fedML_cifardemo-singlemembers.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from __future__ import print_function\n",
    "import keras\n",
    "from keras.datasets import cifar10\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten, BatchNormalization\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "import os\n",
    "\n",
    "\n",
    "batch_size = 32\n",
    "num_classes = 10\n",
    "epochs = 100\n",
    "data_augmentation = True\n",
    "num_predictions = 20\n",
    "save_dir = os.path.join(os.getcwd(), 'saved_models')\n",
    "model_name = 'keras_cifar10_trained_model.h5'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (50000, 32, 32, 3)\n",
      "50000 train samples\n",
      "10000 test samples\n"
     ]
    }
   ],
   "source": [
    "# The data, split between train and test sets:\n",
    "\n",
    "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
    "print('x_train shape:', x_train.shape)\n",
    "print(x_train.shape[0], 'train samples')\n",
    "print(x_test.shape[0], 'test samples')\n",
    "\n",
    "# Convert class vectors to binary class matrices.\n",
    "y_train = keras.utils.to_categorical(y_train, num_classes)\n",
    "y_test = keras.utils.to_categorical(y_test, num_classes)\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "x_train /= 255\n",
    "x_test /= 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape:  (50000, 32, 32, 3)\n",
      "x_train_splits shape:  (8064, 32, 32, 3)\n",
      "x_train_splits shape:  (8065, 32, 32, 3)\n",
      "x_train_splits shape:  (6451, 32, 32, 3)\n",
      "x_train_splits shape:  (6452, 32, 32, 3)\n",
      "x_train_splits shape:  (4838, 32, 32, 3)\n",
      "x_train_splits shape:  (4839, 32, 32, 3)\n",
      "x_train_splits shape:  (3226, 32, 32, 3)\n",
      "x_train_splits shape:  (3226, 32, 32, 3)\n",
      "x_train_splits shape:  (1613, 32, 32, 3)\n",
      "x_train_splits shape:  (1613, 32, 32, 3)\n",
      "x_train_splits shape:  (806, 32, 32, 3)\n",
      "x_train_splits shape:  (807, 32, 32, 3)\n",
      "tot len:  50000\n"
     ]
    }
   ],
   "source": [
    "from fedml import Alliance, AllianceMember,KerasSequentialBaseLearner, FedAveragingClassifier\n",
    "from read_data import read_training_data\n",
    "import numpy as np\n",
    "\n",
    "alliance = Alliance()\n",
    "M=10\n",
    "def split_data(x,y,fractions):\n",
    "    s = len(x)\n",
    "    new_order = np.random.permutation(s)\n",
    "    x = x[new_order]\n",
    "    y = y[new_order]\n",
    "    if type(fractions) is int:\n",
    "        fraction_splits = np.int32(np.linspace(0,s,fractions+1))\n",
    "        nr = fractions\n",
    "    elif type(fractions)==np.ndarray or type(fractions)==list:\n",
    "        parts = np.array(fractions)*s/np.sum(fractions)\n",
    "        fraction_splits = np.int32([np.sum(parts[:i]) for i in range(len(parts)+1)])\n",
    "        nr = len(parts)\n",
    "        \n",
    "    x_list = []\n",
    "    y_list = []\n",
    "    for i in range(nr):\n",
    "        x_list.append(x[fraction_splits[i]:fraction_splits[i+1]]) \n",
    "        y_list.append(y[fraction_splits[i]:fraction_splits[i+1]])\n",
    "\n",
    "    return x_list,y_list\n",
    "F = [10,10,8,8,6,6,4,4,2,2,1,1]\n",
    "M = len(F)\n",
    "x_train_splits, y_train_splits = split_data(x_train,y_train,fractions=F)\n",
    "x_test_splits, y_test_splits = split_data(x_test,y_test,fractions=F)\n",
    "print(\"x_train shape: \", np.array(x_train).shape)\n",
    "tot_len = 0\n",
    "for split in x_train_splits:\n",
    "    print(\"x_train_splits shape: \", split.shape)\n",
    "    tot_len += len(split)\n",
    "print(\"tot len: \", tot_len)\n",
    "alliance.set_validation_dataset(x_test,y_test)\n",
    "\n",
    "classes = np.arange(10)\n",
    "i = 1\n",
    "member = AllianceMember(x_train_splits[i], y_train_splits[i], classes=classes)\n",
    "alliance.add_member(member)\n",
    "epochs = int(1563*500/int(np.ceil(len(x_train_splits[i])/32)))\n",
    "print(\"epochs: \", epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/ubuntu/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:4070: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from cifarmodel import KerasSequentialCifar\n",
    "\n",
    "base_learner = KerasSequentialCifar()\n",
    "fedavg_model = FedAveragingClassifier(alliance=alliance,base_learner=base_learner)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fit starts!\n",
      "else: alliance temp model:  <cifarmodel.KerasSequentialCifar object at 0x7f0a3cf7b2b0>\n",
      "global epoch:  0\n",
      "virtual memory used:  63.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2122 - accuracy: 0.9263\n",
      "training time:  16.1599 s.\n",
      "average model starts\n",
      "average model ends:  0.1865 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.8   0.786]\n",
      "global epoch:  1\n",
      "virtual memory used:  64.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2116 - accuracy: 0.9306\n",
      "training time:  12.4008 s.\n",
      "average model starts\n",
      "average model ends:  0.1071 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.786 0.794]\n",
      "global epoch:  2\n",
      "virtual memory used:  65.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2149 - accuracy: 0.9267\n",
      "training time:  12.5005 s.\n",
      "average model starts\n",
      "average model ends:  0.0729 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.794 0.794]\n",
      "global epoch:  3\n",
      "virtual memory used:  66.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2209 - accuracy: 0.9235\n",
      "training time:  11.7159 s.\n",
      "average model starts\n",
      "average model ends:  0.0806 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.791]\n",
      "global epoch:  4\n",
      "virtual memory used:  66.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2355 - accuracy: 0.9239\n",
      "training time:  13.5763 s.\n",
      "average model starts\n",
      "average model ends:  0.1497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.798]\n",
      "global epoch:  5\n",
      "virtual memory used:  67.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2156 - accuracy: 0.9266\n",
      "training time:  13.3012 s.\n",
      "average model starts\n",
      "average model ends:  0.0778 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.791]\n",
      "global epoch:  6\n",
      "virtual memory used:  65.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2295 - accuracy: 0.9242\n",
      "training time:  14.066 s.\n",
      "average model starts\n",
      "average model ends:  0.0652 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.796]\n",
      "global epoch:  7\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2079 - accuracy: 0.9318\n",
      "training time:  12.8647 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.786]\n",
      "global epoch:  8\n",
      "virtual memory used:  67.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2148 - accuracy: 0.9292\n",
      "training time:  13.1275 s.\n",
      "average model starts\n",
      "average model ends:  0.0659 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.786 0.804]\n",
      "global epoch:  9\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2149 - accuracy: 0.9315\n",
      "training time:  12.4594 s.\n",
      "average model starts\n",
      "average model ends:  0.0728 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.804 0.799]\n",
      "global epoch:  10\n",
      "virtual memory used:  68.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2370 - accuracy: 0.9188\n",
      "training time:  11.1359 s.\n",
      "average model starts\n",
      "average model ends:  0.0769 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.799 0.791]\n",
      "global epoch:  11\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2169 - accuracy: 0.9262\n",
      "training time:  12.8501 s.\n",
      "average model starts\n",
      "average model ends:  0.0583 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.791 0.794]\n",
      "global epoch:  12\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2139 - accuracy: 0.9270\n",
      "training time:  12.7512 s.\n",
      "average model starts\n",
      "average model ends:  0.0348 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.794 0.798]\n",
      "global epoch:  13\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 54ms/step - loss: 0.2208 - accuracy: 0.9252\n",
      "training time:  13.6151 s.\n",
      "average model starts\n",
      "average model ends:  0.0602 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.787]\n",
      "global epoch:  14\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2183 - accuracy: 0.9309\n",
      "training time:  14.0091 s.\n",
      "average model starts\n",
      "average model ends:  0.0608 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.787 0.794]\n",
      "global epoch:  15\n",
      "virtual memory used:  67.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2100 - accuracy: 0.9317\n",
      "training time:  12.9564 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.794 0.8  ]\n",
      "global epoch:  16\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2259 - accuracy: 0.9250\n",
      "training time:  14.0624 s.\n",
      "average model starts\n",
      "average model ends:  0.0468 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.796]\n",
      "global epoch:  17\n",
      "virtual memory used:  68.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2148 - accuracy: 0.9281\n",
      "training time:  13.5288 s.\n",
      "average model starts\n",
      "average model ends:  0.1209 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.798]\n",
      "global epoch:  18\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2403 - accuracy: 0.9199\n",
      "training time:  13.0379 s.\n",
      "average model starts\n",
      "average model ends:  0.0602 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.792]\n",
      "global epoch:  19\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2289 - accuracy: 0.9246\n",
      "training time:  13.225 s.\n",
      "average model starts\n",
      "average model ends:  0.1196 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.796]\n",
      "global epoch:  20\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2276 - accuracy: 0.9251\n",
      "training time:  12.7054 s.\n",
      "average model starts\n",
      "average model ends:  0.1808 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.796 0.791]\n",
      "global epoch:  21\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2119 - accuracy: 0.9320\n",
      "training time:  12.3119 s.\n",
      "average model starts\n",
      "average model ends:  0.0313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.798]\n",
      "global epoch:  22\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2246 - accuracy: 0.9251\n",
      "training time:  13.8913 s.\n",
      "average model starts\n",
      "average model ends:  0.0601 s.\n",
      "alliance_test_loss\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.775]\n",
      "global epoch:  23\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2128 - accuracy: 0.9318\n",
      "training time:  11.7944 s.\n",
      "average model starts\n",
      "average model ends:  0.0393 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.775 0.793]\n",
      "global epoch:  24\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2328 - accuracy: 0.9244\n",
      "training time:  12.2444 s.\n",
      "average model starts\n",
      "average model ends:  0.1186 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.775 0.793 0.79 ]\n",
      "global epoch:  25\n",
      "virtual memory used:  68.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2127 - accuracy: 0.9307\n",
      "training time:  12.7315 s.\n",
      "average model starts\n",
      "average model ends:  0.0783 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.79  0.792]\n",
      "global epoch:  26\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2083 - accuracy: 0.9330\n",
      "training time:  12.2284 s.\n",
      "average model starts\n",
      "average model ends:  0.0606 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.792 0.789]\n",
      "global epoch:  27\n",
      "virtual memory used:  70.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2318 - accuracy: 0.9204\n",
      "training time:  13.2583 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.789 0.801]\n",
      "global epoch:  28\n",
      "virtual memory used:  70.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2212 - accuracy: 0.9247\n",
      "training time:  12.8559 s.\n",
      "average model starts\n",
      "average model ends:  0.0513 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.801 0.8  ]\n",
      "global epoch:  29\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2348 - accuracy: 0.9239\n",
      "training time:  12.7164 s.\n",
      "average model starts\n",
      "average model ends:  0.0365 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.789]\n",
      "global epoch:  30\n",
      "virtual memory used:  69.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2268 - accuracy: 0.9235\n",
      "training time:  11.7883 s.\n",
      "average model starts\n",
      "average model ends:  0.0507 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.789 0.802]\n",
      "global epoch:  31\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2165 - accuracy: 0.9247\n",
      "training time:  13.4046 s.\n",
      "average model starts\n",
      "average model ends:  0.0512 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.802 0.786]\n",
      "global epoch:  32\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2300 - accuracy: 0.9278\n",
      "training time:  12.9654 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.786 0.797]\n",
      "global epoch:  33\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2120 - accuracy: 0.9294\n",
      "training time:  13.1097 s.\n",
      "average model starts\n",
      "average model ends:  0.0944 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.797 0.792]\n",
      "global epoch:  34\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2058 - accuracy: 0.9346\n",
      "training time:  12.6646 s.\n",
      "average model starts\n",
      "average model ends:  0.0465 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.792 0.792]\n",
      "global epoch:  35\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2217 - accuracy: 0.9304\n",
      "training time:  13.0044 s.\n",
      "average model starts\n",
      "average model ends:  0.0486 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.797]\n",
      "global epoch:  36\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2210 - accuracy: 0.9280\n",
      "training time:  13.3885 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.797 0.805]\n",
      "global epoch:  37\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2400 - accuracy: 0.9190\n",
      "training time:  13.4603 s.\n",
      "average model starts\n",
      "average model ends:  0.0512 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.805 0.792]\n",
      "global epoch:  38\n",
      "virtual memory used:  65.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2240 - accuracy: 0.9261\n",
      "training time:  12.8294 s.\n",
      "average model starts\n",
      "average model ends:  0.035 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.792 0.793]\n",
      "global epoch:  39\n",
      "virtual memory used:  65.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.2335 - accuracy: 0.9237\n",
      "training time:  11.6075 s.\n",
      "average model starts\n",
      "average model ends:  0.142 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.793 0.789]\n",
      "global epoch:  40\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2141 - accuracy: 0.9291\n",
      "training time:  12.8044 s.\n",
      "average model starts\n",
      "average model ends:  0.0553 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.789 0.8  ]\n",
      "global epoch:  41\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2154 - accuracy: 0.9273\n",
      "training time:  12.4741 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.8   0.793]\n",
      "global epoch:  42\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2190 - accuracy: 0.9287\n",
      "training time:  11.7485 s.\n",
      "average model starts\n",
      "average model ends:  0.0404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.782]\n",
      "global epoch:  43\n",
      "virtual memory used:  64.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2236 - accuracy: 0.9275\n",
      "training time:  13.7775 s.\n",
      "average model starts\n",
      "average model ends:  0.0865 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.782 0.792]\n",
      "global epoch:  44\n",
      "virtual memory used:  65.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2228 - accuracy: 0.9252\n",
      "training time:  12.2304 s.\n",
      "average model starts\n",
      "average model ends:  0.091 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.782 0.792 0.799]\n",
      "global epoch:  45\n",
      "virtual memory used:  65.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.2313 - accuracy: 0.9249\n",
      "training time:  14.64 s.\n",
      "average model starts\n",
      "average model ends:  0.0634 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.794]\n",
      "global epoch:  46\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2143 - accuracy: 0.9317\n",
      "training time:  12.7541 s.\n",
      "average model starts\n",
      "average model ends:  0.1559 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.786]\n",
      "global epoch:  47\n",
      "virtual memory used:  66.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2281 - accuracy: 0.9302\n",
      "training time:  13.8329 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.786 0.79 ]\n",
      "global epoch:  48\n",
      "virtual memory used:  66.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2180 - accuracy: 0.9270\n",
      "training time:  12.8979 s.\n",
      "average model starts\n",
      "average model ends:  0.0453 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.79  0.805]\n",
      "global epoch:  49\n",
      "virtual memory used:  67.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2300 - accuracy: 0.9189\n",
      "training time:  14.5562 s.\n",
      "average model starts\n",
      "average model ends:  0.0708 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.805 0.781]\n",
      "global epoch:  50\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2306 - accuracy: 0.9247\n",
      "training time:  11.8861 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.781 0.799]\n",
      "global epoch:  51\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2182 - accuracy: 0.9282\n",
      "training time:  13.2408 s.\n",
      "average model starts\n",
      "average model ends:  0.0514 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.799 0.791]\n",
      "global epoch:  52\n",
      "virtual memory used:  66.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2223 - accuracy: 0.9236\n",
      "training time:  13.4357 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.791 0.806]\n",
      "global epoch:  53\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2212 - accuracy: 0.9266\n",
      "training time:  13.2955 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.806 0.802]\n",
      "global epoch:  54\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2333 - accuracy: 0.9237\n",
      "training time:  13.3367 s.\n",
      "average model starts\n",
      "average model ends:  0.0592 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.802 0.802]\n",
      "global epoch:  55\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2174 - accuracy: 0.9291\n",
      "training time:  13.8775 s.\n",
      "average model starts\n",
      "average model ends:  0.1305 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.8  ]\n",
      "global epoch:  56\n",
      "virtual memory used:  66.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2329 - accuracy: 0.9222\n",
      "training time:  12.772 s.\n",
      "average model starts\n",
      "average model ends:  0.1426 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.787]\n",
      "global epoch:  57\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2299 - accuracy: 0.9246\n",
      "training time:  12.8025 s.\n",
      "average model starts\n",
      "average model ends:  0.1075 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.787 0.806]\n",
      "global epoch:  58\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2244 - accuracy: 0.9265\n",
      "training time:  12.3645 s.\n",
      "average model starts\n",
      "average model ends:  0.0611 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.806 0.787]\n",
      "global epoch:  59\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2340 - accuracy: 0.9230\n",
      "training time:  13.5545 s.\n",
      "average model starts\n",
      "average model ends:  0.0444 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.787 0.798]\n",
      "global epoch:  60\n",
      "virtual memory used:  69.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2132 - accuracy: 0.9298\n",
      "training time:  13.4115 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.798 0.801]\n",
      "global epoch:  61\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2136 - accuracy: 0.9313\n",
      "training time:  11.0823 s.\n",
      "average model starts\n",
      "average model ends:  0.1016 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.799]\n",
      "global epoch:  62\n",
      "virtual memory used:  68.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2325 - accuracy: 0.9232\n",
      "training time:  12.8773 s.\n",
      "average model starts\n",
      "average model ends:  0.0469 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.787]\n",
      "global epoch:  63\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2218 - accuracy: 0.9270\n",
      "training time:  14.1428 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.787 0.793]\n",
      "global epoch:  64\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2148 - accuracy: 0.9287\n",
      "training time:  13.5996 s.\n",
      "average model starts\n",
      "average model ends:  0.1243 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.793 0.792]\n",
      "global epoch:  65\n",
      "virtual memory used:  69.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2245 - accuracy: 0.9253\n",
      "training time:  12.9234 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.792 0.794]\n",
      "global epoch:  66\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2214 - accuracy: 0.9260\n",
      "training time:  14.429 s.\n",
      "average model starts\n",
      "average model ends:  0.0458 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.794 0.786]\n",
      "global epoch:  67\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.2189 - accuracy: 0.9257\n",
      "training time:  12.5613 s.\n",
      "average model starts\n",
      "average model ends:  0.0894 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.786 0.787]\n",
      "global epoch:  68\n",
      "virtual memory used:  69.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2235 - accuracy: 0.9267\n",
      "training time:  13.7014 s.\n",
      "average model starts\n",
      "average model ends:  0.0634 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.787 0.799]\n",
      "global epoch:  69\n",
      "virtual memory used:  69.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2204 - accuracy: 0.9281\n",
      "training time:  12.5448 s.\n",
      "average model starts\n",
      "average model ends:  0.0579 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.799 0.787]\n",
      "global epoch:  70\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2175 - accuracy: 0.9296\n",
      "training time:  12.165 s.\n",
      "average model starts\n",
      "average model ends:  0.0567 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.787 0.788]\n",
      "global epoch:  71\n",
      "virtual memory used:  69.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2326 - accuracy: 0.9209\n",
      "training time:  11.1099 s.\n",
      "average model starts\n",
      "average model ends:  0.0954 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.788 0.796]\n",
      "global epoch:  72\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2282 - accuracy: 0.9258\n",
      "training time:  12.8884 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.794]\n",
      "global epoch:  73\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2134 - accuracy: 0.9294\n",
      "training time:  13.6874 s.\n",
      "average model starts\n",
      "average model ends:  0.0647 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.797]\n",
      "global epoch:  74\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2236 - accuracy: 0.9251\n",
      "training time:  13.3471 s.\n",
      "average model starts\n",
      "average model ends:  0.1553 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.798]\n",
      "global epoch:  75\n",
      "virtual memory used:  69.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2247 - accuracy: 0.9253\n",
      "training time:  13.7868 s.\n",
      "average model starts\n",
      "average model ends:  0.0616 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.793]\n",
      "global epoch:  76\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2142 - accuracy: 0.9299\n",
      "training time:  13.3773 s.\n",
      "average model starts\n",
      "average model ends:  0.0482 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.793 0.795]\n",
      "global epoch:  77\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.2352 - accuracy: 0.9205\n",
      "training time:  14.7433 s.\n",
      "average model starts\n",
      "average model ends:  0.044 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.803]\n",
      "global epoch:  78\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2145 - accuracy: 0.9301\n",
      "training time:  13.3319 s.\n",
      "average model starts\n",
      "average model ends:  0.0425 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.803 0.798]\n",
      "global epoch:  79\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2108 - accuracy: 0.9307\n",
      "training time:  14.2089 s.\n",
      "average model starts\n",
      "average model ends:  0.0404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.79 ]\n",
      "global epoch:  80\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2226 - accuracy: 0.9294\n",
      "training time:  13.3782 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.79  0.802]\n",
      "global epoch:  81\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2275 - accuracy: 0.9235\n",
      "training time:  13.1471 s.\n",
      "average model starts\n",
      "average model ends:  0.0324 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.795]\n",
      "global epoch:  82\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2180 - accuracy: 0.9287\n",
      "training time:  12.7632 s.\n",
      "average model starts\n",
      "average model ends:  0.0355 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.798]\n",
      "global epoch:  83\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2168 - accuracy: 0.9260\n",
      "training time:  14.0469 s.\n",
      "average model starts\n",
      "average model ends:  0.0577 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.793]\n",
      "global epoch:  84\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2142 - accuracy: 0.9261\n",
      "training time:  12.1746 s.\n",
      "average model starts\n",
      "average model ends:  0.0699 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.793 0.801]\n",
      "global epoch:  85\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2309 - accuracy: 0.9237\n",
      "training time:  13.0543 s.\n",
      "average model starts\n",
      "average model ends:  0.0369 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.801 0.783]\n",
      "global epoch:  86\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2170 - accuracy: 0.9281\n",
      "training time:  13.0411 s.\n",
      "average model starts\n",
      "average model ends:  0.049 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.783 0.794]\n",
      "global epoch:  87\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2239 - accuracy: 0.9235\n",
      "training time:  13.6055 s.\n",
      "average model starts\n",
      "average model ends:  0.1075 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.794 0.793]\n",
      "global epoch:  88\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2029 - accuracy: 0.9365\n",
      "training time:  14.0614 s.\n",
      "average model starts\n",
      "average model ends:  0.0498 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.789]\n",
      "global epoch:  89\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2192 - accuracy: 0.9249\n",
      "training time:  13.9007 s.\n",
      "average model starts\n",
      "average model ends:  0.0576 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.789 0.791]\n",
      "global epoch:  90\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2132 - accuracy: 0.9268\n",
      "training time:  12.382 s.\n",
      "average model starts\n",
      "average model ends:  0.1252 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.791 0.795]\n",
      "global epoch:  91\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2149 - accuracy: 0.9277\n",
      "training time:  12.875 s.\n",
      "average model starts\n",
      "average model ends:  0.0482 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.795 0.788]\n",
      "global epoch:  92\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2221 - accuracy: 0.9278\n",
      "training time:  13.6424 s.\n",
      "average model starts\n",
      "average model ends:  0.0585 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.788 0.794]\n",
      "global epoch:  93\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2149 - accuracy: 0.9311\n",
      "training time:  12.9872 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.794 0.803]\n",
      "global epoch:  94\n",
      "virtual memory used:  65.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2104 - accuracy: 0.9325\n",
      "training time:  11.9938 s.\n",
      "average model starts\n",
      "average model ends:  0.053 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.803 0.795]\n",
      "global epoch:  95\n",
      "virtual memory used:  66.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2142 - accuracy: 0.9293\n",
      "training time:  12.4241 s.\n",
      "average model starts\n",
      "average model ends:  0.1062 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.794]\n",
      "global epoch:  96\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2103 - accuracy: 0.9318\n",
      "training time:  13.0686 s.\n",
      "average model starts\n",
      "average model ends:  0.0749 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.798]\n",
      "global epoch:  97\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2282 - accuracy: 0.9287\n",
      "training time:  14.4506 s.\n",
      "average model starts\n",
      "average model ends:  0.0547 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.781]\n",
      "global epoch:  98\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2237 - accuracy: 0.9246\n",
      "training time:  11.4215 s.\n",
      "average model starts\n",
      "average model ends:  0.0636 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.781 0.798]\n",
      "global epoch:  99\n",
      "virtual memory used:  66.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2010 - accuracy: 0.9327\n",
      "training time:  13.119 s.\n",
      "average model starts\n",
      "average model ends:  0.0623 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.798 0.78 ]\n",
      "global epoch:  100\n",
      "virtual memory used:  66.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2233 - accuracy: 0.9225\n",
      "training time:  13.0313 s.\n",
      "average model starts\n",
      "average model ends:  0.0789 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.78  0.781]\n",
      "global epoch:  101\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2379 - accuracy: 0.9250\n",
      "training time:  14.0491 s.\n",
      "average model starts\n",
      "average model ends:  0.0315 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.78  0.781 0.8  ]\n",
      "global epoch:  102\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2173 - accuracy: 0.9247\n",
      "training time:  13.5124 s.\n",
      "average model starts\n",
      "average model ends:  0.0489 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.8   0.794]\n",
      "global epoch:  103\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2166 - accuracy: 0.9298\n",
      "training time:  12.8104 s.\n",
      "average model starts\n",
      "average model ends:  0.0607 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.793]\n",
      "global epoch:  104\n",
      "virtual memory used:  66.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2121 - accuracy: 0.9315\n",
      "training time:  13.1783 s.\n",
      "average model starts\n",
      "average model ends:  0.0569 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.792]\n",
      "global epoch:  105\n",
      "virtual memory used:  67.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2197 - accuracy: 0.9266\n",
      "training time:  12.8909 s.\n",
      "average model starts\n",
      "average model ends:  0.0534 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.792 0.799]\n",
      "global epoch:  106\n",
      "virtual memory used:  67.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2113 - accuracy: 0.9284\n",
      "training time:  13.6939 s.\n",
      "average model starts\n",
      "average model ends:  0.1339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.794]\n",
      "global epoch:  107\n",
      "virtual memory used:  68.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2151 - accuracy: 0.9296\n",
      "training time:  14.0984 s.\n",
      "average model starts\n",
      "average model ends:  0.0589 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.796]\n",
      "global epoch:  108\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2102 - accuracy: 0.9298\n",
      "training time:  13.6522 s.\n",
      "average model starts\n",
      "average model ends:  0.0773 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.792]\n",
      "global epoch:  109\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 59ms/step - loss: 0.2108 - accuracy: 0.9315\n",
      "training time:  15.0702 s.\n",
      "average model starts\n",
      "average model ends:  0.112 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.792 0.795]\n",
      "global epoch:  110\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2325 - accuracy: 0.9242\n",
      "training time:  12.9134 s.\n",
      "average model starts\n",
      "average model ends:  0.1331 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.793]\n",
      "global epoch:  111\n",
      "virtual memory used:  68.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2257 - accuracy: 0.9253\n",
      "training time:  13.6599 s.\n",
      "average model starts\n",
      "average model ends:  0.0866 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.797]\n",
      "global epoch:  112\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2125 - accuracy: 0.9339\n",
      "training time:  12.765 s.\n",
      "average model starts\n",
      "average model ends:  0.1042 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.795]\n",
      "global epoch:  113\n",
      "virtual memory used:  68.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2095 - accuracy: 0.9265\n",
      "training time:  12.6038 s.\n",
      "average model starts\n",
      "average model ends:  0.0915 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.795]\n",
      "global epoch:  114\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2264 - accuracy: 0.9277\n",
      "training time:  12.4467 s.\n",
      "average model starts\n",
      "average model ends:  0.0983 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.788]\n",
      "global epoch:  115\n",
      "virtual memory used:  69.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2233 - accuracy: 0.9257\n",
      "training time:  11.7518 s.\n",
      "average model starts\n",
      "average model ends:  0.1016 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.788 0.791]\n",
      "global epoch:  116\n",
      "virtual memory used:  68.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2352 - accuracy: 0.9245\n",
      "training time:  12.935 s.\n",
      "average model starts\n",
      "average model ends:  0.0794 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.791 0.792]\n",
      "global epoch:  117\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2316 - accuracy: 0.9244\n",
      "training time:  13.9504 s.\n",
      "average model starts\n",
      "average model ends:  0.1168 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.792 0.793]\n",
      "global epoch:  118\n",
      "virtual memory used:  69.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2170 - accuracy: 0.9283\n",
      "training time:  12.951 s.\n",
      "average model starts\n",
      "average model ends:  0.1151 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.793 0.796]\n",
      "global epoch:  119\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2329 - accuracy: 0.9234\n",
      "training time:  13.0732 s.\n",
      "average model starts\n",
      "average model ends:  0.0685 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.796]\n",
      "global epoch:  120\n",
      "virtual memory used:  69.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2123 - accuracy: 0.9311\n",
      "training time:  12.7829 s.\n",
      "average model starts\n",
      "average model ends:  0.0423 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.785]\n",
      "global epoch:  121\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2228 - accuracy: 0.9268\n",
      "training time:  11.6908 s.\n",
      "average model starts\n",
      "average model ends:  0.0623 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.785 0.798]\n",
      "global epoch:  122\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2049 - accuracy: 0.9350\n",
      "training time:  13.8282 s.\n",
      "average model starts\n",
      "average model ends:  0.0891 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.798 0.792]\n",
      "global epoch:  123\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2102 - accuracy: 0.9313\n",
      "training time:  14.0658 s.\n",
      "average model starts\n",
      "average model ends:  0.0347 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.781]\n",
      "global epoch:  124\n",
      "virtual memory used:  69.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2030 - accuracy: 0.9320\n",
      "training time:  14.124 s.\n",
      "average model starts\n",
      "average model ends:  0.0725 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.781 0.79 ]\n",
      "global epoch:  125\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2300 - accuracy: 0.9230\n",
      "training time:  13.3126 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.79  0.796]\n",
      "global epoch:  126\n",
      "virtual memory used:  70.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2107 - accuracy: 0.9315\n",
      "training time:  13.1953 s.\n",
      "average model starts\n",
      "average model ends:  0.0543 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.796 0.8  ]\n",
      "global epoch:  127\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2139 - accuracy: 0.9315\n",
      "training time:  14.5424 s.\n",
      "average model starts\n",
      "average model ends:  0.1457 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.79 ]\n",
      "global epoch:  128\n",
      "virtual memory used:  69.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2253 - accuracy: 0.9291\n",
      "training time:  13.3272 s.\n",
      "average model starts\n",
      "average model ends:  0.0313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.79  0.787]\n",
      "global epoch:  129\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2321 - accuracy: 0.9226\n",
      "training time:  14.1496 s.\n",
      "average model starts\n",
      "average model ends:  0.1053 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.787 0.799]\n",
      "global epoch:  130\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 61ms/step - loss: 0.2245 - accuracy: 0.9272\n",
      "training time:  15.4687 s.\n",
      "average model starts\n",
      "average model ends:  0.0926 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.799 0.786]\n",
      "global epoch:  131\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2191 - accuracy: 0.9298\n",
      "training time:  14.153 s.\n",
      "average model starts\n",
      "average model ends:  0.1428 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.786 0.795]\n",
      "global epoch:  132\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1981 - accuracy: 0.9330\n",
      "training time:  14.025 s.\n",
      "average model starts\n",
      "average model ends:  0.045 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.795 0.798]\n",
      "global epoch:  133\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2191 - accuracy: 0.9271\n",
      "training time:  13.6946 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.789]\n",
      "global epoch:  134\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.2221 - accuracy: 0.9263\n",
      "training time:  14.6974 s.\n",
      "average model starts\n",
      "average model ends:  0.0603 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.789 0.795]\n",
      "global epoch:  135\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2264 - accuracy: 0.9260\n",
      "training time:  12.0477 s.\n",
      "average model starts\n",
      "average model ends:  0.0642 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.795 0.781]\n",
      "global epoch:  136\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2101 - accuracy: 0.9309\n",
      "training time:  12.2853 s.\n",
      "average model starts\n",
      "average model ends:  0.0697 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.781 0.796]\n",
      "global epoch:  137\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2196 - accuracy: 0.9288\n",
      "training time:  12.9257 s.\n",
      "average model starts\n",
      "average model ends:  0.0658 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.796 0.791]\n",
      "global epoch:  138\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2150 - accuracy: 0.9256\n",
      "training time:  13.6184 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.794]\n",
      "global epoch:  139\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2339 - accuracy: 0.9239\n",
      "training time:  12.9196 s.\n",
      "average model starts\n",
      "average model ends:  0.0349 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.794 0.807]\n",
      "global epoch:  140\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2154 - accuracy: 0.9272\n",
      "training time:  13.0411 s.\n",
      "average model starts\n",
      "average model ends:  0.1164 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.807 0.786]\n",
      "global epoch:  141\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2135 - accuracy: 0.9314\n",
      "training time:  12.2196 s.\n",
      "average model starts\n",
      "average model ends:  0.0507 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.786 0.791]\n",
      "global epoch:  142\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2037 - accuracy: 0.9301\n",
      "training time:  12.5528 s.\n",
      "average model starts\n",
      "average model ends:  0.0431 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.791 0.78 ]\n",
      "global epoch:  143\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2141 - accuracy: 0.9266\n",
      "training time:  14.4885 s.\n",
      "average model starts\n",
      "average model ends:  0.0583 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.78  0.792]\n",
      "global epoch:  144\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.2204 - accuracy: 0.9246\n",
      "training time:  10.5793 s.\n",
      "average model starts\n",
      "average model ends:  0.0516 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.78  0.792 0.79 ]\n",
      "global epoch:  145\n",
      "virtual memory used:  66.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2108 - accuracy: 0.9342\n",
      "training time:  12.3537 s.\n",
      "average model starts\n",
      "average model ends:  0.0505 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.79  0.791]\n",
      "global epoch:  146\n",
      "virtual memory used:  66.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1976 - accuracy: 0.9355\n",
      "training time:  13.5869 s.\n",
      "average model starts\n",
      "average model ends:  0.0482 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.791 0.802]\n",
      "global epoch:  147\n",
      "virtual memory used:  67.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2258 - accuracy: 0.9249\n",
      "training time:  12.8434 s.\n",
      "average model starts\n",
      "average model ends:  0.1219 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.802 0.784]\n",
      "global epoch:  148\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2296 - accuracy: 0.9244\n",
      "training time:  12.9719 s.\n",
      "average model starts\n",
      "average model ends:  0.0791 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.784 0.796]\n",
      "global epoch:  149\n",
      "virtual memory used:  66.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2139 - accuracy: 0.9312\n",
      "training time:  12.2198 s.\n",
      "average model starts\n",
      "average model ends:  0.0921 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.796 0.796]\n",
      "global epoch:  150\n",
      "virtual memory used:  66.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2008 - accuracy: 0.9358\n",
      "training time:  11.1997 s.\n",
      "average model starts\n",
      "average model ends:  0.0975 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.798]\n",
      "global epoch:  151\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2165 - accuracy: 0.9280\n",
      "training time:  11.8932 s.\n",
      "average model starts\n",
      "average model ends:  0.0567 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.795]\n",
      "global epoch:  152\n",
      "virtual memory used:  68.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2241 - accuracy: 0.9266\n",
      "training time:  12.7098 s.\n",
      "average model starts\n",
      "average model ends:  0.0652 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.795 0.796]\n",
      "global epoch:  153\n",
      "virtual memory used:  68.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2116 - accuracy: 0.9288\n",
      "training time:  11.1178 s.\n",
      "average model starts\n",
      "average model ends:  0.0683 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.794]\n",
      "global epoch:  154\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2277 - accuracy: 0.9267\n",
      "training time:  12.8783 s.\n",
      "average model starts\n",
      "average model ends:  0.055 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.792]\n",
      "global epoch:  155\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2283 - accuracy: 0.9261\n",
      "training time:  13.8987 s.\n",
      "average model starts\n",
      "average model ends:  0.0781 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.792 0.795]\n",
      "global epoch:  156\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1987 - accuracy: 0.9332\n",
      "training time:  13.8631 s.\n",
      "average model starts\n",
      "average model ends:  0.0764 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.788]\n",
      "global epoch:  157\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2213 - accuracy: 0.9284\n",
      "training time:  13.6485 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.788 0.796]\n",
      "global epoch:  158\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2240 - accuracy: 0.9241\n",
      "training time:  12.4361 s.\n",
      "average model starts\n",
      "average model ends:  0.122 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.792]\n",
      "global epoch:  159\n",
      "virtual memory used:  67.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2123 - accuracy: 0.9308\n",
      "training time:  12.2682 s.\n",
      "average model starts\n",
      "average model ends:  0.0651 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.792 0.799]\n",
      "global epoch:  160\n",
      "virtual memory used:  68.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2209 - accuracy: 0.9301\n",
      "training time:  13.0463 s.\n",
      "average model starts\n",
      "average model ends:  0.0722 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.795]\n",
      "global epoch:  161\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 15s 58ms/step - loss: 0.2137 - accuracy: 0.9311\n",
      "training time:  14.6937 s.\n",
      "average model starts\n",
      "average model ends:  0.0405 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.797]\n",
      "global epoch:  162\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2150 - accuracy: 0.9286\n",
      "training time:  13.5217 s.\n",
      "average model starts\n",
      "average model ends:  0.1101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.792]\n",
      "global epoch:  163\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2209 - accuracy: 0.9284\n",
      "training time:  14.2367 s.\n",
      "average model starts\n",
      "average model ends:  0.0631 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.792 0.795]\n",
      "global epoch:  164\n",
      "virtual memory used:  68.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2204 - accuracy: 0.9257\n",
      "training time:  13.3743 s.\n",
      "average model starts\n",
      "average model ends:  0.0396 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.792]\n",
      "global epoch:  165\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2102 - accuracy: 0.9332\n",
      "training time:  13.68 s.\n",
      "average model starts\n",
      "average model ends:  0.0943 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.792 0.786]\n",
      "global epoch:  166\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2274 - accuracy: 0.9231\n",
      "training time:  13.5574 s.\n",
      "average model starts\n",
      "average model ends:  0.0446 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.786 0.786]\n",
      "global epoch:  167\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2034 - accuracy: 0.9297\n",
      "training time:  12.4926 s.\n",
      "average model starts\n",
      "average model ends:  0.0656 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.786 0.779]\n",
      "global epoch:  168\n",
      "virtual memory used:  68.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2037 - accuracy: 0.9327\n",
      "training time:  13.4453 s.\n",
      "average model starts\n",
      "average model ends:  0.0369 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.779 0.794]\n",
      "global epoch:  169\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2243 - accuracy: 0.9213\n",
      "training time:  12.9884 s.\n",
      "average model starts\n",
      "average model ends:  0.1154 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.779 0.794 0.799]\n",
      "global epoch:  170\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2167 - accuracy: 0.9283\n",
      "training time:  13.7377 s.\n",
      "average model starts\n",
      "average model ends:  0.0405 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.786]\n",
      "global epoch:  171\n",
      "virtual memory used:  70.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2092 - accuracy: 0.9312\n",
      "training time:  14.1107 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.786 0.795]\n",
      "global epoch:  172\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2116 - accuracy: 0.9315\n",
      "training time:  14.3973 s.\n",
      "average model starts\n",
      "average model ends:  0.1103 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.795 0.794]\n",
      "global epoch:  173\n",
      "virtual memory used:  69.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2206 - accuracy: 0.9302\n",
      "training time:  12.9707 s.\n",
      "average model starts\n",
      "average model ends:  0.0521 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.79 ]\n",
      "global epoch:  174\n",
      "virtual memory used:  69.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2009 - accuracy: 0.9324\n",
      "training time:  12.7793 s.\n",
      "average model starts\n",
      "average model ends:  0.12 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.788]\n",
      "global epoch:  175\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2081 - accuracy: 0.9308\n",
      "training time:  11.1589 s.\n",
      "average model starts\n",
      "average model ends:  0.0421 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.788 0.79 ]\n",
      "global epoch:  176\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2094 - accuracy: 0.9304\n",
      "training time:  11.3811 s.\n",
      "average model starts\n",
      "average model ends:  0.0512 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.79  0.805]\n",
      "global epoch:  177\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2282 - accuracy: 0.9239\n",
      "training time:  11.7358 s.\n",
      "average model starts\n",
      "average model ends:  0.1155 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.805 0.798]\n",
      "global epoch:  178\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2117 - accuracy: 0.9263\n",
      "training time:  12.3192 s.\n",
      "average model starts\n",
      "average model ends:  0.031 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.798 0.791]\n",
      "global epoch:  179\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2088 - accuracy: 0.9301\n",
      "training time:  13.3603 s.\n",
      "average model starts\n",
      "average model ends:  0.1027 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.787]\n",
      "global epoch:  180\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2260 - accuracy: 0.9289\n",
      "training time:  12.6494 s.\n",
      "average model starts\n",
      "average model ends:  0.0692 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.787 0.799]\n",
      "global epoch:  181\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2153 - accuracy: 0.9306\n",
      "training time:  13.2312 s.\n",
      "average model starts\n",
      "average model ends:  0.0384 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.799 0.796]\n",
      "global epoch:  182\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2137 - accuracy: 0.9291\n",
      "training time:  11.2918 s.\n",
      "average model starts\n",
      "average model ends:  0.1045 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.799]\n",
      "global epoch:  183\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2219 - accuracy: 0.9263\n",
      "training time:  12.1423 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.797]\n",
      "global epoch:  184\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2188 - accuracy: 0.9293\n",
      "training time:  12.1849 s.\n",
      "average model starts\n",
      "average model ends:  0.0412 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.793]\n",
      "global epoch:  185\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2214 - accuracy: 0.9246\n",
      "training time:  13.4762 s.\n",
      "average model starts\n",
      "average model ends:  0.0402 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.794]\n",
      "global epoch:  186\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2202 - accuracy: 0.9301\n",
      "training time:  11.4411 s.\n",
      "average model starts\n",
      "average model ends:  0.0349 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.799]\n",
      "global epoch:  187\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2138 - accuracy: 0.9334\n",
      "training time:  12.148 s.\n",
      "average model starts\n",
      "average model ends:  0.0561 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.794]\n",
      "global epoch:  188\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2132 - accuracy: 0.9265\n",
      "training time:  11.2142 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.788]\n",
      "global epoch:  189\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2058 - accuracy: 0.9332\n",
      "training time:  11.5602 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.788 0.793]\n",
      "global epoch:  190\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2086 - accuracy: 0.9287\n",
      "training time:  11.1259 s.\n",
      "average model starts\n",
      "average model ends:  0.0539 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.786]\n",
      "global epoch:  191\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2181 - accuracy: 0.9283\n",
      "training time:  11.9327 s.\n",
      "average model starts\n",
      "average model ends:  0.0644 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.786 0.795]\n",
      "global epoch:  192\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2038 - accuracy: 0.9314\n",
      "training time:  11.8606 s.\n",
      "average model starts\n",
      "average model ends:  0.0754 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.795 0.793]\n",
      "global epoch:  193\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2167 - accuracy: 0.9287\n",
      "training time:  13.5769 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.787]\n",
      "global epoch:  194\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2213 - accuracy: 0.9270\n",
      "training time:  12.8724 s.\n",
      "average model starts\n",
      "average model ends:  0.0446 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.787 0.795]\n",
      "global epoch:  195\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2103 - accuracy: 0.9313\n",
      "training time:  12.7609 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.795 0.8  ]\n",
      "global epoch:  196\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2244 - accuracy: 0.9249\n",
      "training time:  11.9232 s.\n",
      "average model starts\n",
      "average model ends:  0.0369 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.796]\n",
      "global epoch:  197\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2210 - accuracy: 0.9287\n",
      "training time:  11.9913 s.\n",
      "average model starts\n",
      "average model ends:  0.1156 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.788]\n",
      "global epoch:  198\n",
      "virtual memory used:  72.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1999 - accuracy: 0.9325\n",
      "training time:  12.5827 s.\n",
      "average model starts\n",
      "average model ends:  0.0519 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.79 ]\n",
      "global epoch:  199\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1964 - accuracy: 0.9374\n",
      "training time:  13.3112 s.\n",
      "average model starts\n",
      "average model ends:  0.1101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.79  0.802]\n",
      "global epoch:  200\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2084 - accuracy: 0.9296\n",
      "training time:  11.9408 s.\n",
      "average model starts\n",
      "average model ends:  0.0359 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.794]\n",
      "global epoch:  201\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1991 - accuracy: 0.9314\n",
      "training time:  11.6461 s.\n",
      "average model starts\n",
      "average model ends:  0.0344 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.786]\n",
      "global epoch:  202\n",
      "virtual memory used:  66.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1925 - accuracy: 0.9354\n",
      "training time:  12.5431 s.\n",
      "average model starts\n",
      "average model ends:  0.102 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.786 0.799]\n",
      "global epoch:  203\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2071 - accuracy: 0.9324\n",
      "training time:  13.1716 s.\n",
      "average model starts\n",
      "average model ends:  0.0775 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.799 0.795]\n",
      "global epoch:  204\n",
      "virtual memory used:  67.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2294 - accuracy: 0.9257\n",
      "training time:  12.5238 s.\n",
      "average model starts\n",
      "average model ends:  0.0825 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.801]\n",
      "global epoch:  205\n",
      "virtual memory used:  66.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2049 - accuracy: 0.9301\n",
      "training time:  12.2743 s.\n",
      "average model starts\n",
      "average model ends:  0.0424 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.794]\n",
      "global epoch:  206\n",
      "virtual memory used:  67.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2194 - accuracy: 0.9278\n",
      "training time:  12.3917 s.\n",
      "average model starts\n",
      "average model ends:  0.0779 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.788]\n",
      "global epoch:  207\n",
      "virtual memory used:  68.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2053 - accuracy: 0.9334\n",
      "training time:  11.0985 s.\n",
      "average model starts\n",
      "average model ends:  0.0896 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.788 0.796]\n",
      "global epoch:  208\n",
