IkmlaAV7\n3YyxwcVSbcklyQAMCRnSbmMSopGTrZjzDjU9tnOpCngBfr0JosY1Pm6tgfcXObZD6lKAtr6sWjGf\n2AboqhSbEEII0YPIjHEr2MpUO2hXUynSy9MxaAb6+8uCJNEBaiuhsq4B5fGf1OyxucxxPC/R8fz4\nj02vz4lXjx6+qsJE8XG1nbUXls6DlB/A6AkxU9pl+EIIIURnkRnjVrCV1aVSBLk2Y5xRnkFf3754\nGF0r7SaEW3Qdtr0K/SZB7AxHx7mLnoMdb8CKX6vtxV/CkPlQkgb9p0F5NmTsanq/k62c792lAuC9\n78PgcyFtM6x5DPIS1Gyxh0+HvD0hhBCio8iMcSvUp1IEuJZjnFGeQUxATHsOSfRm5TkqYH33QijL\ngqK6Gd6YKXDpCxAYrbZPbFOPxWkQHAsxUyH1Z7BZHPey1qpgOmIEBEWDfwTMfhD6TYDxv3ScN+zC\njnlvQgghRAeSwLgVbKUnq1K4OGNcIYGxaEcnZ4gB9n4IGTvAYILwoTBoNjyYAKFxUJAEyeuh9ASE\nxMK4a6CqAJK+c1yfuUst2Jv7SNPX8Q2FkQtV9Ylx17b72xJCCCE6mqRStIKtrAyDvz+a0Xjac5OK\nkiioLpCKFKL9lKSpx6D+arbX6KlSHbwafKMRPhwSVkLC12p70ByInQm+YZD4LXgHQc4BMNSl+wyY\n4fy1rn2//d6HEEII0ckkMG4FW3ExxpAQl85deWwlXkYvLo27tJ1HJXqt4rrA+PJXVDUJr0A4+97G\n50SNhyPfATpcskQFxgBDzoMDn6pfJ3n4QUDfDhm6EEII0ZVIYNwKtqIijKGuBcaZ5Zn0D+hPkJdr\nFSyEaKKyEPzCwG5XjTRObStelAIBURA3Fx49AZ7+Tc+ZeR9EjoTUTTD2asf+iYshfYej8gTAiIub\nXi+EEEL0AhIYt4K1uBiPPn1cOrfIXESYd1g7j0j0WMnfwwe/gOs/gx//CV7+cNNKx3FrLRxdo6pG\nQOP0iYY8fWH0IvWroUFz4L59agGe0UMF30IIIUQvJYvvWkHNGIe6dG6huZBQH9fOFb3Axmdh/V8a\nV4Jozo6lKigG2PeRWhiX8oPj2upi+P5p9diwYkRrnCwlaDCoX0IIIUQvJDPGbtJ1HVtxMSYXUykK\nqwtlxlgolQXw4z/U84gRMO7q5s9N+k6VYPMNV5UjEr5yHMs+ADGTYd2TsGe52hc3r/3GLYQQQvQS\nMjXkJntlFXptrUuL76qt1VRZqwjzkcBYANn7Hc83PgN5h1X7ZWutY39eoupc9/MLEDwAfrsLJt0M\nmlEtmkOD5HVQeAyO/aCuueg5MMpnXCGEEOJMyU9TN9mKiwAwhpw+PaLIrM6VGWMBqHJoAOf+ETY8\nA69OV9tDL4AbPoPSTHh1mqpBbLfCrAfBJwTmPwlTblNNNuL/Cz/8Tf0CuORfcNbtnfN+hBBCiB5G\nZozdZKvreudKO+j8qnwAmTEWSvoOCBkIM+6F2FmO/Skb1eOx9erRblWPQ89Xj37hKigGGHGJ47qZ\n98OkW9pzxEIIIUSvIoGxm3SzGQCDj89pzz1achSAuKC4dh2T6OKK01R6xPGfVC6whw/c8q2aKQ6N\nA1stVBWpBXb+ddVOAvrBgOlN7zVwtuP5gj9JCoUQQgjRhuSnqpvs1Sow1rxdCIyLj+Ln4Uc//37t\nPSzR2Tb+DWrK4MK/Nd5fkg7/GQf9p0NtBQyuWySnaSp94vA38Oli2PQvOLEVFr4MQ+ar4NlZLeE+\nY9Tj6Cva9/0IIYQQvZAExm7SzdUAGHy8T3vukeIjDAkegkGTifkexWaFN+aoqhKzHlC1f3cuhapC\nmHgjhA5SgS3ArnfUY/o29Rg1vvG9Qgerx60vq5bOE25ouVyawaCaeJhO/8FMCCGEEO6RwNhNdnMN\nAJp3y4GxruscKT7ChQMv7IhhiY5gt0NZBhQdh7xD8P0hqClXs70nfboYio7B7Rug71g48FnjewQN\naLwdMRxipkLGDjj7/1yrIewtXRSFEEKI9iCBsZscM8Ytz9jlVuVSXlvOsJBhHTEs0RG++72aGW7o\nZFDs4QtjroS9H6jtt851nDPjXjUjDE0DX4MRFv8XCo5AzJT2GbcQQgghXCKBsZvqc4y9vFo870jx\nEQCGhgxt9zGJDmAubRwUT7wR9r6vnp//Vxh3jao1HL8CLJWO84YsgPlPqfSKgCjn9/YOlKBYCCGE\n6AIkMHaT3cUZYwmMe5jUn9Xj9Z+rBXRGD5Uqodth+l2O8+7cCH4RqgpFQJSaEdY0qTUshBBCdAMS\nGLtJd2PGuK9fXwI9T1/vWHRxBcmqsYbJB+LmqqAYYNqvm54bMVw9+p6+AYwQQgghuhYJjN1kN5vR\nfHzQnJXSOnmObudo8VHJL+4JdB1enqyeD78YTC1/IBJCCCFE9yV1xNykm6sxtFCR4ljJMcYvH09y\nSTJDgyWNotsrz3E8H3V5541DCCGEEO1OAmM32avNaC3UMD6Qf6D++SVxlzR7nugmcur+PGfeB2Ov\n6dyxCCGEEKJdSSqFm+zmagwtdL0rNBcC8OysZ2XhXXdWngtr/qByi9FgzsOu1RgWQgghRLclP+nd\npFebW0ylyKnMIdAzkMsGX9aBoxJt7uBndUExMOMe8Aro3PEIIYQQot3JjLGbTi6+a05uZS5Rfs3U\nqxVdX+rPYLfBsQ0QMQKu+whCBnb2qIQQQgjRASQwdpO9ogJjUPMtebMrs+nr17cDRyScSvkBDCYY\nOMv1ayoLYVmDvPDp90DY4DYfmhBCCCG6JgmM3WAtKMCckEDYHXc0e05OVQ4TIid04KhEE8WpsLyu\ngoRnAAw5F65c2nKptdJM+PCqxvtGX9FuQxRCCCFE1yOBsRsqfvwR7HYCL7nY6fFqazWlNaUyY9zZ\nEv+nHn3DIGIkJKyEUYtgzJXNX7P2j1CcBjethOoSOPy1tGkWQgghehkJjN1gycwEgwGvuDinx3Mq\nVc3bPr59OnJYoqGk1bDmMdWW+eFklS+8ZCSsewrKs9VCulOZyyBpFUy8EeLOUftGL+rIUQshhBCi\nC5CqFG6w5OZiCgtDMzn/PHEyMJYZ40708bXq0aNugaTBCFNug9ITKmBu2LDjpLQtYDVLAw8hhBCi\nl5PA2A3W3DxMfZzPBtt1O49uehSQwLhDHVkD2fvV81W/d+wPjHY8n3k/xNYtwkvf0fQeGTvUQr3o\nye03TiGEEEJ0eRIYu8Gam9tsYHyi7ARF5iIA+vpKYNwhaivho2vgjTmQnwQ73lD75z8FV73jOM/D\nG278EoyecPzHxvc48BnsfBv6jAFP344buxBCCCG6HAmM3WDJy8OjmcA4qzILgFfmv4KH0aMjh9V7\nHV3reB7/JaDBw8dg9oMQ2K/xuSYvGPML2P2eqloBkL4TvrwDzCUwcXFHjVoIIYQQXZQExi6yV1dj\nLy1tdsY4uyIbgMHBUve2w6RtdTzf9xH0GQ1+4c2fP+8xsFvg4OdgLoVdb6v95/1J5SELIYQQoleT\nqhQusubmAmDqE9nk2OHCwzy99WkAIn2bHhdtyGaFpfOgqlBVmQiIUo+lJ2DoaYLb4AHQdyxseAY2\n/RssVTDmKph1f8eMXQghhBBdmgTGLrLk5gE4TaXYkeNY0OVhkDSKdnXoS8g54NiOPVvVK078H0y+\n5fTXX/RPVZpty0tqe/C8dhmmEEIIIbofCYxdZM07OWPcNDC22q0A3DZGvo5vd6mbwCcUbv1OLbob\ndTmMXAgXP+fa9bFnq182i0qnGN1C0w8hhBBC9CoSGLvIkqPq35oimwbGReYifEw+PDD5gY4eVu9T\nkAzhwyByBDyeA4ZWpslf9I+2HZcQQgghuj1ZfOcia24eBn9/jP5+TY4VmgsJ9Q7thFH1QoXJED5E\nPW9tUCyEEEII4YREFi6ypKfjER3t9FhRdRFh3mEdPKIeym6DgqPOj2Xugco8CBvasWMSQgghRK8g\ngbGLalKP4zlokNNjMmPcBiry4L+3wyvT4OUpsP4vYK1pfM7XvwXvIBhxaeeMUQghhBA9muQYu0Cv\nrcWSkUngxRc3OVZlqSK7Ipux4WM7YWTdnLkMvr5XlVGz21V94ZM2PQ+hcVCaDv2nQUBfyI2HC//h\nSKUQQgghhGhDEhi7oDY9HWw2vJzMGH+U+BHllnIWDVnUCSPr5g5+Bgkr6zY0FQCnb4ez/w+2vgwr\n71aH+oyF4ReCZoDRV3TacIUQQgjRs0lgfBq61Ur5uu8B8Bo+osnxjSc2Mi58HBMiJ3T00Lq/E9sb\nbOhw9m8hZjn4hsOBT6FClcgj/7Bq2zxwNgQ47zwohBBCCHGmJMe4Bbquc+KOO8h/4QUAvIY2/gq/\ntKaUgwUHmRU9qzOG1z1VFsDPL0DeYUj6DkZeBoPng3cwxJ2jUiaMJoio+xAy52GwW6E8B2Y/1Jkj\nF0IIIUQPJzPGLbBkZlG1dRsAvtOmoZ1SHiytLA0dndHhoztjeN3TZzdB2mbY8qIKgOc8DFHjobYK\nPH0d513xhmr1HByrAuhzHoW4uZ03biGEEEL0eBIYt6A2NRWA6CX/wm/27CbHsyuzAejjK1/vu6Sm\nQgXFAFWFcMkSFRRD46AYIDBK/QK4a3PHjVEIIYQQvZYExi04GRj7TJmCMSCgyfGcStUNL8o/qiOH\n1b3YrPDBFTDtLkB37A+Ng0k3ddqwhBBCCCFOJYFxC2pTUzH4+mKKiHB6PKcyB1+TLwEeTYPmXs1m\nVWkSAJm74fhP6pdmAK8gNQNsMIHRo3PHKYQQQgjRgCy+a0HNsWQ84+LQNM3p8ezKbKL8opo93itl\n7IJno2DHUtB1OPSl45huh0uXQHB/R5qEEEIIIUQXITPGzdB1nZrDiQQsOK/Zc9LK0ogJiOnAUXUD\n8f8FWy2seQyy98HeDxzHblsDA6Z33tiEEEIIIVoggXEzrLm52EpK8Bo50unxGlsNx0uPM6//vA4e\nWTdj+mYAACAASURBVBdkt0PKBtj1LiR+q/bZalVQPPhcuOpdOLFNgmIhhBBCdGmSStGM8u/XA+Az\nZozT4yklKdh0G8NDh3fksLqmfR/AB79wBMULX1Y5xACjFoFPsOpcJ4QQQgjRhcmMcTMK33kb3ylT\n8B471unxlNIUAIYED3F6vFcoToO3F6imHaA600VPhgk3qM51qZsgclTnjlEIIYQQwkUSGDthyc3F\nmpVN2C23NruwrshcBEC4T3hHDq1r2fqKo23zWbfDJf9yHBvzC8g+AJHOU1GEEEIIIboaCYydMB88\nCIDPOOezxQDF5mKMmpEAz15cqi1pFYQOhsm3wJRbGx+bfAuMu7Zp4w4hhBBCiC5KcoydqD5wEEym\nZhfeAZTUlBDkFYRB66W/hWVZUJoOU++Amf8HXqd8QNA0CYqFEEII0a3IjLET1QcP4D18OAYvr2bP\nKakpIdgruANH1YUUHIWfnlfPY6Z27liEEEIIIdqIBMan0O12zAfjCbzs0hbPKzYX997A+PVZYDWr\n51HjO3csQgghhBBtpJfmATSvJjkZe0UFPmPHtXheSU0JId4hHTSqLuTEdkdQHDrY0fpZCCGEEKKb\nk6jmFFXbdwDgN63lFIFiczHjI3rZbKnFDCt+DcGxMO3XMPT8zh6REEIIIUSbkcD4FFW7duERHY1H\ndHSz59jsNkpqSgj1Du3AkXUBBz6F4uNw4wrV0U4IIYQQogeRVIpTWNLT8RwyuMVzimuKsek2Inwj\nOmhUXYDNCikbISAK4qQNthBCCCF6HgmMT2HJycGjT98Wz8mvygcg0ieyI4bUNax9HA6tgKgJqhSb\nEEIIIUQP06sC4z25e5j0/iR2ZO9wetxeU4OtqAhT3z4t3ie/WgXGvWbGWNdhx5vq+amNPIQQQggh\neoheFRgbNAMWuwWr3er0uDUvD6DFGWOLzcKWrC0ARPr2khnjwmOg2+GSJTDsgs4ejRBCCCFEu+hV\ni++MmhEAq95MYJyTA4BHVPOB8QM/PMCPGT8CEOYd1sYj7KKOq/dL3DmdOQohhBBCiHbVq2aMTQb1\nOaC5GWNLTq46r6/zwLjYXMxPGT8BMCpsFB5Gj3YYZRd0/CcIjIbQuM4eiRBCCCFEu+ldM8YGNWNs\n021Oj1tysgEwRTrPMd6atRUdnY8u/ogx4WPaZ5BdTXkOHF0LY6+SRXdCCCGE6NFcmjHWNO1CTdOS\nNE1L1jTtUSfH52iatkfTNKumaVe1/TDbhklrecbYmpOLISAAo7+f0+OJxYl4GDwYGTYSracHidvf\nhLV/hKXzwVINsx7o7BEJIYQQQrSr084Ya5pmBF4BFgAZwE5N077WdT2hwWkngFuA37XHINtKS6kU\nRcvfp/jDD/GMaz5d4FjJMQYGDay/T4+k61BwBL572LEvbKikUQghhBCix3MlwpsKJOu6ngKgadon\nwOVAfWCs63pq3TF7O4yxzZxMpXAWGBe8+qo6lpvb7PXHSo4xNnxs+wyuq/jiNjj0ZeN9ns5n0IUQ\nQgghehJXUimigfQG2xl1+7qdk6kUznKMfSZOBMAYEuL02lpbLZkVmcQF9eCZ0w1/dQTFwQPUY/gw\nuPTfnTcmIYQQQogO4sqMsbNkWr01L6Zp2p3AnQADBgxozS3OSP3iO7uTxXcG9RlhwNtvOb02r0rV\nOO7r13JXvG6nphw8fCFrH/z0T7Xv4udh0s1gt4Knb+eOTwghhBCig7gSGGcA/RtsxwBZrXkxXdff\nBN4EmDJlSquC6zNRv/jOSR1je0UFPpMm4Rkb6/Tak4Fxj2rqYbPC32JUEKzbweQDt62GfhPqTvDs\n1OEJIYQQQnQkVwLjncBQTdMGAZnAdcD17TqqdtLS4jt7ZSXGUOdpFAB51Sow7lFtoEtPqMc976nH\noRc0CIqFEEIIIXqX0+YY67puBe4F1gCHgc90XT+kadqfNU1bCKBp2lmapmUAVwNvaJp2qD0H3Vot\nLb6zV1Rg9Pdv9trsClXjuI+v8xrH3VJhSuPtiOGdMw4hhBBCiC7ApTrGuq6v0nV9mK7rg3Vd/2vd\nvid1Xf+67vlOXddjdF3303U9TNf10e056NY6OWPsbPGdrbICg5/zwDi1NJUlu5cAEOgZ6NZr2uw6\nv/t8P5uTC9wcbRvY/CL8JQLKHJkv+9JLeHH9UfLKzFB0TO00eaOP+QVMvwuAD7enseiVzVhsXbrI\niBBCCCFEm+rBBXmbqq9K4WTxnb2yCoOf87JkR4qPABDpE+l2Y491CTl8sTuDnalF/PjwPDdH7LA7\nrYj+Ib6ggY+HkQDvlttRW212TOueUBtLRsKISykOGcu1m8ZRY9XJS/iJvxjeQfP054aIL7AWwqeB\n/SittvD4ingAfkzKx2CA1344xvLbpuHjaWz1+IUQQgghurpeFRhrmoZRM2KxWxrt12029KoqDM2k\nUmRXqjSKFYtWuPV6S9Ym8eKGZACqam3ouu5WYJ1eVMXSTSlcMLovN7y1nVFRgSRkl+HraWT/U+fj\nYWx+wv+3H+/lWd2fEK1C7Uj8lhC+5Qnms9B/H4GFhWr/1F+z+aeiujFaScopr7/HNwey2JdeQlph\nFZ/uPMEtMwe59f6FEEIIIboTl1IpehKjZmySSmGvrATA0Ewr6JzKHPw8/AjwCHD5dRJzyuqDYqNB\nI7+8hpc2JLMuofkGIoeySvl054n6FIb7P93H8q1p3PDWdgASsssAFWSnFlQ2uf7D7Wms3JfJzL9v\n4Pv4DPypBmDvsPvqz1lsWk+gtZBa3cjWIQ9gnvdU/bHNyYUcy1eB9FkDQ1i5L4u0wioA1hxqftxC\nCCGEED1Br5oxBrUA79TFd/WBcTOpFNmV2UT5Rbk127sxMR+A6GAfXl88mfs+3cuSdSol44VrJ7Bo\nYjQ3v7OD8TFBXD2lPxuT8vh2fzY7UotYHZ/DwcxSCipq6+8X7u9FQUVN/XZSbjnFVRaWbTnOC9dO\npNZmr0+BAJikpeCh2fiz98O8c2AiZxse4yPPZwHQr36Py1ZqJMWbmFS2r/6alzYc5UBGKQDXTxvA\nztRihvcJIDzAk30nSrDZdYwG91JJhBBCCCG6i14XGJsMpiYzxtUHDwLg2b+/s0vIrMikj5971Shy\nSqsJ8Dax+dFzAVh5z0yO5Vey6JXN3P/pPjxNBn48ks+PR/J5f1saxVWO9I6NSSqoXjShHzllZkqq\nLCy9aQovrj+Kn5eJ5VtTeX5NEql1s7mZxVvYXxfQApiwcpfpG6yY+KJkGAB77EPrj2ujLucfgaW8\ntyWVFXszATWrfaDBPRZNiCY62JfJsSGs3JfJ5uRCjuSWMzLKvcWHQgghhBDdRe8LjDVTkxnj8tWr\nMYaF4TtlSpPzS8wlHCk+wu0xt7v1OtmlZqKCvOu3A7w9mNA/mPdum8rN7+zg7g/31B9rGBSH+nlS\nVFnL8tumMmdYRKO85OeuHg/AlmMFHMmtqL+mYVAMOks8XmOBcTd5cVdQlqBmwc14OU7RNCb0D2b0\nVeMoqKghJb+SX0yK5sUNyUwbFMrvLxyOpmlMHRQKwLS4MAB+PloggbEQQggheqzeFxgbmgbGNcnH\n8Bk/Hs3U9Ldja/ZW7LqdOTFz3Hqd3DIzfYN8muyfOyyCz349g2ve2Npo/4MLhuFlMnDdWQPYdryQ\n2UPDAZymb4zuF8SR3AruPmcw+eU1fL47gxunx3Lx2CjeePs1Fhq3wtxHCJ/zCDy+GoC/XzmWqrDV\n+IY4Zr49jAbe/9W0+u1bZg4iyMejSbpEdLAPw/r4szEpjzvmxLn1+yCEEEII0V30usDYaGi6+M6a\nl4fPpIlOz08pTUFDY1TYKLdeJ7vUzIi+zmdXpw4K5b93zeD7w3lcPqEfpVWW+llZgAtG923x3nOH\nRbBibyZTB4XiaTLw+e4MLhzTl+kDAxkTtgK7x2AMs3+HwWjk50fmsTutmIXj+6FpA1q8b6hf8y2g\nZ8SF8cXuDHRdp7TagofRgJ+X+uuTXlRFqJ9n/bYQQgghRHfU6yIZo9Z48Z29thZbSQmmyEin5+dX\n5RPiHYKHoeW6wQ2ZLTbyK2ro0yCV4lSTY0OZHBvq+sAbuHxCP8ZEBzEkUpWX2/i7cxgU7ge7lxFQ\nngzXvA8mFeTGhPgSE+LbqtdpaGifACprbWxMyuO2ZbuICfHh+wfn8vsvDvD1/iyunBjNkmulnbQQ\nQgghuq9eV67Nw+DRqMGHLV8tdDNFRDg9v6C6gAgf58ea80NSHroOk2NDWj/QFmiaVh8UAyooBtj+\nJkRPhpGXtflrDuujStXdtmwXABnF1Yx4YjVf71dd9Tad0tnvw+1pTP3r93y0/USj/Xa7zt4TxWxJ\nLuDf646g6zoVNVa2HCvAbtfbfNxCCCGEEK7qnTPGumPG2JKXB4BHczPG1fmE+4a79RrfHMgm3N+L\nmYPDTn9yW6mthPzDMOf34GZ3PlcM6+MIxMdEBxKfqWoqj48JYlifAD7fnYHZYsPbw0hplYU/fZNA\nrdXOEyvjOWtgCEPrAutPd6Xzhy8PNrp3XnkNH+84wT3zBtM3yIf5IyLpF9w0P1sIIYQQoj31usD4\n1MV31tPNGFcVMCxkmMv3t9t1NicXMH9EH0wtdKZrc9kHQLdDP+e50mcq2NeTt2+ewrA+AfQN8ub3\nXxzg7MFhXDIuiq3HCvl8dwZTnvme6GAfRkQFUGu188K1E7j/031sOVZYHxhvOaY67l07pT9H8sr5\nz/qj9a/x3pY0Kmqs5MwbzMMXjGiX9yGEEEII0ZxeFxifuvjOkqVSAUxRUfX77LqdCksFK46uIK86\nz61UivisUkqqLPVVJTpMTt0sbNT4dnuJ+SMdFS3+3SCf+NwRkbx6wyTu/nAPSbnlJOWWYzRoXDim\nL+H/86qvj6zrOjuOF3LZ+H7846px6LrOB9vS2HuiBKtdr0/LaNiWWgghhBCio/S6wPjUGWNLRiYG\nPz+MwcH1+z5O/Ji/7/g7ACFeIUzp07S+cXO+i8/BZNCYM8y9vOQzVpgMnv4Q0HJFi/agaRoXj41i\n7QNzeHH9UbYeK+SG6bF4exgZFxPEjtRC8srNZBRXk1tWwzl1vzeapnHjjIHcOAO+2pvJ1/uzMBk0\nknIlMBZCCCFEx+t9gbFmarT4zpKejkdMTKN6wbtzdwNw9bCreXLGky7f22bX+XpfFjOHhLdY+qxd\nFCZDaFy75Be7alifAF6+flKjfQvH9+P+T/cx9a/rGRkViKfJwILRTbsInjsykltnDsTDaODNn1JI\nyilneN+AFl/ParPz4fYThPl7MrxPQH26hhBCCCFEa/S6qhRGg2PxnbWggIoff8QjOhqAg/kHGfve\nWBIKExgdNtqtoBhgY2IemSXVXHuW89bS7aroGIQN7vjXPY2F4/txxUT1+3s4u4z/O3cIgd5NS98F\nenvw1GWj+dWsQYT7e3LpS5u48e3trEvI5cHP9nGirv11Q29uSuGprw9x70d7WfDvn9B1qWohhBBC\niNbrlTPGVZZKKjZvpujdZQB4DhwIwIb0DQBkVmQyMnSkW/f926rDvPFTCtHBPiwY1XRGtE3UVIBu\nA++gxvutNVByAsZc1T6vewYMBo1/XzuBswaGYrHZuWlGbIvn9wn0ZsXdM7lj+S42HS1g01FVBk7X\nG+c12+w6H+9oXAruaF4FA8P8qLbY8PEw8snOE1w6rl/Hz94LIYQQolvqdYGx0WBk1O4C0j++vX5f\n2B3quZ+HX/2+vn7u5ep+sTsDgGcWjcGjtdUodiyF/CS45HnHvrzDoBkgYji8Mg3KMuDp0sbX5Sep\nihR93OvO15Gun9Zy172G+of68vlvZrA6Poev9mXi62ni2wNZ/PWKMfh6migzW7jnwz2kF1Xz6g2T\n0IC7PtzDtpRC/rginh2pRSwc34+v92fx89EC3rhxMh9uP8HFY6MkSBZCCCFEs3pdKoXJYMK7orZ+\nO+DCCzGFqEYchdWF9fsbBsmnU2O1UVJt4d55Q5g3wnk95BaZS2HXu7Dqd7BzKRxdB5/cAEXH4dXp\n8MpUsNaqoBhg7RNqChXAZoGfl6jnfca4/9pdVIC3B1dP6c+Ht09n8fRYLDadnanFAKw9lMumowVM\niQ3hojF9uWhsFNHBPjy58hA7UosA6itcrE3I5e2fj/PHr+J5YmV8k9eprrVJCoYQQgghgF44Y2zS\nTBgsjsV3plDVlrmitoLdubuJ9o/mlyN+ycWDLnb5nscLKrHZdYY2aILhlo+ugxNbHNsf1qVE5CU4\n9j3ToMrFlhdh/HXw5Z2Q2yDYC+16OcZtYerAUDyMGluPFTJ3WASHs8vwMhn45M7p9Ysm+wR6kVlS\n3ei6+D9dwLznf+CZ/x0G4Hh+ZaPju9OKuPmdnUQFeXPrzEFcPSWm9bP9QgghhOj2el0UYDQYMTYI\njHWrWoj3h01/4HDRYQI8A7h59M1E+Lpebu1k3d3TVVFwqiyrcVA8YbHjeVFK89e9drYjKO43Ec79\nIxh75uccH08jw/sGEJ+pUkgOZ5cxvG9AowYq9583jBF9A/j4jun1+/y9TPzxEkeueEJ2GRe+8BPL\nt6YC8ML3R6mosXI0r4LHVhzkm7pZ5obSi6qkVbUQQgjRS/TMSKoZ1uJiBu3Px6/EkUrhN3sWAHvz\n9wKQVpbm9n2P5lZgNGgMCnc9/aJe6s/qcf5TMHQBGL3UdtYeNWPs4Qt2GwRGwawHIXoSvK7GTPQU\nuPTfEDXO/dftZsb0C+KTnemMeWoNFTXWJjnLc4ZFMGdYBFW16oNOoLf6q335hGjGRAdRWm3hv7sz\nWB2fw7OrDrNgVB82HS3glrMHMjk2hN9+vJdDWWVc2aDaXH55DbP/uZHzRkby1s1n8dXeTD7cnsYr\n108iMtC7w967EEIIITpGrwqMa48d49xXd3AiQkPz92Pohg0YAwMBGBAwgIM1BwnzDnPrnrquczi7\njIFhvniZjO4PKnWTqjIx8z4w1F2/6BXVye71WTBwNly9DDx8HDWKr3oHggZAzJROrVvckQaE+QJQ\nUWMlzM+Te+YNcXqer6eJv14xhmmDQuv3DY5QKS6TBoRw3sg+3LpsJzP+piqQzB0WwbwRkbz183H2\np5dgttjw9lB/DvFZaob6+8N5HMgo4f5P9wHw99WJLLlmAkIIIYToWXpVYKx5qdlYX7OO1WSoD4oB\nKi2V+Jh8WHr+UrfuuWTdEdYn5nHh6FZ0nCvNgCNrIXamIyg+qe9YuPEriBwFnr6Nj435hfuv1c3d\nMDUWD4OBm88eCICnqfksoBumNV8Sbnqc44PPvOERzK3rwjdpQDDvbk5l3J/WsuGhucSE+HI4u6z+\n3JNB8YBQX/all5zJWxFCCCFEF9UrA2O/GrAENM4bLagu4PLBlxMTEOPy/b7en8UrG5PxMhm4ffYg\n9wf079HqMXam8+OD57l/zx4qyNeDO+bEnfF9fDwdH0B+OXUABoOacX/o/OHoOizbkspVr23F28NA\namEV0cE+eHsYOJZfyYBQXxZNjOblDUeprrU1uldDZouNW97dwYMLhjO1wcy1EEIIIbq2XrX4zlAX\nGPvUgsXkSEGosdVQVlvm1oI7m13nmW8TGBkVyI7Hz2PKQDcDIEuDCgpDz3fvWnFGJg4IBmB0tKNR\nir+XiacXjmZAqC85ZWZS6zrtTR0UyrgYdf7k2BBGRQVg1+FIbnmz999+vIhtKUU86aQ8nBBCCCG6\nrt41Y+ztWDBVW/fOD+YfxKarKhXhPuEu32tnahF55TU8edkognyatjg+rZJ09Xj5qxAxzP3rRau9\nsXgym44WEB3s0+TY8tumsjYhh2dXJQIwe2h4fUpFdLAPI/qq9JvEnDK+2J1BWlEVL103kSBfx9+B\nncdVLeUA7171z0sIIYTo9nrVT27N09H1rFqzMPa9sfXbJs3E9Kjpzi5zan9dnumsIa4H042U1FW/\nCOuZtYe7sshAb34x2XnKzMBwP+6cM5jrp8XyyY4TXDa+HzEhvizddJz5IyMZEOqLr6eRhKwy3t+m\n/gw/353O7bNVmofZYuPLPaoRy87UYg5llTK6X5DT1xJCCCFE19KrAmNDgxnjcsw0fPuXxF1CP/9+\nLt/rWH4F4f6eBPu2ssXwycA42PVWyaLj+HuZ6oPdqYNCOfLMRfUL/ob3DWD1oZz6czcm5eFpMvBD\nUj4VNVaySs0sGNWHdQm5/GN1Estvm9op76GtVNVaeefn45wzPJIx0RLkC9FRiitr8fE01lfKEUK0\nv16VY9xwxrhhjrFJM3H72NvduldyXkV9GbBWyU9SNYr9W1HNQnS4hlUwxscEk1tWA6jKFpuTC3ly\n5SE2JOax43gRt84cyKs3TGJcTBDpRSpXubTKQkWNtVPGfqb+dyCb59ce4dZlOzt7KMJF3x7IYkty\nQWcPQ5yBqlorE/+yjgc/29fZQxGiV+ldgXFxChhVQGyp+wD+yFmPsGPxDgYGDXT5Prquk5xXwZDI\nMwiMj2+CAdPB0Kv+CHqEW+pKxg3vE8CzV6p0nOhgH1b932z+smgMT1wyCg+jgbnDIjhRVEVptYVL\nX97E2X9bT1phZQt37pp2pRYDUFJVK10Au4F96SXc+9Fern9rOyn5FZ09HNEKW44V8KevEwBYdTDn\nNGeLnsZu19mcXECt1d7ZQ+mVeldUVlmAwaAW2lnqsiimRk3Fw+De4rn8ihrKzNbWB8YlJyD/sGre\nIbqdgeF+fP/gHL66ZyZRQT6sf2guax6Yw6h+gdw4Pba+BNzgCH9sdp37PtlLelE1ZWYr3x7I7uTR\nu8du19maUgiAxaaTV17T7Lk1Vlv9DLnoHHa7zt+/O1y//fLG5E4cTff27KrDvLj+aLu+RnWtjRqr\nrdG+H5LyuH7pdj7dpRZo+3oa0XX5QNqb/GNNIje8tZ2XNrTv3z/hXO8KjA1GNIP6D+ZkYOxtdL+1\nb3KemoVxKTC2WcBcqto673oXaipg0xIweMDYq91+bdE1DIkMqK9jPDjCH3+vpun6Zw8OIzrYhx+S\n8rl8Qj+G9wlgW12Q2V2sTcjhRFEVV06KBiC92Hngq+s69360l9n/3Fi/+LAr0HWd7SmFHGvDmdOC\nihpSCyrRdZ1DWaVsTi6goKL5Dwwdacm6I2xLKeKpy0Zx28xBrNibya7Uos4eVpdXeUqa0+HsMt78\nKYUl645wzetbsdt1vtqbSXxmKZkl1bzz83F+tWwnS39K4cmV8a0OXH/x2hYWvrS50fU/JOU3Oqeq\n1saKvZmN9uWVmSms+ztntdnZe6KYXalF9ftE92W363y2U30oen9bGlW1nZeCl1Fc1e1+ZrWFXrX4\nDoMRg1HHhiOVwtvkfmB8LF99He5SjvH3T8PWl+GCZ2HNY2q2eO8HMOkmCO7v9muL7iMy0JsfHz6H\nnanFjIsJ4rk1SXyy8wS1VnuLnfu6gs92pRPh78WGxDxCfD24a+5gvtyTSWpBJWedUrNb13WWbUll\nXUIuAGsO5XDlpBh0XWdbShHT40LROrh1ud2uYzBovLcllae/ScDTZOBfV4/Hw2igssbabFWS0zFb\nbMz/14+UVluYEhvCrjSVZnLeyEjeuvmstnwLrbIhMY/JsSHccvZAcsrMvL8tlate38qm38+jf6jv\n6W/QC1TUWLnp7e1cNr4ft84cxLqEXH7zwW5W3jOTMdFBrI7P4aEGeb07Uov4++pE3vwppcm91ifm\nAWpBrrOOmycKq/A0Gegb5M36w7lklZq5cbo673hBJQl1pSB/OJLPvOGRgCoFGhfhR0p+JbfOHMi7\nm1P523eJXDnJ8Xd26rPrCfQ2ceDpC3jkvwf5b92H0aggb1bcPZO+Qe7/XBMdb196CasOZnPPvCH4\neRoxGQ3EZ5VSXGXhphmxLN+axqc707l1ZisaiLWB25bt5EhuBf+5bgKXjetX/22oOzYk5rIhMY9B\n4f6E+3ty+YTodhhp2+plgbEJzdh4xtjL6OX2bY7lVeDnaSTKlf98jq5Vj2seU48/L1GPM//P7dcV\n3Y/JaGDGYNWGesbgMJZtSWVfekmX7ohnsdn5/RcH6rfnDotgYLgf0cE+vLjhKFdMjMZkNJBTamb+\nv36gslZ9FTxrSDjeHgZS6j44Lt+axlNfH+Ktm6Zw3qg+ZJVUk1+XihHs60FsmF+7vYc7399NYk4Z\nBRU1+Hoaqaq18duP99Yfv3R8FF4m91f637ZsJ6XVFoD6oBggs8R85oM+Q2aLjSO55dw5Jw5N04gK\n8uG926Zy/dLtfHMgi7vPGdLZQ+x0eWVmlqw7wp4TJcRnlfHLqQN4+utD2Ow6GxLzGBMdxEsbjhLs\n68nq+6ejaTD/Xz/y5k8peHsYMFscOZ+eRgMLJ/QjMaeMx1fEMz0ujMER/tRYbWxMzGddQi7/3ZPB\nyKhAvrtvNr96bxcAz69JYvtj81l10JFW9WOSCoyzS6s5lFXGQwuGcfvsOLw9DEQFefPsqkTyy2sI\n8fUgpUD9+yozW4nPLGXNoRxG9A1gzrAIlm1O5YmV8Sy9acppfy8+3nECi83OTTMGtu1vcit8tiud\nswaGMijc/f8THvpsP54mA3+7cuzpT+5CSqpqufr1LVhsOpuOFnA4uwyjQSPC3wtfTyP3nzeMw9ll\nvLXpOIunx+JhNJBfXkO4vye6rv7/mRwbgrEVwaordqcVcyRXfdN23yf7eO2HY1w2vh/zR0bW1/Nv\nKD6zlOfXJnHlpBgWjlcVvqprbdy2bFf9OQYNwv29mNnaMrf/3959h0dVpQ8c/56ZSe+FFBISQhJI\n6E26gBRpKqxlLai46iqurqi7PxXLWnZXd1d37RW7rqICCq4gCApIh9BCSSAJkAKpJCGN1PP7404m\nCUkgiRAI836eJ08yd24mZ3Lnzrz3nPe8p53YYWBs/Fi7wEdbeoyTsouJDHBvWS9YdWXjbd5h4NO1\n1X9XdGzDIvwwKbjh3Y38bmQET17R83w3qUmn5gn37OyJg9nEQxO786evd5GUU0xMkCc70/JtmFRF\nnwAAIABJREFUQfEzV/Vi5tAwXlxxgLUHcqmsruGHPcakoRzr8O5Vr6+3pRwMDPNm0R+aWQr9Vyou\nr2LlfqP3urOXM1/NHk5+SSVXvr7Ots+G5DxbD11LlZRXsTElj2sGhjKtbxCxwZ5UVmlunLeJPGt6\nxa70Aq7q17lF7w2bU/LwcXOke6BH655gM+IzCqmq0fSpV1JvRKQ//bt48+O+rBYHxl9uTWVTynFe\nur7/WWlXUzIKysgsPMmgcJ8G2w9mFbFiXxaDw30Y2s3vrP7NtOOlXPqvn223K6pquPK1dWQUlOHt\n6sD8LansSitg79ETPDolxtbD3iPIg93phdw7NorR3Tvx1upknrgilhBvF5RSZJ84yfB//MSCuHQe\nmRzD/V/sYPle4/UX5OnM/mMnSK+XglRYVsmSnUf5z48HGBTug5uThf9uPsLyvZkcKzQusKb1Dbal\natWWSLzl/c2k5JQ0uKi+4jXjNX3fuCiu6NsZN0cLL608wJoDOYzp3vxKrknZRcxdFA/A5T2D8HN3\nxMHcslGs4vKqJlPH2qqwtJKHF+zG3cnC+kfHUVhaSZhf06MbWmteXZXE5N5B9AjyICWnmIXb03E0\nm3h8WuxZbVdraa1Pe95XVNUwd1E8d43uRvdAd95ak0xVjcbDyWJbRMrRbCLEx4VHh8Xg6+bI7DGR\n3PHxNv65LIEgL2f+9v1+vFwcbBfnD0yI5oEJTS8QlpRdRF5xRZvPow/WH8LT2UKQlzMHsopJyCwi\nITORF5YnctPQMB6ZFNNgYat/LEtgXVIuaw/kMDjchwAPJ37Ya1z8zRoezqReQTy1ZK9xfjw4Gn/3\n1ndKtpcLezz3bDNZMFl7jKusAbKjqfV1iJOyi4lqSRpFTTUUptXdDrJe0Yac+WpeXHy8XB1495bB\n9OzsyfK97T/TPKOgjDnzd5CY2fxy1gCHrZUz3r55EL8dHMoNlxgpP/26GB/QezOMN/Hanquv7h7O\nrBFdsZhNdOvkRkV1Dde+vdE2aS+nqJyik5UN8nB3phVQfY4qXKyw/m8/uG0wax6+jFAfV/qEenHb\niK6M7WEEC23Ju9137ARaw9Q+QYyLCSTYy4UwP1dmjQgnu6icK19fx5z5O3l/3aHTPo7Wmvd+SeH6\ndzdx+Utr6f7EMh5duJuMgrLT/t7ppOaV8siC3Xg6WxhxSm/M0G6+7Mko5GRldaPf+2FPpu1DFox8\n1UcWxvPNjgyyTjTdC661bnVO7fwtqUx/Yz2Z1sDvhR8SuPm9zbbc3k0pefzxix3c/P5mXlieyPXv\nbmJjcstyG/ceLeSTjYdJtS7jnl9SwaLt6Y0qqNQ/564bFMrgcB8OZhfTrZMb94yJ5GjhSVYlZDOl\nd5Atpx7g4UkxeDhZuHpQKP26ePP2LYMI9XG1BUEBns6Mjwng882pFJdXEXekwPa7cyZEA0bQAHD/\nOOPi5OGFu6mu0dw0JIz7x0Xh5eJoC4ov7xlIt3qfL706e2ExKRIyi6iormFdUi5DI3z56u7h1pEX\nVy6NMl7XNw8Lw93JwuxP4zhxsolOGatnvttn+3nY86uY8J81LZobsPZADr2fWs7cRbu546OtzHhj\nPTfN28SO1Pwz/m5z9mca7yfF5VX0e2YFE15aw8nKasoqqvlyaypJ2cWcrKwmPr2QuCP5vLTyAHPm\n7+BIXgkPfGmkvFRU1/CTNa3lfNiTUUjPvywnIfMExeVV/GXxHtLzS3n8m3jGvvAzC+PS2ZlWwMLt\n6Ux6eS0Rc5fyzpoUZvQP4avZwxkXE8C/r+vH+kfHsfCeEcwYYLz+LusRQPdAd95bd4i/fb+fbp3c\nuKJvMIGeRlD50YbDzb6P3vL+Fq5/dxPrW1m2MSHzBBuScvlxXxZXDwzlnrHGImQf3nYJc6fEYDEp\nPt+cyru/JAOwNP4Yv3lzPeuTcxnezY8aDVNe+YU+T6/gwS93EerjwlNX9mJElD9v3TyI2WMi8XNr\n4/oP7cS+eoyVGWon31nLtrU297HoZCWZJ04S2ZKJdycyoKbKyCfOToCZX8OR9RB87npixIVtQs9A\nErOKeGF5IidOVuLp3IblxNto3toUFu88ytZDx1n3yLhm88XirCkCQyJ8mdy7rs52hL87Lg5m4jMK\nuaSrL7vTCgnwcGrQgzWpZxBvd0q2rQwJcCSvlBlvrLfd7t/Fm51pBTy6cDfPTu9t6xk7G0rKq3ji\n2z2E+7kyItK/QS/Y01f1Aow37XhrcN8S//ohgXA/V+ZbJ8ScusjJ2B4BPLc0gaKTRpD33i+HuM16\noVBfdY3mheWJpOWX8v3uY3g4WfjtJV14f90h5m9N49udGSy+dxQ9glrXg1xcXsXfl+4jJbeE124c\n0GiJ+sHhvryzxphI9tjUWNv2vUcLmf1ZHBNiA3lvlnGxXr+KxaaUvCbzAX//yTYyCk6ybE7Lqupo\nrXnU2ju5KiGLXp292HfsBGWV1SzeeZS4I/m2HFmAP4yN5M3VySzbc8yWhnQ6zyzZx5bDx/FyOcCm\nueN5ZdVBPtpwmANZxTw6Jca2X0JmEb5ujtw/LooZA0Ioq6zm6jc38MjkGMbFBJBbXM7UPsEMCGvY\niz0q2p/4Zyadtg33XhbFj/vXc89ncQ0uAGf0D2Hb4brnN61vZ95bd4jSimpuHxlhy3X/6c9j2Hb4\nOH5uTsQGNxym9nJx4Pv7L+V4SQVLdmXwxRYj53RIhC/rHxmHk8Vke635uTvxn9/2465P49idVsio\n6MZD1vklFfxyMJcHJkTj7eLA09/t40heKQ99tYth3fzo7O3Cwawigryc8XB2YMuh4zz57R4enNid\nFfuMi4svtqQ1eMw3Vye3KH2jKfuOGudinxAvTpys5EheKRuSc1kan8mCOOP/FurjQnp+3YVjQmYR\nl724mhoNM/p3Zl1SHsv3ZNqG8JuyPTWfUB8XAjzOfv71/K2plFVWsz4pD2eHfD7ZeITPN6dSVaPx\nd3fk0UW7uWNUt0a/9/DkHgR7ufDBbU3PTzCZFPNuHczsz7az/9gJPrl9CKE+rvx1em+W7DrKA1/u\nZE9GIUFezhzJK7W9F69OzLZdaK1OzG42dWHe2hS+jktj2ZzR5JWUM29tCvN+qbuwn9w7iKERvlzR\ntzMOZhOXxQQwpkcnbn5vM99sz+CesVH86atdlFkvuq8bHEphWSXZReW2C/HbR0bYPmuiAtx/XZnb\ndmJfgbHJTFWp8SGc08YFvI5YeyW6tSQXKv+w8b33NdBtrPFzzLS2/WFx0YgNNgKfxTsyWLLrKPNu\nHdz2FRRbobZU3NHCk2w+dLzJoGPv0ULeXJ1MVIA7Pq4NAyyzSTEyyp+PNhzmow2HAaPyRn1erg48\nMjmGuz+NY2LPQLKLym1BwTNX9WLWiK6k5pUy+oWf+TounfzSCubdOpgvt6axI7WAv1zZEw289OMB\nKqtruH1kBF1bkXcYdySf0opqnrmqV7OrhfUJ8WTl/mwqq2vOOHx8rLCMN1cn224/NLE7gZ4NP1i7\nB3owpnsnqms0t43oyp2fbOPTTUcaTZhZGJfO22uMx4oKcOfz3w+lk7sTk3oFUVVTw03zNjN/aypP\nXdmrxc935b4s7vzEyOG7Z2wkVzYRGAy3Vkd5d20Kd43uhr+7E+uTcpn53mbACICLy6t4YP4OVu7P\nZkb/zqxKyGZjcuPAuKyimpX7jZ656hrdovzGo4V1Pc+Pf7OnwX2PfWMEzNP6BnNl385YTIoJPQNJ\nyi5mxd4snr6y12kn/GitOZBdhMWkKCyr5K/f7+PzzakAvL0mmbKKKp6Z3huAxMwiegZ7cpv1uHgD\nG+eOr2vbtLanNvXr4s2fL+/BC8sTAfjH1X0I