      "virtual memory used:  68.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2179 - accuracy: 0.9278\n",
      "training time:  12.5618 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.788]\n",
      "global epoch:  209\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2281 - accuracy: 0.9219\n",
      "training time:  11.2291 s.\n",
      "average model starts\n",
      "average model ends:  0.1064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.801]\n",
      "global epoch:  210\n",
      "virtual memory used:  67.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2302 - accuracy: 0.9218\n",
      "training time:  12.6739 s.\n",
      "average model starts\n",
      "average model ends:  0.0798 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.801 0.801]\n",
      "global epoch:  211\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2169 - accuracy: 0.9288\n",
      "training time:  12.5932 s.\n",
      "average model starts\n",
      "average model ends:  0.0638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.796]\n",
      "global epoch:  212\n",
      "virtual memory used:  69.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2062 - accuracy: 0.9307\n",
      "training time:  12.8592 s.\n",
      "average model starts\n",
      "average model ends:  0.0313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.793]\n",
      "global epoch:  213\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2062 - accuracy: 0.9345\n",
      "training time:  11.2957 s.\n",
      "average model starts\n",
      "average model ends:  0.0382 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.79 ]\n",
      "global epoch:  214\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2216 - accuracy: 0.9277\n",
      "training time:  11.9792 s.\n",
      "average model starts\n",
      "average model ends:  0.0621 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.79  0.784]\n",
      "global epoch:  215\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2101 - accuracy: 0.9302\n",
      "training time:  13.3064 s.\n",
      "average model starts\n",
      "average model ends:  0.1057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.784 0.795]\n",
      "global epoch:  216\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2059 - accuracy: 0.9309\n",
      "training time:  13.4287 s.\n",
      "average model starts\n",
      "average model ends:  0.0431 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.795 0.799]\n",
      "global epoch:  217\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2096 - accuracy: 0.9291\n",
      "training time:  14.2111 s.\n",
      "average model starts\n",
      "average model ends:  0.0886 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.794]\n",
      "global epoch:  218\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2139 - accuracy: 0.9258\n",
      "training time:  12.9112 s.\n",
      "average model starts\n",
      "average model ends:  0.0342 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.79 ]\n",
      "global epoch:  219\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1877 - accuracy: 0.9379\n",
      "training time:  12.9946 s.\n",
      "average model starts\n",
      "average model ends:  0.0992 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.793]\n",
      "global epoch:  220\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2187 - accuracy: 0.9319\n",
      "training time:  13.3091 s.\n",
      "average model starts\n",
      "average model ends:  0.0566 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.793 0.788]\n",
      "global epoch:  221\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2126 - accuracy: 0.9293\n",
      "training time:  12.0721 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.788 0.796]\n",
      "global epoch:  222\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2223 - accuracy: 0.9260\n",
      "training time:  12.8046 s.\n",
      "average model starts\n",
      "average model ends:  0.0663 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.791]\n",
      "global epoch:  223\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2159 - accuracy: 0.9287\n",
      "training time:  11.3209 s.\n",
      "average model starts\n",
      "average model ends:  0.0344 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.79 ]\n",
      "global epoch:  224\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2125 - accuracy: 0.9303\n",
      "training time:  11.6987 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.79  0.796]\n",
      "global epoch:  225\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2165 - accuracy: 0.9309\n",
      "training time:  12.9637 s.\n",
      "average model starts\n",
      "average model ends:  0.0496 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.796 0.787]\n",
      "global epoch:  226\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2044 - accuracy: 0.9344\n",
      "training time:  11.0477 s.\n",
      "average model starts\n",
      "average model ends:  0.0806 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.787 0.792]\n",
      "global epoch:  227\n",
      "virtual memory used:  69.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2144 - accuracy: 0.9330\n",
      "training time:  12.2223 s.\n",
      "average model starts\n",
      "average model ends:  0.0338 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.792 0.792]\n",
      "global epoch:  228\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2091 - accuracy: 0.9315\n",
      "training time:  13.8551 s.\n",
      "average model starts\n",
      "average model ends:  0.0848 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.789]\n",
      "global epoch:  229\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2214 - accuracy: 0.9317\n",
      "training time:  12.836 s.\n",
      "average model starts\n",
      "average model ends:  0.0317 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.789 0.797]\n",
      "global epoch:  230\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2142 - accuracy: 0.9271\n",
      "training time:  12.353 s.\n",
      "average model starts\n",
      "average model ends:  0.1085 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.797 0.797]\n",
      "global epoch:  231\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2094 - accuracy: 0.9313\n",
      "training time:  11.6557 s.\n",
      "average model starts\n",
      "average model ends:  0.06 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.791]\n",
      "global epoch:  232\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2100 - accuracy: 0.9291\n",
      "training time:  12.8817 s.\n",
      "average model starts\n",
      "average model ends:  0.098 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.796]\n",
      "global epoch:  233\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2133 - accuracy: 0.9314\n",
      "training time:  13.02 s.\n",
      "average model starts\n",
      "average model ends:  0.1038 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.783]\n",
      "global epoch:  234\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2181 - accuracy: 0.9311\n",
      "training time:  13.5883 s.\n",
      "average model starts\n",
      "average model ends:  0.0484 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.783 0.784]\n",
      "global epoch:  235\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2148 - accuracy: 0.9267\n",
      "training time:  11.9709 s.\n",
      "average model starts\n",
      "average model ends:  0.065 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.784 0.785]\n",
      "global epoch:  236\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2026 - accuracy: 0.9315\n",
      "training time:  13.5734 s.\n",
      "average model starts\n",
      "average model ends:  0.0569 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.785 0.793]\n",
      "global epoch:  237\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2186 - accuracy: 0.9291\n",
      "training time:  12.81 s.\n",
      "average model starts\n",
      "average model ends:  0.05 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.793 0.795]\n",
      "global epoch:  238\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2238 - accuracy: 0.9265\n",
      "training time:  13.1898 s.\n",
      "average model starts\n",
      "average model ends:  0.0351 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.801]\n",
      "global epoch:  239\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2060 - accuracy: 0.9315\n",
      "training time:  11.7843 s.\n",
      "average model starts\n",
      "average model ends:  0.0697 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.796]\n",
      "global epoch:  240\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1992 - accuracy: 0.9345\n",
      "training time:  11.6853 s.\n",
      "average model starts\n",
      "average model ends:  0.0528 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.799]\n",
      "global epoch:  241\n",
      "virtual memory used:  72.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2125 - accuracy: 0.9289\n",
      "training time:  11.6578 s.\n",
      "average model starts\n",
      "average model ends:  0.1372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.797]\n",
      "global epoch:  242\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2130 - accuracy: 0.9294\n",
      "training time:  11.9958 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.801]\n",
      "global epoch:  243\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2237 - accuracy: 0.9250\n",
      "training time:  13.5401 s.\n",
      "average model starts\n",
      "average model ends:  0.0646 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.796]\n",
      "global epoch:  244\n",
      "virtual memory used:  74.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2046 - accuracy: 0.9297\n",
      "training time:  11.4299 s.\n",
      "average model starts\n",
      "average model ends:  0.0822 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.797]\n",
      "global epoch:  245\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2059 - accuracy: 0.9312\n",
      "training time:  13.7621 s.\n",
      "average model starts\n",
      "average model ends:  0.0446 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.788]\n",
      "global epoch:  246\n",
      "virtual memory used:  72.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2088 - accuracy: 0.9325\n",
      "training time:  13.3659 s.\n",
      "average model starts\n",
      "average model ends:  0.1018 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.788 0.797]\n",
      "global epoch:  247\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2169 - accuracy: 0.9303\n",
      "training time:  11.5817 s.\n",
      "average model starts\n",
      "average model ends:  0.0527 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.797 0.785]\n",
      "global epoch:  248\n",
      "virtual memory used:  73.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2005 - accuracy: 0.9333\n",
      "training time:  11.4654 s.\n",
      "average model starts\n",
      "average model ends:  0.034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.785 0.794]\n",
      "global epoch:  249\n",
      "virtual memory used:  74.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.2107 - accuracy: 0.9282\n",
      "training time:  10.7315 s.\n",
      "average model starts\n",
      "average model ends:  0.0576 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.794 0.785]\n",
      "global epoch:  250\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2125 - accuracy: 0.9291\n",
      "training time:  14.214 s.\n",
      "average model starts\n",
      "average model ends:  0.049 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.785 0.794]\n",
      "global epoch:  251\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2149 - accuracy: 0.9332\n",
      "training time:  11.658 s.\n",
      "average model starts\n",
      "average model ends:  0.0494 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.794 0.788]\n",
      "global epoch:  252\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2127 - accuracy: 0.9281\n",
      "training time:  11.9572 s.\n",
      "average model starts\n",
      "average model ends:  0.1327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.788 0.796]\n",
      "global epoch:  253\n",
      "virtual memory used:  67.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2104 - accuracy: 0.9277\n",
      "training time:  12.6782 s.\n",
      "average model starts\n",
      "average model ends:  0.0583 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.8  ]\n",
      "global epoch:  254\n",
      "virtual memory used:  67.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2061 - accuracy: 0.9327\n",
      "training time:  12.4384 s.\n",
      "average model starts\n",
      "average model ends:  0.0572 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.798]\n",
      "global epoch:  255\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2126 - accuracy: 0.9293\n",
      "training time:  12.3385 s.\n",
      "average model starts\n",
      "average model ends:  0.0969 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.797]\n",
      "global epoch:  256\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2165 - accuracy: 0.9280\n",
      "training time:  12.698 s.\n",
      "average model starts\n",
      "average model ends:  0.0601 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.789]\n",
      "global epoch:  257\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2172 - accuracy: 0.9298\n",
      "training time:  13.094 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.789 0.788]\n",
      "global epoch:  258\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1949 - accuracy: 0.9359\n",
      "training time:  11.4993 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.788 0.798]\n",
      "global epoch:  259\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2007 - accuracy: 0.9343\n",
      "training time:  11.118 s.\n",
      "average model starts\n",
      "average model ends:  0.098 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.798 0.795]\n",
      "global epoch:  260\n",
      "virtual memory used:  68.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.2176 - accuracy: 0.9308\n",
      "training time:  10.7324 s.\n",
      "average model starts\n",
      "average model ends:  0.0503 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.795 0.793]\n",
      "global epoch:  261\n",
      "virtual memory used:  69.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2029 - accuracy: 0.9332\n",
      "training time:  11.1148 s.\n",
      "average model starts\n",
      "average model ends:  0.0761 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.798]\n",
      "global epoch:  262\n",
      "virtual memory used:  68.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 40ms/step - loss: 0.2158 - accuracy: 0.9268\n",
      "training time:  10.3598 s.\n",
      "average model starts\n",
      "average model ends:  0.0393 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.794]\n",
      "global epoch:  263\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.2097 - accuracy: 0.9298\n",
      "training time:  10.4244 s.\n",
      "average model starts\n",
      "average model ends:  0.0778 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.794 0.792]\n",
      "global epoch:  264\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2147 - accuracy: 0.9258\n",
      "training time:  12.5688 s.\n",
      "average model starts\n",
      "average model ends:  0.0419 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.792 0.79 ]\n",
      "global epoch:  265\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1924 - accuracy: 0.9354\n",
      "training time:  13.1205 s.\n",
      "average model starts\n",
      "average model ends:  0.1033 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.79  0.785]\n",
      "global epoch:  266\n",
      "virtual memory used:  68.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2105 - accuracy: 0.9328\n",
      "training time:  13.4 s.\n",
      "average model starts\n",
      "average model ends:  0.0778 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.785 0.798]\n",
      "global epoch:  267\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2050 - accuracy: 0.9329\n",
      "training time:  12.5578 s.\n",
      "average model starts\n",
      "average model ends:  0.0686 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.798 0.786]\n",
      "global epoch:  268\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2090 - accuracy: 0.9314\n",
      "training time:  11.8702 s.\n",
      "average model starts\n",
      "average model ends:  0.0349 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.786 0.79 ]\n",
      "global epoch:  269\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2176 - accuracy: 0.9307\n",
      "training time:  13.5555 s.\n",
      "average model starts\n",
      "average model ends:  0.0972 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.79  0.8  ]\n",
      "global epoch:  270\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2087 - accuracy: 0.9298\n",
      "training time:  11.6712 s.\n",
      "average model starts\n",
      "average model ends:  0.0379 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.8   0.798]\n",
      "global epoch:  271\n",
      "virtual memory used:  70.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2126 - accuracy: 0.9304\n",
      "training time:  12.3967 s.\n",
      "average model starts\n",
      "average model ends:  0.04 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.803]\n",
      "global epoch:  272\n",
      "virtual memory used:  70.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2049 - accuracy: 0.9337\n",
      "training time:  12.4173 s.\n",
      "average model starts\n",
      "average model ends:  0.1086 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.803 0.802]\n",
      "global epoch:  273\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2058 - accuracy: 0.9318\n",
      "training time:  11.8024 s.\n",
      "average model starts\n",
      "average model ends:  0.0987 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.802]\n",
      "global epoch:  274\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2046 - accuracy: 0.9320\n",
      "training time:  11.6998 s.\n",
      "average model starts\n",
      "average model ends:  0.0764 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.795]\n",
      "global epoch:  275\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2154 - accuracy: 0.9297\n",
      "training time:  13.2951 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.792]\n",
      "global epoch:  276\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2220 - accuracy: 0.9284\n",
      "training time:  12.0483 s.\n",
      "average model starts\n",
      "average model ends:  0.0395 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.792 0.778]\n",
      "global epoch:  277\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2019 - accuracy: 0.9334\n",
      "training time:  11.5811 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.778 0.797]\n",
      "global epoch:  278\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1963 - accuracy: 0.9337\n",
      "training time:  12.5956 s.\n",
      "average model starts\n",
      "average model ends:  0.0913 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.778 0.797 0.783]\n",
      "global epoch:  279\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2232 - accuracy: 0.9298\n",
      "training time:  12.7883 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.783 0.794]\n",
      "global epoch:  280\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2033 - accuracy: 0.9313\n",
      "training time:  12.7089 s.\n",
      "average model starts\n",
      "average model ends:  0.0326 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.794 0.794]\n",
      "global epoch:  281\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2136 - accuracy: 0.9311\n",
      "training time:  12.3327 s.\n",
      "average model starts\n",
      "average model ends:  0.061 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.798]\n",
      "global epoch:  282\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2120 - accuracy: 0.9282\n",
      "training time:  11.5852 s.\n",
      "average model starts\n",
      "average model ends:  0.0695 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.791]\n",
      "global epoch:  283\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.2111 - accuracy: 0.9294\n",
      "training time:  13.7794 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.783]\n",
      "global epoch:  284\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1956 - accuracy: 0.9369\n",
      "training time:  13.9382 s.\n",
      "average model starts\n",
      "average model ends:  0.0672 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.783 0.797]\n",
      "global epoch:  285\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1944 - accuracy: 0.9366\n",
      "training time:  12.6821 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.797 0.793]\n",
      "global epoch:  286\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1860 - accuracy: 0.9359\n",
      "training time:  13.9895 s.\n",
      "average model starts\n",
      "average model ends:  0.0332 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.794]\n",
      "global epoch:  287\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2147 - accuracy: 0.9276\n",
      "training time:  12.9179 s.\n",
      "average model starts\n",
      "average model ends:  0.0619 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.8  ]\n",
      "global epoch:  288\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1988 - accuracy: 0.9353\n",
      "training time:  12.1967 s.\n",
      "average model starts\n",
      "average model ends:  0.0574 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.797]\n",
      "global epoch:  289\n",
      "virtual memory used:  71.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2029 - accuracy: 0.9296\n",
      "training time:  12.2887 s.\n",
      "average model starts\n",
      "average model ends:  0.0618 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.798]\n",
      "global epoch:  290\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2062 - accuracy: 0.9319\n",
      "training time:  13.3744 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.792]\n",
      "global epoch:  291\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2148 - accuracy: 0.9293\n",
      "training time:  12.3811 s.\n",
      "average model starts\n",
      "average model ends:  0.0736 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.8  ]\n",
      "global epoch:  292\n",
      "virtual memory used:  73.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2186 - accuracy: 0.9229\n",
      "training time:  11.1493 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.8   0.794]\n",
      "global epoch:  293\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2031 - accuracy: 0.9356\n",
      "training time:  12.5002 s.\n",
      "average model starts\n",
      "average model ends:  0.104 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.801]\n",
      "global epoch:  294\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.2064 - accuracy: 0.9298\n",
      "training time:  10.8067 s.\n",
      "average model starts\n",
      "average model ends:  0.0329 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.786]\n",
      "global epoch:  295\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1991 - accuracy: 0.9332\n",
      "training time:  11.6274 s.\n",
      "average model starts\n",
      "average model ends:  0.0414 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.786 0.796]\n",
      "global epoch:  296\n",
      "virtual memory used:  73.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2219 - accuracy: 0.9273\n",
      "training time:  11.2731 s.\n",
      "average model starts\n",
      "average model ends:  0.0438 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.796 0.796]\n",
      "global epoch:  297\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2191 - accuracy: 0.9311\n",
      "training time:  13.3715 s.\n",
      "average model starts\n",
      "average model ends:  0.0333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.791]\n",
      "global epoch:  298\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2081 - accuracy: 0.9289\n",
      "training time:  12.7224 s.\n",
      "average model starts\n",
      "average model ends:  0.0759 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.792]\n",
      "global epoch:  299\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1968 - accuracy: 0.9325\n",
      "training time:  12.1979 s.\n",
      "average model starts\n",
      "average model ends:  0.0782 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.792 0.802]\n",
      "global epoch:  300\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2212 - accuracy: 0.9275\n",
      "training time:  12.1348 s.\n",
      "average model starts\n",
      "average model ends:  0.0905 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.802 0.799]\n",
      "global epoch:  301\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2045 - accuracy: 0.9327\n",
      "training time:  11.9892 s.\n",
      "average model starts\n",
      "average model ends:  0.0744 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.8  ]\n",
      "global epoch:  302\n",
      "virtual memory used:  75.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2077 - accuracy: 0.9297\n",
      "training time:  12.9101 s.\n",
      "average model starts\n",
      "average model ends:  0.0953 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.796]\n",
      "global epoch:  303\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.2325 - accuracy: 0.9237\n",
      "training time:  10.4833 s.\n",
      "average model starts\n",
      "average model ends:  0.0488 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.791]\n",
      "global epoch:  304\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 40ms/step - loss: 0.2088 - accuracy: 0.9289\n",
      "training time:  10.168 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.789]\n",
      "global epoch:  305\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2160 - accuracy: 0.9262\n",
      "training time:  11.4346 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.789 0.792]\n",
      "global epoch:  306\n",
      "virtual memory used:  75.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2119 - accuracy: 0.9296\n",
      "training time:  13.0403 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.792 0.805]\n",
      "global epoch:  307\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2069 - accuracy: 0.9309\n",
      "training time:  12.9645 s.\n",
      "average model starts\n",
      "average model ends:  0.0694 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.805 0.788]\n",
      "global epoch:  308\n",
      "virtual memory used:  67.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2202 - accuracy: 0.9265\n",
      "training time:  12.2161 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.788 0.793]\n",
      "global epoch:  309\n",
      "virtual memory used:  68.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2088 - accuracy: 0.9297\n",
      "training time:  12.5823 s.\n",
      "average model starts\n",
      "average model ends:  0.0398 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.798]\n",
      "global epoch:  310\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1828 - accuracy: 0.9390\n",
      "training time:  12.9653 s.\n",
      "average model starts\n",
      "average model ends:  0.115 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.785]\n",
      "global epoch:  311\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2120 - accuracy: 0.9286\n",
      "training time:  13.077 s.\n",
      "average model starts\n",
      "average model ends:  0.0374 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.785 0.779]\n",
      "global epoch:  312\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2156 - accuracy: 0.9289\n",
      "training time:  12.0287 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.779 0.79 ]\n",
      "global epoch:  313\n",
      "virtual memory used:  69.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2164 - accuracy: 0.9273\n",
      "training time:  12.1393 s.\n",
      "average model starts\n",
      "average model ends:  0.1225 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.779 0.79  0.787]\n",
      "global epoch:  314\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1989 - accuracy: 0.9348\n",
      "training time:  11.0831 s.\n",
      "average model starts\n",
      "average model ends:  0.0389 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.787 0.788]\n",
      "global epoch:  315\n",
      "virtual memory used:  69.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2065 - accuracy: 0.9332\n",
      "training time:  11.4109 s.\n",
      "average model starts\n",
      "average model ends:  0.0593 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.788 0.799]\n",
      "global epoch:  316\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2053 - accuracy: 0.9327\n",
      "training time:  13.1592 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.799 0.793]\n",
      "global epoch:  317\n",
      "virtual memory used:  70.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1988 - accuracy: 0.9351\n",
      "training time:  11.5748 s.\n",
      "average model starts\n",
      "average model ends:  0.055 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.796]\n",
      "global epoch:  318\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2027 - accuracy: 0.9311\n",
      "training time:  10.9127 s.\n",
      "average model starts\n",
      "average model ends:  0.05 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.796]\n",
      "global epoch:  319\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2083 - accuracy: 0.9309\n",
      "training time:  12.2444 s.\n",
      "average model starts\n",
      "average model ends:  0.0666 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.784]\n",
      "global epoch:  320\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2108 - accuracy: 0.9320\n",
      "training time:  11.472 s.\n",
      "average model starts\n",
      "average model ends:  0.0827 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.784 0.797]\n",
      "global epoch:  321\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2018 - accuracy: 0.9356\n",
      "training time:  13.2052 s.\n",
      "average model starts\n",
      "average model ends:  0.1137 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.797 0.791]\n",
      "global epoch:  322\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2024 - accuracy: 0.9332\n",
      "training time:  13.1686 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.791]\n",
      "global epoch:  323\n",
      "virtual memory used:  70.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1977 - accuracy: 0.9350\n",
      "training time:  11.9679 s.\n",
      "average model starts\n",
      "average model ends:  0.047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.791 0.799]\n",
      "global epoch:  324\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2093 - accuracy: 0.9315\n",
      "training time:  12.411 s.\n",
      "average model starts\n",
      "average model ends:  0.0843 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.789]\n",
      "global epoch:  325\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2110 - accuracy: 0.9325\n",
      "training time:  12.1542 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.789 0.796]\n",
      "global epoch:  326\n",
      "virtual memory used:  72.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2175 - accuracy: 0.9319\n",
      "training time:  11.8999 s.\n",
      "average model starts\n",
      "average model ends:  0.0716 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.796]\n",
      "global epoch:  327\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2058 - accuracy: 0.9302\n",
      "training time:  11.8454 s.\n",
      "average model starts\n",
      "average model ends:  0.0889 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.798]\n",
      "global epoch:  328\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2046 - accuracy: 0.9328\n",
      "training time:  12.2471 s.\n",
      "average model starts\n",
      "average model ends:  0.0824 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.799]\n",
      "global epoch:  329\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2153 - accuracy: 0.9282\n",
      "training time:  12.1669 s.\n",
      "average model starts\n",
      "average model ends:  0.0876 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.794]\n",
      "global epoch:  330\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1999 - accuracy: 0.9343\n",
      "training time:  13.359 s.\n",
      "average model starts\n",
      "average model ends:  0.1352 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.785]\n",
      "global epoch:  331\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2033 - accuracy: 0.9325\n",
      "training time:  12.9187 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.785 0.791]\n",
      "global epoch:  332\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2018 - accuracy: 0.9350\n",
      "training time:  12.7253 s.\n",
      "average model starts\n",
      "average model ends:  0.0621 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.791 0.804]\n",
      "global epoch:  333\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2014 - accuracy: 0.9337\n",
      "training time:  12.9586 s.\n",
      "average model starts\n",
      "average model ends:  0.0898 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.804 0.798]\n",
      "global epoch:  334\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1945 - accuracy: 0.9381\n",
      "training time:  11.0503 s.\n",
      "average model starts\n",
      "average model ends:  0.0483 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.798 0.799]\n",
      "global epoch:  335\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1958 - accuracy: 0.9365\n",
      "training time:  11.5083 s.\n",
      "average model starts\n",
      "average model ends:  0.077 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.794]\n",
      "global epoch:  336\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2207 - accuracy: 0.9236\n",
      "training time:  12.1232 s.\n",
      "average model starts\n",
      "average model ends:  0.0324 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.79 ]\n",
      "global epoch:  337\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1833 - accuracy: 0.9356\n",
      "training time:  13.0694 s.\n",
      "average model starts\n",
      "average model ends:  0.0428 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.8  ]\n",
      "global epoch:  338\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1958 - accuracy: 0.9363\n",
      "training time:  13.5153 s.\n",
      "average model starts\n",
      "average model ends:  0.0951 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.8   0.793]\n",
      "global epoch:  339\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2066 - accuracy: 0.9320\n",
      "training time:  12.7288 s.\n",
      "average model starts\n",
      "average model ends:  0.0529 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.79 ]\n",
      "global epoch:  340\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2003 - accuracy: 0.9334\n",
      "training time:  12.0533 s.\n",
      "average model starts\n",
      "average model ends:  0.0766 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.79  0.79 ]\n",
      "global epoch:  341\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2057 - accuracy: 0.9311\n",
      "training time:  12.4609 s.\n",
      "average model starts\n",
      "average model ends:  0.0883 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.79  0.805]\n",
      "global epoch:  342\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1913 - accuracy: 0.9358\n",
      "training time:  12.7921 s.\n",
      "average model starts\n",
      "average model ends:  0.1179 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.805 0.798]\n",
      "global epoch:  343\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2015 - accuracy: 0.9368\n",
      "training time:  12.4425 s.\n",
      "average model starts\n",
      "average model ends:  0.0306 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.798 0.797]\n",
      "global epoch:  344\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1946 - accuracy: 0.9364\n",
      "training time:  12.868 s.\n",
      "average model starts\n",
      "average model ends:  0.0459 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.791]\n",
      "global epoch:  345\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1967 - accuracy: 0.9348\n",
      "training time:  11.822 s.\n",
      "average model starts\n",
      "average model ends:  0.0445 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.788]\n",
      "global epoch:  346\n",
      "virtual memory used:  73.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1972 - accuracy: 0.9333\n",
      "training time:  12.301 s.\n",
      "average model starts\n",
      "average model ends:  0.0437 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.788 0.794]\n",
      "global epoch:  347\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2168 - accuracy: 0.9275\n",
      "training time:  12.9132 s.\n",
      "average model starts\n",
      "average model ends:  0.1532 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.794 0.79 ]\n",
      "global epoch:  348\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2182 - accuracy: 0.9292\n",
      "training time:  13.5487 s.\n",
      "average model starts\n",
      "average model ends:  0.0331 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.798]\n",
      "global epoch:  349\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2009 - accuracy: 0.9349\n",
      "training time:  12.7875 s.\n",
      "average model starts\n",
      "average model ends:  0.0937 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.798 0.797]\n",
      "global epoch:  350\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2076 - accuracy: 0.9338\n",
      "training time:  12.1498 s.\n",
      "average model starts\n",
      "average model ends:  0.044 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.788]\n",
      "global epoch:  351\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1945 - accuracy: 0.9356\n",
      "training time:  11.8629 s.\n",
      "average model starts\n",
      "average model ends:  0.0338 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.788 0.797]\n",
      "global epoch:  352\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2136 - accuracy: 0.9301\n",
      "training time:  12.4656 s.\n",
      "average model starts\n",
      "average model ends:  0.0479 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.797 0.799]\n",
      "global epoch:  353\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2046 - accuracy: 0.9315\n",
      "training time:  14.1478 s.\n",
      "average model starts\n",
      "average model ends:  0.146 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.793]\n",
      "global epoch:  354\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2101 - accuracy: 0.9287\n",
      "training time:  12.3383 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.794]\n",
      "global epoch:  355\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1930 - accuracy: 0.9387\n",
      "training time:  11.4386 s.\n",
      "average model starts\n",
      "average model ends:  0.0352 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.802]\n",
      "global epoch:  356\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1876 - accuracy: 0.9376\n",
      "training time:  12.6157 s.\n",
      "average model starts\n",
      "average model ends:  0.0914 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.802 0.794]\n",
      "global epoch:  357\n",
      "virtual memory used:  68.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.2107 - accuracy: 0.9311\n",
      "training time:  10.5349 s.\n",
      "average model starts\n",
      "average model ends:  0.0691 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.794]\n",
      "global epoch:  358\n",
      "virtual memory used:  69.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2084 - accuracy: 0.9307\n",
      "training time:  11.3235 s.\n",
      "average model starts\n",
      "average model ends:  0.0661 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.797]\n",
      "global epoch:  359\n",
      "virtual memory used:  69.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2030 - accuracy: 0.9358\n",
      "training time:  11.1294 s.\n",
      "average model starts\n",
      "average model ends:  0.0989 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.791]\n",
      "global epoch:  360\n",
      "virtual memory used:  68.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2083 - accuracy: 0.9319\n",
      "training time:  12.6881 s.\n",
      "average model starts\n",
      "average model ends:  0.0666 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.803]\n",
      "global epoch:  361\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2005 - accuracy: 0.9328\n",
      "training time:  12.7643 s.\n",
      "average model starts\n",
      "average model ends:  0.0806 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.803 0.798]\n",
      "global epoch:  362\n",
      "virtual memory used:  69.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1941 - accuracy: 0.9377\n",
      "training time:  11.7434 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.795]\n",
      "global epoch:  363\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1877 - accuracy: 0.9401\n",
      "training time:  12.3078 s.\n",
      "average model starts\n",
      "average model ends:  0.0595 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.795 0.8  ]\n",
      "global epoch:  364\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2042 - accuracy: 0.9319\n",
      "training time:  11.0699 s.\n",
      "average model starts\n",
      "average model ends:  0.0348 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.802]\n",
      "global epoch:  365\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1978 - accuracy: 0.9349\n",
      "training time:  13.1729 s.\n",
      "average model starts\n",
      "average model ends:  0.0623 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.796]\n",
      "global epoch:  366\n",
      "virtual memory used:  70.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2081 - accuracy: 0.9284\n",
      "training time:  11.3877 s.\n",
      "average model starts\n",
      "average model ends:  0.0462 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.79 ]\n",
      "global epoch:  367\n",
      "virtual memory used:  70.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.2017 - accuracy: 0.9320\n",
      "training time:  11.6186 s.\n",
      "average model starts\n",
      "average model ends:  0.1626 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.79  0.804]\n",
      "global epoch:  368\n",
      "virtual memory used:  70.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1939 - accuracy: 0.9360\n",
      "training time:  11.4873 s.\n",
      "average model starts\n",
      "average model ends:  0.0412 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.804 0.794]\n",
      "global epoch:  369\n",
      "virtual memory used:  70.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.2013 - accuracy: 0.9344\n",
      "training time:  12.6176 s.\n",
      "average model starts\n",
      "average model ends:  0.0536 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.794 0.796]\n",
      "global epoch:  370\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1956 - accuracy: 0.9349\n",
      "training time:  12.5046 s.\n",
      "average model starts\n",
      "average model ends:  0.0646 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.791]\n",
      "global epoch:  371\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2061 - accuracy: 0.9343\n",
      "training time:  12.6504 s.\n",
      "average model starts\n",
      "average model ends:  0.0373 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.801]\n",
      "global epoch:  372\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1937 - accuracy: 0.9346\n",
      "training time:  11.7384 s.\n",
      "average model starts\n",
      "average model ends:  0.0345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.801 0.8  ]\n",
      "global epoch:  373\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1951 - accuracy: 0.9325\n",
      "training time:  13.1106 s.\n",
      "average model starts\n",
      "average model ends:  0.1393 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.799]\n",
      "global epoch:  374\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1975 - accuracy: 0.9342\n",
      "training time:  12.9043 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.799 0.793]\n",
      "global epoch:  375\n",
      "virtual memory used:  71.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.2030 - accuracy: 0.9323\n",
      "training time:  11.5931 s.\n",
      "average model starts\n",
      "average model ends:  0.0804 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.795]\n",
      "global epoch:  376\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2024 - accuracy: 0.9337\n",
      "training time:  12.1575 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.794]\n",
      "global epoch:  377\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2006 - accuracy: 0.9349\n",
      "training time:  12.9486 s.\n",
      "average model starts\n",
      "average model ends:  0.1076 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.797]\n",
      "global epoch:  378\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1933 - accuracy: 0.9381\n",
      "training time:  11.977 s.\n",
      "average model starts\n",
      "average model ends:  0.1146 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.785]\n",
      "global epoch:  379\n",
      "virtual memory used:  71.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.2042 - accuracy: 0.9342\n",
      "training time:  12.6144 s.\n",
      "average model starts\n",
      "average model ends:  0.0554 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.785 0.791]\n",
      "global epoch:  380\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2040 - accuracy: 0.9328\n",
      "training time:  12.9106 s.\n",
      "average model starts\n",
      "average model ends:  0.0763 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.791 0.791]\n",
      "global epoch:  381\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.2126 - accuracy: 0.9306\n",
      "training time:  11.6171 s.\n",
      "average model starts\n",
      "average model ends:  0.1488 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.791 0.796]\n",
      "global epoch:  382\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2003 - accuracy: 0.9348\n",
      "training time:  12.0794 s.\n",
      "average model starts\n",
      "average model ends:  0.0851 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.799]\n",
      "global epoch:  383\n",
      "virtual memory used:  72.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1961 - accuracy: 0.9332\n",
      "training time:  13.8004 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.799]\n",
      "global epoch:  384\n",
      "virtual memory used:  73.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1948 - accuracy: 0.9342\n",
      "training time:  13.1625 s.\n",
      "average model starts\n",
      "average model ends:  0.0573 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.793]\n",
      "global epoch:  385\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1892 - accuracy: 0.9382\n",
      "training time:  12.202 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.796]\n",
      "global epoch:  386\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1883 - accuracy: 0.9384\n",
      "training time:  13.4302 s.\n",
      "average model starts\n",
      "average model ends:  0.0549 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.788]\n",
      "global epoch:  387\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2221 - accuracy: 0.9303\n",
      "training time:  11.9604 s.\n",
      "average model starts\n",
      "average model ends:  0.0687 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.798]\n",
      "global epoch:  388\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.2023 - accuracy: 0.9343\n",
      "training time:  11.6132 s.\n",
      "average model starts\n",
      "average model ends:  0.0714 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.798 0.788]\n",
      "global epoch:  389\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2033 - accuracy: 0.9366\n",
      "training time:  14.1165 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.788 0.785]\n",
      "global epoch:  390\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1858 - accuracy: 0.9413\n",
      "training time:  12.0822 s.\n",
      "average model starts\n",
      "average model ends:  0.1139 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.785 0.791]\n",
      "global epoch:  391\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1995 - accuracy: 0.9338\n",
      "training time:  11.2072 s.\n",
      "average model starts\n",
      "average model ends:  0.0378 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.791 0.802]\n",
      "global epoch:  392\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.2009 - accuracy: 0.9354\n",
      "training time:  10.8223 s.\n",
      "average model starts\n",
      "average model ends:  0.0792 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.802 0.797]\n",
      "global epoch:  393\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2043 - accuracy: 0.9314\n",
      "training time:  11.3249 s.\n",
      "average model starts\n",
      "average model ends:  0.0874 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.795]\n",
      "global epoch:  394\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1936 - accuracy: 0.9384\n",
      "training time:  12.6639 s.\n",
      "average model starts\n",
      "average model ends:  0.0749 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.802]\n",
      "global epoch:  395\n",
      "virtual memory used:  75.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1978 - accuracy: 0.9340\n",
      "training time:  12.0751 s.\n",
      "average model starts\n",
      "average model ends:  0.0507 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.796]\n",
      "global epoch:  396\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2101 - accuracy: 0.9338\n",
      "training time:  12.7052 s.\n",
      "average model starts\n",
      "average model ends:  0.0762 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.798]\n",
      "global epoch:  397\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2171 - accuracy: 0.9291\n",
      "training time:  12.2632 s.\n",
      "average model starts\n",
      "average model ends:  0.0437 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.794]\n",
      "global epoch:  398\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1950 - accuracy: 0.9350\n",
      "training time:  12.2424 s.\n",
      "average model starts\n",
      "average model ends:  0.1164 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.794 0.798]\n",
      "global epoch:  399\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1986 - accuracy: 0.9339\n",
      "training time:  11.3124 s.\n",
      "average model starts\n",
      "average model ends:  0.0611 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.789]\n",
      "global epoch:  400\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2043 - accuracy: 0.9322\n",
      "training time:  12.0567 s.\n",
      "average model starts\n",
      "average model ends:  0.0802 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.789 0.796]\n",
      "global epoch:  401\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2040 - accuracy: 0.9356\n",
      "training time:  11.3444 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.798]\n",
      "global epoch:  402\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2072 - accuracy: 0.9349\n",
      "training time:  11.5476 s.\n",
      "average model starts\n",
      "average model ends:  0.0584 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.791]\n",
      "global epoch:  403\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2007 - accuracy: 0.9361\n",
      "training time:  14.3423 s.\n",
      "average model starts\n",
      "average model ends:  0.0534 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.802]\n",
      "global epoch:  404\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1993 - accuracy: 0.9366\n",
      "training time:  13.0368 s.\n",
      "average model starts\n",
      "average model ends:  0.085 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.802 0.79 ]\n",
      "global epoch:  405\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1978 - accuracy: 0.9359\n",
      "training time:  12.9959 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.79  0.794]\n",
      "global epoch:  406\n",
      "virtual memory used:  75.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1949 - accuracy: 0.9351\n",
      "training time:  14.6068 s.\n",
      "average model starts\n",
      "average model ends:  0.047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.794 0.793]\n",
      "global epoch:  407\n",
      "virtual memory used:  76.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2116 - accuracy: 0.9323\n",
      "training time:  12.1482 s.\n",
      "average model starts\n",
      "average model ends:  0.0384 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.802]\n",
      "global epoch:  408\n",
      "virtual memory used:  68.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1946 - accuracy: 0.9374\n",
      "training time:  11.0527 s.\n",
      "average model starts\n",
      "average model ends:  0.0349 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.802 0.804]\n",
      "global epoch:  409\n",
      "virtual memory used:  69.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1944 - accuracy: 0.9361\n",
      "training time:  11.6665 s.\n",
      "average model starts\n",
      "average model ends:  0.0862 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.804 0.802]\n",
      "global epoch:  410\n",
      "virtual memory used:  70.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2061 - accuracy: 0.9298\n",
      "training time:  12.4029 s.\n",
      "average model starts\n",
      "average model ends:  0.074 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.802 0.786]\n",
      "global epoch:  411\n",
      "virtual memory used:  70.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1971 - accuracy: 0.9366\n",
      "training time:  13.0699 s.\n",
      "average model starts\n",
      "average model ends:  0.0598 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.786 0.804]\n",
      "global epoch:  412\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1932 - accuracy: 0.9351\n",
      "training time:  13.8088 s.\n",
      "average model starts\n",
      "average model ends:  0.0375 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.804 0.792]\n",
      "global epoch:  413\n",
      "virtual memory used:  69.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1969 - accuracy: 0.9344\n",
      "training time:  12.97 s.\n",
      "average model starts\n",
      "average model ends:  0.068 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.792 0.795]\n",
      "global epoch:  414\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2062 - accuracy: 0.9350\n",
      "training time:  10.8516 s.\n",
      "average model starts\n",
      "average model ends:  0.0501 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.796]\n",
      "global epoch:  415\n",
      "virtual memory used:  70.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2048 - accuracy: 0.9340\n",
      "training time:  12.4066 s.\n",
      "average model starts\n",
      "average model ends:  0.0546 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.802]\n",
      "global epoch:  416\n",
      "virtual memory used:  71.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1889 - accuracy: 0.9390\n",
      "training time:  13.5883 s.\n",
      "average model starts\n",
      "average model ends:  0.0625 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.802]\n",
      "global epoch:  417\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1952 - accuracy: 0.9376\n",
      "training time:  11.9444 s.\n",
      "average model starts\n",
      "average model ends:  0.0385 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.799]\n",
      "global epoch:  418\n",
      "virtual memory used:  70.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2082 - accuracy: 0.9320\n",
      "training time:  12.1475 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.803]\n",
      "global epoch:  419\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2140 - accuracy: 0.9306\n",
      "training time:  12.562 s.\n",
      "average model starts\n",
      "average model ends:  0.0331 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.803 0.794]\n",
      "global epoch:  420\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.2177 - accuracy: 0.9261\n",
      "training time:  10.8364 s.\n",
      "average model starts\n",
      "average model ends:  0.0402 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.794 0.791]\n",
      "global epoch:  421\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1992 - accuracy: 0.9351\n",
      "training time:  10.8713 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.799]\n",
      "global epoch:  422\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1883 - accuracy: 0.9389\n",
      "training time:  12.6636 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.794]\n",
      "global epoch:  423\n",
      "virtual memory used:  71.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2122 - accuracy: 0.9308\n",
      "training time:  12.0966 s.\n",
      "average model starts\n",
      "average model ends:  0.0685 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.79 ]\n",
      "global epoch:  424\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1906 - accuracy: 0.9368\n",
      "training time:  13.4755 s.\n",
      "average model starts\n",
      "average model ends:  0.1578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.792]\n",
      "global epoch:  425\n",
      "virtual memory used:  70.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2154 - accuracy: 0.9324\n",
      "training time:  12.906 s.\n",
      "average model starts\n",
      "average model ends:  0.0806 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.792 0.788]\n",
      "global epoch:  426\n",
      "virtual memory used:  71.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2061 - accuracy: 0.9325\n",
      "training time:  10.9741 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.788 0.793]\n",
      "global epoch:  427\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2049 - accuracy: 0.9320\n",
      "training time:  11.9374 s.\n",
      "average model starts\n",
      "average model ends:  0.0643 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.797]\n",
      "global epoch:  428\n",
      "virtual memory used:  72.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2044 - accuracy: 0.9327\n",
      "training time:  12.642 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.781]\n",
      "global epoch:  429\n",
      "virtual memory used:  73.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1952 - accuracy: 0.9389\n",
      "training time:  12.9048 s.\n",
      "average model starts\n",
      "average model ends:  0.1822 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.781 0.799]\n",
      "global epoch:  430\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2040 - accuracy: 0.9330\n",
      "training time:  12.9107 s.\n",
      "average model starts\n",
      "average model ends:  0.0331 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.799 0.789]\n",
      "global epoch:  431\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2034 - accuracy: 0.9342\n",
      "training time:  11.0054 s.\n",
      "average model starts\n",
      "average model ends:  0.0323 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.789 0.796]\n",
      "global epoch:  432\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2141 - accuracy: 0.9278\n",
      "training time:  12.908 s.\n",
      "average model starts\n",
      "average model ends:  0.0551 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.799]\n",
      "global epoch:  433\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1948 - accuracy: 0.9353\n",
      "training time:  13.5012 s.\n",
      "average model starts\n",
      "average model ends:  0.12 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.799]\n",
      "global epoch:  434\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.2044 - accuracy: 0.9306\n",
      "training time:  14.3924 s.\n",
      "average model starts\n",
      "average model ends:  0.0494 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.797]\n",
      "global epoch:  435\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2004 - accuracy: 0.9333\n",
      "training time:  11.519 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.798]\n",
      "global epoch:  436\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1951 - accuracy: 0.9360\n",
      "training time:  11.6915 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.791]\n",
      "global epoch:  437\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1898 - accuracy: 0.9369\n",
      "training time:  12.5862 s.\n",
      "average model starts\n",
      "average model ends:  0.0516 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.801]\n",
      "global epoch:  438\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1965 - accuracy: 0.9358\n",
      "training time:  12.8732 s.\n",
      "average model starts\n",
      "average model ends:  0.0696 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.801 0.789]\n",
      "global epoch:  439\n",
      "virtual memory used:  73.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1907 - accuracy: 0.9365\n",
      "training time:  13.4749 s.\n",
      "average model starts\n",
      "average model ends:  0.0792 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.789 0.797]\n",
      "global epoch:  440\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2013 - accuracy: 0.9327\n",
      "training time:  10.9893 s.\n",
      "average model starts\n",
      "average model ends:  0.1126 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.797 0.794]\n",
      "global epoch:  441\n",
      "virtual memory used:  73.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1972 - accuracy: 0.9346\n",
      "training time:  14.5494 s.\n",
      "average model starts\n",
      "average model ends:  0.0697 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.802]\n",
      "global epoch:  442\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1953 - accuracy: 0.9351\n",
      "training time:  13.9893 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.802 0.792]\n",
      "global epoch:  443\n",
      "virtual memory used:  75.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1918 - accuracy: 0.9358\n",
      "training time:  12.2336 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.798]\n",
      "global epoch:  444\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1889 - accuracy: 0.9358\n",
      "training time:  13.5236 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.798 0.798]\n",
      "global epoch:  445\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2051 - accuracy: 0.9333\n",
      "training time:  12.292 s.\n",
      "average model starts\n",
      "average model ends:  0.07 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.798]\n",
      "global epoch:  446\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2007 - accuracy: 0.9338\n",
      "training time:  11.0934 s.\n",
      "average model starts\n",
      "average model ends:  0.0328 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.802]\n",
      "global epoch:  447\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2029 - accuracy: 0.9351\n",
      "training time:  12.2811 s.\n",
      "average model starts\n",
      "average model ends:  0.0574 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.803]\n",
      "global epoch:  448\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1962 - accuracy: 0.9381\n",
      "training time:  13.8494 s.\n",
      "average model starts\n",
      "average model ends:  0.045 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.803 0.794]\n",
      "global epoch:  449\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1982 - accuracy: 0.9339\n",
      "training time:  10.7038 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.794 0.798]\n",
      "global epoch:  450\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1993 - accuracy: 0.9361\n",
      "training time:  10.8401 s.\n",
      "average model starts\n",
      "average model ends:  0.0436 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.789]\n",
      "global epoch:  451\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2076 - accuracy: 0.9313\n",
      "training time:  12.3419 s.\n",
      "average model starts\n",
      "average model ends:  0.0386 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.789 0.796]\n",
      "global epoch:  452\n",
      "virtual memory used:  75.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2002 - accuracy: 0.9335\n",
      "training time:  11.6022 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.787]\n",
      "global epoch:  453\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1964 - accuracy: 0.9381\n",
      "training time:  12.0276 s.\n",
      "average model starts\n",
      "average model ends:  0.0454 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.787 0.793]\n",
      "global epoch:  454\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1956 - accuracy: 0.9333\n",
      "training time:  12.8608 s.\n",
      "average model starts\n",
      "average model ends:  0.0636 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.793 0.802]\n",
      "global epoch:  455\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1882 - accuracy: 0.9361\n",
      "training time:  13.7284 s.\n",
      "average model starts\n",
      "average model ends:  0.0585 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.802 0.796]\n",
      "global epoch:  456\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1948 - accuracy: 0.9345\n",
      "training time:  14.154 s.\n",
      "average model starts\n",
      "average model ends:  0.1101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.792]\n",
      "global epoch:  457\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1980 - accuracy: 0.9332\n",
      "training time:  12.8857 s.\n",
      "average model starts\n",
      "average model ends:  0.1475 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.792 0.794]\n",
      "global epoch:  458\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1934 - accuracy: 0.9344\n",
      "training time:  12.6879 s.\n",
      "average model starts\n",
      "average model ends:  0.0543 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.794 0.801]\n",
      "global epoch:  459\n",
      "virtual memory used:  77.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1915 - accuracy: 0.9353\n",
      "training time:  12.8699 s.\n",
      "average model starts\n",
      "average model ends:  0.0608 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.794]\n",
      "global epoch:  460\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2028 - accuracy: 0.9322\n",
      "training time:  12.4011 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.796]\n",
      "global epoch:  461\n",
      "virtual memory used:  69.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.2096 - accuracy: 0.9303\n",
      "training time:  12.2849 s.\n",
      "average model starts\n",
      "average model ends:  0.0573 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.794]\n",
      "global epoch:  462\n",
      "virtual memory used:  69.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.2098 - accuracy: 0.9322\n",
      "training time:  12.616 s.\n",
      "average model starts\n",
      "average model ends:  0.0558 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.797]\n",
      "global epoch:  463\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1793 - accuracy: 0.9391\n",
      "training time:  13.4464 s.\n",
      "average model starts\n",
      "average model ends:  0.1707 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.791]\n",
      "global epoch:  464\n",
      "virtual memory used:  70.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2002 - accuracy: 0.9349\n",
      "training time:  10.9455 s.\n",
      "average model starts\n",
      "average model ends:  0.037 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.793]\n",
      "global epoch:  465\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1976 - accuracy: 0.9351\n",
      "training time:  10.3284 s.\n",
      "average model starts\n",
      "average model ends:  0.0903 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.793 0.799]\n",
      "global epoch:  466\n",
      "virtual memory used:  71.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1910 - accuracy: 0.9350\n",
      "training time:  11.7941 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.799 0.786]\n",
      "global epoch:  467\n",
      "virtual memory used:  71.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1860 - accuracy: 0.9376\n",
      "training time:  12.5417 s.\n",
      "average model starts\n",
      "average model ends:  0.0581 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.786 0.794]\n",
      "global epoch:  468\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2099 - accuracy: 0.9294\n",
      "training time:  12.6365 s.\n",
      "average model starts\n",
      "average model ends:  0.0877 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.794 0.796]\n",
      "global epoch:  469\n",
      "virtual memory used:  70.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2134 - accuracy: 0.9323\n",
      "training time:  11.3852 s.\n",
      "average model starts\n",
      "average model ends:  0.0534 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.8  ]\n",
      "global epoch:  470\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1836 - accuracy: 0.9385\n",
      "training time:  11.69 s.\n",
      "average model starts\n",
      "average model ends:  0.0545 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.799]\n",
      "global epoch:  471\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.2046 - accuracy: 0.9334\n",
      "training time:  10.6172 s.\n",
      "average model starts\n",
      "average model ends:  0.1032 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.799 0.794]\n",
      "global epoch:  472\n",
      "virtual memory used:  71.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2000 - accuracy: 0.9318\n",
      "training time:  13.4261 s.\n",
      "average model starts\n",
      "average model ends:  0.0639 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.8  ]\n",
      "global epoch:  473\n",
      "virtual memory used:  71.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1931 - accuracy: 0.9360\n",
      "training time:  12.1898 s.\n",
      "average model starts\n",
      "average model ends:  0.1027 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.8  ]\n",
      "global epoch:  474\n",
      "virtual memory used:  72.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.2035 - accuracy: 0.9333\n",
      "training time:  13.9806 s.\n",
      "average model starts\n",
      "average model ends:  0.0485 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.794]\n",
      "global epoch:  475\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1972 - accuracy: 0.9340\n",
      "training time:  13.3135 s.\n",
      "average model starts\n",
      "average model ends:  0.1491 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.8  ]\n",
      "global epoch:  476\n",
      "virtual memory used:  72.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1889 - accuracy: 0.9366\n",
      "training time:  14.2893 s.\n",
      "average model starts\n",
      "average model ends:  0.086 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.793]\n",
      "global epoch:  477\n",
      "virtual memory used:  72.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2098 - accuracy: 0.9315\n",
      "training time:  13.6011 s.\n",
      "average model starts\n",
      "average model ends:  0.0547 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.785]\n",
      "global epoch:  478\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1901 - accuracy: 0.9377\n",
      "training time:  11.566 s.\n",
      "average model starts\n",
      "average model ends:  0.0424 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.785 0.794]\n",
      "global epoch:  479\n",
      "virtual memory used:  73.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1933 - accuracy: 0.9351\n",
      "training time:  12.5973 s.\n",
      "average model starts\n",
      "average model ends:  0.0719 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.794 0.794]\n",
      "global epoch:  480\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2003 - accuracy: 0.9329\n",
      "training time:  12.3781 s.\n",
      "average model starts\n",
      "average model ends:  0.0625 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.789]\n",
      "global epoch:  481\n",
      "virtual memory used:  72.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1998 - accuracy: 0.9304\n",
      "training time:  12.9366 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.789 0.797]\n",
      "global epoch:  482\n",
      "virtual memory used:  73.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1981 - accuracy: 0.9349\n",
      "training time:  13.6539 s.\n",
      "average model starts\n",
      "average model ends:  0.0422 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.797 0.789]\n",
      "global epoch:  483\n",
      "virtual memory used:  73.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1996 - accuracy: 0.9344\n",
      "training time:  12.4688 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.789 0.794]\n",
      "global epoch:  484\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2008 - accuracy: 0.9338\n",
      "training time:  13.4269 s.\n",
      "average model starts\n",
      "average model ends:  0.0849 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.794 0.795]\n",
      "global epoch:  485\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1901 - accuracy: 0.9369\n",
      "training time:  12.7393 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.802]\n",
      "global epoch:  486\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1885 - accuracy: 0.9396\n",
      "training time:  11.4383 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.793]\n",
      "global epoch:  487\n",
      "virtual memory used:  73.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1919 - accuracy: 0.9328\n",
      "training time:  10.7519 s.\n",
      "average model starts\n",
      "average model ends:  0.0402 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.793 0.796]\n",
      "global epoch:  488\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1994 - accuracy: 0.9340\n",
      "training time:  12.6695 s.\n",
      "average model starts\n",
      "average model ends:  0.033 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.79 ]\n",
      "global epoch:  489\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2051 - accuracy: 0.9294\n",
      "training time:  12.9145 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.79  0.796]\n",
      "global epoch:  490\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1980 - accuracy: 0.9333\n",
      "training time:  12.6044 s.\n",
      "average model starts\n",
      "average model ends:  0.0802 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.796 0.788]\n",
      "global epoch:  491\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1845 - accuracy: 0.9380\n",
      "training time:  13.0724 s.\n",
      "average model starts\n",
      "average model ends:  0.037 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.797]\n",
      "global epoch:  492\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2064 - accuracy: 0.9318\n",
      "training time:  13.0088 s.\n",
      "average model starts\n",
      "average model ends:  0.0819 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.797 0.793]\n",
      "global epoch:  493\n",
      "virtual memory used:  75.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1823 - accuracy: 0.9411\n",
      "training time:  12.7651 s.\n",
      "average model starts\n",
      "average model ends:  0.1203 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.793]\n",
      "global epoch:  494\n",
      "virtual memory used:  73.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1941 - accuracy: 0.9363\n",
      "training time:  11.1705 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.793 0.793]\n",
      "global epoch:  495\n",
      "virtual memory used:  74.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1961 - accuracy: 0.9349\n",
      "training time:  13.0059 s.\n",
      "average model starts\n",
      "average model ends:  0.0374 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.793 0.793]\n",
      "global epoch:  496\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1976 - accuracy: 0.9330\n",
      "training time:  13.9389 s.\n",
      "average model starts\n",
      "average model ends:  0.105 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.793 0.796]\n",
      "global epoch:  497\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.2013 - accuracy: 0.9319\n",
      "training time:  13.4443 s.\n",
      "average model starts\n",
      "average model ends:  0.0641 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.802]\n",
      "global epoch:  498\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.2026 - accuracy: 0.9346\n",
      "training time:  13.3194 s.\n",
      "average model starts\n",
      "average model ends:  0.0405 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.788]\n",
      "global epoch:  499\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.2012 - accuracy: 0.9330\n",
      "training time:  14.2437 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.788 0.793]\n",
      "global epoch:  500\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1793 - accuracy: 0.9413\n",
      "training time:  12.3017 s.\n",
      "average model starts\n",
      "average model ends:  0.0377 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.794]\n",
      "global epoch:  501\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2096 - accuracy: 0.9327\n",
      "training time:  12.5542 s.\n",
      "average model starts\n",
      "average model ends:  0.0594 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.805]\n",
      "global epoch:  502\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2104 - accuracy: 0.9340\n",
      "training time:  12.0221 s.\n",
      "average model starts\n",
      "average model ends:  0.0411 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.805 0.801]\n",
      "global epoch:  503\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1807 - accuracy: 0.9399\n",
      "training time:  11.4123 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.801 0.801]\n",
      "global epoch:  504\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1913 - accuracy: 0.9351\n",
      "training time:  12.1142 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.795]\n",
      "global epoch:  505\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.2066 - accuracy: 0.9334\n",
      "training time:  11.01 s.\n",
      "average model starts\n",
      "average model ends:  0.052 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.786]\n",
      "global epoch:  506\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1792 - accuracy: 0.9427\n",
      "training time:  10.6398 s.\n",
      "average model starts\n",
      "average model ends:  0.103 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.786 0.791]\n",
      "global epoch:  507\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1877 - accuracy: 0.9400\n",
      "training time:  11.1896 s.\n",
      "average model starts\n",
      "average model ends:  0.035 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.791 0.796]\n",
      "global epoch:  508\n",
      "virtual memory used:  76.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1875 - accuracy: 0.9379\n",
      "training time:  11.5763 s.\n",
      "average model starts\n",
      "average model ends:  0.0351 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.8  ]\n",
      "global epoch:  509\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1871 - accuracy: 0.9353\n",
      "training time:  11.7712 s.\n",
      "average model starts\n",
      "average model ends:  0.1138 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.804]\n",
      "global epoch:  510\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1960 - accuracy: 0.9330\n",
      "training time:  11.02 s.\n",
      "average model starts\n",
      "average model ends:  0.0852 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.793]\n",
      "global epoch:  511\n",
      "virtual memory used:  69.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1911 - accuracy: 0.9390\n",
      "training time:  10.9997 s.\n",
      "average model starts\n",
      "average model ends:  0.0866 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.793 0.789]\n",
      "global epoch:  512\n",
      "virtual memory used:  70.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1962 - accuracy: 0.9350\n",
      "training time:  13.801 s.\n",
      "average model starts\n",
      "average model ends:  0.0604 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.789 0.797]\n",
      "global epoch:  513\n",
      "virtual memory used:  71.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1987 - accuracy: 0.9371\n",
      "training time:  13.7333 s.\n",
      "average model starts\n",
      "average model ends:  0.0851 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.797 0.794]\n",
      "global epoch:  514\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1918 - accuracy: 0.9363\n",
      "training time:  13.4297 s.\n",
      "average model starts\n",
      "average model ends:  0.1366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.794]\n",
      "global epoch:  515\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2082 - accuracy: 0.9334\n",
      "training time:  11.286 s.\n",
      "average model starts\n",
      "average model ends:  0.0346 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.795]\n",
      "global epoch:  516\n",
      "virtual memory used:  71.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1960 - accuracy: 0.9364\n",
      "training time:  12.2989 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.798]\n",
      "global epoch:  517\n",
      "virtual memory used:  71.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1797 - accuracy: 0.9405\n",
      "training time:  11.9287 s.\n",
      "average model starts\n",
      "average model ends:  0.0315 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.797]\n",
      "global epoch:  518\n",
      "virtual memory used:  72.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2007 - accuracy: 0.9307\n",
      "training time:  11.6657 s.\n",
      "average model starts\n",
      "average model ends:  0.1317 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.791]\n",
      "global epoch:  519\n",
      "virtual memory used:  71.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1963 - accuracy: 0.9344\n",
      "training time:  12.1692 s.\n",
      "average model starts\n",
      "average model ends:  0.1278 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.795]\n",
      "global epoch:  520\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.2048 - accuracy: 0.9306\n",
      "training time:  12.5979 s.\n",
      "average model starts\n",
      "average model ends:  0.0474 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.795 0.795]\n",
      "global epoch:  521\n",
      "virtual memory used:  72.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1931 - accuracy: 0.9355\n",
      "training time:  11.5405 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.803]\n",
      "global epoch:  522\n",
      "virtual memory used:  72.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2098 - accuracy: 0.9306\n",
      "training time:  11.9361 s.\n",
      "average model starts\n",
      "average model ends:  0.038 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.803 0.796]\n",
      "global epoch:  523\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1987 - accuracy: 0.9349\n",
      "training time:  12.7422 s.\n",
      "average model starts\n",
      "average model ends:  0.0325 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.788]\n",
      "global epoch:  524\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1839 - accuracy: 0.9413\n",
      "training time:  12.5186 s.\n",
      "average model starts\n",
      "average model ends:  0.0615 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.797]\n",
      "global epoch:  525\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1828 - accuracy: 0.9386\n",
      "training time:  12.7757 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.797 0.794]\n",
      "global epoch:  526\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1957 - accuracy: 0.9339\n",
      "training time:  12.4481 s.\n",
      "average model starts\n",
      "average model ends:  0.0352 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.791]\n",
      "global epoch:  527\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1961 - accuracy: 0.9350\n",
      "training time:  12.4353 s.\n",
      "average model starts\n",
      "average model ends:  0.0958 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.801]\n",
      "global epoch:  528\n",
      "virtual memory used:  72.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.2060 - accuracy: 0.9330\n",
      "training time:  11.7067 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.801 0.797]\n",
      "global epoch:  529\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1874 - accuracy: 0.9368\n",
      "training time:  11.4745 s.\n",
      "average model starts\n",
      "average model ends:  0.0427 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.794]\n",
      "global epoch:  530\n",
      "virtual memory used:  73.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1943 - accuracy: 0.9385\n",
      "training time:  11.5705 s.\n",
      "average model starts\n",
      "average model ends:  0.0503 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.798]\n",
      "global epoch:  531\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1843 - accuracy: 0.9358\n",
      "training time:  11.4259 s.\n",
      "average model starts\n",
      "average model ends:  0.084 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.795]\n",
      "global epoch:  532\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1964 - accuracy: 0.9376\n",
      "training time:  11.3636 s.\n",
      "average model starts\n",
      "average model ends:  0.0361 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.795 0.797]\n",
      "global epoch:  533\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1922 - accuracy: 0.9344\n",
      "training time:  13.8892 s.\n",
      "average model starts\n",
      "average model ends:  0.0595 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.8  ]\n",
      "global epoch:  534\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2091 - accuracy: 0.9317\n",
      "training time:  12.5107 s.\n",
      "average model starts\n",
      "average model ends:  0.0696 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.787]\n",
      "global epoch:  535\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1972 - accuracy: 0.9349\n",
      "training time:  11.8194 s.\n",
      "average model starts\n",
      "average model ends:  0.0494 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.787 0.799]\n",
      "global epoch:  536\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1943 - accuracy: 0.9382\n",
      "training time:  12.5768 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.799 0.794]\n",
      "global epoch:  537\n",
      "virtual memory used:  73.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1950 - accuracy: 0.9381\n",
      "training time:  11.6751 s.\n",
      "average model starts\n",
      "average model ends:  0.0436 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.796]\n",
      "global epoch:  538\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1914 - accuracy: 0.9348\n",
      "training time:  12.8896 s.\n",
      "average model starts\n",
      "average model ends:  0.0642 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.797]\n",
      "global epoch:  539\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1971 - accuracy: 0.9350\n",
      "training time:  12.0249 s.\n",
      "average model starts\n",
      "average model ends:  0.1173 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.798]\n",
      "global epoch:  540\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1904 - accuracy: 0.9359\n",
      "training time:  12.9366 s.\n",
      "average model starts\n",
      "average model ends:  0.0618 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.801]\n",
      "global epoch:  541\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1862 - accuracy: 0.9392\n",
      "training time:  14.5656 s.\n",
      "average model starts\n",
      "average model ends:  0.0498 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.794]\n",
      "global epoch:  542\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1925 - accuracy: 0.9360\n",
      "training time:  12.0483 s.\n",
      "average model starts\n",
      "average model ends:  0.1551 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.796]\n",
      "global epoch:  543\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1968 - accuracy: 0.9360\n",
      "training time:  13.3203 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.798]\n",
      "global epoch:  544\n",
      "virtual memory used:  75.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.2005 - accuracy: 0.9337\n",
      "training time:  11.3783 s.\n",
      "average model starts\n",
      "average model ends:  0.052 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.78 ]\n",
      "global epoch:  545\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2146 - accuracy: 0.9301\n",
      "training time:  12.5943 s.\n",
      "average model starts\n",
      "average model ends:  0.0486 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.78  0.795]\n",
      "global epoch:  546\n",
      "virtual memory used:  76.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1876 - accuracy: 0.9374\n",
      "training time:  10.8869 s.\n",
      "average model starts\n",
      "average model ends:  0.0326 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.78  0.795 0.798]\n",
      "global epoch:  547\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1953 - accuracy: 0.9350\n",
      "training time:  10.4087 s.\n",
      "average model starts\n",
      "average model ends:  0.0554 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.791]\n",
      "global epoch:  548\n",
      "virtual memory used:  75.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1901 - accuracy: 0.9377\n",
      "training time:  10.795 s.\n",
      "average model starts\n",
      "average model ends:  0.0295 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.794]\n",
      "global epoch:  549\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1933 - accuracy: 0.9366\n",
      "training time:  12.7671 s.\n",
      "average model starts\n",
      "average model ends:  0.0386 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.794 0.796]\n",
      "global epoch:  550\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1902 - accuracy: 0.9390\n",
      "training time:  12.2022 s.\n",
      "average model starts\n",
      "average model ends:  0.0432 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.796]\n",
      "global epoch:  551\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1841 - accuracy: 0.9402\n",
      "training time:  13.7478 s.\n",
      "average model starts\n",
      "average model ends:  0.0539 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.8  ]\n",
      "global epoch:  552\n",
      "virtual memory used:  75.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1941 - accuracy: 0.9354\n",
      "training time:  12.7893 s.\n",
      "average model starts\n",
      "average model ends:  0.0984 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.796]\n",
      "global epoch:  553\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1964 - accuracy: 0.9343\n",
      "training time:  11.2719 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.802]\n",
      "global epoch:  554\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1823 - accuracy: 0.9411\n",
      "training time:  11.6645 s.\n",
      "average model starts\n",
      "average model ends:  0.084 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.803]\n",
      "global epoch:  555\n",
      "virtual memory used:  77.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1941 - accuracy: 0.9359\n",
      "training time:  12.17 s.\n",
      "average model starts\n",
      "average model ends:  0.1403 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.803 0.781]\n",
      "global epoch:  556\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1906 - accuracy: 0.9360\n",
      "training time:  11.6625 s.\n",
      "average model starts\n",
      "average model ends:  0.0587 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.781 0.795]\n",
      "global epoch:  557\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1987 - accuracy: 0.9396\n",
      "training time:  11.9476 s.\n",
      "average model starts\n",
      "average model ends:  0.1382 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.781 0.795 0.794]\n",
      "global epoch:  558\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1906 - accuracy: 0.9380\n",
      "training time:  11.4163 s.\n",
      "average model starts\n",
      "average model ends:  0.12 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.797]\n",
      "global epoch:  559\n",
      "virtual memory used:  77.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1889 - accuracy: 0.9371\n",
      "training time:  10.8822 s.\n",
      "average model starts\n",
      "average model ends:  0.0951 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.797]\n",
      "global epoch:  560\n",
      "virtual memory used:  76.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1889 - accuracy: 0.9380\n",
      "training time:  12.9144 s.\n",
      "average model starts\n",
      "average model ends:  0.1333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.796]\n",
      "global epoch:  561\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1972 - accuracy: 0.9343\n",
      "training time:  13.665 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.799]\n",
      "global epoch:  562\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1998 - accuracy: 0.9334\n",
      "training time:  12.994 s.\n",
      "average model starts\n",
      "average model ends:  0.0394 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.803]\n",
      "global epoch:  563\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2057 - accuracy: 0.9306\n",
      "training time:  12.7618 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.803 0.795]\n",
      "global epoch:  564\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1762 - accuracy: 0.9396\n",
      "training time:  12.8794 s.\n",
      "average model starts\n",
      "average model ends:  0.066 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.798]\n",
      "global epoch:  565\n",
      "virtual memory used:  70.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1863 - accuracy: 0.9365\n",
      "training time:  11.4092 s.\n",
      "average model starts\n",
      "average model ends:  0.0768 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.803]\n",
      "global epoch:  566\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1867 - accuracy: 0.9376\n",
      "training time:  13.0762 s.\n",
      "average model starts\n",
      "average model ends:  0.0466 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.803 0.798]\n",
      "global epoch:  567\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1973 - accuracy: 0.9382\n",
      "training time:  12.9963 s.\n",
      "average model starts\n",
      "average model ends:  0.0635 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.797]\n",
      "global epoch:  568\n",
      "virtual memory used:  72.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1891 - accuracy: 0.9384\n",
      "training time:  12.2237 s.\n",
      "average model starts\n",
      "average model ends:  0.0584 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.798]\n",
      "global epoch:  569\n",
      "virtual memory used:  72.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1915 - accuracy: 0.9379\n",
      "training time:  12.4335 s.\n",
      "average model starts\n",
      "average model ends:  0.0421 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.805]\n",
      "global epoch:  570\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1983 - accuracy: 0.9379\n",
      "training time:  13.2259 s.\n",
      "average model starts\n",
      "average model ends:  0.065 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.805 0.795]\n",
      "global epoch:  571\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1918 - accuracy: 0.9369\n",
      "training time:  13.317 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.795 0.794]\n",
      "global epoch:  572\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1765 - accuracy: 0.9446\n",
      "training time:  11.9714 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.788]\n",
      "global epoch:  573\n",
      "virtual memory used:  73.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1882 - accuracy: 0.9363\n",
      "training time:  12.3294 s.\n",
      "average model starts\n",
      "average model ends:  0.0471 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.788 0.791]\n",
      "global epoch:  574\n",
      "virtual memory used:  73.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1929 - accuracy: 0.9363\n",
      "training time:  12.8878 s.\n",
      "average model starts\n",
      "average model ends:  0.0414 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.791 0.803]\n",
      "global epoch:  575\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1852 - accuracy: 0.9420\n",
      "training time:  11.5776 s.\n",
      "average model starts\n",
      "average model ends:  0.0901 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.803 0.801]\n",
      "global epoch:  576\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1909 - accuracy: 0.9364\n",
      "training time:  12.9327 s.\n",
      "average model starts\n",
      "average model ends:  0.0364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.802]\n",
      "global epoch:  577\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1910 - accuracy: 0.9384\n",
      "training time:  13.1962 s.\n",
      "average model starts\n",
      "average model ends:  0.085 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.806]\n",
      "global epoch:  578\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1882 - accuracy: 0.9390\n",
      "training time:  12.0063 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.806 0.793]\n",
      "global epoch:  579\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1935 - accuracy: 0.9377\n",
      "training time:  13.1231 s.\n",
      "average model starts\n",
      "average model ends:  0.1315 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.793 0.792]\n",
      "global epoch:  580\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1904 - accuracy: 0.9382\n",
      "training time:  12.2067 s.\n",
      "average model starts\n",
      "average model ends:  0.0319 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.792 0.798]\n",
      "global epoch:  581\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1976 - accuracy: 0.9375\n",
      "training time:  13.6463 s.\n",
      "average model starts\n",
      "average model ends:  0.072 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.798 0.796]\n",
      "global epoch:  582\n",
      "virtual memory used:  73.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1991 - accuracy: 0.9340\n",
      "training time:  13.8879 s.\n",
      "average model starts\n",
      "average model ends:  0.0601 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.803]\n",
      "global epoch:  583\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1864 - accuracy: 0.9390\n",
      "training time:  12.6753 s.\n",
      "average model starts\n",
      "average model ends:  0.0375 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.799]\n",
      "global epoch:  584\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1923 - accuracy: 0.9386\n",
      "training time:  10.7245 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.799 0.793]\n",
      "global epoch:  585\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2016 - accuracy: 0.9339\n",
      "training time:  11.9851 s.\n",
      "average model starts\n",
      "average model ends:  0.0822 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.793]\n",
      "global epoch:  586\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1919 - accuracy: 0.9369\n",
      "training time:  13.6036 s.\n",
      "average model starts\n",
      "average model ends:  0.0638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.793 0.801]\n",
      "global epoch:  587\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2008 - accuracy: 0.9330\n",
      "training time:  12.5897 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.801 0.797]\n",
      "global epoch:  588\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1857 - accuracy: 0.9416\n",
      "training time:  12.1427 s.\n",
      "average model starts\n",
      "average model ends:  0.066 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.793]\n",
      "global epoch:  589\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1865 - accuracy: 0.9386\n",
      "training time:  12.3361 s.\n",
      "average model starts\n",
      "average model ends:  0.0603 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.794]\n",
      "global epoch:  590\n",
      "virtual memory used:  73.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.2006 - accuracy: 0.9379\n",
      "training time:  12.825 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.802]\n",
      "global epoch:  591\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1865 - accuracy: 0.9396\n",
      "training time:  13.92 s.\n",
      "average model starts\n",
      "average model ends:  0.0311 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.802 0.793]\n",
      "global epoch:  592\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1817 - accuracy: 0.9400\n",
      "training time:  13.0014 s.\n",
      "average model starts\n",
      "average model ends:  0.1065 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.793 0.803]\n",
      "global epoch:  593\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1942 - accuracy: 0.9354\n",
      "training time:  13.2987 s.\n",
      "average model starts\n",
      "average model ends:  0.0787 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.803 0.792]\n",
      "global epoch:  594\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1779 - accuracy: 0.9390\n",
      "training time:  11.91 s.\n",
      "average model starts\n",
      "average model ends:  0.0633 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.792 0.787]\n",
      "global epoch:  595\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1929 - accuracy: 0.9360\n",
      "training time:  12.1933 s.\n",
      "average model starts\n",
      "average model ends:  0.0888 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.787 0.798]\n",
      "global epoch:  596\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1873 - accuracy: 0.9396\n",
      "training time:  10.6034 s.\n",
      "average model starts\n",
      "average model ends:  0.0699 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.798 0.796]\n",
      "global epoch:  597\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.2056 - accuracy: 0.9359\n",
      "training time:  11.3166 s.\n",
      "average model starts\n",
      "average model ends:  0.0892 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.8  ]\n",
      "global epoch:  598\n",
      "virtual memory used:  75.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1839 - accuracy: 0.9368\n",
      "training time:  13.0921 s.\n",
      "average model starts\n",
      "average model ends:  0.0628 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.788]\n",
      "global epoch:  599\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1963 - accuracy: 0.9332\n",
      "training time:  11.6351 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.788 0.792]\n",
      "global epoch:  600\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1803 - accuracy: 0.9404\n",
      "training time:  13.1071 s.\n",
      "average model starts\n",
      "average model ends:  0.0577 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.792 0.797]\n",
      "global epoch:  601\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1860 - accuracy: 0.9389\n",
      "training time:  12.2293 s.\n",
      "average model starts\n",
      "average model ends:  0.057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.797 0.797]\n",
      "global epoch:  602\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1931 - accuracy: 0.9370\n",
      "training time:  11.318 s.\n",
      "average model starts\n",
      "average model ends:  0.0338 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.797]\n",
      "global epoch:  603\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2054 - accuracy: 0.9323\n",
      "training time:  11.9799 s.\n",
      "average model starts\n",
      "average model ends:  0.0739 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.789]\n",
      "global epoch:  604\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1977 - accuracy: 0.9350\n",
      "training time:  12.8066 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.789 0.8  ]\n",
      "global epoch:  605\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1748 - accuracy: 0.9408\n",
      "training time:  12.6251 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.8   0.793]\n",
      "global epoch:  606\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1838 - accuracy: 0.9392\n",
      "training time:  11.8115 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.798]\n",
      "global epoch:  607\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1956 - accuracy: 0.9324\n",
      "training time:  12.7079 s.\n",
      "average model starts\n",
      "average model ends:  0.0647 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.798]\n",
      "global epoch:  608\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1929 - accuracy: 0.9376\n",
      "training time:  12.0516 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.799]\n",
      "global epoch:  609\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1842 - accuracy: 0.9371\n",
      "training time:  12.3703 s.\n",
      "average model starts\n",
      "average model ends:  0.0544 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.798]\n",
      "global epoch:  610\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1772 - accuracy: 0.9400\n",
      "training time:  11.5088 s.\n",
      "average model starts\n",
      "average model ends:  0.0593 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.798 0.792]\n",
      "global epoch:  611\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1860 - accuracy: 0.9377\n",
      "training time:  12.749 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.805]\n",
      "global epoch:  612\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1742 - accuracy: 0.9416\n",
      "training time:  11.1359 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.805 0.787]\n",
      "global epoch:  613\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1801 - accuracy: 0.9407\n",
      "training time:  11.8971 s.\n",
      "average model starts\n",
      "average model ends:  0.0592 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.787 0.802]\n",
      "global epoch:  614\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1824 - accuracy: 0.9374\n",
      "training time:  11.971 s.\n",
      "average model starts\n",
      "average model ends:  0.0555 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.802 0.801]\n",
      "global epoch:  615\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.2090 - accuracy: 0.9342\n",
      "training time:  11.8494 s.\n",
      "average model starts\n",
      "average model ends:  0.0619 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.783]\n",
      "global epoch:  616\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1844 - accuracy: 0.9376\n",
      "training time:  11.4731 s.\n",
      "average model starts\n",
      "average model ends:  0.0549 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.783 0.799]\n",
      "global epoch:  617\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1919 - accuracy: 0.9358\n",
      "training time:  11.2281 s.\n",
      "average model starts\n",
      "average model ends:  0.1057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.799 0.8  ]\n",
      "global epoch:  618\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1876 - accuracy: 0.9373\n",
      "training time:  13.7446 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.793]\n",
      "global epoch:  619\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1815 - accuracy: 0.9418\n",
      "training time:  11.872 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.788]\n",
      "global epoch:  620\n",
      "virtual memory used:  71.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1984 - accuracy: 0.9343\n",
      "training time:  13.5066 s.\n",
      "average model starts\n",
      "average model ends:  0.1138 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.788 0.799]\n",
      "global epoch:  621\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1911 - accuracy: 0.9368\n",
      "training time:  11.864 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.799 0.798]\n",
      "global epoch:  622\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1767 - accuracy: 0.9435\n",
      "training time:  11.1402 s.\n",
      "average model starts\n",
      "average model ends:  0.1293 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.798 0.799]\n",
      "global epoch:  623\n",
      "virtual memory used:  71.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1895 - accuracy: 0.9390\n",
      "training time:  12.3145 s.\n",
      "average model starts\n",
      "average model ends:  0.0576 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.795]\n",
      "global epoch:  624\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1914 - accuracy: 0.9364\n",
      "training time:  12.6692 s.\n",
      "average model starts\n",
      "average model ends:  0.0775 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.793]\n",
      "global epoch:  625\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1892 - accuracy: 0.9408\n",
      "training time:  11.481 s.\n",
      "average model starts\n",
      "average model ends:  0.0394 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.802]\n",
      "global epoch:  626\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1933 - accuracy: 0.9363\n",
      "training time:  12.0838 s.\n",
      "average model starts\n",
      "average model ends:  0.0469 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.802 0.792]\n",
      "global epoch:  627\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1918 - accuracy: 0.9376\n",
      "training time:  12.9716 s.\n",
      "average model starts\n",
      "average model ends:  0.0718 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.794]\n",
      "global epoch:  628\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1840 - accuracy: 0.9394\n",
      "training time:  12.5212 s.\n",
      "average model starts\n",
      "average model ends:  0.1176 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.794 0.795]\n",
      "global epoch:  629\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1864 - accuracy: 0.9406\n",
      "training time:  11.9422 s.\n",
      "average model starts\n",
      "average model ends:  0.0314 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.794]\n",
      "global epoch:  630\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1872 - accuracy: 0.9373\n",
      "training time:  11.8234 s.\n",
      "average model starts\n",
      "average model ends:  0.1162 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.788]\n",
      "global epoch:  631\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.2085 - accuracy: 0.9307\n",
      "training time:  13.0277 s.\n",
      "average model starts\n",
      "average model ends:  0.0826 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.788 0.795]\n",
      "global epoch:  632\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1908 - accuracy: 0.9364\n",
      "training time:  12.1693 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.795 0.79 ]\n",
      "global epoch:  633\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1887 - accuracy: 0.9358\n",
      "training time:  11.6013 s.\n",
      "average model starts\n",
      "average model ends:  0.0404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.79  0.799]\n",
      "global epoch:  634\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1901 - accuracy: 0.9381\n",
      "training time:  12.3914 s.\n",
      "average model starts\n",
      "average model ends:  0.0659 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.799 0.796]\n",
      "global epoch:  635\n",
      "virtual memory used:  75.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1818 - accuracy: 0.9397\n",
      "training time:  13.1154 s.\n",
      "average model starts\n",
      "average model ends:  0.042 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.793]\n",
      "global epoch:  636\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1765 - accuracy: 0.9410\n",
      "training time:  12.8259 s.\n",
      "average model starts\n",
      "average model ends:  0.0717 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.798]\n",
      "global epoch:  637\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1913 - accuracy: 0.9360\n",
      "training time:  12.6359 s.\n",
      "average model starts\n",
      "average model ends:  0.0455 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.799]\n",
      "global epoch:  638\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1797 - accuracy: 0.9415\n",
      "training time:  13.6644 s.\n",
      "average model starts\n",
      "average model ends:  0.0517 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.797]\n",
      "global epoch:  639\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1981 - accuracy: 0.9355\n",
      "training time:  13.7987 s.\n",
      "average model starts\n",
      "average model ends:  0.0563 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.787]\n",
      "global epoch:  640\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1843 - accuracy: 0.9399\n",
      "training time:  11.4569 s.\n",
      "average model starts\n",
      "average model ends:  0.0861 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.787 0.796]\n",
      "global epoch:  641\n",
      "virtual memory used:  74.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1973 - accuracy: 0.9320\n",
      "training time:  12.2338 s.\n",
      "average model starts\n",
      "average model ends:  0.1367 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.796 0.79 ]\n",
      "global epoch:  642\n",
      "virtual memory used:  74.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1938 - accuracy: 0.9364\n",
      "training time:  13.1449 s.\n",
      "average model starts\n",
      "average model ends:  0.0631 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.79  0.793]\n",
      "global epoch:  643\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1982 - accuracy: 0.9364\n",
      "training time:  12.7357 s.\n",
      "average model starts\n",
      "average model ends:  0.0369 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.793 0.794]\n",
      "global epoch:  644\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1903 - accuracy: 0.9356\n",
      "training time:  11.654 s.\n",
      "average model starts\n",
      "average model ends:  0.0589 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.799]\n",
      "global epoch:  645\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1917 - accuracy: 0.9350\n",
      "training time:  11.6143 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.793]\n",
      "global epoch:  646\n",
      "virtual memory used:  75.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1833 - accuracy: 0.9387\n",
      "training time:  10.538 s.\n",
      "average model starts\n",
      "average model ends:  0.0932 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.796]\n",
      "global epoch:  647\n",
      "virtual memory used:  75.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1971 - accuracy: 0.9355\n",
      "training time:  13.1519 s.\n",
      "average model starts\n",
      "average model ends:  0.1056 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.801]\n",
      "global epoch:  648\n",
      "virtual memory used:  75.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1861 - accuracy: 0.9361\n",
      "training time:  12.7648 s.\n",
      "average model starts\n",
      "average model ends:  0.0459 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.795]\n",
      "global epoch:  649\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1865 - accuracy: 0.9374\n",
      "training time:  12.5024 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.802]\n",
      "global epoch:  650\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1758 - accuracy: 0.9449\n",
      "training time:  13.3025 s.\n",
      "average model starts\n",
      "average model ends:  0.098 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.794]\n",
      "global epoch:  651\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1834 - accuracy: 0.9389\n",
      "training time:  11.8233 s.\n",
      "average model starts\n",
      "average model ends:  0.1014 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.806]\n",
      "global epoch:  652\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1930 - accuracy: 0.9391\n",
      "training time:  12.2441 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.806 0.795]\n",
      "global epoch:  653\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1990 - accuracy: 0.9325\n",
      "training time:  12.1961 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.795 0.797]\n",
      "global epoch:  654\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1982 - accuracy: 0.9325\n",
      "training time:  13.7428 s.\n",
      "average model starts\n",
      "average model ends:  0.0604 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.795]\n",
      "global epoch:  655\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1952 - accuracy: 0.9376\n",
      "training time:  14.1829 s.\n",
      "average model starts\n",
      "average model ends:  0.0912 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.8  ]\n",
      "global epoch:  656\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1892 - accuracy: 0.9366\n",
      "training time:  13.1743 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.797]\n",
      "global epoch:  657\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1815 - accuracy: 0.9379\n",
      "training time:  13.1131 s.\n",
      "average model starts\n",
      "average model ends:  0.0476 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.794]\n",
      "global epoch:  658\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1768 - accuracy: 0.9402\n",
      "training time:  13.7576 s.\n",
      "average model starts\n",
      "average model ends:  0.0479 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.799]\n",
      "global epoch:  659\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1870 - accuracy: 0.9360\n",
      "training time:  11.9219 s.\n",
      "average model starts\n",
      "average model ends:  0.1113 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.779]\n",
      "global epoch:  660\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1931 - accuracy: 0.9355\n",
      "training time:  14.2099 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.779 0.793]\n",
      "global epoch:  661\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1774 - accuracy: 0.9417\n",
      "training time:  12.0167 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.779 0.793 0.797]\n",
      "global epoch:  662\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1788 - accuracy: 0.9416\n",
      "training time:  11.8984 s.\n",
      "average model starts\n",
      "average model ends:  0.0889 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.797]\n",
      "global epoch:  663\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1936 - accuracy: 0.9365\n",
      "training time:  12.9588 s.\n",
      "average model starts\n",
      "average model ends:  0.0632 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.793]\n",
      "global epoch:  664\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1772 - accuracy: 0.9446\n",
      "training time:  12.5161 s.\n",
      "average model starts\n",
      "average model ends:  0.0882 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.795]\n",
      "global epoch:  665\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1915 - accuracy: 0.9382\n",
      "training time:  12.5636 s.\n",
      "average model starts\n",
      "average model ends:  0.0566 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.795]\n",
      "global epoch:  666\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1897 - accuracy: 0.9399\n",
      "training time:  12.815 s.\n",
      "average model starts\n",
      "average model ends:  0.0683 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.783]\n",
      "global epoch:  667\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1817 - accuracy: 0.9408\n",
      "training time:  13.2861 s.\n",
      "average model starts\n",
      "average model ends:  0.1121 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.783 0.798]\n",
      "global epoch:  668\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1962 - accuracy: 0.9371\n",
      "training time:  11.123 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.798 0.79 ]\n",
      "global epoch:  669\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1955 - accuracy: 0.9349\n",
      "training time:  13.306 s.\n",
      "average model starts\n",
      "average model ends:  0.0414 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.79  0.796]\n",
      "global epoch:  670\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1751 - accuracy: 0.9421\n",
      "training time:  13.1079 s.\n",