93PDxdHMc1f3JsTbmR5BnvQI8uCBCdE8tzSBaX3r\nLjo9nR0YFxPY7OPXXiwNDPfmkckxtvcLtyZSB2qHzm9+fzNbHhtPwCmv163W0ZIRkf4MifClT6g3\n17y1AYAZb6zn8WmxzJm/Ew8nC1ufmMC9n28np6ic2Z/FAaAU1A4Y3DikC8Xl1Xy36yj/XpHIny7v\nccb/VVV1DT/szWRAmA/px0t5a00ynTycWHLfSCqqaxj47I/MW3uIzYfqRgzS88sI83Ul9XgpPQI9\nSMwq4rIeAdw0NIzB4b78a3kCi7ZnUFBagberY6O0hs0peVz/7iYsJsXyB0fbJs6vTsxmSIQvro5t\nC4UqqmooLq8rw7krrYCj1pGfqhrNzcPCmDk0nCmv/MLba5Lxc3PkjZkDiQ5wp7i8imAvlzP+jXA/\nN5beP4qC0kp8rD2tJpOyXfQs3XOMd9YYE0MP/n0K+SUVzP4sDj83R5QyLiKaSvOoqdH8falR2nF9\nUi5zF8U3GLUaFeXP4HAflFI4mOt+NybIk9duHMjM9zYx/fV1lFVW0zfUi3A/N8bFBDCxZyAOZhN5\nJRUs3pnBzcMaT0q90NlZYGyhssR4ysnBbUtYrx0KDPY+8wua/NplnzveC0OcO7U9jk8u3gsY+a5T\n+wSf079ZWGakMjw4oTtvr0lmaXzTvXFvrk7G3cnC53cObXI05YEJ0WxIzsXT2YF7L4tkUHjjSYTj\nYwL4w9hIbhwSxksrD7ArrYCpfYK4dbhxHoT5ubLnmUl8vOEwLyxPNHr2rD2KwyP92H/shC0dwcXR\nzNwpsY3+RnM2H8rDbFKNKmfUN6lXEF9tS+ft1cn8fnQ3krKLbcfkuaVGDt/K/Vn06uyJxboAz5Te\nQVzS1ZfbRzU9O/yj312C1kbQMDTCl482HG4UGH+3+yh+bo48Pi2WCT0DbaMFtb08k3oFsjT+GH+5\nomeLRrK+3ZFhG0oGmDm06eXl3Z0svDlzINPfWM+sD7Zwz9hIPtt0xHZ/cXkVd3y0lW1H8pkzPpo/\njoti9mdxzN+ahq+bI1sPH+eq/iHcMiycxTvryoZl5Jc1mwsKxuSi/25ObbYToZu/Gym5JfTr4s0b\nNw1scN+kXkGs2JfF7oxC+nfxtm1fFn+MNQdyeHZ6b35OzOZkZTUFpZXcPz6aV1cdtAXFY3t0YnVi\nDh9vPMLcqbFUVteQmFXErOHn7r149phIftyXxc60Ai6LCbBdQDlZzDxUL2C8a3Qktwzr2qaREieL\n+YyTRr1cHIjs5EZyTgnL92ay9+gJuvi6cu9lRhrHtiP5OJpN9A01XvODwn049PxU4jMK+eMXO5gz\n33hNFZVX8diieNuE2Vpf3z2cp5bs5fGpsYyI8qekvIqth47zv93HzhgY5xWXc89n29lySirT/eOi\nUErhZDEzMsqfFfuycHEws/r/xrIzrYCxPTrhZDGTV1yOj6sjWUUnGwSVtwwP57+bU7nh3U0cKzxJ\nYVklFpMizM+V128cyDvWiiImk2L2p3EsmD2CrKKT3PbhVrr5u/HZnUMpKK0kKaeY/6xI5L1Zl5yx\nZ7OwtJIpr6y1Xfj5uzuyZNfRBvvcNqIrUQEe3DEqgoqqGqb0CWKY9cLFrxU5tkopW1Bcy9/diT4h\nXragGCA5p5jPNh2hqlqz8J4R/O37/azcn8VzS/fbLvyKTlbi4ezQYPLwP5YlkFFQxrQ+wcyZEE2o\njwsuDuZm34eGR/oxc2g4n246gqezha/uHt6oIyLE26XDTvi1u8C4VkYb53VkWvPugjxbMBxTcMRI\n3/BqW2kocXEK8HCmk4eT7QMn4dgJhkT44uFsaVOlhNPZlJLH00v28n+TjA+sHkEeXBrtz6ebjmA2\nKZ6+qhfrk3LxcnHA2cHMsvhj3DgkrFEvU63eIV5seXwCNVo3mwZiMZt4eLIxhP3s9N7cOCSM/l28\nG7zJujtZGB3diReWJ7IhuS4Hrn6gFx3gTrK1ZnhJeRXfxx9jRv+QRqXuSsqrqKyuwcvFgR/2ZNom\nNDVnfGwgQyJ8WbEvC41R+/eVG/rzwbpD7EovtO23I9VIBwnxduHNmQNPG6wqpai9e1KvIJ793z6u\nfnM9C+8ZgVIKrTXxGYVM6hnUoOzWqe1avjeLhMwiYoM9Sc4p5tsdGTw4obut11RrzSMLd7M6Mce2\n2IqHs4XbRnQl1Kf5ILVPiBdeLg7sPXqC+z43qnNM6xvMPWMiueHdTWw+dJwpvYN4cKIxkeeSrr6s\n3J9t6y3fejifn/Zn8XO9GrtJOUWE+bmitaasstrW67Z4ZwY7UgvYmWZ8Gf8fuGt0N/YfK+Lf1/Vj\n1f4sxscG8r/dR7mib+Ne7vGxAUbv3t5M4o7kk3a8lKev6sXzyxJIPV5Ken4Z6+rlTg4K9+GmoWF4\nuTjg5eLAFX2D2ZNRyOzPtrM6MYf4jAIqqmqa/Ftni9mkWHTPCE6crDzjCNDZTB9qysqHxhD52FLb\nxTfATUPC8HFzZMuh4/Tr4tUgkFFK0TfUmz+Oi+bPX++iq58rh/NKWbQjgwFh3lRU1aA1fD17OG5O\nFr6/vy6Nxs3Jwu9GduX5ZQlMfnkttw7vSnSgu+3i9HhJBQnHTjC0mx9zF8U3CorvHt2Nu8dE2m5P\n7m1cFD0wIZpAT2cm9arrWa8NJk/taY0J8rSlaNXqFeLFwawipr76C2AEqRN7BjLzvc0s2pFum6iX\nklvCyH/+RP20+XlrU/jntX2b/f+eOGlM5q0/GjL/rmE8+e1eth05zpNX9CQ1r5SoAKOn/1xNtJ7a\nJ5j4jELbheHKfVnM35LGDUO60NXfjTHd/Vm5P4t5vxziNwNC0WimvbqOP46L4n+7j+HqaGZ8bCDf\nWQP6x6fF0rklnX4YnSTODiYm9w5qdnSuo7K7wLjrhBwqomehTSva9BBZJ05aS460YOg7/zB4hYC5\n/fJIRcfw2NQYFm3P4FBuCa/+lMSrPyUxvJsfn905lKMFZYT6uPzq2r+V1TXc/WkchWWVtlJRXf1d\nuX1UBCv2ZfHRhsOMiPTjrk/jGvze6WazA62a+e3uZGm297a2t7E22HJ3slBsnYx16/Bw8koq2JNh\nBKofbzzMv35IZPuRfP5xTd0H1ks/HuAV6+pk3947kuSckmZ7desbFuHLqz8l2VYdq+0lq2U2Kduk\nlnA/11Ydi9p8vu2pBRzMLqZ7oAfp+WUUlFbSJ7T5HK6xPTrhYFY8t3Q/n9w+hMcWxbP50HGGRvgx\nKtqf/6xIZHtqAeuScgnwMAKECbEBzLt18BnbZzIp4p6YQNTjy2zbrurXmd4hXswcFsY7a1Ia5KPe\nMjycCH838koqbNULao/TGzcN5L4vtvP7T+K4pKsPacfLOF5SwY8PjaaThxNPfLvHlm8N4Ggx8cjk\nGO6od1xuGGL0bjdXn9Xb1ZFh3fx4q14aS2lFFanWiinrknIxmxR/m9GbVfuzGRjm3eh1W/s/qU0B\nGNLV19ZLeq6YTKpd0qLORCnFrcO78tGGw3QPdOdAVjG/fWcj/azB4x/GRjb5e78ZEGJctHb35/p3\nNhGfUcjvRkYwrU8wWutGefO1RnfvxPPLEkjILOKxb+LxcnFg+5MTKa+q5uo313M4rxQ3RzMlFdVc\n0TeYv07vzdqDOfQO8Wq0HsBvBoTQv4t3q0u33TQkjJ1pBTx/dR+uGRiKg1nx6qokXlp5ADAmNY6I\n8ifC3421B3IIq1fX+9S5pMv2HOOZ6b0oKK2kWmtC6gWLFVU1vP5TEun5pbw/azBfbEnl9lERRAV4\n8Pnvh1JUXtVuc0fuvDSCS6P9iQ325IN1h3hxxQGUgrtHG8d35tBwgrxc+P0n22wXCACv/ZSEo9nE\nw5N7MC4mgP/tPoqns0PLStBa+bk7/ar0owuZnQXGZlz8K3EZGg2JbQuMMwtP0snDqdk3CJvqKsjc\nI2kUokm/GRDKbwaE8vW2NF5eeZCMgjI2puRx96dxrNyfxW0jutomi7XVh+sPUVhWiaezhRPWQCXc\n142YIDP7n53MiH+sahQUD43wbbcak14uDvi4OrD2gBFwLfrDCDILTzIw3AdXBzOv/nSQpfHHKCmv\nIt7akzt/axpT+wQzunsnCssqbUExGPmRHs6W007AqXVp9068+lMSB7KMmuRVNZq7x0Qyc2gY1769\ngX9f15+YYA/u/iSuwYS1lugR5MG/runLwwt3s+5gLtEB7rbVo+qnBZwqwMOZv1zRkycX7+Wq19cT\nb70oePCrncy/axiv/mRMjHNxMPPnST14eMFuzCbV4qDdYjax8qHRTPjPWgAGW0ul3T06kuKTVUyr\nl87j6mjhcmtP3SVdfXjoq11c0tWXyE7uTO0TxKRdQfywN5NNKccZFO5DRkEZo/75M7eN6ErRySre\nvnkg0YEehPm6Ul2j29SjdMvwcDam5NkuUL7aZuSqe7s6UFBaycyhYdw4xPhqSmcvZ1sN6yBPZ169\ncUC7LzRzPj0xLZYnpsViMZv4alsaDy/YzUHrCExzqVtmk7JNuP3sjqGk5ZfSq7On9f/W/P8uNtiT\n/c9OJjmnmK+3pfHxxiOsPZhDeWU1h/NKueGSLpRX1RAb7MEdo7phNqlmF3pQSjWoytFS1w0OZXik\nX4OFbO69LJKr+nfGyWKy9YSO7dGJD9cfxtnBxIhIP6NHO7OIJfeNpG+oNxuSc7lp3mYe/HIny6wl\nJ+8e040546NxdbRw/xc7+GFvJhN7BjI+1viq3/b2nFDtYDbZ0sBevK4ff/t+H6O7d7L9D0wmxfiY\nhmUpHS0mZo+J5PKegXWlAIeFU1ldY1fnx+mo87UG++DBg/W2bdvOvOPZdLIQ/hEGl/+dPgffASB+\nVnyrHuKW9zdzoqySxfeNOv2O2z+FJffBjLeg/01tbbGwA5XVNRwtKGPGG+vJLzVKLJkULJszmu6B\nLayXfYrFOzP489e7GNsjgHdvGURKbglxR/L57eC61RZfXnmAl1fWBZbDu/nxxV3Dfv0TaoUZb6xn\nZ1oBfm6ObHtiQoPnWvsB1ZTeIZ64O1nYlFI3LOtoNvHs9F623sgzWbwzgznzd3LDJV3464zeLa7j\n2lJTXvmFA1lFuDiYbT3hKc9NPe1ksqrqGq54bR0J1pJ6MUEeJGQWMSDM25ba8fL1/ZnYM5A583fy\nyOQeRLeyDnLa8VK2p+b/qhWoCkorWJ+UR6CnE/27ePPC8kRbHmeAhxMb544/KwsPnDhZiVkprn93\nI1P7BDO5VxC5xRU8snA3n/9+6BknL8WnF/LeuhT+eU3fi264tzW01sxdFM8vB3NZ9IcRjSaQnk0V\nVTUMf34Vzg5mIvzd2JVewPYnJ57186utDmYVMfEl4+Lw69nD6d/Fm2MFJ20jWDU1mnv+G2erRV1r\nVJQ/L17Xj2HPrwKwBdIdweKdGWw9fByLycSgcJ8mJ+naA6VUnNb6jOVT7CswriiB5zrDxGfpk/Qe\n0PrA+LIXVxMT5MFbNw9qfOeuL2HVs/DHOPj577D5HXgiC+QqTLTA6sRs5m9J48p+nbn38+2AMaGn\nfsmplqip0Qx5biX+7k7Mv2tYs0O7+SUVDHluJXeN7saXW9OZd+ugRqWqzrWP1h/i6e/2cf3gLo1y\n+qprNJGPLbXd/vtvevPPZQm23m+TAk8XB57/TR8GdfXBz82p1cHYD3syGRLhi+85qKuZkHmCyS8b\nw5cWkzG0/Zcrzzz0WFxexbqDObg7OdA/zJtpr/7CkbxS/N2dWPN/Y0+bP30+3fHRVlYlZPPHcVEt\nqk4gLl6v/3SQF1cYKQw3Dgm74FalW7LrKCHezk1OHgZjFclhz6+id2cvWy672aRwdTBTVF7F53cO\nbVQvXFz4WhoYX5jvsOdK7eS7mioWXLkAF0vLksxrlVdVcySvhCv6NlNBYPEfjLrFqRugMN3IL5ag\nWLTQ2B4BjO0R0GDxhHfWJjOxZwCDwn0pr6om+0Q5oT4uHMkzJiA5WkyNlpfed+wEucUVzJ0Se9p8\nRx83RxL+OgWzSfF/k1oXfJ8tNw8LN2qRDmjce2k2Kf73x1Hkl1ZgMZlsM6FrajTp+WWE+LhQUVXz\nqyYy1a/TfLbFBHny1d3DKSitYGLPwEZ5jM1xd7IwuXfde8zlPQOZ98shLu8VeMEGxQBPXdmL6QNC\nmHoO/6eiY7hvXDRDIvz4aMMh/nx50yuznU9nSrdydjCz4dFxmE2KkvJqPJ0tvLAi0VYBYvBpqt6I\nju/CfZc9F2yBcTU9fFvfo3Ekr5QaTdNlXGqqjQoUVEHSKmNxD6lGIdqgfjpBoIczL688yNwpsVz3\n9gZKKqq597JI5v1yiIqqGgD+dW3fBtUaavNZL22iuP+pzsZw969hMZtOO1nu1MU0wMibqx32PNez\n+3+t+hctbb1GfmRyDIPCfRtdAF1owvxcT1u+TdiXIREX/mv2dGqrrNRWCrplWDjvrEkh0NOpUWUc\ncXGxr8BYWV/MNVWn368ZB7OMiQunzqIFIPlnqLbWe9z4uvG97w1t+jtCfD17OIdzS9iRVsB3O4/y\n3roUSiqMCgpv/GzM1PdwslBUXsXDC3bzysqD/OOaPjhZzOxKLyTYy7nZkmuiY7GYTee0Z1sIcWah\nPq68ckN/enX2PPPOokOzr8sepYxe3TYGxmsP5ODuZCE68JTAWGvY+Bq4dYJBv6u3vfpXNFbYs0u6\n+nLd4C4M6OJNUXkVi7ZncO2guhGIv07vxY6/TLTdLq2o4pb3t/Dbdzby3a6j9Op8bstSCSGEvZne\nP8RWm1hcvOyrxxiMdIqa1gesldU1LN+XyfjYgMaLMKRthpTVMOl5qCyp2z7m0V/XVmH3Rkb5E+Dh\nxPjYAB6ZHMOCOKNkVUywJxazifl3DSPAwwlvV0cm/mcNeSUVAOe8XqsQQghxMbLTwLj1PcYbk/Mo\nKK1suv5j6ibje/8boaoC0rbCVa+BR2DjfYVohc7eLmx5fEKj7d2tvRa1y4sCrH90HDFP/gDAyKg2\nLu0ohBBC2DE7DYxb32O8fG8mro7mplcFy9oDXl3AxVrqauZXv7KRQjRt/l3D+DkhGy/XxkXknR3M\nBHg4kV1UTr8OUl9TCCGEuJDYYWBsalOP8bbD+VzS1bfpIvGZ8RDY+yw0TojTG9bNr0Eq35KQAAAM\nJklEQVQv8akW3zeSnKLyM6/MKIQQQohG7O/Tsw2pFJtT8kjMKmJQ+CmLH2gNpcchJxFCBp7FRgrR\nNsFeLh1mNSYhhBDiQmOHPcaWVlWL0Fpz//wdgDERyqa6Cj6dAYfXARoixpzlhgohhBBCiPZkn4Fx\nK3KMMwrKyDpRzpzx0Q17jPcvgcO/1N2WHmMhhBBCiA7NDlMpWlfHeMuh4wBc3uuUChNZe+p+HjgL\nzI0nQwkhhBBCiI7D/nqMW7nAx6r92fi7OxITdMpqNzmJ4N8drnwVQgef5UYKIYQQQoj2Zoc9xi2f\nfFdSXsXPidlM6hWE2aQa3pmTCJ16QPhw6S0WQgghhLgI2Glg3LIc4w/WHaK0opqrB4Y2vOPACsg7\nCCHSUyyEEEIIcbGww8DY3KLAeENSLv/+8QBDInwZGHZK+as9C8AtAIbdc44aKYQQQggh2psdBsYt\nS6VYuD0DdycLH952CUopKMmFdS9DXrLxFRALFqd2aLAQQgghhGgP9jf5rgVVKbTWrDmQzcSegbg5\nWaC8GD6cArkHIHufkUbR+9p2arAQQgghhGgPdhgYn7nHOKeonNziCvqGehkbDq0xgmJnL9j9pbHN\nL/IcN1QIIYQQQrQn+0yl0DWn3WV/ZhEAscGeUJwNKauNMm/XvF+3U8igc9hIIYQQQgjR3uywx9gM\nlRVN3rU7vYDZn8ZRWaMBiPWqgpf7QVWZMdkueiL0vR6cPCBsWHu2WgghhBBCnGN2GBg3n0rxycYj\nHC08CcAT02LxOrbOCIrdg2DUg8ZOV7/bXi0VQgghhBDtyP4C42ZWvludmM2CuHQA5k6J4c5Lu8E3\n/wJnb3hon9HTLIQQQgghLlr2Fxg3s8DH4p1HAVh4zwgGhftARSns/w56zpCgWAghhBDCDtjh5Lum\nF/g4mF3E6O6djKAYIHUDVBRD76vbuYFCCCGEEOJ8sMPAuGGO8YvLE9l6+DhJ2cVEdXKv268ww/ju\nH93ODRRCCCGEEOeDnaZSGIHxscIyXv85idd/TgIgOrBeYFx0zPjuHtTeLRRCCCGEEOeB/fUYW5yg\nyqg8se1wfoO7ogPqBcYnjoJbJ7A4tmfrhBBCCCHEeWJ/PcYuPlB6HLRmR2pBg7uiAtxBa2N1u+z9\n4BF8nhophBBCCCHam30GxtXlUFlGck5xg7u8XR3h8Hr45m5jQ/fJ56GBQgghhBDifLC/VApXX+N7\n2XEO55UwITYAFwczY7p3Mrbv/rJu3+B+7d8+IYQQQghxXthhj7ERGFcW55GeX8aVfTvz+k0D6+5P\n3WhMuIu9EkY9dJ4aKYQQQggh2pvd9hjnZh+jukbT1d8NZwczzg5mqCiB3IMw6DaY9iI4OJ/ftgoh\nhBBCiHbTosBYKTVZKZWolEpSSj3axP1OSqkvrfdvVkp1PdsNPWusPcbHMo2V7rrXL9GWtRfQkkIh\nhBBCCGGHzhgYK6XMwBvAFKAncKNSqucpu90B5Guto4CXgH+e7YaeNS7GynY52Zk4mBU9gjzq7sve\nZ3wP7HUeGiaEEEIIIc6nlvQYDwGStNYpWusKYD4w/ZR9pgMfW39eAIxXSqmz18yzyNUPlImyvDS6\nB3rgZDHX3Zd7ECwu4NXl/LVPCCGEEEKcFy0JjEOAtHq3063bmtxHa10FFAJ+Z6OBZ53FkepOPfEr\n2MOQCN+G9+UeAL8oMNlf6rUQQgghhL1rSVWKpnp+dRv2QSl1F3CX9WaxUiqxBX//XPAHcqE3Tzd1\n7x8uzM5u0SrWYywucnKc7YMcZ/sgx9k+nK/jHN6SnVoSGKcD9XMLQoGjzeyTrpSyAF7A8VMfSGv9\nLvBuSxp2LimltmmtB5/vdohzR46xfZDjbB/kONsHOc724UI/zi3JGdgKRCulIpRSjsANwJJT9lkC\nzLL+fC3wk9a6UY+xEEIIIYQQF6oz9hhrrauUUvcBywEz8IHWeq9S6llgm9Z6CfA+8KlSKgmjp/iG\nc9loIYQQQgghzrYWrXyntV4KLD1l21/q/XwSuO7sNu2cOu/pHOKck2NsH+Q42wc5zvZBjrN9uKCP\ns5KMByGEEEIIIexxSWghhBBCCCGaYFeB8ZmWthYdh1Kqi1LqZ6XUfqXUXqXUHOt2X6XUj0qpg9bv\nPtbtSin1qvXY71ZKDTy/z0C0lFLKrJTaoZT6n/V2hHXp+YPWpegdrds7ztL0ogGllLdSaoFSKsF6\nTg+Xc/nio5R60Pp+vUcp9YVSylnO545PKfWBUipbKbWn3rZWn79KqVnW/Q8qpWY19bfag90Exi1c\n2lp0HFXAn7TWscAw4F7r8XwUWKW1jgZWWW+DcdyjrV93AW+1f5NFG80B9te7/U/gJesxzsdYkh46\n0tL04lSvAD9orWOAfhjHW87li4hSKgS4Hxiste6NMZn/BuR8vhh8BEw+ZVurzl+llC/wFDAUY8Xl\np2qD6fZmN4ExLVvaWnQQWutjWuvt1p+LMD5IQ2i4PPnHwAzrz9OBT7RhE+CtlApu52aLVlJKhQLT\ngPestxUwDmPpeWh8jDvG0vTCRinlCYzGqG6E1rpCa12AnMsXIwvgYl3vwBU4hpzPHZ7Wei2N165o\n7fk7CfhRa31ca50P/EjjYLtd2FNg3JKlrUUHZB1iGwBsBgK11sfACJ6BAOtucvw7ppeBh4Ea620/\noMC69Dw0PI4dZ2l6UV83IAf40Joy855Syg05ly8qWusM4EUgFSMgLgTikPP5YtXa8/eCOa/tKTBu\n0bLVomNRSrkDC4EHtNYnTrdrE9vk+F/AlFJXANla67j6m5vYVbfgPnHhsgADgbe01gOAEuqGXZsi\nx7kDsg6LTwcigM6AG8aw+qnkfL64NXdcL5jjbU+BcUuWthYdiFLKASMo/q/WepF1c1btsKr1e7Z1\nuxz/jmckcJVS6jBG6tM4jB5kb+tQLDQ8jrZjrE6zNL244KQD6VrrzdbbCzACZTmXLy4TgENa6xyt\ndSWwCBiBnM8Xq9aevxfMeW1PgXFLlrYWHYQ11+x9YL/W+j/17qq/PPksYHG97bdaZ8QOAwprh3nE\nhUlrPVdrHaq17opxvv6ktZ4J/Iyx9Dw0PsayNH0Ho7XOBNKUUj2sm8YD+5Bz+WKTCgxTSrla379r\nj7Oczxen1p6/y4HLlVI+1tGFy63b2p1dLfChlJqK0eNUu7T1389zk0QbKaVGAb8A8dTlnz6GkWf8\nFRCG8UZ8ndb6uPWN+HWMZP5S4Hda623t3nDRJkqpscCftdZXKKW6YfQg+wI7gJu11uVKKWfgU4x8\n8+PADVrrlPPVZtFySqn+GBMsHYEU4HcYHTdyLl9ElFLPANdjVBXaAdyJkUcq53MHppT6AhgL+ANZ\nGNUlvqWV569S6naMz3GAv2utP2zP51HLrgJjIYQQQgghmmNPqRRCCCGEEEI0SwJjIYQQQgghkMBY\nCCGEEEIIQAJjIYQQQgghAAmMhRBCCCGEACQwFkKIi4pSaqxS6n/nux1CCNERSWAshBBCCCEEEhgL\nIcR5oZS6WSm1RSm1Uyn1jlLKrJQqVkr9Wym1XSm1SinVybpvf6XUJqXUbqXUN9aVoVBKRSmlViql\ndll/J9L68O5KqQVKqQSl1H+tRfWFEEKcgQTGQgjRzpRSsRgrgI3UWvcHqoGZgBuwXWs9EFiDsYIU\nwCfAI1rrvhirPdZu/y/whta6HzACqF0aeQDwANAT6AaMPOdPSgghLgKW890AIYSwQ+OBQcBWa2eu\nC5CNsbz5l9Z9PgMWKaW8AG+t9Rrr9o+Br5VSHkCI1vobAK31SQDr423RWqdbb+8EugLrzv3TEkKI\njk0CYyGEaH8K+FhrPbfBRqWePGU/fYbHaE55vZ+rkfd6IYRoEUmlEEKI9rcKuFYpFQCglPJVSoVj\nvCdfa93nJmCd1roQyFdKXWrdfguwRmt9AkhXSs2wPoaTUsq1XZ+FEEJcZKQXQQgh2pnWep9S6glg\nhVLKBFQC9wIlQC+lVBxQiJGHDDALeNsa+KYAv7NuvwV4Ryn1rPUxrmvHpyGEEBcdpfXpRuqEEEK0\nF6VUsdba/Xy3Qwgh7JWkUgghhBBCCIH0GAshhBBCCAFIj7EQQgghhBCABMZCCCGEEEIAEhgLIYQQ\nQggBSGAshBBCCCEEIIGxEEIIIYQQgATGQgghhBBCAPD/gzJFqBOCCt8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdb4c42c748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "\n",
    "for num_filters in num_filters_alts:\n",
    "    curve = DataFrame(np.array(train_accs[num_filters]) - np.array(test_accs[num_filters])).ewm(alpha=0.1).mean()\n",
    "    plt.plot(curve)\n",
    "plt.ylim(0.0, 0.5)\n",
    "plt.legend(['{} params'.format(param_counts[num_filters]) for num_filters in num_filters_alts])\n",
    "plt.xlabel('epoch')\n",
    "plt.title('CIFAR10, m=4000; risk difference')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}