      "average model starts\n",
      "average model ends:  0.1135 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.796 0.793]\n",
      "global epoch:  671\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1816 - accuracy: 0.9376\n",
      "training time:  11.9814 s.\n",
      "average model starts\n",
      "average model ends:  0.1795 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.798]\n",
      "global epoch:  672\n",
      "virtual memory used:  71.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1938 - accuracy: 0.9379\n",
      "training time:  11.8691 s.\n",
      "average model starts\n",
      "average model ends:  0.0482 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.792]\n",
      "global epoch:  673\n",
      "virtual memory used:  72.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1801 - accuracy: 0.9404\n",
      "training time:  13.199 s.\n",
      "average model starts\n",
      "average model ends:  0.0549 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.792]\n",
      "global epoch:  674\n",
      "virtual memory used:  72.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1906 - accuracy: 0.9386\n",
      "training time:  12.9927 s.\n",
      "average model starts\n",
      "average model ends:  0.1228 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.798]\n",
      "global epoch:  675\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1700 - accuracy: 0.9443\n",
      "training time:  11.9782 s.\n",
      "average model starts\n",
      "average model ends:  0.0423 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.798 0.799]\n",
      "global epoch:  676\n",
      "virtual memory used:  73.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1897 - accuracy: 0.9370\n",
      "training time:  12.8439 s.\n",
      "average model starts\n",
      "average model ends:  0.1382 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.794]\n",
      "global epoch:  677\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1909 - accuracy: 0.9381\n",
      "training time:  12.8271 s.\n",
      "average model starts\n",
      "average model ends:  0.0467 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.797]\n",
      "global epoch:  678\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1876 - accuracy: 0.9356\n",
      "training time:  13.0098 s.\n",
      "average model starts\n",
      "average model ends:  0.04 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.804]\n",
      "global epoch:  679\n",
      "virtual memory used:  75.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1815 - accuracy: 0.9402\n",
      "training time:  12.1619 s.\n",
      "average model starts\n",
      "average model ends:  0.1266 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.804 0.8  ]\n",
      "global epoch:  680\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1820 - accuracy: 0.9402\n",
      "training time:  13.042 s.\n",
      "average model starts\n",
      "average model ends:  0.1371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.8   0.796]\n",
      "global epoch:  681\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1794 - accuracy: 0.9426\n",
      "training time:  12.5155 s.\n",
      "average model starts\n",
      "average model ends:  0.0556 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.788]\n",
      "global epoch:  682\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1898 - accuracy: 0.9400\n",
      "training time:  13.1363 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.802]\n",
      "global epoch:  683\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1889 - accuracy: 0.9361\n",
      "training time:  11.4225 s.\n",
      "average model starts\n",
      "average model ends:  0.0759 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.802 0.797]\n",
      "global epoch:  684\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1853 - accuracy: 0.9374\n",
      "training time:  11.3443 s.\n",
      "average model starts\n",
      "average model ends:  0.0709 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.795]\n",
      "global epoch:  685\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1937 - accuracy: 0.9351\n",
      "training time:  13.152 s.\n",
      "average model starts\n",
      "average model ends:  0.1087 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.8  ]\n",
      "global epoch:  686\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1880 - accuracy: 0.9360\n",
      "training time:  11.4637 s.\n",
      "average model starts\n",
      "average model ends:  0.0403 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.799]\n",
      "global epoch:  687\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1913 - accuracy: 0.9358\n",
      "training time:  11.5468 s.\n",
      "average model starts\n",
      "average model ends:  0.0523 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.799 0.786]\n",
      "global epoch:  688\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1918 - accuracy: 0.9405\n",
      "training time:  12.0363 s.\n",
      "average model starts\n",
      "average model ends:  0.0931 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.786 0.79 ]\n",
      "global epoch:  689\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1843 - accuracy: 0.9381\n",
      "training time:  12.5409 s.\n",
      "average model starts\n",
      "average model ends:  0.0611 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.79  0.801]\n",
      "global epoch:  690\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1749 - accuracy: 0.9444\n",
      "training time:  14.3843 s.\n",
      "average model starts\n",
      "average model ends:  0.0478 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.801 0.795]\n",
      "global epoch:  691\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1903 - accuracy: 0.9374\n",
      "training time:  12.9359 s.\n",
      "average model starts\n",
      "average model ends:  0.083 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.801]\n",
      "global epoch:  692\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1848 - accuracy: 0.9407\n",
      "training time:  10.6411 s.\n",
      "average model starts\n",
      "average model ends:  0.0319 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.802]\n",
      "global epoch:  693\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.2012 - accuracy: 0.9355\n",
      "training time:  12.5413 s.\n",
      "average model starts\n",
      "average model ends:  0.1068 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.799]\n",
      "global epoch:  694\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1838 - accuracy: 0.9402\n",
      "training time:  13.8383 s.\n",
      "average model starts\n",
      "average model ends:  0.0601 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.793]\n",
      "global epoch:  695\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1827 - accuracy: 0.9401\n",
      "training time:  11.2172 s.\n",
      "average model starts\n",
      "average model ends:  0.0307 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.8  ]\n",
      "global epoch:  696\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1879 - accuracy: 0.9358\n",
      "training time:  11.6903 s.\n",
      "average model starts\n",
      "average model ends:  0.0493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.802]\n",
      "global epoch:  697\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1827 - accuracy: 0.9420\n",
      "training time:  13.0424 s.\n",
      "average model starts\n",
      "average model ends:  0.0526 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.792]\n",
      "global epoch:  698\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1853 - accuracy: 0.9415\n",
      "training time:  13.5744 s.\n",
      "average model starts\n",
      "average model ends:  0.0465 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.798]\n",
      "global epoch:  699\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1861 - accuracy: 0.9376\n",
      "training time:  13.7368 s.\n",
      "average model starts\n",
      "average model ends:  0.0359 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.798 0.8  ]\n",
      "global epoch:  700\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1940 - accuracy: 0.9335\n",
      "training time:  12.8556 s.\n",
      "average model starts\n",
      "average model ends:  0.0544 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.801]\n",
      "global epoch:  701\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1890 - accuracy: 0.9360\n",
      "training time:  12.4112 s.\n",
      "average model starts\n",
      "average model ends:  0.0946 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.8  ]\n",
      "global epoch:  702\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1867 - accuracy: 0.9368\n",
      "training time:  12.9055 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.796]\n",
      "global epoch:  703\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1867 - accuracy: 0.9374\n",
      "training time:  12.1423 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.797]\n",
      "global epoch:  704\n",
      "virtual memory used:  75.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1810 - accuracy: 0.9405\n",
      "training time:  12.4991 s.\n",
      "average model starts\n",
      "average model ends:  0.0685 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.794]\n",
      "global epoch:  705\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1941 - accuracy: 0.9385\n",
      "training time:  11.3875 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.796]\n",
      "global epoch:  706\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1815 - accuracy: 0.9413\n",
      "training time:  13.1287 s.\n",
      "average model starts\n",
      "average model ends:  0.0675 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.801]\n",
      "global epoch:  707\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1773 - accuracy: 0.9421\n",
      "training time:  12.9989 s.\n",
      "average model starts\n",
      "average model ends:  0.0668 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.794]\n",
      "global epoch:  708\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1968 - accuracy: 0.9350\n",
      "training time:  12.9016 s.\n",
      "average model starts\n",
      "average model ends:  0.1345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.797]\n",
      "global epoch:  709\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1853 - accuracy: 0.9370\n",
      "training time:  13.084 s.\n",
      "average model starts\n",
      "average model ends:  0.0429 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.798]\n",
      "global epoch:  710\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1924 - accuracy: 0.9363\n",
      "training time:  12.4055 s.\n",
      "average model starts\n",
      "average model ends:  0.0982 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.785]\n",
      "global epoch:  711\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1873 - accuracy: 0.9385\n",
      "training time:  13.1259 s.\n",
      "average model starts\n",
      "average model ends:  0.0808 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.785 0.801]\n",
      "global epoch:  712\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1745 - accuracy: 0.9425\n",
      "training time:  10.9878 s.\n",
      "average model starts\n",
      "average model ends:  0.1063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.801 0.798]\n",
      "global epoch:  713\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1743 - accuracy: 0.9425\n",
      "training time:  10.8361 s.\n",
      "average model starts\n",
      "average model ends:  0.0704 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.791]\n",
      "global epoch:  714\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1788 - accuracy: 0.9425\n",
      "training time:  12.9988 s.\n",
      "average model starts\n",
      "average model ends:  0.0559 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.795]\n",
      "global epoch:  715\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1868 - accuracy: 0.9404\n",
      "training time:  11.1622 s.\n",
      "average model starts\n",
      "average model ends:  0.0577 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.795 0.798]\n",
      "global epoch:  716\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1890 - accuracy: 0.9404\n",
      "training time:  11.3795 s.\n",
      "average model starts\n",
      "average model ends:  0.0994 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.791]\n",
      "global epoch:  717\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1871 - accuracy: 0.9396\n",
      "training time:  13.2321 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.799]\n",
      "global epoch:  718\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1788 - accuracy: 0.9401\n",
      "training time:  13.4799 s.\n",
      "average model starts\n",
      "average model ends:  0.0598 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.79 ]\n",
      "global epoch:  719\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1781 - accuracy: 0.9428\n",
      "training time:  13.2273 s.\n",
      "average model starts\n",
      "average model ends:  0.0655 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.79  0.792]\n",
      "global epoch:  720\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1901 - accuracy: 0.9394\n",
      "training time:  13.3481 s.\n",
      "average model starts\n",
      "average model ends:  0.0545 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.792 0.8  ]\n",
      "global epoch:  721\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1933 - accuracy: 0.9348\n",
      "training time:  12.6042 s.\n",
      "average model starts\n",
      "average model ends:  0.1005 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.8   0.792]\n",
      "global epoch:  722\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1855 - accuracy: 0.9389\n",
      "training time:  12.2065 s.\n",
      "average model starts\n",
      "average model ends:  0.0633 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.792 0.795]\n",
      "global epoch:  723\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1815 - accuracy: 0.9386\n",
      "training time:  11.9364 s.\n",
      "average model starts\n",
      "average model ends:  0.0688 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.799]\n",
      "global epoch:  724\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1915 - accuracy: 0.9364\n",
      "training time:  12.1486 s.\n",
      "average model starts\n",
      "average model ends:  0.1264 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.802]\n",
      "global epoch:  725\n",
      "virtual memory used:  72.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1766 - accuracy: 0.9400\n",
      "training time:  12.4827 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.797]\n",
      "global epoch:  726\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1799 - accuracy: 0.9406\n",
      "training time:  13.4651 s.\n",
      "average model starts\n",
      "average model ends:  0.0469 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.797]\n",
      "global epoch:  727\n",
      "virtual memory used:  73.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1872 - accuracy: 0.9384\n",
      "training time:  12.3186 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.798]\n",
      "global epoch:  728\n",
      "virtual memory used:  73.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1830 - accuracy: 0.9400\n",
      "training time:  13.8237 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.797]\n",
      "global epoch:  729\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1887 - accuracy: 0.9391\n",
      "training time:  13.3339 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.785]\n",
      "global epoch:  730\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1736 - accuracy: 0.9422\n",
      "training time:  12.1009 s.\n",
      "average model starts\n",
      "average model ends:  0.0431 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.785 0.794]\n",
      "global epoch:  731\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1723 - accuracy: 0.9423\n",
      "training time:  13.7634 s.\n",
      "average model starts\n",
      "average model ends:  0.0736 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.794 0.795]\n",
      "global epoch:  732\n",
      "virtual memory used:  74.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1911 - accuracy: 0.9379\n",
      "training time:  14.1987 s.\n",
      "average model starts\n",
      "average model ends:  0.102 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.796]\n",
      "global epoch:  733\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1870 - accuracy: 0.9356\n",
      "training time:  13.1347 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.789]\n",
      "global epoch:  734\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1849 - accuracy: 0.9387\n",
      "training time:  12.1316 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.789 0.795]\n",
      "global epoch:  735\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1869 - accuracy: 0.9404\n",
      "training time:  12.7658 s.\n",
      "average model starts\n",
      "average model ends:  0.1013 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.795 0.793]\n",
      "global epoch:  736\n",
      "virtual memory used:  74.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1809 - accuracy: 0.9404\n",
      "training time:  12.693 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.796]\n",
      "global epoch:  737\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1827 - accuracy: 0.9390\n",
      "training time:  13.0077 s.\n",
      "average model starts\n",
      "average model ends:  0.1052 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.798]\n",
      "global epoch:  738\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1863 - accuracy: 0.9381\n",
      "training time:  12.2234 s.\n",
      "average model starts\n",
      "average model ends:  0.0328 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.798]\n",
      "global epoch:  739\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1700 - accuracy: 0.9420\n",
      "training time:  12.5066 s.\n",
      "average model starts\n",
      "average model ends:  0.0426 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.794]\n",
      "global epoch:  740\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1883 - accuracy: 0.9408\n",
      "training time:  12.6951 s.\n",
      "average model starts\n",
      "average model ends:  0.0566 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.794 0.793]\n",
      "global epoch:  741\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1706 - accuracy: 0.9437\n",
      "training time:  11.5066 s.\n",
      "average model starts\n",
      "average model ends:  0.0489 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.8  ]\n",
      "global epoch:  742\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1878 - accuracy: 0.9376\n",
      "training time:  12.1151 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.793]\n",
      "global epoch:  743\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1646 - accuracy: 0.9447\n",
      "training time:  12.2036 s.\n",
      "average model starts\n",
      "average model ends:  0.098 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.795]\n",
      "global epoch:  744\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1877 - accuracy: 0.9360\n",
      "training time:  13.2241 s.\n",
      "average model starts\n",
      "average model ends:  0.1434 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.796]\n",
      "global epoch:  745\n",
      "virtual memory used:  75.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1816 - accuracy: 0.9410\n",
      "training time:  11.396 s.\n",
      "average model starts\n",
      "average model ends:  0.077 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.791]\n",
      "global epoch:  746\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1895 - accuracy: 0.9361\n",
      "training time:  13.2431 s.\n",
      "average model starts\n",
      "average model ends:  0.0599 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.785]\n",
      "global epoch:  747\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1880 - accuracy: 0.9387\n",
      "training time:  13.8799 s.\n",
      "average model starts\n",
      "average model ends:  0.0383 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.785 0.783]\n",
      "global epoch:  748\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1863 - accuracy: 0.9408\n",
      "training time:  13.8874 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.783 0.797]\n",
      "global epoch:  749\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1867 - accuracy: 0.9389\n",
      "training time:  12.6723 s.\n",
      "average model starts\n",
      "average model ends:  0.1017 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.797 0.797]\n",
      "global epoch:  750\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1905 - accuracy: 0.9386\n",
      "training time:  12.5281 s.\n",
      "average model starts\n",
      "average model ends:  0.0638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.798]\n",
      "global epoch:  751\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1740 - accuracy: 0.9405\n",
      "training time:  11.7363 s.\n",
      "average model starts\n",
      "average model ends:  0.0562 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.79 ]\n",
      "global epoch:  752\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1845 - accuracy: 0.9418\n",
      "training time:  12.8036 s.\n",
      "average model starts\n",
      "average model ends:  0.062 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.79  0.795]\n",
      "global epoch:  753\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1893 - accuracy: 0.9395\n",
      "training time:  13.1926 s.\n",
      "average model starts\n",
      "average model ends:  0.0551 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.795 0.795]\n",
      "global epoch:  754\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1937 - accuracy: 0.9353\n",
      "training time:  11.3053 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.799]\n",
      "global epoch:  755\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1793 - accuracy: 0.9426\n",
      "training time:  12.4946 s.\n",
      "average model starts\n",
      "average model ends:  0.0355 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.79 ]\n",
      "global epoch:  756\n",
      "virtual memory used:  77.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1832 - accuracy: 0.9415\n",
      "training time:  13.1321 s.\n",
      "average model starts\n",
      "average model ends:  0.0424 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.79  0.8  ]\n",
      "global epoch:  757\n",
      "virtual memory used:  77.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1789 - accuracy: 0.9394\n",
      "training time:  11.9649 s.\n",
      "average model starts\n",
      "average model ends:  0.0417 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.8   0.792]\n",
      "global epoch:  758\n",
      "virtual memory used:  78.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1889 - accuracy: 0.9407\n",
      "training time:  12.4242 s.\n",
      "average model starts\n",
      "average model ends:  0.0492 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.792 0.793]\n",
      "global epoch:  759\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1825 - accuracy: 0.9394\n",
      "training time:  11.2189 s.\n",
      "average model starts\n",
      "average model ends:  0.0376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.793 0.799]\n",
      "global epoch:  760\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1751 - accuracy: 0.9435\n",
      "training time:  11.762 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.799 0.793]\n",
      "global epoch:  761\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1851 - accuracy: 0.9382\n",
      "training time:  12.389 s.\n",
      "average model starts\n",
      "average model ends:  0.0649 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.791]\n",
      "global epoch:  762\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1797 - accuracy: 0.9431\n",
      "training time:  12.5692 s.\n",
      "average model starts\n",
      "average model ends:  0.0541 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.791 0.791]\n",
      "global epoch:  763\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1821 - accuracy: 0.9394\n",
      "training time:  13.2752 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.791 0.799]\n",
      "global epoch:  764\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1798 - accuracy: 0.9411\n",
      "training time:  12.7427 s.\n",
      "average model starts\n",
      "average model ends:  0.0463 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.805]\n",
      "global epoch:  765\n",
      "virtual memory used:  77.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1839 - accuracy: 0.9413\n",
      "training time:  13.4454 s.\n",
      "average model starts\n",
      "average model ends:  0.0775 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.805 0.788]\n",
      "global epoch:  766\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1818 - accuracy: 0.9406\n",
      "training time:  13.519 s.\n",
      "average model starts\n",
      "average model ends:  0.08 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.788 0.795]\n",
      "global epoch:  767\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1683 - accuracy: 0.9438\n",
      "training time:  12.5608 s.\n",
      "average model starts\n",
      "average model ends:  0.0436 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.795 0.805]\n",
      "global epoch:  768\n",
      "virtual memory used:  79.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1724 - accuracy: 0.9401\n",
      "training time:  11.1319 s.\n",
      "average model starts\n",
      "average model ends:  0.0515 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.805 0.794]\n",
      "global epoch:  769\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1703 - accuracy: 0.9449\n",
      "training time:  11.1572 s.\n",
      "average model starts\n",
      "average model ends:  0.0353 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.794 0.802]\n",
      "global epoch:  770\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1755 - accuracy: 0.9418\n",
      "training time:  11.559 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.802 0.799]\n",
      "global epoch:  771\n",
      "virtual memory used:  79.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1831 - accuracy: 0.9411\n",
      "training time:  11.5932 s.\n",
      "average model starts\n",
      "average model ends:  0.0628 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.802]\n",
      "global epoch:  772\n",
      "virtual memory used:  79.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1792 - accuracy: 0.9417\n",
      "training time:  11.5691 s.\n",
      "average model starts\n",
      "average model ends:  0.0548 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.793]\n",
      "global epoch:  773\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1836 - accuracy: 0.9411\n",
      "training time:  11.0926 s.\n",
      "average model starts\n",
      "average model ends:  0.0516 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.793 0.796]\n",
      "global epoch:  774\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1913 - accuracy: 0.9355\n",
      "training time:  13.5629 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.806]\n",
      "global epoch:  775\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1743 - accuracy: 0.9423\n",
      "training time:  11.4495 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.806 0.803]\n",
      "global epoch:  776\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1806 - accuracy: 0.9428\n",
      "training time:  11.0483 s.\n",
      "average model starts\n",
      "average model ends:  0.0445 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.803 0.795]\n",
      "global epoch:  777\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1905 - accuracy: 0.9387\n",
      "training time:  13.0506 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.795]\n",
      "global epoch:  778\n",
      "virtual memory used:  73.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1861 - accuracy: 0.9397\n",
      "training time:  11.2566 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.802]\n",
      "global epoch:  779\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1874 - accuracy: 0.9381\n",
      "training time:  13.1122 s.\n",
      "average model starts\n",
      "average model ends:  0.0392 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.791]\n",
      "global epoch:  780\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1967 - accuracy: 0.9361\n",
      "training time:  13.2559 s.\n",
      "average model starts\n",
      "average model ends:  0.1263 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.791 0.796]\n",
      "global epoch:  781\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1876 - accuracy: 0.9443\n",
      "training time:  12.0875 s.\n",
      "average model starts\n",
      "average model ends:  0.0314 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.788]\n",
      "global epoch:  782\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1749 - accuracy: 0.9420\n",
      "training time:  13.3772 s.\n",
      "average model starts\n",
      "average model ends:  0.0318 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.788 0.798]\n",
      "global epoch:  783\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1739 - accuracy: 0.9406\n",
      "training time:  12.9461 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.798 0.804]\n",
      "global epoch:  784\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1756 - accuracy: 0.9426\n",
      "training time:  12.3737 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.804 0.8  ]\n",
      "global epoch:  785\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1908 - accuracy: 0.9375\n",
      "training time:  13.3946 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.8   0.791]\n",
      "global epoch:  786\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1814 - accuracy: 0.9416\n",
      "training time:  12.1778 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.791 0.795]\n",
      "global epoch:  787\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1789 - accuracy: 0.9410\n",
      "training time:  12.9403 s.\n",
      "average model starts\n",
      "average model ends:  0.0774 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.795 0.799]\n",
      "global epoch:  788\n",
      "virtual memory used:  74.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1878 - accuracy: 0.9397\n",
      "training time:  10.9859 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.796]\n",
      "global epoch:  789\n",
      "virtual memory used:  74.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1740 - accuracy: 0.9404\n",
      "training time:  13.3612 s.\n",
      "average model starts\n",
      "average model ends:  0.0566 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.797]\n",
      "global epoch:  790\n",
      "virtual memory used:  75.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1797 - accuracy: 0.9377\n",
      "training time:  13.1954 s.\n",
      "average model starts\n",
      "average model ends:  0.0608 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.801]\n",
      "global epoch:  791\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1871 - accuracy: 0.9390\n",
      "training time:  13.0889 s.\n",
      "average model starts\n",
      "average model ends:  0.0493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.796]\n",
      "global epoch:  792\n",
      "virtual memory used:  76.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1813 - accuracy: 0.9438\n",
      "training time:  13.5961 s.\n",
      "average model starts\n",
      "average model ends:  0.0652 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.803]\n",
      "global epoch:  793\n",
      "virtual memory used:  74.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1724 - accuracy: 0.9453\n",
      "training time:  12.8746 s.\n",
      "average model starts\n",
      "average model ends:  0.0313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.801]\n",
      "global epoch:  794\n",
      "virtual memory used:  75.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1781 - accuracy: 0.9396\n",
      "training time:  11.9904 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.802]\n",
      "global epoch:  795\n",
      "virtual memory used:  75.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1878 - accuracy: 0.9384\n",
      "training time:  12.5328 s.\n",
      "average model starts\n",
      "average model ends:  0.1064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.797]\n",
      "global epoch:  796\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1781 - accuracy: 0.9413\n",
      "training time:  13.1426 s.\n",
      "average model starts\n",
      "average model ends:  0.1104 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.791]\n",
      "global epoch:  797\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1762 - accuracy: 0.9415\n",
      "training time:  12.1947 s.\n",
      "average model starts\n",
      "average model ends:  0.0491 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.785]\n",
      "global epoch:  798\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1873 - accuracy: 0.9385\n",
      "training time:  13.3979 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.785 0.785]\n",
      "global epoch:  799\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1830 - accuracy: 0.9392\n",
      "training time:  13.4912 s.\n",
      "average model starts\n",
      "average model ends:  0.1073 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.785 0.801]\n",
      "global epoch:  800\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1909 - accuracy: 0.9363\n",
      "training time:  13.2333 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.801 0.799]\n",
      "global epoch:  801\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1823 - accuracy: 0.9396\n",
      "training time:  11.4746 s.\n",
      "average model starts\n",
      "average model ends:  0.0523 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.795]\n",
      "global epoch:  802\n",
      "virtual memory used:  75.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1951 - accuracy: 0.9368\n",
      "training time:  12.6021 s.\n",
      "average model starts\n",
      "average model ends:  0.0764 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.798]\n",
      "global epoch:  803\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1815 - accuracy: 0.9374\n",
      "training time:  12.8056 s.\n",
      "average model starts\n",
      "average model ends:  0.0504 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.793]\n",
      "global epoch:  804\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1798 - accuracy: 0.9411\n",
      "training time:  13.8768 s.\n",
      "average model starts\n",
      "average model ends:  0.1099 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.793 0.8  ]\n",
      "global epoch:  805\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1886 - accuracy: 0.9364\n",
      "training time:  12.5022 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.795]\n",
      "global epoch:  806\n",
      "virtual memory used:  76.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1894 - accuracy: 0.9390\n",
      "training time:  12.2435 s.\n",
      "average model starts\n",
      "average model ends:  0.0638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.794]\n",
      "global epoch:  807\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 54ms/step - loss: 0.1762 - accuracy: 0.9420\n",
      "training time:  13.611 s.\n",
      "average model starts\n",
      "average model ends:  0.071 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.797]\n",
      "global epoch:  808\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1879 - accuracy: 0.9386\n",
      "training time:  12.0272 s.\n",
      "average model starts\n",
      "average model ends:  0.0729 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.8  ]\n",
      "global epoch:  809\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1798 - accuracy: 0.9433\n",
      "training time:  12.7471 s.\n",
      "average model starts\n",
      "average model ends:  0.0564 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.798]\n",
      "global epoch:  810\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1766 - accuracy: 0.9396\n",
      "training time:  12.2187 s.\n",
      "average model starts\n",
      "average model ends:  0.0382 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.792]\n",
      "global epoch:  811\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1845 - accuracy: 0.9407\n",
      "training time:  13.1378 s.\n",
      "average model starts\n",
      "average model ends:  0.0523 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.795]\n",
      "global epoch:  812\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1946 - accuracy: 0.9384\n",
      "training time:  12.7404 s.\n",
      "average model starts\n",
      "average model ends:  0.0438 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.801]\n",
      "global epoch:  813\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1750 - accuracy: 0.9446\n",
      "training time:  11.9267 s.\n",
      "average model starts\n",
      "average model ends:  0.0644 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.799]\n",
      "global epoch:  814\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1741 - accuracy: 0.9417\n",
      "training time:  12.1609 s.\n",
      "average model starts\n",
      "average model ends:  0.0862 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.804]\n",
      "global epoch:  815\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1784 - accuracy: 0.9396\n",
      "training time:  11.4017 s.\n",
      "average model starts\n",
      "average model ends:  0.0774 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.804 0.795]\n",
      "global epoch:  816\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1845 - accuracy: 0.9412\n",
      "training time:  11.5698 s.\n",
      "average model starts\n",
      "average model ends:  0.034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.795 0.792]\n",
      "global epoch:  817\n",
      "virtual memory used:  78.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1930 - accuracy: 0.9380\n",
      "training time:  12.4549 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.792 0.792]\n",
      "global epoch:  818\n",
      "virtual memory used:  79.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1901 - accuracy: 0.9412\n",
      "training time:  11.4809 s.\n",
      "average model starts\n",
      "average model ends:  0.1191 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.799]\n",
      "global epoch:  819\n",
      "virtual memory used:  78.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1857 - accuracy: 0.9400\n",
      "training time:  12.0846 s.\n",
      "average model starts\n",
      "average model ends:  0.0527 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.789]\n",
      "global epoch:  820\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1856 - accuracy: 0.9389\n",
      "training time:  11.9164 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.789 0.784]\n",
      "global epoch:  821\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1698 - accuracy: 0.9436\n",
      "training time:  13.0342 s.\n",
      "average model starts\n",
      "average model ends:  0.0356 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.784 0.799]\n",
      "global epoch:  822\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1886 - accuracy: 0.9397\n",
      "training time:  11.3522 s.\n",
      "average model starts\n",
      "average model ends:  0.0495 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.799 0.793]\n",
      "global epoch:  823\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1854 - accuracy: 0.9381\n",
      "training time:  11.0933 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.798]\n",
      "global epoch:  824\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1687 - accuracy: 0.9479\n",
      "training time:  11.6702 s.\n",
      "average model starts\n",
      "average model ends:  0.0537 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.792]\n",
      "global epoch:  825\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1806 - accuracy: 0.9418\n",
      "training time:  13.2354 s.\n",
      "average model starts\n",
      "average model ends:  0.0444 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.8  ]\n",
      "global epoch:  826\n",
      "virtual memory used:  80.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1866 - accuracy: 0.9358\n",
      "training time:  13.3346 s.\n",
      "average model starts\n",
      "average model ends:  0.0364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.8   0.802]\n",
      "global epoch:  827\n",
      "virtual memory used:  80.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1865 - accuracy: 0.9368\n",
      "training time:  11.3751 s.\n",
      "average model starts\n",
      "average model ends:  0.0807 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.8  ]\n",
      "global epoch:  828\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1779 - accuracy: 0.9407\n",
      "training time:  11.6748 s.\n",
      "average model starts\n",
      "average model ends:  0.0394 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.802]\n",
      "global epoch:  829\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1864 - accuracy: 0.9404\n",
      "training time:  12.1304 s.\n",
      "average model starts\n",
      "average model ends:  0.04 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.797]\n",
      "global epoch:  830\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1979 - accuracy: 0.9379\n",
      "training time:  12.5284 s.\n",
      "average model starts\n",
      "average model ends:  0.1077 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.798]\n",
      "global epoch:  831\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1756 - accuracy: 0.9410\n",
      "training time:  13.1704 s.\n",
      "average model starts\n",
      "average model ends:  0.0616 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.797]\n",
      "global epoch:  832\n",
      "virtual memory used:  81.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1899 - accuracy: 0.9412\n",
      "training time:  13.7586 s.\n",
      "average model starts\n",
      "average model ends:  0.0345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.801]\n",
      "global epoch:  833\n",
      "virtual memory used:  73.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1797 - accuracy: 0.9418\n",
      "training time:  12.7753 s.\n",
      "average model starts\n",
      "average model ends:  0.0519 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.804]\n",
      "global epoch:  834\n",
      "virtual memory used:  74.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1753 - accuracy: 0.9428\n",
      "training time:  11.1018 s.\n",
      "average model starts\n",
      "average model ends:  0.0696 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.804]\n",
      "global epoch:  835\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1833 - accuracy: 0.9405\n",
      "training time:  11.1407 s.\n",
      "average model starts\n",
      "average model ends:  0.0338 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.804 0.801]\n",
      "global epoch:  836\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1811 - accuracy: 0.9430\n",
      "training time:  12.5068 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.799]\n",
      "global epoch:  837\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1883 - accuracy: 0.9370\n",
      "training time:  12.6771 s.\n",
      "average model starts\n",
      "average model ends:  0.0421 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.796]\n",
      "global epoch:  838\n",
      "virtual memory used:  74.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1917 - accuracy: 0.9376\n",
      "training time:  11.1959 s.\n",
      "average model starts\n",
      "average model ends:  0.0491 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.787]\n",
      "global epoch:  839\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1699 - accuracy: 0.9431\n",
      "training time:  13.2265 s.\n",
      "average model starts\n",
      "average model ends:  0.0506 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.787 0.799]\n",
      "global epoch:  840\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1807 - accuracy: 0.9400\n",
      "training time:  13.7841 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.799 0.802]\n",
      "global epoch:  841\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1931 - accuracy: 0.9361\n",
      "training time:  11.377 s.\n",
      "average model starts\n",
      "average model ends:  0.0345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.804]\n",
      "global epoch:  842\n",
      "virtual memory used:  74.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1760 - accuracy: 0.9422\n",
      "training time:  11.0108 s.\n",
      "average model starts\n",
      "average model ends:  0.0345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.804 0.794]\n",
      "global epoch:  843\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1941 - accuracy: 0.9381\n",
      "training time:  11.2472 s.\n",
      "average model starts\n",
      "average model ends:  0.0537 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.794 0.793]\n",
      "global epoch:  844\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1778 - accuracy: 0.9415\n",
      "training time:  12.8128 s.\n",
      "average model starts\n",
      "average model ends:  0.067 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.789]\n",
      "global epoch:  845\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1866 - accuracy: 0.9407\n",
      "training time:  13.3298 s.\n",
      "average model starts\n",
      "average model ends:  0.1407 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.789 0.802]\n",
      "global epoch:  846\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1852 - accuracy: 0.9370\n",
      "training time:  13.0634 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.802 0.794]\n",
      "global epoch:  847\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1835 - accuracy: 0.9404\n",
      "training time:  11.6242 s.\n",
      "average model starts\n",
      "average model ends:  0.0477 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.791]\n",
      "global epoch:  848\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1720 - accuracy: 0.9410\n",
      "training time:  12.3979 s.\n",
      "average model starts\n",
      "average model ends:  0.1327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.806]\n",
      "global epoch:  849\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1643 - accuracy: 0.9447\n",
      "training time:  12.3985 s.\n",
      "average model starts\n",
      "average model ends:  0.0471 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.806 0.801]\n",
      "global epoch:  850\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1904 - accuracy: 0.9390\n",
      "training time:  12.9772 s.\n",
      "average model starts\n",
      "average model ends:  0.0505 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.801 0.8  ]\n",
      "global epoch:  851\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1867 - accuracy: 0.9405\n",
      "training time:  12.8511 s.\n",
      "average model starts\n",
      "average model ends:  0.0754 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.795]\n",
      "global epoch:  852\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1816 - accuracy: 0.9443\n",
      "training time:  12.0433 s.\n",
      "average model starts\n",
      "average model ends:  0.0485 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.801]\n",
      "global epoch:  853\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1688 - accuracy: 0.9436\n",
      "training time:  11.1401 s.\n",
      "average model starts\n",
      "average model ends:  0.0524 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.804]\n",
      "global epoch:  854\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1833 - accuracy: 0.9376\n",
      "training time:  11.342 s.\n",
      "average model starts\n",
      "average model ends:  0.0486 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.797]\n",
      "global epoch:  855\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1756 - accuracy: 0.9411\n",
      "training time:  12.9217 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.797 0.803]\n",
      "global epoch:  856\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1756 - accuracy: 0.9444\n",
      "training time:  12.0279 s.\n",
      "average model starts\n",
      "average model ends:  0.0989 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.803 0.794]\n",
      "global epoch:  857\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1771 - accuracy: 0.9427\n",
      "training time:  13.0837 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.794 0.802]\n",
      "global epoch:  858\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1713 - accuracy: 0.9453\n",
      "training time:  13.3682 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.802 0.799]\n",
      "global epoch:  859\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1796 - accuracy: 0.9408\n",
      "training time:  12.6978 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.8  ]\n",
      "global epoch:  860\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1824 - accuracy: 0.9394\n",
      "training time:  12.9641 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.79 ]\n",
      "global epoch:  861\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1772 - accuracy: 0.9415\n",
      "training time:  11.9182 s.\n",
      "average model starts\n",
      "average model ends:  0.053 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.79  0.797]\n",
      "global epoch:  862\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1878 - accuracy: 0.9382\n",
      "training time:  13.3092 s.\n",
      "average model starts\n",
      "average model ends:  0.1391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.797 0.796]\n",
      "global epoch:  863\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1700 - accuracy: 0.9464\n",
      "training time:  12.6051 s.\n",
      "average model starts\n",
      "average model ends:  0.0676 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.791]\n",
      "global epoch:  864\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1666 - accuracy: 0.9480\n",
      "training time:  13.7216 s.\n",
      "average model starts\n",
      "average model ends:  0.0804 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.783]\n",
      "global epoch:  865\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1774 - accuracy: 0.9413\n",
      "training time:  11.7674 s.\n",
      "average model starts\n",
      "average model ends:  0.105 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.783 0.795]\n",
      "global epoch:  866\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1786 - accuracy: 0.9400\n",
      "training time:  12.6791 s.\n",
      "average model starts\n",
      "average model ends:  0.0364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.795 0.794]\n",
      "global epoch:  867\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1773 - accuracy: 0.9432\n",
      "training time:  12.1537 s.\n",
      "average model starts\n",
      "average model ends:  0.0575 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.795]\n",
      "global epoch:  868\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1807 - accuracy: 0.9368\n",
      "training time:  13.0 s.\n",
      "average model starts\n",
      "average model ends:  0.0427 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.805]\n",
      "global epoch:  869\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1735 - accuracy: 0.9408\n",
      "training time:  12.1371 s.\n",
      "average model starts\n",
      "average model ends:  0.0844 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.805 0.799]\n",
      "global epoch:  870\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1852 - accuracy: 0.9390\n",
      "training time:  11.8762 s.\n",
      "average model starts\n",
      "average model ends:  0.0501 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.799 0.805]\n",
      "global epoch:  871\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1836 - accuracy: 0.9377\n",
      "training time:  11.202 s.\n",
      "average model starts\n",
      "average model ends:  0.0412 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.805 0.801]\n",
      "global epoch:  872\n",
      "virtual memory used:  79.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1812 - accuracy: 0.9406\n",
      "training time:  12.7876 s.\n",
      "average model starts\n",
      "average model ends:  0.0517 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.801 0.798]\n",
      "global epoch:  873\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1783 - accuracy: 0.9417\n",
      "training time:  11.4895 s.\n",
      "average model starts\n",
      "average model ends:  0.0921 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.797]\n",
      "global epoch:  874\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1761 - accuracy: 0.9417\n",
      "training time:  12.1172 s.\n",
      "average model starts\n",
      "average model ends:  0.0874 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.79 ]\n",
      "global epoch:  875\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1699 - accuracy: 0.9426\n",
      "training time:  12.5743 s.\n",
      "average model starts\n",
      "average model ends:  0.0543 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.79  0.802]\n",
      "global epoch:  876\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1844 - accuracy: 0.9435\n",
      "training time:  13.6444 s.\n",
      "average model starts\n",
      "average model ends:  0.0937 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.792]\n",
      "global epoch:  877\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1665 - accuracy: 0.9472\n",
      "training time:  13.7301 s.\n",
      "average model starts\n",
      "average model ends:  0.0541 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.796]\n",
      "global epoch:  878\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1932 - accuracy: 0.9385\n",
      "training time:  12.4466 s.\n",
      "average model starts\n",
      "average model ends:  0.0646 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.796 0.796]\n",
      "global epoch:  879\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1740 - accuracy: 0.9420\n",
      "training time:  11.2 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.783]\n",
      "global epoch:  880\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1755 - accuracy: 0.9437\n",
      "training time:  11.6083 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.783 0.795]\n",
      "global epoch:  881\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1888 - accuracy: 0.9337\n",
      "training time:  12.6627 s.\n",
      "average model starts\n",
      "average model ends:  0.1761 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.783 0.795 0.794]\n",
      "global epoch:  882\n",
      "virtual memory used:  74.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1697 - accuracy: 0.9443\n",
      "training time:  12.83 s.\n",
      "average model starts\n",
      "average model ends:  0.0576 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.798]\n",
      "global epoch:  883\n",
      "virtual memory used:  74.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1796 - accuracy: 0.9395\n",
      "training time:  12.3494 s.\n",
      "average model starts\n",
      "average model ends:  0.0605 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.801]\n",
      "global epoch:  884\n",
      "virtual memory used:  75.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1720 - accuracy: 0.9415\n",
      "training time:  14.5849 s.\n",
      "average model starts\n",
      "average model ends:  0.0637 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.8  ]\n",
      "global epoch:  885\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1822 - accuracy: 0.9399\n",
      "training time:  11.0825 s.\n",
      "average model starts\n",
      "average model ends:  0.0815 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.791]\n",
      "global epoch:  886\n",
      "virtual memory used:  74.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1809 - accuracy: 0.9411\n",
      "training time:  10.8594 s.\n",
      "average model starts\n",
      "average model ends:  0.0332 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.791 0.801]\n",
      "global epoch:  887\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1795 - accuracy: 0.9408\n",
      "training time:  14.3637 s.\n",
      "average model starts\n",
      "average model ends:  0.0421 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.801 0.787]\n",
      "global epoch:  888\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1798 - accuracy: 0.9425\n",
      "training time:  12.4601 s.\n",
      "average model starts\n",
      "average model ends:  0.0411 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.787 0.796]\n",
      "global epoch:  889\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1788 - accuracy: 0.9395\n",
      "training time:  10.9731 s.\n",
      "average model starts\n",
      "average model ends:  0.0928 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.796 0.801]\n",
      "global epoch:  890\n",
      "virtual memory used:  75.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1713 - accuracy: 0.9458\n",
      "training time:  10.8878 s.\n",
      "average model starts\n",
      "average model ends:  0.0353 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.793]\n",
      "global epoch:  891\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1841 - accuracy: 0.9384\n",
      "training time:  12.0374 s.\n",
      "average model starts\n",
      "average model ends:  0.0813 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.793 0.796]\n",
      "global epoch:  892\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1779 - accuracy: 0.9410\n",
      "training time:  11.3481 s.\n",
      "average model starts\n",
      "average model ends:  0.0801 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.799]\n",
      "global epoch:  893\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1766 - accuracy: 0.9416\n",
      "training time:  13.869 s.\n",
      "average model starts\n",
      "average model ends:  0.0512 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.795]\n",
      "global epoch:  894\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1801 - accuracy: 0.9411\n",
      "training time:  11.872 s.\n",
      "average model starts\n",
      "average model ends:  0.0515 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.805]\n",
      "global epoch:  895\n",
      "virtual memory used:  75.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1770 - accuracy: 0.9420\n",
      "training time:  11.4007 s.\n",
      "average model starts\n",
      "average model ends:  0.0758 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.805 0.794]\n",
      "global epoch:  896\n",
      "virtual memory used:  76.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1870 - accuracy: 0.9400\n",
      "training time:  12.7323 s.\n",
      "average model starts\n",
      "average model ends:  0.0386 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.794 0.79 ]\n",
      "global epoch:  897\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1749 - accuracy: 0.9418\n",
      "training time:  12.0717 s.\n",
      "average model starts\n",
      "average model ends:  0.1059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.798]\n",
      "global epoch:  898\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1729 - accuracy: 0.9432\n",
      "training time:  11.9009 s.\n",
      "average model starts\n",
      "average model ends:  0.0537 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.798 0.8  ]\n",
      "global epoch:  899\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1711 - accuracy: 0.9446\n",
      "training time:  13.0055 s.\n",
      "average model starts\n",
      "average model ends:  0.0594 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.801]\n",
      "global epoch:  900\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1782 - accuracy: 0.9400\n",
      "training time:  13.1729 s.\n",
      "average model starts\n",
      "average model ends:  0.071 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.801]\n",
      "global epoch:  901\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1817 - accuracy: 0.9374\n",
      "training time:  12.2586 s.\n",
      "average model starts\n",
      "average model ends:  0.0943 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.795]\n",
      "global epoch:  902\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1867 - accuracy: 0.9385\n",
      "training time:  12.2893 s.\n",
      "average model starts\n",
      "average model ends:  0.0475 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.799]\n",
      "global epoch:  903\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1841 - accuracy: 0.9420\n",
      "training time:  11.9566 s.\n",
      "average model starts\n",
      "average model ends:  0.0428 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.801]\n",
      "global epoch:  904\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1595 - accuracy: 0.9448\n",
      "training time:  12.7758 s.\n",
      "average model starts\n",
      "average model ends:  0.0598 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.789]\n",
      "global epoch:  905\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1794 - accuracy: 0.9400\n",
      "training time:  10.6475 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.789 0.797]\n",
      "global epoch:  906\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1713 - accuracy: 0.9438\n",
      "training time:  12.5556 s.\n",
      "average model starts\n",
      "average model ends:  0.0611 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.797 0.801]\n",
      "global epoch:  907\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1725 - accuracy: 0.9402\n",
      "training time:  13.6989 s.\n",
      "average model starts\n",
      "average model ends:  0.0429 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.802]\n",
      "global epoch:  908\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1791 - accuracy: 0.9404\n",
      "training time:  12.7178 s.\n",
      "average model starts\n",
      "average model ends:  0.0377 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.801]\n",
      "global epoch:  909\n",
      "virtual memory used:  78.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1807 - accuracy: 0.9404\n",
      "training time:  12.451 s.\n",
      "average model starts\n",
      "average model ends:  0.1239 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.79 ]\n",
      "global epoch:  910\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1770 - accuracy: 0.9432\n",
      "training time:  10.9106 s.\n",
      "average model starts\n",
      "average model ends:  0.1284 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.79  0.804]\n",
      "global epoch:  911\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1742 - accuracy: 0.9447\n",
      "training time:  12.2157 s.\n",
      "average model starts\n",
      "average model ends:  0.0919 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.804 0.797]\n",
      "global epoch:  912\n",
      "virtual memory used:  78.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1758 - accuracy: 0.9435\n",
      "training time:  12.2999 s.\n",
      "average model starts\n",
      "average model ends:  0.0769 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.797 0.799]\n",
      "global epoch:  913\n",
      "virtual memory used:  78.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1804 - accuracy: 0.9371\n",
      "training time:  12.0372 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.796]\n",
      "global epoch:  914\n",
      "virtual memory used:  79.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1804 - accuracy: 0.9408\n",
      "training time:  11.1246 s.\n",
      "average model starts\n",
      "average model ends:  0.0897 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.797]\n",
      "global epoch:  915\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1895 - accuracy: 0.9377\n",
      "training time:  13.0798 s.\n",
      "average model starts\n",
      "average model ends:  0.0823 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.798]\n",
      "global epoch:  916\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1688 - accuracy: 0.9402\n",
      "training time:  13.4113 s.\n",
      "average model starts\n",
      "average model ends:  0.0483 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.797]\n",
      "global epoch:  917\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1754 - accuracy: 0.9422\n",
      "training time:  12.3206 s.\n",
      "average model starts\n",
      "average model ends:  0.0629 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.796]\n",
      "global epoch:  918\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1842 - accuracy: 0.9387\n",
      "training time:  12.2987 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.795]\n",
      "global epoch:  919\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1729 - accuracy: 0.9433\n",
      "training time:  12.9139 s.\n",
      "average model starts\n",
      "average model ends:  0.0781 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.797]\n",
      "global epoch:  920\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1884 - accuracy: 0.9380\n",
      "training time:  12.1508 s.\n",
      "average model starts\n",
      "average model ends:  0.1213 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.792]\n",
      "global epoch:  921\n",
      "virtual memory used:  79.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1742 - accuracy: 0.9441\n",
      "training time:  11.9738 s.\n",
      "average model starts\n",
      "average model ends:  0.1137 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.792 0.796]\n",
      "global epoch:  922\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1735 - accuracy: 0.9439\n",
      "training time:  11.8772 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.796 0.799]\n",
      "global epoch:  923\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1805 - accuracy: 0.9416\n",
      "training time:  11.6976 s.\n",
      "average model starts\n",
      "average model ends:  0.0865 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.794]\n",
      "global epoch:  924\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1829 - accuracy: 0.9381\n",
      "training time:  12.5621 s.\n",
      "average model starts\n",
      "average model ends:  0.1142 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.794 0.801]\n",
      "global epoch:  925\n",
      "virtual memory used:  79.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1881 - accuracy: 0.9406\n",
      "training time:  11.9275 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.802]\n",
      "global epoch:  926\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1693 - accuracy: 0.9430\n",
      "training time:  10.7903 s.\n",
      "average model starts\n",
      "average model ends:  0.0798 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.801]\n",
      "global epoch:  927\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1636 - accuracy: 0.9459\n",
      "training time:  11.1461 s.\n",
      "average model starts\n",
      "average model ends:  0.0688 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.795]\n",
      "global epoch:  928\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1837 - accuracy: 0.9417\n",
      "training time:  10.7759 s.\n",
      "average model starts\n",
      "average model ends:  0.043 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.807]\n",
      "global epoch:  929\n",
      "virtual memory used:  80.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1880 - accuracy: 0.9380\n",
      "training time:  13.0471 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.807 0.788]\n",
      "global epoch:  930\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1856 - accuracy: 0.9373\n",
      "training time:  10.581 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.788 0.803]\n",
      "global epoch:  931\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1778 - accuracy: 0.9433\n",
      "training time:  12.7608 s.\n",
      "average model starts\n",
      "average model ends:  0.0799 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.803 0.795]\n",
      "global epoch:  932\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1645 - accuracy: 0.9456\n",
      "training time:  12.3902 s.\n",
      "average model starts\n",
      "average model ends:  0.086 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.8  ]\n",
      "global epoch:  933\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1700 - accuracy: 0.9430\n",
      "training time:  11.6501 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.796]\n",
      "global epoch:  934\n",
      "virtual memory used:  74.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1781 - accuracy: 0.9416\n",
      "training time:  12.9977 s.\n",
      "average model starts\n",
      "average model ends:  0.0463 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.8  ]\n",
      "global epoch:  935\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1798 - accuracy: 0.9402\n",
      "training time:  11.6196 s.\n",
      "average model starts\n",
      "average model ends:  0.0546 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.802]\n",
      "global epoch:  936\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1693 - accuracy: 0.9435\n",
      "training time:  12.3358 s.\n",
      "average model starts\n",
      "average model ends:  0.0703 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.794]\n",
      "global epoch:  937\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1796 - accuracy: 0.9386\n",
      "training time:  11.9281 s.\n",
      "average model starts\n",
      "average model ends:  0.0539 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.793]\n",
      "global epoch:  938\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1762 - accuracy: 0.9415\n",
      "training time:  13.6095 s.\n",
      "average model starts\n",
      "average model ends:  0.0505 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.799]\n",
      "global epoch:  939\n",
      "virtual memory used:  75.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1803 - accuracy: 0.9405\n",
      "training time:  13.2033 s.\n",
      "average model starts\n",
      "average model ends:  0.0425 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.799 0.8  ]\n",
      "global epoch:  940\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1755 - accuracy: 0.9427\n",
      "training time:  12.535 s.\n",
      "average model starts\n",
      "average model ends:  0.0726 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.793]\n",
      "global epoch:  941\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1690 - accuracy: 0.9447\n",
      "training time:  13.2263 s.\n",
      "average model starts\n",
      "average model ends:  0.0303 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.804]\n",
      "global epoch:  942\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1835 - accuracy: 0.9394\n",
      "training time:  12.6671 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.804 0.801]\n",
      "global epoch:  943\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1750 - accuracy: 0.9430\n",
      "training time:  12.6036 s.\n",
      "average model starts\n",
      "average model ends:  0.0811 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.801]\n",
      "global epoch:  944\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1850 - accuracy: 0.9391\n",
      "training time:  12.351 s.\n",
      "average model starts\n",
      "average model ends:  0.0315 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.792]\n",
      "global epoch:  945\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1736 - accuracy: 0.9438\n",
      "training time:  11.8309 s.\n",
      "average model starts\n",
      "average model ends:  0.1263 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.792 0.794]\n",
      "global epoch:  946\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1748 - accuracy: 0.9441\n",
      "training time:  12.4705 s.\n",
      "average model starts\n",
      "average model ends:  0.0503 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.794 0.801]\n",
      "global epoch:  947\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1760 - accuracy: 0.9428\n",
      "training time:  12.8931 s.\n",
      "average model starts\n",
      "average model ends:  0.0399 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.796]\n",
      "global epoch:  948\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1776 - accuracy: 0.9411\n",
      "training time:  12.7295 s.\n",
      "average model starts\n",
      "average model ends:  0.0622 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.801]\n",
      "global epoch:  949\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1799 - accuracy: 0.9420\n",
      "training time:  10.6627 s.\n",
      "average model starts\n",
      "average model ends:  0.0518 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.803]\n",
      "global epoch:  950\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1801 - accuracy: 0.9446\n",
      "training time:  10.8897 s.\n",
      "average model starts\n",
      "average model ends:  0.0389 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.803 0.796]\n",
      "global epoch:  951\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1756 - accuracy: 0.9406\n",
      "training time:  12.208 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.799]\n",
      "global epoch:  952\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1709 - accuracy: 0.9432\n",
      "training time:  12.9211 s.\n",
      "average model starts\n",
      "average model ends:  0.0388 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.802]\n",
      "global epoch:  953\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1706 - accuracy: 0.9418\n",
      "training time:  11.7267 s.\n",
      "average model starts\n",
      "average model ends:  0.091 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.793]\n",
      "global epoch:  954\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1832 - accuracy: 0.9413\n",
      "training time:  12.5944 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.793 0.796]\n",
      "global epoch:  955\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1743 - accuracy: 0.9435\n",
      "training time:  11.6731 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.791]\n",
      "global epoch:  956\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1662 - accuracy: 0.9457\n",
      "training time:  12.5351 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.797]\n",
      "global epoch:  957\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1711 - accuracy: 0.9446\n",
      "training time:  13.0749 s.\n",
      "average model starts\n",
      "average model ends:  0.0819 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.797 0.796]\n",
      "global epoch:  958\n",
      "virtual memory used:  77.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1665 - accuracy: 0.9469\n",
      "training time:  12.1567 s.\n",
      "average model starts\n",
      "average model ends:  0.0352 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.795]\n",
      "global epoch:  959\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1753 - accuracy: 0.9422\n",
      "training time:  12.2917 s.\n",
      "average model starts\n",
      "average model ends:  0.0607 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.802]\n",
      "global epoch:  960\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1772 - accuracy: 0.9412\n",
      "training time:  11.5805 s.\n",
      "average model starts\n",
      "average model ends:  0.0344 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.801]\n",
      "global epoch:  961\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1840 - accuracy: 0.9385\n",
      "training time:  12.7074 s.\n",
      "average model starts\n",
      "average model ends:  0.1246 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.8  ]\n",
      "global epoch:  962\n",
      "virtual memory used:  80.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1758 - accuracy: 0.9433\n",
      "training time:  11.4254 s.\n",
      "average model starts\n",
      "average model ends:  0.052 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.803]\n",
      "global epoch:  963\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1645 - accuracy: 0.9442\n",
      "training time:  13.5844 s.\n",
      "average model starts\n",
      "average model ends:  0.0466 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.803 0.796]\n",
      "global epoch:  964\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1811 - accuracy: 0.9423\n",
      "training time:  11.8678 s.\n",
      "average model starts\n",
      "average model ends:  0.0753 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.797]\n",
      "global epoch:  965\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1709 - accuracy: 0.9437\n",
      "training time:  12.241 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.803]\n",
      "global epoch:  966\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1879 - accuracy: 0.9406\n",
      "training time:  11.1597 s.\n",
      "average model starts\n",
      "average model ends:  0.0674 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.803 0.799]\n",
      "global epoch:  967\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1857 - accuracy: 0.9411\n",
      "training time:  13.2776 s.\n",
      "average model starts\n",
      "average model ends:  0.034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.799 0.8  ]\n",
      "global epoch:  968\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1700 - accuracy: 0.9442\n",
      "training time:  12.5123 s.\n",
      "average model starts\n",
      "average model ends:  0.0315 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.8  ]\n",
      "global epoch:  969\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1693 - accuracy: 0.9441\n",
      "training time:  12.6406 s.\n",
      "average model starts\n",
      "average model ends:  0.0536 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.799]\n",
      "global epoch:  970\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1914 - accuracy: 0.9370\n",
      "training time:  11.468 s.\n",
      "average model starts\n",
      "average model ends:  0.0461 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.799 0.793]\n",
      "global epoch:  971\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1723 - accuracy: 0.9420\n",
      "training time:  11.3345 s.\n",
      "average model starts\n",
      "average model ends:  0.0976 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.798]\n",
      "global epoch:  972\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1888 - accuracy: 0.9430\n",
      "training time:  11.3994 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.801]\n",
      "global epoch:  973\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1695 - accuracy: 0.9436\n",
      "training time:  10.6281 s.\n",
      "average model starts\n",
      "average model ends:  0.0469 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.799]\n",
      "global epoch:  974\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1753 - accuracy: 0.9433\n",
      "training time:  12.5636 s.\n",
      "average model starts\n",
      "average model ends:  0.0757 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.802]\n",
      "global epoch:  975\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 40ms/step - loss: 0.1626 - accuracy: 0.9463\n",
      "training time:  10.213 s.\n",
      "average model starts\n",
      "average model ends:  0.0347 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.803]\n",
      "global epoch:  976\n",
      "virtual memory used:  81.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1884 - accuracy: 0.9418\n",
      "training time:  12.0416 s.\n",
      "average model starts\n",
      "average model ends:  0.0561 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.803 0.795]\n",
      "global epoch:  977\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1751 - accuracy: 0.9446\n",
      "training time:  11.2758 s.\n",
      "average model starts\n",
      "average model ends:  0.0382 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.791]\n",
      "global epoch:  978\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1694 - accuracy: 0.9464\n",
      "training time:  10.5867 s.\n",
      "average model starts\n",
      "average model ends:  0.0764 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.791 0.801]\n",
      "global epoch:  979\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1740 - accuracy: 0.9411\n",
      "training time:  11.4605 s.\n",
      "average model starts\n",
      "average model ends:  0.0646 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.801 0.797]\n",
      "global epoch:  980\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1683 - accuracy: 0.9457\n",
      "training time:  13.3393 s.\n",
      "average model starts\n",
      "average model ends:  0.051 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.798]\n",
      "global epoch:  981\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1611 - accuracy: 0.9514\n",
      "training time:  13.1376 s.\n",
      "average model starts\n",
      "average model ends:  0.053 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.789]\n",
      "global epoch:  982\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1656 - accuracy: 0.9444\n",
      "training time:  13.2852 s.\n",
      "average model starts\n",
      "average model ends:  0.0693 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.789 0.793]\n",
      "global epoch:  983\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1726 - accuracy: 0.9436\n",
      "training time:  11.6625 s.\n",
      "average model starts\n",
      "average model ends:  0.0978 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.793 0.801]\n",
      "global epoch:  984\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1683 - accuracy: 0.9451\n",
      "training time:  12.6414 s.\n",
      "average model starts\n",
      "average model ends:  0.0318 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.801 0.8  ]\n",
      "global epoch:  985\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1674 - accuracy: 0.9437\n",
      "training time:  12.1014 s.\n",
      "average model starts\n",
      "average model ends:  0.0422 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.795]\n",
      "global epoch:  986\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1674 - accuracy: 0.9423\n",
      "training time:  12.0965 s.\n",
      "average model starts\n",
      "average model ends:  0.0474 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.808]\n",
      "global epoch:  987\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1749 - accuracy: 0.9406\n",
      "training time:  12.0463 s.\n",
      "average model starts\n",
      "average model ends:  0.06 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.808 0.794]\n",
      "global epoch:  988\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1769 - accuracy: 0.9399\n",
      "training time:  12.8144 s.\n",
      "average model starts\n",
      "average model ends:  0.0919 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.808 0.794 0.793]\n",
      "global epoch:  989\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1818 - accuracy: 0.9433\n",
      "training time:  13.3369 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.802]\n",
      "global epoch:  990\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1607 - accuracy: 0.9438\n",
      "training time:  12.9871 s.\n",
      "average model starts\n",
      "average model ends:  0.0454 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.802 0.79 ]\n",
      "global epoch:  991\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1806 - accuracy: 0.9425\n",
      "training time:  13.4013 s.\n",
      "average model starts\n",
      "average model ends:  0.1638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.79  0.795]\n",
      "global epoch:  992\n",
      "virtual memory used:  75.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1822 - accuracy: 0.9421\n",
      "training time:  12.0312 s.\n",
      "average model starts\n",
      "average model ends:  0.0467 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.795 0.794]\n",
      "global epoch:  993\n",
      "virtual memory used:  75.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1763 - accuracy: 0.9444\n",
      "training time:  12.71 s.\n",
      "average model starts\n",
      "average model ends:  0.0528 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.803]\n",
      "global epoch:  994\n",
      "virtual memory used:  76.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1640 - accuracy: 0.9466\n",
      "training time:  12.0534 s.\n",
      "average model starts\n",
      "average model ends:  0.0643 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.803 0.798]\n",
      "global epoch:  995\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1746 - accuracy: 0.9436\n",
      "training time:  12.4894 s.\n",
      "average model starts\n",
      "average model ends:  0.1559 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.803]\n",
      "global epoch:  996\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1884 - accuracy: 0.9404\n",
      "training time:  12.3486 s.\n",
      "average model starts\n",
      "average model ends:  0.057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.803 0.802]\n",
      "global epoch:  997\n",
      "virtual memory used:  76.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1673 - accuracy: 0.9449\n",
      "training time:  11.5054 s.\n",
      "average model starts\n",
      "average model ends:  0.051 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.791]\n",
      "global epoch:  998\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1728 - accuracy: 0.9446\n",
      "training time:  13.0904 s.\n",
      "average model starts\n",
      "average model ends:  0.0622 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.791 0.798]\n",
      "global epoch:  999\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1770 - accuracy: 0.9406\n",
      "training time:  12.488 s.\n",
      "average model starts\n",
      "average model ends:  0.0954 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.797]\n",
      "global epoch:  1000\n",
      "virtual memory used:  76.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1711 - accuracy: 0.9439\n",
      "training time:  12.8194 s.\n",
      "average model starts\n",
      "average model ends:  0.0513 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.799]\n",
      "global epoch:  1001\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1743 - accuracy: 0.9432\n",
      "training time:  11.5987 s.\n",
      "average model starts\n",
      "average model ends:  0.0437 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.8  ]\n",
      "global epoch:  1002\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1727 - accuracy: 0.9425\n",
      "training time:  11.5645 s.\n",
      "average model starts\n",
      "average model ends:  0.0745 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.795]\n",
      "global epoch:  1003\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1751 - accuracy: 0.9430\n",
      "training time:  11.7611 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.798]\n",
      "global epoch:  1004\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1640 - accuracy: 0.9453\n",
      "training time:  11.1695 s.\n",
      "average model starts\n",
      "average model ends:  0.0526 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.799]\n",
      "global epoch:  1005\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1673 - accuracy: 0.9472\n",
      "training time:  10.6386 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.8  ]\n",
      "global epoch:  1006\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1792 - accuracy: 0.9408\n",
      "training time:  11.9035 s.\n",
      "average model starts\n",
      "average model ends:  0.091 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.795]\n",
      "global epoch:  1007\n",
      "virtual memory used:  78.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1666 - accuracy: 0.9454\n",
      "training time:  11.313 s.\n",
      "average model starts\n",
      "average model ends:  0.0333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.794]\n",
      "global epoch:  1008\n",
      "virtual memory used:  78.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1714 - accuracy: 0.9443\n",
      "training time:  13.7044 s.\n",
      "average model starts\n",
      "average model ends:  0.0634 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.794 0.792]\n",
      "global epoch:  1009\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1672 - accuracy: 0.9458\n",
      "training time:  12.6484 s.\n",
      "average model starts\n",
      "average model ends:  0.0579 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.792 0.793]\n",
      "global epoch:  1010\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1656 - accuracy: 0.9495\n",
      "training time:  12.7011 s.\n",
      "average model starts\n",
      "average model ends:  0.0398 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.793 0.795]\n",
      "global epoch:  1011\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1699 - accuracy: 0.9467\n",
      "training time:  13.3471 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.789]\n",
      "global epoch:  1012\n",
      "virtual memory used:  78.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1662 - accuracy: 0.9483\n",
      "training time:  12.2109 s.\n",
      "average model starts\n",
      "average model ends:  0.1148 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.789 0.796]\n",
      "global epoch:  1013\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1670 - accuracy: 0.9416\n",
      "training time:  12.3934 s.\n",
      "average model starts\n",
      "average model ends:  0.0585 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.794]\n",
      "global epoch:  1014\n",
      "virtual memory used:  78.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1643 - accuracy: 0.9472\n",
      "training time:  12.1963 s.\n",
      "average model starts\n",
      "average model ends:  0.068 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.791]\n",
      "global epoch:  1015\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1645 - accuracy: 0.9444\n",
      "training time:  12.4902 s.\n",
      "average model starts\n",
      "average model ends:  0.0893 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.803]\n",
      "global epoch:  1016\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1754 - accuracy: 0.9439\n",
      "training time:  12.882 s.\n",
      "average model starts\n",
      "average model ends:  0.049 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.803 0.798]\n",
      "global epoch:  1017\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1782 - accuracy: 0.9442\n",
      "training time:  10.7247 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.803]\n",
      "global epoch:  1018\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1708 - accuracy: 0.9467\n",
      "training time:  13.1772 s.\n",
      "average model starts\n",
      "average model ends:  0.0581 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.803 0.797]\n",
      "global epoch:  1019\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1729 - accuracy: 0.9442\n",
      "training time:  11.4725 s.\n",
      "average model starts\n",
      "average model ends:  0.1011 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.797 0.801]\n",
      "global epoch:  1020\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1658 - accuracy: 0.9449\n",
      "training time:  12.4072 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.799]\n",
      "global epoch:  1021\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1701 - accuracy: 0.9427\n",
      "training time:  12.1667 s.\n",
      "average model starts\n",
      "average model ends:  0.0404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.802]\n",
      "global epoch:  1022\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1779 - accuracy: 0.9404\n",
      "training time:  12.7676 s.\n",
      "average model starts\n",
      "average model ends:  0.0465 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.801]\n",
      "global epoch:  1023\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1837 - accuracy: 0.9427\n",
      "training time:  12.2199 s.\n",
      "average model starts\n",
      "average model ends:  0.1 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.796]\n",
      "global epoch:  1024\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1714 - accuracy: 0.9420\n",
      "training time:  11.8745 s.\n",
      "average model starts\n",
      "average model ends:  0.043 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.801]\n",
      "global epoch:  1025\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1622 - accuracy: 0.9513\n",
      "training time:  12.7222 s.\n",
      "average model starts\n",
      "average model ends:  0.0453 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.8  ]\n",
      "global epoch:  1026\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1743 - accuracy: 0.9412\n",
      "training time:  13.6831 s.\n",
      "average model starts\n",
      "average model ends:  0.061 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.782]\n",
      "global epoch:  1027\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1801 - accuracy: 0.9462\n",
      "training time:  12.9448 s.\n",
      "average model starts\n",
      "average model ends:  0.0312 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.782 0.799]\n",
      "global epoch:  1028\n",
      "virtual memory used:  80.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1746 - accuracy: 0.9417\n",
      "training time:  12.3492 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.782 0.799 0.796]\n",
      "global epoch:  1029\n",
      "virtual memory used:  80.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1772 - accuracy: 0.9418\n",
      "training time:  12.563 s.\n",
      "average model starts\n",
      "average model ends:  0.0942 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.796]\n",
      "global epoch:  1030\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1776 - accuracy: 0.9413\n",
      "training time:  12.248 s.\n",
      "average model starts\n",
      "average model ends:  0.0807 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.801]\n",
      "global epoch:  1031\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1682 - accuracy: 0.9431\n",
      "training time:  12.5409 s.\n",
      "average model starts\n",
      "average model ends:  0.0626 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.792]\n",
      "global epoch:  1032\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1694 - accuracy: 0.9456\n",
      "training time:  14.6083 s.\n",
      "average model starts\n",
      "average model ends:  0.0365 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.792 0.793]\n",
      "global epoch:  1033\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1647 - accuracy: 0.9458\n",
      "training time:  10.6578 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.793 0.794]\n",
      "global epoch:  1034\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1713 - accuracy: 0.9420\n",
      "training time:  12.1588 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.794 0.793]\n",
      "global epoch:  1035\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1658 - accuracy: 0.9469\n",
      "training time:  13.2609 s.\n",
      "average model starts\n",
      "average model ends:  0.0745 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.804]\n",
      "global epoch:  1036\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1804 - accuracy: 0.9421\n",
      "training time:  13.0894 s.\n",
      "average model starts\n",
      "average model ends:  0.0515 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.804 0.797]\n",
      "global epoch:  1037\n",
      "virtual memory used:  82.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1867 - accuracy: 0.9399\n",
      "training time:  11.9339 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.797 0.798]\n",
      "global epoch:  1038\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1605 - accuracy: 0.9462\n",
      "training time:  14.089 s.\n",
      "average model starts\n",
      "average model ends:  0.0893 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.789]\n",
      "global epoch:  1039\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1834 - accuracy: 0.9386\n",
      "training time:  12.9037 s.\n",
      "average model starts\n",
      "average model ends:  0.1376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.789 0.798]\n",
      "global epoch:  1040\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1706 - accuracy: 0.9447\n",
      "training time:  12.0137 s.\n",
      "average model starts\n",
      "average model ends:  0.0548 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.798 0.791]\n",
      "global epoch:  1041\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1819 - accuracy: 0.9416\n",
      "training time:  13.128 s.\n",
      "average model starts\n",
      "average model ends:  0.0428 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.794]\n",
      "global epoch:  1042\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1833 - accuracy: 0.9411\n",
      "training time:  12.9419 s.\n",
      "average model starts\n",
      "average model ends:  0.0555 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.794 0.797]\n",
      "global epoch:  1043\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1786 - accuracy: 0.9422\n",
      "training time:  13.4529 s.\n",
      "average model starts\n",
      "average model ends:  0.0575 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.799]\n",
      "global epoch:  1044\n",
      "virtual memory used:  76.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1656 - accuracy: 0.9456\n",
      "training time:  13.8112 s.\n",
      "average model starts\n",
      "average model ends:  0.1029 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.803]\n",
      "global epoch:  1045\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1784 - accuracy: 0.9447\n",
      "training time:  12.32 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.803 0.806]\n",
      "global epoch:  1046\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1835 - accuracy: 0.9402\n",
      "training time:  13.1698 s.\n",
      "average model starts\n",
      "average model ends:  0.055 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.806 0.8  ]\n",
      "global epoch:  1047\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1752 - accuracy: 0.9415\n",
      "training time:  11.7733 s.\n",
      "average model starts\n",
      "average model ends:  0.0905 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.8   0.787]\n",
      "global epoch:  1048\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1656 - accuracy: 0.9474\n",
      "training time:  12.3523 s.\n",
      "average model starts\n",
      "average model ends:  0.0498 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.787 0.797]\n",
      "global epoch:  1049\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1769 - accuracy: 0.9412\n",
      "training time:  12.0531 s.\n",
      "average model starts\n",
      "average model ends:  0.0738 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.797 0.8  ]\n",
      "global epoch:  1050\n",
      "virtual memory used:  78.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1618 - accuracy: 0.9462\n",
      "training time:  11.6628 s.\n",
      "average model starts\n",
      "average model ends:  0.0671 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.794]\n",
      "global epoch:  1051\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1750 - accuracy: 0.9433\n",
      "training time:  12.991 s.\n",
      "average model starts\n",
      "average model ends:  0.1176 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.791]\n",
      "global epoch:  1052\n",
      "virtual memory used:  77.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1710 - accuracy: 0.94880s - loss: 0.1703 - accuracy: 0.94\n",
      "training time:  11.7536 s.\n",
      "average model starts\n",
      "average model ends:  0.1249 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.791 0.796]\n",
      "global epoch:  1053\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1810 - accuracy: 0.9427\n",
      "training time:  13.6798 s.\n",
      "average model starts\n",
      "average model ends:  0.066 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.796 0.8  ]\n",
      "global epoch:  1054\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1676 - accuracy: 0.9474\n",
      "training time:  13.6925 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.793]\n",
      "global epoch:  1055\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1803 - accuracy: 0.9415\n",
      "training time:  13.2155 s.\n",
      "average model starts\n",
      "average model ends:  0.0625 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.793 0.798]\n",
      "global epoch:  1056\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1719 - accuracy: 0.9415\n",
      "training time:  12.0951 s.\n",
      "average model starts\n",
      "average model ends:  0.0689 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.79 ]\n",
      "global epoch:  1057\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1719 - accuracy: 0.9438\n",
      "training time:  12.0545 s.\n",
      "average model starts\n",
      "average model ends:  0.084 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.79  0.792]\n",
      "global epoch:  1058\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1864 - accuracy: 0.9407\n",
      "training time:  13.3272 s.\n",
      "average model starts\n",
      "average model ends:  0.1311 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.792 0.799]\n",
      "global epoch:  1059\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1698 - accuracy: 0.9442\n",
      "training time:  12.3493 s.\n",
      "average model starts\n",
      "average model ends:  0.1146 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.796]\n",
      "global epoch:  1060\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1732 - accuracy: 0.9411\n",
      "training time:  12.2344 s.\n",
      "average model starts\n",
      "average model ends:  0.0422 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.805]\n",
      "global epoch:  1061\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1643 - accuracy: 0.9437\n",
      "training time:  13.7126 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.805 0.8  ]\n",
      "global epoch:  1062\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1763 - accuracy: 0.9400\n",
      "training time:  12.2441 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.8   0.798]\n",
      "global epoch:  1063\n",
      "virtual memory used:  80.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1824 - accuracy: 0.9418\n",
      "training time:  12.8136 s.\n",
      "average model starts\n",
      "average model ends:  0.1424 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.785]\n",
      "global epoch:  1064\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1718 - accuracy: 0.9443\n",
      "training time:  11.1842 s.\n",
      "average model starts\n",
      "average model ends:  0.0763 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.785 0.796]\n",
      "global epoch:  1065\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1782 - accuracy: 0.9442\n",
      "training time:  12.5512 s.\n",
      "average model starts\n",
      "average model ends:  0.1068 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.796 0.797]\n",
      "global epoch:  1066\n",
      "virtual memory used:  80.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1633 - accuracy: 0.9463\n",
      "training time:  12.4541 s.\n",
      "average model starts\n",
      "average model ends:  0.0505 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.795]\n",
      "global epoch:  1067\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1752 - accuracy: 0.9427\n",
      "training time:  13.9947 s.\n",
      "average model starts\n",
      "average model ends:  0.0652 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.796]\n",
      "global epoch:  1068\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1809 - accuracy: 0.9375\n",
      "training time:  12.8063 s.\n",
      "average model starts\n",
      "average model ends:  0.0878 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.8  ]\n",
      "global epoch:  1069\n",
      "virtual memory used:  79.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1694 - accuracy: 0.9408\n",
      "training time:  11.2034 s.\n",
      "average model starts\n",
      "average model ends:  0.0333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.802]\n",
      "global epoch:  1070\n",
      "virtual memory used:  80.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1678 - accuracy: 0.9435\n",
      "training time:  12.3411 s.\n",
      "average model starts\n",
      "average model ends:  0.0796 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.804]\n",
      "global epoch:  1071\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1762 - accuracy: 0.9423\n",
      "training time:  11.6484 s.\n",
      "average model starts\n",
      "average model ends:  0.0727 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.804 0.801]\n",
      "global epoch:  1072\n",
      "virtual memory used:  79.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - ETA: 0s - loss: 0.1775 - accuracy: 0.94 - 13s 51ms/step - loss: 0.1773 - accuracy: 0.9412\n",
      "training time:  12.9826 s.\n",
      "average model starts\n",
      "average model ends:  0.0947 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.794]\n",
      "global epoch:  1073\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1729 - accuracy: 0.9437\n",
      "training time:  12.8517 s.\n",
      "average model starts\n",
      "average model ends:  0.0531 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.797]\n",
      "global epoch:  1074\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1878 - accuracy: 0.9392\n",
      "training time:  12.7253 s.\n",
      "average model starts\n",
      "average model ends:  0.0389 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.801]\n",
      "global epoch:  1075\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1699 - accuracy: 0.9444\n",
      "training time:  13.4537 s.\n",
      "average model starts\n",
      "average model ends:  0.0649 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.794]\n",
      "global epoch:  1076\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1744 - accuracy: 0.9416\n",
      "training time:  12.8413 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.798]\n",
      "global epoch:  1077\n",
      "virtual memory used:  80.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1678 - accuracy: 0.9475\n",
      "training time:  11.832 s.\n",
      "average model starts\n",
      "average model ends:  0.0399 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.805]\n",
      "global epoch:  1078\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1790 - accuracy: 0.9400\n",
      "training time:  11.0454 s.\n",
      "average model starts\n",
      "average model ends:  0.1007 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.805 0.791]\n",
      "global epoch:  1079\n",
      "virtual memory used:  81.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1771 - accuracy: 0.9416\n",
      "training time:  12.6713 s.\n",
      "average model starts\n",
      "average model ends:  0.0602 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.791 0.795]\n",
      "global epoch:  1080\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1701 - accuracy: 0.9431\n",
      "training time:  10.9475 s.\n",
      "average model starts\n",
      "average model ends:  0.0552 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.795 0.793]\n",
      "global epoch:  1081\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1870 - accuracy: 0.9397\n",
      "training time:  11.9897 s.\n",
      "average model starts\n",
      "average model ends:  0.034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.793 0.8  ]\n",
      "global epoch:  1082\n",
      "virtual memory used:  81.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1656 - accuracy: 0.9473\n",
      "training time:  11.8828 s.\n",
      "average model starts\n",
      "average model ends:  0.0549 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.798]\n",
      "global epoch:  1083\n",
      "virtual memory used:  81.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1653 - accuracy: 0.9438\n",
      "training time:  12.822 s.\n",
      "average model starts\n",
      "average model ends:  0.0483 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.796]\n",
      "global epoch:  1084\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1802 - accuracy: 0.9408\n",
      "training time:  11.6145 s.\n",
      "average model starts\n",
      "average model ends:  0.0522 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.795]\n",
      "global epoch:  1085\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - ETA: 0s - loss: 0.1788 - accuracy: 0.94 - 11s 45ms/step - loss: 0.1800 - accuracy: 0.9408\n",
      "training time:  11.5515 s.\n",
      "average model starts\n",
      "average model ends:  0.0932 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.792]\n",
      "global epoch:  1086\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1848 - accuracy: 0.9412\n",
      "training time:  12.3034 s.\n",
      "average model starts\n",
      "average model ends:  0.0303 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.792 0.792]\n",
      "global epoch:  1087\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1689 - accuracy: 0.9457\n",
      "training time:  12.8042 s.\n",
      "average model starts\n",
      "average model ends:  0.0826 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.801]\n",
      "global epoch:  1088\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1748 - accuracy: 0.9442\n",
      "training time:  11.9555 s.\n",
      "average model starts\n",
      "average model ends:  0.0334 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.801 0.794]\n",
      "global epoch:  1089\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1804 - accuracy: 0.9402\n",
      "training time:  10.5352 s.\n",
      "average model starts\n",
      "average model ends:  0.0466 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.798]\n",
      "global epoch:  1090\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1755 - accuracy: 0.9423\n",
      "training time:  12.6557 s.\n",
      "average model starts\n",
      "average model ends:  0.1293 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.796]\n",
      "global epoch:  1091\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1734 - accuracy: 0.9421\n",
      "training time:  10.6349 s.\n",
      "average model starts\n",
      "average model ends:  0.0824 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.796]\n",
      "global epoch:  1092\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1719 - accuracy: 0.9430\n",
      "training time:  12.202 s.\n",
      "average model starts\n",
      "average model ends:  0.049 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.787]\n",
      "global epoch:  1093\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1791 - accuracy: 0.9406\n",
      "training time:  12.1531 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.787 0.803]\n",
      "global epoch:  1094\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1816 - accuracy: 0.9411\n",
      "training time:  12.1456 s.\n",
      "average model starts\n",
      "average model ends:  0.1106 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.803 0.798]\n",
      "global epoch:  1095\n",
      "virtual memory used:  76.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1610 - accuracy: 0.9489\n",
      "training time:  11.5065 s.\n",
      "average model starts\n",
      "average model ends:  0.0437 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.797]\n",
      "global epoch:  1096\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1565 - accuracy: 0.9487\n",
      "training time:  12.8991 s.\n",
      "average model starts\n",
      "average model ends:  0.0455 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.798]\n",
      "global epoch:  1097\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1697 - accuracy: 0.9442\n",
      "training time:  12.202 s.\n",
      "average model starts\n",
      "average model ends:  0.0574 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.804]\n",
      "global epoch:  1098\n",
      "virtual memory used:  77.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1790 - accuracy: 0.9415\n",
      "training time:  12.5266 s.\n",
      "average model starts\n",
      "average model ends:  0.0379 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.804 0.794]\n",
      "global epoch:  1099\n",
      "virtual memory used:  77.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1701 - accuracy: 0.9453\n",
      "training time:  12.246 s.\n",
      "average model starts\n",
      "average model ends:  0.0933 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.794 0.796]\n",
      "global epoch:  1100\n",
      "virtual memory used:  78.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1753 - accuracy: 0.9462\n",
      "training time:  12.2363 s.\n",
      "average model starts\n",
      "average model ends:  0.0377 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.797]\n",
      "global epoch:  1101\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1638 - accuracy: 0.9447\n",
      "training time:  12.0922 s.\n",
      "average model starts\n",
      "average model ends:  0.0924 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.804]\n",
      "global epoch:  1102\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1663 - accuracy: 0.9467\n",
      "training time:  12.4374 s.\n",
      "average model starts\n",
      "average model ends:  0.0484 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.804 0.803]\n",
      "global epoch:  1103\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1780 - accuracy: 0.9397\n",
      "training time:  12.2878 s.\n",
      "average model starts\n",
      "average model ends:  0.0722 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.803 0.801]\n",
      "global epoch:  1104\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1688 - accuracy: 0.9444\n",
      "training time:  12.301 s.\n",
      "average model starts\n",
      "average model ends:  0.048 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.792]\n",
      "global epoch:  1105\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1748 - accuracy: 0.9412\n",
      "training time:  11.3862 s.\n",
      "average model starts\n",
      "average model ends:  0.042 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.792 0.791]\n",
      "global epoch:  1106\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1752 - accuracy: 0.9412\n",
      "training time:  12.7781 s.\n",
      "average model starts\n",
      "average model ends:  0.1292 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.791 0.797]\n",
      "global epoch:  1107\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1859 - accuracy: 0.9402\n",
      "training time:  12.6668 s.\n",
      "average model starts\n",
      "average model ends:  0.0697 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.797 0.794]\n",
      "global epoch:  1108\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1680 - accuracy: 0.9439\n",
      "training time:  13.225 s.\n",
      "average model starts\n",
      "average model ends:  0.031 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.799]\n",
      "global epoch:  1109\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1754 - accuracy: 0.9442\n",
      "training time:  13.1903 s.\n",
      "average model starts\n",
      "average model ends:  0.0693 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.795]\n",
      "global epoch:  1110\n",
      "virtual memory used:  79.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1767 - accuracy: 0.9420\n",
      "training time:  11.9829 s.\n",
      "average model starts\n",
      "average model ends:  0.0497 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.799]\n",
      "global epoch:  1111\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1722 - accuracy: 0.9427\n",
      "training time:  11.4653 s.\n",
      "average model starts\n",
      "average model ends:  0.0658 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.79 ]\n",
      "global epoch:  1112\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1601 - accuracy: 0.9456\n",
      "training time:  12.5391 s.\n",
      "average model starts\n",
      "average model ends:  0.1002 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.79  0.801]\n",
      "global epoch:  1113\n",
      "virtual memory used:  79.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1890 - accuracy: 0.9364\n",
      "training time:  12.4545 s.\n",
      "average model starts\n",
      "average model ends:  0.1454 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.801 0.8  ]\n",
      "global epoch:  1114\n",
      "virtual memory used:  80.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1816 - accuracy: 0.9406\n",
      "training time:  11.6818 s.\n",
      "average model starts\n",
      "average model ends:  0.0339 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.8  ]\n",
      "global epoch:  1115\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1716 - accuracy: 0.9396\n",
      "training time:  11.3587 s.\n",
      "average model starts\n",
      "average model ends:  0.0364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.8  ]\n",
      "global epoch:  1116\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1642 - accuracy: 0.9463\n",
      "training time:  11.4833 s.\n",
      "average model starts\n",
      "average model ends:  0.0535 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.795]\n",
      "global epoch:  1117\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1905 - accuracy: 0.9370\n",
      "training time:  12.4433 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.802]\n",
      "global epoch:  1118\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1773 - accuracy: 0.9411\n",
      "training time:  10.6806 s.\n",
      "average model starts\n",
      "average model ends:  0.0383 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.79 ]\n",
      "global epoch:  1119\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1634 - accuracy: 0.9452\n",
      "training time:  12.3023 s.\n",
      "average model starts\n",
      "average model ends:  0.0442 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.79  0.793]\n",
      "global epoch:  1120\n",
      "virtual memory used:  81.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1712 - accuracy: 0.9439\n",
      "training time:  13.5804 s.\n",
      "average model starts\n",
      "average model ends:  0.048 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.793 0.797]\n",
      "global epoch:  1121\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1676 - accuracy: 0.9437\n",
      "training time:  12.442 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.802]\n",
      "global epoch:  1122\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1745 - accuracy: 0.9407\n",
      "training time:  11.0348 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.799]\n",
      "global epoch:  1123\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1741 - accuracy: 0.9410\n",
      "training time:  11.2107 s.\n",
      "average model starts\n",
      "average model ends:  0.0404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.808]\n",
      "global epoch:  1124\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1746 - accuracy: 0.9421\n",
      "training time:  11.5095 s.\n",
      "average model starts\n",
      "average model ends:  0.037 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.808 0.792]\n",
      "global epoch:  1125\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1699 - accuracy: 0.9433\n",
      "training time:  11.3918 s.\n",
      "average model starts\n",
      "average model ends:  0.0378 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.808 0.792 0.795]\n",
      "global epoch:  1126\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1762 - accuracy: 0.9430\n",
      "training time:  11.9496 s.\n",
      "average model starts\n",
      "average model ends:  0.0649 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.798]\n",
      "global epoch:  1127\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1685 - accuracy: 0.9420\n",
      "training time:  11.6003 s.\n",
      "average model starts\n",
      "average model ends:  0.0893 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.798]\n",
      "global epoch:  1128\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1781 - accuracy: 0.9404\n",
      "training time:  10.658 s.\n",
      "average model starts\n",
      "average model ends:  0.0879 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.8  ]\n",
      "global epoch:  1129\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1749 - accuracy: 0.9444\n",
      "training time:  10.7926 s.\n",
      "average model starts\n",
      "average model ends:  0.0488 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.8  ]\n",
      "global epoch:  1130\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1649 - accuracy: 0.9451\n",
      "training time:  10.8707 s.\n",
      "average model starts\n",
      "average model ends:  0.0358 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.796]\n",
      "global epoch:  1131\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1684 - accuracy: 0.9457\n",
      "training time:  11.6636 s.\n",
      "average model starts\n",
      "average model ends:  0.0768 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.806]\n",
      "global epoch:  1132\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1704 - accuracy: 0.9466\n",
      "training time:  10.9851 s.\n",
      "average model starts\n",
      "average model ends:  0.0518 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.806 0.794]\n",
      "global epoch:  1133\n",
      "virtual memory used:  82.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1639 - accuracy: 0.9438\n",
      "training time:  12.7799 s.\n",
      "average model starts\n",
      "average model ends:  0.0853 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.794 0.8  ]\n",
      "global epoch:  1134\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1703 - accuracy: 0.9411\n",
      "training time:  12.525 s.\n",
      "average model starts\n",
      "average model ends:  0.0441 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.8  ]\n",
      "global epoch:  1135\n",
      "virtual memory used:  81.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1545 - accuracy: 0.9490\n",
      "training time:  10.8112 s.\n",
      "average model starts\n",
      "average model ends:  0.0919 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.797]\n",
      "global epoch:  1136\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1620 - accuracy: 0.9458\n",
      "training time:  12.9322 s.\n",
      "average model starts\n",
      "average model ends:  0.0824 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.797]\n",
      "global epoch:  1137\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1765 - accuracy: 0.9432\n",
      "training time:  11.9307 s.\n",
      "average model starts\n",
      "average model ends:  0.1168 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.797]\n",
      "global epoch:  1138\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1789 - accuracy: 0.9433\n",
      "training time:  11.7732 s.\n",
      "average model starts\n",
      "average model ends:  0.0555 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.798]\n",
      "global epoch:  1139\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1625 - accuracy: 0.9458\n",
      "training time:  11.5518 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.799]\n",
      "global epoch:  1140\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1665 - accuracy: 0.9437\n",
      "training time:  11.7095 s.\n",
      "average model starts\n",
      "average model ends:  0.0546 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.802]\n",
      "global epoch:  1141\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1654 - accuracy: 0.9433\n",
      "training time:  11.9017 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.801]\n",
      "global epoch:  1142\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1666 - accuracy: 0.9469\n",
      "training time:  12.2428 s.\n",
      "average model starts\n",
      "average model ends:  0.0493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.798]\n",
      "global epoch:  1143\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1771 - accuracy: 0.9420\n",
      "training time:  11.79 s.\n",
      "average model starts\n",
      "average model ends:  0.0967 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.805]\n",
      "global epoch:  1144\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1742 - accuracy: 0.9437\n",
      "training time:  13.3205 s.\n",
      "average model starts\n",
      "average model ends:  0.0596 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.805 0.804]\n",
      "global epoch:  1145\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1617 - accuracy: 0.9472\n",
      "training time:  11.4792 s.\n",
      "average model starts\n",
      "average model ends:  0.0555 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.804 0.806]\n",
      "global epoch:  1146\n",
      "virtual memory used:  84.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1746 - accuracy: 0.9416\n",
      "training time:  12.0554 s.\n",
      "average model starts\n",
      "average model ends:  0.032 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.806 0.796]\n",
      "global epoch:  1147\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1742 - accuracy: 0.9447\n",
      "training time:  12.586 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.796 0.789]\n",
      "global epoch:  1148\n",
      "virtual memory used:  76.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1720 - accuracy: 0.9458\n",
      "training time:  12.2758 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.789 0.796]\n",
      "global epoch:  1149\n",
      "virtual memory used:  77.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1553 - accuracy: 0.9483\n",
      "training time:  12.7185 s.\n",
      "average model starts\n",
      "average model ends:  0.0932 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.797]\n",
      "global epoch:  1150\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1658 - accuracy: 0.9427\n",
      "training time:  13.9923 s.\n",
      "average model starts\n",
      "average model ends:  0.0764 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.789]\n",
      "global epoch:  1151\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1691 - accuracy: 0.9441\n",
      "training time:  13.6066 s.\n",
      "average model starts\n",
      "average model ends:  0.1041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.789 0.799]\n",
      "global epoch:  1152\n",
      "virtual memory used:  77.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1563 - accuracy: 0.9498\n",
      "training time:  11.517 s.\n",
      "average model starts\n",
      "average model ends:  0.0846 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.799 0.802]\n",
      "global epoch:  1153\n",
      "virtual memory used:  78.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1659 - accuracy: 0.9461\n",
      "training time:  12.113 s.\n",
      "average model starts\n",
      "average model ends:  0.0812 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.805]\n",
      "global epoch:  1154\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1802 - accuracy: 0.9441\n",
      "training time:  12.0014 s.\n",
      "average model starts\n",
      "average model ends:  0.0529 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.805 0.801]\n",
      "global epoch:  1155\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1579 - accuracy: 0.9488\n",
      "training time:  12.4536 s.\n",
      "average model starts\n",
      "average model ends:  0.051 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.801 0.784]\n",
      "global epoch:  1156\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1700 - accuracy: 0.9473\n",
      "training time:  13.065 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.784 0.802]\n",
      "global epoch:  1157\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1684 - accuracy: 0.9475\n",
      "training time:  13.2184 s.\n",
      "average model starts\n",
      "average model ends:  0.1087 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.802 0.799]\n",
      "global epoch:  1158\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1762 - accuracy: 0.9435\n",
      "training time:  13.0837 s.\n",
      "average model starts\n",
      "average model ends:  0.0357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.79 ]\n",
      "global epoch:  1159\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1613 - accuracy: 0.9449\n",
      "training time:  12.4727 s.\n",
      "average model starts\n",
      "average model ends:  0.0976 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.79  0.792]\n",
      "global epoch:  1160\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1619 - accuracy: 0.9436\n",
      "training time:  12.2577 s.\n",
      "average model starts\n",
      "average model ends:  0.1145 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.792 0.796]\n",
      "global epoch:  1161\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1800 - accuracy: 0.9415\n",
      "training time:  12.0682 s.\n",
      "average model starts\n",
      "average model ends:  0.0608 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.796 0.798]\n",
      "global epoch:  1162\n",
      "virtual memory used:  80.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1745 - accuracy: 0.9456\n",
      "training time:  10.6954 s.\n",
      "average model starts\n",
      "average model ends:  0.0988 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.8  ]\n",
      "global epoch:  1163\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1749 - accuracy: 0.9422\n",
      "training time:  10.901 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.801]\n",
      "global epoch:  1164\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1724 - accuracy: 0.9425\n",
      "training time:  11.5954 s.\n",
      "average model starts\n",
      "average model ends:  0.0446 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.805]\n",
      "global epoch:  1165\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1763 - accuracy: 0.9417\n",
      "training time:  12.7793 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.805 0.79 ]\n",
      "global epoch:  1166\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1743 - accuracy: 0.9438\n",
      "training time:  11.9456 s.\n",
      "average model starts\n",
      "average model ends:  0.105 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.79  0.802]\n",
      "global epoch:  1167\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1548 - accuracy: 0.9493\n",
      "training time:  11.4288 s.\n",
      "average model starts\n",
      "average model ends:  0.0397 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.796]\n",
      "global epoch:  1168\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1646 - accuracy: 0.9459\n",
      "training time:  12.6752 s.\n",
      "average model starts\n",
      "average model ends:  0.0594 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.789]\n",
      "global epoch:  1169\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1663 - accuracy: 0.9464\n",
      "training time:  12.7607 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.789 0.796]\n",
      "global epoch:  1170\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1612 - accuracy: 0.9463\n",
      "training time:  12.4069 s.\n",
      "average model starts\n",
      "average model ends:  0.0359 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.796 0.798]\n",
      "global epoch:  1171\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1753 - accuracy: 0.9442\n",
      "training time:  13.5612 s.\n",
      "average model starts\n",
      "average model ends:  0.0695 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.797]\n",
      "global epoch:  1172\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1711 - accuracy: 0.9452\n",
      "training time:  13.7938 s.\n",
      "average model starts\n",
      "average model ends:  0.0518 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.797]\n",
      "global epoch:  1173\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1787 - accuracy: 0.9428\n",
      "training time:  13.2783 s.\n",
      "average model starts\n",
      "average model ends:  0.1306 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.794]\n",
      "global epoch:  1174\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1814 - accuracy: 0.9400\n",
      "training time:  12.5279 s.\n",
      "average model starts\n",
      "average model ends:  0.0423 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.794]\n",
      "global epoch:  1175\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1714 - accuracy: 0.9438\n",
      "training time:  12.0401 s.\n",
      "average model starts\n",
      "average model ends:  0.0612 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.797]\n",
      "global epoch:  1176\n",
      "virtual memory used:  81.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1645 - accuracy: 0.9448\n",
      "training time:  14.0922 s.\n",
      "average model starts\n",
      "average model ends:  0.0643 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.802]\n",
      "global epoch:  1177\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1865 - accuracy: 0.9386\n",
      "training time:  13.5772 s.\n",
      "average model starts\n",
      "average model ends:  0.0675 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.794]\n",
      "global epoch:  1178\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1641 - accuracy: 0.9485\n",
      "training time:  12.5957 s.\n",
      "average model starts\n",
      "average model ends:  0.1752 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.797]\n",
      "global epoch:  1179\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1772 - accuracy: 0.9431\n",
      "training time:  13.2 s.\n",
      "average model starts\n",
      "average model ends:  0.1404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.805]\n",
      "global epoch:  1180\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1609 - accuracy: 0.9444\n",
      "training time:  12.5079 s.\n",
      "average model starts\n",
      "average model ends:  0.0431 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.805 0.798]\n",
      "global epoch:  1181\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1626 - accuracy: 0.9469\n",
      "training time:  12.5199 s.\n",
      "average model starts\n",
      "average model ends:  0.0864 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.798 0.802]\n",
      "global epoch:  1182\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1713 - accuracy: 0.9404\n",
      "training time:  12.6917 s.\n",
      "average model starts\n",
      "average model ends:  0.0468 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.797]\n",
      "global epoch:  1183\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1960 - accuracy: 0.9368\n",
      "training time:  12.3766 s.\n",
      "average model starts\n",
      "average model ends:  0.0826 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.804]\n",
      "global epoch:  1184\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1751 - accuracy: 0.9442\n",
      "training time:  13.6512 s.\n",
      "average model starts\n",
      "average model ends:  0.0593 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.804 0.804]\n",
      "global epoch:  1185\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1635 - accuracy: 0.9469\n",
      "training time:  13.231 s.\n",
      "average model starts\n",
      "average model ends:  0.0527 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.804 0.796]\n",
      "global epoch:  1186\n",
      "virtual memory used:  83.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1654 - accuracy: 0.9490\n",
      "training time:  11.9985 s.\n",
      "average model starts\n",
      "average model ends:  0.0506 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.796 0.796]\n",
      "global epoch:  1187\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1696 - accuracy: 0.9466\n",
      "training time:  12.9168 s.\n",
      "average model starts\n",
      "average model ends:  0.0985 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.794]\n",
      "global epoch:  1188\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1666 - accuracy: 0.9436\n",
      "training time:  13.3981 s.\n",
      "average model starts\n",
      "average model ends:  0.0996 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.797]\n",
      "global epoch:  1189\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1720 - accuracy: 0.9426\n",
      "training time:  13.0481 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.799]\n",
      "global epoch:  1190\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1746 - accuracy: 0.9417\n",
      "training time:  14.1832 s.\n",
      "average model starts\n",
      "average model ends:  0.0347 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.799]\n",
      "global epoch:  1191\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1717 - accuracy: 0.9462\n",
      "training time:  12.9967 s.\n",
      "average model starts\n",
      "average model ends:  0.1225 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.792]\n",
      "global epoch:  1192\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1767 - accuracy: 0.9448\n",
      "training time:  11.885 s.\n",
      "average model starts\n",
      "average model ends:  0.0702 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.792 0.801]\n",
      "global epoch:  1193\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1537 - accuracy: 0.9511\n",
      "training time:  11.9174 s.\n",
      "average model starts\n",
      "average model ends:  0.1211 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.801 0.796]\n",
      "global epoch:  1194\n",
      "virtual memory used:  84.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1608 - accuracy: 0.9451\n",
      "training time:  13.3997 s.\n",
      "average model starts\n",
      "average model ends:  0.0743 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.797]\n",
      "global epoch:  1195\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1699 - accuracy: 0.9466\n",
      "training time:  13.7215 s.\n",
      "average model starts\n",
      "average model ends:  0.0642 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.797 0.798]\n",
      "global epoch:  1196\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1735 - accuracy: 0.9425\n",
      "training time:  12.0353 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.8  ]\n",
      "global epoch:  1197\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1659 - accuracy: 0.9456\n",
      "training time:  11.9316 s.\n",
      "average model starts\n",
      "average model ends:  0.0458 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.795]\n",
      "global epoch:  1198\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1667 - accuracy: 0.9443\n",
      "training time:  11.5342 s.\n",
      "average model starts\n",
      "average model ends:  0.0453 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.802]\n",
      "global epoch:  1199\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1595 - accuracy: 0.9448\n",
      "training time:  12.4808 s.\n",
      "average model starts\n",
      "average model ends:  0.0933 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.796]\n",
      "global epoch:  1200\n",
      "virtual memory used:  77.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1710 - accuracy: 0.9446\n",
      "training time:  13.4524 s.\n",
      "average model starts\n",
      "average model ends:  0.0897 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.8  ]\n",
      "global epoch:  1201\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1568 - accuracy: 0.9467\n",
      "training time:  12.7966 s.\n",
      "average model starts\n",
      "average model ends:  0.044 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.797]\n",
      "global epoch:  1202\n",
      "virtual memory used:  79.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1585 - accuracy: 0.9472\n",
      "training time:  11.9577 s.\n",
      "average model starts\n",
      "average model ends:  0.0444 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.796]\n",
      "global epoch:  1203\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1745 - accuracy: 0.9411\n",
      "training time:  11.4122 s.\n",
      "average model starts\n",
      "average model ends:  0.1335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.79 ]\n",
      "global epoch:  1204\n",
      "virtual memory used:  78.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1629 - accuracy: 0.9466\n",
      "training time:  12.0788 s.\n",
      "average model starts\n",
      "average model ends:  0.0376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.79  0.799]\n",
      "global epoch:  1205\n",
      "virtual memory used:  78.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1701 - accuracy: 0.9446\n",
      "training time:  12.3424 s.\n",
      "average model starts\n",
      "average model ends:  0.0313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.799 0.791]\n",
      "global epoch:  1206\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1674 - accuracy: 0.9421\n",
      "training time:  11.5106 s.\n",
      "average model starts\n",
      "average model ends:  0.0495 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.791 0.799]\n",
      "global epoch:  1207\n",
      "virtual memory used:  80.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1584 - accuracy: 0.9480\n",
      "training time:  12.1055 s.\n",
      "average model starts\n",
      "average model ends:  0.0645 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.797]\n",
      "global epoch:  1208\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1643 - accuracy: 0.9441\n",
      "training time:  13.376 s.\n",
      "average model starts\n",
      "average model ends:  0.0567 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.799]\n",
      "global epoch:  1209\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1658 - accuracy: 0.9461\n",
      "training time:  12.7803 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.795]\n",
      "global epoch:  1210\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1573 - accuracy: 0.9473\n",
      "training time:  11.8588 s.\n",
      "average model starts\n",
      "average model ends:  0.0426 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.795]\n",
      "global epoch:  1211\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1712 - accuracy: 0.9442\n",
      "training time:  12.3447 s.\n",
      "average model starts\n",
      "average model ends:  0.1493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.798]\n",
      "global epoch:  1212\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1719 - accuracy: 0.9438\n",
      "training time:  12.2429 s.\n",
      "average model starts\n",
      "average model ends:  0.0543 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.799]\n",
      "global epoch:  1213\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1680 - accuracy: 0.9452\n",
      "training time:  12.7994 s.\n",
      "average model starts\n",
      "average model ends:  0.0367 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.801]\n",
      "global epoch:  1214\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1669 - accuracy: 0.9433\n",
      "training time:  12.9165 s.\n",
      "average model starts\n",
      "average model ends:  0.1221 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.789]\n",
      "global epoch:  1215\n",
      "virtual memory used:  81.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1700 - accuracy: 0.9467\n",
      "training time:  13.3769 s.\n",
      "average model starts\n",
      "average model ends:  0.0456 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.789 0.795]\n",
      "global epoch:  1216\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1700 - accuracy: 0.9456\n",
      "training time:  12.5724 s.\n",
      "average model starts\n",
      "average model ends:  0.0523 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.795 0.797]\n",
      "global epoch:  1217\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1677 - accuracy: 0.9489\n",
      "training time:  12.6937 s.\n",
      "average model starts\n",
      "average model ends:  0.0752 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.787]\n",
      "global epoch:  1218\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.1801 - accuracy: 0.9413\n",
      "training time:  14.8647 s.\n",
      "average model starts\n",
      "average model ends:  0.0722 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.787 0.801]\n",
      "global epoch:  1219\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1630 - accuracy: 0.9492\n",
      "training time:  12.7412 s.\n",
      "average model starts\n",
      "average model ends:  0.1169 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.787 0.801 0.794]\n",
      "global epoch:  1220\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1673 - accuracy: 0.9456\n",
      "training time:  12.8207 s.\n",
      "average model starts\n",
      "average model ends:  0.0402 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.793]\n",
      "global epoch:  1221\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1702 - accuracy: 0.9446\n",
      "training time:  12.9063 s.\n",
      "average model starts\n",
      "average model ends:  0.1034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.793 0.803]\n",
      "global epoch:  1222\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1571 - accuracy: 0.9509\n",
      "training time:  12.2698 s.\n",
      "average model starts\n",
      "average model ends:  0.1278 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.803 0.803]\n",
      "global epoch:  1223\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1734 - accuracy: 0.9406\n",
      "training time:  14.0676 s.\n",
      "average model starts\n",
      "average model ends:  0.052 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.803 0.804]\n",
      "global epoch:  1224\n",
      "virtual memory used:  81.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1706 - accuracy: 0.9461\n",
      "training time:  13.4369 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.804 0.803]\n",
      "global epoch:  1225\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1803 - accuracy: 0.9386\n",
      "training time:  13.0178 s.\n",
      "average model starts\n",
      "average model ends:  0.0691 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.803 0.804]\n",
      "global epoch:  1226\n",
      "virtual memory used:  82.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1714 - accuracy: 0.9463\n",
      "training time:  13.2409 s.\n",
      "average model starts\n",
      "average model ends:  0.0581 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.804 0.801]\n",
      "global epoch:  1227\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1619 - accuracy: 0.9482\n",
      "training time:  14.4193 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.798]\n",
      "global epoch:  1228\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1739 - accuracy: 0.9423\n",
      "training time:  13.9299 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.788]\n",
      "global epoch:  1229\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1575 - accuracy: 0.9470\n",
      "training time:  12.1979 s.\n",
      "average model starts\n",
      "average model ends:  0.0379 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.788 0.798]\n",
      "global epoch:  1230\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1691 - accuracy: 0.9421\n",
      "training time:  12.6223 s.\n",
      "average model starts\n",
      "average model ends:  0.038 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.798 0.794]\n",
      "global epoch:  1231\n",
      "virtual memory used:  84.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1775 - accuracy: 0.9417\n",
      "training time:  12.8122 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.794 0.797]\n",
      "global epoch:  1232\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1725 - accuracy: 0.9442\n",
      "training time:  12.9634 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.802]\n",
      "global epoch:  1233\n",
      "virtual memory used:  82.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1775 - accuracy: 0.9430\n",
      "training time:  14.0973 s.\n",
      "average model starts\n",
      "average model ends:  0.0705 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.8  ]\n",
      "global epoch:  1234\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1730 - accuracy: 0.9432\n",
      "training time:  12.7005 s.\n",
      "average model starts\n",
      "average model ends:  0.053 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.802]\n",
      "global epoch:  1235\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1593 - accuracy: 0.9461\n",
      "training time:  12.5386 s.\n",
      "average model starts\n",
      "average model ends:  0.1061 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.807]\n",
      "global epoch:  1236\n",
      "virtual memory used:  82.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1679 - accuracy: 0.9443\n",
      "training time:  14.0364 s.\n",
      "average model starts\n",
      "average model ends:  0.1404 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.807 0.795]\n",
      "global epoch:  1237\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1529 - accuracy: 0.9535\n",
      "training time:  14.1643 s.\n",
      "average model starts\n",
      "average model ends:  0.0596 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.795 0.796]\n",
      "global epoch:  1238\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1564 - accuracy: 0.9473\n",
      "training time:  12.5409 s.\n",
      "average model starts\n",
      "average model ends:  0.073 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.805]\n",
      "global epoch:  1239\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1665 - accuracy: 0.9439\n",
      "training time:  11.4654 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.805 0.796]\n",
      "global epoch:  1240\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.1854 - accuracy: 0.9387\n",
      "training time:  14.7111 s.\n",
      "average model starts\n",
      "average model ends:  0.0695 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.796 0.793]\n",
      "global epoch:  1241\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1636 - accuracy: 0.9456\n",
      "training time:  13.3368 s.\n",
      "average model starts\n",
      "average model ends:  0.0563 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.8  ]\n",
      "global epoch:  1242\n",
      "virtual memory used:  84.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1722 - accuracy: 0.9456\n",
      "training time:  13.5566 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.785]\n",
      "global epoch:  1243\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1598 - accuracy: 0.9452\n",
      "training time:  13.6995 s.\n",
      "average model starts\n",
      "average model ends:  0.1047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.785 0.8  ]\n",
      "global epoch:  1244\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1662 - accuracy: 0.9453\n",
      "training time:  13.7323 s.\n",
      "average model starts\n",
      "average model ends:  0.1435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.785 0.8   0.8  ]\n",
      "global epoch:  1245\n",
      "virtual memory used:  84.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1866 - accuracy: 0.9382\n",
      "training time:  13.457 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.802]\n",
      "global epoch:  1246\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1693 - accuracy: 0.9473\n",
      "training time:  14.4101 s.\n",
      "average model starts\n",
      "average model ends:  0.0915 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.804]\n",
      "global epoch:  1247\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1741 - accuracy: 0.9433\n",
      "training time:  13.4276 s.\n",
      "average model starts\n",
      "average model ends:  0.0602 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.804 0.804]\n",
      "global epoch:  1248\n",
      "virtual memory used:  85.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1739 - accuracy: 0.9430\n",
      "training time:  12.5977 s.\n",
      "average model starts\n",
      "average model ends:  0.0547 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.804 0.797]\n",
      "global epoch:  1249\n",
      "virtual memory used:  78.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1566 - accuracy: 0.9475\n",
      "training time:  14.5368 s.\n",
      "average model starts\n",
      "average model ends:  0.1266 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.797 0.8  ]\n",
      "global epoch:  1250\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1594 - accuracy: 0.9485\n",
      "training time:  13.8475 s.\n",
      "average model starts\n",
      "average model ends:  0.103 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.789]\n",
      "global epoch:  1251\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1821 - accuracy: 0.9408\n",
      "training time:  11.6917 s.\n",
      "average model starts\n",
      "average model ends:  0.0929 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.789 0.8  ]\n",
      "global epoch:  1252\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1767 - accuracy: 0.9387\n",
      "training time:  13.9746 s.\n",
      "average model starts\n",
      "average model ends:  0.1172 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.8   0.802]\n",
      "global epoch:  1253\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1673 - accuracy: 0.9442\n",
      "training time:  12.0751 s.\n",
      "average model starts\n",
      "average model ends:  0.1364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.802]\n",
      "global epoch:  1254\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1628 - accuracy: 0.9463\n",
      "training time:  14.2151 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.801]\n",
      "global epoch:  1255\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1796 - accuracy: 0.9406\n",
      "training time:  11.7525 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.801]\n",
      "global epoch:  1256\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1597 - accuracy: 0.9493\n",
      "training time:  13.0482 s.\n",
      "average model starts\n",
      "average model ends:  0.1448 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.797]\n",
      "global epoch:  1257\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1682 - accuracy: 0.9464\n",
      "training time:  13.8464 s.\n",
      "average model starts\n",
      "average model ends:  0.0389 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.794]\n",
      "global epoch:  1258\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1670 - accuracy: 0.9466\n",
      "training time:  13.039 s.\n",
      "average model starts\n",
      "average model ends:  0.0338 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.799]\n",
      "global epoch:  1259\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1701 - accuracy: 0.9447\n",
      "training time:  14.4151 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.799]\n",
      "global epoch:  1260\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1713 - accuracy: 0.9428\n",
      "training time:  14.027 s.\n",
      "average model starts\n",
      "average model ends:  0.091 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.795]\n",
      "global epoch:  1261\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1700 - accuracy: 0.9449\n",
      "training time:  14.5086 s.\n",
      "average model starts\n",
      "average model ends:  0.1106 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.803]\n",
      "global epoch:  1262\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1709 - accuracy: 0.9452\n",
      "training time:  11.9601 s.\n",
      "average model starts\n",
      "average model ends:  0.0344 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.803 0.797]\n",
      "global epoch:  1263\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1718 - accuracy: 0.9464\n",
      "training time:  13.1021 s.\n",
      "average model starts\n",
      "average model ends:  0.1187 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.797 0.799]\n",
      "global epoch:  1264\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1718 - accuracy: 0.9439\n",
      "training time:  12.2986 s.\n",
      "average model starts\n",
      "average model ends:  0.0399 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.795]\n",
      "global epoch:  1265\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1626 - accuracy: 0.9468\n",
      "training time:  11.5948 s.\n",
      "average model starts\n",
      "average model ends:  0.0917 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.79 ]\n",
      "global epoch:  1266\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1818 - accuracy: 0.9385\n",
      "training time:  13.7337 s.\n",
      "average model starts\n",
      "average model ends:  0.0539 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.79  0.79 ]\n",
      "global epoch:  1267\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1604 - accuracy: 0.9453\n",
      "training time:  13.5059 s.\n",
      "average model starts\n",
      "average model ends:  0.0943 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.79  0.793]\n",
      "global epoch:  1268\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1526 - accuracy: 0.9499\n",
      "training time:  14.1109 s.\n",
      "average model starts\n",
      "average model ends:  0.0592 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.793 0.803]\n",
      "global epoch:  1269\n",
      "virtual memory used:  82.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1619 - accuracy: 0.9494\n",
      "training time:  14.02 s.\n",
      "average model starts\n",
      "average model ends:  0.0807 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.803 0.805]\n",
      "global epoch:  1270\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1589 - accuracy: 0.9492\n",
      "training time:  14.2519 s.\n",
      "average model starts\n",
      "average model ends:  0.0782 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.805 0.793]\n",
      "global epoch:  1271\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1619 - accuracy: 0.9458\n",
      "training time:  13.3172 s.\n",
      "average model starts\n",
      "average model ends:  0.0472 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.793 0.796]\n",
      "global epoch:  1272\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1682 - accuracy: 0.9405\n",
      "training time:  14.3423 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.796 0.798]\n",
      "global epoch:  1273\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1672 - accuracy: 0.9433\n",
      "training time:  13.1483 s.\n",
      "average model starts\n",
      "average model ends:  0.09 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.796]\n",
      "global epoch:  1274\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1642 - accuracy: 0.9452\n",
      "training time:  13.6896 s.\n",
      "average model starts\n",
      "average model ends:  0.0376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.793]\n",
      "global epoch:  1275\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1665 - accuracy: 0.9431\n",
      "training time:  14.1425 s.\n",
      "average model starts\n",
      "average model ends:  0.0413 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.801]\n",
      "global epoch:  1276\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 54ms/step - loss: 0.1692 - accuracy: 0.9436\n",
      "training time:  13.6777 s.\n",
      "average model starts\n",
      "average model ends:  0.0427 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.801 0.79 ]\n",
      "global epoch:  1277\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1781 - accuracy: 0.9415\n",
      "training time:  13.944 s.\n",
      "average model starts\n",
      "average model ends:  0.1409 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.79  0.802]\n",
      "global epoch:  1278\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1678 - accuracy: 0.9415\n",
      "training time:  13.9222 s.\n",
      "average model starts\n",
      "average model ends:  0.1101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.801]\n",
      "global epoch:  1279\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1646 - accuracy: 0.9453\n",
      "training time:  13.56 s.\n",
      "average model starts\n",
      "average model ends:  0.0567 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.803]\n",
      "global epoch:  1280\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1624 - accuracy: 0.9468\n",
      "training time:  13.7087 s.\n",
      "average model starts\n",
      "average model ends:  0.0493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.803 0.796]\n",
      "global epoch:  1281\n",
      "virtual memory used:  83.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1740 - accuracy: 0.9396\n",
      "training time:  11.2812 s.\n",
      "average model starts\n",
      "average model ends:  0.0718 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.794]\n",
      "global epoch:  1282\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1695 - accuracy: 0.9453\n",
      "training time:  13.1351 s.\n",
      "average model starts\n",
      "average model ends:  0.0383 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.794 0.797]\n",
      "global epoch:  1283\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1663 - accuracy: 0.9473\n",
      "training time:  13.9308 s.\n",
      "average model starts\n",
      "average model ends:  0.061 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.793]\n",
      "global epoch:  1284\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1581 - accuracy: 0.9456\n",
      "training time:  11.9701 s.\n",
      "average model starts\n",
      "average model ends:  0.0407 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.793 0.797]\n",
      "global epoch:  1285\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1674 - accuracy: 0.9441\n",
      "training time:  12.6914 s.\n",
      "average model starts\n",
      "average model ends:  0.0985 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.791]\n",
      "global epoch:  1286\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1531 - accuracy: 0.9514\n",
      "training time:  13.8655 s.\n",
      "average model starts\n",
      "average model ends:  0.0675 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.793]\n",
      "global epoch:  1287\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1584 - accuracy: 0.9482\n",
      "training time:  13.8327 s.\n",
      "average model starts\n",
      "average model ends:  0.0503 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.793 0.79 ]\n",
      "global epoch:  1288\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 15s 58ms/step - loss: 0.1770 - accuracy: 0.9441\n",
      "training time:  14.8233 s.\n",
      "average model starts\n",
      "average model ends:  0.1184 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.79  0.796]\n",
      "global epoch:  1289\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1727 - accuracy: 0.9452\n",
      "training time:  14.1027 s.\n",
      "average model starts\n",
      "average model ends:  0.0506 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.796 0.791]\n",
      "global epoch:  1290\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1811 - accuracy: 0.9404\n",
      "training time:  14.2382 s.\n",
      "average model starts\n",
      "average model ends:  0.0668 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.791 0.802]\n",
      "global epoch:  1291\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1621 - accuracy: 0.9478\n",
      "training time:  14.0222 s.\n",
      "average model starts\n",
      "average model ends:  0.0576 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.802 0.804]\n",
      "global epoch:  1292\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.1572 - accuracy: 0.9470\n",
      "training time:  14.748 s.\n",
      "average model starts\n",
      "average model ends:  0.0685 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.804 0.801]\n",
      "global epoch:  1293\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1667 - accuracy: 0.9469\n",
      "training time:  13.9981 s.\n",
      "average model starts\n",
      "average model ends:  0.1109 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.795]\n",
      "global epoch:  1294\n",
      "virtual memory used:  84.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1555 - accuracy: 0.9493\n",
      "training time:  12.2304 s.\n",
      "average model starts\n",
      "average model ends:  0.0615 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.801]\n",
      "global epoch:  1295\n",
      "virtual memory used:  85.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1707 - accuracy: 0.9480\n",
      "training time:  10.5917 s.\n",
      "average model starts\n",
      "average model ends:  0.062 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.8  ]\n",
      "global epoch:  1296\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1586 - accuracy: 0.9463\n",
      "training time:  12.4946 s.\n",
      "average model starts\n",
      "average model ends:  0.0367 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.798]\n",
      "global epoch:  1297\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1611 - accuracy: 0.9487\n",
      "training time:  13.6936 s.\n",
      "average model starts\n",
      "average model ends:  0.0697 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.801]\n",
      "global epoch:  1298\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1689 - accuracy: 0.9416\n",
      "training time:  13.3153 s.\n",
      "average model starts\n",
      "average model ends:  0.0397 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.797]\n",
      "global epoch:  1299\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1653 - accuracy: 0.9449\n",
      "training time:  13.0129 s.\n",
      "average model starts\n",
      "average model ends:  0.0353 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.799]\n",
      "global epoch:  1300\n",
      "virtual memory used:  79.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1690 - accuracy: 0.9477\n",
      "training time:  13.9505 s.\n",
      "average model starts\n",
      "average model ends:  0.0367 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.79 ]\n",
      "global epoch:  1301\n",
      "virtual memory used:  79.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1640 - accuracy: 0.9454\n",
      "training time:  13.1783 s.\n",
      "average model starts\n",
      "average model ends:  0.1452 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.79  0.798]\n",
      "global epoch:  1302\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1633 - accuracy: 0.9451\n",
      "training time:  12.0415 s.\n",
      "average model starts\n",
      "average model ends:  0.0388 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.798 0.809]\n",
      "global epoch:  1303\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1730 - accuracy: 0.9421\n",
      "training time:  13.2488 s.\n",
      "average model starts\n",
      "average model ends:  0.0822 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.809 0.804]\n",
      "global epoch:  1304\n",
      "virtual memory used:  79.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1715 - accuracy: 0.9437\n",
      "training time:  12.3013 s.\n",
      "average model starts\n",
      "average model ends:  0.0586 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.809 0.804 0.788]\n",
      "global epoch:  1305\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1651 - accuracy: 0.9475\n",
      "training time:  14.1472 s.\n",
      "average model starts\n",
      "average model ends:  0.0854 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.788 0.796]\n",
      "global epoch:  1306\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1725 - accuracy: 0.9408\n",
      "training time:  14.2377 s.\n",
      "average model starts\n",
      "average model ends:  0.0583 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.796 0.792]\n",
      "global epoch:  1307\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1672 - accuracy: 0.9466\n",
      "training time:  14.0025 s.\n",
      "average model starts\n",
      "average model ends:  0.0656 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.792 0.795]\n",
      "global epoch:  1308\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1661 - accuracy: 0.9458\n",
      "training time:  13.5914 s.\n",
      "average model starts\n",
      "average model ends:  0.1534 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.796]\n",
      "global epoch:  1309\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1605 - accuracy: 0.9493\n",
      "training time:  12.6056 s.\n",
      "average model starts\n",
      "average model ends:  0.1027 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.806]\n",
      "global epoch:  1310\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1621 - accuracy: 0.9464\n",
      "training time:  12.7854 s.\n",
      "average model starts\n",
      "average model ends:  0.0851 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.806 0.801]\n",
      "global epoch:  1311\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1740 - accuracy: 0.9441\n",
      "training time:  13.0987 s.\n",
      "average model starts\n",
      "average model ends:  0.0431 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.801 0.79 ]\n",
      "global epoch:  1312\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1549 - accuracy: 0.9482\n",
      "training time:  13.1581 s.\n",
      "average model starts\n",
      "average model ends:  0.0659 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.79  0.802]\n",
      "global epoch:  1313\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1653 - accuracy: 0.9492\n",
      "training time:  13.4312 s.\n",
      "average model starts\n",
      "average model ends:  0.1 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.795]\n",
      "global epoch:  1314\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1578 - accuracy: 0.9484\n",
      "training time:  13.6644 s.\n",
      "average model starts\n",
      "average model ends:  0.1066 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.801]\n",
      "global epoch:  1315\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1718 - accuracy: 0.9466\n",
      "training time:  13.0992 s.\n",
      "average model starts\n",
      "average model ends:  0.1189 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.799]\n",
      "global epoch:  1316\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1659 - accuracy: 0.9472\n",
      "training time:  13.7034 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.797]\n",
      "global epoch:  1317\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1586 - accuracy: 0.9484\n",
      "training time:  14.0714 s.\n",
      "average model starts\n",
      "average model ends:  0.1259 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.794]\n",
      "global epoch:  1318\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1679 - accuracy: 0.9422\n",
      "training time:  11.8652 s.\n",
      "average model starts\n",
      "average model ends:  0.0355 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.795]\n",
      "global epoch:  1319\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1627 - accuracy: 0.9504\n",
      "training time:  12.5614 s.\n",
      "average model starts\n",
      "average model ends:  0.1105 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.795 0.804]\n",
      "global epoch:  1320\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1651 - accuracy: 0.9451\n",
      "training time:  13.0993 s.\n",
      "average model starts\n",
      "average model ends:  0.0433 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.804 0.798]\n",
      "global epoch:  1321\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1634 - accuracy: 0.9470\n",
      "training time:  14.523 s.\n",
      "average model starts\n",
      "average model ends:  0.0702 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.798 0.805]\n",
      "global epoch:  1322\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1575 - accuracy: 0.9488\n",
      "training time:  12.1248 s.\n",
      "average model starts\n",
      "average model ends:  0.0446 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.805 0.797]\n",
      "global epoch:  1323\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1588 - accuracy: 0.9474\n",
      "training time:  13.5867 s.\n",
      "average model starts\n",
      "average model ends:  0.0816 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.797 0.8  ]\n",
      "global epoch:  1324\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1701 - accuracy: 0.9441\n",
      "training time:  12.3954 s.\n",
      "average model starts\n",
      "average model ends:  0.0457 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.796]\n",
      "global epoch:  1325\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1541 - accuracy: 0.9500\n",
      "training time:  11.5562 s.\n",
      "average model starts\n",
      "average model ends:  0.0668 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.802]\n",
      "global epoch:  1326\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1598 - accuracy: 0.9457\n",
      "training time:  13.0477 s.\n",
      "average model starts\n",
      "average model ends:  0.043 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.796]\n",
      "global epoch:  1327\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1794 - accuracy: 0.9422\n",
      "training time:  13.8827 s.\n",
      "average model starts\n",
      "average model ends:  0.0558 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.8  ]\n",
      "global epoch:  1328\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1644 - accuracy: 0.9480\n",
      "training time:  13.454 s.\n",
      "average model starts\n",
      "average model ends:  0.0459 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.808]\n",
      "global epoch:  1329\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1643 - accuracy: 0.9469\n",
      "training time:  11.1378 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.808 0.793]\n",
      "global epoch:  1330\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1691 - accuracy: 0.9444\n",
      "training time:  12.6402 s.\n",
      "average model starts\n",
      "average model ends:  0.0537 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.808 0.793 0.793]\n",
      "global epoch:  1331\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1751 - accuracy: 0.9439\n",
      "training time:  13.29 s.\n",
      "average model starts\n",
      "average model ends:  0.0646 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.793 0.788]\n",
      "global epoch:  1332\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1725 - accuracy: 0.9452\n",
      "training time:  14.1904 s.\n",
      "average model starts\n",
      "average model ends:  0.1498 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.788 0.793]\n",
      "global epoch:  1333\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1588 - accuracy: 0.9483\n",
      "training time:  13.992 s.\n",
      "average model starts\n",
      "average model ends:  0.1129 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.797]\n",
      "global epoch:  1334\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1739 - accuracy: 0.9408\n",
      "training time:  13.5082 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.801]\n",
      "global epoch:  1335\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1630 - accuracy: 0.9470\n",
      "training time:  12.1533 s.\n",
      "average model starts\n",
      "average model ends:  0.0347 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.794]\n",
      "global epoch:  1336\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1597 - accuracy: 0.9463\n",
      "training time:  14.3023 s.\n",
      "average model starts\n",
      "average model ends:  0.1248 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.805]\n",
      "global epoch:  1337\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1680 - accuracy: 0.9469\n",
      "training time:  13.3318 s.\n",
      "average model starts\n",
      "average model ends:  0.0865 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.805 0.802]\n",
      "global epoch:  1338\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1549 - accuracy: 0.9482\n",
      "training time:  11.3024 s.\n",
      "average model starts\n",
      "average model ends:  0.0638 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.802 0.802]\n",
      "global epoch:  1339\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1628 - accuracy: 0.9474\n",
      "training time:  13.8738 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.799]\n",
      "global epoch:  1340\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1623 - accuracy: 0.9475\n",
      "training time:  13.5908 s.\n",
      "average model starts\n",
      "average model ends:  0.0364 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.798]\n",
      "global epoch:  1341\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1627 - accuracy: 0.9473\n",
      "training time:  14.3289 s.\n",
      "average model starts\n",
      "average model ends:  0.0347 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.798 0.791]\n",
      "global epoch:  1342\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1699 - accuracy: 0.9418\n",
      "training time:  12.9248 s.\n",
      "average model starts\n",
      "average model ends:  0.0783 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.806]\n",
      "global epoch:  1343\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1651 - accuracy: 0.9458\n",
      "training time:  14.1443 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.806 0.791]\n",
      "global epoch:  1344\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1671 - accuracy: 0.9444\n",
      "training time:  13.0094 s.\n",
      "average model starts\n",
      "average model ends:  0.0875 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.791 0.797]\n",
      "global epoch:  1345\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1634 - accuracy: 0.9456\n",
      "training time:  14.0809 s.\n",
      "average model starts\n",
      "average model ends:  0.0821 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.797 0.795]\n",
      "global epoch:  1346\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1669 - accuracy: 0.9444\n",
      "training time:  13.1374 s.\n",
      "average model starts\n",
      "average model ends:  0.0355 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.796]\n",
      "global epoch:  1347\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1733 - accuracy: 0.9466\n",
      "training time:  13.2346 s.\n",
      "average model starts\n",
      "average model ends:  0.0829 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.796 0.801]\n",
      "global epoch:  1348\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1577 - accuracy: 0.9487\n",
      "training time:  13.8689 s.\n",
      "average model starts\n",
      "average model ends:  0.0504 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.792]\n",
      "global epoch:  1349\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 15s 58ms/step - loss: 0.1808 - accuracy: 0.9395\n",
      "training time:  14.6511 s.\n",
      "average model starts\n",
      "average model ends:  0.1179 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.792 0.804]\n",
      "global epoch:  1350\n",
      "virtual memory used:  79.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1693 - accuracy: 0.9431\n",
      "training time:  12.7516 s.\n",
      "average model starts\n",
      "average model ends:  0.043 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.804 0.799]\n",
      "global epoch:  1351\n",
      "virtual memory used:  80.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1690 - accuracy: 0.9428\n",
      "training time:  11.7941 s.\n",
      "average model starts\n",
      "average model ends:  0.0573 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.799 0.795]\n",
      "global epoch:  1352\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1582 - accuracy: 0.9514\n",
      "training time:  11.8413 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.795 0.802]\n",
      "global epoch:  1353\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1658 - accuracy: 0.9464\n",
      "training time:  14.1123 s.\n",
      "average model starts\n",
      "average model ends:  0.094 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.8  ]\n",
      "global epoch:  1354\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1610 - accuracy: 0.9444\n",
      "training time:  11.6238 s.\n",
      "average model starts\n",
      "average model ends:  0.0581 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.794]\n",
      "global epoch:  1355\n",
      "virtual memory used:  79.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1589 - accuracy: 0.9497\n",
      "training time:  13.2061 s.\n",
      "average model starts\n",
      "average model ends:  0.084 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.803]\n",
      "global epoch:  1356\n",
      "virtual memory used:  80.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1501 - accuracy: 0.9519\n",
      "training time:  14.5926 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.803 0.803]\n",
      "global epoch:  1357\n",
      "virtual memory used:  81.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1505 - accuracy: 0.9518\n",
      "training time:  12.9067 s.\n",
      "average model starts\n",
      "average model ends:  0.1345 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.803 0.801]\n",
      "global epoch:  1358\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1678 - accuracy: 0.9453\n",
      "training time:  13.2125 s.\n",
      "average model starts\n",
      "average model ends:  0.0665 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.805]\n",
      "global epoch:  1359\n",
      "virtual memory used:  80.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1653 - accuracy: 0.9477\n",
      "training time:  13.1591 s.\n",
      "average model starts\n",
      "average model ends:  0.0962 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.805 0.799]\n",
      "global epoch:  1360\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1634 - accuracy: 0.9461\n",
      "training time:  13.2517 s.\n",
      "average model starts\n",
      "average model ends:  0.0889 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.799 0.796]\n",
      "global epoch:  1361\n",
      "virtual memory used:  81.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1831 - accuracy: 0.9427\n",
      "training time:  11.6061 s.\n",
      "average model starts\n",
      "average model ends:  0.0397 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.795]\n",
      "global epoch:  1362\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1593 - accuracy: 0.9484\n",
      "training time:  12.0949 s.\n",
      "average model starts\n",
      "average model ends:  0.101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.801]\n",
      "global epoch:  1363\n",
      "virtual memory used:  81.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1608 - accuracy: 0.9464\n",
      "training time:  12.4223 s.\n",
      "average model starts\n",
      "average model ends:  0.07 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.793]\n",
      "global epoch:  1364\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1575 - accuracy: 0.9485\n",
      "training time:  13.0553 s.\n",
      "average model starts\n",
      "average model ends:  0.0502 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.793 0.797]\n",
      "global epoch:  1365\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1740 - accuracy: 0.9437\n",
      "training time:  13.5522 s.\n",
      "average model starts\n",
      "average model ends:  0.1316 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.797 0.795]\n",
      "global epoch:  1366\n",
      "virtual memory used:  81.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1601 - accuracy: 0.9467\n",
      "training time:  12.946 s.\n",
      "average model starts\n",
      "average model ends:  0.0597 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.798]\n",
      "global epoch:  1367\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1629 - accuracy: 0.9475\n",
      "training time:  12.5254 s.\n",
      "average model starts\n",
      "average model ends:  0.0842 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.793]\n",
      "global epoch:  1368\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1608 - accuracy: 0.9477\n",
      "training time:  12.6743 s.\n",
      "average model starts\n",
      "average model ends:  0.0409 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.793 0.8  ]\n",
      "global epoch:  1369\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1483 - accuracy: 0.9520\n",
      "training time:  12.4525 s.\n",
      "average model starts\n",
      "average model ends:  0.1329 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.804]\n",
      "global epoch:  1370\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1676 - accuracy: 0.9470\n",
      "training time:  13.2605 s.\n",
      "average model starts\n",
      "average model ends:  0.0689 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.796]\n",
      "global epoch:  1371\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1590 - accuracy: 0.9477\n",
      "training time:  11.8136 s.\n",
      "average model starts\n",
      "average model ends:  0.0471 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.796 0.796]\n",
      "global epoch:  1372\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1622 - accuracy: 0.9463\n",
      "training time:  13.1655 s.\n",
      "average model starts\n",
      "average model ends:  0.0695 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.795]\n",
      "global epoch:  1373\n",
      "virtual memory used:  84.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1668 - accuracy: 0.9448\n",
      "training time:  13.8019 s.\n",
      "average model starts\n",
      "average model ends:  0.1156 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.8  ]\n",
      "global epoch:  1374\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1725 - accuracy: 0.9422\n",
      "training time:  14.0442 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.798]\n",
      "global epoch:  1375\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1710 - accuracy: 0.9456\n",
      "training time:  12.8384 s.\n",
      "average model starts\n",
      "average model ends:  0.074 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.798]\n",
      "global epoch:  1376\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1742 - accuracy: 0.9426\n",
      "training time:  12.9599 s.\n",
      "average model starts\n",
      "average model ends:  0.0665 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.795]\n",
      "global epoch:  1377\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1632 - accuracy: 0.9452\n",
      "training time:  12.7093 s.\n",
      "average model starts\n",
      "average model ends:  0.0963 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.795 0.8  ]\n",
      "global epoch:  1378\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1739 - accuracy: 0.9443\n",
      "training time:  13.2431 s.\n",
      "average model starts\n",
      "average model ends:  0.0835 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.8   0.804]\n",
      "global epoch:  1379\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1626 - accuracy: 0.9478\n",
      "training time:  13.683 s.\n",
      "average model starts\n",
      "average model ends:  0.1027 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.809]\n",
      "global epoch:  1380\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1614 - accuracy: 0.9482\n",
      "training time:  13.851 s.\n",
      "average model starts\n",
      "average model ends:  0.122 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.809 0.799]\n",
      "global epoch:  1381\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1653 - accuracy: 0.9474\n",
      "training time:  13.4815 s.\n",
      "average model starts\n",
      "average model ends:  0.0392 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.809 0.799 0.8  ]\n",
      "global epoch:  1382\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1596 - accuracy: 0.9459\n",
      "training time:  11.4379 s.\n",
      "average model starts\n",
      "average model ends:  0.0438 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.79 ]\n",
      "global epoch:  1383\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1475 - accuracy: 0.9504\n",
      "training time:  12.1052 s.\n",
      "average model starts\n",
      "average model ends:  0.0886 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.79  0.802]\n",
      "global epoch:  1384\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1567 - accuracy: 0.9473\n",
      "training time:  11.8463 s.\n",
      "average model starts\n",
      "average model ends:  0.0754 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.802 0.792]\n",
      "global epoch:  1385\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1616 - accuracy: 0.9459\n",
      "training time:  11.8284 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.799]\n",
      "global epoch:  1386\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1740 - accuracy: 0.9439\n",
      "training time:  11.5991 s.\n",
      "average model starts\n",
      "average model ends:  0.1193 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.799 0.799]\n",
      "global epoch:  1387\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1643 - accuracy: 0.9461\n",
      "training time:  13.1398 s.\n",
      "average model starts\n",
      "average model ends:  0.0851 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.796]\n",
      "global epoch:  1388\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1419 - accuracy: 0.9530\n",
      "training time:  13.242 s.\n",
      "average model starts\n",
      "average model ends:  0.0579 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.798]\n",
      "global epoch:  1389\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1664 - accuracy: 0.9438\n",
      "training time:  11.5283 s.\n",
      "average model starts\n",
      "average model ends:  0.056 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.802]\n",
      "global epoch:  1390\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1585 - accuracy: 0.9485\n",
      "training time:  12.1507 s.\n",
      "average model starts\n",
      "average model ends:  0.106 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.8  ]\n",
      "global epoch:  1391\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1565 - accuracy: 0.9493\n",
      "training time:  13.2962 s.\n",
      "average model starts\n",
      "average model ends:  0.142 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.795]\n",
      "global epoch:  1392\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1643 - accuracy: 0.9441\n",
      "training time:  13.4311 s.\n",
      "average model starts\n",
      "average model ends:  0.0618 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.803]\n",
      "global epoch:  1393\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1547 - accuracy: 0.9504\n",
      "training time:  11.6291 s.\n",
      "average model starts\n",
      "average model ends:  0.0777 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.803 0.797]\n",
      "global epoch:  1394\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1600 - accuracy: 0.9478\n",
      "training time:  13.4274 s.\n",
      "average model starts\n",
      "average model ends:  0.0489 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.797 0.8  ]\n",
      "global epoch:  1395\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1632 - accuracy: 0.9493\n",
      "training time:  12.4937 s.\n",
      "average model starts\n",
      "average model ends:  0.0666 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.802]\n",
      "global epoch:  1396\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1709 - accuracy: 0.9452\n",
      "training time:  12.209 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.797]\n",
      "global epoch:  1397\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1677 - accuracy: 0.9474\n",
      "training time:  11.3715 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.806]\n",
      "global epoch:  1398\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1583 - accuracy: 0.9501\n",
      "training time:  11.5891 s.\n",
      "average model starts\n",
      "average model ends:  0.1203 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.806 0.803]\n",
      "global epoch:  1399\n",
      "virtual memory used:  79.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1741 - accuracy: 0.9451\n",
      "training time:  11.2841 s.\n",
      "average model starts\n",
      "average model ends:  0.0398 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.803 0.804]\n",
      "global epoch:  1400\n",
      "virtual memory used:  80.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1728 - accuracy: 0.9461\n",
      "training time:  14.0885 s.\n",
      "average model starts\n",
      "average model ends:  0.0949 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.804 0.796]\n",
      "global epoch:  1401\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1625 - accuracy: 0.9459\n",
      "training time:  13.7572 s.\n",
      "average model starts\n",
      "average model ends:  0.0647 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.796 0.802]\n",
      "global epoch:  1402\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1537 - accuracy: 0.9500\n",
      "training time:  12.1764 s.\n",
      "average model starts\n",
      "average model ends:  0.0757 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.798]\n",
      "global epoch:  1403\n",
      "virtual memory used:  80.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1651 - accuracy: 0.9485\n",
      "training time:  13.717 s.\n",
      "average model starts\n",
      "average model ends:  0.0455 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.798 0.802]\n",
      "global epoch:  1404\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1729 - accuracy: 0.9430\n",
      "training time:  12.7338 s.\n",
      "average model starts\n",
      "average model ends:  0.0762 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.8  ]\n",
      "global epoch:  1405\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1768 - accuracy: 0.9459\n",
      "training time:  14.5062 s.\n",
      "average model starts\n",
      "average model ends:  0.0346 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.794]\n",
      "global epoch:  1406\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1649 - accuracy: 0.9444\n",
      "training time:  13.1039 s.\n",
      "average model starts\n",
      "average model ends:  0.0551 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.801]\n",
      "global epoch:  1407\n",
      "virtual memory used:  81.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1583 - accuracy: 0.9484\n",
      "training time:  13.6269 s.\n",
      "average model starts\n",
      "average model ends:  0.0982 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.796]\n",
      "global epoch:  1408\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1681 - accuracy: 0.9438\n",
      "training time:  13.4726 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.803]\n",
      "global epoch:  1409\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1596 - accuracy: 0.9479\n",
      "training time:  11.7197 s.\n",
      "average model starts\n",
      "average model ends:  0.101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.803]\n",
      "global epoch:  1410\n",
      "virtual memory used:  82.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1503 - accuracy: 0.9487\n",
      "training time:  14.0803 s.\n",
      "average model starts\n",
      "average model ends:  0.1069 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.803 0.8  ]\n",
      "global epoch:  1411\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1664 - accuracy: 0.9444\n",
      "training time:  13.6583 s.\n",
      "average model starts\n",
      "average model ends:  0.1437 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.8   0.797]\n",
      "global epoch:  1412\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1511 - accuracy: 0.9492\n",
      "training time:  13.3768 s.\n",
      "average model starts\n",
      "average model ends:  0.0506 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.798]\n",
      "global epoch:  1413\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 56ms/step - loss: 0.1515 - accuracy: 0.9498\n",
      "training time:  14.2298 s.\n",
      "average model starts\n",
      "average model ends:  0.0523 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.797]\n",
      "global epoch:  1414\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1614 - accuracy: 0.9493\n",
      "training time:  11.3648 s.\n",
      "average model starts\n",
      "average model ends:  0.0477 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.799]\n",
      "global epoch:  1415\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1645 - accuracy: 0.9452\n",
      "training time:  13.4938 s.\n",
      "average model starts\n",
      "average model ends:  0.0649 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.801]\n",
      "global epoch:  1416\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1597 - accuracy: 0.9469\n",
      "training time:  13.5487 s.\n",
      "average model starts\n",
      "average model ends:  0.0675 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.796]\n",
      "global epoch:  1417\n",
      "virtual memory used:  82.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1623 - accuracy: 0.9441\n",
      "training time:  13.1311 s.\n",
      "average model starts\n",
      "average model ends:  0.1459 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.792]\n",
      "global epoch:  1418\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1636 - accuracy: 0.9485\n",
      "training time:  11.6104 s.\n",
      "average model starts\n",
      "average model ends:  0.0352 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.792 0.795]\n",
      "global epoch:  1419\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1639 - accuracy: 0.9469\n",
      "training time:  13.3793 s.\n",
      "average model starts\n",
      "average model ends:  0.1093 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.795 0.802]\n",
      "global epoch:  1420\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1788 - accuracy: 0.9415\n",
      "training time:  12.8175 s.\n",
      "average model starts\n",
      "average model ends:  0.0607 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.806]\n",
      "global epoch:  1421\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1526 - accuracy: 0.9474\n",
      "training time:  13.4365 s.\n",
      "average model starts\n",
      "average model ends:  0.034 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.806 0.796]\n",
      "global epoch:  1422\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1713 - accuracy: 0.9456\n",
      "training time:  13.297 s.\n",
      "average model starts\n",
      "average model ends:  0.1057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.796 0.802]\n",
      "global epoch:  1423\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1649 - accuracy: 0.9446\n",
      "training time:  12.683 s.\n",
      "average model starts\n",
      "average model ends:  0.1393 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.802 0.796]\n",
      "global epoch:  1424\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1577 - accuracy: 0.9490\n",
      "training time:  13.9444 s.\n",
      "average model starts\n",
      "average model ends:  0.0829 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.804]\n",
      "global epoch:  1425\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1622 - accuracy: 0.9452\n",
      "training time:  13.4721 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.804 0.803]\n",
      "global epoch:  1426\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1691 - accuracy: 0.9437\n",
      "training time:  12.8535 s.\n",
      "average model starts\n",
      "average model ends:  0.0624 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.803 0.8  ]\n",
      "global epoch:  1427\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1610 - accuracy: 0.9446\n",
      "training time:  11.3638 s.\n",
      "average model starts\n",
      "average model ends:  0.0472 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.8   0.803]\n",
      "global epoch:  1428\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1617 - accuracy: 0.9459\n",
      "training time:  13.6906 s.\n",
      "average model starts\n",
      "average model ends:  0.0898 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.803 0.803]\n",
      "global epoch:  1429\n",
      "virtual memory used:  84.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1598 - accuracy: 0.9478\n",
      "training time:  11.8531 s.\n",
      "average model starts\n",
      "average model ends:  0.0487 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.803 0.803]\n",
      "global epoch:  1430\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1682 - accuracy: 0.9480\n",
      "training time:  12.5887 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.803 0.809]\n",
      "global epoch:  1431\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1491 - accuracy: 0.9518\n",
      "training time:  12.575 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.809 0.797]\n",
      "global epoch:  1432\n",
      "virtual memory used:  86.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1545 - accuracy: 0.9505\n",
      "training time:  13.5887 s.\n",
      "average model starts\n",
      "average model ends:  0.1008 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.809 0.797 0.788]\n",
      "global epoch:  1433\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1756 - accuracy: 0.9436\n",
      "training time:  13.5423 s.\n",
      "average model starts\n",
      "average model ends:  0.0792 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.788 0.793]\n",
      "global epoch:  1434\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1664 - accuracy: 0.9444\n",
      "training time:  12.4616 s.\n",
      "average model starts\n",
      "average model ends:  0.0624 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.793 0.801]\n",
      "global epoch:  1435\n",
      "virtual memory used:  86.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1751 - accuracy: 0.9421\n",
      "training time:  13.5151 s.\n",
      "average model starts\n",
      "average model ends:  0.047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.801 0.806]\n",
      "global epoch:  1436\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1617 - accuracy: 0.9487\n",
      "training time:  12.301 s.\n",
      "average model starts\n",
      "average model ends:  0.0377 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.806 0.802]\n",
      "global epoch:  1437\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1640 - accuracy: 0.9461\n",
      "training time:  10.847 s.\n",
      "average model starts\n",
      "average model ends:  0.0397 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.802 0.803]\n",
      "global epoch:  1438\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1733 - accuracy: 0.9437\n",
      "training time:  11.9778 s.\n",
      "average model starts\n",
      "average model ends:  0.0613 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.803 0.794]\n",
      "global epoch:  1439\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1564 - accuracy: 0.9457\n",
      "training time:  13.3252 s.\n",
      "average model starts\n",
      "average model ends:  0.0491 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.794 0.796]\n",
      "global epoch:  1440\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1726 - accuracy: 0.9467\n",
      "training time:  12.8974 s.\n",
      "average model starts\n",
      "average model ends:  0.0625 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.796 0.798]\n",
      "global epoch:  1441\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1573 - accuracy: 0.9468\n",
      "training time:  13.2221 s.\n",
      "average model starts\n",
      "average model ends:  0.1268 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.798]\n",
      "global epoch:  1442\n",
      "virtual memory used:  86.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1532 - accuracy: 0.9487\n",
      "training time:  12.6243 s.\n",
      "average model starts\n",
      "average model ends:  0.0718 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.799]\n",
      "global epoch:  1443\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1610 - accuracy: 0.9470\n",
      "training time:  12.253 s.\n",
      "average model starts\n",
      "average model ends:  0.0855 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.802]\n",
      "global epoch:  1444\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1584 - accuracy: 0.9482\n",
      "training time:  11.8245 s.\n",
      "average model starts\n",
      "average model ends:  0.1186 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.8  ]\n",
      "global epoch:  1445\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1625 - accuracy: 0.9442\n",
      "training time:  11.956 s.\n",
      "average model starts\n",
      "average model ends:  0.0716 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.8  ]\n",
      "global epoch:  1446\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1669 - accuracy: 0.9441\n",
      "training time:  12.5315 s.\n",
      "average model starts\n",
      "average model ends:  0.0971 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.801]\n",
      "global epoch:  1447\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1614 - accuracy: 0.9482\n",
      "training time:  13.1601 s.\n",
      "average model starts\n",
      "average model ends:  0.2181 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.792]\n",
      "global epoch:  1448\n",
      "virtual memory used:  80.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1608 - accuracy: 0.9485\n",
      "training time:  12.8626 s.\n",
      "average model starts\n",
      "average model ends:  0.06 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.792 0.796]\n",
      "global epoch:  1449\n",
      "virtual memory used:  81.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1661 - accuracy: 0.9461\n",
      "training time:  11.4787 s.\n",
      "average model starts\n",
      "average model ends:  0.0514 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.796 0.803]\n",
      "global epoch:  1450\n",
      "virtual memory used:  82.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1546 - accuracy: 0.9490\n",
      "training time:  13.1119 s.\n",
      "average model starts\n",
      "average model ends:  0.1123 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.789]\n",
      "global epoch:  1451\n",
      "virtual memory used:  82.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1621 - accuracy: 0.9459\n",
      "training time:  12.2734 s.\n",
      "average model starts\n",
      "average model ends:  0.0726 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.789 0.786]\n",
      "global epoch:  1452\n",
      "virtual memory used:  83.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1662 - accuracy: 0.9449\n",
      "training time:  12.5195 s.\n",
      "average model starts\n",
      "average model ends:  0.0938 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.786 0.807]\n",
      "global epoch:  1453\n",
      "virtual memory used:  81.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1598 - accuracy: 0.9500\n",
      "training time:  11.9334 s.\n",
      "average model starts\n",
      "average model ends:  0.0623 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.807 0.803]\n",
      "global epoch:  1454\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1654 - accuracy: 0.9430\n",
      "training time:  12.5434 s.\n",
      "average model starts\n",
      "average model ends:  0.0409 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.803 0.801]\n",
      "global epoch:  1455\n",
      "virtual memory used:  82.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1476 - accuracy: 0.9518\n",
      "training time:  12.1484 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.798]\n",
      "global epoch:  1456\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1627 - accuracy: 0.9489\n",
      "training time:  12.0865 s.\n",
      "average model starts\n",
      "average model ends:  0.0405 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.802]\n",
      "global epoch:  1457\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1699 - accuracy: 0.9449\n",
      "training time:  13.0748 s.\n",
      "average model starts\n",
      "average model ends:  0.0609 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.802]\n",
      "global epoch:  1458\n",
      "virtual memory used:  82.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1653 - accuracy: 0.9466\n",
      "training time:  12.8537 s.\n",
      "average model starts\n",
      "average model ends:  0.0443 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.79 ]\n",
      "global epoch:  1459\n",
      "virtual memory used:  82.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1604 - accuracy: 0.9441\n",
      "training time:  12.4013 s.\n",
      "average model starts\n",
      "average model ends:  0.0624 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.79  0.794]\n",
      "global epoch:  1460\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1569 - accuracy: 0.9479\n",
      "training time:  12.1678 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.794 0.803]\n",
      "global epoch:  1461\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1652 - accuracy: 0.9453\n",
      "training time:  13.3131 s.\n",
      "average model starts\n",
      "average model ends:  0.0731 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.803 0.795]\n",
      "global epoch:  1462\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1587 - accuracy: 0.9483\n",
      "training time:  13.939 s.\n",
      "average model starts\n",
      "average model ends:  0.0556 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.795 0.798]\n",
      "global epoch:  1463\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1569 - accuracy: 0.9479\n",
      "training time:  13.313 s.\n",
      "average model starts\n",
      "average model ends:  0.0674 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.803]\n",
      "global epoch:  1464\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1535 - accuracy: 0.9511\n",
      "training time:  11.3714 s.\n",
      "average model starts\n",
      "average model ends:  0.0454 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.803 0.801]\n",
      "global epoch:  1465\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1490 - accuracy: 0.9483\n",
      "training time:  11.4931 s.\n",
      "average model starts\n",
      "average model ends:  0.0412 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.801 0.795]\n",
      "global epoch:  1466\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1683 - accuracy: 0.9447\n",
      "training time:  11.9728 s.\n",
      "average model starts\n",
      "average model ends:  0.1097 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.795 0.801]\n",
      "global epoch:  1467\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1576 - accuracy: 0.9463\n",
      "training time:  12.8003 s.\n",
      "average model starts\n",
      "average model ends:  0.0399 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.801]\n",
      "global epoch:  1468\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1554 - accuracy: 0.9480\n",
      "training time:  12.8554 s.\n",
      "average model starts\n",
      "average model ends:  0.0394 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.804]\n",
      "global epoch:  1469\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1533 - accuracy: 0.9488\n",
      "training time:  11.8877 s.\n",
      "average model starts\n",
      "average model ends:  0.0376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.795]\n",
      "global epoch:  1470\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1655 - accuracy: 0.9452\n",
      "training time:  11.9543 s.\n",
      "average model starts\n",
      "average model ends:  0.0388 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.795 0.795]\n",
      "global epoch:  1471\n",
      "virtual memory used:  83.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1604 - accuracy: 0.9478\n",
      "training time:  11.3885 s.\n",
      "average model starts\n",
      "average model ends:  0.0355 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.789]\n",
      "global epoch:  1472\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1553 - accuracy: 0.9513\n",
      "training time:  13.3264 s.\n",
      "average model starts\n",
      "average model ends:  0.0595 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.789 0.805]\n",
      "global epoch:  1473\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1564 - accuracy: 0.9466\n",
      "training time:  12.3244 s.\n",
      "average model starts\n",
      "average model ends:  0.0506 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.805 0.801]\n",
      "global epoch:  1474\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1558 - accuracy: 0.9468\n",
      "training time:  11.2589 s.\n",
      "average model starts\n",
      "average model ends:  0.046 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.801 0.798]\n",
      "global epoch:  1475\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1680 - accuracy: 0.9426\n",
      "training time:  10.9999 s.\n",
      "average model starts\n",
      "average model ends:  0.047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.786]\n",
      "global epoch:  1476\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1644 - accuracy: 0.9488\n",
      "training time:  11.0039 s.\n",
      "average model starts\n",
      "average model ends:  0.0472 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.786 0.802]\n",
      "global epoch:  1477\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1584 - accuracy: 0.9479\n",
      "training time:  10.9107 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.802 0.802]\n",
      "global epoch:  1478\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1657 - accuracy: 0.9504\n",
      "training time:  10.7559 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.8  ]\n",
      "global epoch:  1479\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1697 - accuracy: 0.9461\n",
      "training time:  11.1639 s.\n",
      "average model starts\n",
      "average model ends:  0.0388 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.79 ]\n",
      "global epoch:  1480\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1599 - accuracy: 0.9478\n",
      "training time:  10.7773 s.\n",
      "average model starts\n",
      "average model ends:  0.0583 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.79  0.794]\n",
      "global epoch:  1481\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1648 - accuracy: 0.9478\n",
      "training time:  11.3698 s.\n",
      "average model starts\n",
      "average model ends:  0.0481 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.794 0.79 ]\n",
      "global epoch:  1482\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1490 - accuracy: 0.9534\n",
      "training time:  11.8586 s.\n",
      "average model starts\n",
      "average model ends:  0.058 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.79  0.789]\n",
      "global epoch:  1483\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1589 - accuracy: 0.9474\n",
      "training time:  11.985 s.\n",
      "average model starts\n",
      "average model ends:  0.0423 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.789 0.804]\n",
      "global epoch:  1484\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1701 - accuracy: 0.9444\n",
      "training time:  12.006 s.\n",
      "average model starts\n",
      "average model ends:  0.101 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.804 0.795]\n",
      "global epoch:  1485\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1548 - accuracy: 0.9472\n",
      "training time:  12.9327 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.795 0.801]\n",
      "global epoch:  1486\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1578 - accuracy: 0.9480\n",
      "training time:  12.9009 s.\n",
      "average model starts\n",
      "average model ends:  0.0481 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.801]\n",
      "global epoch:  1487\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1581 - accuracy: 0.9452\n",
      "training time:  11.6506 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.807]\n",
      "global epoch:  1488\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1597 - accuracy: 0.9516\n",
      "training time:  12.0342 s.\n",
      "average model starts\n",
      "average model ends:  0.0324 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.807 0.797]\n",
      "global epoch:  1489\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1643 - accuracy: 0.9474\n",
      "training time:  13.0061 s.\n",
      "average model starts\n",
      "average model ends:  0.0551 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.797 0.797]\n",
      "global epoch:  1490\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1550 - accuracy: 0.9484\n",
      "training time:  11.1529 s.\n",
      "average model starts\n",
      "average model ends:  0.0644 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.791]\n",
      "global epoch:  1491\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1599 - accuracy: 0.9478\n",
      "training time:  12.1252 s.\n",
      "average model starts\n",
      "average model ends:  0.0493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.791 0.792]\n",
      "global epoch:  1492\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1616 - accuracy: 0.9479\n",
      "training time:  12.0379 s.\n",
      "average model starts\n",
      "average model ends:  0.0402 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.792 0.792]\n",
      "global epoch:  1493\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1527 - accuracy: 0.9511\n",
      "training time:  12.5736 s.\n",
      "average model starts\n",
      "average model ends:  0.0527 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.792 0.802]\n",
      "global epoch:  1494\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1640 - accuracy: 0.9449\n",
      "training time:  12.6788 s.\n",
      "average model starts\n",
      "average model ends:  0.0536 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.802 0.799]\n",
      "global epoch:  1495\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1661 - accuracy: 0.9421\n",
      "training time:  13.5037 s.\n",
      "average model starts\n",
      "average model ends:  0.117 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.804]\n",
      "global epoch:  1496\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1623 - accuracy: 0.9477\n",
      "training time:  10.9549 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.804 0.799]\n",
      "global epoch:  1497\n",
      "virtual memory used:  89.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1597 - accuracy: 0.9480\n",
      "training time:  12.2559 s.\n",
      "average model starts\n",
      "average model ends:  0.0985 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.799 0.799]\n",
      "global epoch:  1498\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1553 - accuracy: 0.9485\n",
      "training time:  12.6164 s.\n",
      "average model starts\n",
      "average model ends:  0.0448 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.801]\n",
      "global epoch:  1499\n",
      "virtual memory used:  87.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1527 - accuracy: 0.9509\n",
      "training time:  11.2369 s.\n",
      "average model starts\n",
      "average model ends:  0.0468 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.799]\n",
      "global epoch:  1500\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1703 - accuracy: 0.9448\n",
      "training time:  11.173 s.\n",
      "average model starts\n",
      "average model ends:  0.0784 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.804]\n",
      "global epoch:  1501\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1641 - accuracy: 0.9493\n",
      "training time:  11.4678 s.\n",
      "average model starts\n",
      "average model ends:  0.0346 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.804 0.802]\n",
      "global epoch:  1502\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1559 - accuracy: 0.9485\n",
      "training time:  11.6136 s.\n",
      "average model starts\n",
      "average model ends:  0.0606 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.802 0.8  ]\n",
      "global epoch:  1503\n",
      "virtual memory used:  81.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1599 - accuracy: 0.9478\n",
      "training time:  11.2614 s.\n",
      "average model starts\n",
      "average model ends:  0.0421 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.801]\n",
      "global epoch:  1504\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1589 - accuracy: 0.9488\n",
      "training time:  11.1264 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.797]\n",
      "global epoch:  1505\n",
      "virtual memory used:  82.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1642 - accuracy: 0.9477\n",
      "training time:  11.6821 s.\n",
      "average model starts\n",
      "average model ends:  0.0607 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.799]\n",
      "global epoch:  1506\n",
      "virtual memory used:  81.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1694 - accuracy: 0.9475\n",
      "training time:  11.7235 s.\n",
      "average model starts\n",
      "average model ends:  0.0356 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.802]\n",
      "global epoch:  1507\n",
      "virtual memory used:  82.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1587 - accuracy: 0.9461\n",
      "training time:  12.8248 s.\n",
      "average model starts\n",
      "average model ends:  0.0663 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.801]\n",
      "global epoch:  1508\n",
      "virtual memory used:  83.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1555 - accuracy: 0.94930s - loss: 0.1551 - accu\n",
      "training time:  12.9064 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.804]\n",
      "global epoch:  1509\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1579 - accuracy: 0.9469\n",
      "training time:  12.7572 s.\n",
      "average model starts\n",
      "average model ends:  0.103 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.793]\n",
      "global epoch:  1510\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1579 - accuracy: 0.9510\n",
      "training time:  12.4995 s.\n",
      "average model starts\n",
      "average model ends:  0.0612 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.793 0.8  ]\n",
      "global epoch:  1511\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1672 - accuracy: 0.9492\n",
      "training time:  12.6472 s.\n",
      "average model starts\n",
      "average model ends:  0.0923 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.807]\n",
      "global epoch:  1512\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1570 - accuracy: 0.9490\n",
      "training time:  13.2687 s.\n",
      "average model starts\n",
      "average model ends:  0.1624 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.807 0.801]\n",
      "global epoch:  1513\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1604 - accuracy: 0.9495\n",
      "training time:  12.4804 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.801 0.797]\n",
      "global epoch:  1514\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1629 - accuracy: 0.9462\n",
      "training time:  12.4411 s.\n",
      "average model starts\n",
      "average model ends:  0.0596 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.799]\n",
      "global epoch:  1515\n",
      "virtual memory used:  83.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1535 - accuracy: 0.9524\n",
      "training time:  11.8608 s.\n",
      "average model starts\n",
      "average model ends:  0.0933 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.799 0.8  ]\n",
      "global epoch:  1516\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1705 - accuracy: 0.9446\n",
      "training time:  12.1005 s.\n",
      "average model starts\n",
      "average model ends:  0.0418 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.801]\n",
      "global epoch:  1517\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1686 - accuracy: 0.9446\n",
      "training time:  12.5476 s.\n",
      "average model starts\n",
      "average model ends:  0.0597 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.809]\n",
      "global epoch:  1518\n",
      "virtual memory used:  85.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1557 - accuracy: 0.9513\n",
      "training time:  11.6698 s.\n",
      "average model starts\n",
      "average model ends:  0.0831 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.809 0.812]\n",
      "global epoch:  1519\n",
      "virtual memory used:  83.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1522 - accuracy: 0.9495\n",
      "training time:  12.8478 s.\n",
      "average model starts\n",
      "average model ends:  0.0611 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.809 0.812 0.804]\n",
      "global epoch:  1520\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1596 - accuracy: 0.9480\n",
      "training time:  12.4993 s.\n",
      "average model starts\n",
      "average model ends:  0.0397 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.812 0.804 0.802]\n",
      "global epoch:  1521\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1579 - accuracy: 0.9487\n",
      "training time:  11.0415 s.\n",
      "average model starts\n",
      "average model ends:  0.0412 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.802 0.798]\n",
      "global epoch:  1522\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1573 - accuracy: 0.9458\n",
      "training time:  11.4163 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.798 0.806]\n",
      "global epoch:  1523\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1500 - accuracy: 0.9528\n",
      "training time:  11.7885 s.\n",
      "average model starts\n",
      "average model ends:  0.0835 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.806 0.802]\n",
      "global epoch:  1524\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1506 - accuracy: 0.9511\n",
      "training time:  13.3567 s.\n",
      "average model starts\n",
      "average model ends:  0.1014 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.802 0.799]\n",
      "global epoch:  1525\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1917 - accuracy: 0.9407\n",
      "training time:  12.5318 s.\n",
      "average model starts\n",
      "average model ends:  0.0633 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.799]\n",
      "global epoch:  1526\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1549 - accuracy: 0.9468\n",
      "training time:  11.1232 s.\n",
      "average model starts\n",
      "average model ends:  0.1501 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.799 0.796]\n",
      "global epoch:  1527\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1561 - accuracy: 0.9456\n",
      "training time:  12.2409 s.\n",
      "average model starts\n",
      "average model ends:  0.0792 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.8  ]\n",
      "global epoch:  1528\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1581 - accuracy: 0.9467\n",
      "training time:  12.2817 s.\n",
      "average model starts\n",
      "average model ends:  0.048 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.8  ]\n",
      "global epoch:  1529\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1547 - accuracy: 0.9490\n",
      "training time:  13.0515 s.\n",
      "average model starts\n",
      "average model ends:  0.037 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.801]\n",
      "global epoch:  1530\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1559 - accuracy: 0.9488\n",
      "training time:  10.875 s.\n",
      "average model starts\n",
      "average model ends:  0.0361 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.797]\n",
      "global epoch:  1531\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1551 - accuracy: 0.9492\n",
      "training time:  10.8928 s.\n",
      "average model starts\n",
      "average model ends:  0.0498 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.795]\n",
      "global epoch:  1532\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1564 - accuracy: 0.9492\n",
      "training time:  13.2056 s.\n",
      "average model starts\n",
      "average model ends:  0.0616 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.804]\n",
      "global epoch:  1533\n",
      "virtual memory used:  85.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1500 - accuracy: 0.9524\n",
      "training time:  13.8355 s.\n",
      "average model starts\n",
      "average model ends:  0.0642 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.804 0.793]\n",
      "global epoch:  1534\n",
      "virtual memory used:  86.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1664 - accuracy: 0.9462\n",
      "training time:  13.6079 s.\n",
      "average model starts\n",
      "average model ends:  0.0685 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.793 0.795]\n",
      "global epoch:  1535\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1499 - accuracy: 0.9515\n",
      "training time:  11.688 s.\n",
      "average model starts\n",
      "average model ends:  0.0321 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.795 0.795]\n",
      "global epoch:  1536\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1767 - accuracy: 0.9439\n",
      "training time:  11.4787 s.\n",
      "average model starts\n",
      "average model ends:  0.0648 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.795 0.797]\n",
      "global epoch:  1537\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1508 - accuracy: 0.9500\n",
      "training time:  11.7131 s.\n",
      "average model starts\n",
      "average model ends:  0.0647 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.797 0.796]\n",
      "global epoch:  1538\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1568 - accuracy: 0.9447\n",
      "training time:  12.167 s.\n",
      "average model starts\n",
      "average model ends:  0.0754 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.8  ]\n",
      "global epoch:  1539\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1580 - accuracy: 0.9511\n",
      "training time:  12.4719 s.\n",
      "average model starts\n",
      "average model ends:  0.0566 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.8   0.795]\n",
      "global epoch:  1540\n",
      "virtual memory used:  86.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1571 - accuracy: 0.9444\n",
      "training time:  11.1237 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.802]\n",
      "global epoch:  1541\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1713 - accuracy: 0.9444\n",
      "training time:  12.3986 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.8  ]\n",
      "global epoch:  1542\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1402 - accuracy: 0.9525\n",
      "training time:  12.6158 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.801]\n",
      "global epoch:  1543\n",
      "virtual memory used:  88.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1603 - accuracy: 0.9477\n",
      "training time:  11.3247 s.\n",
      "average model starts\n",
      "average model ends:  0.0878 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.794]\n",
      "global epoch:  1544\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1599 - accuracy: 0.9457\n",
      "training time:  10.793 s.\n",
      "average model starts\n",
      "average model ends:  0.0958 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.799]\n",
      "global epoch:  1545\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1697 - accuracy: 0.9458\n",
      "training time:  11.1051 s.\n",
      "average model starts\n",
      "average model ends:  0.0677 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.799 0.802]\n",
      "global epoch:  1546\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1533 - accuracy: 0.9485\n",
      "training time:  11.4321 s.\n",
      "average model starts\n",
      "average model ends:  0.0398 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.792]\n",
      "global epoch:  1547\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1577 - accuracy: 0.9482\n",
      "training time:  12.2829 s.\n",
      "average model starts\n",
      "average model ends:  0.036 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.792 0.8  ]\n",
      "global epoch:  1548\n",
      "virtual memory used:  88.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1543 - accuracy: 0.9521\n",
      "training time:  10.6239 s.\n",
      "average model starts\n",
      "average model ends:  0.0828 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.8   0.804]\n",
      "global epoch:  1549\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1638 - accuracy: 0.9463\n",
      "training time:  10.9452 s.\n",
      "average model starts\n",
      "average model ends:  0.0393 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.801]\n",
      "global epoch:  1550\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1535 - accuracy: 0.9519\n",
      "training time:  11.2354 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.802]\n",
      "global epoch:  1551\n",
      "virtual memory used:  88.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1557 - accuracy: 0.9534\n",
      "training time:  10.9785 s.\n",
      "average model starts\n",
      "average model ends:  0.0468 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.799]\n",
      "global epoch:  1552\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1787 - accuracy: 0.9441\n",
      "training time:  11.8701 s.\n",
      "average model starts\n",
      "average model ends:  0.1161 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.788]\n",
      "global epoch:  1553\n",
      "virtual memory used:  81.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1572 - accuracy: 0.9483\n",
      "training time:  12.395 s.\n",
      "average model starts\n",
      "average model ends:  0.0365 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.788 0.8  ]\n",
      "global epoch:  1554\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1662 - accuracy: 0.9456\n",
      "training time:  12.7516 s.\n",
      "average model starts\n",
      "average model ends:  0.0536 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.788 0.8   0.798]\n",
      "global epoch:  1555\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1673 - accuracy: 0.9479\n",
      "training time:  12.2222 s.\n",
      "average model starts\n",
      "average model ends:  0.1064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.807]\n",
      "global epoch:  1556\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1541 - accuracy: 0.9504\n",
      "training time:  12.381 s.\n",
      "average model starts\n",
      "average model ends:  0.0391 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.807 0.799]\n",
      "global epoch:  1557\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1667 - accuracy: 0.9489\n",
      "training time:  11.2769 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.799 0.802]\n",
      "global epoch:  1558\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1537 - accuracy: 0.9503\n",
      "training time:  13.6048 s.\n",
      "average model starts\n",
      "average model ends:  0.0436 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.8  ]\n",
      "global epoch:  1559\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1587 - accuracy: 0.9449\n",
      "training time:  11.9823 s.\n",
      "average model starts\n",
      "average model ends:  0.0513 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.802]\n",
      "global epoch:  1560\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1544 - accuracy: 0.9466\n",
      "training time:  12.0806 s.\n",
      "average model starts\n",
      "average model ends:  0.0811 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.797]\n",
      "global epoch:  1561\n",
      "virtual memory used:  83.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1468 - accuracy: 0.9498\n",
      "training time:  12.1926 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.797]\n",
      "global epoch:  1562\n",
      "virtual memory used:  83.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 40ms/step - loss: 0.1612 - accuracy: 0.9466\n",
      "training time:  10.2113 s.\n",
      "average model starts\n",
      "average model ends:  0.0387 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.797 0.794]\n",
      "global epoch:  1563\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1605 - accuracy: 0.9475\n",
      "training time:  11.5568 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.8  ]\n",
      "global epoch:  1564\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1620 - accuracy: 0.9469\n",
      "training time:  10.5056 s.\n",
      "average model starts\n",
      "average model ends:  0.068 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.8   0.8  ]\n",
      "global epoch:  1565\n",
      "virtual memory used:  83.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1626 - accuracy: 0.9458\n",
      "training time:  12.4055 s.\n",
      "average model starts\n",
      "average model ends:  0.1107 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.8  ]\n",
      "global epoch:  1566\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1532 - accuracy: 0.9485\n",
      "training time:  12.8895 s.\n",
      "average model starts\n",
      "average model ends:  0.0621 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.803]\n",
      "global epoch:  1567\n",
      "virtual memory used:  84.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1441 - accuracy: 0.9505\n",
      "training time:  12.6603 s.\n",
      "average model starts\n",
      "average model ends:  0.1323 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.803 0.798]\n",
      "global epoch:  1568\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1428 - accuracy: 0.9566\n",
      "training time:  11.8145 s.\n",
      "average model starts\n",
      "average model ends:  0.0417 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.798 0.794]\n",
      "global epoch:  1569\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1598 - accuracy: 0.9500\n",
      "training time:  12.8356 s.\n",
      "average model starts\n",
      "average model ends:  0.0701 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.794 0.798]\n",
      "global epoch:  1570\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1564 - accuracy: 0.9467\n",
      "training time:  13.233 s.\n",
      "average model starts\n",
      "average model ends:  0.0844 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.802]\n",
      "global epoch:  1571\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1714 - accuracy: 0.9416\n",
      "training time:  11.2793 s.\n",
      "average model starts\n",
      "average model ends:  0.1048 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.8  ]\n",
      "global epoch:  1572\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1517 - accuracy: 0.9478\n",
      "training time:  11.0603 s.\n",
      "average model starts\n",
      "average model ends:  0.0363 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.791]\n",
      "global epoch:  1573\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1659 - accuracy: 0.9477\n",
      "training time:  12.4688 s.\n",
      "average model starts\n",
      "average model ends:  0.0747 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.791 0.798]\n",
      "global epoch:  1574\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1634 - accuracy: 0.9462\n",
      "training time:  11.1102 s.\n",
      "average model starts\n",
      "average model ends:  0.0358 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.801]\n",
      "global epoch:  1575\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1560 - accuracy: 0.9479\n",
      "training time:  11.9734 s.\n",
      "average model starts\n",
      "average model ends:  0.039 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.801 0.797]\n",
      "global epoch:  1576\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1573 - accuracy: 0.9480\n",
      "training time:  11.8861 s.\n",
      "average model starts\n",
      "average model ends:  0.0322 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.802]\n",
      "global epoch:  1577\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1490 - accuracy: 0.9516\n",
      "training time:  11.1682 s.\n",
      "average model starts\n",
      "average model ends:  0.0358 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.8  ]\n",
      "global epoch:  1578\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1386 - accuracy: 0.9509\n",
      "training time:  11.2188 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.791]\n",
      "global epoch:  1579\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1617 - accuracy: 0.9473\n",
      "training time:  13.493 s.\n",
      "average model starts\n",
      "average model ends:  0.063 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.791 0.799]\n",
      "global epoch:  1580\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1487 - accuracy: 0.9485\n",
      "training time:  11.5213 s.\n",
      "average model starts\n",
      "average model ends:  0.0713 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.797]\n",
      "global epoch:  1581\n",
      "virtual memory used:  86.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1573 - accuracy: 0.9492\n",
      "training time:  11.9154 s.\n",
      "average model starts\n",
      "average model ends:  0.0734 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.803]\n",
      "global epoch:  1582\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1605 - accuracy: 0.9493\n",
      "training time:  11.0383 s.\n",
      "average model starts\n",
      "average model ends:  0.0847 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.803 0.796]\n",
      "global epoch:  1583\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1625 - accuracy: 0.9463\n",
      "training time:  11.9501 s.\n",
      "average model starts\n",
      "average model ends:  0.0356 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.804]\n",
      "global epoch:  1584\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1570 - accuracy: 0.9457\n",
      "training time:  13.0641 s.\n",
      "average model starts\n",
      "average model ends:  0.0648 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.804 0.794]\n",
      "global epoch:  1585\n",
      "virtual memory used:  86.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1544 - accuracy: 0.9472\n",
      "training time:  12.8791 s.\n",
      "average model starts\n",
      "average model ends:  0.045 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.794 0.798]\n",
      "global epoch:  1586\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1565 - accuracy: 0.9483\n",
      "training time:  12.2335 s.\n",
      "average model starts\n",
      "average model ends:  0.0492 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.798]\n",
      "global epoch:  1587\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 14s 57ms/step - loss: 0.1530 - accuracy: 0.9492\n",
      "training time:  14.4586 s.\n",
      "average model starts\n",
      "average model ends:  0.0635 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.798]\n",
      "global epoch:  1588\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1514 - accuracy: 0.9505\n",
      "training time:  11.7295 s.\n",
      "average model starts\n",
      "average model ends:  0.0335 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.798]\n",
      "global epoch:  1589\n",
      "virtual memory used:  89.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1627 - accuracy: 0.9469\n",
      "training time:  11.9367 s.\n",
      "average model starts\n",
      "average model ends:  0.0964 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.791]\n",
      "global epoch:  1590\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1582 - accuracy: 0.9489\n",
      "training time:  12.9301 s.\n",
      "average model starts\n",
      "average model ends:  0.1236 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.798]\n",
      "global epoch:  1591\n",
      "virtual memory used:  88.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1632 - accuracy: 0.9421\n",
      "training time:  11.751 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.797]\n",
      "global epoch:  1592\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1642 - accuracy: 0.9478\n",
      "training time:  12.4253 s.\n",
      "average model starts\n",
      "average model ends:  0.046 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.801]\n",
      "global epoch:  1593\n",
      "virtual memory used:  88.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1579 - accuracy: 0.9475\n",
      "training time:  11.0841 s.\n",
      "average model starts\n",
      "average model ends:  0.0471 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.799]\n",
      "global epoch:  1594\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1696 - accuracy: 0.9435\n",
      "training time:  10.899 s.\n",
      "average model starts\n",
      "average model ends:  0.0395 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.792]\n",
      "global epoch:  1595\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1594 - accuracy: 0.9478\n",
      "training time:  11.8569 s.\n",
      "average model starts\n",
      "average model ends:  0.0917 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.792 0.798]\n",
      "global epoch:  1596\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1462 - accuracy: 0.9547\n",
      "training time:  11.3327 s.\n",
      "average model starts\n",
      "average model ends:  0.0308 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.798 0.799]\n",
      "global epoch:  1597\n",
      "virtual memory used:  89.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1548 - accuracy: 0.9474\n",
      "training time:  13.5106 s.\n",
      "average model starts\n",
      "average model ends:  0.1242 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.793]\n",
      "global epoch:  1598\n",
      "virtual memory used:  89.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1582 - accuracy: 0.9503\n",
      "training time:  12.4888 s.\n",
      "average model starts\n",
      "average model ends:  0.1313 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.793 0.799]\n",
      "global epoch:  1599\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1633 - accuracy: 0.9444\n",
      "training time:  11.2755 s.\n",
      "average model starts\n",
      "average model ends:  0.0585 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.799 0.8  ]\n",
      "global epoch:  1600\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1577 - accuracy: 0.9463\n",
      "training time:  12.108 s.\n",
      "average model starts\n",
      "average model ends:  0.0343 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.794]\n",
      "global epoch:  1601\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1635 - accuracy: 0.9468\n",
      "training time:  11.0348 s.\n",
      "average model starts\n",
      "average model ends:  0.0519 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.805]\n",
      "global epoch:  1602\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1609 - accuracy: 0.9484\n",
      "training time:  11.88 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.805 0.797]\n",
      "global epoch:  1603\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1572 - accuracy: 0.9463\n",
      "training time:  11.2958 s.\n",
      "average model starts\n",
      "average model ends:  0.0403 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.797 0.794]\n",
      "global epoch:  1604\n",
      "virtual memory used:  82.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1524 - accuracy: 0.9467\n",
      "training time:  11.1509 s.\n",
      "average model starts\n",
      "average model ends:  0.0362 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.794 0.798]\n",
      "global epoch:  1605\n",
      "virtual memory used:  83.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1723 - accuracy: 0.9438\n",
      "training time:  11.8613 s.\n",
      "average model starts\n",
      "average model ends:  0.055 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.798 0.802]\n",
      "global epoch:  1606\n",
      "virtual memory used:  83.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1555 - accuracy: 0.9473\n",
      "training time:  12.7934 s.\n",
      "average model starts\n",
      "average model ends:  0.0627 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.798]\n",
      "global epoch:  1607\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1610 - accuracy: 0.9458\n",
      "training time:  12.3191 s.\n",
      "average model starts\n",
      "average model ends:  0.1259 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.798 0.802]\n",
      "global epoch:  1608\n",
      "virtual memory used:  82.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1666 - accuracy: 0.9469\n",
      "training time:  12.9759 s.\n",
      "average model starts\n",
      "average model ends:  0.0954 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.798]\n",
      "global epoch:  1609\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1695 - accuracy: 0.9428\n",
      "training time:  11.3824 s.\n",
      "average model starts\n",
      "average model ends:  0.0749 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.798 0.806]\n",
      "global epoch:  1610\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1576 - accuracy: 0.9479\n",
      "training time:  11.4218 s.\n",
      "average model starts\n",
      "average model ends:  0.0672 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.806 0.8  ]\n",
      "global epoch:  1611\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1476 - accuracy: 0.9505\n",
      "training time:  11.736 s.\n",
      "average model starts\n",
      "average model ends:  0.0415 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.8   0.795]\n",
      "global epoch:  1612\n",
      "virtual memory used:  84.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1491 - accuracy: 0.9500\n",
      "training time:  12.3315 s.\n",
      "average model starts\n",
      "average model ends:  0.1201 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.795 0.802]\n",
      "global epoch:  1613\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1624 - accuracy: 0.9500\n",
      "training time:  11.8384 s.\n",
      "average model starts\n",
      "average model ends:  0.0896 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.795]\n",
      "global epoch:  1614\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1477 - accuracy: 0.9519\n",
      "training time:  11.3008 s.\n",
      "average model starts\n",
      "average model ends:  0.0676 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.798]\n",
      "global epoch:  1615\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1662 - accuracy: 0.9474\n",
      "training time:  10.6243 s.\n",
      "average model starts\n",
      "average model ends:  0.0395 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.796]\n",
      "global epoch:  1616\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1516 - accuracy: 0.9516\n",
      "training time:  11.4444 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.796 0.803]\n",
      "global epoch:  1617\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1495 - accuracy: 0.9500\n",
      "training time:  10.6153 s.\n",
      "average model starts\n",
      "average model ends:  0.0604 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.8  ]\n",
      "global epoch:  1618\n",
      "virtual memory used:  84.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1572 - accuracy: 0.9484\n",
      "training time:  12.3234 s.\n",
      "average model starts\n",
      "average model ends:  0.0578 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.8   0.8  ]\n",
      "global epoch:  1619\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1475 - accuracy: 0.9518\n",
      "training time:  13.864 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.798]\n",
      "global epoch:  1620\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1478 - accuracy: 0.9525\n",
      "training time:  13.3898 s.\n",
      "average model starts\n",
      "average model ends:  0.0533 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.798 0.807]\n",
      "global epoch:  1621\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1447 - accuracy: 0.9523\n",
      "training time:  12.7331 s.\n",
      "average model starts\n",
      "average model ends:  0.1149 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.807 0.798]\n",
      "global epoch:  1622\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1654 - accuracy: 0.9437\n",
      "training time:  12.626 s.\n",
      "average model starts\n",
      "average model ends:  0.0405 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.798 0.797]\n",
      "global epoch:  1623\n",
      "virtual memory used:  85.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1572 - accuracy: 0.9493\n",
      "training time:  12.7158 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.8  ]\n",
      "global epoch:  1624\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1563 - accuracy: 0.9494\n",
      "training time:  12.9162 s.\n",
      "average model starts\n",
      "average model ends:  0.0487 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.797]\n",
      "global epoch:  1625\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1676 - accuracy: 0.9457\n",
      "training time:  12.8851 s.\n",
      "average model starts\n",
      "average model ends:  0.0684 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.797 0.796]\n",
      "global epoch:  1626\n",
      "virtual memory used:  85.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1545 - accuracy: 0.9520\n",
      "training time:  10.9753 s.\n",
      "average model starts\n",
      "average model ends:  0.0361 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.796 0.793]\n",
      "global epoch:  1627\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1462 - accuracy: 0.9523\n",
      "training time:  12.3772 s.\n",
      "average model starts\n",
      "average model ends:  0.0641 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.793 0.8  ]\n",
      "global epoch:  1628\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1646 - accuracy: 0.9488\n",
      "training time:  13.3887 s.\n",
      "average model starts\n",
      "average model ends:  0.057 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.8   0.8  ]\n",
      "global epoch:  1629\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1579 - accuracy: 0.9449\n",
      "training time:  12.2064 s.\n",
      "average model starts\n",
      "average model ends:  0.0893 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.8   0.801]\n",
      "global epoch:  1630\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1536 - accuracy: 0.9506\n",
      "training time:  13.2382 s.\n",
      "average model starts\n",
      "average model ends:  0.031 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.802]\n",
      "global epoch:  1631\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1495 - accuracy: 0.9508\n",
      "training time:  13.168 s.\n",
      "average model starts\n",
      "average model ends:  0.1357 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.8  ]\n",
      "global epoch:  1632\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1657 - accuracy: 0.9473\n",
      "training time:  13.0755 s.\n",
      "average model starts\n",
      "average model ends:  0.107 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.796]\n",
      "global epoch:  1633\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1578 - accuracy: 0.9468\n",
      "training time:  11.4706 s.\n",
      "average model starts\n",
      "average model ends:  0.0323 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.796 0.803]\n",
      "global epoch:  1634\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1675 - accuracy: 0.9452\n",
      "training time:  10.3757 s.\n",
      "average model starts\n",
      "average model ends:  0.0384 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.789]\n",
      "global epoch:  1635\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1609 - accuracy: 0.9472\n",
      "training time:  12.3206 s.\n",
      "average model starts\n",
      "average model ends:  0.0351 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.789 0.802]\n",
      "global epoch:  1636\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1516 - accuracy: 0.9490\n",
      "training time:  10.7263 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.802 0.806]\n",
      "global epoch:  1637\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1442 - accuracy: 0.9511\n",
      "training time:  12.7759 s.\n",
      "average model starts\n",
      "average model ends:  0.0449 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.806 0.798]\n",
      "global epoch:  1638\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1648 - accuracy: 0.9473\n",
      "training time:  12.4135 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.798 0.791]\n",
      "global epoch:  1639\n",
      "virtual memory used:  87.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1563 - accuracy: 0.9479\n",
      "training time:  10.3543 s.\n",
      "average model starts\n",
      "average model ends:  0.0741 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.791 0.799]\n",
      "global epoch:  1640\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1441 - accuracy: 0.9504\n",
      "training time:  11.1854 s.\n",
      "average model starts\n",
      "average model ends:  0.038 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.799 0.8  ]\n",
      "global epoch:  1641\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1627 - accuracy: 0.9500\n",
      "training time:  10.4886 s.\n",
      "average model starts\n",
      "average model ends:  0.0388 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.8   0.792]\n",
      "global epoch:  1642\n",
      "virtual memory used:  89.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1491 - accuracy: 0.9492\n",
      "training time:  11.1835 s.\n",
      "average model starts\n",
      "average model ends:  0.0372 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.792 0.804]\n",
      "global epoch:  1643\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1560 - accuracy: 0.9493\n",
      "training time:  11.349 s.\n",
      "average model starts\n",
      "average model ends:  0.1281 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.804 0.79 ]\n",
      "global epoch:  1644\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1653 - accuracy: 0.9452\n",
      "training time:  11.8784 s.\n",
      "average model starts\n",
      "average model ends:  0.0673 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.79  0.803]\n",
      "global epoch:  1645\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1599 - accuracy: 0.9466\n",
      "training time:  13.3068 s.\n",
      "average model starts\n",
      "average model ends:  0.0903 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.79  0.803 0.802]\n",
      "global epoch:  1646\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1746 - accuracy: 0.9458\n",
      "training time:  12.8379 s.\n",
      "average model starts\n",
      "average model ends:  0.098 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.799]\n",
      "global epoch:  1647\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1586 - accuracy: 0.9489\n",
      "training time:  12.3619 s.\n",
      "average model starts\n",
      "average model ends:  0.1126 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.798]\n",
      "global epoch:  1648\n",
      "virtual memory used:  88.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1638 - accuracy: 0.9452\n",
      "training time:  12.1023 s.\n",
      "average model starts\n",
      "average model ends:  0.0422 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.798 0.806]\n",
      "global epoch:  1649\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1643 - accuracy: 0.9439\n",
      "training time:  11.5747 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.806 0.799]\n",
      "global epoch:  1650\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1597 - accuracy: 0.9479\n",
      "training time:  12.0347 s.\n",
      "average model starts\n",
      "average model ends:  0.0692 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.799 0.797]\n",
      "global epoch:  1651\n",
      "virtual memory used:  83.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1540 - accuracy: 0.9500\n",
      "training time:  13.2782 s.\n",
      "average model starts\n",
      "average model ends:  0.1258 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.802]\n",
      "global epoch:  1652\n",
      "virtual memory used:  83.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1464 - accuracy: 0.9501\n",
      "training time:  13.489 s.\n",
      "average model starts\n",
      "average model ends:  0.0538 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.795]\n",
      "global epoch:  1653\n",
      "virtual memory used:  84.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1675 - accuracy: 0.9462\n",
      "training time:  11.7923 s.\n",
      "average model starts\n",
      "average model ends:  0.0525 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.799]\n",
      "global epoch:  1654\n",
      "virtual memory used:  84.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1501 - accuracy: 0.9508\n",
      "training time:  11.8461 s.\n",
      "average model starts\n",
      "average model ends:  0.0449 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.804]\n",
      "global epoch:  1655\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1617 - accuracy: 0.9468\n",
      "training time:  10.9817 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.804 0.795]\n",
      "global epoch:  1656\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1489 - accuracy: 0.9499\n",
      "training time:  13.3106 s.\n",
      "average model starts\n",
      "average model ends:  0.0392 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.795 0.801]\n",
      "global epoch:  1657\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1607 - accuracy: 0.9484\n",
      "training time:  11.9269 s.\n",
      "average model starts\n",
      "average model ends:  0.0407 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.799]\n",
      "global epoch:  1658\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1591 - accuracy: 0.9513\n",
      "training time:  12.7992 s.\n",
      "average model starts\n",
      "average model ends:  0.0501 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.804]\n",
      "global epoch:  1659\n",
      "virtual memory used:  85.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1587 - accuracy: 0.9501\n",
      "training time:  11.2551 s.\n",
      "average model starts\n",
      "average model ends:  0.0791 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.804 0.801]\n",
      "global epoch:  1660\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1518 - accuracy: 0.9506\n",
      "training time:  13.0584 s.\n",
      "average model starts\n",
      "average model ends:  0.0517 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.804]\n",
      "global epoch:  1661\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1661 - accuracy: 0.9447\n",
      "training time:  13.8776 s.\n",
      "average model starts\n",
      "average model ends:  0.0649 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.805]\n",
      "global epoch:  1662\n",
      "virtual memory used:  85.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1568 - accuracy: 0.9494\n",
      "training time:  12.9309 s.\n",
      "average model starts\n",
      "average model ends:  0.0643 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.805 0.796]\n",
      "global epoch:  1663\n",
      "virtual memory used:  86.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1573 - accuracy: 0.9521\n",
      "training time:  12.3366 s.\n",
      "average model starts\n",
      "average model ends:  0.1132 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.796 0.805]\n",
      "global epoch:  1664\n",
      "virtual memory used:  85.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 50ms/step - loss: 0.1526 - accuracy: 0.9524\n",
      "training time:  12.7159 s.\n",
      "average model starts\n",
      "average model ends:  0.0769 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.805 0.793]\n",
      "global epoch:  1665\n",
      "virtual memory used:  85.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1539 - accuracy: 0.9501\n",
      "training time:  12.117 s.\n",
      "average model starts\n",
      "average model ends:  0.0336 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.793 0.798]\n",
      "global epoch:  1666\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1363 - accuracy: 0.9546\n",
      "training time:  12.2787 s.\n",
      "average model starts\n",
      "average model ends:  0.0366 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.793 0.798 0.797]\n",
      "global epoch:  1667\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1555 - accuracy: 0.9508\n",
      "training time:  13.6016 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.797 0.804]\n",
      "global epoch:  1668\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1594 - accuracy: 0.9463\n",
      "training time:  14.0582 s.\n",
      "average model starts\n",
      "average model ends:  0.0407 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.804 0.801]\n",
      "global epoch:  1669\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1577 - accuracy: 0.9484\n",
      "training time:  10.7456 s.\n",
      "average model starts\n",
      "average model ends:  0.0859 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.786]\n",
      "global epoch:  1670\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 46ms/step - loss: 0.1706 - accuracy: 0.9436\n",
      "training time:  11.6048 s.\n",
      "average model starts\n",
      "average model ends:  0.0777 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.786 0.798]\n",
      "global epoch:  1671\n",
      "virtual memory used:  86.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1526 - accuracy: 0.9504\n",
      "training time:  12.112 s.\n",
      "average model starts\n",
      "average model ends:  0.0912 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.786 0.798 0.802]\n",
      "global epoch:  1672\n",
      "virtual memory used:  87.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1499 - accuracy: 0.9509\n",
      "training time:  12.2482 s.\n",
      "average model starts\n",
      "average model ends:  0.0794 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.801]\n",
      "global epoch:  1673\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1579 - accuracy: 0.9473\n",
      "training time:  12.2698 s.\n",
      "average model starts\n",
      "average model ends:  0.059 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.805]\n",
      "global epoch:  1674\n",
      "virtual memory used:  86.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1533 - accuracy: 0.9485\n",
      "training time:  11.7535 s.\n",
      "average model starts\n",
      "average model ends:  0.0325 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.805 0.796]\n",
      "global epoch:  1675\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1657 - accuracy: 0.9478\n",
      "training time:  11.616 s.\n",
      "average model starts\n",
      "average model ends:  0.0483 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.796 0.796]\n",
      "global epoch:  1676\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1607 - accuracy: 0.9459\n",
      "training time:  12.2752 s.\n",
      "average model starts\n",
      "average model ends:  0.0731 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.796 0.799]\n",
      "global epoch:  1677\n",
      "virtual memory used:  85.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1592 - accuracy: 0.9468\n",
      "training time:  10.6378 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.798]\n",
      "global epoch:  1678\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1513 - accuracy: 0.9499\n",
      "training time:  11.8055 s.\n",
      "average model starts\n",
      "average model ends:  0.0801 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.798 0.798]\n",
      "global epoch:  1679\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1512 - accuracy: 0.9500\n",
      "training time:  11.3852 s.\n",
      "average model starts\n",
      "average model ends:  0.0802 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.798 0.792]\n",
      "global epoch:  1680\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1556 - accuracy: 0.9464\n",
      "training time:  11.6964 s.\n",
      "average model starts\n",
      "average model ends:  0.0318 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.792 0.806]\n",
      "global epoch:  1681\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1593 - accuracy: 0.9475\n",
      "training time:  10.6926 s.\n",
      "average model starts\n",
      "average model ends:  0.0406 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.792 0.806 0.796]\n",
      "global epoch:  1682\n",
      "virtual memory used:  86.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1507 - accuracy: 0.9494\n",
      "training time:  11.8323 s.\n",
      "average model starts\n",
      "average model ends:  0.0455 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.796 0.807]\n",
      "global epoch:  1683\n",
      "virtual memory used:  87.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1538 - accuracy: 0.9495\n",
      "training time:  12.2788 s.\n",
      "average model starts\n",
      "average model ends:  0.0425 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.807 0.803]\n",
      "global epoch:  1684\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1399 - accuracy: 0.9521\n",
      "training time:  12.9915 s.\n",
      "average model starts\n",
      "average model ends:  0.0389 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.803 0.796]\n",
      "global epoch:  1685\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1561 - accuracy: 0.9493\n",
      "training time:  12.2294 s.\n",
      "average model starts\n",
      "average model ends:  0.0342 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.796 0.795]\n",
      "global epoch:  1686\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1608 - accuracy: 0.9479\n",
      "training time:  12.7377 s.\n",
      "average model starts\n",
      "average model ends:  0.0408 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.795 0.801]\n",
      "global epoch:  1687\n",
      "virtual memory used:  87.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1549 - accuracy: 0.9511\n",
      "training time:  13.5543 s.\n",
      "average model starts\n",
      "average model ends:  0.0327 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.801 0.794]\n",
      "global epoch:  1688\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 41ms/step - loss: 0.1598 - accuracy: 0.9467\n",
      "training time:  10.5684 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.806]\n",
      "global epoch:  1689\n",
      "virtual memory used:  88.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1568 - accuracy: 0.9472\n",
      "training time:  12.6946 s.\n",
      "average model starts\n",
      "average model ends:  0.0333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.806 0.8  ]\n",
      "global epoch:  1690\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1709 - accuracy: 0.9442\n",
      "training time:  11.1967 s.\n",
      "average model starts\n",
      "average model ends:  0.0474 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.8   0.804]\n",
      "global epoch:  1691\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1451 - accuracy: 0.9509\n",
      "training time:  13.8275 s.\n",
      "average model starts\n",
      "average model ends:  0.0955 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.802]\n",
      "global epoch:  1692\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1703 - accuracy: 0.9467\n",
      "training time:  12.7991 s.\n",
      "average model starts\n",
      "average model ends:  0.1493 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.802 0.798]\n",
      "global epoch:  1693\n",
      "virtual memory used:  89.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1494 - accuracy: 0.9523\n",
      "training time:  13.0049 s.\n",
      "average model starts\n",
      "average model ends:  0.0925 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.798 0.802]\n",
      "global epoch:  1694\n",
      "virtual memory used:  89.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1530 - accuracy: 0.9511\n",
      "training time:  13.0574 s.\n",
      "average model starts\n",
      "average model ends:  0.0341 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.802 0.799]\n",
      "global epoch:  1695\n",
      "virtual memory used:  90.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1476 - accuracy: 0.9516\n",
      "training time:  12.1985 s.\n",
      "average model starts\n",
      "average model ends:  0.0687 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.801]\n",
      "global epoch:  1696\n",
      "virtual memory used:  88.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1528 - accuracy: 0.9500\n",
      "training time:  12.3655 s.\n",
      "average model starts\n",
      "average model ends:  0.0726 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.801]\n",
      "global epoch:  1697\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1628 - accuracy: 0.9498\n",
      "training time:  12.3273 s.\n",
      "average model starts\n",
      "average model ends:  0.0386 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.8  ]\n",
      "global epoch:  1698\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1486 - accuracy: 0.9505\n",
      "training time:  10.8311 s.\n",
      "average model starts\n",
      "average model ends:  0.0354 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.794]\n",
      "global epoch:  1699\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1564 - accuracy: 0.9489\n",
      "training time:  12.3848 s.\n",
      "average model starts\n",
      "average model ends:  0.0447 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.794 0.797]\n",
      "global epoch:  1700\n",
      "virtual memory used:  90.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1613 - accuracy: 0.9479\n",
      "training time:  12.3834 s.\n",
      "average model starts\n",
      "average model ends:  0.0464 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.797 0.803]\n",
      "global epoch:  1701\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1532 - accuracy: 0.9510\n",
      "training time:  12.9669 s.\n",
      "average model starts\n",
      "average model ends:  0.1144 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.803 0.799]\n",
      "global epoch:  1702\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1580 - accuracy: 0.9477\n",
      "training time:  12.4564 s.\n",
      "average model starts\n",
      "average model ends:  0.0376 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.799 0.797]\n",
      "global epoch:  1703\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1504 - accuracy: 0.9506\n",
      "training time:  12.4328 s.\n",
      "average model starts\n",
      "average model ends:  0.0668 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.802]\n",
      "global epoch:  1704\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1594 - accuracy: 0.9447\n",
      "training time:  10.8825 s.\n",
      "average model starts\n",
      "average model ends:  0.1071 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.802 0.799]\n",
      "global epoch:  1705\n",
      "virtual memory used:  83.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1582 - accuracy: 0.9478\n",
      "training time:  11.0422 s.\n",
      "average model starts\n",
      "average model ends:  0.0766 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.799 0.807]\n",
      "global epoch:  1706\n",
      "virtual memory used:  84.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1556 - accuracy: 0.9487\n",
      "training time:  13.2061 s.\n",
      "average model starts\n",
      "average model ends:  0.0614 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.807 0.797]\n",
      "global epoch:  1707\n",
      "virtual memory used:  84.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1567 - accuracy: 0.9477\n",
      "training time:  13.0888 s.\n",
      "average model starts\n",
      "average model ends:  0.0744 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.797 0.804]\n",
      "global epoch:  1708\n",
      "virtual memory used:  85.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1568 - accuracy: 0.9468\n",
      "training time:  11.4588 s.\n",
      "average model starts\n",
      "average model ends:  0.0579 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.804 0.804]\n",
      "global epoch:  1709\n",
      "virtual memory used:  84.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1608 - accuracy: 0.9485\n",
      "training time:  12.1144 s.\n",
      "average model starts\n",
      "average model ends:  0.0957 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.804 0.797]\n",
      "global epoch:  1710\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1489 - accuracy: 0.9515\n",
      "training time:  13.3238 s.\n",
      "average model starts\n",
      "average model ends:  0.0827 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.797 0.8  ]\n",
      "global epoch:  1711\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1462 - accuracy: 0.95130s - loss: 0.1470 \n",
      "training time:  12.3428 s.\n",
      "average model starts\n",
      "average model ends:  0.0401 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.801]\n",
      "global epoch:  1712\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1456 - accuracy: 0.9534\n",
      "training time:  12.0457 s.\n",
      "average model starts\n",
      "average model ends:  0.0877 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.799]\n",
      "global epoch:  1713\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1463 - accuracy: 0.9528\n",
      "training time:  12.2555 s.\n",
      "average model starts\n",
      "average model ends:  0.0874 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.797]\n",
      "global epoch:  1714\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 10s 42ms/step - loss: 0.1549 - accuracy: 0.9495\n",
      "training time:  10.5604 s.\n",
      "average model starts\n",
      "average model ends:  0.041 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.803]\n",
      "global epoch:  1715\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1478 - accuracy: 0.9509\n",
      "training time:  11.4738 s.\n",
      "average model starts\n",
      "average model ends:  0.0563 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.803 0.802]\n",
      "global epoch:  1716\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1519 - accuracy: 0.9495\n",
      "training time:  11.3412 s.\n",
      "average model starts\n",
      "average model ends:  0.0555 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.796]\n",
      "global epoch:  1717\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1721 - accuracy: 0.9423\n",
      "training time:  11.5346 s.\n",
      "average model starts\n",
      "average model ends:  0.0818 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.798]\n",
      "global epoch:  1718\n",
      "virtual memory used:  85.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1588 - accuracy: 0.9484\n",
      "training time:  13.0875 s.\n",
      "average model starts\n",
      "average model ends:  0.0715 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.804]\n",
      "global epoch:  1719\n",
      "virtual memory used:  85.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1476 - accuracy: 0.9528\n",
      "training time:  13.1053 s.\n",
      "average model starts\n",
      "average model ends:  0.0647 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.804 0.805]\n",
      "global epoch:  1720\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1499 - accuracy: 0.9531\n",
      "training time:  12.8488 s.\n",
      "average model starts\n",
      "average model ends:  0.0379 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.805 0.799]\n",
      "global epoch:  1721\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1634 - accuracy: 0.9452\n",
      "training time:  12.1631 s.\n",
      "average model starts\n",
      "average model ends:  0.0823 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.799 0.802]\n",
      "global epoch:  1722\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1624 - accuracy: 0.9437\n",
      "training time:  12.383 s.\n",
      "average model starts\n",
      "average model ends:  0.0377 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.802]\n",
      "global epoch:  1723\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1496 - accuracy: 0.9504\n",
      "training time:  10.7849 s.\n",
      "average model starts\n",
      "average model ends:  0.0628 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.802 0.794]\n",
      "global epoch:  1724\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1564 - accuracy: 0.9499\n",
      "training time:  12.3224 s.\n",
      "average model starts\n",
      "average model ends:  0.0314 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.794 0.794]\n",
      "global epoch:  1725\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1630 - accuracy: 0.9466\n",
      "training time:  11.5455 s.\n",
      "average model starts\n",
      "average model ends:  0.0637 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.794 0.801]\n",
      "global epoch:  1726\n",
      "virtual memory used:  86.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1581 - accuracy: 0.9483\n",
      "training time:  12.1327 s.\n",
      "average model starts\n",
      "average model ends:  0.0429 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.801 0.796]\n",
      "global epoch:  1727\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1744 - accuracy: 0.9452\n",
      "training time:  12.8202 s.\n",
      "average model starts\n",
      "average model ends:  0.0623 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.803]\n",
      "global epoch:  1728\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1573 - accuracy: 0.9473\n",
      "training time:  12.1906 s.\n",
      "average model starts\n",
      "average model ends:  0.0934 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.803 0.806]\n",
      "global epoch:  1729\n",
      "virtual memory used:  87.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1538 - accuracy: 0.9509\n",
      "training time:  11.5057 s.\n",
      "average model starts\n",
      "average model ends:  0.1047 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.806 0.806]\n",
      "global epoch:  1730\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1513 - accuracy: 0.9510\n",
      "training time:  12.7819 s.\n",
      "average model starts\n",
      "average model ends:  0.0703 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.806 0.801]\n",
      "global epoch:  1731\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1504 - accuracy: 0.9503\n",
      "training time:  13.0191 s.\n",
      "average model starts\n",
      "average model ends:  0.048 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.801 0.8  ]\n",
      "global epoch:  1732\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1475 - accuracy: 0.9482\n",
      "training time:  13.0953 s.\n",
      "average model starts\n",
      "average model ends:  0.0818 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.8   0.806]\n",
      "global epoch:  1733\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1521 - accuracy: 0.9499\n",
      "training time:  10.9042 s.\n",
      "average model starts\n",
      "average model ends:  0.0326 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.806 0.797]\n",
      "global epoch:  1734\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1664 - accuracy: 0.9453\n",
      "training time:  11.108 s.\n",
      "average model starts\n",
      "average model ends:  0.0935 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.797 0.795]\n",
      "global epoch:  1735\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1573 - accuracy: 0.9484\n",
      "training time:  10.7949 s.\n",
      "average model starts\n",
      "average model ends:  0.0786 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.795 0.804]\n",
      "global epoch:  1736\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1519 - accuracy: 0.9526\n",
      "training time:  13.025 s.\n",
      "average model starts\n",
      "average model ends:  0.0624 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.804 0.801]\n",
      "global epoch:  1737\n",
      "virtual memory used:  90.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1539 - accuracy: 0.9511\n",
      "training time:  11.793 s.\n",
      "average model starts\n",
      "average model ends:  0.0427 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.801 0.797]\n",
      "global epoch:  1738\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1667 - accuracy: 0.9456\n",
      "training time:  11.7475 s.\n",
      "average model starts\n",
      "average model ends:  0.0478 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.797 0.8  ]\n",
      "global epoch:  1739\n",
      "virtual memory used:  89.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1603 - accuracy: 0.9492\n",
      "training time:  11.6494 s.\n",
      "average model starts\n",
      "average model ends:  0.0856 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.8   0.799]\n",
      "global epoch:  1740\n",
      "virtual memory used:  89.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1523 - accuracy: 0.9501\n",
      "training time:  11.5306 s.\n",
      "average model starts\n",
      "average model ends:  0.0648 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.799 0.796]\n",
      "global epoch:  1741\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1672 - accuracy: 0.9480\n",
      "training time:  13.0345 s.\n",
      "average model starts\n",
      "average model ends:  0.0745 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.798]\n",
      "global epoch:  1742\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1574 - accuracy: 0.9483\n",
      "training time:  12.5261 s.\n",
      "average model starts\n",
      "average model ends:  0.0392 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.798 0.8  ]\n",
      "global epoch:  1743\n",
      "virtual memory used:  89.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1562 - accuracy: 0.9509\n",
      "training time:  13.0224 s.\n",
      "average model starts\n",
      "average model ends:  0.0369 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.801]\n",
      "global epoch:  1744\n",
      "virtual memory used:  89.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1462 - accuracy: 0.9513\n",
      "training time:  11.6439 s.\n",
      "average model starts\n",
      "average model ends:  0.0723 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.801 0.784]\n",
      "global epoch:  1745\n",
      "virtual memory used:  90.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 55ms/step - loss: 0.1465 - accuracy: 0.9514\n",
      "training time:  13.9397 s.\n",
      "average model starts\n",
      "average model ends:  0.056 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.784 0.801]\n",
      "global epoch:  1746\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1578 - accuracy: 0.9489\n",
      "training time:  11.062 s.\n",
      "average model starts\n",
      "average model ends:  0.0507 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.784 0.801 0.804]\n",
      "global epoch:  1747\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1461 - accuracy: 0.9524\n",
      "training time:  10.6753 s.\n",
      "average model starts\n",
      "average model ends:  0.0844 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.804 0.791]\n",
      "global epoch:  1748\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1539 - accuracy: 0.9511\n",
      "training time:  10.9201 s.\n",
      "average model starts\n",
      "average model ends:  0.075 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.791 0.798]\n",
      "global epoch:  1749\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1547 - accuracy: 0.9511\n",
      "training time:  11.399 s.\n",
      "average model starts\n",
      "average model ends:  0.0592 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.791 0.798 0.806]\n",
      "global epoch:  1750\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1665 - accuracy: 0.9435\n",
      "training time:  11.3232 s.\n",
      "average model starts\n",
      "average model ends:  0.0395 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.806 0.802]\n",
      "global epoch:  1751\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1511 - accuracy: 0.9488\n",
      "training time:  12.1191 s.\n",
      "average model starts\n",
      "average model ends:  0.1109 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.802 0.801]\n",
      "global epoch:  1752\n",
      "virtual memory used:  90.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1619 - accuracy: 0.9504\n",
      "training time:  12.6246 s.\n",
      "average model starts\n",
      "average model ends:  0.0422 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.801 0.801]\n",
      "global epoch:  1753\n",
      "virtual memory used:  91.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1552 - accuracy: 0.9504\n",
      "training time:  11.579 s.\n",
      "average model starts\n",
      "average model ends:  0.0333 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.801 0.802]\n",
      "global epoch:  1754\n",
      "virtual memory used:  91.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1520 - accuracy: 0.9489\n",
      "training time:  11.1933 s.\n",
      "average model starts\n",
      "average model ends:  0.0765 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.802 0.795]\n",
      "global epoch:  1755\n",
      "virtual memory used:  84.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1526 - accuracy: 0.9510\n",
      "training time:  13.3016 s.\n",
      "average model starts\n",
      "average model ends:  0.042 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.795 0.802]\n",
      "global epoch:  1756\n",
      "virtual memory used:  84.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1561 - accuracy: 0.9503\n",
      "training time:  11.4388 s.\n",
      "average model starts\n",
      "average model ends:  0.0781 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.802 0.797]\n",
      "global epoch:  1757\n",
      "virtual memory used:  85.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1424 - accuracy: 0.9537\n",
      "training time:  12.8969 s.\n",
      "average model starts\n",
      "average model ends:  0.0712 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.797 0.798]\n",
      "global epoch:  1758\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1539 - accuracy: 0.9516\n",
      "training time:  11.9624 s.\n",
      "average model starts\n",
      "average model ends:  0.0337 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.798 0.799]\n",
      "global epoch:  1759\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1515 - accuracy: 0.9488\n",
      "training time:  12.1121 s.\n",
      "average model starts\n",
      "average model ends:  0.0371 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.806]\n",
      "global epoch:  1760\n",
      "virtual memory used:  85.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1578 - accuracy: 0.9499\n",
      "training time:  12.5958 s.\n",
      "average model starts\n",
      "average model ends:  0.0952 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.806 0.801]\n",
      "global epoch:  1761\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1548 - accuracy: 0.9505\n",
      "training time:  12.4628 s.\n",
      "average model starts\n",
      "average model ends:  0.0494 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.806 0.801 0.798]\n",
      "global epoch:  1762\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1497 - accuracy: 0.9515\n",
      "training time:  12.3203 s.\n",
      "average model starts\n",
      "average model ends:  0.0514 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.798 0.8  ]\n",
      "global epoch:  1763\n",
      "virtual memory used:  86.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1474 - accuracy: 0.9544\n",
      "training time:  12.3237 s.\n",
      "average model starts\n",
      "average model ends:  0.0958 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.8   0.804]\n",
      "global epoch:  1764\n",
      "virtual memory used:  85.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1504 - accuracy: 0.9504\n",
      "training time:  12.0087 s.\n",
      "average model starts\n",
      "average model ends:  0.0368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.804 0.805]\n",
      "global epoch:  1765\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1474 - accuracy: 0.9526\n",
      "training time:  11.9647 s.\n",
      "average model starts\n",
      "average model ends:  0.049 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.804 0.805 0.801]\n",
      "global epoch:  1766\n",
      "virtual memory used:  86.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1581 - accuracy: 0.9499\n",
      "training time:  10.8349 s.\n",
      "average model starts\n",
      "average model ends:  0.0994 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.801 0.794]\n",
      "global epoch:  1767\n",
      "virtual memory used:  86.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 43ms/step - loss: 0.1480 - accuracy: 0.9531\n",
      "training time:  11.0555 s.\n",
      "average model starts\n",
      "average model ends:  0.0451 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.794 0.807]\n",
      "global epoch:  1768\n",
      "virtual memory used:  86.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1519 - accuracy: 0.9506\n",
      "training time:  10.6878 s.\n",
      "average model starts\n",
      "average model ends:  0.075 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.794 0.807 0.801]\n",
      "global epoch:  1769\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1525 - accuracy: 0.9505\n",
      "training time:  11.4355 s.\n",
      "average model starts\n",
      "average model ends:  0.0584 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.807 0.801 0.796]\n",
      "global epoch:  1770\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - ETA: 0s - loss: 0.1435 - accuracy: 0.95 - 11s 44ms/step - loss: 0.1432 - accuracy: 0.9541\n",
      "training time:  11.3437 s.\n",
      "average model starts\n",
      "average model ends:  0.1368 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.801]\n",
      "global epoch:  1771\n",
      "virtual memory used:  86.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 11s 44ms/step - loss: 0.1674 - accuracy: 0.9432\n",
      "training time:  11.2343 s.\n",
      "average model starts\n",
      "average model ends:  0.0858 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.801 0.799]\n",
      "global epoch:  1772\n",
      "virtual memory used:  87.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1472 - accuracy: 0.9525\n",
      "training time:  11.6062 s.\n",
      "average model starts\n",
      "average model ends:  0.0795 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.799 0.801]\n",
      "global epoch:  1773\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1576 - accuracy: 0.9483\n",
      "training time:  12.0918 s.\n",
      "average model starts\n",
      "average model ends:  0.0877 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.801 0.803]\n",
      "global epoch:  1774\n",
      "virtual memory used:  88.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1596 - accuracy: 0.9462\n",
      "training time:  13.724 s.\n",
      "average model starts\n",
      "average model ends:  0.0586 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.803 0.802]\n",
      "global epoch:  1775\n",
      "virtual memory used:  87.4 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1500 - accuracy: 0.9498\n",
      "training time:  11.624 s.\n",
      "average model starts\n",
      "average model ends:  0.0456 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.796]\n",
      "global epoch:  1776\n",
      "virtual memory used:  88.0 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1537 - accuracy: 0.9489\n",
      "training time:  12.6169 s.\n",
      "average model starts\n",
      "average model ends:  0.0557 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.796 0.799]\n",
      "global epoch:  1777\n",
      "virtual memory used:  88.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1498 - accuracy: 0.9514\n",
      "training time:  12.8564 s.\n",
      "average model starts\n",
      "average model ends:  0.054 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.796]\n",
      "global epoch:  1778\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 45ms/step - loss: 0.1565 - accuracy: 0.9519\n",
      "training time:  11.4395 s.\n",
      "average model starts\n",
      "average model ends:  0.0483 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.796 0.789]\n",
      "global epoch:  1779\n",
      "virtual memory used:  87.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1531 - accuracy: 0.9489\n",
      "training time:  12.6762 s.\n",
      "average model starts\n",
      "average model ends:  0.0524 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.789 0.8  ]\n",
      "global epoch:  1780\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1566 - accuracy: 0.9492\n",
      "training time:  12.4026 s.\n",
      "average model starts\n",
      "average model ends:  0.1228 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.789 0.8   0.802]\n",
      "global epoch:  1781\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 11s 42ms/step - loss: 0.1464 - accuracy: 0.9500\n",
      "training time:  10.6644 s.\n",
      "average model starts\n",
      "average model ends:  0.0351 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.802 0.8  ]\n",
      "global epoch:  1782\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1675 - accuracy: 0.9437\n",
      "training time:  12.5307 s.\n",
      "average model starts\n",
      "average model ends:  0.0379 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.8   0.805]\n",
      "global epoch:  1783\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1599 - accuracy: 0.9474\n",
      "training time:  12.6877 s.\n",
      "average model starts\n",
      "average model ends:  0.0481 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.8   0.805 0.795]\n",
      "global epoch:  1784\n",
      "virtual memory used:  88.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 51ms/step - loss: 0.1516 - accuracy: 0.9531\n",
      "training time:  13.1149 s.\n",
      "average model starts\n",
      "average model ends:  0.0541 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.795 0.799]\n",
      "global epoch:  1785\n",
      "virtual memory used:  89.1 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1601 - accuracy: 0.9489\n",
      "training time:  12.5654 s.\n",
      "average model starts\n",
      "average model ends:  0.0473 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.799 0.797]\n",
      "global epoch:  1786\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1485 - accuracy: 0.9525\n",
      "training time:  11.6905 s.\n",
      "average model starts\n",
      "average model ends:  0.061 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.805]\n",
      "global epoch:  1787\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1462 - accuracy: 0.9509\n",
      "training time:  13.8107 s.\n",
      "average model starts\n",
      "average model ends:  0.0456 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.805 0.795]\n",
      "global epoch:  1788\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1567 - accuracy: 0.9470\n",
      "training time:  13.7053 s.\n",
      "average model starts\n",
      "average model ends:  0.0596 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.795 0.798]\n",
      "global epoch:  1789\n",
      "virtual memory used:  89.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 49ms/step - loss: 0.1546 - accuracy: 0.9482\n",
      "training time:  12.5633 s.\n",
      "average model starts\n",
      "average model ends:  0.1152 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.795 0.798 0.799]\n",
      "global epoch:  1790\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 14s 54ms/step - loss: 0.1484 - accuracy: 0.9493\n",
      "training time:  13.8235 s.\n",
      "average model starts\n",
      "average model ends:  0.1196 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.798 0.799 0.797]\n",
      "global epoch:  1791\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1495 - accuracy: 0.9474\n",
      "training time:  12.1132 s.\n",
      "average model starts\n",
      "average model ends:  0.1346 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.797 0.801]\n",
      "global epoch:  1792\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1486 - accuracy: 0.9492\n",
      "training time:  11.861 s.\n",
      "average model starts\n",
      "average model ends:  0.1683 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.797 0.801 0.796]\n",
      "global epoch:  1793\n",
      "virtual memory used:  89.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1574 - accuracy: 0.9506\n",
      "training time:  11.8831 s.\n",
      "average model starts\n",
      "average model ends:  0.0937 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.801 0.796 0.799]\n",
      "global epoch:  1794\n",
      "virtual memory used:  90.2 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "252/252 [==============================] - 13s 50ms/step - loss: 0.1649 - accuracy: 0.9452\n",
      "training time:  12.8445 s.\n",
      "average model starts\n",
      "average model ends:  0.1435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.796 0.799 0.802]\n",
      "global epoch:  1795\n",
      "virtual memory used:  90.8 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 53ms/step - loss: 0.1465 - accuracy: 0.9528\n",
      "training time:  13.4215 s.\n",
      "average model starts\n",
      "average model ends:  0.0424 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.799 0.802 0.805]\n",
      "global epoch:  1796\n",
      "virtual memory used:  91.5 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1591 - accuracy: 0.9483\n",
      "training time:  12.3364 s.\n",
      "average model starts\n",
      "average model ends:  0.0812 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.805 0.803]\n",
      "global epoch:  1797\n",
      "virtual memory used:  89.7 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 13s 52ms/step - loss: 0.1573 - accuracy: 0.9514\n",
      "training time:  13.1151 s.\n",
      "average model starts\n",
      "average model ends:  0.0435 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.803 0.802]\n",
      "global epoch:  1798\n",
      "virtual memory used:  90.3 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 47ms/step - loss: 0.1572 - accuracy: 0.9474\n",
      "training time:  12.0229 s.\n",
      "average model starts\n",
      "average model ends:  0.0593 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.803 0.802 0.805]\n",
      "global epoch:  1799\n",
      "virtual memory used:  90.9 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 46ms/step - loss: 0.1585 - accuracy: 0.9489\n",
      "training time:  11.641 s.\n",
      "average model starts\n",
      "average model ends:  0.064 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.802 0.805 0.8  ]\n",
      "global epoch:  1800\n",
      "virtual memory used:  91.6 %\n",
      "training starts model size:  8064\n",
      "training steps:  None\n",
      "Epoch 1/1\n",
      "252/252 [==============================] - 12s 48ms/step - loss: 0.1519 - accuracy: 0.9510\n",
      "training time:  12.13 s.\n",
      "average model starts\n",
      "average model ends:  0.0673 s.\n",
      "alliance_test_loss\n",
      "test_loss:  [0.202 0.364 0.409 ... 0.805 0.8   0.805]\n"
     ]
    }
   ],
   "source": [
    "parameters = {\"nr_global_iterations\": 1801, \"nr_local_iterations\":1, \"training_steps\": None, \"data_augmentation\": True, \"model_size_averaging\": True} \n",
    "training_loss, test_loss = fedavg_model.fit(parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "itterations:  1300 , tot epochs:  3101 , epochs_left:  1801\n"
     ]
    }
   ],
   "source": [
    "itterations = len(fedavg_model.test_loss)\n",
    "tot_epochs = int(1563*500/int(np.ceil(len(x_train_splits[0])/32)))\n",
    "epochs_left = tot_epochs - itterations\n",
    "print(\"itterations: \", itterations, \", tot epochs: \", tot_epochs, \", epochs_left: \", epochs_left)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib import pylab as plt\n",
    "import pickle\n",
    "pickle.dump(fedavg_model.test_loss,open('test_loss_3101e_bigestmember_12members_diffsizes.p','wb'))\n",
    "\n",
    "\n",
    "plt.plot(fedavg_model.test_loss)\n",
    "plt.savefig('accuracy_progress_b')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}